NI 43-101 TECHNICAL REPORT

Updated Preliminary Economic Assessment for the Gold Project,

CSA Global Report Nº R302.2019 Effective date: 30 April 2019 Signature Date: 29 August 2019

Qualified Persons Maria O’Connor, MAIG, MAusIMM (CSA Global) Gary Patrick, MAusIMM CP (Met) (CSA Global) David Muir, MAIG (CSA Global) Alex Veresezan, P.Eng. (CSA Global) Greg Trout, P.Eng. (AGP)

www.csaglobal.com DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Report prepared for

Client Name Dundee Precious Metals Inc. Project Name/Job Code DPMNIR01 Contact Name Ian Lipchak Contact Title Project Manager Dundee Precious Metals, 1 Adelaide Street East, Suite 500, P.O. Box 195, Toronto, Ontario, M5C Office Address 2V9, Canada

Report issued by CSA Global (UK) Ltd (UK Office) First Floor, Suite 2 Springfield House, Springfield Road, Horsham, West Sussex RH12 2RG CSA Global Office UNITED KINGDOM T +44 1403 255 969 F +44 1403 240 896 E [email protected]

Division Resources

Report information

File Name R302.2019 DPMNIR01 Dundee Timok NI 43-101 PEA 20190829.docx Last Edited 27 August 2019 Report Date 29 August 2019 Report Effective Date 30 April 2019 Report Status Final

Author/Qualified Person (QP) and Reviewer Signatures

Maria O’Connor “signed” Author and QP Signature: B.Sc. (Hons), MAIG at Horsham, UK

Gary Patrick “signed” Author and QP Signature: BSc, MAusIMM, CP (Met) at Horsham, UK

David Muir “signed” Author and QP Signature: B.Sc. (Hons), MAIG at Horsham, UK

“signed and sealed” Author and QP Alex Veresezan, M.Sc., P.Eng. Signature: at Toronto, ON, Canada

CSA Global Report Nº R302.2019 I DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Greg Trout, P.Eng. Author and QP Signature: “signed and sealed” B.Sc. Civil Engineering

Stan Wholley Peer review B.App. Sc. Grad Dip (Oen) Signature: “signed” MAIG MAICD

© Copyright 2019

CSA Global Report Nº R302.2019 II DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Certificate of Qualified Person – Maria O’Connor I, Maria O’Connor, as a Qualified Person of this Technical Report titled “NI 43-101 Technical Report – Updated Preliminary Economic Assessment for the Timok Project, Serbia” (the “Technical Report”) prepared for Dundee Precious Metals Inc. with an effective date of 30 April 2019 and report date of 29 August 2019, do hereby certify that: 1. I am a Principal Resource Geologist and Director of CSA Global (UK) Ltd and carried out this assignment for CSA Global (UK) Ltd, Springfield House, Springfield Road, Horsham, West Sussex, RH12 2RG, UK (telephone: +44 1403 255 969, email: [email protected]). 2. I hold a BSc (Hons) degree in Environmental Geochemistry from University College Dublin, Ireland (2004) and am a registered Member in good standing of the Australian Institute of Geoscientists (MAIG Membership Number 5931) and a Member of the Australasian Institute of Mining and Metallurgy (MAusIMM Membership Number 307704). I am familiar with NI 43-101 and, by reason of education, experience in exploration, evaluation and mining of gold deposits, and professional registration; I fulfil the requirements of a Qualified Person as defined in NI 43-101. My experience includes 13 years in mineral exploration and resource evaluation. 3. I visited the Timok Gold Project from 28 February to 1 March 2017 for two days. 4. I am responsible for the following sections of this Technical Report: Sections 2 to 10, 12.1.1, 12.1.4, 12.1.7, 12.2, 14, 23, 24, 25.1, 25.8.1, 25.8.2, 25.9.1, 26.1, 26.2, 26.8, 27 and associated sections in Section 1 (Executive Summary). I am responsible for their accuracy and validity. 5. I am independent of the issuer as described in Section 1.5 of NI 43-101. 6. I have had prior involvement with the property that is the subject of this Technical Report, having: a. visited the Project in 2017; b. updated the Mineral Resource in 2017 and 2018 and; c. co-authored the NI 43-101 Technical Report. Timok Gold Project, Serbia dated 31 March 2017 and the NI 43-101 Technical Report – Mineral Resource Estimate Update for the Timok Project, Serbia, dated 7 November 2018 with an effective date of 15 May 2018. 7. I have read NI 43-101 and the parts of the Technical Report I am responsible for have been prepared in compliance with N1 43-101. 8. As of the effective date of the Technical Report, to the best of my knowledge, information and belief, the parts of the Technical Report that I am responsible for contain all scientific and technical information that is required to be disclosed to make the Technical Report not misleading.

Dated this 29th day of August 2019 at Horsham, UK. “Signed” Maria O’Connor, BSc (Hons), MAIG Director/Principal Resource Geologist CSA Global UK Ltd

CSA Global Report Nº R302.2019 III DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Certificate of Qualified Person – David Muir I, David Muir, as a Qualified Person of this Technical Report titled “NI 43-101 Technical Report – Updated Preliminary Economic Assessment for the Timok Project, Serbia” (the “Technical Report”) prepared for Dundee Precious Metals Inc. with an effective date of 30 April 2019 and report date of 29 August 2019, do hereby certify that: 1. I am a Principal Data Geologist at CSA Global (UK) Ltd and carried out this assignment for CSA Global (UK) Ltd, Springfield House, Springfield Road, Horsham, West Sussex, RH12 2RG, UK (telephone: +44 1403 255 969, email: [email protected]). 2. I hold a BSc (Hons) degree in Geology from the University of Natal, Durban, South Africa and am a registered Member in good standing of the Australian Institute of Geoscientists (MAIG Membership Number 9102). I am familiar with NI 43-101 and, by reason of education, experience in exploration, evaluation and data management, and professional registration; I fulfil the requirements of a Qualified Person as defined in NI 43-101. My experience includes 11 continuous years in the exploration and mining industry. 3. I visited the Timok Gold Project from 28 February to 1 March 2017. 4. I am responsible for the following sections of this Technical Report: Section 11, 12.1.2, 12.1.3, 12.1.5, 12.1.6, 12.1.8 and the associated sections in Section 1 (Executive Summary). I am responsible for their accuracy and validity. 5. I am independent of the issuer as described in Section 1.5 of NI 43-101. 6. I have had prior involvement with the property that is the subject of this Technical Report, having: a. visited the Project in 2017; b. co-authored the NI 43-101 Technical Report. Timok Gold Project, Serbia dated 31 March 2017 and the NI 43-101 Technical Report – Mineral Resource Estimate Update for the Timok Project, Serbia, dated 7 November 2018 with an effective date of 15 May 2018. 7. I have read NI 43-101 and the parts of the Technical Report I am responsible for have been prepared in compliance with N1 43-101. 8. As of the effective date of the Technical Report, to the best of my knowledge, information and belief, the parts of the Technical Report that I am responsible for contain all scientific and technical information that is required to be disclosed to make the Technical Report not misleading.

Dated this 29th day of August 2019 at Horsham, UK. “Signed” David Muir, BSc (Hons), MAIG Principal Data Geologist CSA Global (UK) Ltd

CSA Global Report Nº R302.2019 IV DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Certificate of Qualified Person – Gary Patrick I, Gary Patrick, as a Qualified Person of this Technical Report titled “NI 43-101 Technical Report – Updated Preliminary Economic Assessment for the Timok Project, Serbia” (the “Technical Report”) prepared for Dundee Precious Metals Inc. with an effective date of 30 April 2019 and report date of 29 August 2019, do hereby certify that: 1. I am a Senior Associate Metallurgist at CSA Global (UK) Ltd and carried out this assignment for CSA Global (UK) Ltd, First Floor, Suite 2, Springfield House, Springfield Road, Horsham, West Sussex, RH12 2RG, UK (telephone +44 1403 255 969, email: [email protected]). 2. I hold a BSc. (Chemistry/Extractive Metallurgy) and am a registered Member of the Australasian Institute of Mining and Metallurgy (MAusIMM, CP, #108090). I am familiar with NI 43-101 and, by reason of education, experience in exploration, evaluation and mining of gold deposits and professional registration; I fulfil the requirements of a Qualified Person as defined in NI 43-101. My experience includes 25 years in operations, metallurgical testwork supervision, flowsheet development, and study work. 3. I visited the Timok Gold Project for three days from 15 to 17 November 2018. 4. I am responsible for the following sections of this Technical Report: Section 13, 17, 21.3, 21.8, 21.10, 21.11, 25.4, 25.8.3, 25.9.2, 26.5 and the associated sections in Section 1 (Executive Summary). I am responsible for their accuracy and validity. 5. I am independent of the issuer as described in Section 1.5 of NI 43-101. 6. I have had prior involvement with the property that is the subject of this Technical Report, having co- authored the NI 43-101 Technical Report – Mineral Resource Estimate Update for the Timok Project, Serbia, dated 7 November 2018 with an effective date of 15 May 2018. 7. I have read NI 43-101 and the parts of the Technical Report I am responsible for have been prepared in compliance with N1 43-101. 8. As of the effective date of the Technical Report, to the best of my knowledge, information and belief, the parts of the Technical Report that I am responsible for contain all scientific and technical information that is required to be disclosed to make the Technical Report not misleading.

Dated this 29th day of August 2019 at Horsham, UK.

“Signed”

Gary Patrick, BSc., MAusIMM, CP (Met) Senior Associate Metallurgist CSA Global (UK) Ltd

CSA Global Report Nº R302.2019 V DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Certificate of Qualified Person –Alex Veresezan, P.Eng. I, Alex Veresezan, P.Eng., as a Qualified Person of this Technical Report titled “NI 43-101 Technical Report – Updated Preliminary Economic Assessment for the Timok Project, Serbia” (the “Technical Report”) prepared for Dundee Precious Metals Inc. with an effective date of 30 April 2019 and report date of 29 August 2019, do hereby certify that: 1. I am a Principal Mining Engineer at CSA Global Consultants Canada Inc. and carried out this assignment for CSA Global Consultants Canada Inc., 15 Toronto Street, Toronto, Ontario, M5C 2E3, Canada (telephone: +1 416 368 7041, email: [email protected]). 2. I hold a MSc degree in Mining Engineering from the University of Petrosani, Petrosani, Romania and am a registered Member in good standing of the Professional Engineers Ontario (Membership Number 100078587). I am familiar with NI 43-101 and, by reason of education, experience in exploration, evaluation and data management, and professional registration; I fulfil the requirements of a Qualified Person as defined in NI 43-101. My experience includes 25 continuous years in the engineering and mining industry. 3. I have not visited the Timok Gold Project. 4. I am responsible for the following sections of this Technical Report: Sections 19, 22, 25.7 and the associated text in Section 1 (Executive Summary). I am responsible for their accuracy and validity. 5. I am independent of the issuer as described in Section 1.5 of NI 43-101. 6. I did not have prior involvement with the property that is the subject of this Technical Report. 7. I have read NI 43-101 and the parts of the Technical Report I am responsible for have been prepared in compliance with N1 43-101. 8. As of the effective date of the Technical Report, to the best of my knowledge, information and belief, the parts of the Technical Report that I am responsible for contain all scientific and technical information that is required to be disclosed to make the Technical Report not misleading.

Dated this 29th day of August 2019 at Toronto, Ontario, Canada.

“Signed and Sealed”

Alex Veresezan, M.Sc., P.Eng. Principal Mining Engineer CSA Global Consultants Canada Inc.

CSA Global Report Nº R302.2019 VI DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Certificate of Qualified Person – Greg Trout I, Greg Trout, P.Eng., as a Qualified Person of this Technical Report titled “NI 43-101 Technical Report – Updated Preliminary Economic Assessment for the Timok Project, Serbia” prepared for Dundee Precious Metals Inc. with an effective date of 30 April 2019 and report date of 29 August 2019, do hereby certify that: 1. I am employed as a Principal Mine Engineer with AGP Mining Consultants Inc. located at #246-132K Commerce Park Drive, Barrie Ontario L4N 0Z7. 2. I am a member in good standing of the Association of Professional Engineers and Geoscientists of Alberta, (Membership Number 97358). 3. I graduated from the University of Saskatchewan, (B.Sc. Civil Engineering) in 1982. 4. I have practiced my profession in the mining industry continuously since graduation. 5. I have been directly involved in the mining industry, in design, operation and evaluation of open pit mines, for more than 35 years. As a result of my experience and qualifications, I am a Qualified Person as defined in NI 43–101. 6. I have visited the project site from 15 to 17 November 2018. 7. I have read the definition of “Qualified Persons” set out in National Instrument 43–101 (NI 43-101) and the sections of the technical report for which I am responsible have been prepared in compliance with that Instrument and Form 43-101F1. 8. I am responsible for sections 15, 16, 18, 20, 21.1, 21.2, 21.4 to 21.7, 21.8, 21.9, 21.12, 25.2, 25.3, 25.5, 25.6, 25.8.4, 25.8.5, 25.9.3, 25.9.4, 26.3, 26.4, 26.6, 26.7 and the associated sections in Section 1 (Executive Summary) of the Technical Report. 9. I am independent of Dundee Precious Metals as described by Section 1.5 of the instrument. 10. I have had no previous involvement with the Timok Gold Project. 11. As of the effective date of the technical report, to the best of my knowledge, information, and belief, the sections of the technical report that I am responsible for, contain all scientific and technical information that is required to be disclosed to make those sections of the technical report not misleading.

Dated this 29th day of August 2019.

“Signed and Sealed”

Greg Trout, P.Eng. AGP Mining Consultants Inc.

CSA Global Report Nº R302.2019 VII DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Contents

Report prepared for ...... I Report issued by ...... I Report information ...... I Author/Qualified Person (QP) and Reviewer Signatures ...... I Certificate of Qualified Person – Maria O’Connor ...... III Certificate of Qualified Person – David Muir ...... IV Certificate of Qualified Person – Gary Patrick ...... V Certificate of Qualified Person –Alex Veresezan, P.Eng...... VI Certificate of Qualified Person – Greg Trout ...... VII

1 EXECUTIVE SUMMARY ...... 1 1.1 Introduction ...... 1 1.2 Project Description and Location ...... 1 1.3 Accessibility, Climate, Local Resources, Infrastructure and Physiography ...... 1 1.4 History...... 2 1.5 Geological Setting ...... 2 1.5.1 Regional Geology ...... 2 1.5.2 Project Geology ...... 2 1.6 Exploration and Drilling ...... 3 1.7 Sampling and Analysis ...... 4 1.8 Data Verification ...... 4 1.9 Metallurgy ...... 4 1.10 Mineral Resource Estimates ...... 5 1.11 Mining Methods ...... 7 1.12 Recovery Methods ...... 8 1.13 Project Infrastructure and Site Layout ...... 8 1.14 Environmental Studies, Permitting and Social or Community Impact ...... 9 1.14.1 Permitting ...... 9 1.14.2 Environmental and Social Constraints ...... 9 1.14.3 Social Licence...... 10 1.14.4 Mineral Wastes ...... 10 1.14.5 Mine Closure and Aftercare ...... 10 1.15 Capital and Operating Costs ...... 10 1.15.1 Capital Costs ...... 10 1.15.2 Operating Costs ...... 11 1.16 Economic Analysis...... 12 1.17 Interpretation and Conclusions ...... 13 1.17.1 Geology and Mineral Resource Estimates ...... 13 1.17.2 Geotechnical ...... 14 1.17.3 Mining ...... 14 1.17.4 Mineral Processing and Metallurgical Testing ...... 15 1.17.5 Infrastructure and Site Layout ...... 15

CSA Global Report Nº R302.2019 VIII DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

1.17.6 Environmental, Permitting, Social and Community...... 15 1.17.7 Economic Analysis ...... 15 1.17.8 Risks ...... 16 1.18 Recommendations ...... 17 1.18.1 Geology and Resources ...... 17 1.18.2 Geotechnical ...... 17 1.18.3 Mining ...... 17 1.18.4 Mineral Processing and Metallurgical Testing ...... 18 1.18.5 Environmental ...... 18 1.18.6 Estimated Budget ...... 18

2 INTRODUCTION ...... 19 2.1 Issuer...... 19 2.2 Terms of Reference...... 19 2.2.1 Scope of Work ...... 19 2.2.2 Principal Sources of Information ...... 20 2.2.3 Independence ...... 20 2.3 Qualified Person Section Responsibility ...... 20 2.4 Site Visits ...... 21 2.4.1 CSA Global Site Visits ...... 21 2.4.2 AGP Site Visit ...... 22 2.5 Units and Datum ...... 22 2.6 Report Effective Date:...... 22 2.7 Forward Looking Statements ...... 22

3 RELIANCE ON OTHER EXPERTS ...... 24

4 PROPERTY DESCRIPTION AND LOCATION ...... 25 4.1 Location ...... 25 4.2 Property Description ...... 26 4.2.1 Ownership ...... 27 4.3 Mineral Tenure ...... 28 4.4 Royalties ...... 28

5 ACCESSIBILITY, CLIMATE, LOCAL RESOURCES, INFRASTRUCTURE AND PHYSIOGRAPHY ...... 29 5.1 Accessibility ...... 29 5.2 Physiography ...... 29 5.3 Climate ...... 30 5.4 Infrastructure ...... 31

6 HISTORY ...... 32 6.1 Prior and Current Ownership ...... 32 6.2 Exploration History ...... 32

7 GEOLOGICAL SETTING AND MINERALISATION ...... 33 7.1 Regional Geology ...... 33 7.2 Regional Structural Geology ...... 34 7.3 Local Geology ...... 35

CSA Global Report Nº R302.2019 IX DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

7.4 Project Stratigraphy ...... 38 7.4.1 Palaeozoic and Proterozoic Basement ...... 38 7.4.2 Carbonate Sequence, JLS and KLS ...... 38 7.4.3 Calcareous Clastic Sedimentary Rocks, S1 and S2 ...... 39 7.4.4 Marl ...... 41 7.4.5 Andesitic Epiclastics and Diorite Intrusions ...... 42 7.4.6 Potaj Čuka Monzonite ...... 42 7.5 Structural Geology ...... 42 7.5.1 Structure ...... 42 7.5.2 Tectonic-Stratigraphic Relationships ...... 43 7.6 Metamorphism ...... 44 7.7 Alteration ...... 44 7.8 Mineralisation ...... 44 7.8.1 Bigar Hill Deposit ...... 46 7.8.2 Korkan Deposit ...... 47 7.8.3 Korkan West Deposit ...... 48 7.8.4 Kraku Pester ...... 49 7.9 Metallogeny and Paragenesis ...... 50 7.10 Weathering Profiles ...... 51

8 DEPOSIT TYPES ...... 53

9 EXPLORATION ...... 54 9.1 Introduction ...... 54 9.2 Geological Mapping ...... 54 9.3 Outcrop Sampling ...... 54 9.4 Soil Geochemistry ...... 55 9.5 Trenching ...... 57 9.6 Exploration Drilling ...... 59 9.7 Topographic Surveys ...... 59 9.8 Conclusions ...... 59

10 DRILLING ...... 60 10.1 Introduction ...... 60 10.2 Methodology and Planning, Site Preparation, Setup and Rehabilitation ...... 62 10.3 Collar and Downhole Surveying ...... 63 10.4 Drill-Hole Logging, Data Acquisition and Sampling ...... 63 10.4.1 Reverse Circulation Drilling ...... 63 10.4.2 Diamond Drilling ...... 64 10.5 Deposit Drilling ...... 65 10.6 2017/2018 Metallurgical Drill-Holes ...... 67

11 SAMPLE PREPARATION, ANALYSES AND SECURITY ...... 68 11.1 Field Sample Preparations ...... 68 11.1.1 Soil and Trench Samples ...... 68 11.1.2 Reverse Circulation Hole Samples ...... 68 11.1.3 Diamond Drill Core Hole Samples ...... 68

CSA Global Report Nº R302.2019 X DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

11.2 Laboratory Sample Preparation and Analyses ...... 69 11.2.1 Laboratory Sample Preparation ...... 69 11.2.2 Laboratory Analyses ...... 69 11.2.3 Bottle Roll Testwork Program ...... 70 11.2.4 Dry Bulk Density Measurements ...... 70 11.3 Avala Assay QAQC Procedures ...... 71 11.4 Conclusions and Recommendations ...... 71

12 DATA VERIFICATION ...... 73 12.1 Data Verification Completed by CSA Global ...... 73 12.1.1 Collar Locations ...... 73 12.1.2 Source Data Verification ...... 73 12.1.3 Database Validation ...... 73 12.1.4 Core Inspection ...... 74 12.1.5 QAQC Review ...... 74 12.1.6 Inspection of Procedures ...... 80 12.1.7 Twinned Hole Review ...... 80 12.1.8 Laboratory Audit...... 80 12.2 Conclusions ...... 80

13 MINERAL PROCESSING AND METALLURGICAL TESTING ...... 81 13.1 Metallurgical Testing Summary (2017 to 2019) ...... 81 13.2 Sample Selection and Representivity (2017 to 2019) ...... 82 13.3 SGS Testwork Program (2018) ...... 92 13.3.1 Head Assays ...... 93 13.3.2 Coarse Sample Bottle Roll Leach Tests ...... 95 13.3.3 Percolation Testing ...... 98 13.3.4 Column Leach Testing...... 99 13.3.5 Size-by-Size Analysis ...... 102 13.4 SGS Testwork Program (2019) ...... 104 13.4.1 Head Assays ...... 105 13.4.2 Coarse Sample Bottle Roll Leach Tests ...... 107 13.4.3 Column Leach Testing...... 109 13.5 COBR Test Results Summary (2019) ...... 112 13.6 Metallurgical Data Interpretation and Predictions (2018 to 2019) ...... 114 13.6.1 Preferred Process Option ...... 114 13.6.2 Predicted Metallurgical Recovery ...... 114 13.6.3 Predicted Reagent Consumption ...... 114 13.6.4 Predicted Leach Cycle Time ...... 115 13.7 Sulphide Mineralised Material Testwork (2012 to 2013) ...... 119 13.7.1 Mineralogical Characterisation (2012 to 2013) ...... 120 13.7.2 Flotation Testwork (2012 to 2013) ...... 121 13.8 Chlorination Laboratory Testwork (2016) ...... 127 13.9 Conclusions ...... 127 13.9.1 Oxide/Transitional Mineralised Material ...... 127 13.9.2 Sulphide Mineralised Material ...... 128

CSA Global Report Nº R302.2019 XI DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

14 MINERAL RESOURCE ESTIMATES ...... 129 14.1 Introduction ...... 129 14.2 2018 Mineral Resource Update ...... 129 14.3 Factors that may affect the Mineral Resource ...... 133 14.4 Previous Mineral Resource Estimates ...... 134

15 MINERAL RESERVE ESTIMATES ...... 136

16 MINING METHODS ...... 137 16.1 Overview ...... 137 16.2 Geologic Model Importation ...... 137 16.3 Geotechnical ...... 138 16.4 Economic Pit Shell Development ...... 138 16.5 Mine Planning ...... 141 16.5.1 Mine Design ...... 141 16.6 Mining Cut-Off Grades ...... 147 16.7 Resource Loss and Dilution ...... 147 16.8 Pit Phase Mill Feed Tonnages ...... 147 16.9 Mine Production Schedule ...... 148 16.10 Mine Rock Management ...... 149 16.11 Mine Equipment ...... 150

17 RECOVERY METHODS ...... 152 17.1 Heap Leach Facility ...... 152 17.1.1 Site Layout ...... 152 17.1.2 Heap Leach Pad Site Selection ...... 152 17.1.3 Process Description Summary ...... 153 17.1.4 Heap Leach Facility Design Criteria ...... 157 17.1.5 Heap Leach Facility Design Approach ...... 158 17.1.6 Heap Leach Facility Design ...... 159 17.1.7 Solution Application and Leaching ...... 161 17.1.8 Adsorption-Desorption-Recovery Plant...... 164 17.1.9 Process Reagents and Consumables ...... 171 17.2 Sulphide Concentrator ...... 173 17.2.1 Design Basis ...... 173 17.2.2 Process Design Criteria ...... 173 17.2.3 Production Criteria ...... 173 17.2.4 Metallurgical Design Criteria ...... 174 17.2.5 Unit Process Consumables ...... 174 17.2.6 Process Flowsheet and Plant Description ...... 175

18 PROJECT INFRASTRUCTURE ...... 183 18.1 Overall Site ...... 183 18.2 Roads ...... 183 18.3 Waste Storage Facilities ...... 183 18.4 Heap Leach Facility ...... 183 18.5 Water Management and Supply ...... 185

CSA Global Report Nº R302.2019 XII DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

18.6 Fuel Supply ...... 185 18.7 Camps and Buildings ...... 185 18.8 Power and Electrical ...... 185

19 MARKET STUDIES AND CONTRACTS ...... 186

20 ENVIRONMENTAL STUDIES, PERMITTING AND SOCIAL OR COMMUNITY IMPACT ...... 187 20.1.1 Regulatory Constraints ...... 187 20.1.2 Environmental and Social Constraints ...... 188 20.1.3 Social Licence...... 188 20.2 Permitting ...... 188 20.2.1 Permitting Requirements ...... 188 20.2.2 Current Status of Permitting ...... 189 20.3 Environmental and Social Studies, Setting and Issues ...... 197 20.3.1 Environmental and Social Studies ...... 197 20.3.2 Potential Environmental and Social Risks ...... 197 20.4 Social and Community Engagement ...... 203 20.4.1 Stakeholder Engagement ...... 203 20.5 Mineral Wastes ...... 203 20.5.1 Non-Mineral Wastes...... 203 20.6 Mine Closure and Aftercare ...... 204

21 CAPITAL AND OPERATING COSTS ...... 205 21.1 Capital Cost Summary ...... 205 21.2 Mine Capital Costs ...... 206 21.2.1 Pre-Production Stripping ...... 207 21.3 Process Capital Costs ...... 207 21.3.1 Direct Capital Costs ...... 207 21.4 Infrastructure Capital Costs ...... 208 21.5 Environmental Capital Costs ...... 209 21.6 Indirect Capital Costs ...... 209 21.7 Contingency ...... 209 21.8 Operating Costs Summary ...... 209 21.9 Mining Operating Costs ...... 210 21.10 Processing Operating Costs ...... 214 21.10.1 Heap Leach Facility ...... 214 21.10.2 Mill/Flotation Plant ...... 216 21.11 Tailings Management Operating Costs ...... 217 21.12 General and Administrative Operating Costs ...... 217

22 ECONOMIC ANALYSIS ...... 218 22.1 Caution to the Reader ...... 218 22.2 Model Assumptions ...... 218 22.3 Metal Pricing ...... 219 22.4 Mine Production Summary ...... 220 22.5 Operating and Capital Cost Summary ...... 221

CSA Global Report Nº R302.2019 XIII DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

22.6 Gold Doré and Concentrate Transport, Refining, Insurance and Treatment Charges ...... 222 22.7 Royalties and Taxation ...... 222 22.8 Working Capital and Cash Flow Treatment ...... 223 22.9 Results ...... 223 22.10 Sensitivity Analysis ...... 226 22.11 Operating Costs per Ounce of Gold ...... 228

23 ADJACENT PROPERTIES ...... 229

24 OTHER RELEVANT DATA AND INFORMATION ...... 230

25 INTERPRETATIONS AND CONCLUSIONS ...... 231 25.1 Geology and Resources ...... 231 25.2 Geotechnical ...... 232 25.3 Mining ...... 232 25.4 Mineral Processing ...... 232 25.5 Infrastructure and Site Layout ...... 233 25.6 Environmental, Permitting, Social and Community ...... 233 25.7 Economic Analysis...... 234 25.8 Risks ...... 234 25.8.1 General ...... 234 25.8.2 Mineral Resource Estimate ...... 234 25.8.3 Metallurgy/Mineral Processing ...... 235 25.8.4 Mining ...... 235 25.8.5 Environment, Permitting, Social and Community Risks ...... 235 25.9 Opportunities ...... 236 25.9.1 Geology/Mineral Resource Estimate ...... 236 25.9.2 Metallurgy/Mineral Processing ...... 236 25.9.3 Mining ...... 236 25.9.4 Environment, Permitting, Social and Community ...... 236

26 RECOMMENDATIONS ...... 237 26.1 Introduction ...... 237 26.2 Geology and Resources ...... 237 26.3 Geotechnical ...... 237 26.4 Mining ...... 238 26.5 Mineral Processing and Metallurgy ...... 238 26.6 Infrastructure ...... 238 26.7 Environmental ...... 239 26.8 Estimated Budget ...... 239

27 REFERENCES ...... 240

APPENDIX 1: GLOSSARY OF TECHNICAL TERMS AND ABBREVIATIONS ...... 241

CSA Global Report Nº R302.2019 XIV DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Figures Figure 1-1: Overall site layout ...... 9 Figure 4-1: Location map – Timok Gold Project ...... 25 Figure 4-2: Timok Gold Project exploration licences ...... 26 Figure 5-1: Typical landscape of Timok Gold Project, looking south towards Bigar Hill deposit ...... 29 Figure 5-2: Typical physiographic landscape and climatic contrasts (summer, top vs winter, bottom) – Bigar Hill project entrance ...... 30 Figure 7-1: Tectonics and chronology of the ABCD province ...... 33 Figure 7-2: Exploration licences with the TMC ...... 36 Figure 7-3: Schematic stratigraphy of the Western TMC ...... 37 Figure 7-4: Typical contact between Upper Jurassic (T) and Lower Cretaceous (V) Limestones with black chert nodules ...... 39 Figure 7-5: Typical S1 unit: Fine-grained calcirudite with stylolites ...... 40 Figure 7-6: Typical S2 unit: Conglomeratic sandstone with characteristic red fragmental clasts ...... 41 Figure 7-7: Example of Marl Unit: Grey marlstone with deformed laminations, drill-hole BHDD044, 59.2 m ...... 41 Figure 7-8: Example of andesite intrusive unit sill with phenocrysts of hornblende and plagioclase, drill-hole BHD010, 101 m ...... 42 Figure 7-9: Exploration areas and geology of TMC (inset: Mineral Resource plots of the deposits) ...... 45 Figure 7-10: Cross section of Bigar Hill deposit ...... 46 Figure 7-11: Mineralisation cross-sections of Korkan deposit (the Korkan East extension is shown in the cross-hatched area on the lower cross section) ...... 47 Figure 7-12: Cross-section of the Korkan West deposit ...... 49 Figure 7-13: Cross-section of the Kraku Pester deposit...... 50 Figure 7-14: SEM imaging of mineralised goethite taken from S1 sandstones in Bigar Hill (sample taken at 30.2 m depth from drill-hole BHDDMET001)...... 51 Figure 9-1: Location of soil sampling lines ...... 56 Figure 9-2: Trenching at the Timok Project ...... 58 Figure 10-1: Diamond (left) and RC drilling (right) at Bigar Hill ...... 60 Figure 10-2: Drilling completed at Bigar Hill (left) and Korkan (including Korkan West) (right) ...... 61 Figure 10-3: Drilling completed at Kraku Pester ...... 62 Figure 10-4: Cross section showing drilling and interpreted mineralisation at Bigar Hill ...... 65 Figure 10-5: Cross section showing drilling and interpreted mineralisation at Korkan ...... 66 Figure 10-6: Cross section showing drilling and interpreted mineralisation at Korkan West...... 66 Figure 10-7: Cross section showing drilling and interpreted mineralisation at Kraku Pester ...... 67 Figure 12-1: Results of Geostats gold CRM G308-8 showing failures ...... 75 Figure 12-2: Q-Q plot for silver external check assays (umpires) showing bias to duplicate sample) – pre-2017 samples ...... 78 Figure 13-1: Location of drill holes sampled for metallurgical bulk composite samples – all samples fall within the conceptual pit shells (light orange) used to constrain Mineral Resources ...... 84 Figure 13-2: Plan view of the location of the Korkan oxide, transitional and sulphide samples with drill holes and mineralisation outlines ...... 85 Figure 13-3: Section view of drill holes for Korkan oxide and transitional mineralisation samples (MET_KO_01/02)...... 86 Figure 13-4: Section view of drill hole for Korkan transitional mineralisation sample (KO_P1_01) ...... 86 Figure 13-5: Section view of drill holes for Korkan sulphide mineralisation sample (KO_P1_02) ...... 87 Figure 13-6: Plan view of the location of the Bigar Hill oxide, transitional and sulphide samples with drill holes and mineralisation outlines ...... 88 Figure 13-7: Section view of drill hole for Bigar Hill oxide mineralisation sample (Met18_BH_01) ...... 89 Figure 13-8: Section view of drill hole for Bigar Hill transitional mineralisation sample (BH_P1_01) ...... 89 Figure 13-9: Section view of drill hole for Bigar Hill oxide and transitional mineralisation samples (BH_P1_02/BH_P1_03) ...... 90 Figure 13-10: Section view of drill hole for Bigar Hill sulphide mineralisation sample (BH_P1_04) ...... 90 Figure 13-11: Plan view of the location of the Korkan West oxide and transitional mineralisation samples with drill hole and mineralisation outlines ...... 91 Figure 13-12: Section view of drill hole for Korkan West oxide mineralisation sample (MET18_KW_01) ...... 91 Figure 13-13: Section view of drill hole for Korkan West oxide mineralisation sample (KW_P1_01) ...... 92

CSA Global Report Nº R302.2019 XV DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Figure 13-14: Section view of drill hole for Korkan West oxide mineralisation sample (KW_P1_02) ...... 92 Figure 13-15: Coarse bottle roll test leach curves – Korkan oxide mineralisation sample ...... 97 Figure 13-16: Coarse bottle roll test leach curves – Korkan transitional mineralisation sample ...... 97 Figure 13-17: Coarse bottle roll test leach curves – Korkan West oxide mineralisation sample ...... 98 Figure 13-18: Coarse bottle roll test leach curves – Bigar Hill oxide mineralisation sample ...... 98 Figure 13-19: Column leach curves ...... 101 Figure 13-20: Residue assay/feed assay by size fraction ...... 102 Figure 13-21: Size-by-size recovery – Korkan oxide ...... 103 Figure 13-22: Size-by-size recovery – Korkan transitional ...... 103 Figure 13-23: Size-by-size recovery – Bigar Hill oxide ...... 104 Figure 13-24: Size-by-size recovery – Korkan West oxide ...... 104 Figure 13-25: Column leach curves ...... 111 Figure 13-26: Crush size vs %Au extraction ...... 113 Figure 13-27: %TS vs %Au extraction (all samples) ...... 113 Figure 13-28: Gold flux rate curves (oxide mineralised material) ...... 116 Figure 13-29: Gold flux rate curves (transitional) ...... 116 Figure 13-30: Field leach days vs %Au extraction (oxide) ...... 118 Figure 13-31: Field leach days vs %Au extraction (transitional) ...... 119 Figure 13-32: OCC grade-recovery curves ...... 125 Figure 16-1: Korkan West sensitivities ...... 140 Figure 16-2: Bigar Hill sensitivities ...... 140 Figure 16-3: Korkan pits sensitivities ...... 141 Figure 16-4: Korkan – Pit 1 design ...... 142 Figure 16-5: Korkan – Pit 2 design ...... 143 Figure 16-6: Korkan – Pit 3 design ...... 143 Figure 16-7: Korkan pit layouts ...... 144 Figure 16-8: Korkan West design ...... 144 Figure 16-9: Bigar Hill – Phase 1 design ...... 145 Figure 16-10: Bigar Hill – Phase 2 design ...... 146 Figure 16-11: Bigar Hill – Phase 3 design ...... 146 Figure 16-12: Mine tonnage by year and phase ...... 148 Figure 16-13: Heap and plant feed grade and recovered ounces ...... 149 Figure 16-14: Overall site layout with waste rock storage facilities ...... 150 Figure 17-1: Overall process flow diagram ...... 154 Figure 17-2: Crushing circuit configuration ...... 156 Figure 17-3: HLF general arrangement ...... 160 Figure 17-4: Leach solution irrigation schematic diagram...... 161 Figure 17-5: Overall HLF/ADR flow diagram ...... 165 Figure 17-6: Concentrator schematic ...... 176 Figure 18-1: HLF – cross section ...... 184 Figure 20-1: Permitting overview ...... 190 Figure 20-2: Watercourses and wells in Project area ...... 200 Figure 20-3: Bigar hydrograph ...... 200 Figure 20-4: Cultural heritage features ...... 202 Figure 22-1 Monthly gold price and three-year moving average before inflation ...... 220 Figure 22-2 Monthly gold price and three-year moving average in 2019 US$ ...... 220 Figure 22-3: Pre-tax sensitivity of Project IRR to changes in gold price and capital and operating costs ...... 226 Figure 22-4: Pre-tax sensitivity of Project NPV discounted at 5% to changes in gold price and capital and operating costs (US$ M) ...... 226 Figure 22-5: Post-tax sensitivity of Project IRR to changes in gold price and capital and operating costs ...... 227 Figure 22-6: Post-tax sensitivity of Project NPV discounted at 5% to changes in gold price and capital and operating costs (US$M) ...... 227

CSA Global Report Nº R302.2019 XVI DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Tables Table 1-1: Summary of gold recoveries used in the PEA ...... 5 Table 1-2: Mineral Resource reporting cut-off grades ...... 6 Table 1-3: MRE for Timok Gold Project as at 15 May 2018 ...... 7 Table 1-4: Capital cost estimate ...... 11 Table 1-5: Indirect costs and percentages ...... 11 Table 1-6: Contingency costs and percentages ...... 11 Table 1-7: Operating cost summary ...... 12 Table 1-8: PEA results overview ...... 13 Table 1-9: Summary of recommended budgets to complete PFS ...... 18 Table 2-1: Qualified Person section responsibility ...... 20 Table 4-1: Tenement details for Timok Gold Project exploration licences ...... 27 Table 9-1: Soil and outcrop sampling ...... 54 Table 10-1: Summary of drilling for the main resource areas of the Timok Gold Project ...... 60 Table 12-1: Ag standard results (absolute bias ≥5% in red) – pre-2017 samples ...... 76 Table 12-2: Duplicate types ...... 76 Table 12-3: Gold duplicate precision errors (with acceptable limits) – 2017/2018 samples ...... 77 Table 12-4: S duplicate precision errors – 2017/2018 samples...... 77 Table 12-5: Gold duplicate precision errors (with acceptable limits) – pre-2017 samples ...... 77 Table 12-6: Silver duplicate precision errors) – pre-2017 samples ...... 78 Table 12-7: Sulphur duplicate precision errors) – pre-2017 samples ...... 79 Table 12-8: Copper duplicate precision errors) – pre-2017 samples ...... 79 Table 13-1: Column leach testwork results summary (2018) ...... 81 Table 13-2: Column leach testwork results summary (2019) ...... 82 Table 13-3: Metallurgical drill-hole summary ...... 82 Table 13-4: Head assay results (2018) ...... 93 Table 13-5: Detailed head assay results (2018) ...... 93 Table 13-6: Summary of coarse sample bottle roll test results (2018) ...... 96 Table 13-7: Summary of column leach test results ...... 100 Table 13-8: Comparison of column leach test vs coarse bottle roll leach test results ...... 100 Table 13-9: Gold leach recoveries based on column calculated head grades ...... 101 Table 13-10: Gold residue assay/feed assay by size fraction ...... 102 Table 13-11: Head assay results (2019) ...... 105 Table 13-12: Detailed head assay results (2019) ...... 106 Table 13-13: Gold extraction from master composites 100% passing -16 mm ...... 107 Table 13-14: Summary of coarse sample bottle roll test results (2019) ...... 108 Table 13-15: Summary of column leach test results ...... 110 Table 13-16: Comparison of column leach test vs coarse bottle roll leach test results ...... 110 Table 13-17: Gold leach recoveries based on column calculated head grades ...... 111 Table 13-18: Oxide mineralised material coarse sample bottle roll test results ...... 112 Table 13-19: Transitional mineralised material coarse sample bottle roll test results ...... 112 Table 13-20: Sulphide mineralised material coarse sample bottle roll test results ...... 112 Table 13-21: Summary of discounted column leach test results ...... 114 Table 13-22: Gold recovery assumptions used in economic modelling ...... 114 Table 13-23: Summary of reagent consumptions (oxide samples) ...... 114 Table 13-24: Summary of reagent consumptions (transitional samples) ...... 115 Table 13-25: Metallurgical testwork sample summary (2013) ...... 122 Table 13-26: Rougher flotation optimisation results summary (2013) ...... 123 Table 13-27: Kraku Pester rougher flotation results (2013) ...... 124 Table 13-28: OCC test results summary ...... 124 Table 14-1: Parameters used in pit optimisations ...... 131

CSA Global Report Nº R302.2019 XVII DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Table 14-2: Mineral Resource reporting cut-off grades ...... 132 Table 14-3: Mineral Resource estimate for Timok Gold Project as at 15 May 2018 ...... 133 Table 14-4: MREs as at 31 March 2017 – Timok Gold Project, Serbia, CSA Global ...... 134 Table 14-5: Pit optimisation parameters used to constrain the 2017 MRE ...... 135 Table 16-1: LOM plan key results ...... 137 Table 16-2: Recommended wall slope design parameters ...... 138 Table 16-3: Open pit optimisation parameters ...... 138 Table 16-4: Timok pit phase design basis RFs ...... 139 Table 16-5: Open pit design parameters ...... 141 Table 16-6: Processing cut-off grades used ...... 147 Table 16-7: Feed by phase ...... 147 Table 16-8: Mine production schedule ...... 148 Table 16-9: Waste rock facility design parameters ...... 149 Table 16-10: Major mine equipment requirements ...... 151 Table 17-1: Process design criteria ...... 152 Table 17-2: Key crushing circuit design criteria ...... 157 Table 17-3: HLF general criteria ...... 159 Table 17-4: Key heap leaching design criteria ...... 162 Table 17-5: Key adsorption design criteria ...... 166 Table 17-6: Key production criteria (concentrator)...... 174 Table 17-7: Key metallurgical criteria ...... 174 Table 17-8: Unit process consumables ...... 175 Table 20-1: Key permits required for Timok Gold Project ...... 191 Table 20-2: Overview of permitting schedule to mine operation ...... 196 Table 20-3: Environmental and social studies completed and planned ...... 197 Table 20-4: Initial assessment of interactions between Project activities and environment ...... 198 Table 21-1: Capital cost estimate ...... 205 Table 21-2: Indirect costs and percentages ...... 205 Table 21-3: Contingency costs and percentages ...... 205 Table 21-4: Major mining equipment – capital cost and estimated equipment life ...... 206 Table 21-5: Mining capital cost estimate ...... 207 Table 21-6: Summary of process plant direct capital costs ...... 207 Table 21-7: Infrastructure capital ...... 208 Table 21-8: Indirect capital costs ...... 209 Table 21-9: Contingency cost by category and percentages ...... 209 Table 21-10: Operating cost summary ...... 209 Table 21-11: Reference mining costs ...... 210 Table 21-12: Mine staffing requirement and annual salaries ...... 210 Table 21-13: Average annual salaries ...... 211 Table 21-14: Maintenance labour factors (maintenance per operator) ...... 211 Table 21-15: Major equipment operating costs – no labour ($/hr) ...... 212 Table 21-16: Drill pattern specifications ...... 212 Table 21-17: Drill productivity criteria ...... 212 Table 21-18: Support equipment operating factors ...... 213 Table 21-19: Open pit mine base operating costs by pit area ($/tonne total material moved) ...... 214 Table 21-20: Operating cost estimate summary breakdown ...... 214 Table 21-21: Labour cost summary ...... 214 Table 21-22: Plant power cost estimate...... 215 Table 21-23: Summary of estimated consumable operating costs ...... 215 Table 21-24: Operating cost estimate summary breakdown ...... 216 Table 21-25: Plant labour cost summary...... 216 Table 21-26: Mill / Flotation Plant power cost estimate ...... 217

CSA Global Report Nº R302.2019 XVIII DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Table 21-27: Summary of estimated consumable operating costs ...... 217 Table 22-1: PEA results summary ...... 219 Table 22-2: Mineral Resources in PEA Mining Scenario ...... 221 Table 22-3: Project operating costs...... 221 Table 22-4: Project total capital costs ...... 222 Table 22-5: EBITDA and net profit to Project with tax credit ...... 224 Table 22-6: Project pre-tax results ...... 224 Table 22-7: Project post-tax results with tax credit ...... 224 Table 22-8: Summary DCF model ...... 225 Table 22-9: Sensitivity of Project at a base gold price of $1,250/oz ±30% post-tax with a 5% Serbian gold royalty ...... 228 Table 22-10: Operating cost per ounce of gold (World Gold Council definition) ...... 228 Table 26-1: Recommended metallurgical testwork budget ...... 238 Table 26-2: Summary of recommendation budgets for PFS ...... 239

CSA Global Report Nº R302.2019 XIX DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

1 Executive Summary

1.1 Introduction In April 2016, Dundee Precious Metals Inc. (DPM or the Company) completed the acquisition of 49.9% of the common shares of Avala Resources Ltd (Avala) not already owned by the Company, thus taking full ownership of, amongst others, the Timok Gold Project (Timok or the Project) in Serbia. Since that time, DPM has undertaken a series of exploration and evaluation programs to better understand the nature of the deposits culminating in the reporting of Mineral Resources. This Technical Report summarises the results of the exploration and evaluation programs completed to date and discloses the results of a Preliminary Economic Assessment (PEA) of the Timok Project. The PEA is based on the 2018 Mineral Resource reported in the previous NI 43-101 Technical Report on the Project filed on SEDAR on 7 November 2018. The PEA was commissioned to assess the viability of extracting gold from the oxide and transitional material via heap leaching as it had been established that a larger proportion of the Mineral Resource estimate contained oxide and transitional mineralisation than previously thought. The bulk of work outlined in this document is focused on standard open pit mining methods and heap leaching to extract the gold from oxide and transitional mineralisation. However, it was noted during mine planning that significant tonnes of gold- bearing sulphide mineralisation were included in the pit shells during the oxide and transitional mining and, as such, a small flotation plant was added to the processing facility to realise value from this material.

1.2 Project Description and Location The Project is located in the central-eastern region of the Republic of Serbia, approximately 270 km southeast of the capital, . It comprises three exploration licences (Potaj Čuka Tisnica, Umka and Bigar Istok licences) covering an aggregate area of 131.21 km2. The northern boundary is positioned about 25 km southwest from the River, and the Project area extends 24 km southwards to a point approximately 14 km west of Bor at its southern boundary. The Bigar Hill, Korkan, Korkan West and Kraku Pester deposits, which are the focus of this report, are located within the boundary of the Potaj Čuka Tisnica exploration licence. The exploration licences for the Project are held by Avala Resources d.o.o., a Serbian registered company. Since 2016, Avala Resources d.o.o., has been a wholly owned subsidiary of DPM following the acquisition of the remaining shares of Avala Resources Ltd. and its amalgamation with DPM.

1.3 Accessibility, Climate, Local Resources, Infrastructure and Physiography The Project is accessible by regional asphalt roads between Bor, Žagubica, Krepoljin, and Zlot, and well- developed unsealed forestry roads. The area is also linked via Bor to Zaječar and Paraćin and via Žagubica to Požarevac (and further to Belgrade). There is a railroad from Bor to Belgrade through Požarevac. Terrain in the Timok area is hilly to mountainous, ranging from about 500 metres above sea level (masl) to 944 masl at Čoka Rakita, the highest peak in the area. The most important drainage is the Jagnjilo River, which drains into the River Veliki Pek, and further on to the Danube, and incorporates the Bigar Hill and Korkan areas. The lower slopes and valleys are largely given over to seasonal farming, while forests dominate the higher slopes and peaks.

CSA Global Report Nº R302.2019 1 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

The Timok area is characterised by moderate continental climate, with some influence of high mountainous climate. Winters are long and cold, with abundant snow cover, and summers are usually hot. The Bigar Hill, Korkan and Korkan West, and Kraku Pester mineralised areas are located approximately 3 km, 4 km and 2 km respectively from the 110 kV Serbian national power grid, which extends from Bor to Petrovac and passes through the Project. The town of Bor is connected by rail to Belgrade (via Požarevac); this same rail network is part of European Transportation Corridor 10, which extends southwards through the Republic of North Macedonia to Greece and the Mediterranean, and also eastwards through to ports on the (and further to Turkey). Bor is accessible via the national highway grid (Paraćin turnpike), leading to sealed roads through Boljevac to Bor.

1.4 History The Timok region has a long history of exploration, dating back to Roman times. Geological mapping of the region was initiated in 1933 by the Serbian government, while regional geophysical surveys in the region were initially undertaken during the 1930s and then over various periods until 1985 by governmental geological and geophysical agencies. Geochemical surveys over the region were undertaken by Geozavod, Belgrade, and Geology Institute Bor. Small-scale adits were excavated in several localities prior to World War II.

1.5 Geological Setting

1.5.1 Regional Geology The Project is located immediately to the west of the Timok Magmatic Complex (TMC) in eastern Serbia. This complex is part of the larger tectonic Alpine-Balkan-Carpathian-Dinaride metallogenic-geodynamic province (ABCD) that extends from Western Europe to South-East Asia and comprises the Tethyan orogenic system. The most economically significant segment comprises the Late Cretaceous subduction-related magmatic rocks and mineral deposits, referred to as the as the Apuseni–Banat–Timok–Srednogorie Magmatic and Metallogenic Belt, which extends from Romania, through Serbia, and into Bulgaria. The easternmost magmatic complex in the Serbian part of the belt is the TMC. The Project is located on the western margin of the TMC and is subdivided into a western sequence of Proterozoic metamorphic basement rocks, Late Jurassic and Early Cretaceous limestones and an eastern sequence of epiclastic and diorite intrusive rocks of Late Cretaceous age. In the Project area, the interface between these two sequences consists of six rock units: Jurassic to Cretaceous limestone rocks (JLS and KLS units) are unconformably overlain by calcareous clastic sedimentary rocks consisting of a basal sandstone (S1 unit) and an overlying red sandstone or conglomerate unit (S2 unit), a marl unit which overlies the clastic units which, in turn, is overlain by magmatic and derivative clastic rocks (andesite epiclastic unit). The Jurassic and Cretaceous limestones are intensely karstified towards the upper unconformity.

1.5.2 Project Geology Gold mineralisation at the Project is classified as relatively low-temperature auriferous deposits that share many characteristics with Carlin-type gold deposits. The interpretation of the sediment-hosted gold prospects within the Project area as Carlin-type is based upon the following criteria: • The character of the sedimentary host; • The metal association of gold, arsenic, mercury, thallium, sulphur and antimony;

CSA Global Report Nº R302.2019 2 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

• The fine-grained nature of the gold, high gold-to-silver ratio and alteration types, including argillisation, decarbonisation, and locally, addition of quartz. Four important mineralised areas have been defined in the Potaj Čuka Tisnica exploration licence, comprising of the Bigar Hill deposit, the Korkan and Korkan West deposits and the Kraku Pester deposit. All four zones share a similarity of mineralisation style, which has been most clearly defined at the Bigar Hill deposit and are associated with a large hydrothermal system that has been identified within the Project. Gold mineralisation at Bigar Hill is located principally along two stratigraphic horizons, with lesser amounts present along peripheral steeply dipping fracture zones within the clastic rocks and an andesite sill. A lower zone is localised along the unconformable and brecciated lower contact between the clastic S1 and isolated karst-infill zones above the KLS unit. The most continuous horizons lie at shallow stratigraphic levels along the contact between the S1 and S2 units, forming a middle zone. Above this zone, gold mineralisation occurs within the andesite intrusive unit. Mineralisation at the Korkan deposit is generally southeast-northwest trending and shares similar characteristics with the Bigar Hill deposit. Unlike Bigar Hill, stratiform gold mineralisation at Korkan occurs primarily along the unconformable and breccia-like lower contact zone of the clastic S1 sequence against the underlying KLS limestone unit, and in karst-infill zones at the upper boundary of the KLS limestone unit. The Korkan West deposit is the newest discovery within the Project. It lies between the Bigar Hill and Korkan deposits, along a northwest trending structural corridor. The Korkan West deposit shares many characteristics with the Bigar Hill deposit, located approximately 1 km to the southeast, and the Korkan deposit located approximately 1 km to the northeast. Almost all mineralised intervals are manifested as oxide and transitional weathering states. Host rocks for gold mineralisation are: (1) oxidised fine to very coarse-grained (0.1 mm to 2 mm) sandstone belonging to the S1 or S2 units; (2) conglomerate layers containing quartzite clasts and/or not limestone clasts (S1 or S2 units). Mineralisation at S2/S1 contact can commonly be observed. The Kraku Pester deposit is located in an embayment at the north-western tip of the Potaj Čuka monzonite, consisting of a thermal aureole across a variably disrupted stratigraphic sequence of metamorphosed shale, marls and limestone metamorphosed to calc-silicate phyllite and marble, and tuffaceous rocks. Unlike Bigar Hill, gold mineralisation at Kraku Pester is hosted in brittle fault rocks composed of pyritised fault breccia to cataclasite, with relatively higher gold concentrations being associated with finer-grained cataclasite. Gold deposition is interpreted as being relatively late in the geological-structural evolution, post-dating the emplacement of the monzonite.

1.6 Exploration and Drilling Exploration of the Project has been carried out since 2007 by DPM and Avala teams. Extensive soil sampling and surface trenching was completed by DPM from 2007 to 2009 and four diamond drill-holes were completed on the Project during this period. From 2010 onwards, Avala completed geological mapping, outcrop sampling, soil geochemistry surveys and trenching over a large part of the Project. Exploration completed on Avala licences produced 11,683 soil samples, 2,104 rock chip samples, and 35.5 km of trenching. Drilling campaigns from 2010 to 2012 have been focused on the Potaj Čuka Tisnica licence to outline mineralisation across the Bigar Hill, Korkan, and Kraku Pester mineralised areas. Avala has completed 369 diamond drill-holes (100,936 m), 722 reverse circulation (RC) drill-holes (136,053 m) and 47 drill-holes (14,018 m) that comprised a RC pre-collar and diamond tail on the three deposits in this licence.

CSA Global Report Nº R302.2019 3 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

1.7 Sampling and Analysis Sample preparation for all samples (soil, channel sample, RC and diamond core) is undertaken at SGS Bor (SGS) sample preparation facility in Bor. This facility is owned by Avala, but independently managed by SGS, such that the chain of custody is transferred from Avala to SGS at the laboratory door. The SGS facility is located adjacent to Avala’s core shed facilities in Bor. All submissions to the sample preparation facility have appropriate chain of custody records, which are maintained until reject sample pulps are returned to Avala’s jurisdiction. Routine analyses of all samples are currently performed at the SGS analytical laboratory in Bor, or previously at the SGS analytical laboratory in Chelopech, Bulgaria. All laboratory methods, procedures, and quality control/quality assurance (QAQC) protocols are consistent with standards adopted by SGS worldwide. Gold analysis methodology is conventional 50 g fire assay, with an atomic absorption finish. Silver and base metal analyses are performed by aqua regia digestion and atomic absorption analysis. Sulphur samples are analysed by combustion with an infrared finish. The Bor and Chelopech laboratories, however, are not ISO 9002 or ISO 17025 accredited for the above analytical procedures. All soil samples are assayed by ALS Chemex in Perth and SGS Vancouver, using methods Au-TL43 (gold by aqua regia digestion with inductively coupled plasma and mass spectrometry (ICP-MS)), and ME-MS41 (multi- elements by aqua regia digestion with combined ICP-MS and ICP-AES (atomic emission spectrometry). The ALS Chemex laboratory in Perth is certified to ISO 9002, but is not ISO 17025 accredited for this technique. CSA Global concludes that the sample preparation, security and analytical procedures are robust and in line with industry best practices. The Avala QAQC procedures are comprehensive and should be suitable to monitor assay contamination, accuracy and precision.

1.8 Data Verification CSA Global completed a site visit on 28 February and 1 March 2017. Qualified Persons Maria O’Connor and David Muir undertook verification of aspects related to geology and mineral resource estimation. CSA Global staff again visited the site in November 2018. Qualified Persons Greg Trout and Gary Patrick visited to site to meet with key technical staff of DPM and to review data related to metallurgy and mining. CSA Global is satisfied that DPM (and before them, Avala) has conducted its exploration and evaluation programs at a high standard, and data verification completed by CSA Global has indicated no issues. CSA Global therefore concludes that data derived from them can be reliably used in the Mineral Resource estimate (MRE) and to support the PEA.

1.9 Metallurgy Coarse sample bottle roll and column leach tests have been carried out on representative oxide and transitional samples representing material from the Bigar Hill, Korkan and Korkan West deposits. Variables tested included crush size, leach cycle time, and cyanide concentration. Comparative results of coarse bottle roll leach and column leach tests are summarised in Table 1-1. Results of column leach tests showed that the oxide and transitional material types are amenable to processing using heap leach technology, at a fairly coarse crush size (80% passing 12.5 mm). Gold leach extraction appears to be dependent on sulphide sulphur content, and independent of crush size up to 1 inch (25 mm).

CSA Global Report Nº R302.2019 4 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Coarse sample bottle roll tests carried out on representative sulphide material resulted in low gold extraction ranging from 6% to 7%, at a crush size of 80% passing 12.5 mm. These results show the refractory nature of the sulphide material. Various flotation test programs carried out from 2011 to 2013 indicate that gold recoveries to a bulk sulphide concentrate ranged from 68% to 80%, achieving a concentrate grade ranging from 30 g/t to 50 g/t Au. At these concentrate grades, the gold-bearing concentrate would be deemed to be saleable to smelters/roasters. Table 1-1: Summary of gold recoveries used in the PEA Domain Material type Average Bigar Hill Korkan Korkan West Oxide material (to doré) 91% 91% 73% 88% Transitional material (to doré) 69% 69% 69% 69% Sulphide material (to sulphide concentrate) 75% 75% 75% 75%

1.10 Mineral Resource Estimates CSA Global has previously reported a Mineral Resource estimation for the Bigar Hill, Korkan and Korkan West, and Kraku Pester deposits dated 7 November 2018 (CSA Global, 2018). The 2018 MRE remains current and a summary of all relevant methodology, parameters and key assumptions regarding the preparation of the MRE is reported in Section 14. The MRE was based on interpretations using integrated geological and grade information recorded from RC and diamond core logging and assaying. DPM geologists conducted the geological interpretation and modelling work using the Leapfrog software package. CSA Global reviewed these models and found them suitable for use in the MRE. The estimation work was completed using the Datamine Studio and Isatis software packages by CSA Global. The effective date of the MRE is 15 May 2018. Mineral Resources are reported constrained within conceptual pit optimisation shells for each deposit, for the purposes of demonstrating “reasonable prospects for eventual economic extraction” (CIM, 2014). Mineral Resources that are not Mineral Reserves do not have demonstrated economic viability. The open-pit shells have been determined via consideration of various cut-off grades for material types that were calculated based upon, among other things, the material type, haulage distance and recoveries derived from metallurgical testwork. The Mineral Resource statement for each deposit and the cut-off grades for each material type, reported using these various cut-off grades, is presented in Table 1-2 and

CSA Global Report Nº R302.2019 5 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Table 1-3. Table 1-2: Mineral Resource reporting cut-off grades Cut-off grade (Au g/t)

Deposit Cut-off for Cut-off for Cut-off for Cut-off for Cut-off for Cut-off for oxide in oxide transitional in transitional sulphide in sulphide Whittle rounded Whittle rounded Whittle rounded Bigar Hill 0.178 0.20 0.235 0.25 0.603 0.60 Korkan 0.178 0.20 0.235 0.25 0.65 0.65 Korkan West 0.223 0.20 0.235 0.25 0.65 0.65 Kraku Pester 0.351 0.35 0.369 0.40 1.065 1.05

CSA Global Report Nº R302.2019 6 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Table 1-3: MRE for Timok Gold Project as at 15 May 2018 Indicated Mineral Resource Inferred Mineral Resource Deposit Material Type Tonnage Au Tonnage Au (Mt) (g/t) koz (Mt) (g/t) koz Oxide 12.4 1.14 455 0.7 0.7 16 Transitional 5.9 1.21 229 0.4 1.0 12 Bigar Hill Sulphide 11.1 1.72 615 0.1 1.6 7 Subtotal 29.4 1.38 1,299 1.2 0.9 34 Oxide 5.8 0.90 166 0.2 0.5 4 Transitional 2.8 1.06 97 0.1 0.7 3 Korkan Sulphide 3.3 1.91 205 0.0 1.1 0 Subtotal 11.9 1.22 468 0.4 0.6 7 Oxide 2.9 1.03 98 1.0 0.8 24 Transitional 0.3 0.85 8 0.2 0.8 6 Korkan West Sulphide 0.0 1.33 1 0.0 0.9 0 Subtotal 3.2 1.02 106 1.2 0.8 31 Oxide 0.7 0.95 22 0.1 1.3 5 Transitional 0.1 0.95 4 0.0 1.2 0 Kraku Pester Sulphide 1.5 2.01 95 0.0 1.8 0 Subtotal 2.3 1.61 122 0.1 1.3 6 Total – Oxide 21.8 1.06 742 2.0 0.7 48 Total – Transitional 9.2 1.15 338 0.7 0.9 22 Total – Sulphide 15.9 1.79 916 0.2 1.5 8 GRAND TOTAL 46.9 1.32 1,996 2.9 0.8 78 Notes: • The effective date of the MREs is 15 May 2018. • Mineral Resources are reported in accordance with CIM guidelines. • A cut-off of 0.20 g/t Au for the oxide material, 0.25 g/t Au for the transitional material, and 0.60 g/t Au for the sulphide material is applied at Bigar Hill. • A cut-off of 0.20 g/t Au for the oxide material, 0.25 g/t Au for the transitional material, and 0.65 g/t Au for the sulphide material is applied at Korkan and Korkan West. • A cut-off of 0.35 g/t Au for the oxide material, 0.40 g/t Au for the transitional material, and 1.05 g/t Au for the sulphide material is applied at Kraku Pester. • Figures have been rounded to the appropriate level of precision for the reporting of Mineral Resources. • Due to rounding, some columns or rows may not compute exactly as shown. • The Mineral Resources are stated as in situ dry tonnes. All figures are in metric tonnes. • The models are reported above surfaces based on conceptual US$1,400 gold price pit shells to support assumptions relating to reasonable prospects of eventual economic extraction. • Mineral Resources that are not Mineral Reserves do not have demonstrated economic viability. The estimate of Mineral Resources may be materially affected by environmental, permitting, legal, title, taxation, socio-political, marketing, or other relevant issues.

1.11 Mining Methods Readers are cautioned that the PEA is preliminary in nature. It includes Inferred Mineral Resources considered too speculative geologically to have the economic considerations applied to them that would enable them to be categorised as Mineral Reserves, and there is no certainty the PEA will be realised. Mineral Resources that are not Mineral Reserves do not have demonstrated economic viability. The estimate of Mineral Resources

CSA Global Report Nº R302.2019 7 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

may be materially affected by environmental, permitting, legal, title, socio-political, marketing or other relevant issues. The PEA study focused on the potential development of the Timok Project, targeting primarily the oxide portions of the deposit with higher gold recoveries. Three distinct pit areas are designed to mine the three deposits considered for this PEA: Bigar Hill, Korkan and Korkan West. These pits are estimated to provide a total of 18.9 million tonnes (Mt) of feed material grading 1.36 g/t Au over a nine-year mine life. These pits are also estimated to produce 49.7 Mt of waste for a delivered strip ratio of 2.63:1 (waste:feed). The feed material is comprised of 15.4 Mt of oxide and transitional material grading 1.18 g/t delivered to a heap leach pad, plus 3.5 Mt of sulphide material grading 2.17 g/t delivered to a small flotation concentrator facility. Conventional mining practices will be employed in developing the Timok pits. Drilling will use 200 mm rotary blasthole drills. Loading of mill feed and waste will utilise two 13 m3 production loaders and two 6.7 m3 hydraulic excavators. The haulage fleet will peak at 12 rigid body trucks of 63-t capacity. Normal support equipment including track dozers, graders and water trucks will be part of the mine equipment fleet. Three waste dump locations are envisioned to accommodate the mine waste: one east of Bigar Hill, one south of the Korkan pits and one west of Korkan West. These facilities are designed to accommodate the full amount of waste and total 35.6 million cubic metres (Mm3) of capacity: 25.5 Mm3 from Bigar Hill, 8.2 Mm3 from Korkan, and 1.9 Mm3 from Korkan West. These facilities can be expanded in the future, should it be required. The waste material is assumed to be non-acid generating (NAG) for this study, based on the abundance of carbonates in the rock, but additional testing is required to confirm that assumption.

1.12 Recovery Methods The flowsheet developed for processing oxide and transitional material from the Timok deposit is that of heap leach technology. The process flowsheet will involve three-stage crushing to produce a product size of 80% passing 12.5 mm. Crushed material will then be trucked to the heap leach pad and stacked in 8 m lifts. The heap will then be irrigated with dilute cyanide solution and gold leached from the stacked material. Leached solution will flow into the pregnant solution pond from where it will be pumped through carbon contactors, whereby the gold will be adsorbed onto activated carbon. Periodically, the loaded carbon will be sent to elution where the gold will be stripped from the carbon at high temperature and pressure. The pregnant solution will then be sent to electrowinning whereby gold will be plated onto steel wool cathodes. Steel wool will then be harvested, calcined, mixed with fluxes and smelted to produce doré. During the mining of oxide and transitional material, fresh, gold-bearing sulphide mineralisation will also be mined. The sulphide material will be treated by bulk sulphide flotation to produce a gold-bearing sulphide concentrate, which can subsequently be sold to smelters/roasters for an agreed payable on gold metal content.

1.13 Project Infrastructure and Site Layout The overall site plan is shown in Figure 1-1 and includes major facilities of the Project including the open pits, plant/shops, waste storage facilities, heap leach facility and mine access roads.

CSA Global Report Nº R302.2019 8 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Grid power will be provided by a new 5 km power line that connects to the adjacent high voltage power line to the east of the Project. Raw water will be drawn from wells in the area. The heap leach facility will be constructed to the east of the plant location within a valley. Filtered tails from the sulphide mill will also be stored in this facility.

Figure 1-1: Overall site layout Source: CSA Global 2019

1.14 Environmental Studies, Permitting and Social or Community Impact

1.14.1 Permitting The Serbian regulatory and permitting system requires a range of permits and permissions to be issued for mining projects. A permitting schedule has been developed, which aligns with Serbian requirements and international good practice. The permitting strategy will need to be flexible to allow for anticipated changes as the Serbian permitting system evolves to align with European Union (EU) requirements.

1.14.2 Environmental and Social Constraints The Timok Gold Project is located in a rural, hilly area with steep valleys, characterised by seasonally grazed pastures, woodlands and isolated farms and houses. There are no designated protected areas for biodiversity or cultural heritage in the Project footprint.

CSA Global Report Nº R302.2019 9 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

DPM has completed a range of hydrological, habitat and species, soil, land ownership and heritage baseline surveys for the site. Further surveys are planned to meet Serbian permitting requirements and align with international good practice. Potential environmental and social risks have been identified. None of these are likely to prevent the Project progressing. Key risk mitigation measures are similar to those associated with other gold mining projects and include safeguarding rivers, groundwater and biodiversity, and those associated with acquiring land.

1.14.3 Social Licence The Project is undertaking active engagement with stakeholders in line with its stakeholder engagement and communications plan. Continued engagement with regulators, directly affected communities and other interested parties before and as part of formal engagement processes will be key to maintaining the Project’s social licence to operate.

1.14.4 Mineral Wastes Waste rock will be stored at waste dumps adjacent to each of the pits. Tailings from the flotation plant will be added to the lined heap leach pad. There will be no separate tailings storage facility. The approach to mine waste management, including the potential for acid rock drainage, will be further investigated at the PFS stage. The approach to mineral waste management will be set out in the statutory project mine waste management plan.

1.14.5 Mine Closure and Aftercare The Project’s approach to closure will be to rehabilitate the mine site so that it is physically and chemically stable and compatible with the intended future land use, which has yet to be determined. The Project will conduct a full closure planning and costing exercise at the PFS stage.

1.15 Capital and Operating Costs Readers are cautioned that the PEA is preliminary in nature. It includes Inferred Mineral Resources considered too speculative geologically to have the economic considerations applied to them that would enable them to be categorised as Mineral Reserves, and there is no certainty the PEA will be realised. Mineral Resources that are not Mineral Reserves do not have demonstrated economic viability. The estimate of Mineral Resources may be materially affected by environmental, permitting, legal, title, socio-political, marketing or other relevant issues.

1.15.1 Capital Costs The initial and life-of-mine (LOM) capital cost estimates for the Timok Gold Project are summarised in Table 1-4. All costs are expressed in United States dollars (US$) unless otherwise stated and are based on Q1 2019 pricing with a deemed overall accuracy of ±40%.

CSA Global Report Nº R302.2019 10 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Table 1-4: Capital cost estimate Initial Sulphide Sustaining Capital LOM Capital Costs ($million) Processing ($million) ($million) ($ million) Mining 35 – 5 40 Processing 33 30 16 80 Infrastructure 34 – 4 38 Total Direct Costs 103 30 25 158 Indirect & Owner's Costs 18 – 18 35 Total Indirect Costs 18 – 18 35 Contingency 15 – 13 28 Reclamation – – 10 10 Total Capital 136 30 65 232 Note: Mining includes $2.206 million in pre-production stripping. Infrastructure includes $10.7 million in land acquisition cost. The Indirect and Contingency amounts are based on various percentage factors that are outlined in Table 1-5 and Table 1-6. Within the Indirects is a sum of $8.3 million for owners’ costs in the pre-production period and the remaining $4.4 million in sustaining. Indirects within the Infrastructure category are included in the base cost estimate. Table 1-5: Indirect costs and percentages Capital category Indirect cost ($‘000) Indirect cost (%) Open Pit Mining 380 1 Process Plant 22,395 28 Infrastructure - - Environmental - - Owners Cost 12,700 -

Table 1-6: Contingency costs and percentages Capital category Contingency cost ($‘000) Contingency cost (%) Open Pit Mining 1,899 5 Process Plant 19,995 25 Infrastructure 5,383 20 Environmental 1,000 10 Owners Cost - -

1.15.2 Operating Costs Operating costs have been developed for a 2.5 million tonnes per annum (Mt/a) leach operation and a 0.5 Mt/a milling and flotation operation with a nine-year life. Total LOM operating costs are summarised in Table 1-7.

CSA Global Report Nº R302.2019 11 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Table 1-7: Operating cost summary $/tonne of LOM $/oz Au $/tonne of Operating Costs(1) Oxide & ($ million) Recovered Sulphide Feed Transitional Feed Mining costs 162 245 9 9 Processing costs 138 209 5 16 G&A costs 31 47 2 2 Cash Costs 331 501 15 27 Royalty (5% NSR to Serbian Gov’t) 40 60 2 3 Offsite costs (Treatment and Refining Charges) 38 57 0 10 Total Cash Costs 409 618 18 39 Sustaining capital 65 99 3 3 AISC(2) 474 717 21 43 (1) Due to rounding, some columns may not total exactly as shown. (2) All-in sustaining cost per ounce of gold represents mining, processing and site general and administrative costs, royalty, offsite costs and sustaining capital expenditures, divided by payable gold of 661,000 ounces. All prices in the PEA study are quoted in Q1 2019 US$ unless otherwise noted. Diesel fuel pricing is estimated at $1.51/L. This estimate was derived from a price quotation for off-road diesel fuel delivered to site with applicable taxes considered. The price for electrical power was set at $70 per MWh, based on current local industrial pricing.

1.16 Economic Analysis Readers are cautioned that the PEA is preliminary in nature. It includes Inferred Mineral Resources considered too speculative geologically to have the economic considerations applied to them that would enable them to be categorised as Mineral Reserves, and there is no certainty the PEA will be realised. Mineral Resources that are not Mineral Reserves do not have demonstrated economic viability. The estimate of Mineral Resources may be materially affected by environmental, permitting, legal, title, socio-political, marketing or other relevant issues. The mine schedule developed by the mining team for this PEA was used as the basis for a discounted cash flow (DCF) model for the Project. All dollar values discussed in the report are US$ values unless specifically stated otherwise. Table 1-8 provides a high-level summary of key pricing assumptions used for the DCF and summary of the key pre-tax and post- tax financial results for the Timok Gold Project. The corporate income tax rate in Serbia is 15%; however, a 10-year tax credit is available to reduce corporate taxes by up to 100%. In practice, tax reductions in the 90% range are typical. The model assumes a tax reduction of 91% for the initial 10-year period in the DCF. Serbian corporate taxes are calculated as EBITDA (earnings before interest taxes, depreciation and amortisation), less: • Accelerated depreciation using the decline balance method with annual rates of 15% (i.e. nine years) for all capital investment. • All mining royalties are deductible. • Tax losses may be carried forward for five years, but tax losses may not be carried back to previous years. Serbia has a 20% value-added tax (VAT). The Project does not collect VAT on its production as all gold production is exported and is therefore VAT exempt. The Project however pays VAT on all supplies and

CSA Global Report Nº R302.2019 12 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

services once production has begun. The model assumes that the Government of Serbia will refund all VAT expended by the Project within four weeks. Note that VAT is not refundable on capital expenditures. The basic royalty rate for gold is 5% based on net payable ounces produced either as doré or concentrate on a net smelter payable basis. The Project is most sensitive to metal prices, followed by operating costs and finally capital costs. Table 1-8: PEA results overview Assumptions Units Gold price $/oz 1,250 Production Profile Total tonnes of mineralized material mined and processed Million tonnes 18.9 Total tonnes waste mined Million tonnes 49.7 Strip ratio waste:feed 2.6:1 Head grade g/t Au 1.36 Peak tonnes per day mineralized material mined Tonnes 8,219 Average gold recovery % 81.5 Total gold ounces mined Oz 826,000 Total gold ounces recovered Oz 673,000 Average annual gold production Oz 75,000 Peak annual gold production Oz 132,000 Mine life Years 9 Unit Operating Costs LOM average cash cost $/oz Au 618 AISC(1) $/oz Au 717 Project Economics Royalties % 5.0 Average annual EBITDA $M 47 Pre-tax NPV 5% / After-tax NPV 5% $M 108 / 105 Pre-tax NPV 7.5% / After-tax NPV 7.5% $M 78 / 75 Pre-tax IRR / After-tax IRR % 18.9 / 18.6 Undiscounted operating pre-tax cash flow / after-tax cash flow $M 195 / 191 After-tax payback period Years 4.1 (1) All-in sustaining cost per ounce of gold represents mining, processing and site general and administrative costs, royalty, offsite costs and sustaining capital expenditures, divided by payable gold of 661,000 ounces.

1.17 Interpretation and Conclusions

1.17.1 Geology and Mineral Resource Estimates The Bigar Hill, Korkan, Korkan West and Kraku Pester sediment-hosted gold deposits have been defined as a result of a systematic sequence of exploration activities from soil sampling, trenching, and mapping, through geophysical evaluation and structural and stratigraphic interpretation, RC and diamond core drilling, metallurgical testwork and, finally, estimation of Mineral Resources. CSA Global has reviewed procedures, visited site, viewed core, verified the locations of several drill-holes, conducted spot checks between hard copy data and digital data, reviewed QAQC results and had extensive discussions with site personnel as part of data verification work. CSA Global has found the site to be well run,

CSA Global Report Nº R302.2019 13 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

with excellent procedures, a good understanding of the deposit geology and an emphasis on data quality that has contributed to a high degree of confidence in the data used in the MRE. Drilling at Bigar Hill and Korkan have served to confirm the structural setting, the stratigraphy, and the geometric, spatial and lithological relationships of the gold mineralisation. The controls on the mineralisation at a local (sample interval) level remain less well understood, and this translates into uncertainties regarding the estimates of gold at the mining scale. This local uncertainty is unlikely to be material in an open pit mining scenario, with a relatively low level of mining selectivity. The level of uncertainty will likely increase under circumstances where cut-off grades are raised and where more selective mining regimes are applied. This Technical Report includes the 2018 MRE update with an effective date of 15 May 2018 and first reported 7 November 2018 (CSA Global, 2018). The 2018 MRE update includes Inferred and Indicated Mineral Resources. Mineral Resources that are not Mineral Reserves do not have demonstrated economic viability. There are no Mineral Reserves defined over the Project. Most of the Mineral Resources defined at the Timok Gold Project are Indicated Mineral Resources supported by good geological knowledge, drill coverage, robust standard operating procedures and data quality, and have been classified under the guidelines of the CIM Standards on Mineral Resources and Reserves, Definitions and Guidelines prepared by the CIM Standing Committee on Reserve Definitions and adopted by CIM Council, and procedures for classifying the reported Mineral Resources were undertaken within the context of reported in accordance with the Canadian Securities Administrators NI 43-101.

1.17.2 Geotechnical A previous geotechnical assessment of the Project was completed in 2014. The wall angles proposed were based on pit heights of 230 m but the final pits for this study only had a small portion of the pit perimeter where the pit depth exceeded 200 m. Based on the previous review, the analysis showed that for the planned pit depths and a factor of safety of 1.2, overall slope angles can be in the region of 45° to 52.5°. The oxidised zone varies by pit area and requires an inter-ramp slope of 45°.

1.17.3 Mining The PEA is based solely on a scenario of the open pit mining of three deposits: Bigar Hill, Korkan and Korkan West. These provide an estimated total of 18.9 Mt of run-of-mine (ROM) process feed material grading 1.36 g/t gold, and an estimated 49.7 Mt of waste, for a delivered strip ratio of 2.63:1 (waste:feed). The heap leach feed marginal cut-off grades varied by pit area due to metallurgical response and material type. The oxide material cut-off grades are 0.19 g/t for Bigar Hill and Korkan, and 0.24 g/t for Korkan West. The transitional material cut-off grades were 0.25 g/t for all deposits. The sulphide concentrator feed material used a cut-off grade of 0.69 g/t for all deposits. The pits are designed based on 5 m bench heights, with safety benches of 8 m every 15 m, vertically. The bench face angles were 55° in oxidised material and varied between 70° and 75° in the fresh rock domain. This achieved inter-ramp angles of 45° in the oxidised material and 52.5° in the fresh rock domain. Ramps are designed at a 10% gradient and are 22.7 m wide to accommodate double lanes for 63-t haulage trucks. The LOM mining operating cost is estimated to be $2.40 per tonne of material moved. Pre-production stripping costs of $2.2 million have been included in the initial capital cost. Initial mine capital costs total $35.3 million and sustaining capital costs total $4.8 million.

CSA Global Report Nº R302.2019 14 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

1.17.4 Mineral Processing and Metallurgical Testing Adequate testwork has been completed to support the assumptions used for gold recoveries for heap leaching of the oxide and transition material for a PEA level study. Gold recoveries for oxide material ranged from 82.1% to 94.4%, whilst gold recoveries for transitional material ranged from 60.3% to 67.9% (uncorrected for full scale heap leach). Additionally, sufficient testwork has been done to support gold recovery estimates through the flotation plant to support a PEA – level study. Flotation tests carried out indicate that gold recoveries to a bulk sulphide concentrate ranged from 68% to 80%, at a concentrate grade ranging from 30 g/t Au to 50 g/t Au. At these concentrate grades, the gold-bearing concentrate would be deemed to be saleable to smelters/roasters.

1.17.5 Infrastructure and Site Layout The process plant is envisioned to be located to the east of Bigar Hill, on a flatter area of ground and west of the proposed heap leach facility. Mine shop facilities are located north of the process plant facilities in order to provide easy access to the pit. Water for the process plant will come from wells in the area and power will come from the main high voltage line 5 km to the east of the Project. Various roads need to be constructed for access to the Project. These include the main access road to the process plant and heap leach facility, as well as to the various pit areas. Korkan West will require the temporary placement of a culvert over the valley bottom to avoid disturbance to natural drainage. Filtered tailings from the 0.5 Mt/a sulphide plant will be stored on the heap leach facility. These tailings will be stored along the edges or in the back to avoid disrupting the leaching process.

1.17.6 Environmental, Permitting, Social and Community The permitting strategy for the Project will need to be flexible as the Serbian permitting system evolves to align with EU requirements. Potential environmental and social risks have been identified during this phase of the Project, although none of these are considered likely to prevent the Project progressing. Key risk mitigation measures are similar to those associated with other gold mining projects and include safeguarding rivers, groundwater and biodiversity and planning for land acquisition. Management of the use and transportation of cyanide will be particularly important, which DPM will do through the International Cyanide Management Code. Further assessment of environmental and social risks will be undertaken through the environmental impact assessment. Details of water management, mine closure and the potential for acid generating rock will be developed at the PFS stage.

1.17.7 Economic Analysis The LOM capital cost for the Project is estimated at $231.6 million, with an initial capital expenditure of $136.1 million, which includes $35.4 million of capitalised pre-stripping. At a gold price of $1,250/oz, the Project is estimated to have an after-tax internal rate of return (IRR) of 18.6% and an undiscounted payback period of 4.1 years from start of production. With a discount rate of 5%, the after-tax net present value (NPV) is estimated to be $105.5 million.

CSA Global Report Nº R302.2019 15 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

1.17.8 Risks General Environmental, permitting, legal, title, taxation, socio-economic, marketing, and political or other relevant issues could potentially materially affect access, title, or the right or ability to perform the work recommended in this report on the Project. However, at the time of this report, the authors and Qualified Persons are unaware of any such potential issues affecting the Project and work programs recommended in this report.

Mineral Resource Estimate As noted in Section 14.2, the MRE could be affected by: • Future, yet unknown environmental, permitting, legal, title, taxation, socio-economic, marketing, political, or other relevant issues. • Metal price and valuation assumptions. • Changes to geological interpretation. • The Mineral Resource estimate has been completed using uniform conditioning (UC) which results in an estimate of tonnage and grade about a set of cut-offs per 20 m x 20 m x 10 m panel. The model provided for further downstream work has been localised which results in a single grade per selective mining unit (SMU) (5 m x 5 m x 5 m). This is for ease of use in downstream processes. However, it is important to note that the estimates of SMUs can be considered reliable within a panel (taking classification into account), but the exact location of an individual SMU remains unknown until pre-production drilling.

Metallurgy/Mineral Processing • Selective mining of the different mineralisation types separately may not be fully achievable. • Adverse effect of degree of sulphidation on gold extraction. • Cold climate effect on leach kinetics. • Degree of clay content on percolation rates.

Mining • Possible flatter highwall angles especially where competent rock overlays weathered in Bigar Hill. • Karst areas not previously identified under waste rock sites or in mining areas. • Additional recovery loss and dilution (due to slow results from sampling or analyses). • Elevated water tables requiring highwall dewatering. Environment, Permitting, Social and Community • Changes to the Serbian permitting system as part of alignment with EU requirements. • Additional environmental or social risks identified in PFS phase, as a result of further design development or baseline studies. • Stakeholder concerns slowing or stopping the permitting process. • Delays and costs associated with acquiring land, particularly if compulsory purchase is needed. Potential for acquisition to be unsuccessful.

CSA Global Report Nº R302.2019 16 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

1.18 Recommendations Based on the results of the PEA study, CSA Global recommends that DPM proceed forward with additional studies, including a PFS for its Timok Gold Project. The recommendations and associated budgets for this next phase of work are described in the sections below.

1.18.1 Geology and Resources • Maintain strong QAQC procedures; CSA Global recommends ongoing vigilance to ensure that standards and blanks are correctly identified and labelled. • When drilling is taking place, it is recommended that a site visit be arranged for a Qualified Person to inspect drilling and sampling practices as they are occurring. • Sampling of core should continue in 1 m increments but should break to honour geological boundaries to enable enhanced analysis and effective modelling of the geology, Mineral Resources and geometallurgy. • Complete a geometallurgical study of the deposits to better define the oxide, transitional and sulphide mineralisation at the three deposit areas, this will be essential prior to commencing mining to ensure the correct materials are directed to the heap leach or the flotation plant. • Future bottle roll testwork should be undertaken on shorter intervals (1 m or 2 m) with the selection of the intervals prioritised in mineralised areas that reflect the geometallurgical domains of the deposit, this will allow improved understanding of material types and allow optimal mine planning. • DPM’s planned 2019 exploration activities at the Timok Gold Project include infill soil sampling, geological mapping, trenching and up to 5,000 m of exploration drilling on near-resource targets, with the aim of increasing near-surface oxide Mineral Resources. In addition to these planned exploration activities, further drilling during 2019 of the existing Bigar Hill, Korkan and Korkan West Mineral Resources includes 7,000 m of infill drilling and 4,300 m of condemnation drilling.

1.18.2 Geotechnical A comprehensive geotechnical and hydrogeological study of the Project will be required, consisting of: • Drilling focused on anticipated final wall positions. • Pit hydrogeological drilling of an estimated eight holes. These holes can be the same as some of the geotechnical holes estimated above. • Infrastructure geotechnical drilling in the area of the process plant, crusher, tailings facilities and waste dumps. These holes can also be used to identify any areas of karst.

1.18.3 Mining Additional work is required to advance the mine plan to a PFS level: • An analysis of the impact of changes in the wall slope angles resulting from the geotechnical study noted above. • A study of the impact of improved sulphide recoveries on overall pit shapes and sizes. • A study of the relative process throughput rates for heap leach and sulphide concentrator feed materials to optimise the mine life and reduce G&A costs. • A detailed analysis of pit dewatering requirements as part of site overall water balance.

CSA Global Report Nº R302.2019 17 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

• An analysis of the mineralised material to determine appropriate grade control procedures for classification of run-of-mine material as either waste, heap leach feed, or sulphide floatation feed materials. This analysis would include sample sizes, methodology of sample selection and assay procedures. • Equipment selection and sourcing. • Blasting analysis with the objective to reduce mining unit costs.

1.18.4 Mineral Processing and Metallurgical Testing Based on the preliminary scoping tests carried out, further testing is required to establish: • The optimum crush size, and whether high gold extraction can be obtained at a coarse crush size of 1 inch (25 mm); for both oxide and transitional material. • Any variability in metallurgical performance associated with testing of different domain types. • Relationship between sulphide sulphur and gold leach extraction.

1.18.5 Environmental • Continue with the program of baseline data collection. • Progress with delivering the permitting plan. • Keep environmental and social risk register up to date; develop/integrate with environmental and social management system. • Continue regular engagement with stakeholders.

1.18.6 Estimated Budget A summary of the estimated costs per discipline of these recommendations for this next phase of work, leading up to the completion of a PFS, is shown in Table 1-9.

Table 1-9: Summary of recommended budgets to complete PFS Area Recommended budget Geology and drilling $3,500,000 Geotechnical $550,000 Mining $500,000 Metallurgy/Processing $400,000 Infrastructure $400,000 Environmental $250,000 PFS $1,000,000 Total $6,600,000

CSA Global Report Nº R302.2019 18 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

2 Introduction

2.1 Issuer This Technical Report has been prepared for Dundee Precious Metals Inc. (DPM), a Canadian-based international mining company engaged in the acquisition, exploration, development, mining and processing of precious metal properties. DPM is a producing issuer in Canada, as defined in NI 43-101. DPM’s operating assets include the Chelopech operation, which produces a gold-copper concentrate containing gold, copper and silver and a pyrite concentrate containing gold, located east of Sofia, Bulgaria; the Ada Tepe operation, which produces a gold concentrate containing gold and silver, located in southern Bulgaria; and the Tsumeb smelter, a complex copper concentrate processing facility located in Namibia. DPM also holds interests in a number of developing gold and exploration properties located in Canada and Serbia, and its 10.3% interest in Sabina Gold & Silver Corp. DPM’s head office is located at 1 Adelaide Street East, Suite 500, Toronto, Ontario, Canada, M5C 2V9. Its regional office in Serbia is located at Kralja Milana 6/4, 11000 Belgrade. In April 2016, DPM completed the acquisition of 49.9% of the common shares of Avala Resources Ltd (Avala) not already owned by the Company. With its acquisition of Avala, DPM obtained a 100% interest in the Timok Gold Project in Serbia, which comprises several sediment-hosted gold deposits, including Bigar Hill, Korkan, and Korkan West.

2.2 Terms of Reference CSA Global (UK) Ltd (CSA Global) was engaged by DPM to complete a Preliminary Economic Assessment (PEA) of the Timok Gold Project utilising the 2018 Mineral Resource estimate (MRE) previously reported by CSA Global (2018) which remains current. CSA Global was initially requested to assess the economic viability of mining only the oxide and transitional mineralisation at the Project through open pit mining, and then processing it using a heap leach operation. The work was conducted in accordance with CIM guidelines and is reported in accordance with NI 43-101.

2.2.1 Scope of Work The scope of technical work to underpin the Project’s PEA level included: • Use of the 2018 MRE previously reported by CSA Global (2018). • A site inspection by CSA Global engineering staff. • Open pit optimisation and mine planning (no underground mine planning will be completed as part of this work). • Development of PEA-level operating expenditure and capital expenditure estimates. • Initiating vendor contact for PEA level pricing. • Preliminary pit designs and conceptual schedules to support the mining plan. • Metallurgical assessment and conceptual processing options and costs. • Heap leach pad design and infrastructure estimates. • Geotechnical review and advice on future programs. • Assessment of environmental and social licence requirements required. • Develop a PEA-level economic model for the Project. • Report the PEA work in accordance with NI 43-101 and NI 43-101 Form F1.

CSA Global Report Nº R302.2019 19 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

2.2.2 Principal Sources of Information The authors of this Technical Report have reviewed available company documentation and other public and private information relating to the Project. In addition, this information has been supported by first-hand reviews, on-site observations and data collection conducted by certain authors (see Section 2.4). Principal sources of information are: • CSA Global (2018), Estimation of Mineral Resources and NI 43-101 Technical Report Gold Project, Serbia – 7 November 2018. • CSA Global (2017), CSA Avala Timok Summary Diary Notes – Site Visit – March 2017. • CSA Global (2017), NI 43-101 Technical Report, Timok Gold Project, Serbia – 31 March 2017. • DST (2016), Results of Laboratory Tests on Avala mineralisation. • AMEC (2014), Preliminary Economic Assessment and Updated Mineral Resource, 1 May 2014. • AMEC (2013), Indicative Pit Slope Angles Assessment, Timok Gold Project, Serbia – 4 December 2013. • Documents and electronic data files provided by DPM. • Information gathered during site visits to the Timok Project. • Information gathered from mining technical literature. • Information gathered from SEDAR (System for Electronic Document Analysis and Retrieval). A list of references, including relevant reports, articles, documents and websites, is provided in Section 27 of this report.

2.2.3 Independence Neither CSA Global, nor the authors of this report, have any material present or contingent interest in the outcome of this report, nor do they have any pecuniary or other interest that could be reasonably regarded as being capable of affecting their independence in the preparation of this report. The report has been prepared in return for professional fees based upon agreed commercial rates and the payment of these fees is in no way contingent on the results of this report. No member or employee of CSA Global is, or is intended to be, a director, officer or other direct employee of DPM. No member or employee of CSA Global has, or has had, any shareholding in DPM. There is no formal agreement between CSA Global and DMP as to CSA Global providing further work for DPM.

2.3 Qualified Person Section Responsibility The various sections of this report have been prepared by or under the supervision of the respective Qualified Persons identified in Table 2-1. Table 2-1: Qualified Person section responsibility Section Section title Qualified Person(s) 1 Summary Maria O’Connor, David Muir, Gary Patrick, Greg Trout, Alex Veresezan 2 Introduction Maria O’Connor 3 Reliance on Other Experts Maria O’Connor 4 Property Description and Location Maria O’Connor 5 Accessibility, Climate, Local Resources, Maria O’Connor Infrastructure and Physiography 6 History Maria O’Connor 7 Geological Setting and Mineralisation Maria O’Connor

CSA Global Report Nº R302.2019 20 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Section Section title Qualified Person(s) 8 Deposit Types Maria O’Connor 9 Exploration Maria O’Connor 10 Drilling Maria O’Connor 11 Sample Preparation Analyses and Security David Muir 12 Data Verification Maria O’Connor (Sections 12.1.1, 12.1.4, 12.1.7, 12.2) David Muir (Sections 12.1.2, 12.1.3, 12.1.5, 12.1.6, 12.1.8) 13 Mineral Processing and Metallurgical Testing Gary Patrick 14 Mineral Resource Estimates Maria O’Connor 15 Mineral Reserve Estimates Greg Trout 16 Mining Methods Greg Trout 17 Recovery Methods Gary Patrick 18 Project Infrastructure Greg Trout 19 Market Studies and Contracts Alex Veresezan 20 Environmental Studies, Permitting, and Social Greg Trout or Community Impact 21 Capital and Operating Costs Greg Trout (Sections 21.1, 21.2, 21.4 to 21.7, 21.8, 21.9, 21.12) Gary Patrick (Section 21.3, 21.8, 21.10, 21.11) 22 Economic Analysis Alex Veresezan 23 Adjacent Properties Maria O’Connor 24 Other Relevant Data and Information Maria O’Connor 25 Interpretation and Conclusions Maria O’Connor (Sections 25.1, 25.8.1, 25.8.2, 25.9.1) Greg Trout (Sections 25.2, 25.3, 25.5, 25.6, 25.8.4, 25.8.5, 25.9.3, 25.9.4) Gary Patrick (Sections 25.4, 25.8.3, 25.9.2) Alex Veresezan (Section 25.7) 26 Recommendations Maria O’Connor (Sections 26.1, 26.2, 26.8) Gary Patrick (Section 26. 5) Greg Trout (Sections 26.3, 26.4, 26.6, 26.7) 27 References Maria O’Connor

2.4 Site Visits

2.4.1 CSA Global Site Visits CSA Global completed a two-day site visit from 28 February to 1 March 2017. Qualified Persons, Ms Maria O’Connor and Mr David Muir, completed the following during the site visit: • Discussions with Justin van der Toorn (Exploration Manager), Dragana Davidović (Senior Geologist) and Mladen Zdravković (Regional Geologist) regarding procedures, geology, interpretation, exploration, tenure and assumptions made for the MRE. • Review of Exploration Method Policies compiled in 2005 and used throughout Avala’s drilling and exploration programs. • Field trip to drill sites at Bigar Hill. Seven drill collars were located but due to snowy conditions, roads to most drill sites at Bigar Hill and all at Korkan and Kraku Pester were inaccessible. • Visual review of mineralised portions of three diamond drill-holes in the core shed. • Visits of the pulp library where returned sample pulps are securely kept.

CSA Global Report Nº R302.2019 21 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

• Visits and audits of the sample preparation and analytical laboratories at SGS Bor. • Spot checks of the database using hard copy data. • Spot checks of assay certificates against assays stored in the database. • Independent reporting and evaluation of quality assurance/quality control (QAQC). Mr Gary Patrick, Qualified Person for Metallurgy and Processing aspects, visited the site for three days from 15 to 17 November 2018. The visit included the Project offices, Timok exploration site, surrounding infrastructure, and core storage. During the site visit, in-depth discussions took place with the Project’s main personnel, data and information was studied and reviewed, and information was freely exchanged.

2.4.2 AGP Site Visit AGP Mining Consultants Inc (AGP) completed a three-day site visit from 15 to 17 November 2018. Qualified Person, Mr Greg Trout, P.Eng completed the following during the site visit: • Discussions with Ian Lipchak (DPM Manager), Paul Cromie (DPM Director Exploration), Richard Gosse (DPM SVP Exploration), Ross Overall (DPM Corporate Senior Resource Geologist) and Samuel Amoh (DPM Corporate Senior Mining Engineer) regarding geology, exploration, tenure and assumptions made for the mine planning. • Field trip to Bigar Hill, Korkan and Korkan West sites. • Visual review of several drill-hole cores in the core shed. The visit included the Project offices, Timok exploration site, surrounding infrastructure, and core storage. During the site visit, in-depth discussions took place with the Project’s main personnel, and data and information was studied and reviewed, and information freely exchanged.

2.5 Units and Datum All units in this report use the International System of Units (SI), i.e. are metric unless stated otherwise. All surveying on the Project area has been undertaken using the Universal Transverse Mercator (UTM) coordinate system, specifically Zone 34 North in WGS 84 datum, on the EGM96 Geoid model. A primary survey control network was implemented using AUSPOS, an online global positioning system (GPS) processing service provided by Geoscience Australia. A secondary control network was observed from the primary control network to locate control around the actual prospect areas using static surveys.

2.6 Report Effective Date: The report is based on information known to CSA Global and the authors as of 30 April 2019, the Effective Date of this report. The MRE that forms the basis of the PEA has an effective date of 15 May 2018 as reported 7 November 2018 (CSA Global, 2018).

2.7 Forward Looking Statements This report contains “forward-looking information” or "forward-looking statements" that involve a number of risks and uncertainties. Forward-looking information and forward-looking statements include, but are not limited to, statements with respect to the future prices of gold and other metals, the estimation of Mineral Resources, the realisation of mineral estimates, anticipated exploration activities, permitting time lines, currency fluctuations, government regulation of mining operations, mining and operating parameters, the PEA and related parameters, and the commencement of a PFS.

CSA Global Report Nº R302.2019 22 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Often, but not always, forward-looking statements can be identified by the use of words such as “plans”, “expects”, or “does not expect”, “is expected”, “budget”, “scheduled”, “estimates”, “forecasts”, “intends”, “anticipates”, or “does not anticipate”, or “believes”, or variations of such words and phrases or state that certain actions, events or results “may”, “could”, “would”, “might” or “will” be taken, occur or be achieved. Forward-looking statements are based on the opinions, estimates and assumptions of contributors to this report. Certain key assumptions are discussed in more detail. Forward Looking statements involve known and unknown risks, uncertainties and other factors which may cause the actual results, performance or achievements of DPM to be materially different from any other future results, performance or achievements expressed or implied by the forward-looking statements. Such factors include, among others: the actual results of current exploration activities; conclusions of economic evaluations; changes in project parameters as plans continue to be refined; future prices of gold and other metals; possible variations in grade or recovery rates; failure of plant, equipment or processes to operate as anticipated; accidents, labour disputes and other risks of the mining industry; delays in obtaining governmental approvals or financing or in the completion of development or construction activities, fluctuations in metal prices, as well as those risk factors discussed or referred to in this report and in DPM’s latest annual information form under the heading "Risk Factors" and other documents filed from time to time with the securities regulatory authorities in all provinces and territories of Canada and available at www.sedar.com. There may be other factors than those identified that could cause actual actions, events or results to differ materially from those described in forward-looking statements, there may be other factors that cause actions, events or results not to be anticipated, estimated or intended. There can be no assurance that forward- looking statements will prove to be accurate, as actual results and future events could differ materially from those anticipated in such statements. Accordingly, readers are cautioned not to place undue reliance on forward looking statements. Unless required by securities laws, the authors and DPM undertake no obligation to update Forward-looking statements if circumstances or estimates or opinions should change.

CSA Global Report Nº R302.2019 23 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

3 Reliance on Other Experts

The authors of this report are not qualified to provide extensive comment on any legal, political, environmental or tax matters associated with the Timok Gold Project included in Section 4 of this report. Reporting of these aspects relies on information provided by DPM and has not been independently verified by the authors. The authors have not verified the status of DPM’s tenure or joint venture agreements pertaining to the Property beyond viewing the tenure agreement and have relied on information provided by DPM with regard to the legal title to the mineral concessions (Section 4.3). Section 20 of this report relies upon a Microsoft Word document prepared and provided to CSA Global by Sarah Sanders Hewitt of ERM International. Greg Trout, the Engineering Qualified Person takes responsibility for the presentation of this work in this report.

CSA Global Report Nº R302.2019 24 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

4 Property Description and Location

4.1 Location The Project is located in the eastern part of the Republic of Serbia, approximately 270 km southeast of its capital, Belgrade, as shown in Figure 4-1. Its northern boundary is positioned about 25 km south from the Danube River and the Project area extends 24 km southwards to a point approximately 14 km west and southwest of Bor at its southern boundary. The main deposits on the Project are located approximately 25 km northwest of the town of Bor, Serbia. Bor is a historical centre for copper mining and smelting in Serbia.

Figure 4-1: Location map – Timok Gold Project Source: Avala, 2018

CSA Global Report Nº R302.2019 25 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

4.2 Property Description The Timok Project comprises three exploration licences (Potaj Čuka Tisnica, Umka and Bigar Istok) covering an aggregate area of 131.21 km2. Locations of the exploration licences are shown in Figure 4-2. The Bigar Hill, Korkan, Korkan West, and Kraku Pester deposits, which are the subject of this Technical Report, are located within the boundary of the Potaj Čuka Tisnica exploration licence.

Figure 4-2: Timok Gold Project exploration licences Source: Avala, 2017

CSA Global Report Nº R302.2019 26 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

The Lenovac exploration licence is no longer considered to be part of the Project due to the different style of mineralisation occurring on this licence. Exploration licences are currently granted by decisions of the Serbian Ministry of Mining and Energy (MoM&E); they are generally issued on an initial three-year basis and are twice renewable for a further period of three years (first renewal), followed by a period of two years (second renewal). An integral part of the exploration licence application and renewal process is submission of a detailed exploration work program. Supporting documentation is also required from the Institute for the Preservation of Cultural Heritage and the Institute for Nature Conservation of Serbia, to the effect that the proposed exploration activity is in accordance with Republic of Serbia’s environmental and cultural legislation. The obligations of the licence holder are to complete the submitted and approved work program, provide annual exploration activity reports to the Serbian MoM&E, and to advance the geological knowledge of the property. Exploration licences can be renewed if the exploration licence holder fulfils its obligations, including the completion of at least 75% of the planned work program. The legislation provides for a clear development process, from discovery through to mine development and operation. To retain the licences beyond the final two-year extension period, a similar application can be made to request a reservation of the exploration licences for a further three-year period, during which permitting activities may take place.

4.2.1 Ownership The exploration licences for the Project are held by Avala Resources d.o.o., a Serbian registered, wholly owned subsidiary of DPM, following the amalgamation of a wholly owned subsidiary of DPM with Avala Resources Ltd. in April 2016. The Potaj Čuka Tisnica and Bigar Istok exploration licences were renewed (second renewal) in July 2019 and are valid until July 2021. The Umka exploration licence was renewed in August 2019 for a further three years (first renewal). Details of each of the exploration licences are outlined in Table 4-1. The expenditure commitments for keeping the exploration licences in good standing and eligible for renewal at the end of each respective licence period are summarised in Table 4-1. DPM fulfilled all its commitments on the licences renewed in 2019 and fully expects to fulfil all obligated commitments on the licence extensions in order to maintain the Timok exploration licences in good standing. Table 4-1: Tenement details for Timok Gold Project exploration licences Expenditure Area Licence Licence number Holder Grant date Expiry date commitment* (km²) (EUR) Potaj Čuka 310-02-0121/2006-06 Avala Resources d.o.o. 20-Jun-2006 22 Jul 2021 80.38 5,617,333 Tisnica Bigar Istok 310-02-0262/2013-03 Avala Resources d.o.o. 05-Mar-2014 22 Jul 2021 15 460,740 Umka 310-02-01413/2015-02 Avala Resources d.o.o. 25-Mar-2016 14 Aug 2022 35.83 833,855 *Expenditure commitment relates to the full work program (covering the period from the grant date to the expiry date) as submitted to the Serbian Ministry. The company is required to meet 75% of this commitment for the licence to be eligible for renewal after the expiry date. Source: Avala, 2019

CSA Global Report Nº R302.2019 27 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

4.3 Mineral Tenure There are no other known agreements or encumbrances on the properties. DPM operates with the permission of the MoM&E, in conjunction with the Ministry of Environmental Protection, and the Ministry of Culture and the Media of the Republic of Serbia. DPM does not currently own the surface rights to any of the land parcels located on the exploration licences. To gain access to the land to conduct exploration activities, land access agreements are negotiated with the local landowners, in the case of privately held land, or with the state, in the case of state land. These land access agreements follow Serbian legislative requirements in terms of proscribed compensation for access and land disturbance, etc. The land access agreements are recorded in a master register in order to document and maintain transparency in negotiating and maintaining land access compensation. The Qualified Person knows of no other significant factors and risks that may affect access, title, or the right or ability to perform work on the property. The Qualified Person knows of no environmental liabilities to which the property is subject. No additional permits are required if the work program associated with the licence application does not fall below or exceed the plan by 25%. An addendum must be filed detailing the work program if the 25% tolerance is exceeded. The Lenovac exploration licence is still held by Avala Resources d.o.o., but is no longer considered to be part of the Timok Gold Project. The Lenovac licence is situated within the Timok Magmatic Complex (TMC), but does not appear to possess the sedimentary-hosted gold mineralisation identified within the remaining exploration licences of the Timok Gold Project. The Lenovac licence is also no longer subject to a joint venture agreement with Rio Tinto Mining and Exploration Limited, which expired as of 31 December 2018 and was not extended or renewed.

4.4 Royalties The Serbian government levies a royalty of 5% net smelter return (NSR) for production of metallic raw materials.

CSA Global Report Nº R302.2019 28 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

5 Accessibility, Climate, Local Resources, Infrastructure and Physiography

5.1 Accessibility The Project is accessible by regional asphalt roads between Bor, Žagubica, Krepoljin, and Zlot, and well- developed unsealed forestry roads. The area is also linked via Bor to Zaječar and Paraćin and via Žagubica to Požarevac (and further to Belgrade). There is a railroad from Bor to Belgrade through Požarevac. Avala operated its exploration programs on a year-round basis. Exploration activity during the winter period was generally limited to drilling operations (both diamond and reverse circulation (RC)), and, provided that adequate preparation works are completed in the fall, year-round access is possible. During the 2010/2011 and 2011/2012 winter periods, Avala was unable to access the Timok areas for a cumulative total of 10 to 15 days on an annual basis, due to extreme weather conditions (very low daily temperature maximums and/or high snowfall).

5.2 Physiography Terrain in the Project area is hilly to mountainous, ranging from about 500 metres above sea level (masl) to 944 masl at Coka Rakita, the highest peak in the region. Other high peaks are Coka Berbjesce (817 masl), Strez (731 masl), and Coka-Unuk (741 masl). The most important drainage is the Jagnjilo River, which drains into the Veliki Pek, and further on to the Danube, and incorporates the Bigar Hill and Korkan areas. The lower slopes and valleys are largely given over to seasonal farming, while forests dominate the higher slopes and peaks. Figure 5-1 shows a view of the Project area from Korkan, looking southwards towards Bigar. Seasonal (summer) pastures, together with forested areas, are characteristic of the Project area. Typical physiographic landscapes and climate contrasts at the Project are shown in Figure 5-2.

Figure 5-1: Typical landscape of Timok Gold Project, looking south towards Bigar Hill deposit Source: Avala, 2014

CSA Global Report Nº R302.2019 29 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Figure 5-2: Typical physiographic landscape and climatic contrasts (summer, top vs winter, bottom) – Bigar Hill project entrance Source: Avala, 2014

5.3 Climate The Timok area is characterised by moderate continental climate, with some influence of high mountainous climate. Winters are long and cold, with abundant snow cover, and summers are usually hot. First seasonal frosts occur in October and the last frosts are in April. Based on long-term observations from the Crni Vrh weather station, the coldest month is January, with an average temperature of -2.4°C, and the hottest month is July, with an average temperature of +19°C. The annual precipitation is in the range of 540 mm to 820 mm, according to governmental information.

CSA Global Report Nº R302.2019 30 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

5.4 Infrastructure The Bigar Hill, Korkan and Korkan West, and Kraku Pester mineralised areas are located approximately 3 km, 4 km and 2 km respectively from the 110 kV Serbian national power grid, which extends from Bor to Petrovac and passes through the Project. The main road between Bor and Petrovac, which is sealed, also passes through the Project, near the power line. Located some 30 km west of Bigar Hill is an operating aggregate plant, which is in good condition and currently supplies customers with aggregate for concrete. The town of Bor is connected by rail to Belgrade (via Požarevac); this same rail network is part of European Transportation Corridor 10, which extends southwards through the Republic of North Macedonia to Greece and the Mediterranean, and also eastwards through Bulgaria to ports on the Black Sea (and further to Turkey). Bor is accessible via the national highway grid (Paraćin turnpike), leading to sealed roads through Boljevac to Bor. While there is limited infrastructure within the mineral deposit area, there are existing power lines and networks of well-developed, unsealed forestry roads. Habitation within the Project area is sparse and generally restricted to summer-months seasonal occupancy. Avala has an operational base in the town of Bor (population approximately 40,000). Bor is a historic mining centre within eastern Serbia, which has been in near-continuous operation since 1902. Currently, majority of the population is employed by the state-owned mining group, RTB Bor (and its subsidiaries), which operates the Veliki Krivelj and Cerovo open pit copper mines and the underground Borska-Jama copper-gold operation, together with the Bor smelter, all located proximal to the town. A large proportion of the population has experience in work activities associated with mining operations, and the local availability of technical staff for any future mining operations within the region should be considered high.

CSA Global Report Nº R302.2019 31 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

6 History

6.1 Prior and Current Ownership DPM has been active in minerals exploration in Serbia since 2004 and had acquired several exploration licences and concessions between 2004 and 2010. In July 2010, Avala Resources Ltd. acquired Avala Resources d.o.o. (formerly named Dundee Plemeniti Metali d.o.o.) from DPM through a reverse takeover transaction, in which DPM retained a 51% share. In April 2016, DPM completed the acquisition of the remaining 49.9% of Avala Resources that it did not currently own, effectively re-acquiring full ownership of the project. The exploration licences for the Timok Gold Project are held by Avala Resources d.o.o., a Serbian registered wholly owned subsidiary of DPM. Avala Resources d.o.o. became a wholly owned subsidiary of DPM in April 2016, when a wholly owned subsidiary of DPM amalgamated with Avala Resources Ltd.

6.2 Exploration History The Timok region has a long history of exploration and mining, dating back to Roman times. Key periods include: • Mining during Roman times, as demonstrated by the discovery of slag and mining tools. • Geological mapping commenced in 1933 by Geozavod, Belgrade, and Geology Institute Bor. • Geophysical exploration undertaken by French prospectors in the 1930s and during various periods until 1985 by the Institute for Geological and Geophysical Exploration, Belgrade. • Several geochemical surveys, commencing in 1958, undertaken by Geozavod, Belgrade, and Geology Institute Bor. • Small-scale adits developed prior to World War II. • Limited exploration, including drilling, which commenced post-World War II, by RTB Bor. • Pits and adits of unknown chronology are scattered through the eastern and southern portions of the exploration licences. • No production of any significance from the property has been undertaken. Previous exploration at the Project, undertaken from 2007 to 2009, has been summarised by Coffey Mining (2010). DPM is not aware of any exploration for gold taking place within the Project area prior to 2007. Extensive soil sampling and surface trenching programs were carried out during the 2007 to 2009 period. Four (581.7 m) diamond core drill-holes and 152 trenches (28,014.6 m for 14,138 samples) were completed on the Project, though much of this was outside the four deposits that are the subject of this report (Bigar Hill, Korkan and Korkan West, and Kraku Pester). Avala then focused exploration drilling campaigns from 2010 to 2013 on the Potaj Čuka Tisnica licence to outline mineralisation on the Bigar Hill, Korkan, Kraku Pester and Umka areas. The drilling that relates to Bigar Hill, Korkan and Kraku Pester is covered in more detail in Section 10. Along with drilling, from 2010 onwards, outcrop, soil and trench sampling were conducted. After 2014, a number of exploration trenches, channels and drill-holes were completed on wide-spaced grids on areas peripheral to the mineralised prospects. After Avala was fully re-acquired by DPM in 2016, a near-resource target generation exercise was undertaken, which led to the discovery of the Korkan West deposit during winter 2016/2017.

CSA Global Report Nº R302.2019 32 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

7 Geological Setting and Mineralisation

7.1 Regional Geology The Project is located within the north-western part of the Timok Magmatic Complex (TMC) in eastern Serbia. The TMC is part of the greater Alpine-Balkan-Carpathian-Dinaride metallogenic-geodynamic (ABCD) province (Figure 7-1), which, in turn, is part of the Tethyan (or Alpine-Himalayan) orogenic system that extends from Western Europe to South-East Asia. The orogen resulted from the convergence and collision of the Indian, Arabian, and African plates with Eurasia, initially in the Cretaceous and continuing today. The complex arcuate geometry of the collision interface, and the presence of several micro-plates within the orogenic collage, resulted in a variety of collision products (Gallhofer et al., 2015). Some segments are characterised by extensive regional metamorphism, whereas others by calc-alkaline igneous activity. The structural complexity and present-day geometry of the region reflects large-scale oroclinal bending during post-collision tectonics throughout the Tertiary, including major transcurrent fault systems with overall dextral displacements exceeding 100 km (Knaak et al., 2016).

Figure 7-1: Tectonics and chronology of the ABCD province Source: AMEC, 2014 Orogenic segmentation resulted in a discontinuous distribution of mineral deposits within the ABCD province and limited the lateral extents of the various metallogenic belts along the trace of the orogen. These Late Cretaceous to Miocene belts and adjacent segments host significant porphyry copper-gold deposits with

CSA Global Report Nº R302.2019 33 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

related high sulphidation copper-gold mineralisation. The major deposits within these belts are Skouries, Chelopech, Bor, Veliki Krivelj, and Majdanpek, as well as many deposits in the Golden Quadrilateral of Romania. Within the ABCD province, the most economically significant segment comprises the Upper Cretaceous subduction-related magmatic rocks and mineral deposits, referred as the Apuseni-Banat-Timok-Srednogorie Belt (ABTSB). This L-shaped belt extends from Romania, through Serbia, and into Bulgaria. Plate reconstructions show that the ABTSB originally had an east-west orientation in Late Cretaceous times (Gallhofer et al., 2015 and references therein). The structural complexity, the present-day L-shape geometry of the region and clockwise rotation (~30°) of the TMC segment reflects large-scale oroclinal bending during post-collision escape tectonics throughout the Tertiary, including major transcurrent fault systems with an overall dextral displacement in excess of 100 km and associated alternating transpressive and transtensional episodes. Intrusive and extrusive rocks of the ABTSB were emplaced during a 30 million-year (Ma) period from ~90 Ma to 60 Ma and may have been associated with several different subduction zones of varying polarity (Gallhofer et al., 2015). The easternmost magmatic complex in Serbia, the TMC, bounds the Project area on the east.

7.2 Regional Structural Geology Several fault populations of various inferred ages-of-formation have been identified in the TMC, characterised by relatively more intense development of strike length and density on the western margin of the TMC. All these fault populations are interpreted as products of Cenozoic (Alpine) transpression. From oldest to youngest, the populations constitute: • Palaeozoic/Mesozoic faulting of metamorphic basement rocks. These faults were undoubtedly reactivated during syn-sedimentary TMC basin formation and subsequent emplacement of igneous intrusions. • Early (?) Cretaceous, currently northwest-striking, dislocations that appear to have controlled basin opening. These structures are interpreted as major accommodation-structures during Eocene-Oligocene deformation. • Late Cretaceous strike-extensive reverse faults, trending north-south to northeast-southwest. These faults were reactivated by Alpine transpression that resulted in accommodation of dextral strike-slip motion. A discontinuous easterly-dipping subpopulation of these faults is developed through the sediment-hosted gold prospects and is interpreted as having been a single structure prior to disruption by subsequent deformation. This feature is defined as a domain-bounding structure and is discussed below. Geology maps at 1:25,000 scale show north-trending, east-dipping reverse faults as part of a larger north-trending reverse fault system at the north-western margin of the TMC. • Evidence for reverse movement is expressed as repetition/imbrication of stratigraphy and is also associated with local folding and variation in the dip of stratigraphic layering. Northeast-striking faults locally post-date sedimentary rock-hosted mineralisation, as evidenced by their intersection and offset of the margins of the Potaj Čuka monzonite, although the degree to which this can be attributed to fault reactivation is unknown. • Eocene to Oligocene northwest-striking, strike-slip faults that hosted sinistral movement as a result of oroclinal bending. These structures constrain numerous regionally pervasive, short strike-length northeast-trending faults that are typically expressed as topographic lows. • Late normal faults are responsible for the geometry of features such as the Miocene Žagubica Basin, which contains approximately 2,000 m of sedimentary infill. These structures extend eastward into Bigar Hill

CSA Global Report Nº R302.2019 34 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

and offset the mineralised system. Similar faults are present at Korkan, but their trace is complicated due to the presence of numerous northwest-striking faults that are also post-mineral. Regionally developed east-west striking faults of variable strike length are expressed as discrete brittle structures at all scales and crosscut all other structural features. Despite the age relationships indicated above, the assignment of individual faults to populations of particular ages is difficult. Surface expressions of faults are uncommon, and crosscutting relationships are rarely conclusive. Furthermore, a diversity of fault orientations is present, due to different ages of faulting, shifting far-field stress geometries over time, reactivation of older faults, and the role of pre-existing architecture during the formation of each successive stage of faulting. A critical element in the identification of faults has been the resolution of a consistent stratigraphic framework – the components of which can be identified regionally.

7.3 Local Geology In eastern Serbia, magmatic activity of the Late Cretaceous ABTSB is developed along two subparallel north- trending branches; the narrow Ridanj-Krepoljin Belt to the west, and the wider TMC to the east. The latter branch contains the Bor and Cukaru Peki high-sulphidation type epithermal copper-gold deposits, and hosts major porphyry copper deposits (Majdanpek, Veliki Krivelj) and several Late Cretaceous epithermal occurrences (e.g. Lipa). The TMC is approximately 85 km long and extends from Majdanpek in the north to Bučje in the south. The disposition of the Project’s exploration licences and the local geology are shown in Figure 7-2. The Late Cretaceous TMC developed on a continental crust composed of different fault-bounded terranes composed of Proterozoic metamorphic to Lower Cretaceous rocks. The area is now incorporated in the Getic Nappe or the Kučaj Terrane, as part of the complex Carpathian-Balkan Terrane in eastern Serbia. Upper Jurassic and Lower Cretaceous shallow marine sedimentary rocks, dominated by homogeneous, massive to bedded limestone and marl, unconformably overlie a metamorphic basement. Carbonate sedimentation terminated in the Early Cretaceous due to the impact of the Austrian deformational phase, which caused weak deformation, uplift, erosion, and subsequent paleokarst formation. Clastic sedimentation commenced with an Albian transgression, unconformably burying the partially eroded and faulted carbonate platform rocks. These calcareous clastic rocks mark the start of the evolution of the TMC, beginning with Austrian deformation and followed by deformation in the Late Cretaceous (Albian). They outcrop along the eastern and western boundary of the TMC but rarely in the central part. Sedimentation continued through the Cenomanian, with an increasingly volcanic detrital component becoming important with decreasing age. During the Turonian, volcanism commenced, and progressed from east-to-west across the TMC. At this time, the TMC became a topographically positive volcanic area. Contemporaneous sedimentation, magmatism, and hydrothermal activity were relatively continuous within the TMC throughout the entire Late Cretaceous, as illustrated in Figure 7-3. The sedimentation persisted from the Albian to the Maastrichtian. Late Cretaceous magmatic activity has been documented during a 10-millon- year period from ~89 Ma to 78 Ma and has been interpreted to generally progress from east to west, younging across strike towards the subduction zone. This process can be related to an arc under extension and gradual steepening and rollback of a northward subducting lithosphere slab, derived from the Vardar ocean.

CSA Global Report Nº R302.2019 35 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Figure 7-2: Exploration licences with the TMC Source: Avala, 2018

CSA Global Report Nº R302.2019 36 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Figure 7-3: Schematic stratigraphy of the Western TMC Source: AMEC, 2014 The TMC is dominated by alkaline to high-potassium calc-alkaline magmatic rocks, classically divided into three successive volcanic sequences (commonly referred to as V1 to V3 or Phase 1 to Phase 3) that are intercalated with Late Cretaceous volcaniclastic sedimentary rocks. Diorite dykes and sills are common, but locally difficult to distinguish from the volcanic supracrustal rocks. The first phase of volcanism commenced during the lower Turonian with mainly hornblende andesitic magmatic rocks in the easternmost (present coordinates) parts of the TMC. Cessation of volcanism in the early Campanian and uplift and erosion of the eastern part of the TMC were followed by local turbiditic deposition of the Bor pelites. Magmatism shifted westward into the central and western parts of the TMC during the Santonian. A second phase of magmatism is represented by two compositionally and geographically distinct assemblages: • Pyroxene-bearing, subaqueous to subaerial andesitic rocks. • A sequence of latites, trachytes and trachy-basalt dykes, restricted to the south-western part of the TMC. During Late Cretaceous (Campanian), diorite, quartz-diorite, and monzonite plutonic rocks were emplaced. Magmatism and sedimentation terminated in the upper Campanian and Maastrichtian. The coarse-grained Bor conglomerate records exhumation of the basement within the eastern TMC. Calcareous rocks were deposited in the central part of the TMC at this time. The Upper Cretaceous rocks of the TMC are overlain by Paleogene to Neogene sedimentary rocks and deposits of quaternary sediments.

CSA Global Report Nº R302.2019 37 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

The structural complexity and present-day asymmetric lozenge-shaped geometry of the TMC area resulted from oroclinal bending during post-collision tectonics throughout the Tertiary. This has led to tectonic modifications of lithological contacts, including those that represent syn-depositional features, beds, or faults. The extent of deformation is commonly difficult to assess due to variable responses of different rock types to the same deformation event. Much of the deformation has been absorbed by argillaceous horizons due to their ability to accommodate shearing and shortening, whereas sandstone beds have resisted much of the deformation. Similarly, competent massive limestone units forming the base of the sequence exhibit minor deformation and much of this is expressed as fracturing near the contact with the overlying clastic sedimentary rocks.

7.4 Project Stratigraphy The Project area at the western margin of the TMC can be subdivided into two northerly-trending domains: • Western Domain – dominated by Proterozoic metamorphic basement, Upper Jurassic and Lower Cretaceous limestones. • Eastern Domain – dominated by the Cretaceous volcanic, epiclastic rocks, and associated diorite intrusive rocks of the TMC, including the known porphyry copper-gold centres at Valja Strz and Dumitru Potok. The boundary between the two domains is dominated by calcareous clastic sedimentary rocks, including sandstone, conglomerate, and marl, and is partly defined by a domain-bounding structure. The four identified sediment-hosted gold zones within the Potaj Čuka Tisnica licence are Bigar Hill, Korkan, Korkan West and Kraku Pester. These prospects are hosted by calcareous clastic sedimentary rocks that outcrop within the boundary zone between the two domains. The overall east-west cross-sectional geometry of the TMC is that of a complexly faulted synclinorium. The stratigraphy generally dips moderately to the east at approximately 20° to 30°, along the western margin of the TMC. This general tilting of stratigraphy at the western margin indicates that the oldest rocks outcrop in the west, and that farther east the stratigraphy becomes younger, with stratigraphically higher units dominating the outcrops. Mapping by Avala, building upon public domain geologic maps and knowledge, has defined litho-stratigraphic interpretive units which are recognised as being important to the Project and are outlined below. In Figure 7-3, these units are summarised in a stratigraphic column and correlated with geological time series, deformation, and magmatic events.

7.4.1 Palaeozoic and Proterozoic Basement Within the Potaj Čuka Tisnica licence, the oldest outcropping rocks are Palaeozoic phyllites, a meta- sedimentary sequence composed of sandstone, shale, and conglomerate protolith, and a variety of heterogeneous Proterozoic greenschist-facies schistose quartzo-feldspathic schists and gneisses. These units, which have not been further differentiated, commonly outcrop in the cores of anticlines, but have also been encountered in the bottom of some exploration drill-holes within the Project area.

7.4.2 Carbonate Sequence, JLS and KLS Two units constitute the Upper Jurassic (JLS) and Lower Cretaceous (KLS) carbonate rocks within the project area. The older Jurassic age unit is characterised by massive limestone, most which is dominated by bedded and massive bioclastic and micritic, white, light-grey, and light brownish reef limestone of Tithonian age. The lower parts are commonly composed of micritic limestone with concretional chert nodules, and the contact with the underlying basement is commonly faulted. Unconformably overlying the Jurassic limestone is Upper Cretaceous dark grey limestone with black concretional chert nodules, deposited during the Valanginian- Hauterivian (Vasic, 2012). Most the Lower Cretaceous rocks are well-bedded bioclastic, nodular, and

CSA Global Report Nº R302.2019 38 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

stromatolitic, and locally sandy limestones deposited during the Barremian and Aptian; these are referred to as the Urgonian limestone. The limestone units are karsted, with the massive Jurassic limestone being more susceptible to karstification than the well-bedded Urgonian limestone. Some paleokarst formed prior to deposition of the younger and unconformably overlying clastic sedimentary rocks. A typical assemblage of the units is shown in Figure 7-4. These karst areas are partly filled by syn-karst fine-grained sedimentary rocks, as well as along the upper contact with finely laminated upper Lower Cretaceous (Albian) calcareous clastic sedimentary rocks. Locally, paleokarst collapse breccia is developed, and the karsted zones are a host to gold. Recent karst is also evident.

Figure 7-4: Typical contact between Upper Jurassic (T) and Lower Cretaceous (V) Limestones with black chert nodules Source: AMEC, 2014

7.4.3 Calcareous Clastic Sedimentary Rocks, S1 and S2 Three distinct units of calcareous clastic rocks unconformably overlie the carbonate sequence. Various carbonate units lie beneath the unconformity, indicating exhumation and accompanying faulting during the depositional hiatus in the Early Cretaceous. Formation of the unconformity reflects the effect of the Cretaceous Austrian orogenic event. The clastic units, stratigraphically from lowest-to-highest, include calcareous sandstone with lesser siltstone-dominated sequence, overlain by reddish and iron-rich sandstone

CSA Global Report Nº R302.2019 39 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

containing abundant andesitic volcanic detritus, capped by thinly bedded ferruginous marl. Total stratigraphic thickness of this sequence ranges from 365 m to 840 m: • S1 unit is a basal clastic sequence that was deposited during the Albian to Cenomanian time (Vasic, 2012), and consists of well-bedded, coarse- and medium-grained calcareous, weakly glauconitic sandstones and conglomerates. Clasts are dominantly angular to sub-rounded limestone fragments sourced locally, but also include a variety of well-rounded metamorphic and igneous clasts from distal sources. Conspicuous rounded white quartz pebbles form a major detrital component. Intercalated with, and locally forming a significant thickness, are black, laminated, fine-grained clastic siltstone and sandstone. A chaotic breccia, termed the “basal breccia”, is common along the basal unconformable contact. The breccia is composed dominantly of coarse blocks and smaller cobbles and pebbles of limestone in a black, commonly sulphide- rich, fine- to medium-grained calcareous sandy matrix. Locally, the basal breccia appears bedded. Similar angular clastic rock types are present within the underlying limestone sequence at various depths below, but close to, the uppermost limestone contact. These clastic rocks have very irregular thicknesses, are not laterally continuous, and are inferred to represent infill of karst features. The thickness of this S1 unit can vary from between 50 m and 250 m above the unconformity. A typical core specimen from the S1 unit is presented in Figure 7-5. • S2 unit overlies the basal clastic sandstone (S1) and is comprised of reddish, coarse- and medium-grained sandstones and conglomerates, tuffaceous clastic rocks, and air-fall tuff (S2) containing varying abundances of detrital magnetite, mafic silicate minerals, and common volcanic fragments. Pyrite, presumably diagenetic in origin, is also present. This clastic sequence, deposited at the western margin of the TMC during Cenomanian time, records the approach of the volcanic arc to the east. The thickness of the S2 sandstone unit is between 15 m and 90 m. A typical core specimen from the S2 sandstone and conglomerate is shown in Figure 7-6. • The S1 and S2 units form the principal host to gold in the Bigar Hill, Korkan and Korkan West deposits.

Figure 7-5: Typical S1 unit: Fine-grained calcirudite with stylolites Source: AMEC, 2014

CSA Global Report Nº R302.2019 40 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Figure 7-6: Typical S2 unit: Conglomeratic sandstone with characteristic red fragmental clasts Source: AMEC, 2014

7.4.4 Marl This unit, overlying the S2 and deposited during Santonian time, is a grey marlstone that is typically finely laminated. The marl unit is interbedded with locally present sandstone, andesite, and andesite volcaniclastic rocks (Vasic, 2012). These rocks, ranging in thickness from 50 m to 500 m, are rarely mineralised, and an example is shown in Figure 7-7. In the Bigar Hill deposit area, pyroxene-hornblende diorite sills are emplaced into the marl.

Figure 7-7: Example of Marl Unit: Grey marlstone with deformed laminations, drill-hole BHDD044, 59.2 m Source: AMEC, 2014

CSA Global Report Nº R302.2019 41 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

7.4.5 Andesitic Epiclastics and Diorite Intrusions This unit, of Late Cretaceous age and overlying clastic units S1, S2, and the Marl, is comprised of andesitic shallow intrusive and derivative epiclastic rocks. Rapid facies changes along strike characterise the sequence. The lower part of the epiclastic unit is characterised by polymictic basaltic andesite conglomerate and breccia, whereas the upper part is dominated by monomictic basaltic andesite breccia and conglomerate, which are interpreted being products of epiclastic debris flow deposits. A finer grained sedimentary rock unit, consisting of well-bedded tuff, marl, sandstone, and volcaniclastic breccia that locally forms a thin but mappable horizon between the debris flow deposits. The dioritic stocks, dikes, and sill-like intrusions are generally aligned along a north-westerly trend, which most likely represents a structural fabric in the subsurface that controlled their emplacement. An example of this unit is presented in Figure 7-8.

Figure 7-8: Example of andesite intrusive unit sill with phenocrysts of hornblende and plagioclase, drill-hole BHD010, 101 m Source: AMEC, 2014

7.4.6 Potaj Čuka Monzonite This unit comprises coarse-grained equigranular monzonite with visible alkali feldspar phenocrysts, biotite, and minor magnetite and pyroxene. This monzonite is part of the Late Cretaceous Potaj Čuka pluton which, in the latest Cretaceous (79.8±0.6 Ma; uranium-lead in zircon), intrudes the clastic sedimentary units in the region. The Potaj Čuka pluton is located immediately east of the western margin of the TMC and is elongated in a north-westerly orientation.

7.5 Structural Geology This subsection contains descriptions of the regional geological structure and tectonic-stratigraphic relationships of the region.

7.5.1 Structure The formation of the basin which hosts the TMC is associated with the Alpine Orogeny, which occurred during oblique convergence of the Indian, Arabian, and African plates with Eurasia. Convergence began in the early Cretaceous resulting in an orogenic collage that is characterised by discrete segments that have undergone a distinct geologic evolution. Major phases of mountain building associated with the Alpine Orogeny were ongoing in the Late Cretaceous to Miocene.

CSA Global Report Nº R302.2019 42 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

The TMC is generally considered to represent a basin which has an overall disjointed, elongate lozenge shape, with apparent sinistral, northwest-striking dislocations. These dislocations appear to have controlled basin opening as well as modified the geometry of the TMC. The regional-scale northwest dislocations were second- order structures to an overall dextral, orogen-scale motion resulting from Eocene-Miocene oroclinal bending. The interpreted overall east-west cross-sectional geometry of the TMC is that of a synclinorium. Tertiary Alpine deformation was accommodated by several suites of fault zones that are developed across the entire TMC. Regional cross-sections confirm that the bulk of Alpine deformation was concentrated on the TMC margins, whereas the central part of the magmatic complex was only affected by gentle folding and fault reactivation. Accommodation of deformation by ductile deformation is largely restricted to the eastern and western margins of the TMC and was long-lived, as indicated by open folds and rotation of bedding to sub- vertical dips. Marl units north of the Korkan prospect display vertical dips in road exposures. An east-west cross section of the basin displays strain accumulation toward the eastern and western TMC margins, with inferred synclines and anticlines separated by faults that have accommodated complex movement histories. Post-mineral structures are interpreted as being active during Cenozoic transpressional deformation. These structures include pre-sedimentary rock-hosted gold-bearing structures that were reactivated in addition to newly formed, post-mineral faults. Late orogen-parallel extension produced early Miocene normal faults that controlled the architecture of Miocene coal basins, such as the Žagubica Basin, and numerous regional east- west trending normal faults extending into the TMC. Examples of all these structures occur at Bigar Hill and Korkan, where northwest-trending strike-slip faults, reactivated north-south to north-northeast striking strike-slip faults, and east-west trending normal faults are developed.

7.5.2 Tectonic-Stratigraphic Relationships The spatial relationships between mineralisation styles in the Timok region suggest that the region might be composed of successively westward-migrating metallogenic belts. The belts are temporally distinct events, lying from east-to-west, beginning with the Bor-Veliki Krivelj Belt, succeeded by younger Kuruga high- sulphidation belt, and the still younger Timok diorite porphyry belt. These younger porphyry copper-gold prospects, including the Valja Štrz, the Dumitru Potok, Crna Reka porphyry copper-gold and Čoka Rakita porphyry gold prospects, are present in the eastern part of the Project area. Evidence for sedimentary-hosted gold along the western boundary of the TMC extends over a strike length of more than 30 km and is up to 8 km wide. The mineralised belt was initially identified by soil geochemistry programs conducted by DPM. The geology, geochemistry, and available drill intersections of known prospects suggest a strong similarity to the sedimentary rock-hosted or Carlin-style deposits. Most of Avala’s exploration property is located on the margins of the TMC, namely the Potaj Čuka Tisnica, Bigar Istok and Umka licences (Figure 7-2) on the western margin, and the Lenovac licence on the eastern margin of the TMC. Upper Jurassic to Upper Cretaceous calcareous rocks, including limestone, marl, and calcareous clastic and volcaniclastic rocks, capped by TMC volcanic and derivative clastic rocks, underlie most of the licences. Several overprinting generations of fault systems disrupt the stratigraphy and caused structural complexity, including local reverse faulting and thrusting of stratigraphic units of the TMC area. The sediment-hosted gold prospects: Bigar Hill, Korkan, Kraku Pester and Korkan West, are all part of the Potaj Čuka Tisnica licence and are located on north-south to north-northwest trending segments of the western margin of the TMC, centred around the Late Cretaceous Potaj Čuka monzonite batholith. Exploration within the other licences has been limited to soil geochemistry, trench sampling and scout diamond drilling.

CSA Global Report Nº R302.2019 43 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

7.6 Metamorphism Thermally metamorphosed rocks are present in the contact aureole of the Late Cretaceous Potaj Čuka monzonitic pluton. Calc-silicate skarns are locally present. The most evident thermal effect is present at the Kraku Pester prospect and south of Bigar Hill. At Kraku Pester, the fine-grained clastic sequences adjacent to the pluton are converted to biotite-magnetite and calc-silicate hornfels, depending upon the protolith composition. South of Bigar Hill and the southern-bounding, post-mineral normal fault, the carbonate rocks are converted to marble near the monzonite, with distal bleaching of the normally grey limestone to white colours in outcrop.

7.7 Alteration Except for the quartz-bearing zones in the andesite sill at Bigar Hill, no macroscopically visible silicate alteration minerals are evident. However, the fine grain size of the horizons, coupled with the common evidence for additional post-mineral brecciation precludes easy identification of silicate alteration minerals. Analysis of the geochemical characteristics of the mineralised horizons at Kraku Pester using 1 m drill-hole data suggests that clay minerals, presumably combinations of kaolinite, illite, and probably smectitic clays, from part of the hydrothermal alteration associated with pyrite deposition. Elevated gold is also associated with rocks containing sufficient iron, thus suggesting that the depositional mechanism for gold was likely the sulphidation of iron present in the rocks. The recognition of auriferous concentrations in karst infill sedimentary rocks is consistent with this interpretation as iron is a common residual element during carbonate dissolution. Decarbonisation of diagenetic and detrital carbonate is associated with gold zones.

7.8 Mineralisation Four important mineralised zones have been identified within the Potaj Čuka Tisnica exploration licence. These areas comprise the Bigar Hill deposit, the Korkan deposit, the Korkan West deposit and the Kraku Pester deposit, and are summarised in Figure 7-9. All four zones share a similarity with the style of mineralisation defined at the Bigar Hill deposit and are associated with a large hydrothermal system that has been identified within the Project.

CSA Global Report Nº R302.2019 44 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Figure 7-9: Exploration areas and geology of TMC (inset: Mineral Resource plots of the deposits) Source: Avala, 2018

CSA Global Report Nº R302.2019 45 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

7.8.1 Bigar Hill Deposit The Bigar Hill deposit is the most advanced Mineral Resource within the Project. The deposit comprises Bigar Hill and the adjacent Bigar Au-polymetallic replacement showing (immediately south and east from Bigar Hill) and is located immediately north and outside of the thermal aureole of the Potaj Čuka monzonite. Rock types beneath Bigar Hill comprise of Proterozoic metamorphic basement, Late Jurassic and Early Cretaceous limestone, which are unconformably overlain by a Late Cretaceous clastic sequence (S1, S2 and marl), capped by Late Cretaceous andesitic volcanic and derivative epiclastic rocks. Diorite porphyry has also intruded the stratigraphic package. Bigar Hill is bounded to the north and south by east-west-striking faults that have brought the Late Cretaceous clastic units in tectonic contact with Late Jurassic limestone. Gold mineralisation at Bigar Hill is located principally along two stratigraphic horizons, with lesser amounts present along peripheral steeply dipping fractures zones within the clastic rocks and andesite sill. A lower zone is localised along the unconformable and brecciated lower contact of S1 and isolated karst-infill zones at the upper boundary of the KLS unit. The most continuous horizons lie at shallow stratigraphic levels along the contact between the S1 and S2 units forming a middle zone. Above this zone, gold mineralisation occurs within the andesite intrusive unit forming a sill where gold is associated with narrow zones of quartz infill. At the Bigar Hill deposit, the highest concentrations of mineralisation are along each of the KLS/S1 and S1/S2 contacts as illustrated in Figure 7-10. Metal distribution and thickness variations of the host-rocks suggest the presence of west-northwest and northeast striking sub-vertical feeder structures.

Figure 7-10: Cross section of Bigar Hill deposit Source: Avala, 2018 Mineralisation is continuous and follows the dips of the stratigraphy. It has a north-south extent of approximately 900 m and an east-west extent of approximately 900 m. Mineralisation is largely from surface, and in the south its depth extent is greatest (approximately 500 m). Depth extent reduces to 200–300 m below surface moving further north. There is a small zone in the centre, where mineralisation starts from approximately 80 m vertical depth from surface. Within the basement of the Bigar Hill area (also known as the Rapture Fault zone), located south of Bigar Hill, is a Palaeozoic phyllite comprising a meta-sedimentary sequence composed of sandstone, shale, and conglomerate protolith. Jurassic and Cretaceous limestone are juxtaposed against the phyllite along steeply dipping, normal separation faults; on a regional basis, these rocks unconformably overlie the phyllite unit. Brecciated horizons at Bigar Hill contain clasts of intense calcite network veining, clasts of ferroan carbonate,

CSA Global Report Nº R302.2019 46 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

and local veins with base metal sulphides. A northwest-southeast trending portion of the contact zone discontinuity has localised emplacement of a dioritic porphyry intrusion. Smaller dioritic intrusions define northwest-southeast trends in the phyllite and both northwest-southeast and north-south trends in the overlying sequence.

7.8.2 Korkan Deposit The Korkan deposit is the second most advanced exploration target within the Project, after the Bigar Hill deposit. The deposit constitutes a generally easterly-trending zone of mineralised rocks and incorporates both the Korkan and adjacent Korkan East zones. Korkan East is located to the east of Korkan, across a braided post-mineralisation, strike-slip fault zone. The Korkan deposit shares similar characteristics with the Bigar Hill deposit, located 2 km to the south. Rock types in the Korkan deposit comprise Late Jurassic and Early Cretaceous limestones, which are unconformably overlain by a Late Cretaceous lower clastic sequence (S1 and lower parts of S2) and farther east also by a Late Cretaceous upper clastic sequence (upper parts of S2 and marl), capped by Late Cretaceous andesitic volcanic and derivative clastic rock. Unlike Bigar Hill, stratiform gold mineralisation at Korkan occurs primarily along the unconformable and breccia-like lower contact zone of the clastic S1 sequence, against the underlying KLS limestone unit, and in karst-infill zones at the upper boundary of the KLS limestone unit. It is presumed that erosion has removed some mineralisation related to the S1/S2 contact that would have sat higher in the stratigraphic sequence. Korkan mineralisation along each of the KLS/S1 and S1/S2 contacts is illustrated in Figure 7-11.

Figure 7-11: Mineralisation cross-sections of Korkan deposit (the Korkan East extension is shown in the cross- hatched area on the lower cross section) Source: Avala, 2018 Mineralisation is less continuous at Korkan compared to Bigar Hill, due to higher structural complexity. As at Bigar Hill, it tends to follow the dips of the stratigraphy. The mineralised footprint has a northeast-southwest extent of approximately 1,100 m and a northwest-southeast extent of approximately 1,100 m. Mineralisation commences from surface and can be traced to a maximum depth of 400 m below surface. Structurally, Korkan is dominated by northwest-striking faults which, though apparently associated with mineralisation, have also dismembered mineralisation and earlier structures during late reactivation. Late

CSA Global Report Nº R302.2019 47 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

east-striking faults such as those found at Bigar Hill have been recorded at Korkan but are less important in forming boundaries to the deposit and in juxtaposing stratigraphy. Unlike Korkan proper, significant base metal sulphide minerals accompany gold mineralisation at Korkan East. Overall, the mineralised zones in this environment have the appearance of carbonate replacement deposits. Local repetition or imbrication of stratigraphy and mineralisation are related to north-northeast striking faults.

7.8.3 Korkan West Deposit The Korkan West deposit is the newest discovery within the Project. It lies between the Bigar Hill and Korkan deposits, along a northwest trending structural corridor. The Korkan West deposit shares many characteristics with the Bigar Hill deposit, located approximately 1 km to the southeast, and the Korkan deposit located approximately 1 km to the northeast. Almost all mineralised intervals are manifested as oxide and transitional weathering states. Host rocks for gold mineralisation are: (1) oxidised fine to very coarse-grained (0.1 mm to 2 mm) sandstone belonging to the S1 or S2 units; (2) conglomerate layers containing quartzite clasts and/or limestone clasts (S1 or S2 units). Mineralisation at the S2/S1 contact can commonly be observed. The presence of several oxidised, discontinuous intervals occurring throughout the S1 or S2 units, and associated higher individual gold assays, suggests gold mineralisation may also locally be associated with structurally controlled zones. Non-oxidised, interbedded medium to dark grey coloured calcareous mudstone and fine-grained sandstone beds, known as the IB unit (Interbedded unit), underlies the S1 unit. This unit typically does not host gold mineralisation. A thin sequence of conglomerate and breccia, with angular clasts of limestone within a clay matrix, occurs at the boundary sandstone-limestone and is known as BBX (Basal breccia unit). This unit usually carries no gold mineralisation, which is contrary to the BBX at Bigar Hill and Korkan. Limestone hosted gold mineralisation can be observed in fractured zones proximal to feeder structures, as well as at the Cretaceous-Jurassic limestone contact, in Jurassic limestone and karstified zones. The orientation of structures in the Korkan West area are currently interpreted to be striking predominantly along a west-northwest to east-southeast orientation. These structures are located within a 300 m wide and 600 m long corridor and were most likely the feeder zones for hydrothermal fluids. Structural modelling has demonstrated the presence of additional fault sets striking either northeast- southwest or east-west. It is assumed that in fault intersection zones, particularly in areas where west- northwest trending deep-seated structures intersect northeast-southwest trending structures, strata-bound mineralisation in the S1/S2 units can be observed. East-west trending structures are interpreted as the youngest and are not mineralised. Post-mineralisation faulting can locally displace mineralisation by up to 10 m in places. Figure 7-12 is an example showing the distribution of Korkan West mineralisation relative to the main lithological units.

CSA Global Report Nº R302.2019 48 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Figure 7-12: Cross-section of the Korkan West deposit Source: Avala, 2018

7.8.4 Kraku Pester The Kraku Pester deposit is the third most advanced exploration target within the Project after the Bigar Hill and the Korkan deposits. Kraku Pester shares similar characteristics with the Bigar Hill deposit, 3.7 km to the north, and is located in an embayment at the north-western tip of the Potaj Čuka monzonite. Gold at Kraku Pester is hosted in a variably disrupted stratigraphic sequence comprising, from base to top, shale metamorphosed to biotite ± magnetite phyllite, calcareous rocks; including marl and limestone metamorphosed to calc-silicate hornfels and marble, and tuffaceous rocks that locally may be calcite-rich and interbedded with coherent hornblende andesite. Metamorphism is due to emplacement of the Potaj Čuka monzonite unit that produced a thermal aureole up to 800 m in width. Direct correlation of the stratigraphic sequence at Kraku Pester with those recognised regionally at Bigar Hill and Korkan is uncertain. Mineralisation is less continuous at Korkan compared to Bigar Hill, due to greater structural complexity. As at Bigar Hill, it tends to follow the dips of the stratigraphy. It has a northeast-southwest extent of approximately 700 m and a north-south extent of approximately 600 m. Mineralisation can generally be traced from approximately 30 m below surface, and has a depth extent of 300 m. Disruption of stratigraphic continuity at Kraku Pester indicates structural complication of the host sequence. Low-dipping structures of appreciable thickness are exposed, and fabric asymmetries associated with these faults indicate accommodation of down-dip extension. The presence of massive Jurassic limestone structurally above the heterogeneous Cretaceous sedimentary sequence suggests that the moderately- dipping structures originated as reverse faults that were reactivated. Steeply dipping fault damage zones have also been recognised, and cataclastic zones noted in the monzonite are locally host to auriferous pyrite. Gold deposition is interpreted as being relatively late in the geological-structural evolution of Kraku Pester, post-dating the emplacement of the monzonite. Unlike Bigar Hill, gold mineralisation at Kraku Pester is hosted in brittle fault rocks composed of pyritised fault breccia to cataclasite, with relatively higher gold concentrations being associated with finer-grained cataclasite. Fluid flow associated with gold mineralisation was controlled by a permeability fabric produced

CSA Global Report Nº R302.2019 49 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

by brittle reactivation of a complicated geometric architecture in a north-westerly trending cross fault and the footwall intrusive contact with the monzonite. Figure 7-13 is an example showing the distribution of Kraku Pester mineralisation relative to the main lithological units.

Figure 7-13: Cross-section of the Kraku Pester deposit Source: Avala, 2018

7.9 Metallogeny and Paragenesis Except for Korkan East and Bigar prospects, there is a common character to the sedimentary rock-hosted horizons, regardless of the prospect. The quantity of fine-grained pyrite increases from the margins toward the central and higher gold-content zone in all mineralised horizons. Except for the quartz-bearing zones in the andesite sill at Bigar Hill, no macroscopically visible silicate alteration minerals are evident. However, the fine-grained mineralisation, coupled with the common evidence of additional post-mineral brecciation, precludes easy identification of silicate alteration minerals. Analysis of the geochemical characteristics of the mineralised horizons at Kraku Pester, using 1 m composites, suggests that clay minerals, presumably combinations of kaolinite, illite, and probably smectitic clays, form part of the hydrothermal alteration associated with pyrite deposition. Elevated gold is also associated with relatively iron-rich rocks, thus suggesting that the depositional mechanism was likely the sulphidation of iron present in the host. The recognition of auriferous concentrations in karst-infill sedimentary rocks is consistent with this interpretation, as iron is a common residual element during carbonate dissolution. Decarbonisation of diagenetic and detrital carbonate is associated with gold zones. Previous petrographic studies and metallurgical testwork on the Bigar Hill, Korkan and Kraku Pester sedimentary rock-hosted gold deposits suggest that gold is present in sulphide mineralisation as 0.5–40 μm grain size native gold, electrum or telluride crystals intergrowths with pyrite and other sulphides/sulfosalts, or as solid solution or submicroscopic scaled colloidal gold locked within As-rich pyrite bands (SGS, 2012a, 2012b, 2013; Pacevski, 2012a, 2012b, 2013; Magyar, 2018 ). Most recently, Magyar (2018) showed on SEM images and EMPA maps the arsenic-rich pyrites (potentially associated to Au-mineralisation) have complex

CSA Global Report Nº R302.2019 50 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

growth-zoning in various pathfinder elements, thus indicating several hydrothermal and supergene Au- mineralisation stages and implying variable Au-liberation metallurgical properties. Two textural types characterise the gold-bearing horizons; breccia, and replacement. The breccia-type consists of the basal breccia and karst horizons localised principally along the lower contact between the subjacent carbonate rocks and the overlying calcareous clastic rocks. Many of the breccia horizons are also the locus for post-mineral faulting, thus complicating the interpretation of the original mineralised rock texture. The sedimentary rock-hosted deposit at Bigar Hill and Korkan are characterised by both textural types. The upper horizon along S1/S2 contact is a mixture of stratabound replacement type textures, and brecciated horizons that may, or may not, have formed post-mineralisation. Brecciated mineralised rocks are concentrated along the lower contact of the clastic rocks with the underlying carbonate.

7.10 Weathering Profiles All four of the prospects at the Timok Gold Project show extensive weathering and oxidation of iron bearing minerals. Weathering characteristics vary within each of the stratigraphic settings. A petrographic study by Magyar (2018) into the variability of weathering at the Timok Gold Project is summarised below. Within higher stratigraphic elevations, late Cretaceous marls, andesites and magmatic derived clastics typically exhibit a shallow weathering profile, detectable up to 15 m below surface, which can be extended further downward when in proximity to faulting. These levels within the Timok Gold Project generally contain limited oxide and transitional mineralisation. The S1 and S2 sandstones and conglomerates show pervasive weathering that can extend hundreds of metres below surface. Structural corridors, such as faults and lithologic contacts, allowed meteoric water to permeate downward. When these waters came into contact within mineralised zones, the oxidation of gold-bearing sulphides such as pyrite resulted in the formation of secondary iron oxides such as goethite. Corollary to this decomposition of sulphides, nanoscale gold particles were either liberated and left in situ or taken in solution and re-precipitated as native gold or electrum in geothites (Figure 7-14).

Figure 7-14: SEM imaging of mineralised goethite taken from S1 sandstones in Bigar Hill (sample taken at 30.2 m depth from drill-hole BHDDMET001) Source: Magyar, 2018

CSA Global Report Nº R302.2019 51 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

In certain locations, the meteoric water has pooled beneath impermeable caps, permitting extensive oxidation to occur. This is most notable within the S1 and S2 horizons that are capped by the relatively impermeable marls. In these locations, tabular bodies of sulphide mineralisation can be underlain by zones of oxidation which can extend for many hundreds of metres down-dip. Within the lower levels of the Timok Gold Project stratigraphy, Early Cretaceous and Late Jurassic limestone may also display oxidation controlled along structural pathways. The breccia horizons that mark the unconformable contact between the S1 unit and Lower limestone almost always exhibit a zone of oxidation. The Limestones of the early Cretaceous are typically karsted beneath the unconformity. The sub-tropical climate of the Cretaceous era, coupled with tectonic uplift, resulted in the formation of karstic cavities. These cavities appear to have been infilled with the overlying S1/S2, which often results in the re-deposition of mineralised pyrite and/or secondary iron minerals such as goethite. Fluctuations in groundwater levels resulted in gold grains being remobilised to microfractures on the edges of weathered pyrite grains.

CSA Global Report Nº R302.2019 52 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

8 Deposit Types

The following is taken from the 2014 Technical Report (AMEC, 2014) and remains current. The dominant mineral prospects in the clastic sedimentary rocks along the western margin of the TMC are relatively low-temperature auriferous deposits that share many characteristics with Carlin-type gold deposits, as outlined by Cline et al. (2005). The interpretation of the sediment -hosted gold prospects within the project area as Carlin-type is based upon the following criteria (Knaak et al., 2016): • Character of the sedimentary host. • The metal association (gold, arsenic, mercury, thallium, sulphur and antimony). • The fine-grained nature of the gold, high gold-to-silver ratio and alteration types including argillisation, decarbonisation, and locally, addition of quartz. Sulphidation reactions appear to have controlled gold deposition, although the potential influence of a simple redox boundary along stratigraphic horizons and a decrease of gold solubility of mineralising fluids due to temperature decrease cannot be discounted. The anomalous prospects associated with sedimentary rocks within the NW Timok area are Korkan East and Bigar, where significant gold is associated with carbonate replacement deposits composed of a variable assemblage of sphalerite-galena-arsenopyrite ± chalcopyrite concentrated along the brecciated contact between limestone and the overlying clastic sequence. Additionally, at NW Timok porphyry copper-gold and gold-only deposits are associated with hornblende-biotite-plagioclase-phyric diorite porphyry intrusions emplaced into the andesitic volcanic and volcaniclastic rocks. The most significant and previously known deposits are the Valja Štrz and the Dumitru Potok porphyry copper-gold deposits. Exploration since 2000 has discovered the Kraku Ridji and Crna Reka porphyry copper-gold and Čoka Rakita porphyry gold prospects, largely as a result of soil and stream sediment geochemical survey by DPM. Although spatially related, the timing and genesis of the sediment-hosted Au systems are uncertain as these deposits are always separated from porphyry gold-copper and polymetallic replacement deposits by faults. The current understanding is that the various Late Cretaceous Au mineralisation types from NW Timok form a continuum and are part of larger magmatic-hydrothermal system(s) and represents various lithological traps (intrusive host, contacts, limestone replacement, clastic sediments replacement) or temperature segments (from porphyry toward epithermal) of the same system (Knaak et al., 2016). The sediment-hosted gold belt lies west of a well-endowed metallogenic belt containing a range of magmatic- related deposits, including high sulphidation copper-gold and porphyry copper-gold. These deposits have formed the basis of significant mining activity at Bor for over 100 years. Exploration by Avala and DPM has defined the previously unrecognised sediment-hosted gold prospects along the western margin of the TMC.

CSA Global Report Nº R302.2019 53 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

9 Exploration

9.1 Introduction Intensive exploration at the Project commenced in July 2010 following the acquisition of the projects by Avala Resources Ltd. A systematic exploration approach has been undertaken with the assembly of the following data sets over the whole Project area: topography, geological mapping, rock chip sampling, and stream sediment geochemistry. Stream sediment sampling was previously completed over the entire Project area, at a nominal density of one sample per square kilometre. Anomalous areas were followed up by rock chip sampling, mapping, and soil sampling, on a first-pass 400 m x 50 m grid in some of the anomalous areas and, very locally, subsequently with 100 m x 50 m grid sampling. Soil anomalies were usually followed up by trenching and drilling over anomalous zones including the Bigar Hill, Korkan and Kraku Pester deposits, and the Umka, and Bigar exploration prospects. During 2016, with the assistance of the MoM&E, a portion of the Potaj Čuka Tisnica licence was incorporated into a new licence called Umka. Alongside this, certain exploration licences have been relinquished since the last Technical Report (as described in Section 4.3). Thus, direct comparison of physical exploration statistics from previous reports (AMEC, 2014; Coffey Mining, 2010) will differ from those presented in Table 9-1. Soil and outcrop sampling have taken place across the whole licence up to present day, and includes the Bigar Hill, Korkan, Korkan West and Kraku Pester deposit areas. Table 9-1: Soil and outcrop sampling Exploration licences Soil samples Rock chips Potaj Čuka Tisnica, Bigar Istok and Umka 6,159 1,117

9.2 Geological Mapping Outcrop exposure over the exploration licences is generally poor. However, in areas with outcrop, ground geological mapping together with rock sampling was undertaken. All existing surface outcrops have been mapped, including those created by earthworks activities associated with drill pad construction and cuttings for access roads. Geological maps were created using available observed lithology, alteration and structure data, followed by interpretation. This has improved the definition of the geology in plan, with cross-checking during three-dimensional (3D) modelling of drill results for the Bigar Hill, Korkan, Korkan West and Kraku Pester areas.

9.3 Outcrop Sampling Rock chip sampling has been conducted by Avala across the project area. Rock samples, representing a wide range of rock types, were taken and analysed for gold by 50 g fire assay with atomic absorption finish. Pathfinder elements were analysed using multi-element inductively coupled plasma-mass spectrometry (ICP- MS) analysis covering 53 elements. All sample locations were surveyed by handheld GPS. Data for each sample, including lithology, sample description, coordinates and assay results, are stored in an acQuire database. A total of 1,117 outcrop rock chip samples were collected by Avala over the Potaj Čuka Tisnica, Bigar Istok and Umka exploration licences.

CSA Global Report Nº R302.2019 54 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

9.4 Soil Geochemistry Soil sampling has proven to be a very effective exploration method for localising potential sediment-hosted mineralisation. Gold, as well as low-temperature pathfinder elements such as arsenic, mercury, and thallium, have been found to be important elements in soil geochemistry surveys. Avala collects soil samples from small pits, which are hand-dug by a sampling team. All samples are collected from the lower B-horizon. In the Potaj Čuka Tisnica licence area, most samples were collected at depths of 0.5 m to 1 m. Sampling was conducted in a grid pattern, beginning with a grid line spacing of 400 m and sample collection at 50 m intervals along each line. Follow-up or detailed sample grids were configured at a line spacing of 100 m, with 50 m samples collected along each line. The sampling approach was based on orientation surveys completed by Avala in a similar environment from the Eastern Rhodope Mountains of Bulgaria. Soil field duplicates are collected at frequency of 1:20. Soil samples are collected by Avala field staff and transported to the Avala core storage facility in Bor on the same day they are sampled. Soil sampling programs were completed from July 2010 to 2018, building on previous soil sampling programs. Avala collected 6,159 soil samples over the Potaj Čuka Tisnica, Bigar Istok and Umka licences. Avala and Dundee soil sampling is shown in Figure 9-1. Details are tabulated in Table 9-1 (above).

CSA Global Report Nº R302.2019 55 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Figure 9-1: Location of soil sampling lines Source: Avala, 2018

CSA Global Report Nº R302.2019 56 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

9.5 Trenching Trenching was used as a follow-up strategy to explore areas with anomalous soil geochemistry and to assist in defining key geological relationships due to the limited outcrop in the Project areas. There was a high success rate in intersecting sediment-hosted gold mineralisation by drilling near extensive and well mineralised trench intercepts. Trenching activity was focused in the Potaj Čuka Tisnica licence area. In the period from 2010 to 2013, approximately 296 trenches (34.5 km) were completed over the Potaj Čuka Tisnica licence. Trenching in this period was concentrated on the Bigar Hill, Korkan, Kraku Pester, and Umka zones. An additional 135 trenches and channels were completed between 2015 and 2018, for 11,236 m on Potaj Čuka Tisnica, Bigar Istok and Umka licences. Avala’s trenching locations are shown in Figure 9-2. Trenches were completed under the supervision of exploration geologists. The dimensions of the trench are set out according to safety regulations, with a maximum depth of 2 m and a minimum width of 0.8 m. During excavation, the upper humus layer is separated from the underlying soil material so that it can be replaced and revegetated during rehabilitation. Trenches were sampled as channels, with channel samples collected just above the trench floor at either 1 m or 2 m intervals. Except where extensive soil cover is encountered, trenches are sampled in their entirety. The samples were routinely weighed prior to final bagging to maintain an even sample size and to avoid sampling bias in harder rock types. An average channel sample weight of 3 kg/m was maintained. Field duplicate samples and certified standards were taken at a frequency of 1:20. All data collected in the field is routinely entered into geology and structural geology spreadsheets using Field Marshal software and later exported to an acQuire database. Channel samples are collected by Avala field staff and transported to the Avala core storage facility in Bor on the same day they are sampled.

CSA Global Report Nº R302.2019 57 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Figure 9-2: Trenching at the Timok Project Source: Avala, 2018

CSA Global Report Nº R302.2019 58 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

9.6 Exploration Drilling In 2009, four diamond drill-holes were drilled at the Project (Potaj Čuka Tisnica exploration licence). Two drill holes were drilled on the Kraku Pester area (PEDD001 and PEDD002), and two in the Bigar (Rapture Fault zone) area (BIDD001 and BIDD002). Avala then focused exploration drilling campaigns from 2010 to 2013 on the Potaj Čuka Tisnica licence to outline mineralisation on the Bigar Hill, Korkan, Kraku Pester, and Umka areas. The drilling that relates to Bigar Hill, Korkan and Kraku Pester is covered in more detail in Section 10. After 2014, several exploration drill-holes were completed at wide space on areas around mineralised prospects which led to the discovery of Korkan West deposit during winter 2016/2017.

9.7 Topographic Surveys All survey activities are conducted using by a licensed third-party surveyor. A base geodesic operational network within the Timok Gold Project has been established that covers the entire exploration tenement areas. This primary survey control network was implemented using AUSPOS, an online GPS processing service provided by Geoscience Australia. High resolution topographic surveys are completed using GPS using a real time kinematic method which provides a centimetre level of precision. The system (Trimble R8 GNSS) uses two receivers; one is centred on control point with known coordinates whilst a second receiver is mobile and is used to determine survey points across the terrain. All coordinates are recorded using UTM coordinate system, specifically Zone 34 North in WGS 84 datum. The Korkan West topographic surface was surveyed in November 2017, whilst the remaining prospects were surveyed between 2011 and 2012.

9.8 Conclusions The author/Qualified Person did not observe sampling while at site, because no exploration activities were underway during the site visit completed in February/March 2017. However, from review of procedures, maps, and discussions with site personnel, the sampling methods and sample quality appear to have not resulted in sample biases and have befit exploration programs that allow for follow-up targeting through drilling, as described in Section 10. The author/Qualified Person believes these exploration programs to be systematic in their nature to provide samples that are representative, and no factors have been identified that may have resulted in sample biases.

CSA Global Report Nº R302.2019 59 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

10 Drilling

10.1 Introduction Avala has employed a combination of diamond drilling and RC drilling across the Bigar Hill, Korkan, Korkan West and Kraku Pester exploration areas, and diamond drilling at Umka. Drilling has been undertaken since July 2010. Drilling was carried out by Serbian contractors using Atlas Copco CS-14 and Atlas Copco Mustang 9/13/18, Alton HD, Coretech, YDX 1300G and Gemex MP 1200 rigs for diamond drilling, and GEMSA 500RC rigs for RC drilling. Examples of drilling activities are shown in Figure 10-1, while drilling operations are summarised by area in Table 10-1.

Figure 10-1: Diamond (left) and RC drilling (right) at Bigar Hill Source: AMEC, 2014 Table 10-1: Summary of drilling for the main resource areas of the Timok Gold Project RC pre-collar/ Diamond drilling RC Total Prospect diamond tail Drill holes m Drill holes m Drill holes m Drill holes m Bigar Hill 91 22,967 333 71,287 30 3,423 454 97,677 Korkan (including 215 50,564 295 49,804 9 1,237 519 101,605 Korkan West) Kraku Pester (including Kraku 51 12,478 94 14,962 7 960 152 28,400 Pester South) Total 357 100,936 722 136,053 46 5,620 1,125 227,682

CSA Global Report Nº R302.2019 60 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Figure 10-2: Drilling completed at Bigar Hill (left) and Korkan (including Korkan West) (right) Source: CSA Global, 2018

CSA Global Report Nº R302.2019 61 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Figure 10-3: Drilling completed at Kraku Pester Source: CSA Global, 2018

10.2 Methodology and Planning, Site Preparation, Setup and Rehabilitation After testing exploration targets with a detailed trenching program, prospects were evaluated with two to three drilling campaign stages, as follows: • First stage: Wide-spaced diamond drill-holes on a nominal grid spacing of 160 m x 160 m, with the objective of outlining the boundaries of the deposit. • Second stage: Infill drilling, using RC drill-holes, at nominal 80 m x 80 m spacing. Diamond drilling is used to support RC holes by twinning about 15% of RC holes. • Third stage: Delineation drilling, using RC drill-holes, at a nominal grid spacing of 40 m x 40 m. The majority of drill holes at the Bigar Hill project are orientated at an azimuth of 270° and inclined approximately 60° to the west. In the case of Kraku Pester, drill holes are mostly inclined at 60° to the east to intersect gently west-dipping mineralisation. At Korkan and Korkan West, however, the orientation of the mineralisation is much more variable than at either Bigar Hill or Kraku Pester, and this is reflected in the greater range of drill-hole orientations. The central portion of the Korkan deposit is dominantly explored using holes inclined between 60° and 70° to the northeast. Toward the north, most of the drilling is inclined

CSA Global Report Nº R302.2019 62 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

between 50° and 60° toward just south of west. A significant number of holes are aligned and inclined in a variety of angles outside of these two patterns.

10.3 Collar and Downhole Surveying Diamond drill-hole downhole surveys are carried out by drilling contractors at 30 m intervals. Typically, a Devi Tool digital multi-shot camera is used for diamond holes. RC holes are surveyed at intervals of about 48 m using a Globaltech Pathfinder S@W survey tool after the drill hole has been completed and drill rods have been extracted. On a few occasions, an Eastman single-shot camera was used on both diamond and RC holes. Survey results indicate downhole deviations from the drill-hole collar azimuth and dip measurements are typically small.

10.4 Drill-Hole Logging, Data Acquisition and Sampling This subsection describes the methods and protocols used for RC and diamond core drilling.

10.4.1 Reverse Circulation Drilling Avala staff and drilling contractors followed a comprehensive set of drilling quality control and safety procedures for all RC drilling programs. All RC drilling was conducted under constant on-site supervision by the Rig Geologist. RC drilling was undertaken using downhole hammers with face sampling drill bits. All drilling and sampling were confined to dry downhole conditions. Predominantly 141 mm and, to lesser extent, 147 mm and 139 mm drill bits were used with a shroud annulus of 2-3 mm to enhance sample recovery. All collars were lined with a 6 m casing of PVC pipe. To ensure sampling was under dry conditions, and to enhance sample recovery, two 1250 cfm compressors and an 870 psi booster were used at each drill site. Pressurised air blowbacks were routinely used after every metre of advance so that all the material within the drill stem was displaced into the sample bag prior to advancing to the next metre. At every rod change, compressed air blowdowns were used for cleaning the air system and for conditioning the hole before drilling resumed. If drilling could not be continued under dry conditions, the RC drill-hole was abandoned and re-entered using a diamond core drill to advance the hole. A dedicated compressed airline from the rig compressor was available at all times for cleaning of the cyclone and the sample splitter. All RC sample splits were collected daily by Avala staff from the drill rigs and transported to a secure core-shed facility in Bor where they were maintained under 24-hour security by Avala staff. RC drilling samples have been routinely collected at 1 m intervals. Drill cuttings for each drilled metre are collected in a new plastic bag and marked with the drill-hole number and interval sampled. Each bag of cuttings is weighed at the drill site using electronic scales. Cutting weights are recorded using handheld data loggers for input into the acQuire database and are monitored in real time during drilling for consistency using expected weights based on drill rods, bit sizes and shroud sizes being used and rock types. Changes in the weight of cuttings are also monitored by evaluating the statistical variations of cutting weights for each drill hole. Routine sampling procedures require that the cyclone be cleaned at each rod change and after a wet sample. At every rod change, any material in the hole is cleared before the first new sample is collected. The riffle splitter is cleaned with compressed air and bottle brushes after each sample is split. The average sample recovery is 88%, with an average 1 m sample bag weighing 38.1 kg.

CSA Global Report Nº R302.2019 63 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Upon arrival at the Avala core shed in Bor, all RC samples are measured for magnetic susceptibility, using a handheld meter. A small sample split is washed, and the chips kept in a chip tray for reference.

10.4.2 Diamond Drilling Avala staff and drilling contractors followed a comprehensive set of drilling quality control and safety procedures for all diamond core drilling programs. Diamond drilling was carried out such that drill holes were always started using PQ core and then reduced to HQ triple tube (HQ3) once competent rock had been intersected. The diamond drill core size was maintained at HQ3 for as long as possible. NQ2 core diameter was used to extend RC holes that had not reached target depth because of drilling difficulties. Core was transferred directly from the core barrel into appropriately labelled aluminium core boxes to ensure that core was correctly placed, and no core was lost. Wooden core blocks were placed between runs, recording the length of the run and any core loss. Forced breaks made by the drillers were marked on the core with a red cross on both sides of the breaks. At the drill site, core was washed clean of surface mud or other drilling fluids. All core boxes were labelled with the drill-hole number, starting and ending depths for the core box, and box number. Drill core orientation procedures were carried out at approximately 3 m intervals, and less in mineralised zones or areas of poor ground conditions. EzyMark, or occasionally spear-orientation equipment was used to mark the orientation of drill core. Core boxes were collected by Avala staff at least once a day from the drilling rigs and transported to the Avala core storage facility in Bor on the same day. For transportation, core boxes lids were fitted by adhesive-coated fastening tape, and boxes were firmly secured with strapping in the transport vehicle. Diamond drilling core recovery averaged 98%. The majority of drill core was HQ3 size, followed by PQ3 and a small proportion of NQ. Specialised drilling muds and polymers were used throughout the program to maximise core recovery and, in areas of poor core recovery, drill runs were reduced to less than 0.5 m. At the Avala core facility in Bor, all core is photographed dry and wet using a digital camera before logging commences. Core photos record the drill-hole number, box number, starting and ending depths, and date. Photo sets are integrated with the Avala acQuire drill-hole database. Logging procedures are initiated with geotechnical logging, during which rock quality designation (RQD), joint strength and roughness, rock strength classification, and detailed core recovery are recorded. Core with drilling orientation marks is aligned with adjacent core intervals so that an orientation line can be drawn more or less consistently over most of the drill core. Geological structures are measured on the basis of alpha, beta, and gamma angles relative to the orientation line. True orientations of features are determined using either a jig or by calculation. Geological logging is recorded using a digital logging form that provides an extensive geological description through a system of codes for lithology, alteration, veins, mineralisation, weathering, and vein descriptors. After core logging has been completed, core is marked up for sampling at regular 1 m intervals corresponding to drilled depths. The 1 m sample intervals may be adjusted at key geological contacts or in sample intervals with significant core loss. These intervals must be less than 1.5 m and greater than 0.5 m long. Core is split along orientation lines using a diamond saw. Half the core is placed in a heavy cotton sample bag, together with a sample tag. Core samples weigh (on average) 3–4 kg. The remaining split core is replaced in the core box and retained at Avala’s core shed facilities in Bor.

CSA Global Report Nº R302.2019 64 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

10.5 Deposit Drilling Diamond drilling and RC drilling form the basis of modelling and tonnage-grade estimation mineralisation at each of the Bigar Hill, Korkan (including Korkan West), and Kraku Pester deposits. Bigar Hill drilling consists of 91 diamond core drill-holes for 22,967 m, and 333 RC drill-holes for 71,287 m. A further 30 holes were drilled with RC pre-collars and diamond drill tails, for 3,423 m. Twenty-four drill holes were twinned to confirm repeatability of drilling methods in identifying mineralisation. Korkan (including Korkan West) drilling consists of 215 diamond core drill-holes for 50,564 m and 295 RC drill holes for 49,804 m, as well as nine drill holes for 1,237 m that were drilled with RC pre-collars and diamond tails. This included 18 drill holes which were twinned. Drill-hole collar locations for Bigar Hill and Korkan (including Korkan West) are shown in Figure 10-2. At the Kraku Pester deposit, drilling consists of 51 diamond core drill-holes for 12,478 m and 94 RC drill-holes for 14,962 m. A further seven drill holes were drilled with RC pre-collars and diamond tails, for 960 m. Seven drill holes were twinned to confirm repeatability of drilling methods in identifying mineralisation. Drill-hole collar locations for Kraku Pester are shown in Figure 10-3. Drilling was generally done perpendicular to the mineral deposits to attempt to intersect the true thickness. The author/QP has identified no drilling, sampling or recovery factors that could materially impact the accuracy and reliability of the results. Representative examples of drill sections through the four mineral deposits are presented in Figure 10-4 to Figure 10-7.

Figure 10-4: Cross section showing drilling and interpreted mineralisation at Bigar Hill Source: CSA Global, 2018

CSA Global Report Nº R302.2019 65 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Figure 10-5: Cross section showing drilling and interpreted mineralisation at Korkan Source: CSA Global, 2018

Figure 10-6: Cross section showing drilling and interpreted mineralisation at Korkan West Source: CSA Global, 2018

CSA Global Report Nº R302.2019 66 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Figure 10-7: Cross section showing drilling and interpreted mineralisation at Kraku Pester Source: CSA Global, 2018

10.6 2017/2018 Metallurgical Drill-Holes Avala completed a series of six metallurgical twin-holes during December 2017 in order to collect sulphide material for initial scoping testwork carried out in 2018. An additional seven drill holes were drilled in 2018 to prepare additional metallurgical composites for testing in 2019. The metallurgical holes and sampling are detailed in Section 13.2.

CSA Global Report Nº R302.2019 67 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

11 Sample Preparation, Analyses and Security

This section of the report includes descriptions of sampling methods for different sampling types, sample preparation and analysis and descriptions of quality control (QC) procedures employed along with the results of the QC sampling.

11.1 Field Sample Preparations Avala collected different types of samples including density, soil and trench samples and samples from RC and diamond core drilling. Sample preparations conducted by Avala prior to delivery to the laboratory are described below.

11.1.1 Soil and Trench Samples Soil field duplicates are collected at frequency of 1:20. Blanks and low-level gold certified reference standards are inserted at the same frequency. Trench field duplicate rock samples and certified standards were taken at a frequency of 1:20. As of Q1 2017, blanks were similarly inserted at a frequency of 1:20 samples.

11.1.2 Reverse Circulation Hole Samples Drill cuttings are split using a Jones three-tier riffle splitter to provide a sample that will be submitted to a laboratory for analysis. A typical split sample weighs approximately 4–5 kg. RC field duplicates, pulp duplicates, and certified standard reference material are submitted at a frequency of 1:20 samples. Blank samples consisting of un-mineralised quartz sand are submitted at a frequency of one for each drilling location at the start of the drill-hole sample sequence. Umpire samples are submitted at a frequency of 1:20 to either Genalysis/Intertek (accredited with the National Association of Testing Authorities, Australia), Perth, or ALS Chemex (accredited to the requirements of ISO/IEC 17025), Vancouver. The labs are independent of Avala, DPM and CSA Global.

11.1.3 Diamond Drill Core Hole Samples Core “field duplicates” are prepared by producing split samples after the jaw crushing stage of sample preparation, with each split being assigned a unique sample number. Pulp duplicates and certified standard reference material are submitted into the assay sequence at a frequency of 1:20 samples. Blank samples of un-mineralised quartz sand were submitted at one in every batch submitted to the analytical laboratory at the beginning of the batch sample sequence. The procedure was updated in 2017, wherein coarse blanks (rocks) are now used instead of sand and blanks are now inserted at a 1:20 frequency. Umpire samples are submitted, at a frequency of 1:20, to either Genalysis/Intertek (accredited with the National Association of Testing Authorities, Australia), Perth, or ALS Chemex (accredited to the requirements of ISO/IEC 17025), Vancouver and ALS Bor during 2019 (5% of samples that have >0.1 ppm Au). The labs are independent of Avala, DPM and CSA Global.

CSA Global Report Nº R302.2019 68 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

11.2 Laboratory Sample Preparation and Analyses

11.2.1 Laboratory Sample Preparation Sample preparation for all samples (soil, trench, channel sample, RC and diamond core) is undertaken at the SGS Bor (SGS) sample preparation facility in Bor. This facility is owned by Avala, but independently managed by SGS, such that the chain of custody is transferred from Avala to SGS at the laboratory door. The SGS facility is located adjacent to Avala’s core shed facilities in Bor. All submissions to the sample preparation facility are accompanied by sample submission forms with instructions for preparation methods, insertion-of-standards protocols, and analytical process codes. Once the samples are delivered to the SGS sample preparation facility, chain of custody records are maintained until reject sample pulps are returned to Avala’s jurisdiction. All samples submitted to the facility are initially dried at 105°C for a minimum of 12 hours. Core, trench, and rock samples are then crushed to 4 mm, using jaw crushers. Crushing is checked by confirming that 85% of the crushed material can pass through a 4 mm sieve. Core “field duplicates” are produced by splitting crushed samples on a 1:20 basis at the jaw crusher output stage. Each “field duplicate” subsample is assigned its own identification number for the remainder of the assay procedure. All crushed sample material is then pulverised using LM5 pulverising mills (of which there is currently a bank of eight). RC drilling samples are pulverised in their entirety using the LM5 pulverising mills. A standard part of the SGS laboratory operating procedures is for 1:10 pulps to be wet-sieved using a motorised sieve bank in order to confirm that the sample passes a P90 of 75 µm. If a sample fails the test, the previous 10 samples are re-pulverised. Pulverised material, from all types of sample, is split into 250 g and 600 g pulps, where the former is used for assay determination, and the latter is stored as part of the reference pulp library which is securely stored within the Avala sample office facility. An additional 250 g pulp duplicate is split from the pulverised material at a frequency of 1:13.

11.2.2 Laboratory Analyses Routine analysis of all samples is currently performed at the SGS analytical laboratory in Bor, or previously at the SGS analytical laboratory in Chelopech, Bulgaria. The labs are independent of Avala, DPM and CSA Global. All laboratory methods, procedures, and QAQC protocols are consistent with standards adopted by SGS worldwide standards. Gold analysis methodology is conventional 50 g fire assay, with an atomic absorption finish. Silver and base metal analyses (copper, molybdenum, arsenic, bismuth, lead, antimony and zinc) are performed using a 0.3 g charge, aqua regia digestion, and atomic absorption analysis. Sulphur samples are analysed by combustion with an infrared finish. The Bor and Chelopech laboratories are not ISO 9002 or ISO 17025 accredited for the above analytical procedures. However, the procedures routinely used at both the SGS laboratories include the following established and standard specifications at all SGS laboratories worldwide: • Cross-referencing of sample identifiers. • Use of compressed air gun and vacuum gun, along with routine barren quartz “washes”, for cleaning of crushing and pulverising equipment. • Routine assaying of quartz washes. • Assaying of SGS-submitted certified standards at a rate of two per batch of 40 original samples. • A minimum of 10% of submitted samples are subject to repeat analysis.

CSA Global Report Nº R302.2019 69 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Second splits generated by the SGS CCLAS system are produced at a rate of 1:13 and represent a second subsample taken from the LM5 pulverised pulp. All soil samples were assayed by ALS Chemex in Perth and SGS Vancouver, using methods Au-TL43 (gold by aqua regia digestion with ICP-MS and ME-MS41 (combined ICP-MS and ICP-AES dependent on concentration). Elements assayed for are silver, aluminium, arsenic, boron, barium, beryllium, bismuth, calcium, cadmium, cerium, cobalt, chromium, caesium, copper, iron, gallium, germanium, hafnium, mercury, indium, potassium, lanthanum, lithium, magnesium, manganese, molybdenum, sodium, niobium, nickel, phosphorous, lead, rubidium, rhenium, sulphur, antimony, scandium, selenium, tin, strontium, tantalum, terbium, thorium, titanium, thallium, uranium, vanadium, tungsten, yttrium, zinc and zirconium. The ALS Chemex laboratory in Perth is certified to ISO 9002, but is not ISO 17025 accredited for this technique. The lab is independent of Avala, DPM and CSA Global. An ICP-MS machine has recently been installed and brought online at the SGS Bor laboratory where relevant samples are analysed for 49 elements. Umpire pulp aliquots, sent to two external accredited labs, are assayed for gold by 50 g fire assay, with an atomic absorption finish. Silver is analysed using a 0.3 g charge, aqua regia digestion, and atomic absorption analysis. Sulphur is analysed by combustion furnace. Pulp aliquots for dispatch to other laboratories (abroad) are packed in boxes which are plastic-wrapped or taped-shut for transport in sealed containers. The sealed sample boxes, accompanied by chain-of-custody documents, are transported door-to-door by an international courier delivery company. Reject pulps, returned to Avala jurisdiction, are stored in an enclosed “pulp library”, with access through secure key card only.

11.2.3 Bottle Roll Testwork Program A cyanide-gold leach bottle roll work program test was conducted in 2014 on 3,930 five-metre composite samples taken from diamond and RC drill-holes from the Bigar Hill (1,810), Korkan (1,201) and Kraku Pester (919) prospects. The samples were ground to 85% passing 75 μm and subsequently 200 g subsamples were analysed by conventional fire assay (“Au-FA505-ppm”) and then processed for agitated cyanide leach using tap water at ambient temperature for four hours using the SGS LeachWELL method. The SGS LeachWELL is an accelerated partial digest technique designed to determine the cyanide extractable gold content of samples. The settled solutions were analysed for gold by AAS (“Au-leached-ppm”, Au- LWL69J_ppm). The post-leach residues were washed, dried, re-ground and analysed by 25 g fire assay with atomic absorption spectroscopy (AAS) finish to determine the undissolved gold contents; each measurement was replicated (“Au-residue-avg-ppm”, Au_FAA303_ppm and Au_FAA303R_ppm). The residue and solution assays were used to calculate the total gold content of the samples and a recovered percentage of gold leached by the cyanide solutions.

11.2.4 Dry Bulk Density Measurements Bulk density measurements are restricted to diamond core only. Half-core samples of 20–30 cm length are collected at an interval frequency of approximately every 3 m of all drilled core. These core lengths are submitted to the SGS sample preparation facility at Bor for determination using a wax-sealed core water immersion method. After measurements have been completed, the core was returned to Avala and replaced in the core boxes.

CSA Global Report Nº R302.2019 70 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

An external check of 188 bulk density measurements was performed by sending samples for retesting to the Evrotest-Control laboratory in Sofia, Bulgaria, which is certified for BS EN ISO 9001:2008. The lab is independent of Avala, DPM and CSA Global.

11.3 Avala Assay QAQC Procedures Avala performs routine checks on every laboratory submission upon import to the drill-hole database, using acQuire QAQC tools. These checks are initially undertaken on receipt of the assay results, in order to determine if the submission has passed the Avala control test. If the submission fails, it is re-assayed. On a monthly basis, the QAQC data in general is assessed using custom acQuire tools to identify any QC issues or trends, so they can be acted on in a timely manner. Failures in QC samples can be immediately discussed with the analytical laboratory and, if needed, batches can be rapidly resubmitted. Avala routinely inserts internationally certified standards, covering a wide grade range, along with blanks, into the sample submission stream. The samples are in standard pulp packets, but the recommended values of the samples are unknown to the SGS laboratories. The standards and blanks are inserted at a rate of 1:20 samples. In addition, Avala has produced, as part of the sample sequence, RC field duplicates, which are also unknown to the SGS laboratory. Coarse crush duplicates have been produced from the diamond core samples by SGS and included for analysis. Avala considers certified reference material (CRM) that assays 10% outside of the expected value for gold, or 15% outside of the base metal expected values to be a failure and will require the laboratory to re-assay 10 samples prior to, and 10 samples following the failed QC assay. This instruction includes the submission of standard reference material control samples. If more than two standards have failed in a submission, the entire submission will be required to be re-assayed. If a failed standard is amid a sequence of results below the detection limit, it is up to the geologist assessing the data to determine if re-assay is required. As part of the Avala’s standard QAQC program, 6,417 umpire samples from the Bigar Hill, Korkan, and Kraku Pester deposits drilling programs were submitted to ALS Chemex in Vancouver BC, and 6,445 samples were submitted to Genalysis/Intertek in Perth for gold analysis. A further 127 samples were submitted to ALS Chemex in Bor in 2019. The results of these QC checks indicated no biases between the umpire laboratories and gold assays from SGS Bor. Duplicate data from the Project, submitted to the SGS Bor laboratory, were analysed using HARD, HRD, Thompson Howarth, scatter and quantile-quantile (Q-Q) plots. The results indicate no bias, along with a high repeatability or precision for duplicate QC data, showing a decrease in variance due to increased sample homogeneity.

11.4 Conclusions and Recommendations The author/Qualified Person did not have an opportunity to review any drilling or sampling during the field visit as this had already been completed, but procedures were reviewed, and discussions held with DPM geologists and the SGS laboratory manager regarding past and current practices. The author/Qualified Person concludes that the sample preparation, security, and analytical procedures are robust and follow industry best practice. The QAQC procedures are comprehensive and are suitable to monitor assay contamination, accuracy and precision. The author/Qualified Person notes that the failure limits used for the standards should be adequate, although it is more common to use the standard deviations to obtain acceptable limits. Any standard result that varies from the expected value by more than three standard deviations, or any two consecutive standards differing more than two standard deviations would constitute a failure. The

CSA Global Report Nº R302.2019 71 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

author/Qualified Person concurs with the conclusions above that no significant bias between labs was observed in the gold external check samples.

CSA Global Report Nº R302.2019 72 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

12 Data Verification

12.1 Data Verification Completed by CSA Global

12.1.1 Collar Locations On Tuesday, 28 February 2017, GPS coordinates from six collar locations were collected at Bigar Hill. Weather conditions were such that most of Bigar Hill and all of Korkan and Kraku Pester were inaccessible due to accumulation of winter snow, and drifts on the roads which made them impassable by four-wheel drive. In most cases, the collar cap and steel tag were missing, and it was not possible to identify the drill-hole name. In those cases, the drill-hole name was inferred from drill-hole plans. In two examples, the cap and steel tag remained intact. In addition, many holes have been rehabilitated, since most drilling was completed in 2012/2013. It was confirmed that the checked drill-hole collars matched the recorded coordinates, and drill-hole plan within an acceptable tolerance, accounting for resolution differences between handheld GPS devices and surveyed collar locations.

12.1.2 Source Data Verification DPM maintain comprehensive hard copy records with a file for each drill hole or trench. These files include the following documents: • Drill site establishment sheet. • Geology log sheet printout. • Drill-hole survey records. • Drill plots. • Chain of custody (drill rig to core yard). • Sample submission sheet/s (fire assay and sulphur, bulk density, multi-element composites). • Collar survey. • Drill site rehabilitation sheets. Eight drill holes (three from Bigar Hill, three from Korkan and two from Kraku Pester) were randomly selected and the database data checked against the hard copy files. Collar coordinates, drill-hole survey records and lithology records were compared, and no issues were noted. Files are maintained for all assay submission batches which contain copies of the assay submission sheets as well as printouts of the assay results, QAQC report and QAQC graphs. These were reviewed for the eight drill holes above, and no issues were noted. Database gold assay results were compared against the laboratory supplied files for 11 drill holes containing significant grade. No issues were noted in this comparison. The author/Qualified Person notes that the source data appears to have been accurately captured in the database and that therefore the database should be able to confidently be used in downstream work.

12.1.3 Database Validation To ensure that the data were validated, the CSV files provided were imported into a SQL (Structured Query Language) relational database, which is an industry best practice standard for exploration project databases.

CSA Global Report Nº R302.2019 73 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

The database schema used is the Maxwell DataShed model; which contains validation constraints and triggers, ensuring that data loaded meets the following validation rules: • Data is captured in the correct format: o Real number: This is a number such as a drill-hole depth, coordinate, etc. In some cases, there can be a constraint on a number (e.g. a number which is a percent should be ≤100). o Date: Set format such as dd/mm/yyyy. o Text: Usually a comment. o Library field: A library field (lookup) has a predetermined list of values and only these values can be entered into that field (e.g. lithotype codes or responsible person). This ensures that there is consistency in the database (e.g. a quartz vein is always captured as “Qv” not as Q-V, Qtz V, etc). • Collar table: Incorrect coordinates (not within known range), unique hole IDs per dataset. Data can only be merged into the database if the drill hole has been entered into the collar table. • Survey table: Duplicate entries, survey intervals past the specified maximum depth in the collar table, overlapping intervals and anomalous dips and azimuths are not merged until corrected. • Geotechnical tables: Core recoveries and RQDs less than 0% or greater than 110% (Recovery) or 100% (RQD), overlapping intervals, negative widths, geotechnical results past the specified maximum depth in the collar table are not merged until corrected. • Geology table: Duplicate entries, lithological intervals past the specified maximum depth in the collar table, overlapping intervals and negative widths are not merged until corrected. Standardised logging codes are required. • Sampling table: Duplicate entries, sampling intervals past the specified maximum depth in the collar table, negative widths, overlapping intervals, sampling widths exceeding tolerance levels, missing intervals and duplicated sample IDs are not merged until corrected. • Assay table: Missing samples (assay results received, but no samples in database) are imported into an incoming assay table, assay metadata such as detection limits, methods, etc. are captured where possible. The author/Qualified Person notes that minor interval issues were observed and corrected, but none were deemed to materially affect the integrity of the dataset.

12.1.4 Core Inspection Sections of mineralised core from three holes were inspected – one from each deposit – PEDD010, BHDD044 and KODD085. The contacts between stratigraphic units was observed and cross-referenced against assay results and geological logging. It was observed that the core has deteriorated following cutting, being friable and incompetent. However, core photos and recovery data confirm that this is as a result of aging and cutting and is not a reflection of condition when drilled.

12.1.5 QAQC Review QAQC reports were produced for three project areas; namely Bigar Hill (BH), Korkan (KO) and Kraku Pester (KP) for the CSA Global 2017 NI 43-101 report. QAQC was reviewed globally (in other words, QC results were not broken down by laboratory, but instead by project area) for the elements of interest which were predominantly Au and Ag with some S and Cu. Results of the 2017/2018 Au and S QC samples were reviewed separately for each sample type and added to the 2017 NI 43-101 Report results below.

CSA Global Report Nº R302.2019 74 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Contamination is monitored with coarse blanks, assay accuracy by inserting standard samples with known concentrations of the relevant elements and precision by comparison of various duplicate sample analyses.

Cross Contamination Two blanks were used prior to the 2017/2018 drilling; a non-certified coarse blank (BLANK_BOR) and a certified pulp blank (GREY BLANK). No issues were observed with the results of the blank analysis for Au and Ag. Failures were noted in the Grey Blank results for S and it appears that some of these failures could be due to mislabelling/misidentification of standards and blanks. No Cu blank results were available. 2018 blank results did not indicate any signs of Au or S contamination. The author/Qualified Person notes that no sample contamination was observed in the assay of the Ag and Au samples, but some potential contamination was noted in the S samples.

Assay Accuracy Geostat’s CRM have been used throughout the drilling campaigns. DPM standards with a prefix of Mo were included with the primary samples prior to 2017 and CRM with the prefix TGP were included with the 2017/2018 drilling. Failures were observed in the pre-2017 results, which in some instances are probably due to misidentification of blanks and CRM. Minor failures were noted in the 2017/2018 CRM results (apart from TGP001 – S results which had a 100% failure rate). Figure 12-1 below is a plot of the results of gold CRM G308-8, showing two outliers which are probably mislabelled/misidentified QC samples. The first failure is probably meant to be G905-7 (expected value 3.92 ppm Au) and the second should probably be a blank.

Figure 12-1: Results of Geostats gold CRM G308-8 showing failures Source: CSA Global, 2018 Once the apparent misidentified samples are filtered out, there are no significant issues with the gold standard results (i.e. no significant or systematic bias and no failures noted). Failures and bias were noted in the pre-2017 Ag standard results. Table 12-1 below lists the results of the silver standard analyses per project area. Absolute biases ≥5% are highlighted in red with most standard

CSA Global Report Nº R302.2019 75 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

results under reporting relative to the expected value. Low grade standards, which include the Mo series, perform worse than the higher grade Geostats standards. Table 12-1: Ag standard results (absolute bias ≥5% in red) – pre-2017 samples BH KO KP Standard Exp. Samples Mean Mean Samples Mean Mean Samples Mean Mean code value CV CV CV (count) Ag bias (count) Ag bias (count) Ag bias GBM303-8 7.00 457 6.54 0.08 -7% 100 6.67 0.07 -5% GBM307-3 0.60 425 0.53 0.73 -11% 107 0.61 1.28 2% GBM309-4 42.30 466 41.56 0.06 -2% 109 42.92 0.07 1% GBM311-11 19.60 13 18.63 0.04 -5% 20 17.83 0.22 -9% GBM398-4 48.70 421 47.16 0.05 -3% 105 46.56 0.11 -4% GBM907-6 26.80 467 26.85 0.04 0% 101 26.47 0.05 -1% GBM909-11 25.50 18 24.56 0.06 -4% 13 23.60 0.10 -7% GBM909-13 127.30 13 125.38 0.05 -2% 16 132.38 0.05 4% GBM910-13 1.90 14 1.57 0.33 -17% 14 1.36 0.34 -28% Mo1 0.39 62 0.37 0.28 -4% 10 0.27 0.45 -31% Mo2 0.85 69 0.91 0.40 7% 7 0.76 0.75 -11% Mo3 2.34 80 2.21 0.18 -6% 7 2.36 0.34 1% Mo4 1.12 63 1.05 0.28 -7% 7 0.95 0.09 -15% Mo5 3.91 71 3.55 0.09 -9% 12 3.15 0.18 -20% GBMS304-3 1.50 9 1.10 0.54 -27% GBMS304-5 0.80 6 0.60 0.78 -25% Geostats S standards generally show acceptable accuracy, but the Mo series all fail. Expected values could be incorrect as failures are by orders of magnitude. The author/Qualified Person notes that once the apparent mislabelled or misidentified standards have been filtered out, Au standard assay results have acceptable accuracy, as do the higher-grade Ag samples and the Geostats S standards.

Precision Precision error can be estimated by measuring the precision error at each stage of the sampling and assay process. Field duplicates contain all sources of error (sampling error, sample reduction error and analytical error), Laboratory coarse duplicates contain sample reduction error and analytical error, pulp duplicates contain analytical error only. Table 12-2: Duplicate types Duplicate type Description Detail description FIELDDUP Field duplicate (sampling stage) Sampling stage duplicate LABDUP Laboratory duplicate (crushing stage) Crushing stage duplicate LABREP Laboratory repeat (instrument stage) Instrumental stage duplicate LABSPLIT Laboratory split (pulverising stage) Pulverising stage duplicate UMPIRE External laboratory check Blind pulp duplicate

The data were assessed using coefficients of variation (CV = std dev/average – also known as relative standard deviation) calculated from individual duplicate pairs and averaged using the RMS (root mean squared) approach. This approach is recommended by Stanley and Lawie (2007) and Abzalov (2008) as a way of defining a fundamental measure of data precision using duplicate paired data.

CSA Global Report Nº R302.2019 76 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Precision errors (CVAVR(%)) were calculated for duplicates with mean values ≥10 times the analytical detection limit and compared to acceptable limits. Acceptable and best practice limits are obtained from Abzalov’s 2008 paper, “Quality Control of Assay Data: A Review of Procedures for Measuring and Monitoring Precision and Accuracy”. Scatterplots, relative difference plots and Q-Q plots were produced. The author/Qualified Person reviewed the precision results for Au and S for the 2017/2018 drilling and results are discussed below: Table 12-3: Gold duplicate precision errors (with acceptable limits) – 2017/2018 samples Precision Bias CV(AVR(%)) CV(AVR(%)) Mean Au Mean Au Duplicate type best acceptable Pairs Count of pairs CV( ) AVR original duplicate Bias practice practice (total) (>10 x DL) % (ppm) (ppm) Field Dup 20 30 455 55 14 1.41 1.44 2% LabDup 10 20 34 7 5 0.95 0.97 1% LabSplit 10 20 184 25 4 1.53 1.55 2% LabRep 10 20 160 20 5 1.28 1.29 1%

Table 12-4: S duplicate precision errors – 2017/2018 samples Precision Bias Duplicate type Pairs Count of pairs CV( ) Mean S Mean S AVR Bias (total) (>10 x DL) % original (%) duplicate (%) Field Dup 455 55 14 1.41 1.44 2% LabRep 184 25 4 1.53 1.55 2% LabSplit 160 20 5 1.28 1.29 1%

Results of the duplicate pair comparisons are summarised below: • Au pairs have good precision (within best practice limits) and no significant bias (2% bias to field duplicates). • S pairs are precise with no significant bias. Results for the pre-2017 drilling are listed in the tables below. Table 12-5: Gold duplicate precision errors (with acceptable limits) – pre-2017 samples BH KO KP CV(AVR(%)) CV(AVR(%)) Duplicate Count of Count of Count of best acceptable Pairs CV( ) Pairs CV( ) Pairs CV( ) type pairs AVR pairs AVR pairs AVR practice practice (total) % (total) % (total) % (>10 x DL) (>10 x DL) (>10 x DL) Field Dup 20 30 3,477 805 33 2,366 580 30 848 223 39 LabDup 10 20 4,400 891 21 4,612 809 16 1,273 287 14 LabSplit 10 20 5,014 1,037 6 5,157 899 5 1,452 336 5 LabRep 10 20 8,290 1,990 6 8,814 1,617 5 2,370 526 5 Umpire 10 20 4,734 4,612 18 0 2,147 1,910 16

Results of the gold duplicate pair comparison are summarised below:

• Field duplicate pairs have CV(AVR)% from 30% (Korkan) to 39% (Kraku Pester) which exceeds the acceptable limit of 30% (Abzalov, 2008) for coarse to medium grained gold, but is within the 40% limit for nuggety gold indicating that pairs have acceptable repeatability. • External checks show no bias.

CSA Global Report Nº R302.2019 77 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

• Lab duplicates have an acceptable precision error. • Lab replicates and lab splits show excellent repeatability with average CV within best practice limits. Table 12-6: Silver duplicate precision errors) – pre-2017 samples BH KO KP

Duplicate type Pairs Count of pairs CV(AVR) Pairs Count of pairs CV(AVR) Pairs Count of pairs CV(AVR) (total) (>10 x DL) % (total) (>10 x DL) % (total) (>10 x DL) % Field Dup 1,323 5 10 0 103 0 LabDup 2,173 11 19 524 5 3 117 0 LabSplit 2,602 13 1 604 3 3 108 1 LabRep 4,934 40 2 1,325 6 19 218 1 Umpire 1,061 156 22 0 151 0

Results of the silver duplicate pair comparison are summarised below: • In most cases, the sample size is too small to make any definitive conclusions. • Precision and repeatability acceptable for lab split and lab replicates. • External checks have bias to duplicates (19% for Bigar Hill) and poorer repeatability.

Figure 12-2: Q-Q plot for silver external check assays (umpires) showing bias to duplicate sample) – pre-2017 samples Source: CSA Global

CSA Global Report Nº R302.2019 78 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Table 12-7: Sulphur duplicate precision errors) – pre-2017 samples BH KO KP Duplicate type Pairs Count of pairs CV(AVR) Pairs Count of pairs CV(AVR) Pairs Count of pairs CV(AVR) (total) (>10 x DL) % (total) (>10 x DL) % (total) (>10 x DL) % Field Dup 1,490 392 26 1,384 457 22 363 176 21 LabDup 2,471 546 11 3,562 956 9 708 392 12 LabSplit 2,940 571 2 3,999 1,088 2 792 428 4 LabRep 5,525 1,136 2 7,903 2,154 2 1,557 859 6 Umpire 1,060 559 18 0 950 720 18

Results of the sulphur duplicate pair comparison are summarised below: • Acceptable precision and no significant bias for duplicates. Table 12-8: Copper duplicate precision errors) – pre-2017 samples BH KO KP Duplicate type Pairs Count of pairs CV(AVR) Pairs Count of pairs CV(AVR) Pairs Count of pairs CV(AVR) (total) (>10 x DL) % (total) (>10 x DL) % (total) (>10 x DL) % Field Dup 0 0 103 95 15 LabDup 12 4 12 0 117 105 8 LabSplit 16 6 4 0 108 97 3 LabRep 31 13 10 0 217 194 3 Umpire 0 0 198 174 13

Results of the copper duplicate pair comparison are summarised below: • In the Bigar Hill and Korkan datasets, the sample size is too small to make any definitive conclusions. • Acceptable precision and no significant bias for Kraku Pester duplicates. • Kraku Pester external duplicates have a 4.5% bias to the duplicate samples (mean grade 80.2 ppm vs 83.8 ppm Cu). The author/Qualified Person notes that the precision errors measured for all the Au and S pairs indicate an acceptable repeatability with no significant bias noted. Ag pairs were mostly too low grade to accurately measure precision error, but the Bigar Hill external duplicates were biased towards the check laboratory. Only Kraku Pester had sufficient Cu pairs to analyse and these showed acceptable repeatability. However, there was a 4.5% bias to the external check laboratory in the external check comparison.

Conclusions and Recommendations The author/Qualified Person considers that the gold assay results provided should accurately represent the underlying samples and therefore can confidently be used in Mineral Resource estimation. Due to sporadic failures or biases, Ag, S and Cu assay results provide less confidence and a greater degree of caution is required if these values were to be used for MREs. However, only Au results are material to this report. The author/Qualified Person recommends ongoing vigilance to ensure that standards and blanks are correctly identified and labelled.

CSA Global Report Nº R302.2019 79 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

12.1.6 Inspection of Procedures DPM and Avala have used standard DPM procedures (Red, Blue, Green and Orange Books) since 2007, updated with some additional RC and trenching procedures in 2010 and 2011, respectively. These procedures are comprehensive and should ensure that best practices are maintained. Discussions were held with DPM’s Exploration Manager (Justin van der Toorn), Senior Geologist (Dragana Davidovic) and Regional Geologist (Mladen Zdravkovic) and it was apparent that these procedures have been applied to the exploration activities completed to date.

12.1.7 Twinned Hole Review Avala completed a twinned hole program to compare DD drilling with RC to ensure RC was representative for use in closer spaced infill drill programs. CSA Global reviewed comparison work on this dataset completed by AMEC, as well as completed additional review of twin holes. The CSA Global review focused on twin holes intersecting the modelled mineralisation solids. Of these, there are 30 holes twinned at Bigar Hill, 20 holes twinned at Korkan, and seven holes twinned at Kraku Pester. None were twinned at Korkan West. The author/Qualified Person notes that broadly, the twinned hole grade comparisons show the same mineralisation trends. While there can be significant local variability, particularly in higher-grade portions at Bigar Hill, this is more likely to be attributed to short-scale grade variability than any bias identified in drilling or sampling methods. The author/Qualified Person concludes that assays derived from RC and DD drilling can be combined for use in the grade estimation.

12.1.8 Laboratory Audit An audit of the onsite, Société Générale de Surveillance (SGS) managed preparation and analytical laboratory was undertaken by the author/Qualified Person on 1 March 2017. The laboratory is managed by SGS and processes DPM samples as well as samples from other clients. In 2016, the breakdown was approximately 70% DPM, 30% other (pers. comm., George Daher, SGS laboratory manager). The laboratory is not independently accredited but operates under the SGS company accreditation and uses SGS accredited methods. SGS Bor takes part in the six monthly Geostats round robin as well as in the monthly SGS round robin. The sample preparation, fire assay and sulphur analytical lab sections were not in use when CSA Global Qualified Persons visited the laboratory, but samples were being analysed for a multi-element suite using the ICP-MS machine. The laboratory was well laid out and clean with no indication of contamination. No significant issues were noted.

12.2 Conclusions Subject to the limitations listed below and based on the outcomes of the above data verification undertaken, as well as discussions with DPM geologists and the SGS laboratory manager; The author/Qualified Person considers the drill-hole database for the Bigar Hill, Korkan and Kraku Pester projects to be sufficiently reliable for Mineral Resource estimation and associated downstream work.

CSA Global Report Nº R302.2019 80 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

13 Mineral Processing and Metallurgical Testing

13.1 Metallurgical Testing Summary (2017 to 2019) Metallurgical testing has focused on supporting the development of the Timok Gold Project as a heap leach operation for processing weathered and transitional mineralised material types from the various deposits. Testing to date has focused on gold recovery at coarse particle sizes. Metallurgical testing was initiated in 2017 using samples from existing exploration diamond drill-holes. Coarse sample bottle roll and column leach tests were conducted in 2018 at SGS Lakefield on composite samples representing the oxide and transitional mineralisation types from the Korkan deposit, and oxide zones from Bigar Hill and Korkan West deposits. Results from coarse sample bottle roll and column leach tests were mainly used to estimate metallurgical recoveries for use in the 2018 updated MRE for the oxide and transitional mineralisation zones. Results of the coarse sample bottle roll leach tests indicated gold leach extractions ranging from 53% for the Korkan transitional mineralised material to 94% for the Bigar Hill oxide mineralised material, after 14 days of leaching, and at a crush size of 100% passing 16 mm. Leach curves indicated that gold leaching was still ongoing after 14 days of leaching when the tests were terminated. Column leach tests carried out at the optimal crush size of 80% - 12.5 mm exhibited fast leach kinetics except for the Korkan transitional mineralised material, where leaching was still ongoing at 63 days when the tests were terminated. Lime consumption is moderate and cyanide consumption is low for all mineralised material types. The projected gold recovery, reagent consumption, leach time and crush size based on the column leach testwork results are summarised in Table 13-1. Table 13-1: Column leach testwork results summary (2018) Calculated Extracted Leach Reagent consumption (kg/t) Sample Crush size Leach Sample description head grade recovery ID (P mm) (days) 80 (Au g/t) (Au g/t) (Au %) Cyanide Lime KO_01 Korkan oxide 1.54 1.46 94.8 0.21 0.88 KO_02 Korkan transitional 1.96 1.34 67.9 0.36 0.90 -12.5 63 BH_01 Bigar Hill oxide 2.01 1.90 94.2 0.36 1.21 KW_01 Korkan West oxide 1.14 0.87 75.5 0.30 0.99 Note: Gold leach recoveries have not been downgraded to consider losses resulting from short-circuiting and on the side of the heaps or adjusted to consider Carbon-In-Column and slag metal losses. Size-by-size analysis of the column leach test feed and tails samples shows gold evenly distributed among the size classes, roughly following the mass splits. Some of the metallurgical samples showed low gold recovery in the coarse size fractions; +19.0 mm. There was generally good correlation between gold extraction obtained from the coarse sample bottle roll leach and column leach tests, apart from the Korkan transitional mineralised material which was still leaching in both tests. Following on from the 2018 testwork program, additional oxide and transitional mineralisation samples were selected for coarse sample bottle roll, and column leach testing in 2019. Based on the findings from the 2018

CSA Global Report Nº R302.2019 81 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

testwork program, the leach cycle time for the coarse bottle roll leach tests was increased to 30 days, whilst the columns were leached to completion (i.e. terminated when the gold extraction plateaued). The projected gold recovery, reagent consumption, leach time and crush size based on the column leach testwork results from the 2019 testwork program are summarised in Table 13-1. Table 13-2: Column leach testwork results summary (2019) Calculated Extracted Leach Reagent consumption (kg/t) Sample Crush size Leach Sample description head grade recovery ID (P80 mm) (days) (Au g/t) (Au g/t) (Au %) Cyanide Lime BH_P1_01 Bigar Hill transitional -11.3 91 1.36 1.46 60.3 0.27 0.22 KW_P1_01 Korkan West oxide -12.3 1.15 1.90 89.3 0.05 0.20 65 KO_P1_01 Korkan oxide -12.3 0.77 0.87 82.1 0.08 0.23

Results of the testing program indicate that oxide and transitional mineralisation samples from the Timok Gold Project are amenable to heap leach processing. Leach rates are relatively fast with high gold recovery for the oxide mineralised zone, whereas leach rates are slower and gold recovery is lower for the transitional mineralised zone.

13.2 Sample Selection and Representivity (2017 to 2019) A series of six metallurgical twin-holes were completed during December 2017 to collect sulphide material for initial scoping testwork carried out in 2018. Further sample intervals were taken from seven drill holes drilled in 2018 to prepare additional metallurgical composites for testing in 2019. The sample composites targeted mineralisation within the S1 stratigraphic horizon, which is the dominant host of mineralisation at Timok. Samples were selected based upon logged weathering style, visual estimates of the percentage of oxidation and review of the sulphur assay data. All sample composites are located within the conceptual pit shells used to constrain the 2018 MREs. Details of the metallurgical drill-holes from which the sample intervals were selected to prepare the master composites representing the various oxide and transitional mineralisation zones are shown in Table 13-3. Table 13-3: Metallurgical drill-hole summary Hole Composite Drill-hole ID Phase Prospect Easting Northing Elevation Azimuth Dip length (m) mineralisation type BHDDMET01 570478 4898645 675 280 -55 80 Oxide Composite Bigar Hill BHDDMET02 570622 4898611 688 275 -64 175 Sulphide Composite KODDMET01 570227 4900437 610 45 -45 70 MRE Oxide/Trans KODDMET02 Korkan 570266 4900472 622 45 -70 60 Composite KODDMET03 570266 4900473 622 40 -50 65 KWDDMET01 Korkan West 569839 4899342 640 190 -60 70 Oxide Composite BHDDMET003 570146 4898529 672 270 -65 70 Oxide + Transitional BHDDMET004 Bigar Hill 570302 4898239 718 270 -55 240 Sulphide BHDDMET005 570143 4898689 640 255 -65 60 Transitional KWDDMET002 Scoping 569802 4899367 653 245 -60 90 Korkan West Oxide KODDMET009 570222 4899140 685 220 -50 60 KODDMET004 570309 4900620 648 55 -55 100 Transitional Korkan KODDMET010 570435 4900442 673 40 -55 150 Sulphide Total metrage 1,290

CSA Global Report Nº R302.2019 82 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

In total, four composites were collected and tested as part of the 2018 testwork program: • Met18_KO_01 – Korkan oxide: o Oxidised, sedimentary breccia-conglomerate with quartz and limestone pebble fragments within a sandy matrix. Taken from the S1 unit from within the Korkan deposit. • Met18_KO_02 – Korkan transitional: o Transitional S1 unit material from the Korkan deposit, comprised of alternating zones oxide and sulphide mineralisation, of equal proportions. The rock type is comprised of a sedimentary breccia- conglomerate with quartz and limestone pebble fragments within a sandy, or to a lesser extent mudstone matrix. • Met18_BH_01 – Bigar Hill oxide: o Oxidised S1/S2 horizon material from the Bigar Hill deposit. Coarse to medium grained sandstone with interbedded mudstone laminas within S1 fraction of the sample. • Met18_KW_01 – Korkan West oxide: o Oxidised S1 calcareous, medium/fine grained sandstone from the Korkan West deposit. A sulphide sample from BH was also collected, but no testwork was conducted on this sample. In total, eight composites were collected and tested as part of the 2019 testwork program: • BH_P1_01 – Bigar Hill transitional: o S1 calcareous, medium/fine grained sandstone and brecciated sandstone from the Bigar Hill; frequent alternation of relatively short, strongly oxidised and non-oxidised intervals, and also intervals with oxidation along fractures. • BH_P1_02 – Bigar Hill oxide: o Strongly to partially oxidised contact zone of skarn altered S2 and S1 sandstone from the Bigar Hill. • BH_P1_03 – Bigar Hill transitional: o Partially oxidised, skarn altered medium to fine-grained sandstone from the S1 unit from the Bigar Hill; oxidation dominantly developed along fracture systems. • BH_P1_04 – Bigar Hill sulphide: o Non-oxidised, sulphidised clastic rocks from the Bigar Hill – mudstone and basal breccia composed dominantly of limestone fragments and subordinately of quartz pebbles in siltstone to fine-grained sandstone matrix. • KW_P1_01 – Korkan West transitional: o Dominantly oxidised fine-grained sandstone, brecciated sandstone and breccia-conglomerate from the Korkan West with scattered intervals of sulphidised material; oxidation is developed along fracture systems, and also as a part of breccia matrix. • KW_P1_02 – Korkan West oxide: o Oxidised contact zone of S1 and S2 sandstones from the Korkan West; oxidation developed in variable intensity, from pervasive, in the entire rock mass, to partial, along the fracture zones. • KO_P1_01 – Korkan transitional: o Small interval of breccia-conglomerate with limestone fragments and quartz pebbles with sandy matrix from the S1 unit (upper part of the sample) and dominantly oxidised, massive, fractured to brecciated limestone from the Korkan; oxidation is along fracture system and within breccia matrix; sample also include small interval of non-oxidised rock.

CSA Global Report Nº R302.2019 83 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

• KO_P1_02 – Korkan sulphide: o Non-oxidised, sulphidised, fractured conglomerate and breccia-conglomerate composed of limestone fragments and quartz pebbles in sandy to silty matrix. Plan and section views showing the location of the metallurgical drill-holes in each of the respective deposits are shown in Figure 13-2 to Figure 13-12.

Figure 13-1: Location of drill holes sampled for metallurgical bulk composite samples – all samples fall within the conceptual pit shells (light orange) used to constrain Mineral Resources Source: Avala, 2018

CSA Global Report Nº R302.2019 84 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Figure 13-2: Plan view of the location of the Korkan oxide, transitional and sulphide samples with drill holes and mineralisation outlines Source Avala 2018

CSA Global Report Nº R302.2019 85 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Figure 13-3: Section view of drill holes for Korkan oxide and transitional mineralisation samples (MET_KO_01/02) Source Avala 2018

Figure 13-4: Section view of drill hole for Korkan transitional mineralisation sample (KO_P1_01) Source Avala 2018

CSA Global Report Nº R302.2019 86 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Figure 13-5: Section view of drill holes for Korkan sulphide mineralisation sample (KO_P1_02) Source Avala 2018

CSA Global Report Nº R302.2019 87 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Figure 13-6: Plan view of the location of the Bigar Hill oxide, transitional and sulphide samples with drill holes and mineralisation outlines Source Avala 2018

CSA Global Report Nº R302.2019 88 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Figure 13-7: Section view of drill hole for Bigar Hill oxide mineralisation sample (Met18_BH_01) Source Avala 2018

Figure 13-8: Section view of drill hole for Bigar Hill transitional mineralisation sample (BH_P1_01) Source Avala 2018

CSA Global Report Nº R302.2019 89 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Figure 13-9: Section view of drill hole for Bigar Hill oxide and transitional mineralisation samples (BH_P1_02/BH_P1_03) Source Avala 2018

Figure 13-10: Section view of drill hole for Bigar Hill sulphide mineralisation sample (BH_P1_04) Source Avala 2018

CSA Global Report Nº R302.2019 90 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Figure 13-11: Plan view of the location of the Korkan West oxide and transitional mineralisation samples with drill hole and mineralisation outlines Source Avala 2018

Figure 13-12: Section view of drill hole for Korkan West oxide mineralisation sample (MET18_KW_01) Source Avala 2018

CSA Global Report Nº R302.2019 91 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Figure 13-13: Section view of drill hole for Korkan West oxide mineralisation sample (KW_P1_01) Source Avala 2018

Figure 13-14: Section view of drill hole for Korkan West oxide mineralisation sample (KW_P1_02) Source Avala 2018

13.3 SGS Testwork Program (2018) SGS’s testwork program in 2018 was conducted on composites selected from PQ drill core intervals taken in 2017. The testwork program commenced in February 2018. Sample intervals were selected to prepare master composites representing:

CSA Global Report Nº R302.2019 92 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

• Korkan oxide: Met18_KO_01. • Korkan transitional: Met18_KO_02. • Bigar Hill oxide: Met18_BH_01. • Korkan West: Met18_KW_01. The scoping testwork program consisted of: • Head assays. • Coarse sample bottle roll leach tests. • Percolation tests. • Column leach tests. • Size-by-size analyses; column heads and tails. The testwork program primarily considered the application and suitability of heap leach technology using coarse sample bottle roll and column leach tests. Coarse sample bottle roll leach tests were conducted on the individual composites to determine the optimum crush size. A total of three bottle roll leach tests were conducted on each of the individual composites at crush sizes of 100% -50 mm (1”), -16 mm (5/8”) and -6.4 mm (1/4”). A single column leach tests was carried out on each of the master composites at the optimum crush size derived from the coarse bottle roll leach tests; 80% -12.5 mm.

13.3.1 Head Assays Detailed head assay was performed on each of the composite samples to determine the level of a range of elements of interest. The analyses were performed on a representative subsample of the -2.0 mm material from each sample which had been pulverised to 100% passing 75 µm. Results are given in Table 13-4. Table 13-4: Head assay results (2018) -150 mesh Calculated Total +150 mesh % Au distribution Composite ID head grade weight Mass Au (Au g/t) (g) Mass (%) Mass (g) Au (g/t) (%) a (g/t) b (g/t) +150 (#) -150 (#) MET18_KO_01 1.44 963.1 3.14 30.3 0.22 96.9 1.45 1.50 0.5 99.5 MET18_KO_02 1.73 997.2 3.05 30.5 0.32 96.9 1.79 1.76 0.6 99.4 MET18_KW_01 1.04 998.5 2.94 29.4 0.34 97.1 1.09 1.04 1.0 99.0 MET18_BH_01 1.86 999.4 2.58 25.8 0.49 97.4 1.89 1.90 0.7 99.3

The results showed gold assays, based on screen fire assay, to range from 1.04 ppm Au in the BH_01 sample to 1.73 ppm Au in the KO_01 sample. A more detailed analysis was also carried out on the different master composites. Results are shown in Table 13-5. Table 13-5: Detailed head assay results (2018) Timok deposit composites Element Units MET18_KO_01 MET18_KO_02 MET18_KW_01 MET18_BH_01 Au g/t 1.44 1.73 1.04 1.86 As % 0.008 0.010 0.007 0.016

CSA Global Report Nº R302.2019 93 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Timok deposit composites Element Units MET18_KO_01 MET18_KO_02 MET18_KW_01 MET18_BH_01 Hg g/t 0.9 1.2 1.1 2.8

ST % 0.02 0.48 0.01 0.04 S= % <0.01 0.41 <0.01 <0.01 S° % <0.05 <0.05 <0.05 <0.05

SO4 % <0.1 <0.1 <0.1 <0.1

CT % 5.08 5.73 6.58 4.05

Cg % <0.05 <0.05 <0.05 <0.05 TOC % <0.05 0.14 0.06 0.07

CO3 % 31.3 26.5 41.0 23.3 Ag g/t 2 <2 <2 3 Al g/t 13,100 18,900 7,480 16,600 Ba g/t 43.3 356 58.0 84.2 Be g/t 0.28 0.42 0.16 0.28 Bi g/t <20 <20 <20 <20 Ca g/t 202,000 173,000 268,000 152,000 Cd g/t <2 <2 <2 <2 Co g/t <4 <4 <4 <4 Cr g/t 23 132 10 35 Cu g/t 6.8 6.0 12.6 11.9 Fe g/t 5,600 8,490 5,510 9,910 K g/t 3,680 5,660 1,960 4,680 Li g/t <5 <5 <5 <5 Mg g/t 1,430 1,910 1,930 1,580 Mn g/t 225 202 922 389 Mo g/t <5 <5 <5 <5 Na g/t 158 347 122 245 Ni g/t <20 <20 <20 <20 P g/t <300 <300 <300 375 Pb g/t <60 <60 <60 <60 Sb g/t <30 <30 <30 <30 Se g/t <40 <40 <40 <40 Sn g/t <30 <30 <30 <30 Sr g/t 105 173.0 167 110 Ti g/t 612 989 509 1,070 Tl g/t <30 <30 <30 <30 U g/t <20 <20 <20 <20 V g/t 16 24 19 32 Y g/t 5.4 6.0 6.3 7.6 Zn g/t 7 14 27 27

Silver levels ranged between 2 g/t and 3 g/t Ag. Total sulphur levels within the samples were generally low, averaging 0.02% in the three oxide samples and higher in the Korkan transitional mineralisation sample at 0.48% TS.

CSA Global Report Nº R302.2019 94 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

The proportion of sulphide sulphur in the KO_02 sample was 85.4% of total sulphur assay.

13.3.2 Coarse Sample Bottle Roll Leach Tests Coarse sample bottle roll testing was conducted to identify the maximum gold recovery achievable from each of the samples at crush sizes typical of conventional heap leach operations. A series of tests were performed to investigate the effect of crush size on leach performance at a fixed cyanide concentration of 0.5 g/L.

Coarse sample bottle roll leach tests were carried out for a leach duration of 14 days; various crush sizes (P100 -50 mm, -16 mm and -6.3 mm) were tested. A summary of the gold recovery achieved during the 2018 coarse sample bottle roll test program are given in Table 13-6.

CSA Global Report Nº R302.2019 95 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Table 13-6: Summary of coarse sample bottle roll test results (2018)

Mineralised Reagent Head (Au g/t) Composite ID/ Reagent addition %Au extraction (CN) CN residue material consumption Deposit Test no. (kg/t of CN feed) (hours/days) grade CN crush size (kg/t of CN feed) CN Oxidation (Au g/t) calc. direct (inches) NaCN CaO NaCN CaO 4 h 1 d 2 d 5 d 7 d 9 d 12 d 14 d MET18_KO-01 -1/4" COBR-1 0.74 0.79 0.51 0.76 76.7 86.6 88.9 85.6 94.2 90.0 89.5 95.6 0.07 1.60 Korkan -5/8" COBR-2 1.39 0.78 0.93 0.70 65.9 81.2 85.3 84.2 88.7 90.6 83.1 93.2 0.11 1.62 1.44 Oxide -1" COBR-3 1.38 0.80 0.90 0.70 53.7 75.7 78.0 79.1 91.2 86.8 85.8 93.4 0.10 1.52 MET18_BH-01 -1/4" COBR-4 0.77 1.18 0.27 1.14 54.6 79.7 79.7 80.9 91.3 88.9 88.2 93.6 0.11 1.65 Bigar Hill -5/8" COBR-5 1.31 1.08 0.92 0.98 46.8 78.3 87.1 81.0 95.4 91.2 88.5 93.7 0.11 1.75 1.86 Oxide -1" COBR-6 1.60 1.01 0.95 0.92 32.2 66.3 82.0 82.7 83.9 89.6 86.5 93.9 0.11 1.72 MET18_KW-01 -1/4" COBR-7 0.77 0.76 0.46 0.71 41.2 61.6 68.4 69.6 76.3 73.1 72.5 75.9 0.26 1.08 Korkan West -5/8" COBR-8 1.17 0.76 0.68 0.62 30.2 56.3 68.2 69.3 73.4 76.7 77.7 75.5 0.26 1.06 1.04 Oxide -1" COBR-9 1.14 0.77 0.71 0.67 24.0 51.0 65.0 66.7 69.5 68.0 69.9 74.1 0.28 1.08 MET18_KO-02 -1/4" COBR-10 0.85 0.76 0.32 0.71 37.7 46.0 51.5 42.7 51.0 50.3 51.1 55.1 0.80 1.77 Korkan -5/8" COBR-11 1.23 0.81 0.86 0.77 30.5 42.3 47.9 44.5 49.8 51.3 50.4 53.1 0.71 1.51 1.73 Transitional -1" COBR-12 1.34 0.87 0.98 0.83 11.6 43.7 48.6 45.5 46.8 50.9 53.7 54.3 0.73 1.60

CSA Global Report Nº R302.2019 96 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Results did show a slight improvement in gold extraction between a crush size of 100% -50 mm, and that achieved at 100% -16 mm. There was no real increase in gold extraction arising from crushing to the finer crush size of 100% -6.3 mm. Gold extraction from testing conducted on the different master composites at the optimum crush size of 100% -16 mm; after 14 days of leaching were: • Korkan – oxide mineralised material: 93.2%. • Korkan – transitional mineralised material: 53.1%. • Bigar Hill – oxide mineralised material: 93.7%. • Korkan West – oxide mineralised material: 75.5%. The effect of crush size is graphically represented in Figure 13-15 to Figure 13-18.

Figure 13-15: Coarse bottle roll test leach curves – Korkan oxide mineralisation sample Source: Avala, 2018

Figure 13-16: Coarse bottle roll test leach curves – Korkan transitional mineralisation sample Source: Avala, 2018

CSA Global Report Nº R302.2019 97 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Figure 13-17: Coarse bottle roll test leach curves – Korkan West oxide mineralisation sample Source: Avala, 2018

Figure 13-18: Coarse bottle roll test leach curves – Bigar Hill oxide mineralisation sample Source: Avala, 2018 In most cases, the leach curves indicate that leaching was still ongoing after 14 days.

13.3.3 Percolation Testing Percolation testing was undertaken to determine the drainage characteristics of each sample and to identify whether it was necessary to agglomerate with cement prior to subsequent column leach testing. Each sample was subjected to a total of three percolation tests; at the three different crush sizes. A 500 g charge of the mineralised material was tested. Each charge was placed in a plastic, 2-inch diameter tube fitted with a filter cloth at the bottom. After the sample was loaded into the column, the column was lightly tapped to settle the material. The charge height in the column was measured. A piece of filter cloth was placed on the top of the material in the column to allow for better solution dispersion. The column apparatus was then placed over top of a four-litre plastic jar containing two litres of water. Using a pump, the water flow was recycled through the column. Initially at a rate of approximately 20 L/m2/hr.

CSA Global Report Nº R302.2019 98 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Dependant on the progress of the test, the water flowrate may be increased. Observations such as water clarity, water flow and mineralised material height in the column were recorded. During the percolation testing some of the column began to flood at the high solution application rate. However, based on testing it was deemed unnecessary to agglomerate the columns.

13.3.4 Column Leach Testing Column leach testing was undertaken to provide confirmation of the achievable metal recoveries and leach rates from each of the samples under heap leaching conditions. A total of four tests were conducted; one on each of the master composite samples at the optimal crush size of 80% -12.5 mm. Samples were leached for a total of 63 days, using a 0.5 g/L cyanide solution at a target solution application rate of 10 L/m2/hr. The pregnant leach solution was passed through activated carbon to adsorb the gold. The activated carbon was changed on days 1, 4 and 7, and weekly thereafter. The results of the column leach tests are summarised in Table 13-7. The results show good correlation between the gold extraction based on back calculated head, and that based on carbon assays and solids leach residue. A comparison of column leach vs coarse bottle roll leach test results is summarised in Table 13-8. Results show a good correlation between column and coarse bottle gold extractions; except for the Korkan transitional mineralisation sample. As discussed in Section 13.3.2, the coarse bottle roll leach tests were still leaching when the test was terminated after 14 days of leaching.

CSA Global Report Nº R302.2019 99 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Table 13-7: Summary of column leach test results

Calculated Extracted Tails Column leach recovery (based on) Crush size Agglom. Head assay Sample ID head grade grade Measured head Calculated head Carbon/Residue (80% mm) stage (Au g/t) (Au g/t) (Au g/t) (Au g/t) % Au Korkan oxide -12.5 No 1.48 1.54 1.46 0.08 98.6 94.8 94.4 Korkan transitional -12.5 No 1.72 1.96 1.34 0.62 77.9 68.4 67.9 Bigar Hill oxide -12.5 No 1.87 2.01 1.90 0.11 101.6 94.5 94.2 Korkan West oxide -12.5 No 1.11 1.14 0.87 0.27 78.4 76.3 75.5

Table 13-8: Comparison of column leach test vs coarse bottle roll leach test results

Leach Column leach recovery (based on) Leach Coarse bottle roll Mineralised Crush size Crush size Sample ID time Measured head Calculated head Loaded carbon/Residue time leach recovery material type (80% mm) (100% mm) (days) % Au (days) (Au %) Oxide -12.5 98.6 94.8 94.4 -16 93.2 Korkan Transitional -12.5 77.9 76.3 67.9 -16 53.1 63 14 Bigar Hill Oxide -12.5 101.6 71.6 94.2 -16 93.7 Korkan West Oxide -12.5 78.4 70.5 75.5 -16 75.5

CSA Global Report Nº R302.2019 100 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

The gold leach recoveries, based on column calculated head grades, as a function of time are shown in Table 13-9. Table 13-9: Gold leach recoveries based on column calculated head grades

Leach C-1 C-2 C-3 C-4 days KO_01 KO_02 BH-01 KW_01 1 44.2 8.0 25.1 4 87.7 35.0 0.0 63.7 7 89.2 41.7 1.4 67.6 14 92.1 50.7 41.1 70.5 21 93.4 56.0 75.9 72.2 28 93.9 59.9 88.1 74.4 35 94.1 62.5 91.5 75.0 42 94.2 63.7 91.9 75.2 49 94.3 64.9 93.3 75.3 56 94.3 65.9 93.8 75.4 63 94.4 67.9 94.2 75.5

The column leach curves for the four master composites are shown in Figure 13-19.

100 y = 9.467ln(x) + 61.062 90 y = 40.11ln(x) - 60.034 y = 10.087ln(x) + 39.103 80

70 y = 13.828ln(x) + 12.531

60

50

40 Au Extraction (%) Extraction Au 30

20 C-1 KO_01 C-2 KO_02 C-3 BH_01 10 C-4 KW_01 Log. (C-1 KO_01) Log. (C-2 KO_02) Log. (C-3 BH_01) Log. (C-4 KW_01) 0 0 10 20 30 40 50 60 70 Time (Days)

Figure 13-19: Column leach curves Source: Avala, 2018 The test results showed gold recoveries to carbon ranged from 67.9% (KO-02), to 94.4% (KO-01). Analysis of the leach kinetics during the tests showed the gold to be fast leaching with between 92.1% (KO-01) and 70.5% (KW-01) of the total gold recovered within the first 14 days of leaching. Gold leach kinetics for the transitional mineralised zone were slower, and leaching was still ongoing after 63 days of leaching.

CSA Global Report Nº R302.2019 101 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

During testing, consumption of lime was moderate, ranging from 0.88 kg/t to 1.21 kg/t whilst the consumption of cyanide was low, ranging from 0.21 kg/t to 0.36 kg/t. Figure 13-19 shows an initial lag in the gold extraction for the Bigar Hill oxide sample, due to the column being plugged. The column was drained, cleared, and solution irrigation restarted. This appears to have had no detrimental effect on the final gold extraction achieved for the Bigar Hill oxide sample, with metallurgical performance after 63 days of leaching being similar to that achieved for the Korkan oxide sample.

13.3.5 Size-by-Size Analysis Subsamples of both the column leach feed and leach residues from the four samples were submitted for size- by-size analysis for gold to determine the distribution of metal within each sample and to allow metal recoveries by size to be calculated. The gold distribution in the leach residue as a function of size is shown in Table 13-10, and graphically represented in Figure 13-20. Table 13-10: Gold residue assay/feed assay by size fraction Sample ID C-1 C-2 C-3 C-4 Size fraction (µm) KO_01 KO_02 BH-01 KW_01 12,700 10% 37% 11% 30% 9,525 7% 39% 8% 30% 4,750 6% 29% 7% 25% 1,700 6% 35% 6% 22% 600 5% 46% 5% 22% 150 6% 37% 4% 21% 75 6% 39% 3% 20% 38 6% 40% 3% 21% -38 2% 21% 3% 13%

Figure 13-20: Residue assay/feed assay by size fraction Source: Avala, 2018 Size-by-size recoveries for the four metallurgical composites are shown in Figure 13-21 to Figure 13-24.

CSA Global Report Nº R302.2019 102 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

The size-by-size recovery curves show a decrease in gold extraction in the coarse size fractions (+19.0 mm). This suggests that a finer crush size of 100% -12.5 mm could result in higher gold leach extractions.

Figure 13-21: Size-by-size recovery – Korkan oxide Source: Avala, 2018

Figure 13-22: Size-by-size recovery – Korkan transitional Source: Avala, 2018

CSA Global Report Nº R302.2019 103 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Figure 13-23: Size-by-size recovery – Bigar Hill oxide Source: Avala, 2018

Figure 13-24: Size-by-size recovery – Korkan West oxide Source: Avala, 2018

13.4 SGS Testwork Program (2019) The SGS testwork program in 2019 was conducted on composites selected from PQ drill core intervals taken in 2018. The testwork program commenced in January 2019. Sample intervals were selected to prepare master composites representing: • Bigar Hill: o Transitional: BH_P1_01. o Oxide: BH_P1_02. o Transitional: BH_P1_03. o Sulphide: BH_P1_04.

CSA Global Report Nº R302.2019 104 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

• Korkan West: o Transitional: KW_P1_01. o Oxide: KW_P1_02. • Korkan: o Transitional: KO_P1_01. o Sulphide: KO_P1_02. The scoping testwork program consisted of: • Head assays. • Coarse sample bottle roll leach tests. • Column leach tests. • Size-by-size analyses; column heads and tails. The 2019 testwork program followed on from the 2018 testwork program and primarily considered the application and suitability of heap leach technology using coarse sample bottle roll and column leach tests on selected oxide and transitional mineralisation samples. Coarse sample bottle roll leach tests were conducted on the individual composites to determine the effect of crush size on gold leach extraction. Crush sizes of 100% -100 mm (2”), -75 mm (1.5”), -50 mm (1”), -16 mm (5/8”) and -6.4 mm (1/4”) were tested. Single column leach tests were carried out on selected master composites at the optimum crush size derived from the 2018 coarse bottle roll leach tests; 80% -12.5 mm.

13.4.1 Head Assays Detailed head assay was performed on each of the composite samples to determine the level of a range of elements of interest. The analyses were performed on a representative subsample of the -2.0 mm material from each sample which had been pulverised to 100% passing 75 µm. Results are given in Table 13-11. Table 13-11: Head assay results (2019) -150 mesh Calculated Total +150 mesh % Au distribution Composite ID head grade weight Mass Au (Au g/t) (g) Mass (%) Mass (g) Au (g/t) (%) a (g/t) b (g/t) +150 (#) -150 (#) BH_P1_01 1.47 993.5 1.89 18.8 0.44 98.1 1.48 1.50 0.6 99.4 BH_P1_02 0.47 999.5 1.79 17.9 0.26 98.2 0.48 0.47 1.0 99.0 BH_P1_03 0.98 1,005.6 2.36 23.7 0.68 97.6 0.98 1.00 1.6 98.4 BH_P1_04 5.32 1,034.1 2.95 30.5 6.65 97.0 5.31 5.25 3.7 96.3 KW_P1_01 1.15 1,030.2 2.26 23.3 0.52 97.7 1.18 1.15 1.0 99.0 KW_P1_02 0.89 1,022.6 2.35 24.1 0.27 97.6 0.92 0.90 0.7 99.3 KO_P1_01 0.77 1,041.1 1.99 20.7 0.46 98.0 0.78 0.78 1.2 98.8 KO_P1_02 3.43 1,043.4 2.67 27.8 2.41 97.3 3.46 3.46 1.9 98.1

The results showed gold assays, based on screen fire assay, to range from 0.47 ppm Au in the BH_P1_02 sample to 5.32 ppm Au in the BH_P1_04 sample. A more detailed analysis was also carried out on the different master composites. Results are shown in Table 13-12.

CSA Global Report Nº R302.2019 105 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Table 13-12: Detailed head assay results (2019) Timok deposit composites Element Units BH_P1_01 BH_P1_02 BH_P1_03 BH_P1_04 KW_P1_01 KW_P1_02 KO_P1_01 KO_P1_02 Au g/t 1.47 0.47 0.98 5.32 1.15 0.89 0.77 3.43 As % 0.014 0.038 0.020 0.018 0.014 0.006 0.006 0.110 Hg g/t 3.4 4.1 6.1 2.9 2.4 0.40 2.0 10.5

ST % 0.45 0.13 0.40 2.34 0.10 0.04 0.04 1.77 S= % 0.39 0.11 0.32 2.10 0.09 < 0.05 < 0.05 1.50 S° %

SO4 %

CT % 6.97 5.58 9.87 5.40 7.50 5.50 11.1 5.72

Cg % TOC %

CO3 % 33.0 26.1 51.8 27.2 38.5 25.7 56.7 28.3 Ag g/t <3 <3 <3 <3 <3 <3 <3 <3 Al g/t 17,600 19,100 14,500 46,600 10,100 15,200 5,140 36,200 Ba g/t 152 248 43.8 238 48.9 45 39.3 254 Be g/t <2 <2 <2 <2 <2 <2 <2 <2 Bi g/t <20 <20 <20 <20 <20 <20 <20 <20 Ca g/t 226,000 198,000 335,000 178,000 278,000 175,000 403,000 169,000 Cd g/t <5 <5 <5 <5 <5 <5 <5 <5 Co g/t <10 <10 <10 <10 <10 <10 <10 <10 Cr g/t 27 22 15 62 16 5 10 51 Cu g/t 10.4 27.7 83.6 28.1 8.9 16.7 7.1 15.2 Fe g/t 12,900 22,200 10,800 24,500 7,800 16,500 3,790 19,000 K g/t 4,560 3,660 667 14,200 3,480 2,940 1,830 13,700 Li g/t <90 <90 <90 <90 <90 <90 <90 <90 Mg g/t 6,940 12,900 22,400 8,980 2,160 1,530 4,360 17,200 Mn g/t 338 637 713 1090 237 407 316 454 Mo g/t <20 <20 <20 <20 <20 <20 <20 <20 Na g/t 247 297 103 778 233 239 205 897 Ni g/t <20 <20 61 37 <20 <20 <20 27 P g/t <700 <700 <700 <700 <700 <700 <700 <700 Pb g/t <100 <100 <100 <100 <100 <100 <100 <100 Sb g/t <70 <70 <70 <70 <70 <70 <70 <70 Se g/t <50 <50 <50 <50 <50 <50 <50 <50 Sn g/t <200 <200 <200 <200 <200 <200 <200 <200 Sr g/t 99.5 185 144 85.4 200 96.6 159 162 Ti g/t 1030 1340 858 2870 579 832 253 2080 Tl g/t <70 <70 <70 <70 <70 <70 <70 <70 U g/t <20 <20 <20 <20 <20 <20 <20 <20 V g/t 31 49 28 70 18 28 13 59 Y g/t 7.60 11.3 9.91 8.94 5.27 9.51 3.95 11.0 Zn g/t <90 <90 <90 <90 <90 <90 <90 <90

CSA Global Report Nº R302.2019 106 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Silver levels ranged between 2 g/t Ag and 3 g/t Ag. Average total sulphur levels for the oxide, transitional and sulphide mineralisation samples were 0.09% TS, 0.25% TS and 2.1% TS respectively.

13.4.2 Coarse Sample Bottle Roll Leach Tests Coarse sample bottle roll testing was conducted to identify the maximum gold recovery achievable from each of the samples at crush sizes typical of conventional heap leach operations. A series of tests were performed to investigate the effect of crush size on leach performance at a fixed cyanide concentration of 0.5 g/L. Coarse sample bottle roll leach tests were carried out for a leach duration of 30 days; various crush sizes were tested on the oxide (P100 -100 mm, -75 mm, -50 mm and -16 mm), and on the transitional mineralisation samples (P100 -75 mm, -16 mm and -6.3 mm). A summary of the gold recovery achieved during the 2019 coarse sample bottle roll test program are given in Table 13-14. Results showed that gold extraction for the oxide mineralisation samples appear to be largely independent of crush size. Whereas for the transitional mineralisation samples the gold extraction appears to be largely independent of crush size up to P100 -16 mm; further testing of transitional mineralisation samples at a P100 crush size of -25 mm is required to be able to complete the crush size matrix effect. Table 13-13: presents data on gold extraction from testing conducted on the different master composites at the optimum crush size of 100% -16 mm; after 30 days of leaching were: Table 13-13: Gold extraction from master composites 100% passing -16 mm

Deposit Material type Crush size (P100 mm) Leach (days) Leach recovery (Au %) Transitional 59 Oxide 89 Bigar Hill Transitional 64 Sulphide 7 16 30 Oxide 87 Korkan West Oxide 52 Korkan Oxide 82 Sulphide 6

CSA Global Report Nº R302.2019 107 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Table 13-14: Summary of coarse sample bottle roll test results (2019) Reagent Head (Au g/t) Mineralised Composite ID/ Reagent addition consumption %Au extraction (CN) CN leach material Deposit Test no. (kg/t of CN feed) (kg/t of CN (days) residue crush size CN CN feed) (Au g/t) Oxidation (inches) calc. direct NaCN CaO NaCN CaO 0.5 1 2 5 7 9 15 21 26 30 BH_P1_01 Bigar Hill -5/8" COBR-1 1.65 1.07 1.21 0.84 32 41 43 47 51 52 56 55 58 59 0.59 1.44 1.47 Transitional BH_P1_02 Bigar Hill -5/8” COBR-2 1.74 1.65 1.30 1.60 44 63 67 76 80 78 87 80 86 89 0.06 0.56 0.47 Oxide BH_P1_03 -5/8" COBR-3 1.68 0.97 1.23 0.96 30 44 49 53 57 57 61 60 62 64 0.37 1.02 Bigar Hill -1.5” COBR-4 2.08 1.00 1.73 0.91 24 36 37 41 44 45 50 46 51 54 0.33 0.71 0.98 Transitional -1/4" COBR-5 1.14 1.17 0.68 1.17 32 45 49 52 56 58 63 60 64 62 0.38 1.01 BH_P1_04 Bigar Hill -5/8" COBR-5 2.33 1.61 1.97 1.61 3 4 4 5 6 7 7 8 8 7 4.78 5.16 5.32 Sulphide KW_P1_01 -1/4" COBR-7 1.70 0.92 1.30 0.60 45 62 67 71 74 78 89 88 90 87 0.15 1.12 Korkan West -5/8" COBR-8 2.13 1.22 1.75 0.83 44 62 68 72 77 80 89 84 86 85 0.19 1.24 1.15 Oxide -1" COBR-9 1.09 0.89 0.56 0.89 49 60 65 66 68 76 84 81 88 86 0.17 1.20 KW_P1_02 -5/8" COBR-10 1.59 1.15 1.14 0.44 45 64 71 71 72 77 117 83 94 82 0.16 0.89 Korkan West -1.5” COBR-11 1.92 1.21 1.53 0.26 33 56 68 68 75 79 57 88 86 81 0.21 1.08 0.89 Oxide -2" COBR-12 2.06 1.23 1.69 0.01 32 60 70 74 73 73 84 76 75 83 0.16 0.92 KO_P1_01 -5/8" COBR-13 1.07 0.92 0.47 0.90 50 66 73 72 77 77 84 82 84 82 0.14 0.74 Korkan -1.5” COBR-14 1.77 1.04 1.32 -0.48 42 60 68 69 72 67 80 78 78 79 0.16 0.76 0.77 Oxide -1/4" COBR-15 0.97 0.91 0.46 0.88 45 61 66 67 67 66 81 78 82 82 0.14 0.78 KO_P1_02 -5/8" COBR-16 1.59 1.46 1.21 1.46 2 2 3 3 4 5 5 5 7 6 3.19 3.39 Korkan -1.5” COBR-17 2.82 1.45 2.51 0.84 2 2 2 2 3 5 6 7 8 8 3.98 4.31 3.43 Sulphide -1/4" COBR-18 1.57 1.46 1.19 1.46 2 2 3 3 4 5 5 5 7 6 3.15 3.34

CSA Global Report Nº R302.2019 108 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

13.4.3 Column Leach Testing Column leach testing was undertaken to provide confirmation of the achievable metal recoveries and leach rates from each of the samples under heap leaching conditions. A total of three tests were conducted on master composite samples representing transitional mineralised material from Bigar Hill, Korkan West and Korkan; at the base case crush size of 80% -12.5 mm. The Korkan West and Korkan samples were leached for a total of 65 days, whilst the Bigar Hill sample was leached for a total of 91 days, using a 0.5 g/L cyanide solution at a target solution application rate of 10 L/m2/hr. The pregnant leach solution was passed through activated carbon to adsorb the gold. The activated carbon was changed on days 1, 4 and 7, and weekly thereafter. The results of the column leach tests are summarised in Table 13-15. The results show good correlation between the gold extraction based on back calculated head, and that based on carbon assays and solids leach residue. A comparison of column leach vs coarse bottle roll leach test results is summarised in Table 13-16. Results show a good correlation between column and coarse bottle gold extractions.

CSA Global Report Nº R302.2019 109 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Table 13-15: Summary of column leach test results Column leach recovery (based on) Crush size Agglom. Head assay Calculated head Extracted grade Tails grade Sample ID Measured head Calculated head Carbon/Residue (80% mm) stage (Au g/t) (Au g/t) (Au g/t) (Au g/t) % Au Bigar Hill transitional -11.3 No 1.35 1.36 0.83 0.53 61.5 61.0 60.3 Korkan West oxide -12.3 No 1.15 1.15 1.04 0.11 90.4 90.4 89.3 Korkan oxide -12.3 No 0.77 0.80 0.66 0.14 85.7 82.5 82.1

Table 13-16: Comparison of column leach test vs coarse bottle roll leach test results

Leach Column leach recovery (based on) Bottle roll Mineralised Crush size Leach time Crush size Sample ID time Measured head Calculated head Loaded carbon/Residue leach recovery material type (80% mm) (days) (100% inch) (days) % Au (Au %) Bigar Hill Transitional 91 -11.3 61.5 61.0 60.3 30 -5/8 59 Korkan West Oxide -12.3 90.4 90.4 89.3 -5/8 87 63 Korkan Oxide -12.3 85.7 82.5 82.1 -5/8 82

CSA Global Report Nº R302.2019 110 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

The gold leach recoveries, based on column calculated head grades, as a function of time are shown in Table 13-17. Table 13-17: Gold leach recoveries based on column calculated head grades

Leach C-1 Leach C-2 C-3 days BH_P1_01 days KW_P1_01 KO_P1_01 1 15.3 1 12.8 38.2 2 31.3 2 69.5 71.4 4 40.2 4 82.3 76.6 7 46.5 7 84.1 78.3 15 53.2 15 87.6 80.2 21 56.3 21 88.4 80.7 28 57.4 28 88.6 81.1 35 58.3 35 88.8 81.4 42 59.0 42 88.9 81.5 49 59.4 49 89.0 81.7 56 59.7 56 89.1 81.9 63 59.9 63 89.1 82.0 70 60.0 65 89.3 82.1 77 60.2 84 60.3 91 60.3

The column leach curves for the three master composites are shown in Figure 13-25.

100

90

80

70

60

50

40 Au Au Extraction (%) 30

20

10 C-1 BH_P1_01 C-2 KW_P1_01 C-3 KO_P1_01 0 0 7 14 21 28 35 42 49 56 63 70 77 84 91 98 105 112 119 Leach Time (Days) Figure 13-25: Column leach curves Source: Avala, 2019

CSA Global Report Nº R302.2019 111 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

The test results showed gold recoveries to carbon ranged from 60.3% (BH_P1_01) to 89.3% (KW_P1_01). Analysis of the leach kinetics during the tests showed the gold to be fast leaching for the Korkan West and Korkan oxide mineralisation samples with between 80.2% (KO_P1_01 KW_P1_01) and 87.6% (KW_P1_01) of the total gold recovered within the first 14 days of leaching. Gold leach kinetics for the transitional mineralised zone was slower, and leaching was terminated after 91 days of leaching. During testing, consumption of lime was moderate, ranging from 0.61 kg/t to 0.90 kg/t whilst the consumption of cyanide was low, ranging from 0.15 kg/t to 0.36 kg/t.

13.5 COBR Test Results Summary (2019) The COBR test results for the oxide, transitional, and sulphide mineralisation samples are summarised in Table 13-18, Table 13-19, and Table 13-20. Table 13-18: Oxide mineralised material coarse sample bottle roll test results

Au S Sulphide Crush size (inches) Composite Classification (g/t) (%) S (%) 1/4 5/8 1 1.5 2 Met18_KO_01 1.44 0.02 0.01 96 93 93 Met18_KW_01 1.04 0.01 0.01 76 76 74 Met18_BH_01 1.86 0.04 0.01 94 94 94 BH_P1_02 0.47 0.13 0.11 89 Oxide KW_P1_01 1.15 0.10 0.09 86 87 85 KW_P1_02 0.89 0.04 0.03 82 81 83 KO_P1_01 0.77 0.04 0.03 82 82 79 Average 1.09 0.05 0.04 86.6 86.1 87.1

Table 13-19: Transitional mineralised material coarse sample bottle roll test results

Au S Sulphide Crush size (inches) Composite Classification (g/t) (%) S (%) 1/4 5/8 1 1.5 2 Met18_KO_02 1.73 0.48 0.41 55 53 54 BH_P1_01 1.47 0.45 0.39 59 Transitional BH_P1_03 0.98 0.40 0.32 62 64 54 Average 1.39 0.44 0.37 58.7

Table 13-20: Sulphide mineralised material coarse sample bottle roll test results

Au S Sulphide Crush size (inches) Composite Classification (g/t) (%) S (%) 1/4 5/8 1 1.5 2 BH_P1_04 5.32 2.34 2.10 7 KO_P1_02 Sulphide 3.43 1.77 1.50 6 6 8 Average 4.38 2.06 1.80 6.5

The effect of crush size on gold leach extractions are shown below in Figure 13-26. The gold leach extraction for the oxide mineralisation samples is largely independent of crush size.

CSA Global Report Nº R302.2019 112 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

30 Day COBR Au Extraction Rates

100 90 80 70 60 50 40

%AuExtraction 30 20 10 0 0 1 2 3 4 5 6 7 8 9 Sample

1/4" 5/8" 1.5" 2"

Figure 13-26: Crush size vs %Au extraction Source Avala 2019 Figure 13-27 shows a direct relationship between %TS and gold extraction. Further testing on oxide and transitional mineralisation samples with variable %TS head grades is required in order to firm up this trend.

COBR Extraction vs %S - All samples to date 100 90 80 70 60 50 40 30 %Au Extraction %Au 20 10 0 -10 0 0.5 1 1.5 2 2.5 %TS

Figure 13-27: %TS vs %Au extraction (all samples) Source Avala 2019

CSA Global Report Nº R302.2019 113 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

13.6 Metallurgical Data Interpretation and Predictions (2018 to 2019)

13.6.1 Preferred Process Option Based on the metallurgical performance obtained from column leach tests, the Timok mineralisation types can be considered amenable to heap leach technology. Capital and operating costs for heap leach technology are lower than those of conventional carbon-in-leach (CIL), for the same recovered gold ounces, therefore resulting in improved project economics.

13.6.2 Predicted Metallurgical Recovery For the purposes of the updated MRE, laboratory column gold extractions are normally discounted by two to three percentage points when estimating field extractions. The column leach recovery for the transitional mineralisation has been discounted by 4% to take into account variability. Gold recoveries to doré are also multiplied by 99% to take into gold solution and metal losses to slag. Based on the above discount values for gold leach extractions the predicted full-scale metal leach extractions are shown in Table 13-21 and Table 13-22 Table 13-21: Summary of discounted column leach test results Crush Column leach Corrected CIC/gold room Au Material type Correction Sample ID size recovery recovery recovery recovery description factor (mm) (% Au) (% Au) (%) to doré KO_01 Korkan oxide 94.4 2 92.4 99 91.5 KO_02 Korkan transitional 75.0 4 71.0 99 70.3 BH_01 Bigar Hill Oxide 94.2 2 92.2 99 91.3 KW_01 Korkan West oxide -12.5 75.5 2 73.5 99 72.8 BH_P1_01 Bigar Hill transitional 60.3 4 56.3 99 55.7 KW_P1_01 Korkan West oxide 89.3 2 87.3 99 86.4 KO_P1_01 Korkan oxide 82.1 2 80.1 99 79.3

Table 13-22: Gold recovery assumptions used in economic modelling Domain Material type Average Bigar Hill Korkan Korkan West Oxide material (to doré) 91% 91% 73% 88% Transitional material (to doré) 69% 69% 69% 69% Sulphide material (to sulphide concentrate) 75% 75% 75% 75%

13.6.3 Predicted Reagent Consumption Based upon typical heap leach operations with mostly clean non-reactive materials, cyanide consumption in production heaps would be only 25% to 33% of the laboratory column test consumptions. The predicted full-scale heap leach reagent consumptions for the oxide and transitional mineralisation samples are shown in Table 13-23 and *Assumed that 33% of the laboratory scale test reagent consumption consumed in full scale heap leach. Table 13-24. Table 13-23: Summary of reagent consumptions (oxide samples)

Crush size Reagent consumption (kg/t) Column ID Deposit ID (mm) Lime NaCN Lime* NaCN*

CSA Global Report Nº R302.2019 114 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

BH-01 Bigar Hill -12.9 1.21 0.36 0.40 0.12 KO_01 Korkan -12.7 0.88 0.21 0.29 0.07 KW_01 Korkan West -12.6 0.99 0.30 0.33 0.10 KW_P1_01 Korkan West -12.3 0.61 0.15 0.20 0.05 KO_P1_01 Korkan -12.3 0.70 0.25 0.23 0.08 Average 1.03 0.29 0.34 0.10 *Assumed that 33% of the laboratory scale test reagent consumption consumed in full scale heap leach. Table 13-24: Summary of reagent consumptions (transitional samples)

Crush size Reagent consumption (kg/t) Column ID Deposit ID (mm) Lime NaCN Lime* NaCN* KO_02 Korkan -12.3 0.90 0.36 0.30 0.12 BH_P1_01 Bigar Hill -11.3 0.67 0.81 0.33 0.10 Average 0.79 0.59 0.31 0.11 *Assumed that 33% of the laboratory scale test reagent consumption consumed in full scale heap leach.

13.6.4 Predicted Leach Cycle Time Leach profiles were plotted to determine leach cycle time. Leach cycle times for full scale heap leach operations are typically measured in tonnes of leach solution applied to tonnes of mineralisation under leach (ts/to ratio). The full leach cycle is not normally completed with a single continuous application of solution. The cycle is usually broken down into the primary leach cycle where solution is directly applied to the mineralisation under leach, and a secondary leach cycle, where solution flows through an area previously leached from a lift above. Figure 13-28 and Figure 13-29 shows gold leach extraction as a function of solution flux rate (ts/to) for the oxide and transitional column leach tests. Full gold leach recovery is achieved between a flux rate (ts/to) of 2.5 (oxide mineralised material) and 3.5:1 (transitional mineralised material).

CSA Global Report Nº R302.2019 115 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Flux Rate vs Au Leach Extraction Oxide Material Types 100

90

80

70

60

50

40

%Au Leach Extraction Leach %Au 30

20 KO-01 BH-01 KW-01 10 KW_P1_01 KO_P1_01 0 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 Flux Rate (ts/to)

Figure 13-28: Gold flux rate curves (oxide mineralised material) Source: CSA Global 2019

Flux Rate vs %Au Extraction Transitional Material Types 100

80

60

40

%Au Leach Extraction Leach %Au 20 KO_02 BH_P1_01 0 0 1 2 3 4 5 6 Flux Rate to/ts Figure 13-29: Gold flux rate curves (transitional) Source: CSA Global 2019

Oxide Mineralised Material The primary leach cycle has been designed for a flux rate of 1.1:1 (14 full scale leach days), whilst the secondary leach cycle has been designed for a flux rate of 1.4:1 (17.5 leach days). The combined design total

CSA Global Report Nº R302.2019 116 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

flux rate is 2.5:1. The remainder of the gold would be leached during leaching of mineralisation in the subsequent lifts above (i.e. lifts two and three). At the design flux rate (ts/to) of 1.1:1 the predicted gold recovery for the metallurgical composites are: • Korkan oxide (KO_01) mineralisation is 92.1%, of the ultimate gold leach recovery of 94.4% (uncorrected). • Bigar Hill (BH_01) oxide mineralisation is 75.9% of the ultimate gold leach recovery of 94.2% (uncorrected). • Korkan West (KW_01) oxide mineralisation is 70.5% of the ultimate gold leach recovery of 75.5% (uncorrected). • Korkan West (KW_P1_01) oxide mineralisation is 84.1% of the ultimate gold leach recovery of 89.3% (uncorrected). • Korkan oxide (KO_P1_01) mineralisation is 78.3% of the ultimate gold leach recovery of 82.1% (uncorrected).

Transitional Mineralised Material The primary leach cycle has been designed for a flux rate of 1.6:1 (20 full scale leach days), whilst the secondary leach cycle has been designed for a flux rate of 2.0:1 (25 leach days). The combined design total flux rate is 3.6:1. The remainder of the gold would be leached during leaching of mineralisation in the subsequent lifts above (i.e. lifts two and three). At the design flux rate (ts/to) of 1.6:1 the predicted gold recovery for the metallurgical composites are: • Korkan (KO_02) transitional mineralisation is 56.0%, of the ultimate gold leach recovery of 67.9% (uncorrected) • Bigar Hill (BH_P1_01) oxide mineralisation is 53.2% of the ultimate gold leach recovery of 60.1% (uncorrected). Predicted Leach Cycle Time Comments There is a correlation between the solution application rate and days of leaching, the latter derived from the heap lift height (8 m), design cell size for each primary leach cycle, and the solution irrigation rate of 10 L/m2/hr. The primary and secondary leach cycles are seven days and 56 days, resulting in a total leach cycle of 63 days. As the column leach tests were not conducted under conditions equivalent to the proposed HLF, i.e. 2 m tall column tests, vs 8 m high lifts, the rate of gold extraction from the columns tests were scaled to industrial conditions by equating the column test extraction rates as a function of cumulative solution to the crushed feed ratio to the proposed industrial flux rate. The gold leach extraction has been scaled to reflect actual field days. Figure 13-30 and Figure 13-31 show gold leach recovery as a function of actual field days for the oxide mineralisation samples and transitional mineralisation samples respectively.

CSA Global Report Nº R302.2019 117 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Field Days vs % Au Extraction Oxide Material Types 100

90

80

70

60

50

40

%Au Leach Extraction Leach %Au 30

20 KO-01 BH-01 KW-01 10 KW_P1_01 KO_P1_01 0 0 50 100 150 200 250 300 Field Leach Days

Figure 13-30: Field leach days vs %Au extraction (oxide) Source CSA Global 2019

CSA Global Report Nº R302.2019 118 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Field Days vs %Au Extn. Transitional Material Types 100

90

80

70

60

50

40

%Au Leach Extraction Leach %Au 30

20

10 BH_P1_01 KO_02

0 0 50 100 150 200 250 300 350 400 Field Leach Days

Figure 13-31: Field leach days vs %Au extraction (transitional) Source CSA Global 2019

13.7 Sulphide Mineralised Material Testwork (2012 to 2013) Avala initiated the scoping-level metallurgical study for the Project during 2011. The primary objective of the scoping-level study during 2011 and 2012 was to determine the potential recovery of gold and identify the potential processing options. The focus of these programs was predominantly cyanide leaching, including refractory gold recovery enhancement techniques such as pressure oxidation (POX). The results of these testwork programs have been described in a previous NI 43-101 report by AMEC (2013). Combinations of flotation, refractory concentrate treatment (roasting, POX or bio-oxidation), with cyanide leaching of oxidised concentrate and flotation tailings was necessary to achieve reasonable gold recovery (72– 76% overall). Given the refractory nature of the mineralised material and concentrate, the capital and operating costs for adopting a pre-oxidative stage were prohibitive. A simplified approach aiming to maximise recovery to a flotation concentrate for third party treatment, either domestically or internationally, offers the lowest on-site capital and operating costs and this approach was adopted for the 2014 PEA study design (AMEC, 2014), and has been taken forward as the base case process flowsheet by DPM for treating the sulphide mineralised material types from the Timok deposit. In 2013, metallurgical testwork focused on demonstrating that milling and flotation could produce a gold-rich sulphide concentrate (for further downstream processing) and form the basis of a viable process flowsheet. Metallurgical testwork focused the two largest deposits, Bigar Hill and Korkan, although some work was also completed on samples from the Kraku Pester deposit. All testing was undertaken on composite samples to minimise any sample variability whilst other testing parameters were under assessment.

CSA Global Report Nº R302.2019 119 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

The relevant testwork completed is in the following reports: • Phase 1 Program: “Preliminary Testing of Various Ore Samples from Timok Deposit, Serbia”, SGS Mineral Services UK Ltd, 26 May 2012: o In general, the Phase 1 testwork program (SGS UK) focused on selective flotation of sulphides and gold associated with non-sulphide gangue to produce a gold-rich concentrate including the optimisation of flotation feed grind size, reagent types and other flotation parameters. • Phase 2 Program: “Phase 2 Testing of 3 Ore Types from Avala Resources (Draft)”, SGS Mineral Services UK Ltd, Project No. 10866-410, 29 October 2013: o The Phase 2 program (SGS UK) further explored the fine grind milling and flotation approach to further confirm the veracity of the production of a saleable gold-rich sulphide concentrate. A wide range of variables were tested including laboratory flotation procedures, flotation froth removal rates and final grind sizes. In general, the optimum flotation reagent types and dosage rates derived from the Phase 1 program were employed for the Phase 2 program in order to minimise testwork variables. In addition, testwork was initiated to establish the veracity of a beneficiation by size approach where a barren reject fraction could be separated from a high-grade fraction after attritioning. • SGS Lakefield Tests: “SGS Lakefield Lab Floats – Avala – 20131104.xls”, SGS Lakefield testwork results summary received from Woodgrove: o Samples were despatched to SGS Lakefield (Canada) for bench scale and Woodgrove “Mini-SFR pilot plant” flotation testwork. Unfortunately, the latter tests were not successful due to procedural issues with the newly commissioned pilot plant equipment but the SGS laboratory testwork results were instructive. • Extra flotation testing: o An additionalflotation program was conducted by SGS (UK) in late 2013 to compare the performance of an alternative (less expensive) flotation reagent regime and confirm final concentrate analyses for marketing purposes. Similarly, additional high intensity scrubbing testwork was undertaken with some encouraging results but not formally reported. In 2016, a metallurgical sample from Bigar Hill was submitted to Dundee Sustainable Technologies (DST) for characterisation work.

13.7.1 Mineralogical Characterisation (2012 to 2013) Mineralogical characterisation testing has been undertaken by SGS UK and key aspects of the studies include the following: • Sulphur grades are relatively low. Head assays for the Phase 2 program composite samples (multi-hole and multi-interval) indicated sulphur to gold ratios of approximately 1.5, 0.87 and 2.80 for the Bigar Hill, Korkan and Kraku Pester samples, respectively. • X-ray diffraction analyses indicate that: o almost the entire sulphides content is present as pyrite, with less than 10% (relative) classed as other sulphides including chalcopyrite and pyrrhotite. o gangue is dominantly quartz, calcite, dolomite (Korkan) and feldspars (Kraku Pester), but some samples also showed significant levels of clays and micaceous minerals. • QEMSEM analysis indicated that mean pyrite grain sizes of 25 μm, 19 μm and 17 μm were applicable for the Bigar Hill, Korkan, and Kraku Pester composite subsamples, respectively.

CSA Global Report Nº R302.2019 120 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

• Approximately 34–50% of the free and liberated sulphide particles are under 25 μm, whereas 29–38% are above 25 μm in size. The total proportion of pyrite classified as free or liberated was 45–92%. The remainder was classified as middlings, where composite particles with quartz/feldspar and calcite represented the main occurrences. • Pyrite exposure (defined as greater than 50% exposed) was reported as 82%, 74% and 59% for the Bigar Hill, Korkan, and Kraku Pester composite subsamples, respectively. These levels of exposure should render sulphide particles amenable to flotation, but samples with pyrite exposure values closer to 50% can be expected to exhibit slower flotation kinetics. • Dynamic secondary ions mass spectrometry (D-SIMS) examinations undertaken in 2012 indicated the presence of substantial quantities of sub-microscopic and solid solution gold within the tested sample. Gold was observed within pyrite, chalcopyrite and iron oxide host minerals. Three different pyrite types were observed (coarse, porous and fine) where each displayed varying gold grades. The mineralogical characterisation studies largely explain the flotation performance observed for the various mineralised material type samples during the combined flotation testing results where it may be concluded that: • Pyrite grain size is relatively fine, and a corresponding fine flotation feed size will be required for optimum sulphur recoveries. • Whilst gold is associated with pyrite, there is a significant proportion associated with the oxide components. Bright phase analysis targeting gold values would be useful to quantify non-sulphide mineral associations. • Gold grades varied within the three pyrite types and, as it can be expected that each pyrite type displays differing flotation rates, provided some insight into the flotation concentrate gold grade kinetic profile.

13.7.2 Flotation Testwork (2012 to 2013) As outlined in Section 13.7, four testwork programs focused on flotation were completed on sulphide mineralisation samples from the Timok deposit in 2012 and 2013.

Phase 1 Flotation Testing (2012) (SGS UK Report 10866-255) The Phase 1 testwork program investigated various parameters to demonstrate the ultra fine grinding (UFG)/flotation concept including some optimisation of flotation feed grind size, reagent types and other flotation parameters. Testwork was completed on several samples representing Bigar Hill, Korkan and Kraku Pester mineralised material types. The program included the following flotation related testing: • Detailed head assays and mineralogical characterisation. • Gravity recovery testing. Comparative flotation testing demonstrated superior gold recoveries and grade relationships and gravity recovery testing was abandoned during the program. • Batch flotation testing at various grind sizes and conditions, including the investigation of several reagent addition regimes. The results from this program have not been used as metallurgical recovery inputs for the updated 2018 MRE for the sulphide mineralisation zone due to the following considerations: • Gold head assays for the “original” samples varied between 0.52 g/t and 1.18 g/t which are significantly lower than the current expected life-of-mine (LOM) flotation feed gold grade of around 2 g/t. The subsequently provided “new blue” samples were more consistent with respect to gold head grade and ranged between 1.54 g/t and 3.34 g/t.

CSA Global Report Nº R302.2019 121 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

• Tested grind P80 size was mainly 37 μm (100% passing 53 μm) although some 53 μm and 25 μm tests were completed. This compares to the target flotation feed P80 size of 20 μm. • Very high rougher concentrate weights (generally around 40% but up to 70%) were recorded for most tests in an attempt to maximise rougher gold and sulphur recovery. Phase 2 Flotation Testing (2013) (SGS UK Report 10866-410) Detailed metallurgical testwork during 2013 focused on the two largest deposits, Bigar Hill and Korkan, although some work was also completed on samples from the Kraku Pester deposit. General information regarding the composite samples used for the Phase 2 testwork program is presented in Table 13-25. Further sample details are available within the AMEC (2013) technical report. Table 13-25: Metallurgical testwork sample summary (2013) Sample ID Deposit No. of holes Intersection (m) Weight (kg) Au (g/t) S (%) MET13_KO_01 Korkan 7 30 62.9 1.51 1.48 MET13_BH_01 Bigar Hill 5 52 113.00 1.45 3.14 MET13_PE_01 Kraku Pester 2 20 50.6 1.41 4.36

The Phase 2 testwork program further explored the fine grind milling and flotation approach to generally confirm the potential for the production of a saleable gold-rich sulphide concentrate. A wide range of flotation parameters were tested including laboratory flotation procedures, flotation froth removal rates and flotation feed grind sizes. In general, the optimum flotation reagent types and dosage rates derived from the Phase 1 program were employed for the Phase 2 program in order to minimise testwork variables. These reagents included proprietary flotation collectors developed specifically for the flotation of gold-containing oxide- based minerals as well as more common sulphide mineralisation. In general, the following flotation related testing was conducted: • Detailed head assays and mineralogical characterisation.

• Investigation of coarse flotation applicability via four-stage sequential flotation tests at reducing P80 grind sizes (i.e. 75 μm, 53 μm, 38 μm and 20 μm). The BH test indicated very good sulphur recoveries (to 92%) but relatively poor gold recoveries (up to 68%) as gold-containing oxide particles were not floated during the procedure due to the lack of flotation time at the finer final tested grind size. Similar results were reported for a KO sequential test and this approach was abandoned for the remainder of the program. • All rougher-scavenger flotation testing was undertaken using the Phase 1 Variability testing FT7 reagent addition regime (i.e. 100 g/t MaxGold 900 and 100 g/t Aero 3418A collectors, no activation and minor dispersant additions). • Several physical rougher flotation test methods were investigated where various grind sizes, scrape rates, cell types and general rougher concentrate pulling procedures were adopted. Testwork demonstrated that the metal recovery rate is related to the percentage pyrite exposed and the offset between the sulphur recovery and the gold recovery being due to the gold that is associated with the gangue minerals. The Bigar Hill mineralised material type is the least mineralogically constrained in terms of pyrite associations, followed secondly by Korkan and thirdly by Kraku Pester. The metallurgical results reported herein, show that this mineralogy is the main driver on the metallurgical responses, which are not surprisingly, best for Bigar Hill and worse for Kraku Pester.

CSA Global Report Nº R302.2019 122 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

It was clear from the results that the grind size required for effective liberation of sulphides was circa 75 µm for Bigar Hill (slightly finer for Korkan and Kraku Pester) but that ultimately a 20 µm grind size is required to maximise rougher gold recovery on Bigar Hill and Korkan mineralised material types. The Kraku Pester mineralised material type is clearly still mineralogically constrained at 20 µm and so some ultra-fine (10 µm and 5 µm) grinds were performed on this mineralised material type alone but were unsuccessful. As a consequence, no further testing was conducted on this mineralised material type, and Bigar Hill and Korkan became the main focus of testing. Flash flotation simulation gave good sulphur recoveries and demonstrates that pyrite and gold can be floated at up to 75 µm, but the overall terminal gold recovery was lower than when the material is ground to 20 µm and then floated. This could be because the pyrite provides a “flotation carrier” for the gold-gangue particles. This needs further investigation before it can be built into the design. The metallurgical and mineralogical results for Bigar Hill show that it has greater potential than the other two mineralised material types. The exposed pyrite surfaces for Bigar Hill is 82%, Korkan is 73%, whilst the Kraku Pester sample is 59%. These exposed pyrite edges are what helps the bubbles attach to the particles during flotation. The lower the percentage of exposed edges the harder it is to conventionally float. By linking the mineralogy to metallurgy, the results follow a similar pattern, the metallurgical comparison of the three mineralised material types is shown in Table 13-26, Bigar Hill has the highest sulphide and gold recovery 96.9% and 86.8% respectively. Korkan is next recoveries of 80% and 71% respectively and Kraku Pester has an excellent sulphur recovery of 74.7% but the gold recovery is extremely poor at 51.6%.

Batch Rougher Tests (2013) Bigar Hill was shown to be the best rougher flotation performer of the samples tested, followed by Korkan and then Kraku Pester, as illustrated in Table 13-26. Table 13-26: Rougher flotation optimisation results summary (2013)

Mineralised Mass pull Grade Recovery Sample ID material type (Wt.%) Au (g/t) S (%) Au (%) S (%) Bigar Hill MET13_BH_01 24.5 5.5 9.7 86.8 96.9 Korkan MET13_KO_01 17.9 6.32 6.7 71.8 80.2 Kraku Pester MET13_KP_01 24.8 2.98 12.7 51.6 74.7

It must be noted that to achieve the 86.8% gold recovery and the 96.9% sulphur recovery the weight needed was 24.5%, which is extremely high. It is lower than the weight pulls achieved in the previous testwork where the froth removal was extracted at a higher more conventional rate. Results of the bulk rougher tests carried out on the Kraku Pester mineralisation sample are shown in Table 13-27.

CSA Global Report Nº R302.2019 123 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Table 13-27: Kraku Pester rougher flotation results (2013) Kraku Pester (grind size 6 µm) Kraku Pester (grind size 11 µm) Mass pull % Au % Cum. Grade Mass pull % Au % Cum. Grade (Wt.%) recovery recovery Au (g/t) Cum. Au (g/t) (Wt.%) recovery recovery Au (g/t) Cum. Au (g/t) 2.5 6.60 6.60 3.58 3.58 2.20 7.38 7.38 4.44 4.44 2.5 7.68 14.28 4.31 3.94 1.9 7.45 14.83 5.1 4.75 1.9 6.62 20.90 4.86 4.2 1.60 7.47 22.30 6.14 5.14 3.6 13.85 34.75 5.33 4.58 2.6 11.83 34.13 5.89 5.38 4.5 19.15 53.90 5.82 5.0 3.8 12.13 46.26 4.24 5.02 3.4 6.53 60.43 2.68 4.54 3.3 7.97 54.23 3.17 4.63 81.60 39.57 100.00 0.67 1.38 84.6 45.77 100.00 0.71 1.31 100.00 100.00 1.38 100.00 100.00 1.31

Results in Table 13-27 show that gold recovery to the bulk sulphide concentrate increases with increasing liberation fineness.

Open Cycle Cleaner Tests (2013) The Phase 2 FT8 flotation tests included bulk rougher-scavenger tests to produce concentrates for subsequent open cycle cleaner (OCC) testing of the Bigar Hill and Korkan composite samples (no cleaning flotation was conducted on Kraku Pester samples). Unfortunately, the gold recoveries to the bulk rougher concentrate were not satisfactory for the FT8BH test at 77% and, particularly, the FT8KO test at 60%. The reported calculated head gold grade of the FT8KO test correlated poorly with the assay head grade. This is due to an error in the FT8KO reported rougher tailings gold grade assay of 0.18 g/t whereas the actual assayed grade was 0.64 g/t which has a detrimental effect on all the reported recoveries for that test. The applicable information has been corrected within this report. In both cases, the relatively low gold recoveries were due to insufficient concentrate weight (around 7%). Notwithstanding the relatively low rougher gold recoveries for the FT8 tests (which were also reflected in the subsequent cleaner testing results), reasonable cleaner concentrate grades were demonstrated as presented in Table 13-28. Table 13-28: OCC test results summary Sample description Bigar Hill (FT88H) Korkan (FT8KO) Mass pull Grade Mass Grade % Au % Cum. % Au % Cum. Product cum. pull recovery recovery Au Cum. recovery recovery Au Cum. (Wt.%) (g/t) Au (g/t) (Wt.%) (g/t) Au (g/t) Cleaner 1 conc. 1.60 22.43 22.43 19.90 19.9 1.00 25.34 25.24 26.43 26.4 Cleaner 1-2 conc. 3.41 27.82 50.25 21.89 20.9 1.71 17.69 41.4 23.69 25.3 Cleaner 1-3 conc. 4.08 10.15 60.40 21.44 21.0 2.10 8.99 50.24 23.55 25 Cleaner 1-4 conc. 4.50 6.58 66.99 22.01 21.1 2.46 4.92 55.61 15.76 23.6 Cleaner 1-5 conc. 4.68 2.18 69.17 17.50 21.0 2.92 8.24 63.98 18.87 22.9 Cleaner tailings 7.79 8.09 3.69 14.1 7.19 14.99 4.89 12.2 Rougher conc. 7.79 77.26 77.3 14.07 7.19 80.16 83.99 12.19 Rougher tailings 22.74 0.35 92.81 19.84 0.18 Calculated head 100.00 100.00 100.00 1.42 100.00 100.00 100.0 1.04

CSA Global Report Nº R302.2019 124 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

The difference in cleaning of the concentrates can be compared in Figure 13-32. The sulphide grades achieved in Korkan and that of Bigar Hill are different. Bigar Hill mineralised material produced a 92% pyrite concentrate compared with 62% pyrite concentrate on Korkan. This is undoubtedly due to the coarser pyrite mineralogical associations on Bigar Hill when compared with Korkan. The very fine pyrite mineral associations observed for Kraku Pester are undoubtedly the reason why this mineralised material type shows the least tendency to upgrade selectively.

Figure 13-32: OCC grade-recovery curves Source: Avala, 2018 Results of the OCC flotation tests demonstrated: • For the FT8BH test, a relatively flat gold grade-recovery relationship was reported where a final cleaner concentrate (1 to 5 combined) gold grade of 21 g/t was obtained. Gold recovery to the final concentrate of 69.2% represented an absolute loss of 8.1% gold recovery to the cleaner tailing. The final concentrate weight was 4.7% of the flotation feed. • For the FT8KO test, the grade-recovery relationship was more typical where the reported final cleaner concentrate (1 to 5 combined) gold grade was 23 g/t. Gold recovery to the final concentrate of 65.0% represented an absolute loss of 20.0% gold recovery to the cleaner tailing. The final concentrate weight was 2.9% of the flotation feed. All FT8 cleaner tests were conducted in open circuit so the recoveries stated above reflect that the cleaner tail gold is lost from the system. In reality, the cleaner tail would be recycled to the head of the cleaner circuit (or similar such as a separate cleaner-scavenger circuit) and a proportion could report to the final concentrate thus improving gold recovery. However, this also implies that the final concentrate weight will increase with consequent reduction in the final concentrate grade. Whilst the BH curve is not typical (it is considered unusual for both grade and recovery to increase on a cumulative basis), a final concentrate weight of approximately 4.7% appears optimum for this sample.

CSA Global Report Nº R302.2019 125 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

The average calculated Au head grades for the MET13 samples used in all Phase 2 and SGS Lakefield flotation tests were 1.47 g/t and 1.44 g/t for the BH and KO mineralised material types. These grades are considered relatively low where corresponding LOM mill feed gold grades of around 2 g/t have been determined for the sulphide ores as per the MRE dated 15 May 2018. As such, and if final concentrate weights remain near the values shown above, final concentrate grades are expected to be higher than demonstrated by these tests and require verification via variability testwork. No locked cycle or similar cleaner testing has been conducted to date. Similarly, no variability style testing has been completed so the effect of higher flotation feed gold and sulphur grades on parameters such as concentrate weights, recoveries and grades is not able to be confirmed at this stage.

Flotation Testwork Results Summary (2012 to 2013) The relevant data available from the rougher-scavenger flotation testwork programs described above was collated and assessed and individual tests selected for further analysis where the following rationale was employed: • Phase 1 results have not been included for the reasons discussed in Section 13.7.2. • Kraku Pester flotation testing during the Stage 2 program was not entirely comparable with that undertaken for the Bigar Hill and Korkan samples, Due to differences in analytical methodology. In addition, Kraku Pester samples were not included within the Lakefield SGS or Extra Flotation testwork programs, and cleaning flotation testing was not undertaken. • In general, some Phase 2 rougher-scavenger flotation test results were not considered relevant as the results are anomalous or not consistent with the final design approach, such as: o FT1 and FT2 procedures produced poor results due to the slower scrape rates, presumably as the gold containing oxide particles dropped out of the froth.

o FT9, FT10 and FT11 flotation test procedures were undertaken at P80 grind sizes of 75 μm, 53 μm and 35 μm, respectively and are thus not comparable with the target grind size basis of a flotation on feed P80 size of 20 μm. • Each of the rougher-scavenger flotation tests used to generate cleaner flotation test feed for the Phase 2 FT8 tests did not perform as well as some of the corresponding rougher-scavenger tests. • The SGS Lakefield BH1 and KO1 tests produced poor results, perhaps due to effects associated with the very fine grind size (~7 μm P80) although BH2 performed much better (albeit at a higher concentrate weight). Elimination of the individual flotation tests described above allowed for a selected Phase 2 and Lakefield SGS flotation testwork results dataset for further detailed analysis to derive preliminary metallurgical parameters suitable for the base case flotation circuit. This data set represented Bigar Hill and Korkan samples only. The flotation performance of the corresponding Kraku Pester sample has not been specifically considered for the flotation circuit design due to the later planned Kraku Pester pit development and mineralised material treatment period. However, Kraku Pester projected flotation performance has been assessed under the conditions expected to apply for treatment of this mineralised material using the base case flowsheet for the estimation of metal recoveries and similar values required for the 2018 MRE input parameters (CSA Global, 2018).

CSA Global Report Nº R302.2019 126 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

13.8 Chlorination Laboratory Testwork (2016) In 2016, Avala submitted 16 bags of core samples from Bigar Hill to DST for characterisation work. The work is outlined in the Avala document “Results of Laboratory Tests on Avala ‘Ore’” and the conclusions have been extracted and presented below. • A composite sample, BH-1, was made up to represent an “average Ca content” sample. The composite sample assays were: 1.98 g/t Au, 24.8 g/t Ag, 1.27% S2-, 2.20% Fe, 10.4% Ca. • A concentrate and tails sample were produced through a locked cycle flotation test of the composite sample having a P80 of 60.1 µm – thus coarser as the Phase 2 flotation work conducted in 2013. • Chlorination with sodium hypochlorite tests were performed on the composite, the flotation concentrate, and tails samples. All samples were subjected to a sulphuric acid leach prior to chlorination. • The Au associated with the sulphides is considered highly refractory and requires an oxidation step to maximise Au recovery from the leaching step. Chlorination feed samples containing more than 1% sulphides, composite and concentrate sample, were subjected to an oxidation step to reduce the sulphur content to <1% prior to the acid leach and chlorination. • Chlorination of the various samples achieved the following recoveries: o BH-1 composite sample without oxidation: 75.0% o BH-1 composite sample with oxidation: 83.8% o Flotation concentrate of BH-1 with oxidation: 61.6% o Flotation tails of BH-1 without oxidation: 82.6%. • Based on the results from these samples, the potential to extract a further ~80% Au from the flotation tails, without oxidation, indicate that the potential to recover 85–90% total Au exists by considering other processing options.

13.9 Conclusions

13.9.1 Oxide/Transitional Mineralised Material Results of the coarse bottle roll leach tests indicated gold leach extractions ranging from 53% for the Korkan transitional mineralised material to 94% for the Bigar Hill oxide mineralised material, after 14 days of leaching, and at a crush size of 100% -16 mm. Leach curves indicated that gold leaching was still ongoing after 14 days of leaching when the tests were terminated. Column leach tests carried out at the optimal crush size of 80% -12.5 mm exhibited fast leach kinetics except for the Korkan transitional mineralised material, where leaching was still ongoing of 63 days when the tests were terminated. Lime consumption is moderate and cyanide consumption is low for all mineralised material types. Extended leach times in both the coarse bottle roll, and column leach tests, appeared to be beneficial with respect to increasing gold extraction for the slower leaching transitional mineralised material types. The projected gold recovery, reagent consumption, leach time and crush size based on the column leach testwork results are summarised in Table 13-1. Size-by-size analysis of the column leach test feed and tails samples shows gold evenly distributed among the size classes, roughly following the mass splits. Some of the metallurgical samples showed low gold recovery in the coarse size fractions; +19.0 mm.

CSA Global Report Nº R302.2019 127 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

There was generally good correlation between gold extraction obtained from the coarse bottle roll leach and column leach tests apart from the Korkan transitional mineralised material which was still leaching in both tests. Results of the testing program indicate that oxide and transitional mineralisation samples from the Timok Gold Project are amenable to heap leach processing. Leach rates are relatively fast with high gold recovery for the oxide mineralised zone, and moderate gold recovery for the transitional mineralised zone. Coarse bottle roll leach tests carried out at coarse crush size tend to indicate that gold extraction for the oxide and transitional mineralised zones is independent of liberation up until a crush size of 100% -1.0 mm; after which the percent gold extraction decreases for the transitional mineralised material type.

13.9.2 Sulphide Mineralised Material Testwork demonstrates that the sulphide mineralised material types are amenable to bulk sulphide flotation to produce a gold-bearing sulphide concentrate. Based on testing, the following predictions on gold and sulphur recoveries to the cleaner concentrate are: • Final concentrate gold recovery of 70%, 65% and 50% for Bigar Hill, Korkan and Kraku Pester mineralised material types, respectively. • Similarly, final concentrate sulphur recoveries of 90%, 80% and 70% for Bigar Hill, Korkan and Kraku Pester mineralised material types, respectively. Recommendations from the characterisation work completed in 2016 by DST are as follows: • All gold chlorination tests were preceded by sulphuric acid leaching. Chlorination tests should be conducted without acid leaching to verify the removal of this step on the gold recovery. If the gold recovery remains stable, the withdrawal of acid leaching will simplify the process and reduce the capital cost. • The chlorination of the oxidised and acid-leaching sulphide concentrate provided a gold recovery of 61.6%. Gold is possibly finely disseminated in sulphides. Generally, an UFG of the sulphide concentrate should improve the gold recovery. Equipment for UFG is available (ISA mill), but their capacity is rather limited. UFG of the sulphide concentrate is feasible because its tonnage represents only 4.25% of the feed. UFG tests of the sulphide concentrate should be performed to verify whether gold recovery will increase.

CSA Global Report Nº R302.2019 128 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

14 Mineral Resource Estimates

14.1 Introduction CSA Global has previously reported a MRE update for the Bigar Hill, Korkan and Korkan West, and Kraku Pester deposits dated 7 November 2018 (CSA Global, 2018). The 2018 MRE remains current and a summary of all relevant methodology, parameters and key assumptions regarding the preparation of the updated MRE is reported below. The reader is referred to the 2018 Technical Report for full details of the MRE.

14.2 2018 Mineral Resource Update The MRE was based on interpretations using integrated geological and grade information recorded from RC and diamond core logging and assaying. DPM geologists conducted the geological interpretation and modelling work using the Leapfrog software package. CSA Global reviewed these models and found them suitable for use in the MRE. The estimation work was completed by CSA Global using the Datamine Studio and Isatis software packages. The date of receipt of final data for the Bigar Hill, Korkan and Korkan West, and Kraku Pester deposits was 15 May 2018, which is considered the effective date of the MRE. The deposits have been evaluated regarding the UTM grid (Zone 34 North in WGS 84 datum), and all directional references in the MRE portions of this report are according to this grid. Solid wireframes were created to represent the geological units at each of the four deposits, as follows: • Bigar Hill: o Overburden, Andesite Sill, Marl, Conglomerates (S2), Sandstones and Conglomerates (S1), Basal Breccia, Jurassic Limestone and Metamorphic Phyllite. • Korkan: o Overburden, Andesite Sill, Hornblende Diorite Porphyry, Marl, Conglomerates (S2), Sandstones and Conglomerates (S1), Basal Breccia, Lower Cretaceous Limestone and Metamorphic Phyllite. • Korkan West: o Overburden, Marl, Conglomerates (S2), Sandstones and Conglomerates (S1), Lower Cretaceous Limestones, Jurassic Limestone and Metamorphic Phyllite. • Kraku Pester: o Overburden, Andesite Sill, Hornblende Diorite Porphyry, Skarn, Marl, Monzonite, Jurassic Limestone and Metamorphic Phyllite. Suites of interpreted fault structures at each deposit were defined as wireframe planes, and solid wireframes were created to represent sulphide (fresh – no oxidation), partial oxidation and complete oxidation. Within each of the deposits, main zones of concentrated gold mineralisation were identified and modelled as solid shapes, in the form of loose mineralised shells corresponding to a broad cut-off grade of 0.1 g/t gold. These zones totalled four at Bigar Hill, three at Korkan, two at Korkan West, and one at Kraku Pester. The geometries of the mineralisation zones are generally aligned with local stratigraphic trends although, at Korkan in particular, there are indications of structural influences on the distribution of mineralisation. The stratigraphic, structural and mineralisation surfaces and solids were used as constraints in the construction of a cell model, based on parent cell XYZ dimensions of 20 m x 20 m x 10 m. To better represent the geometries of the mineralisation, cells were permitted to reduce to 5 m, 5 m, and 5 m in the X, Y and Z dimensions respectively. Models were coded to reflect the stratigraphic units and individual mineralised

CSA Global Report Nº R302.2019 129 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

zones, as well as to distinguish between weathered and unweathered material. Triangulated surfaces of topography were used to constrain the upper bounds of the models. Drill-hole samples were coded by stratigraphic unit, mineralisation zone and weathering in a manner consistent with the cell model. The high coefficients of variation (CV) values within the individual mineralisation zones indicate highly-skewed distributions with large grade ranges, or more than one population within a mineralised shell. These distribution characteristics are consistent with expectations given the loosely-domained mineralised shells based on a notional 0.1 g/t gold cut-off. At Bigar Hill, the mineralised global gold mean grade is relatively high (0.49 g/t) compared to Korkan (0.42 g/t) and Korkan West (0.35 g/t). The mean gold grade for the single identified zone at Kraku Pester is low (0.31 g/t), reflecting the high proportion of very low grades captured within the mineralised shell. Within each deposit area, mineralisation data was grouped by geological unit. Following statistical and visual review of grade distribution and continuity, selected mineralised geological units were combined into estimation domains. Samples were composited to 1 m lengths within these domains, which is the most common sample interval length. Higher grades in the various domains represent relatively small proportions of each complete domain grade distributions and tend to be spatially discontinuous on a local scale, within more continuous trends of elevated grades, at larger scales. Log probability plots and the spatial distribution of higher grades for each estimation domain were examined for high-grade outlier values. Top cutting was applied to various domains to reduce local high-grade bias due to very high-grade samples. Statistical observations, along with visualisation of mineralisation characteristics, were used to guide the selection of grade estimation technique. Ordinary kriging (OK) was considered for the mineralised domains. However, within broad mineralisation zones (defined at approximately 0.1 g/t gold and within geological boundaries), the grade architecture at the Timok deposits is gradational rather than mosaic, i.e. there is a transition between high grades and low grades, rather than extremely sharp contacts, where Multiple Indicator Kriging may be suitable. As such, estimation of Mineral Resources with the potential for economic extraction based on a selective mining unit (SMU) of 5 m x 5 m x 5 m was completed using Uniform Conditioning (UC). The UC estimate was further post-processed to produce single cell grades for each SMU, based on Localised Uniform Conditioning (LUC) where the grade tonnage of the panel gets reconstituted in SMU sized blocks resulting in a block model with single grades. The location of the high and low grades in each panel is an estimate based on the spatial distribution of high- and low-grade samples within the panel, but exact locations of the SMUs remain unknown. Experimental variograms were generated and modelled based on 1 m gold composites within the defined estimation domains. Traditional semi-variograms were used as the spatial model for this study, with variography completed using Supervisor software. The UC method required the estimation, by OK, of gold into 20 m x 20 m x 10 m panels, and gold into SMU- sized cells. Search ellipse orientations were consistent with local stratigraphic trends. Post-processing of the panel estimates was applied to account for change of support, and the kriged SMU estimates were used to guide the distribution of panel estimates into an SMU-sized cell. A review of the database of bulk density determinations showed that the variation in densities between lithological units and mineralised zones is low. In view of the large number of density values available, bulk

CSA Global Report Nº R302.2019 130 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

density estimates were interpolated into the cell models by inverse distance squared weighting, subject to stratigraphic search constraints. Procedures for classifying the reported “Mineral Resource”, “Inferred Mineral Resource” and “Indicated Mineral Resource” were undertaken under the guidelines adopted by the Canadian Institute of Mining, Metallurgy and Petroleum (CIM), as the CIM Definition Standards on Mineral Resources and Mineral Reserves. Mineral Resources estimated have been classified with consideration of the following criteria: • Quality and reliability of raw data (sampling, assaying, surveying). • Confidence in the geological interpretation. • Number, spacing, and orientation of drill-hole intercepts through mineralised zones. • Knowledge of grade continuities gained from observations and geostatistical analyses. • The likelihood of material meeting economic mining constraints over a range of reasonable future scenarios, and expectations of relatively low selectivity of mining. At each deposit, the level of confidence in the mineralisation varies between, and within, individual zones. The Bigar Hill, Korkan and Korkan West, and Kraku Pester deposits have each been classified as containing dominantly Indicated Mineral Resources with subsidiary Inferred Mineral Resources. CSA Global completed the following validation checks on the MRE: • Swath plots depicting model tonnes, input de-clustered composite gold grade, output block model gold grade and drill metres per slice for each domain of each deposit for the purposes of comparing input and output grades and trends. • On-screen visual comparisons of the block model grades (via LUC) for all domains. • Statistical comparison between the input composite grades and output model grades globally and for all domains. Results of the validations of the MRE supports the use of the resource model to underpin mine planning work, once constrained by a pit using appropriate parameters. The Mineral Resources are constrained within pit shells based on the parameters presented in Table 14-1. Table 14-1: Parameters used in pit optimisations Korkan Kraku Units Bigar Hill Korkan West Pester Waste $/t mined 2.39 2.58 2.39 2.45 Feed (oxide and transitional) $/t feed 2.39 2.58 2.39 2.45 Feed (sulphide) $/t feed 3.09 3.28 3.09 3.15 Mining cost 0.045 from 0.045 from 0.045 from 0.045 from Incremental cost per 10 m bench $/t mined 530 RL 560 RL 560 RL 480 RL Rehabilitation $/t mined 0.09 0.09 0.09 0.09 Costs Feed haulage from Kraku Pester $/t feed - - - 3.5 Processing and Feed (oxide and transitional) $/t feed 6.22 administration Feed (sulphide) $/t feed 12.81 Feed (oxide and transitional) $/tr oz 5 Total concentrate and smelter cost Off-site costs $/tr oz 200 (sulphide) Royalty % 5

CSA Global Report Nº R302.2019 131 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Korkan Kraku Units Bigar Hill Korkan West Pester Mining Mining recovery % 95.00 parameters Dilution % 0.00 Feed (oxide) % 91.3 91.5 72.8 72.8 Au processing Parameters Feed (transitional) % 69.3 69.3 69.3 69.3 recovery Feed (sulphide) % 70 65 65 50 Overall slope Oxide zone ° 45 angle Transitional and sulphide ° 52.5 Price of gold $/oz t 1,250 (RF=1). Pit shell at 1,400 Revenue Payable for oxide and transitional % 99 Payable for sulphide % 100.00 Discount rate % 7.50 Analysis Grams in a troy ounce g/oz t 31.1035 Processing rate Mt/a 2.0

Oxide and transitional mineralised material from the Timok Gold Project will be treated using conventional heap leaching technology. Additionally, the sulphide mineralised material will be processed by flotation to produce a saleable gold-bearing concentrate. A nominal production rate of 2 Mt/a has been assumed for treating all mineralised material types. The various mineralised material types (oxide/transitional/sulphide) are processed by different process technologies and metallurgical recoveries are dependent on the type of mineralisation. Only material classified as Indicated and Inferred Mineral Resources is considered to be processable. No Korkan East polymetallic mineralisation falls within the constraining conceptual pit shell. The price adopted for this study is $1,250/oz of gold as a pit shell with revenue factor=1; however, the pit shell generated at $1,400/oz has been selected as a constraining shell for reporting Mineral Resources. Pit optimisations run in Whittle software resulted in varying cut-off grades, dependent on oxidation state, per deposit (Table 14-2). Table 14-2: Mineral Resource reporting cut-off grades Cut-off grade (Au g/t)

Deposit Cut-off for Cut-off for Cut-off for Cut-off for Cut-off for Cut-off for oxide in oxide transitional in transitional sulphide in sulphide Whittle rounded Whittle rounded Whittle rounded Bigar Hill 0.178 0.20 0.235 0.25 0.603 0.60 Korkan 0.178 0.20 0.235 0.25 0.65 0.65 Korkan West 0.223 0.20 0.235 0.25 0.65 0.65 Kraku Pester 0.351 0.35 0.369 0.40 1.065 1.05

Mineral Resources are reported constrained within conceptual pit optimisation shells for each deposit, for the purposes of demonstrating “reasonable chances of eventual economic extraction”, required for Mineral Resource disclosure. Mineral Resources that are not Mineral Reserves do not have demonstrated economic viability. The open-pit shells have been determined via consideration of various cut-off grades for material types that were calculated based upon, among other things, the material type, haulage distance and recoveries derived from metallurgical testwork. The Mineral Resource statement for each deposit is presented in Table 14-3.

CSA Global Report Nº R302.2019 132 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Table 14-3: Mineral Resource estimate for Timok Gold Project as at 15 May 2018 Indicated Mineral Resource Inferred Mineral Resource Deposit Tonnage Au Tonnage Au (Mt) (g/t) koz (Mt) (g/t) koz Oxide 12.4 1.14 455 0.7 0.7 16 Transitional 5.9 1.21 229 0.4 1.0 12 Bigar Hill Sulphide 11.1 1.72 615 0.1 1.6 7 Subtotal 29.4 1.38 1,299 1.2 0.9 34 Oxide 5.8 0.90 166 0.2 0.5 4 Transitional 2.8 1.06 97 0.1 0.7 3 Korkan Sulphide 3.3 1.91 205 0.0 1.1 0 Subtotal 11.9 1.22 468 0.4 0.6 7 Oxide 2.9 1.03 98 1.0 0.8 24 Transitional 0.3 0.85 8 0.2 0.8 6 Korkan West Sulphide 0.0 1.33 1 0.0 0.9 0 Subtotal 3.2 1.02 106 1.2 0.8 31 Oxide 0.7 0.95 22 0.1 1.3 5 Transitional 0.1 0.95 4 0.0 1.2 0 Kraku Pester Sulphide 1.5 2.01 95 0.0 1.8 0 Subtotal 2.3 1.61 122 0.1 1.3 6 Total – Oxide 21.8 1.06 742 2.0 0.7 48 Total – Transitional 9.2 1.15 338 0.7 0.9 22 Total – Sulphide 15.9 1.79 916 0.2 1.5 8 GRAND TOTAL 46.9 1.32 1,996 2.9 0.8 78 Notes: • The effective date of the MREs is 15 May 2018. • Mineral Resources are reported in accordance with CIM guidelines. • A cut-off of 0.20 g/t Au for the oxide material, 0.25 g/t Au for the transitional material, and 0.60 g/t Au for the sulphide material is applied at Bigar Hill. • A cut-off of 0.20 g/t Au for the oxide material, 0.25 g/t Au for the transitional material, and 0.65 g/t Au for the sulphide material is applied at Korkan and Korkan West. • A cut-off of 0.35 g/t Au for the oxide material, 0.40 g/t Au for the transitional material, and 1.05 g/t Au for the sulphide material is applied at Kraku Pester. • Figures have been rounded to the appropriate level of precision for the reporting of Mineral Resources. • Due to rounding, some columns or rows may not compute exactly as shown. • The Mineral Resources are stated as in situ dry tonnes. All figures are in metric tonnes. • The models are reported above surfaces based on conceptual US$1,400 gold price pit shells to support assumptions relating to reasonable prospects of eventual economic extraction. • Mineral Resources that are not Mineral Reserves do not have demonstrated economic viability. The estimate of Mineral Resources may be materially affected by environmental, permitting, legal, title, taxation, socio-political, marketing, or other relevant issues.

14.3 Factors that may affect the Mineral Resource As of the Effective Date, the Qualified Person is not aware of any known environmental, permitting, legal, title, taxation, socio-economic, marketing, political, or other relevant issues that could potentially affect this MRE. The Mineral Resources could potentially be affected by future studies further assessing mining, processing, environmental, and other factors.

CSA Global Report Nº R302.2019 133 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Additional technical factors which may affect the MRE include: • DPM’s ability to obtain all required governmental approvals and permits for the possible development and operation of the Project • Metal price and valuation assumptions • Changes to the technical inputs used to estimate gold content (e.g. bulk density estimation, and grade model methodology) • Geological interpretation • Changes to mining assumptions • Changes to processing assumptions. There is insufficient information at this stage of study to assess the extent to which the Mineral Resources might be affected by these factors.

14.4 Previous Mineral Resource Estimates Mineral Resource estimation for the Bigar Hill, Korkan and Kraku Pester deposits was previously completed in March 2017, as shown in Table 14-4 (CSA Global, 2017). The key changes between the 2017 MRE for the Timok Gold Project deposits (CSA Global, 2017) and this updated Mineral Resource are: • Additional drilling since 2014 informing the 2018 MRE. • Updated interpretations by DPM of the different weathering domains, which was not previously recognised. • Reported at variable cut-offs, dependent on mineralised material types and deposit (Table 14-2), commensurate with differing costs parameters used to define the constrained pits). • The updated Mineral Resource also includes a maiden MRE for the Korkan West prospect of the Timok Gold Project, discovered by DPM in 2017. Due to the substantial changes listed above, it is not relevant to make a comparison to what was reported in 2017. Table 14-4: MREs as at 31 March 2017 – Timok Gold Project, Serbia, CSA Global Indicated Mineral Resource Inferred Mineral Resource Deposit Tonnage Au Tonnage Au (Mt) (g/t) Moz (Mt) (g/t) Moz Bigar Hill 22.97 1.57 1.16 0.3 1.5 0 Korkan 6.71 1.55 0.33 0 0.8 0 Kraku Pester 5.06 1.40 0.23 0.2 1.2 0 Total 34.74 1.54 1.72 0.4 1.4 0 Notes: • The effective date of the MRE is 31 March 2017. • CIM definitions were used for Mineral Resources. • A cut-off of 0.5 g/t Au is applied for Bigar Hill and Korkan. • A cut-off of 0.65 g/t Au is applied for Kraku Pester. • Figures have been rounded to the appropriate level of precision for the reporting of Resources. • Due to rounding, some columns or rows may not compute exactly as shown. • No Mineral Reserves have been estimated for the Bigar Hill, Korkan or Kraku Pester deposits. • Mineral Resources that are not Mineral Reserves do not have demonstrated economic viability. The estimate of Mineral Resources may be materially affected by environmental, permitting, legal, title, taxation, socio-political, marketing, or other relevant issues.

CSA Global Report Nº R302.2019 134 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Table 14-5: Pit optimisation parameters used to constrain the 2017 MRE Units Bigar Hill Korkan Kraku Pester Waste $/t mined $2.36 $2.55 $2.42 Feed $/t feed $3.06 $3.25 $3.12 Mining cost Rehabilitation $/t mined $0.09 $0.09 $0.09 Feed haulage from Kraku $/t feed n/a n/a $3.50 Costs Pester Processing and admin Mill processing costs $/t feed 12.29 12.29 12.29 Total concentrate and $/oz $200.00 Off-site concentrate smelter cost transport and smelter costs Royalty % 5% Mining recovery % 95.00% Mining parameters Dilution % 0.00% Parameters Processing recovery Au % 85% 85% 80% Weathered zone ° 45 Overall slope angle Partially weathered and fresh ° 52.5 Revenue Price of gold $/oz 1,250(RF=1). Pit shell at 1400 Discount rate % 7.50% Analysis Grams in a troy ounce 31.1035 Processing rate Mt/a 1.68

CSA Global Report Nº R302.2019 135 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

15 Mineral Reserve Estimates

DPM has not carried out any prefeasibility or feasibility studies of the Timok Gold Project designed to convert the Mineral Resources described in this report (Section 14) to Mineral Reserves.

CSA Global Report Nº R302.2019 136 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

16 Mining Methods

Mine design and planning for the Timok Gold Project are based on the Mineral Resource estimates as detailed in Section 14 of this report. Mine planning and optimisation results are based on Indicated and Inferred Mineral Resources for gold (Au). In the case of Timok the majority of potential feed material to the heap leach and processing facilities is classified as Indicated Mineral Resources (96%) with the reminder (4%) being Inferred Mineral Resources. This section outlines the parameters and procedures used to perform pit optimisation and subsequent mine planning work. Readers are cautioned that the PEA is preliminary in nature. It includes Inferred Mineral Resources considered too speculative geologically to have the economic considerations applied to them that would enable them to be categorised as Mineral Reserves, and there is no certainty the PEA will be realised. Mineral Resources that are not Mineral Reserves do not have demonstrated economic viability. The estimate of Mineral Resources may be materially affected by environmental, permitting, legal, title, socio-political, marketing or other relevant issues.

16.1 Overview The deposit is planned as a conventional truck-and-shovel operation. Heap leach material placement of approximately 2.5 Mt/a is planned over a 9-year mine life) with an additional 0.5 Mt/a of feed to a concentrator from Year 3 onwards. Pre-stripping activities start in Year -1, with production ramp-up in Year 1. The mine design and planning, cut-off grade reporting, and optimisation were completed using Hexagon’s MineSight software. Optimisation was performed using the MineSight Lerchs-Grossman algorithm to determine economic shells. The ultimate pit and phases were designed to develop the LOM plan. Table 16-1 shows the key results from the LOM plan. Table 16-1: LOM plan key results Description Unit Value Mineral Resource material mined Mt 18.9 Average Au grade g/t 1.36 Waste Mt 49.7 Strip ratio waste:feed 2.63 Heap leach pad placement rate Mt/a 2.5 Milling rate (Year 3+) Mt/a 0.5 Mine life (pre-strip and production) years 10

16.2 Geologic Model Importation The 2018 Mineral Resource models developed by CSA Global and summarised in Section 14 are used as the basis for the design work in the PEA study. The resource models were exported as a comma separated values files (CSV) and imported into Hexagon’s MineSight software for use by the mining team of AGP. There are three separate models covering the Bigar Hill, Korkan and Korkan West deposits. The files contained coordinates, rock types, oxidisation state, density, classification, geological rock description, and gold grade. The resource models are whole block models.

CSA Global Report Nº R302.2019 137 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

16.3 Geotechnical Pit wall slope recommendations were provided in a 4 December 2013 report entitled “Indicative Pit Slope Angles Assessment, Timok Gold Project, Serbia” by AMEC. Overall wall angles applied for pit optimisation and subsequent pit design rely on an interpretation of the preliminary geotechnical investigation undertaken by AMEC. The initial AMEC recommendations for Bigar Hill assumed a 230 m pit height but there is only a small section of the pit perimeter where the pit depth exceeds 200 m. For pit optimisation, uniform overall wall angles assume an average pit depth of 200 m in the partially weathered/fresh zone and a factor of safety of 1.2. For clarity, in the absence of more definitive geotechnical data, the following slope recommendations shown in Table 16-2 are adopted for all three pit areas for use in the pit shell generation and pit design work. Table 16-2: Recommended wall slope design parameters Deposit Weathered Zone Partially Weathered and Fresh Zones Bigar Hill 45° 52.5° Korkan 45° 52.5° Korkan West 45° 52.5°

16.4 Economic Pit Shell Development The open pit ultimate size and phasing requirements were determined with various input parameters including estimates of the expected mining, processing and G&A costs, metallurgical recoveries, pit slopes and reasonable long-term metal price assumptions. AGP worked together with DPM and CSA Global personnel to select appropriate operating cost parameters. The mining costs are estimates based on cost estimates for equipment from vendors. The costs represent what is expected as a blended cost over the life of the mine for all material types to the various dump locations. Process and G&A costs were estimated based on the expected plant throughput to suit the Mineral Resource. The recommended wall slope parameters were modified to consider the number of ramps that crossed the pit slope wall. An overall angle of 45° was used for the three pit areas. The Lerchs-Grossman parameters used for economic shell development are outlined in Table 16-3. Table 16-3: Open pit optimisation parameters Parameters Units Bigar Hill Korkan Korkan West Gold price $/oz 1,250 1,250 1,250 Payable – gold % 99 99 99 Transportation and refining charge – gold $/oz 6 6 6 Royalty % 5 5 5 $/oz 1,169.93 1,169.93 1,169.93 Net gold price $/g 37.61 37.61 37.61 Process information Oxide gold recovery % 91.3 91.5 72.8 Transitional gold recovery % 69.3 69.3 69.3 Sulphide gold recovery % 10.0 10.0 10.0 Process cost $/t 4.88 4.88 4.88 G&A cost $/ t 1.60 1.60 1.60 Total process cost $/t 6.48 6.48 6.48 Mining costs

CSA Global Report Nº R302.2019 138 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Waste – base rate $/t moved 2.14 2.07 2.19 Waste incremental cost – per 5 m bench $/t moved 0.02 0.00 0.03 Leach pad feed – base rate at surface $/t moved 2.41 2.67 2.79 Resource feed incremental cost – per 5 m bench $/t moved 0.01 0.01 0.01 Wall slopes – overall ° 45 45 45

All values are in US$ unless otherwise noted. The mining cost estimates are based on the use of 63-t trucks with the appropriate loading units and normal support equipment requirements. Waste material for Bigar Hill, Korkan and Korkan West are placed near the pits while not covering any known occurrences of mineralisation. The higher mill feed haul costs for Korkan and Korkan West reflects the longer haulage distance to the leach pad, which will be located to the east side of Bigar Hill. Pit shells were generated for the areas to examine sensitivity to metal pricing with revenue factors (RFs) of 0.5 to 1.1 (base case = $1,250/oz). This was to gain an understanding of the deposit and highlight potential opportunities in the design process to follow for phasing or sequencing. The net profit before capital was calculated on an undiscounted basis for each pit shell using the RF = 1.0 metal price. Mill feed tonnage, waste tonnage and net profit were plotted against the appropriate gold price for each RF. The graph for each model area is shown in Figure 16-1 to Figure 16-3. The graph indicated a linear increase in the shell size and net profit as the price increased. This is typical of a deposit(s) that have much greater potential as the profit has not levelled off. The various steps indicate the shells driving for deeper higher-grade material and the subsequent increase in waste stripping required. The graph is for the entire project but individual shells by area were examined and those different shells used for the detailed pit design work. The pit design basis shells used the RFs shown in Table 16-4. Table 16-4: Timok pit phase design basis RFs Design Area/Phase RF Gold price ($/oz) Notes Korkan West (Only 1 Phase) 0.9 $1,125 Korkan – Main Pit 1.0 $1,250 Korkan – East Pit 1.0 $1,250 Korkan – South Pit 1.0 $1,250 Bigar Hill – Phase 1 Pit 0.9 $1,125 Confined to northwest shoulder Bigar Hill – Phase 2 Pit 0.9 $1,125 Hanging wall designed to allow pushback width Bigar Hill – Phase 3 Pit 0.9 $1,125

CSA Global Report Nº R302.2019 139 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Figure 16-1: Korkan West sensitivities Source CSA Global 2019

Figure 16-2: Bigar Hill sensitivities Source CSA Global 2019

CSA Global Report Nº R302.2019 140 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Figure 16-3: Korkan pits sensitivities Source CSA Global 2019

16.5 Mine Planning

16.5.1 Mine Design The final pits were designed based on the economic factors used in the optimisation work, as shown in Table 16-5. Table 16-5: Open pit design parameters Design area Weathering zone Bench face angle (°) Bench height (m) Bench width (m) Inter-ramp angle (°) All Oxidised 55 15 8 45 Bigar Hill Fresh 75 15 8 52.5 Korkan Fresh 75 15 8 52.5 Korkan West Fresh 70 15 8 52.5

The pit phase designs for this study were completed using the RF pits described in Table 16-6 and following the design parameters shown in Table 16-7. Mining will be performed on a bench height of 5 m with safety bench placement every three benches. The ramp design width follows the guidelines set out in the Health, Safety and Reclamation Codes for Mines in British Columbia, some of the most stringent in the world, which calls for “a travel width where dual traffic exists of not less than three times, or where single lane traffic exists, of not less than two times the width of the widest haulage vehicle used on the road, and the shoulder barrier at least three quarters of the height of the largest tire on any vehicle hauling on the road.”

CSA Global Report Nº R302.2019 141 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

The current design uses a Komatsu HD605 (70-t) haul truck as the largest vehicle traveling on the ramp, with a truck width of 6.8 m. Tire size is based on the 27.00R49, with an overall height of 2.7 m. The calculated operating width for dual lane ramps is 22.7 m. These include a 1.5 m ditch for water runoff. Korkan is designed as three separate pit areas. The primary (largest) Korkan Pit 1 is in the northwest portion of the Korkan area. A smaller Korkan Pit 2 is east of Korkan Pit 1 and a smaller Korkan Pit 3 is located south of the previous pits. Korkan Pits 1, 2, and 3 are shown in Figure 16-4 to Figure 16-6. The relational layout of these pits is shown in Figure 16-7. The Korkan West pit is designed only as a single phase due to its size. The design is shown in Figure 16-8. Bigar Hill is the largest pit and is the only pit that has been designed in three phases. All three phases are designed using the 0.9 RF, $1,125 economic shell and limited by a physical area. Phase 1 is limited to the northwest lower slopes. Phase 2 also uses the 0.9 RF shell while leaving adequate mining width for the final phase, which completes the mining of the 0.9 RF shell. The phases have been shown in Figure 16-9, Figure 16-10 and Figure 16-11.

Figure 16-4: Korkan – Pit 1 design Source CSA Global 2019

CSA Global Report Nº R302.2019 142 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Figure 16-5: Korkan – Pit 2 design Source CSA Global 2019

Figure 16-6: Korkan – Pit 3 design Source CSA Global 2019

CSA Global Report Nº R302.2019 143 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Figure 16-7: Korkan pit layouts Source CSA Global 2019

Figure 16-8: Korkan West design Source CSA Global 2019

CSA Global Report Nº R302.2019 144 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Figure 16-9: Bigar Hill – Phase 1 design Source CSA Global 2019

CSA Global Report Nº R302.2019 145 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Figure 16-10: Bigar Hill – Phase 2 design Source CSA Global 2019

Figure 16-11: Bigar Hill – Phase 3 design Source CSA Global 2019

CSA Global Report Nº R302.2019 146 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

16.6 Mining Cut-Off Grades A calculation was completed for each pit area to determine the appropriate cut-off grade to use in reporting the Mineral Resources mined. There are normally two cut-off grades considered: • Mining cut-off grade: Includes all revenue calculations, and costs including the mining cost. • Processing or marginal cut-off grade: Includes all revenue calculations, and costs except the mining cost. The mining cut-off grade is used to determine the boundary of an economic pit shell and the processing cut- off has been used in this case for the determination of the resource contained within that same shell. The mining cut-off grade used for economic pit shell runs used a recovery of 10% for the sulphide material. The processing cut-off grade used a 25% recovery for the sulphide material as the concept of a small mill operation would be added around Year 3 to process this material. The resultant tailings would be filter pressed and placed on the leach pad for final disposal. For the Mineral Resources, the processing cut-off grade was considered as leach pad feed. If the value was less than this, the block was considered as waste. The pit area by pit area cut-off grades are illustrated in Table 16-6. Table 16-6: Processing cut-off grades used Oxide processing cut-off Transitional processing cut-off Sulphide processing cut-off Parameter grades grades grades (Au g/t) (Au g/t) (Au g/t) Bigar Hill 0.19 0.25 0.69 Korkan 0.19 0.25 0.69 Korkan West 0.24 0.25 0.69

16.7 Resource Loss and Dilution No additional estimate has been made for dilution due to the gradational nature of the deposit and the whole block basis of the model. A resource loss of 5% was included for the PEA during the mine scheduling. At this level of study, it was assumed that the current geological model incorporates some level of dilution. It is recommended to further refine mill feed loss and dilution at the next level of study.

16.8 Pit Phase Mill Feed Tonnages The leach pad feed tonnages and grades by phase using the processing cut-offs are shown in Table 16-7. Table 16-7: Feed by phase Indicated Inferred Total Waste Total Strip ratio tonnage tonnage (waste:feed) Design Area/Phase Feed Au Feed Au Feed Au tonnage (g/t) tonnage (g/t) tonnage (g/t) (Mt) (Mt) (Mt) (Mt) (Mt) Bigar Hill – Phase 1 2.8 1.00 0.2 0.72 3.01 0.98 2.1 5.1 0.69 Bigar Hill – Phase 2 2.7 1.56 0.0 1.00 2.68 1.56 8.3 11.0 3.09 Bigar Hill – Phase 3 5.3 1.86 0.1 0.74 5.33 1.85 24.4 29.7 4.59 Korkan Pit 1 3.6 1.11 0.1 0.56 3.70 1.09 8.3 12.0 2.22 Korkan Pit 2 1.3 1.20 1.30 1.20 2.0 3.3 1.50 Korkan Pit 3 0.4 0.75 0.40 0.75 0.9 1.3 2.62 Korkan West 2.1 1.16 0.4 0.91 2.48 1.12 3.6 6.1 1.47 Total 18.1 1.38 0.8 0.79 18.89 1.36 49.7 68.6 2.63

CSA Global Report Nº R302.2019 147 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

16.9 Mine Production Schedule The mining production schedule was developed based on a heap leach plant capacity of approximately 2.5 Mt/a and a sulphide processing plant capacity of 0.5 Mt/a. The Project life is 10 years, with one year of pre-stripping followed by nine years of operations. The pad placement throughput rate is assumed to start at the rated 2.5 Mt/a in Year 1 and the sulphide plant, with a throughput rate of 0.5 Mt/a, would start in Year 3. Table 16-8 and Figure 16-12 outline the mine production schedule by year, and Figure 16-13 outlines the heap and mill production and feed grade by year. Table 16-8: Mine production schedule Leach pad Au Mill tonnage Au Mine to Stockpile to leach Waste Total Year tonnage (kt) (g/t) (kt) (g/t) stockpile (kt) pad or mill (kt) (kt) material (kt) -1 - - - 3 - 1,011 1,014 1 2,500 1.19 60 3 8,028 10,584 2 2,500 1.03 202 - 8,679 11,381 3 2,500 1.26 461 1.82 - 262 7,534 10,233 4 2,500 1.49 500 2.23 154 - 11,303 14,456 5 2,500 1.17 500 2.62 771 - 6,935 10,707 6 2,500 0.98 500 2.87 581 - 5,929 9,510 7 418 0.90 500 2.58 - 490 241 669 8 500 1.52 - 500 9 516 1.51 - 516 Total 15,418 1.18 3,477 2.17 1,770 1,770 49,660 68,555

Figure 16-12: Mine tonnage by year and phase Source CSA Global 2019

CSA Global Report Nº R302.2019 148 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Figure 16-13: Heap and plant feed grade and recovered ounces Source CSA Global 2019

16.10 Mine Rock Management Over the LOM, the open pit will produce approximately 49.7 Mt (19 Mm3) of ex-pit waste rock that equates to a total of approximately 24 Mm3 of storage required. Waste rock storage will be in three locations; west of Korkan West, East of Bigar Hill and South of the Korkan pits. A portion of the Bigar Hill waste rock will be utilised to expand the plant site. The Bigar Hill and Korkan waste facilities have been designed with additional volumes to accommodate possible pit expansions. For this study, the waste rock has been assumed to be non-acid generating. This is to be confirmed by acid- base accounting testing and kinetic cells as the Project advances. The waste rock facilities are designed using the following parameters in Table 16-9. Table 16-9: Waste rock facility design parameters Parameters Value Waste dump face angle 37° (angle of repose) Bench lift 10 m Swell factor 30% Overall slope 26.6° (2:1 Horizontal: Vertical)

CSA Global Report Nº R302.2019 149 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Mine waste swell factor 30%

The proposed site layout including the waste rock facilities is shown in Figure 16-14.

Figure 16-14: Overall site layout with waste rock storage facilities Source CSA Global 2019 The capacities of the waste storage facilities are as follows: • Korkan rock facility – 8.2 Mm3 capacity. • Korkan West rock facility – 1.9 Mm3 capacity. • Bigar Hill rock facility – 25.5 Mm3 capacity. • Leach pad – 19.0 Mm3 capacity. The rock storage facilities have been designed to accommodate 2(H):1(V) slopes to be reclaimed concurrently with mining as available.

16.11 Mine Equipment This operation will be a conventional, open pit, truck-and-shovel operation. The equipment description in this section provides general information of the size and/or capacity of the selected equipment. Standard 63-t class trucks, 6.7 m3 class hydraulic shovels and a 13 m3 front-end wheel-loaders for open pit hauling and loading. Track-mounted rotary drill is proposed for blasthole drilling, capable of drilling 200 mm diameter

CSA Global Report Nº R302.2019 150 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

holes. Due to the size of the operation, all equipment on site will be diesel powered to provide greater mobility within the pit. The mine will operate 24 hours per day, 365 days per year. Table 16-10 lists major mine equipment requirements for the Project. Table 16-10: Major mine equipment requirements Equipment type Pre-production LOM Primary drill (PV271) 3 3 Hydraulic excavator – 6.7 m3 (PC1250) 1 2 Production loader – 13 m3 (WA900) 2 2 Haulage trucks (HD605 63 tonne) 10 12 Haulage trucks (HD405-7 36 tonne) for 2 3 pad

The mine support equipment will consist of: • Tracked dozer (D275AX-5) • Grader (GD655-6) • Support backhoe (PC490LC) • Water truck (Kenworth) • Tire manipulator (WA-500-7) • Lube/Fuel truck • Mechanics truck • Welding truck • Pump truck • Integrated tool carrier • Compactor • Lighting plants • Man bus • Pickup trucks • Ambulance • Fire truck • Rough terrain crane – 50 ton • Lowboy (75–100 ton) and tractor.

CSA Global Report Nº R302.2019 151 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

17 Recovery Methods

17.1 Heap Leach Facility

17.1.1 Site Layout The general arrangement of the mine site including the layout and position of the leach pad is shown on shown in Figure 16-14.

17.1.2 Heap Leach Pad Site Selection An initial high-level siting study has been undertaken for the PEA study. Two possible locations within the Timok project site were considered: one south of the Korkan pits in the area of the current KO dump (HL-01) and the second location east of the Bigar Hill pit and waste rock facility (HL-02). Locating a heap leach facility (HLF) within the confined valleys nearer to the Bigar Hill or Korkan open pits was not considered a valid engineering option. This is due to the complexity of constructing a lined system on the steeper and weathered slopes present adjacent to the pits. The engineering construction costs are considered prohibitive to the Project, and as such have not been considered further in this study. The required size for the Korkan pit site would require straddling two drainage channels necessitating significant preconstruction fill or twinning the ponds and pumping. In addition, with majority of the heap leach feed coming from the Bigar Hill and Korkan West pits, a significantly longer haul including elevation climb would have been required. Based on the high-level siting study the HL-02 location was selected as the preferred site for the leach pad. The HL-02 site can accommodate possible growth of the Project size, is closer to the major source of heap leach feed and has a single drainage point. In addition, this drainage has relatively flatter slopes as the subsequent lifts are placed on the pad.

Process Design Criteria The criteria used for the design of the processing circuit are summarised in Table 17-1. Table 17-1: Process design criteria Item Units Design criteria Annual tonnage processed Mt/a 2.5 Crushing production rate t/d 6,944 Crushing operation 12-hour shifts, 2 shifts/day, 6.5 days/week Crusher availability % 75.0 Crushed product size mm 80% passing 12.5 Feed type Oxide Transitional Primary leach cycle days 56 80 Secondary leach cycle days 70 120 Combined leach cycle days 126 200 Solution application rate L/m2/h 10 10

A three-stage crushing circuit has been selected that will nominally produce a product of 80% passing 12 mm, at the desired throughput.

CSA Global Report Nº R302.2019 152 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Results from compacted permeability tests and column leach tests on core samples tested to-date indicate there is no requirement for agglomeration.

17.1.3 Process Description Summary Processing of oxide and transitional feed types from the different Timok deposits will be carried out using heap leach technology. An overall process flow diagram is shown in Figure 17-1. Crushing is accomplished by a three-stage, closed-circuit crushing system operating 6.5 days per week, 24 hours per day, at a rate of 6,944 tonnes day. Feed to the run-of-mine (ROM) bin will predominantly be by direct truck tipping, with a front-end loader (FEL) to feed in material re-handled from the ROM stockpile. The crushing plant will also be configured to be able to produce gravel and stockpile sized product for use as leach pad drainage material. The final product from the crusher circuit discharges to a truck load-out hopper. The crushed material is loaded on to trucks using a FEL and hauled to the HLF. Prior to laying the irrigation pipework on the stacked feed the surface will be ripped with a dozer. The stacked crushed feed is leached using a drip irrigation system for solution application depending on water balance requirements. After percolating through the crushed feed, the gold-bearing solution drains to a pregnant leach solution tank where it is collected and pumped to an activated carbon Adsorption-Desorption- Recovery (ADR) plant. Pregnant solution is pumped to the plant where adsorption will take place in one train of five cascade columns. Barren discharge from the final columns flows by gravity to a barren tank and is then pumped to the heap for further leaching. High strength cyanide solution is injected into the barren solution to maintain the cyanide concentration in the leach solutions at the desired level. Desorption of gold and silver from loaded carbon utilises a pressurised elution column followed by recovery of gold and silver from pregnant eluate solutions in electrolytic cells containing stainless steel cathodes. Loaded cathodes are washed and the resulting precious metal sludge is smelted in a diesel-fired crucible to produce doré. Acid washing of carbon is done in a fibre-reinforced plastic vessel. Thermal regeneration of the carbon is carried out by a rotary kiln. An excess solution (stormwater) pond is included to contain any leach solutions and/or precipitation events that cannot be managed during normal operations.

CSA Global Report Nº R302.2019 153 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Figure 17-1: Overall process flow diagram Source CSA Global 2019

CSA Global Report Nº R302.2019 154 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Crushing and Stockpile The following modular components are included in the crushing facility: • ROM hopper bin. • A primary crushing plant with a vibrating grizzly, primary jaw crusher, and a crusher product conveyor. • Secondary and tertiary cone crushing stages operating in closed circuit with a double-deck vibrating screen. • A truck load-out hopper. • A belt weightometer, magnet, and metal detector; and associated transfer conveyors. Crushing is accomplished at the HLF by a three-stage, closed-circuit crushing system operating 6.5 days per week, 24 hours per day, at a rate of 6,944 t per day. Feed to the crusher will predominantly be by direct truck tipping, with feed of material re-handled from the ROM stockpile if required. Feed is fed from the hopper bin by a vibrating grizzly feeder. The grizzly oversize is fed to the C130 jaw crusher. The jaw crusher product and vibrating grizzly undersize are recombined on the primary crusher discharge conveyor and transferred to the secondary splitter feed conveyor, which feeds a splitter bin. A tramp metal electromagnet and metal detector are installed on this conveyor to protect the secondary and tertiary crushing stages. Crushed feed is conveyed to the double deck vibrating screen deck. Material from the top deck is fed to the to the secondary cone crusher, whilst material from the bottom deck is fed to the tertiary cone crusher. The secondary and tertiary cone crusher discharge onto a transfer conveyor and are then conveyed back to the screen operating in closed circuit. Screen undersize is final product and falls directly onto the final product conveyor. The final plant product is 100% passing 16 mm (approximately 80% passing 12.5 mm) and is conveyed to a truck load-out hopper. A modular motor control centre is housed in a separate room or container and is located proximal to the crushing area. A crusher operator control cabin is mounted above the motor control centre. All the conveyors are interlocked, so that if one conveyor trips out, all upstream conveyors and the vibrating grizzly feeder will also trip. This interlocking will prevent large spills and equipment damage. Both these features are considered necessary to meet the design utilisation for the system. The crushing circuit described above is shown as Figure 17-2.

CSA Global Report Nº R302.2019 155 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Figure 17-2: Crushing circuit configuration Source Metso 2018 Key crushing circuit design criteria are presented in Table 17-2.

CSA Global Report Nº R302.2019 156 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Table 17-2: Key crushing circuit design criteria Item Units Design criteria Annual tonnage processed Mt/a 2.5 Crushing production rate t/day 6,944 Crushing operation 12-hour shifts, 2 shifts per day, 6.5 days per week Crusher availability % 75 Design throughput rate t/hr 460 Primary crushing Crusher type Jaw Model C130 Crusher size mm x mm 1,000 x 1,300

Crushed feed size (P80) mm 370 Closed side setting mm 110

Particle size (P80) mm 78 Screening Screen type Double Deck Vibrating Motor size kW 10 Screen size m x m 3.0 x 7.3 Top deck 32 Bottom deck 13 Secondary crushing Crusher type Cone Model HP5 Motor size kW 370 Closed side setting mm 35

Product P80 mm 32 Tertiary crushing Crusher type Cone Model HP5 Motor size kW 370 Closed side setting mm 13

Product P80 mm 12.5

Truck Load-Out Hopper The truck load-out hopper has been designed with an approximate capacity of 500 t, or a live volume of 377 m3. Crushed feed from the load-out hopper is discharged via an automatic knife-gate control valve into the trucks and hauled to the heap leach pad.

17.1.4 Heap Leach Facility Design Criteria The proposed HLF is a typical single use pad located within relatively flat terrain. The planned deposition rates for the HLF is 2.5 Mt/a at a dry density of 1.6 t/m3. The LOM capacity of the HLF is 11.5 Mm3 over nine years for a total tonnage of 22.7 Mt.

CSA Global Report Nº R302.2019 157 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

17.1.5 Heap Leach Facility Design Approach Design Features The proposed HLF is a Valley Leach Fill design located to the east of the Bigar Hill deposit, as shown in Figure 17-3. Due to the relatively short mine life of nine years, the full heap leach pad area will be constructed in a phased approach. The pad footprint will be graded to create gradual uniform fall towards the decant location. The entire pad base will be lined with a multi-layer mining system including a leak detection system and engineered sub-base and protected by an over-liner drainage layer. The pad will be surrounded by fully lined perimeter containment bunds to impound the pregnant solution and prevent flooding during storm events. The containment bunds will be constructed from locally sourced material. An overflow spillway to the downstream emergency pond will be located 1.0 m below the crest of the containment bund to allow for the decant of excess solution from the heap during periods of high rainfall and/or extended plant breakdowns.

Pregnant Solution Pond The Pregnant Solution Pond will be a pond integrated with the Valley Leach Fill. The pregnant solution will be funnelled via the above-liner drainage system towards a drainage collection ditch on the downstream western edge of the pad which will flow towards the decant outlet pipe located on the western side of the pad. The pregnant solution will be pumped to the pregnant solution pond tank located at the ADR plant. The perimeter drainage system will allow full drain-down of the pregnant solution to the plant. Valve control on the outlet pipe is to be located within the plant area. An additional outlet (redundant) pipe to serve as backup for the primary outlet system to be considered during future design stages. The pregnant solution pipes will be located within bunded containment channels.

Event Pond The event pond comprises a cut-to-fill fully lined pond including engineered containment embankments. The emergency pond is sized to accommodate excess solution for the heap during periods of high rainfall and/or extended plant breakdowns. The lining system includes a single high-density polyethylene (HDPE) and leak detection system. The emergency pond will be decanted by a submersible pump and pipeline to the process plant. The emergency pond outflow is completely separate from the pregnant solution pond outlet to allow for improved control at the plant. The emergency pond will have a nominal 0.5 m freeboard and an emergency spillway, which will discharge into the downstream environment in the event the capacity is exceeded.

Flood Diversions Measures The aim of the stormwater management system at the HLF is to prevent overflows to the downstream environment whilst maintaining the physical integrity of the containment structures. Surface water at the HLF will be managed by keeping catchment flows separate from incident precipitation.

CSA Global Report Nº R302.2019 158 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Due to the topography and climatic conditions of the site, surface water is considered to be minimal with the exception of rare flash flood events. The diversion channels will intercept clean stormwater flowing in the wadi system and divert around the HLF before release downstream of the facility. Due to the relatively flat laying topography of the site, the flood protection measures are enhanced by the relatively high containment bunds around the HLF. Erosion protection measures on upstream faces of the perimeter containment bunds will be required comprising grading of materials on site and rock riprap.

Heap Leach Facility Placement Method The ROM material will be trucked from the open pit to the ROM pad; once crushed the material is loaded onto trucks via a truck load-out system and hauled to the HLF. Prior to laying the irrigation pipework on the stacked material, it will be ripped with a dozer. Heap feed material will be placed from the western edge of the pad eastward (i.e. upstream direction).

17.1.6 Heap Leach Facility Design General Layout The HLF pad required a total volume of 10 Mm3 of material (current design is 11.5 Mm3) is approximately 750 m x 950 m resulting in a total pad area of approximately 66 ha with 8 m high lifts. The general HLF layout is shown on Drawing H18089-C-001 as Figure 17-3. The lift volumes and phasing are summarised in Table 17-3 below. The facility is capable of handling both the heap leach feed as well as storage of the sulphide tailings. Table 17-3: HLF general criteria Volume Tonnage Plan area Lift height Heap elevation Production life Lift (Mm3) (Mt) (ha) (m) (masl) (years) 1 0.07 0.15 0.1 8 688 0.1 2 0.3 0.7 3.4 8 696 0.2 3 0.6 1.1 7 8 704 0.5 4 1.0 1.9 12 8 712 0.8 5 1.4 2.9 18 8 720 1.0 6 2.2 4.4 27 8 728 1.5 7 2.7 5.4 34 8 736 1.8 8 3.2 6.4 40 8 744 3.2 Total 11.5 22.7 66 64 744 9

CSA Global Report Nº R302.2019 159 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Figure 17-3: HLF general arrangement Source CSA Global 2019

CSA Global Report Nº R302.2019 160 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Other Structures The other structures considered with the study design include: • Containment embankments. • Drainage decant ditch. • Lining systems. • Over-liner and solution collection system. • Emergency storm pond. • Flood diversion channel.

17.1.7 Solution Application and Leaching Leach Concept Due to the leach curves exhibiting fast initial leach kinetics, and a long leach tail, a three-pond system will be adopted for solution application and collection. The proposed leach solution irrigation concept is shown in Figure 17-4.

Figure 17-4: Leach solution irrigation schematic diagram Source CSA Global 2019 Following stacking the crushed material is ripped and is then irrigated with barren leach solution and the resulting gold and silver-bearing solutions are collected into the pregnant solution pond.

CSA Global Report Nº R302.2019 161 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Due to the climate conditions at the Timok site, the stacked material is leached with drip tubes spaced 1 m apart. Dripper tubes are also used on side-slopes. Reusable PVC pipes are used to distribute the solutions to the dripper tubes on top of the heap. To reduce the potential for scaling problems within the irrigation system, it is necessary to continuously add an anti-scalant polymer to the leach solutions. The primary leach cycle of 28 days has been designed into the crushed material leach system. The leach time is based upon metallurgical testwork. Leach solutions are applied to the feed at a nominal application rate of 10 L/m2/hr with a cyanide concentration of 250 ppm to the heap. To reduce cyanide consumption, high concentration cyanide solution is injected directly into the suction side of the barren pumps using metering pumps. This allows for accurate control of cyanide concentration and greatly reduces loss due to natural degradation in the circuit. Key heap leaching design criteria are presented in Table 17-4. Table 17-4: Key heap leaching design criteria Item Units Design criteria Irrigation method - Drippers Solution application rate L/m2/hr 10 Irrigation rate – nominal m3/hr 600 Primary leach cycle days 28 120 Total leach cycle days 112 200 Flux rate primary leach ts/to 1.1 1.6 Total flux rate ts/to 2.0 3.6 Irrigation solution pH - 10.5–11.0 Irrigation solution NaCN concentration mg/L 250–3 00 Average PLS grade Au ppm 2.0 0.6

Leach System Description Double suction pumps (one operating, one standby) at the barren tank are used for barren solution application to the heap leach. These pumps are mounted beside the barren tank and are high-volume, high- head to pump to the HLF. High-strength cyanide and an anti-scalant agent are added to the suction side of the barren leach solution pumps by metering pumps. The nominal flowrate of barren solution is 600 m3/hr with a concentration of 250 ppm cyanide. A steel and HDPE header-pipe from the barren tank pumps supplies the solution to the active irrigation areas on the leach pad. A totalizing magnetic flow meter and continuous drip solution sampler are installed on the leach solution header for metallurgical balance calculations. The leach solution header is installed along the north side of the leach pad area. Valved tees at the header supply leach solution to risers that distribute solution to the top of the stacked material at the active leach- cells. Dripper emitters will be placed with 1 m intervals. The dripper lines connect directly to the main header pipe. Pressure gauges are included on each main header pipeline with valves to enable pressure control. Extra drippers provide solution to the side slopes.

CSA Global Report Nº R302.2019 162 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Gold and silver bearing solutions draining from the leach pad are collected at the bottom of the stack by a network of perforated drainage pipes within a gravel layer and are directed to the pregnant solution pond (tank). Installed submersible pumps in the ponds are used for solution transfer. The pumps are mounted on slides on the pond sidewalls to facilitate placement and extraction of the pumps in the ponds. Additional rough- textured protective liner panels and conveyor belting are installed on the pond sidewalls in the area where the pumps are located to protect the pond liner.

Heap Leach Pad The heap leach pad capacity is approximately 11.5 Mm3 (22.7 Mt). An under-drain system consisting of perforated pipes is installed below the low permeability soil (clay) liner to collect and convey any near surface underground water below the pad. In addition, the under-drains act as an early leak detection system that collects any solution that may leak through the composite liner system and allows it to be captured and pumped back to the circuit. The leach pad consists of a composite liner system utilising 300 mm of compacted clay underlying a 1.5 mm low-density polyethylene (LDPE) welded liner (geomembrane). A 600 mm layer of sized gravel over-liner is placed on the top of the geomembrane to protect the liner and act as a basal drainage layer. Perforated collection pipes are embedded in the gravel layer to enhance solution drainage and provide a rapid return of pregnant solution after it has passed through the stacked material. The piping and collection layer also minimise the depth of solution (head) over the liner system. The collected solution is directed to the pregnant solution pond and then pumped to the carbon ADR process plant for metal extraction.

Solution Storage The solution containment and storage system include the following facilities: • Barren solution tank. • Intermediate solution tank. • Pregnant solution pond. • Event solution pond. The barren and intermediate solution tanks are each 10 m diameter x 6 m high carbon steel tanks and have been sized to provide 45 minutes of storage capacity at the design flow rate of 500 m3/hr. The pregnant and intermediate solution tanks are the same size. The event pond has been sized to ensure that all the leach solutions can be managed in a controlled manner to prevent any unplanned discharges of solution. The storm water pond has been sized to hold the 24 hours of heap drain-down and the 24-hour/100-year storm event over the entire lined area (a total of ~11.5 Mm3 minimum). The event pond liners utilise a double 1.5 mm HDPE liner system on top of 300 mm of compacted clay soil. Leak detection is provided by GeoNet sandwiched between the two HDPE liners on top of a low permeability soil liner and a collection system to detect any solution between the liners in the event there is leakage through the primary liner.

CSA Global Report Nº R302.2019 163 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

There is a second similar leak detection and collection system installed between the bottom HDPE liner and the compacted clay liner. This type of double-redundancy liner and leak detection system significantly reduces the possibility of solution entering the environment below the pond. The leak detection systems are checked and logged for solution each shift during operations. Heavy rain events will result in solution being diverted to the excess solution pond. Stormwater solution in the excess pond is returned to the barren tank as make-up solution as soon as practical. All ponds have a freeboard of 1 m (including a safety berm of 500 mm) around the perimeter of the ponds.

Solution Management The event pond solution collection and storage system has been designed to accommodate and manage seasonal variations of solution from the process, as well as stormwater surges from the leach pad and waste dump areas. This pond system (HDPE double-lined) provides the necessary capacity to store the process solutions during normal operations. The event pond is equipped with interconnecting pumping and piping/canal systems to allow for orderly transfer of water or solution to and from the reservoirs and to the points of use in the HLF. Solution management for the system is generally simple. The excess solution pond should normally be maintained at empty or low levels whenever possible. When solution is diverted to the excess solution pond, it should be pumped back to the leach system as soon as practical. Every effort should be made to avoid storing excess solution over a long period of time.

17.1.8 Adsorption-Desorption-Recovery Plant An overall process flow diagram for the HLF, solution storage and ADR plant are shown in Figure 17-5.

CSA Global Report Nº R302.2019 164 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Figure 17-5: Overall HLF/ADR flow diagram Source CSA Global 2019

CSA Global Report Nº R302.2019 165 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Carbon-in-Column Adsorption The carbon-in-column (CIC) adsorption section of the ADR facility consists of two CIC trains with five cascade type open-top, up-flow, mild-steel carbon adsorption columns. Each of the carbon columns are nominally 2.90 m in diameter x 2.90 m high, with a capacity of 3 t of activated carbon, designed for a solution flow in the range of 500–600 m3/hr. Pregnant leach solution (PLS) is pumped directly to the adsorption circuit using a submersible pump on a pump slide in the pregnant pond. The PLS will be sampled and will pass through a trash screen prior to being distributed to the CIC circuit. The solution will flow by gravity from the first to the fifth contactor of each line. Discharge from the last contactor will report to the carbon safety screen and will be sampled again prior to being pumped to the barren solution tank for recirculation on the heap. Oversize from the carbon safety screen will be recovered in a tote bin. Anti-scalant is added to the pump suctions to prevent scaling of the carbon that can affect carbon loading. Barren solution exiting the last carbon adsorption column flows through a screen to separate and capture any floating carbon from the solution. A magnetic flow meter equipped with a totalizer measures solution flow to the carbon columns. Pregnant and barren solution continuous samplers are installed at the feed and discharge end of the carbon column train. These are used to measure feed and barren solution gold and silver concentrations. Additional sample points for solution and carbon samples are provided on each column to monitor adsorption efficiency and gold/silver loading profiles. Adsorption of gold and silver from pregnant leach solution is a continuous process. Periodically, the carbon contained in the lead column in the series becomes loaded with gold and silver and must be advanced to the acid wash and desorption circuit. Loaded carbon is transferred as a batch from the lead column to the acid wash column using pumps. Carbon in the remaining columns is then advanced, one at a time, and a batch of new (or stripped/regenerated) carbon is transferred into the final empty column using pumps. Drive solution for the carbon transfer pumps comes from the barren tank process solution pump. Loaded carbon transfers approximately every 24 hours and desorption occur approximately four to seven times per week during normal operation. Key adsorption circuit design criteria are presented in Table 17-5. Table 17-5: Key adsorption design criteria Item Units Design criteria Number of trains no. 2 Number of contactors no. 5 Nominal flow rate m3/hr 500 Design flow rate m3/hr 600 Carbon hold-up per column T 3 Combined Au + Ag carbon loading g/t 2,500 Combined Au + Ag barren carbon grade g/t 100

CSA Global Report Nº R302.2019 166 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Carbon Acid Wash Acid washing consists of circulating a dilute acid solution through the bed of carbon to dissolve and remove scale from the carbon. Acid washing the carbon may be done either before or after each desorption cycle. The process is performed on a batch basis. After carbon is transferred into the acid wash column, but before any acid is introduced, fresh water is circulated through the bed of carbon to remove any entrained caustic cyanide solution. This rinse solution is pumped to the pregnant solution pond with the acid wash circulation pump. A dilute acid solution is then prepared in the mix tank, and circulation is established between the acid wash vessel and the acid mix tank. Concentrated acid is injected into the recycle stream to achieve and maintain a pH ranging from 1.0 to 2.0. Completion of the cycle is indicated when the pH stabilises between 1.0 and 2.0 without acid addition for a minimum of one full hour of circulation. After acid washing has been completed, the acid wash pump will pump spent acid solution from the acid mix tank and wash vessel directly to the pregnant pond. The carbon is then rinsed with raw water followed by rinsing with dilute caustic solution to remove any residual acid. Total time required for acid washing a 3-t batch of carbon is four to six hours. After acid washing is complete, a carbon transfer pump will transfer the carbon to the desorption section.

Desorption A Zadra pressure elution, hot caustic desorption circuit has been selected. This type of circuit requires 24 hours or less to complete a cycle and for this reason each strip batch is sized for 3 t of carbon. Each desorption cycle requires the transfer of a 4-t batch from the adsorption circuit to the strip vessel. The desorption circuit is sized to elute, or “strip,” the gold from a 4-t batch of carbon into pregnant eluate solution. During the elution cycle, gold and silver are continuously extracted by electrowinning from the pregnant eluate concurrently with desorption. A complete desorption cycle will require approximately 18 hours. After a batch of carbon has been transferred to the elution vessel, barren strip solution (eluent) containing sodium hydroxide and sodium cyanide is pumped through the heat recovery and primary heat exchangers, and introduced to the elution vessel at a temperature of 135°C and a nominal operating pressure of approximately 340 kPa (50 psig). Final stripped-carbon gold and silver content is typically less than 160 g/t of carbon. Under normal operating conditions, barren eluent solution from the solution storage tank will pass through the heat recovery exchanger to be preheated by hot pregnant eluate leaving the elution column. The barren eluent solution then passes through the primary heat exchanger to raise the temperature up to 135°C using pressurised hot water (~180°C) from the boiler system. The elution column contains three internal stainless-steel inlet screens to hold carbon in the column and to distribute incoming stripping solution evenly in the column. Pregnant eluate solution leaving the elution column passes through two external stainless-steel screens before passing the cooling heat exchanger to reduce the eluate temperature to about 75°C. The cooled pregnant eluate solution is sent to the electrowinning cells. After desorption is complete, the stripped carbon is pumped to dewatering screens to remove water and carbon fines, and then transferred to carbon regeneration or to the carbon storage tank.

CSA Global Report Nº R302.2019 167 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Electrowinning The electrowinning circuit is operated in series with the elution circuit. Solution is pumped continuously from the barren eluent tank through the elution vessel, then through the electrowinning cells, and back to the barren eluent tank in a continuous closed loop process. The gold and silver-laden solution exiting the elution column is filtered to trap any carbon escaping from the column; it then passes through the heat recovery exchanger and the cooling exchanger to reduce the solution temperature to 75°C and flows to the electrowinning circuit. Gold and silver are won from the eluate in the electrowinning cells using stainless steel cathodes and a current density of approximately 50 amperes per square metre of anode surface. Caustic soda (sodium hydroxide) in the eluate solution acts as an electrolyte to encourage free flow of electrons and promote the precious metal winning from solution. To keep the electrical resistance of the solution low during desorption and the electrowinning cycle, make-up caustic soda must sometimes be added to the barren eluent tank. Barren eluent solution leaving the electrolytic cells is pumped back to the eluate storage tank for recycle through the elution column. Periodically, all or part of the barren eluent is dumped to the pregnant solution pond and new solution is added to the tank. Typically, about one-third of the barren eluent is discarded after each elution or strip cycle. Sodium hydroxide and sodium cyanide are added as required from the reagent handling systems to the barren eluent tank during fresh solution make-up. The precious metal-laden cathodes in the electrolytic cells are removed about once per week and processed to produce the final doré product. Loaded cathodes are transferred to a cathode wash box where precipitated precious metals are removed from the cathodes with a pressure washer. The resulting sludge is pumped to a plate-and-frame filter press to remove water and the filter cake is loaded into pans for mercury retorting.

Regeneration Thermal regeneration consists of drying the carbon thoroughly and heating it to approximately 750°C for 10 minutes. It is expected that thermal reactivation every third adsorption cycle will maintain suitable carbon activity. The 3-t carbon batch to be thermally reactivated is dewatered on a vibrating screen, transferred to the regeneration kiln feed hopper and fed to the regeneration kiln by a screw feeder. Hot, regenerated carbon leaving the kiln falls into a water-filled quench tank for cooling and storage. Ultimately, quenched regenerated carbon is pumped to the carbon storage dewatering screen to remove any fines and the coarse carbon will be returned to the adsorption circuit. Carbon fines passing the regeneration feed dewatering screen and/or the carbon storage dewatering screen are stored in the carbon fines tank and are periodically filtered in a carbon fines filter press. The carbon fines recovered in the filter are stored in drums and the filtrate solution is sent to the barren tank.

Smelting Cathode sludge from the filter press is dried and mixed with fluxes and is then fed to a tilting diesel fired furnace. After melting, slag is poured off into 100 kg capacity cast iron moulds until the remaining molten furnace charge is mostly molten metal (doré). Doré is poured off into 20 kg bar moulds, cooled, cleaned, and

CSA Global Report Nº R302.2019 168 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

stored in a vault pending shipment to a third-party refiner. The doré poured from the furnace will represent the final product of the processing circuit. Periodically, slag produced from the smelting operation is re-smelted on a batch basis to recover residual precious metal content. Reprocessed slag will be placed on the HLF. A hood collects the furnace fumes, which will pass through a bag house to remove particulates, then through an induced draft fan. The system will be designed to remove over 99.5% of the particulates present in the exhaust fumes.

Reagent Mixing and Handling The reagent handling system includes equipment used to mix and store sodium cyanide and sodium hydroxide batches, to add both these reagents to the barren eluent solution tank, caustic to the acid wash circuit, and to add cyanide to the barren leach solution system. Reagent mixing and storage are at ambient temperature and pressure. The major equipment required to perform these tasks includes: • Sodium cyanide mix tank with agitator and bag breaker system. • Sodium cyanide transfer pump. • Sodium cyanide storage tank. • Sodium cyanide dosing pumps. • Sodium hydroxide mix/storage tank with agitator. • Sodium hydroxide transfer pump. • Sodium hydroxide metering pump (for acid wash circuit). • Anti-scalant dosing pump.

Sodium Cyanide Solid, sodium cyanide briquettes are delivered to the site in 1-t bulk bags. Raw water or barren solution is used to partially fill the cyanide mix tank and a small amount of sodium hydroxide (pumped from the caustic storage tank) is added to the tank to make a 0.5% NaOH solution prior to the addition of sodium cyanide briquettes. The caustic addition will ensure that proper alkaline pH is maintained, thereby minimising waste of cyanide by dissociation and possible generation of toxic HCN gas. An electric hoist is used to lift the sacks to the top of the cyanide mix tank. A bag breaker system is mounted above the mix tank to discharge cyanide briquettes into the mix tank. The tank is designed to contain and dissolve solid sodium cyanide briquettes and yield a solution containing 25% (by weight) sodium cyanide. Distribution of the high-strength cyanide solution is by metering pumps to points of use at the barren solution pump suctions. If needed in the elution cycle, cyanide solution for the barren eluent tank is pumped from the cyanide storage tank using a separate metering pump. All cyanide distribution lines will be double-containment, either by pipe-within-pipe or pipe-over-liner containment systems.

Sodium Hydroxide Sodium hydroxide (caustic) solution is prepared in an agitated caustic mix tank. Sodium hydroxide is delivered to the site in small sacks (25–50 kg). Raw water or barren solution will be used to fill the mix tank and solid

CSA Global Report Nº R302.2019 169 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

sodium hydroxide will be manually added to the tank for dissolution. The tank is designed to contain and dissolve solid NaOH flakes, or pearls, and to yield a solution containing 20% (by weight) sodium hydroxide. For elution, concentrated caustic solution is pumped from the mix tank to the eluent storage tank where it is mixed with raw water to produce a 1% (by weight) sodium hydroxide eluate solution. The estimated consumption of sodium hydroxide is about 160 kg per elution cycle assuming replacing one-third of the eluate solution batch each cycle. Fresh sodium hydroxide solution for barren eluent make-up is pumped from the caustic mix tank directly into the eluent storage tank. For cyanide mixing, concentrated caustic solution is pumped to the cyanide mix tank where it is mixed with barren solution (or raw water) to produce a 0.5% (by weight) solution. The estimated consumption of sodium hydroxide is about 140 kg per cyanide mix batch. For carbon acid wash neutralisation, concentrated caustic solution is pumped to the acid wash tank where it is mixed with raw water and circulated through the acid wash column. The estimated consumption of sodium hydroxide is about 8 kg per 4-t carbon acid wash batch.

Activated Carbon The carbon handling system will include equipment used to transfer, store and to add/handle carbon. The major equipment required to perform these tasks includes: • Carbon transfer pumps for carbon transfer • Carbon dewatering screens • Carbon fines filter press • Electric hoist for carbon bulk bag handling. Carbon transfer pumps transfer carbon between the various unit operations in the recovery plant. Loss of carbon in the processing circuit is expected to average approximately 2 t per minute. The carbon column train has one carbon transfer pump to transfer carbon between columns and to transport carbon to the elution column. Each carbon column will be valved to a suction and discharge manifold on the pump. “Push” water for carbon transfer will be barren solution pumped from the barren solution pump to the recovery plant manifold. New carbon is added to the circuit after pre-soaking in water. A carbon bulk bag, which contains about 0.5 t of virgin carbon, will be loaded into a soaking vessel by an electric hoist. Fresh water is added to the vessel, and the carbon will be soaked for 24 hours. Once the soaking process is complete, the carbon slurry is pumped to a vibrating dewatering screen for removal of fines. The oversize carbon from this screen drops directly into the carbon quench tank and the carbon is ready for transfer to the adsorption columns. The undersize from the screen drains to the carbon fines tank and is filtered into dry cakes in a filter press and stored in drums or supersacks. Carbon leaving the acid wash circuit passes over a carbon dewatering screen or the regeneration feed dewatering screen. Water and fine carbon passing either screen is sent to the carbon fines filter to recover the fine carbon. Clear solution leaving the filter will be discharged to the excess solution pond or to the barren solution tank. New/washed/regenerated carbon stored in the carbon quench tank is pumped to the carbon trains to replace carbon removed for elution or to replace carbon losses.

CSA Global Report Nº R302.2019 170 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Process Solution and Make-Up Water Process solution is required in the heap leach recovery plant for reagent make-up, wash down, filter cleaning, and other uses. The process solution requirements are met by a separate pipeline from the barren tank. Based on the heap water balance, only minor amounts of fresh raw water will be required in the recovery plant.

17.1.9 Process Reagents and Consumables Usage and Storage Requirements A contained and roofed reagent storage area is included in the process area.

Cyanide The cyanide system includes equipment used to mix, store and distribute sodium cyanide solutions to the agglomeration, heap leach, and elution systems. Reagent mixing and storage is at ambient temperature and pressure. The major equipment required to perform these tasks includes: • Cyanide bag handling and bag breaker systems • Cyanide agitated mix tank • Cyanide transfer pump • Cyanide storage tank • Cyanide metering pumps. Sodium cyanide is delivered as briquettes in 1,000 kg bulk bags inside plywood crates. The sodium cyanide briquettes are dissolved using barren solution in a carbon steel tank equipped with an agitator. A 25% sodium cyanide solution concentration (maximum) is prepared in this manner. After dissolution, the cyanide solution is transferred to a storage tank with a 1.5-batch storage capacity. Usage requirements indicate that one 2-t batch should be mixed every shift. Distribution of the high-strength cyanide solution is by individual metering pumps to points of use at the barren solution pumps, elution circuit, and agglomeration solution pump. All cyanide distribution lines will be double-containment: either pipe-within-pipe or pipe-over-liner systems. A one-month reserve supply of dry cyanide inventory should be kept on-site, in case of supply interruptions, and is to be stored in a secure fenced and roofed area.

Carbon Activated carbon is used to adsorb the precious metal values from the leach solution in the adsorption columns. Make-up carbon is 6 x 12 mesh. Carbon is delivered in 500 kg supersacks. Carbon transfer pumps transfer carbon between the various unit operations in the recovery plant. New carbon is added to the circuit after pre-soaking. The new carbon requirement to replace fine carbon losses is projected to be 3% of the weight of carbon stripped in the elution section.

Hydrochloric Acid Hydrochloric acid (HCl) is used in the acid wash section of the elution circuit prior to return to the adsorption columns, or transfer to the carbon regeneration kiln.

CSA Global Report Nº R302.2019 171 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Hydrochloric acid (approximately 32% by weight in water) is delivered in 200-litre drums or 1,000-litre tote bins. Acid washing consists of circulating a dilute acid solution through the bed of carbon to dissolve and remove scale from the carbon. Acid washing the carbon is usually done after each desorption cycle.

Diesel Fuel Diesel fuel is required for the elution boiler, carbon regeneration kiln, and the smelting furnace. Approximately 20,000 litres (20 m3) of diesel fuel will be consumed in the process area each month.

Anti-Scalant Anti-scalant agents are used to prevent the build-up of scale in the process solution and heap irrigation lines. Anti-scalant agent is normally added to the process pump intakes, or directly into pipelines, and consumption varies depending on the concentration of scale-forming species in the process stream. Delivery is in liquid form in 1 m3 (1 t) bulk containers. Anti-scalant is added directly from the supplier bulk containers into the pregnant and barren pumping systems using variable speed, chemical-metering pumps. On average, anti-scalant consumption is expected to be about 6 kg per 1,000 m3 (6 ppm) of process solution to be treated (pregnant and barren) which equates to 120 kg per day. The recommended minimum inventory should be 10 tote bins.

Fluxes Various fluxes are used in the smelting process to remove impurities from the bullion in the form of a glass slag. The normal flux components are a mix of silica sand, borax, and sodium carbonate (soda ash). The flux mix composition is variable and is adjusted to meet individual project smelting needs: fluorspar and/or potassium nitrate (nitre) are sometimes added to the mix. Dry fluxes will be delivered in 25 kg or 50 kg bags. Average consumption of fluxes is estimated to be 0.075 kg per troy ounce of gold.

Hydrogen Peroxide The amount of hydrogen peroxide required is strongly dependent on the cyanide concentration of the solution to be treated and discharged (5.6 g H2O2 / 1 ppm cyanide). Typically, the solution to be discharged is contaminated stormwater that has been exposed to sunlight while stored in the excess solution pond for some time. Ultraviolet radiation significantly decomposes cyanide fairly rapidly. Under these conditions, the cyanide levels in the excess solution are normally only a few parts per million, and dosage rates are typically in the range of 0.25 kg hydrogen peroxide per cubic metre of solution treated.

Copper Sulphate Copper sulphate is used as a catalyst with the hydrogen peroxide to speed the cyanide destruction process. Only a very small addition is required, and some solutions contain enough copper naturally to eliminate the need for adding the copper sulphate. Provisionally, this small system and reagent inventory is included although it may not be necessary.

Sulphuric Acid Sulphuric acid may be needed during detoxification and discharge to lower pH sufficiently to meet discharge standards. The amount to be added can vary significantly depending on the pH of the solution to be treated.

CSA Global Report Nº R302.2019 172 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

The pH is mostly dictated by how diluted, or how contaminated, the stormwater actually is. Strict solution management procedures can greatly influence the level of contamination of stormwaters stored in the excess solution pond. Typical dosage rates at similar operations are in the range of 0.65 kg sulphuric acid per cubic metre of solution treated.

Crusher Liners Crusher liners require replacement periodically due to abrasive wear. Replacements of liners are expected approximately every four to five months and a spare set for all crushers should be kept in inventory onsite.

17.2 Sulphide Concentrator

17.2.1 Design Basis This section provides design factors and site information used in the design of the Timok Gold Project processing plant and associated facilities for a 0.5 Mt/a bulk sulphide concentrator.

17.2.2 Process Design Criteria The design criteria form the basis for the design of the sulphide mill feed processing facility and the required site services. Given the early stage of the Project’s development, limited testwork data was available to support the design. Database information, vendor advice and assumptions based on experience have been used in lieu of project- specific criteria. The criteria allow for the definition of a preliminary mass balance, as well as the design and specification of equipment for the derivation of the Project capital cost estimate. In addition, it allows for the development of operating cost requirements such as power, water and reagents, included within Section 21 (Capital and Operating Costs).

17.2.3 Production Criteria The concentrator is designed to process 0.5 Mt/a of mill feed over the LOM following ramp-up in Year 3. Being essentially a pyrite concentrator containing gold values, sulphur feed grades are expected to largely dictate concentrate production rates but, for selection of equipment, a final concentrate mass pull of 5.5% has been adopted. The operating regime has been set at 8,000 hr/a, which is typical for a plant of this level of complexity and size. This sets the nominal throughput at 62 dry t/hr. Plant feed starts in Year 3 with Bigar Hill pit the primary source until Year 7. At that time Korkan will provide the bulk of the material until the end of the mine life. ROM pad blending will assist with producing a mill feed stream that contains a sustainable envelope of sulphur grades at target gold grades to achieve smooth plant operation. Recoveries of sulphur and gold to flotation concentrate have been estimated from the available testwork for the Bigar Hill, and Korkan feed sources, as described in Section 13.7. Further investigations will be undertaken to improve gold recovery from all feed sources as the Project develops. Key production criteria are provided in Table 17-6.

CSA Global Report Nº R302.2019 173 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Table 17-6: Key production criteria (concentrator) Criteria Units Value Annual processing capacity Mt/a 0.5 Operating time Primary crushing h/a 6,570 Concentrator h/a 8,000 Nominal processing rate Primary crushing t/h 76 Concentrator t/h 62 Concentrate production Nominal % feed 4.0 Design % feed 5.0 Concentrate sulphur grade (min) %TS 30

17.2.4 Metallurgical Design Criteria Table 17-7 summarises key metallurgical design criteria used for the PEA study. Table 17-7: Key metallurgical criteria Criteria Units Value Flotation Cell type Woodgrove staged flotation reactor Circuit configuration - Recovery control stage Rougher + scavenger

- Grade control stage Cleaner 1 + Cleaner 2 + Cleaner-scavenger Pulp densities - Rougher feed % solids w/w 30 - Cleaner 1 feed % solids w/w 20 - Cleaner 2 feed % solids w/w 20 - Cleaner 1 and 2 concentrates % solids w/w 40 All other concentrates % solids w/w 30 Concentrate mass pulls - Rougher % of new feed 25.1 - Rougher-scavenger % of new feed 8.7 - Cleaner 1 % of new feed 2.2 - Cleaner-scavenger % of new feed 4.3 - Cleaner 2 % of new feed 3.3 Concentrate handling Concentrate thickening - Thickener type High-rate - Specific settling rate t/m2.h 0.51 - Underflow pulp density % w/w 55

17.2.5 Unit Process Consumables Table 17-8 summarises process consumables adopted for the Timok Gold Project PEA process plant design. Values are a combination of criteria supplied by DPM from testwork interpretation (flotation reagents), calculated from specific energy consumption (grinding media) and typical values from the AMEC database (for thickeners).

CSA Global Report Nº R302.2019 174 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Table 17-8: Unit process consumables Criteria Units Value Grinding media - SAG mill kg/t new feed 0.26 - Ball mill kg/t new feed 0.55 - Stirred media detritors kg/t new feed 0.28 Flotation reagents - Frother MIBC g/t new feed 50 - Dispersant (sodium silicate) g/t new feed 150 - Activator (copper sulphate) g/t new feed 100 - Collector (potassium amyl xanthate g/t new feed 105 - Promoter (Aerofloat 404) g/t new feed 105 Flocculant - Concentrate thickener g/t thickener feed 50 - Tailings thickener g/t thickener feed 50

17.2.6 Process Flowsheet and Plant Description Figure 17-6 provides a schematic representation of the proposed Project concentrator adopted for the PEA. The following sections describe the process plant design by facility area.

Primary Crushing Selection Basis The low strength of the feed favours the use of primary crushing to minimise dust generation to feed a semi- autogenous grinding (SAG) mill, rather than adopting multi-stage crushing and ball milling as is more common with the older Serbian operations. No crushability data was available for sizing the jaw crusher, but the degree of fracturing noted in drill core indicates low resistance to impact breakage, and a suitable crushability value was selected for modelling purposes. An apron feeder was selected as the primary feeder as it can handle the impact of large rocks falling into the ROM bin. The vibrating grizzly feeder is used to scalp out fines ahead of the jaw crusher to reduce the risk of packing within the jaw cavity and to reduce wear.

CSA Global Report Nº R302.2019 175 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Tails stored on Heap Leach Facility

Figure 17-6: Concentrator schematic Source CSA Global 2019

CSA Global Report Nº R302.2019 176 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Description ROM feed is crushed to a size suitable for feeding to the SAG mill. The crushing circuit consists of a bin to receive ROM feed from the mine and/or stockpiles, a feeder system, jaw crusher and a conveying system to transfer feed to the primary crushed feed (ore) stockpile (COS). ROM feed is delivered to the ROM bin by FEL from ROM stockpiles for blending purposes. The ROM bin is fitted with a 700 mm aperture static grizzly to prevent the ingress of oversize feed material. A mobile rock breaker will be utilised to manage ROM oversized material. Feed material is withdrawn from the ROM bin by an apron feeder and discharges onto a vibrating grizzly feeder. Oversize from the grizzly feeder discharges directly to the jaw crusher. Jaw crusher product, together with spillage from the apron feeder and undersize from the grizzly feeder, discharge onto the primary crusher discharge conveyor. Crushed feed is transferred from the primary crusher discharge conveyor onto the stockpile feed conveyor en route to the COS. The primary crusher discharge conveyor is fitted with a weightometer to monitor the ROM feed throughput rate. The conveyor is also fitted with a cross-belt magnet to protect the intermediate conveyors from potential damage due to tramp metal ingress. Magnetic objects captured by the magnet are discarded to the trash bin, which is emptied periodically. Dust extraction to a centralised crushing area bag house is used at all transfer points in the crushing area to minimise fugitive dust emissions. Spillage from the primary crushing area is collected by the primary crushing area sump pump and the dust scrubber area clean-up sump pump before being pumped to the dust scrubber area clean-up sump in the crushed feed storage and reclaim area.

Primary Stockpile and Reclaim Selection Basis A live stockpile was selected over a crushed feed silo primarily because of the perceived capital cost impost of the latter if fabricated out of steel. The alternative of using a concrete silo was not examined for this design but could be evaluated in future project stages as formwork is cheaper than fabricated steelwork in the Balkans region. Sixteen hours live capacity (average milling circuit withdrawal rate) was made on the basis of a re-fit of the jaw crusher taking 12 to 14 hours only, which provides a measure of leeway to maintain mill production without resorting to the use of a bulldozer to push in the non-live areas of the COS. Description Primary crushed feed material is conveyed to a conical stockpile which provides 16 hours of mill feed when full. Up to 48 hours additional capacity is available during extended crusher outages by using a dozer to push the crushed feed into the reclaim slots. A reclaim system beneath the crushed feed stockpile consisting of three reclaim hoppers with apron feeders (two duty/one standby) to transfer feed to the SAG mill via the stockpile reclaim conveyor. The stockpile reclaim conveyor is equipped with a weightometer located before the recycle feed conveyor discharge point to measure the net SAG mill feed rate without recycled pebbles.

CSA Global Report Nº R302.2019 177 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

SAG mill grinding media is loaded onto the stockpile reclaim conveyor via the SAG mill ball loading chute. Dust extraction is used at all transfer points around the COS and reclaim area to minimise fugitive dust emissions. Dust generated is collected in the reclaim area dust scrubber. Spillage from the crushed feed storage and reclaim area is collected by the reclaim area sump pump and the dust scrubber area clean-up sump pump before being pumped to the SAG mill discharge hopper.

Grinding and Classification Selection Basis Flotation testwork indicates that it is necessary to grind the feed to a size of 80% passing 20 μm to achieve acceptable flotation recovery of gold values due to the gold deportment described in Section 13. Limited testwork was available to evaluate fine grinding technologies, aside from basic stirred media detritor (SMD) energy values from the completed laboratory testwork. Grinding Circuit Options A trade-off study was undertaken to evaluate the capital and operating costs for various combinations of SAG mills with ball mills, Vertimills and SMDs in order to determine the optimum combination from an economic perspective. The arrangement described in the following section represents the optimum configuration selected from the trade-off study. Description Primary crushed feed is conveyed to a high aspect SAG mill that operates in closed circuit with a vibrating screen deck. Screen oversize pebbles are transferred via a series of conveyors back to the SAG mill feed conveyor for additional grinding. Screen undersize slurry gravitates to a pump hopper, which is shared with the secondary ball mill. The contents of the pump hopper are pumped to a hydrocyclone cluster to produce a cyclone overflow stream (with particle size 80% passing 90 μm) and a coarse cyclone underflow stream, which is returned to the secondary ball mill for further grinding. Ball mill discharge flows down a launder into the common mill discharge hopper. Cyclone overflow slurry is pumped over a trash screen to remove organic and inorganic waste material that may otherwise cause blockages in the fine grinding circuit. Trash screen undersize flows to the first of three SMDs. The primary SMD unit uses 5 mm diameter ceramic media to grind the feed from 80% passing 90 μm to an intermediate size, which is then evenly distributed to two secondary SMD units. These units impart additional grinding via a finer ceramic media size of 2 mm diameter and are also used to condition the flotation activator (copper sulphate) to the freshly exposed mineral surfaces. The discharge from both secondary SMD units gravitates to a common pump hopper, which transfers the slurry to the flotation circuit. Dilution water is added to the hopper to control flotation feed at 30% solids w/w and copper sulphate solution is dosed to activate pyrite ahead of flotation. Spillage within the primary grinding and fine grinding areas is collected in sumps and pumped back to the SAG mill discharge hopper as this circuit has greater flexibility to cope with tramp material that may report tom these sumps and also the intermittent flows of dilute slurry from the sump pumps.

Flotation Selection Basis Liberated free gold and pyrite hosted gold is recovered to a concentrate using flotation techniques. The use of Woodgrove Technologies Inc.’s (Woodgrove) staged flotation reactors (SFRs) has been assumed for the

CSA Global Report Nº R302.2019 178 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

flotation circuit. Woodgrove claims superior ability to increase the flotation kinetics of naturally slow floating minerals more quickly through high shear collision within the particle collection units (PCUs), to increase the recovery of valuable minerals via the subsequent modified column flotation cell (BDU/FRU) which also feature under-froth wash systems to disperse adhering gangue minerals and potentially increase valuable metal grades. Due to the height of the units, SFRs require 200 kPa flotation air delivery pressure at the manifold, hence low- pressure compressors are used instead of conventional blowers. Tank Cells Alternative A trade-off study was undertaken after the initial design with SFRs was completed to determine if conventional tank cells would be more cost effective than SFRs, from the perspectives of capital cost, operating cost and overall comparative net present cost (NPC). The analysis indicated that tank cells offered a lower direct capital cost, but higher operating costs due to the higher power draw of the tank cell rotors compared to SFRs. The tank cell option returned a slightly lower NPC than the SFR base case. This modest incremental benefit offers little incentive to move away from SFRs at this time, hence, SFRs were retained for the PEA study flotation circuit design. Description Diluted SMD discharge slurry is pumped to the rougher flotation conditioning tank and dosed with collector (PAX), dispersant (sodium silicate) and promoter (Aero 404), and the residence time is five minutes. Recycled rougher-scavenger concentrate and cleaner-scavenger tailings slurry also enters this tank for re-conditioning with reagents. Slurry discharges from the rougher conditioning tank to the rougher flotation SFR units, which produce rougher concentrate that is transferred to the cleaner flotation circuit. Rougher tailings slurry is pumped to the head of the Rougher-scavenger flotation bank. Rougher-scavenger concentrate slurry is returned to the rougher flotation conditioning tank whilst rougher-scavenger tailings slurry is pumped to the tailing’s thickener. Frother is dosed to each cell down the two banks of roughing SFR units to aid recovery of gold and pyrite. Rougher concentrate at 30% solids (w/w) is diluted to 20% solids ahead of the Cleaner 1 bank. Cleaner 1 concentrate is pumped directly to the concentrate thickener whilst Cleaner 1 tailings slurry discharges to the Cleaner-Scavenger bank. Cleaner-scavenger concentrate is pumped to the Cleaner 2 bank for further upgrading of gold and sulphur values whilst cleaner-scavenger tailings are pumped back to the rougher flotation conditioning tank. The Cleaner 2 bank upgrades the cleaner-scavenger concentrate to a sufficiently high level of sulphur (minimum 30% S grade). Cleaner 2 concentrate is pumped to the concentrate thickener whilst the Cleaner 1 tailings stream is recycled to the head of the Cleaner-scavenger bank for additional treatment. Spillage from flotation circuit is collected via sump pumps and is generally pumped back to the SMD circuit product discharge hopper as it can more easily accept the dilution of the spillage slurry better than the rougher conditioning tank. Compressed air is injected into the PCU of each SFR unit to produce bubbles for mineral recovery.

CSA Global Report Nº R302.2019 179 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Concentrate Dewatering and Handling Selection Basis For the PEA, no Project testwork data was available for the sizing of the concentrate thickener and pressure filter unit. In lieu of actual data, database values for similar concentrate types and particle size were used for sizing purposes. Vendors provided some of their own database information on moisture content in concentrates, which assisted with establishing the design moisture content of 10% w/w. Description Flocculant diluted with concentrate thickener overflow water via an in-line mixer is added to the combined concentrate slurry to assist with aggregation and settling of the concentrate particles within the concentrate thickener. Flocculant dilution may also be achieved with process water if necessary. The overflow of the concentrate thickener gravitates to the concentrate thickener overflow tank and is transferred to the process water tank for reuse in the concentrator. The dewatered underflow of the concentrate thickener is pumped to the concentrate filter feed tank which has 24 hours storage capacity ahead of the filter unit. Slurry is pumped into the filter press in batch cycles (approximately 12 minutes duration) where water is squeezed out of the cakes by high pressure air which compresses the membranes. High pressure water can also be used for more difficult duties, but it requires a separate membrane squeeze water delivery system. After cake squeezing, air is blown through the cakes to remove additional water. Filtrate is returned to the concentrate thickener to allow eventual collection of any fine concentrate particles. Allowance for washing the cakes with raw water has been made if necessary. Filter cake drops out of the bottom of the filter unit into a surge chute which regulates the discharge of cake via the transfer conveyor to the load-out conveyor for bagging. At this stage, the concentrate is loaded into bags, weighed and tied automatically. Bags are removed from the loading unit by forklift, awaiting loading onto trucks for transport to the processing facility. Spillage from the concentrate thickener and filtration areas is pumped back to the concentrate thickener via an area sump pump.

Tailings Disposal Selection Basis As with the concentrate thickener, no Project sulphide feed testwork data was available for the sizing of the flotation tailings thickener. In lieu of actual data, database values for similar material types and particle size were used for sizing purposes. The design has made provision for an emergency power supply to pump raw water down the line to flush out solids to the Heap Leach Facility where tailings will be stored. Description Flotation tailings slurry is pumped to the feed launder of the flotation tailings thickener. Flocculant diluted with flotation tailings thickener overflow water is added to the flotation tailings slurry promoting aggregation and settling of the particles. As with the concentrate thickener, an in-line static mixer in the line feeding diluted flocculant to the thickener assists with mixing of the flocculant and the thickener overflow water. Flocculant dilution may also be achieved with process water in the event the thickener overflow water turbidity increases excessively.

CSA Global Report Nº R302.2019 180 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

The overflow of the flotation tailings thickener gravitates to the flotation tailings thickener overflow tank and is pumped to the process water tank for reuse in the concentrator. The thickened flotation tailings are then pumped to the filter feed stock tank. Thickened tailings are then filtered in two plate-frame filter presses to a moisture content of approximately 15–20%. Filtered material is then conveyed to a stockpile, whereby it is subsequently loaded into trucks and dumped on a barren part of the HLF.

Plant Services and Reagents Plant Services Plant and instrument air are supplied as a vendor package. Air from compressors will be filtered, dried, filtered again and sent to instrument air receivers before distribution for use as instrument air. Plant air is not dried or filtered and is taken directly from the plant air receiver. Low pressure compressor air is required for the flotation circuit which will be supplied by dedicated duty/standby 200 kPa units. Diesel is unloaded from tankers to the plant diesel storage tank from where it is transferred to the light vehicle refuelling stations. The mine service diesel storage tank receives its own supply of diesel from delivery trucks due to the mine workshop’s remoteness from the process plant. Reagents Reagents used in the concentrator are distributed to the relevant sections of the processing plant from the reagents area. Some reagents are received in a pre-prepared, ready-for-use condition, whilst other reagents require preparation before use. All reagents requiring preparation prior to distribution into the various sections of each plant are prepared in batches, while distribution is continuous and determined by the feed rate into each section. Collector – potassium amyl xanthate (PAX): • PAX is used in the recovery of pyrite during flotation. PAX dosing solution is prepared on site by mixing dry powder delivered in drums with raw water in an agitated mixing tank fitted with an extraction system to purge potentially explosive xanthogen gas that may be generated during the mixing process. The mixed PAX solution (at 10% w/w strength) is pumped to a distribution tank and then dosed continuously to the flotation circuit at the desired rate. Activator – copper sulphate: • Copper sulphate is used to alter the surface chemistry of tarnished sulphides (pyrite). It also activates pyrrhotite and sphalerite if they are present, improving floatability. Copper sulphate solution is prepared on site by mixing dry powder delivered in bags with raw water in an agitated mixing tank. The resultant 15% (w/w) strength solution is pumped to a distribution tank and is dosed continuously to flotation feed at the desired rate. Dispersant – sodium silicate: • Sodium silicate is used to disperse fine non-sulphide gangue minerals from the froth recovery zone of the flotation cells, improving the concentrate grade and effective froth carrying capacity of the cell. It is delivered in bulk tanks at 40% w/w strength and transferred to a day tank for dosing to the rougher flotation conditioning tank without further dilution.

CSA Global Report Nº R302.2019 181 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Frother – metyl isobutyl carbinol (MIBC): • MIBC is an agent added to stabilise bubbles generated in the flotation cells so that adhering gold and pyrite particles can be recovered before the bubbles burst. • MIBC is supplied neat in IBC containers and is transferred by drum pump batch-wise to a day tank for dosing to the flotation cells without any dilution. Promoter – Aero 404: • Aero 404 (mercaptobenzothiazole) is a specialised promoter that works synergistically with PAX collector in the recovery of gold and pyrite, particularly in acidic pulp systems or when pyrite is slightly oxidised. • Aero 404 is supplied neat in IBC containers and is transferred by drum pump batch wise to a day tank for dosing to the flotation circuit without any dilution.

CSA Global Report Nº R302.2019 182 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

18 Project Infrastructure

18.1 Overall Site The overall site plan is shown in Figure 16-14 and includes the major facilities of the Project. These include pits at Korkan, Korkan West and Bigar Hill, process facility, heap leach pad and waste dumps. The undulating terrain was utilised for the waste dump storage and heap leach pad but limited the locations for the plant. The plant site and associated buildings have been placed at a minimum of 500 m from the proposed open pits due to vibration concerns. Grid power will be provided from the power line located to the southeast of the Project.

18.2 Roads Site roads will be developed from the plant site to the three pit areas and to the HLF. The main access road for the Project will come from the south along an existing track. This will be upgraded for Project use.

18.3 Waste Storage Facilities There are three waste storage facilities proposed for the site; Korkan dump, Korkan West dump and Bigar Hill dump. These will be constructed in a wrap-around fashion along the contours. The three waste dumps can accommodate the estimated 49.7 Mt of waste currently in the schedule. The material is likely to below grade oxide or transitional material with a high carbonate content. The carbonate is likely to act as buffer and, as such, acid rock drainage (ARD) issues are not anticipated. Testwork is recommended to support this assumption as part of the PFS.

18.4 Heap Leach Facility The HLF is located in the same valley as the Bigar Hill waste dump. This valley was chosen due to its proximity to Bigar Hill, which represents 58% of the estimated total feed material. Material from the mining of Bigar Hill will be used in the construction of the facility in addition to sourcing of material in the valley itself. The HLF has been designed with a liner system involving a polymer liner and a clay liner together with a leak detection system, typical of current design standards. A cross section through the facility is shown in Figure 18-1. The HLF will also be built in a sequence of lifts. The initial development will set the groundwork; then, annual lifts will raise the liner to sufficient height to provide adequate space for the next season’s construction period. The overall facility is shown in Figure 17-3 with a cross section of the lift sequence as envisaged for the PEA. The HLF will also be used as a tailings storage facility. The sulphide plant that starts in Year 3 will generate a modest quantity of filtered tailings. This material will be placed on the heap leach pad along the liner or in the back end to avoid disturbing the regular leaching cycle. This has the advantage of storing the dewatered tailings in a lined facility. Due to the higher carbonate nature of the feed material, ARD is not considered to be an issue, but it was felt that placing the material on the lined facility mitigated any risk associated with ARD. The HLF has been designed to accommodate the full quantity of heap feed and sulphide tailings (18.9 Mt).

CSA Global Report Nº R302.2019 183 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Figure 18-1: HLF – cross section Source CSA Global 2019

CSA Global Report Nº R302.2019 184 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

18.5 Water Management and Supply Potable and fresh water is expected to be provided by pump-wells. Additional water for plant operations will come from reclaim water via the filtered tailings from the sulphide plant.

18.6 Fuel Supply Diesel fuel will be stored on site, near the mine services area, for heavy and light vehicle refuelling. Diesel fuel storage and supply will be provided by a fuel supplier and will include fuel storage tanks, offloading pumps, dispensing pumps, associated piping, and electronic fuel control/tracking.

18.7 Camps and Buildings No camp facilities are required for this project due to its proximity to neighbouring communities. Employees will be bused to work to minimise traffic on local roads. Administration personnel will be based in the local town, or will be located in buildings near the plant together with plant operations and maintenance personnel. Mine operations, geology and engineering will be together in the truck shop building.

18.8 Power and Electrical Power will be brought to the site from the 35 kV line to the southeast of the Project. This line will be 5 km long and end at the site in the transformer and switchgear area.

CSA Global Report Nº R302.2019 185 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

19 Market Studies and Contracts

As of the Effective Date of this report, DPM has not requisitioned any market studies or entered into any marketing contracts pertaining to the Timok Gold Project. Gold is the primary expected product of the operation and is readily marketable. Gold pyrite concentrates, such as those expected to be produced at the flotation plant, are also readily marketable to local and global smelters. The US$1,250 per troy ounce gold price utilised in the PEA is based on the unadjusted three-year average of the London Gold Market of US$1,259 per troy ounce as of 2 February 2019. Estimated costs for gold smelting and refining included the following: Gold Concentrate Terms • Smelting charge $120 / dmt • Gold deductions 4% • Refining cost $8 / payable Oz • Transportation $10 / dmt Gold Bullion Terms • Gold refining losses 1% • Treatment charge $6 / payable Oz Co-author and Qualified Person, Mr Alex Veresezan, has reviewed the gold price, US-Canadian exchange rates and gold and silver smelting and refining (including transportation and insurance) costs and confirms that the values used are adequate for the purposes used in the PEA and technical report. As of the Effective Date of this report, no contracts material to DPM that are required for property development have been entered into. Contracts with respect to development, transportation, refining and sales will require negotiation in due course.

CSA Global Report Nº R302.2019 186 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

20 Environmental Studies, Permitting and Social or Community Impact

1.1 Overview The Timok gold mine is located in a rural, hilly area with steep valleys, characterised by seasonally grazed pastures, woodlands and isolated farms and houses. There are no designated protected areas for biodiversity or cultural heritage in the Project footprint. Much of the area is underlain by limestone where a karst has developed leading to caverns, springs and sinkholes. DPM has placed the key mine facilities outside known karst-affected areas. DPM will use heap-leach technology to extract gold from coarse-crushed oxide mineralised material. The crushed oxide mineralised material will be supplemented by the tailings generated by the small-scale grinding-flotation plant, used for sulphide mineralised material extracted, to maximise gold recovery. The mixed crushed oxide mineralised material and sulphide tailings will be treated together on a single large heap- leach pad. Overall, the DPM approach presents several advantages in minimising environmental and social impacts, compared to other approaches: • Relatively low footprint, comprising pits, waste rock dumps, small plant area, and the heap leach; which will be rehabilitated at the end of the mine life. • No need for a separate tailings facility. • Low energy demand. • Low water discharge. The transportation and use of cyanide in the heap leach process presents potential risks to surface and groundwater quality. DPM is a signatory to the International Cyanide Management Code1, which provides standards of practice for protection of communities and the environment during transportation of cyanide and specific usage requirements on handling, storage, operation, disposal and decommissioning. The leach pad has been sited away from the karstic limestone zone, where infiltration to groundwater could more readily occur, and leach pad design includes a robust liner system that collects the gold solution and also prevents infiltration to groundwater. The details of how water will be managed will be developed at PFS stage and will include water supply and maximisation of water efficiency in the process. It is currently anticipated that process water supply will come from groundwater resources, including pit dewatering. Contact and non-contact waters will be managed separately, and the discharge to surface water will be minimised. Where discharge is required, it will be managed carefully to maintain appropriate social and ecological water quality and flow limits. The design will be for no discharge or seepage to groundwater. A hydrological study of watercourses and details of planned water management will be submitted as part of the application for water conditions, a key stage in permitting requirements in Serbia.

20.1.1 Regulatory Constraints The Project has developed a permitting schedule to comply with the Serbian system and to be in line with international good practice. The Serbian system is changing and maturing to adopt European Union (EU) requirements and there is a need to consult with regulatory authorities regularly to anticipate and manage

1 https://www.cyanidecode.org

CSA Global Report Nº R302.2019 187 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

future changes to the system. The spatial planning process is critical to the schedule. Flexibility is needed in the permitting plan to allow for uncertainties related to the evolving regulatory system, also noting that there is as yet no precedent for a private sector mining project of this type completing the permitting process, albeit there are two major mining projects progressing through the permitting system ahead of the Timok Gold Project.

20.1.2 Environmental and Social Constraints None of the environmental and social risks identified at this stage are likely to prevent the Project progressing. Key risks are similar to those associated with other gold mining projects and include safeguarding rivers, groundwater and biodiversity, as risks associated with acquiring land. Completion of the baseline survey program is required to fully understand and manage these risks. The program includes: • Conducting surface and groundwater monitoring. • Completing terrestrial and aquatic biodiversity assessments. • Completing cultural heritage baseline data collection (caves, buried archaeology, historic farms and mills, intangible heritage, particularly for certain minority ethnic groups). • Conducting air quality and noise monitoring. • Determining acid-generating potential of rocks and soils to be disturbed. • Identifying local users of water. • Conducting a social baseline and census and socio-economic survey for landowners and those who will be economically displaced by the Project.

20.1.3 Social Licence The Project has fostered good relationships with local stakeholders since 2007 and is undertaking active engagement with stakeholders in line with its stakeholder engagement and communications plan. Continued engagement with regulators, directly affected communities and other interested parties before and as part of formal engagement processes will be key to maintaining the Project’s social licence to operate. The Project design is being developed with environmental and social constraints in mind. Consideration of project alternatives to minimise impact will be key to minimising costs associated with mitigation. Environmental and social impacts will be fully considered as part of the environmental impact assessment process, in line with Serbian regulations and international guidance. DPM has an existing corporate responsibility policy and sustainability framework under which the plans, programs, procedures can be deployed to address environment, social and health and safety risks.

20.2 Permitting

20.2.1 Permitting Requirements The Serbian regulatory and permitting system requires a range of permits and permissions to be issued for mining projects. Mining structures are permitted under the Law on Mining and Geological Explorations whilst supporting and auxiliary structures such as roads andadministrative buildings are separately permitted under the Law on Planning and Construction. There is a range of other approvals and permissions required under ministries including the Ministry of Agriculture, Forestry and Water Management and Ministry of Environmental Protection.

CSA Global Report Nº R302.2019 188 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

The permitting process can be broadly divided into exploration, pre-construction, construction, operation and closure phases (see Figure 20-1). Table 20-1 provides a list of key permits required for the Project by phase. There is no formal permit for mine closure. There is a requirement to include a design for closure in the main mine design, develop a closure plan for the mine waste management plan and consider the environmental impacts of closure as part of the environmental impact assessment. These approvals must be received prior to construction. A financial bond and an insurance deposit are payable prior to construction under the Law on Mining and Geological Exploration and Ordinance on Conditions and Procedure for Issuing Licence for Waste Management and Criteria, Characterisation, Classification and Reporting on Mining Waste, repayable after closure. The permitting system is undergoing change as part of Serbia’s planned accession to the EU. Serbia was granted candidate status in March 2012 and legislation and frameworks are being harmonised progressively to those of the EU. Changes are anticipated, for example, to fully align with the environmental impact assessment directive (2011/92/EU), water and waste framework directive (2008/98/EC) and industrial emission directives (2010/75/EU). No private-sector mining project of this scale has passed through the entire Serbian permitting process in recent years; hence there is a lack or precedent to inform the permitting process as it currently stands. DPM has sought to minimise permitting risks by engaging with regulators and aligning the Timok Gold Project with EU requirements and good international practice such as the performance requirements of the European Bank for Reconstruction and Development (EBRD), Equator Principles and World Bank Group Environmental, Health and Safety Guidelines.

20.2.2 Current Status of Permitting The relevant exploration licences at the Timok Gold Project are currently in their final licence extension period of two years. This will allow for exploration activities to continue until 2021. The subsequent reservation of the exploration field permit will provide up to three additional years to complete the initial permitting activities to obtain a Certificate of Mineral Resources and Reserves and an exploitation licence. Table 20-2 provides a high-level summary of the permitting schedule, showing the key milestones. The spatial plan is the precursor to several other approvals required and is likely to be on the critical path; this is a regulator-led process to be triggered by DPM. The environmental impact assessment and water management approval processes can be lengthy, so the Project has already started collecting baseline data and identifying key issues to help streamline these processes. The timeframes set out for regulator responses to submissions are not always adhered to in Serbia for a number of reasons and flexibility is needed in the program to accommodate this. Continuing to engage with regulators to keep abreast of forthcoming changes to regulations will be important to keeping the permitting schedule on track.

CSA Global Report Nº R302.2019 189 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Figure 20-1: Permitting overview Source CSA Global 2019

CSA Global Report Nº R302.2019 190 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Table 20-1: Key permits required for Timok Gold Project Permit Legislation Permit summary Regulator Prerequisites Exploration Geological Exploration Law on Mining and Permits exploration activities. Ministry of Mining and • Design for geological exploration with Approval Geological Explorations Energy, Sector for Geology report on technical control. (Official Gazette of the and Mining • Cultural heritage and nature protection Republic of Serbia – OGRS conditions (integral part of design for No. 101-15 and 95-18 and geological exploration). oth. laws) • Proof of the land ownership or easements on land for the planned exploration. Reservation of Law on Mining and This is an optional phase where there may be a Ministry of Mining and • Program of activities that investor plans to Exploration Field Geological Explorations request to retain the ground after the expiry of Energy, Sector for Geology undertake during the period of reservation. the final exploration period. This provides and Mining additional time to complete permitting activities required for exploitation. Pre-Construction Certificate of Law on Mining and This certificate confirms Mineral Reserves and Ministry of Mining and • Geological exploration approval. Resources and Geological Explorations Mineral Resources and is issued on the basis of Energy, Sector for Geology • Final report on the results of explorations. Reserves Study (Elaborate) of resources and reserves. and Mining • Study (Elaborate) of resources and reserves. Approved Serbian Law on Mining and The Serbian feasibility study covers various topics Ministry of Mining and • Decision on environmental impact Feasibility Study Geological Explorations including details of exploitation techniques, mine Energy, Sector for Geology assessment screening and scoping. service life, annual production capacity, mine and Mining • Cultural heritage and nature protection environmental and social impacts and economic conditions. assessment. • Water conditions. • Information on the impact of seismic events on the Project. Location Conditions Law on Planning and Location Conditions sets out the environmental, Ministry of Construction, • Approved spatial plan. Construction (OGRS No. social and engineering design requirements from Traffic and Infrastructure/ • Concept design (non-mining structures). 72-09, 81-09 - corr., 64-10 the relevant authority. Žagubica municipality – CC decision, 24-11, 121- (department for urbanism) 12, 42-13 - CC decision, 50- 13 - CC decision, 98-13 - CC decision, 132-14 and 145- 14 and 83-18 and 31-19)

CSA Global Report Nº R302.2019 191 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Permit Legislation Permit summary Regulator Prerequisites Nature Protection Law on Nature Protection These conditions set out the nature protection Institute for Nature • Location layout. Conditions (OGRS, nos. 36-09, 88-10, design requirements from the institute. Protection of Serbia • Excerpt from Serbian approved feasibility 91-10 - corr., 14-16 and 95- study (for mining structures). 18 - oth. law) • Concept design for non-mining structures (for non-mining structures). Cultural Heritage Law on Cultural Heritage These conditions set out the cultural heritage Institute for Protection of • Location layout. Conditions (OGRS, nos. 71-94, 52-11 – design requirements from the institute, including Cultural Heritage • Excerpt from Serbian approved feasibility oth. Laws and 99-11 – oth. areas to be avoided. study (for mining structures). laws) • Concept design for non-mining structures (for non-mining structures). Water Conditions Law on Water (OGRS No. Water conditions set out the water management Ministry of Agriculture, • Excerpt from Serbian approved feasibility 30-10, 93-12 and 101-16, design requirements from the relevant authority. Forestry and Water study (for mining structures). 95-18 and 95-18 – oth. This is the first of three steps to obtaining the Management, Water • Concept design for non-mining structures law) overall water permit. Directorate/Žagubica (for non-mining structures). municipality • 12 months of groundwater and surface water monitoring. • Hydrological study of watercourses. • Opinions of public water management company, hydrometeorological service and Serbian Environmental Protection Agency. Decision on Law on Environmental This decision provides Ministry’s confirmation of Ministry of Environmental • Excerpt from Serbian approved feasibility Environmental Impact Impact Assessment (OGRS whether an environmental impact assessment is Protection study. Assessment Screening No. 135-04 and 36-09) required – all mining projects typically require an • Excerpt from the spatial plan. and Scoping EIA. It also sets out the scope required for the • Water conditions. impact assessment. DPM will prepare a scoping • Nature protection conditions. report setting out the nature of the Project, information on environmental baseline conditions, • Cultural heritage conditions. potential impacts and mitigation approaches, to • Location conditions. inform the Ministry’s decision. Approval for Law on Mining and This approval regulates exploitation of mineral Ministry of Mining and • Certificate on resources and reserves. Exploitation of Geological Explorations reserves. Submitted when project owner has Energy, Sector for Geology • Serbian approved feasibility study. Minerals finalised the geological survey. and Mining • Act confirming compliance with spatial plan.

CSA Global Report Nº R302.2019 192 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Permit Legislation Permit summary Regulator Prerequisites Construction Long-term Program of Law on Mining and In the case of exploitation, the reserves of mineral Ministry of Mining and • Serbian approved Feasibility study. Exploitation Geological Explorations resources that have strategic importance for the Energy, Sector for Geology (OGRS No. 101-15, Art. 88) Republic of Serbia, it is mandatory to develop a and Mining long-term program of exploitation for a period of at least 10 years. The long-term program is the technical basis of the development the spatial plan for special purposes. Main Mine Design Law on Mining and The main mine design must be technically Ministry of Mining and • Serbian approved Feasibility study. Geological Explorations approved by a Serbian company licensed for the Energy, Sector for Geology • Certificate of Resources and Reserves. (OGRS No. 101-15) purpose. and Mining Supplementary Mine Law on Mining and The supplementary mine design is a detailed Ministry of Mining and • Serbian approved Feasibility study. Design Geological Explorations design of. Energy, Sector for Geology • Certificate of Resources and Reserves. (OGRS No. 101-15; Art 89, and Mining 91) Approval of Law on EIA (OGRS No. 135- This approval sets out measures to be Ministry of Environmental • Environmental baseline. Environmental Impact 04 and 36-09) implemented during construction, operation and Protection • Environmental impact assessment. Assessment closure to safeguard the environment. The Project • Feasibility study (for mining structures)/ must prepare an environmental impact Preliminary design (for non-mining assessment for the design for review by the structures). Ministry. The assessment must follow the requirements of the scoping decision, take account of environmental and social baseline data and the findings of stakeholder engagement. Nature Protection Law on Nature Protection Confirms that nature protection conditions have Institute for Nature • Location layout. Consent (OGRS, nos. 36-09, 88-10, been implemented into design. Protection of Serbia • Main mine design (for mining structures). 91-10 - corr., 14-16 and 95- • Design for construction permit (for non- 18 - oth. law) mining structures). Cultural Heritage Law on Cultural Heritage Confirms that cultural heritage conditions have Institute for Protection of • Location layout. Consent (OGRS, nos. 71-94, 52-11 – been implemented into design. Cultural Heritage • Main mine design (for mining structures). oth. Laws and 99-11 – oth. • Design for construction permit (for non- Laws) mining structures).

CSA Global Report Nº R302.2019 193 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Permit Legislation Permit summary Regulator Prerequisites Water Consent Law on Water This is the second phase in water permitting. It Ministry of Agriculture, • Water conditions. confirms that water conditions have been Forestry and Water • Location conditions. implemented into design. Management, Water • Main mine design. Directorate/Žagubica • Design for construction permit (for non- municipality (for structures mining structures). under its jurisdiction) • Decision of Ministry competent for geological exploration on established and classified reserves of groundwater. Mine Waste The Ordinance on the This permit sets out the approach to management Ministry of Mining and • Mine waste management plan. Management Permit Conditions and Procedure of mine waste. Energy, Sector for Geology • Environmental impact assessment approval. and Mining for Issuing License for • Approval for geological Waste Management, and exploration/exploration field area. Criteria, Characterisation, Classification and Reporting on Mining Waste (Official Gazette of the Republic of Serbia, no. 53/17) Approval for Law on Mining and This permit authorises construction of the mine. Ministry of Mining and • Main mine design. Construction of Mine Geological Explorations Energy, Sector for Geology • Certificate of Resources and Reserves. Structures and Mining • Approval of environmental impact (Construction Permit) assessment. • Nature protection consent. • Cultural heritage consent. • Water consent. • Act confirming compliance with spatial plan. • Approval of fire protection design. • Proof of lease/ownership. Approval for Law on Mining and This approval covers testing and trials of Ministry of Mining and • Application for commissioning into a trial Commissioning into a Geological Explorations, installations, devices and plant and commissioning Energy, Sector for Geology operation. Trial Operation (Art. 112) tests. and Mining

CSA Global Report Nº R302.2019 194 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Permit Legislation Permit summary Regulator Prerequisites Approval for Law on Planning and This permit authorises construction of non-mine Ministry of Construction, • Design for construction permit. Construction of Non- Construction (OGRS no. 72- structures (e.g. roads, substations). The design for Traffic and • Location conditions. Mine Structures 09, 81-09 - corr., 64-10 – the construction permit must be technically Infrastructure/Žagubica • Proof of lease/ownership/approval for (Construction Permit) CC decision, 24-11, 121-12, approved by the commission for structures municipality property access by landowner. 42-13 - CC decision, 50-13 - appointed by the Ministry. • Energy permit (applicable for energy CC decision, 98-13 - CC structures). decision, 132-14 and 145- 14 and 8318 and 31-19) Operation Water Permit Law on Water This permit applies to abstraction, use, and Ministry of Agriculture, • Report from the public water management discharge of water. It defines the quantity of Forestry and Water company Srbija Vode. water that can be abstracted and discharged, and Management - Water • Commission’s report on the technical the quality of the discharged water in line with the Directorate inspection of the structure. receiving water, or potential use of wastewater. • Designs used for construction of mine and non-mining structures. Approval for Use of Law on Mining and This permit is required prior to operation of the Ministry of Mining and • Technical examination (by company Mine Structures Geological Explorations mine. It is issued when the mine passes technical Energy, Sector for Geology approved by Ministry). (operating permit) examination. and Mining Approval for Use of Law on Planning and This permit is required prior to operation of non- Ministry of Construction, • Technical examination. Non-Mine Structures Construction mining structures. It is issued when the structures Traffic and Infrastructure (operating permit) pass technical examination. Integrated Pollution Law on IPPC (OGRS No. This permit regulates environmental control of Ministry of Environmental • Approved spatial plan. Prevention and Control 135-04 and 25-15) industrial emissions. Mine design will need to be Protection, Sector for • Construction permit. (IPPC) permit (if reviewed to confirm whether an IPPC permit is Environmental • Water permit. required) required. Management • Environmental impact assessment approval. • Environmental monitoring reports (trial run). Closure Notification to Law on Mining and The Law on Mining and Geological Explorations Ministry of Mining and Ministries Geological Explorations requires that Ministries responsible for mining, Energy, Sector for Geology agriculture and water management are notified of and Mining mine closure. Permit to Demolish Law on Planning and This permit relates to non-mining structures only. Ministry of Construction, Structures Construction Traffic and Infrastructure Note: This table includes key permits only. DPM maintains a register of legislation applicable to the Project.

CSA Global Report Nº R302.2019 195 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Table 20-2: Overview of permitting schedule to mine operation Activity 2019 2020 2021 2022 2023 2024

NI 43-101 Studies

Preliminary Economic Assessment

Prefeasibility Study

Feasibility Study

Permitting for Exploration and Exploitation Spatial Plan (including SEA) Location Conditions

Study (Elaborate) of Resources and Reserves

Certificate of Resources and Reserves

Excerpt from Serbian Feasibility Study

Approved Serbian Feasibility Study

Approved Concept Design (Non-Mining Structures)

Water/Nature Protection/Cultural Heritage Conditions

Decision on EIA Screening and Scoping

Approval for Exploitation of Minerals

Permitting for Construction

Main Mine Design

Approval of EIA

Water/Nature Protection/Cultural Heritage Consent

Mine Waste Management Permit

Approval for Construction of Mine Structures (Construction Permit)

Approval for Construction of Non-Mine Structures (Construction Permit)

Mine Construction

Water Permit

Approval for Use of Mine Structures (Operating Permit)

Approval for Use of Non-Mine Structures (Operating Permit)

Integrated Pollution Prevention and Control (IPPC) Permit

Mine Operation

CSA Global Report Nº R302.2019 196 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

20.3 Environmental and Social Studies, Setting and Issues

20.3.1 Environmental and Social Studies Table 20-3 provides a summary of the environmental and social studies already completed by the Timok Gold Project and those planned in the coming 12 to 18 months. Baseline studies completed to date have provided an understanding of the key environmental and social sensitivities in the area. The schedule of future work has been planned to meet Serbian permitting requirements, align with international good practice and to help further understand or close out environmental and social risks. Table 20-3: Environmental and social studies completed and planned Environment Work completed Work planned Surface water and Hydrological studies undertaken in 2012, 2013 and 12 months of surface water data collection groundwater 2014 of the Bigar-Korkan area. upstream/downstream of Project and 12 months of groundwater sampling up gradient and downgradient of the Project as required under the Law on Water. Terrestrial ecology Habitat and species surveys undertaken in 2013 of Assessment to record data on presence of protected the area around Korkan, Bigar and Kraku Pester, and threatened flora and fauna (e.g. trails, tracks, before the project footprint changed. Surveys sighting) and to update 2013 survey. Dedicated included habitats and flora, mammals, birds, seasonal surveys for flora, bats and large mammals as amphibians and reptiles a minimum. Aquatic ecology Habitat and species surveys undertaken in 2013 of Environmental DNA (eDNA) sampling and analysis to the area around Korkan, Bigar and Kraku Pester, provide a list of fish and crayfish species present. before the project footprint changed. Surveys Targeted surveys to identify macroinvertebrates of included phytobenthos, aquatic macrophytes, conservation interest (e.g. caddisfly Helicopsyche aquatic macroinvertebrates and fish. bascescui). Air quality - 12 months of air quality monitoring (oxides of nitrogen, nitrogen dioxide, sulphur dioxide, particulate matter (PM10 and PM2.5) and dust deposition with associated metals analysis, in line with good international practice. Noise and - One to two weeks longer term and shorter term vibration operator-attended monitoring. Social baseline - Household surveys, focus group discussions, informant interviews. Land acquisition Survey of land to be acquired undertaken in 2012, Census and socio-economic survey in line with good including details of ownership, and potential costs. international practice. Landscape - Observations and photographs at sensitive viewpoints. Soils Study by Belgrade Institute of Soil Science in 2007. Soil sampling for land capacity and chemical quality required by Law on Soils Protection. Heritage Broad, high-level study of archaeology across Targeted walkover survey to identify surface traces of licence area in 2012. buried remains, significant historical buildings and caves/rock shelters (documentary surveys) and identify whether further studies are needed.

20.3.2 Potential Environmental and Social Risks Site Setting The Timok Gold Project is located in eastern Serbia, on the mountain range between Bor to the southeast and Žagubica and Laznica to the west (Figure 4-1). Mjajdanpek and the Danube River are to the north. State roads 164 and 161 run east-west through the exploration area (Figure 4-2 and Figure 20-4). There are no designated

CSA Global Report Nº R302.2019 197 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

protected areas for biodiversity or cultural heritage in or around the three northern pits that make up the current Timok Gold Project (Korkan, Korkan West and Bigar Hill). The area is characterised by wooded valleys and seasonally grazed pastures with isolated settlements.

Key Environmental and Social Risks Environmental and social risks associated with the Project were identified through a risk review in December 2018. These will be further investigated and assessed as part of the environmental impact assessment process. The key risks are around surface water and groundwater, biodiversity and economic displacement associated with land acquisition. The risks identified are typical of similar gold mining projects. Table 20-4 summarises interactions between project activities and the environment. Table 20-4: Initial assessment of interactions between Project activities and environment

Aspect Receptor

heritage

missions

e

Light

Water

Land use Land

Land take Land

Wateruse

Air quality Air

Landscape

Discharges

Livelihoods

Flora /fauna Flora

Soils, geology Soils,

GHG GHG

Emissions to air to Emissions

Noise/vibration

Fishing, hunting Fishing,

Cultural Cultural

Health and safety and Health Hazardous materials Hazardous Construction Change in land use to mining ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ Pit construction ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ Construction of infrastructure ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ Road construction ✓ ✓ ✓ ✓ Operation Pits ✓ ✓ ✓ ✓ ✓ ✓ Heap leach process ✓ ✓ ✓ Flotation process ✓ ✓ ✓ ✓ ✓ ✓ Water supply ✓ ✓ ✓ ✓ ✓ Dewatering ✓ ✓ ✓ ✓ Discharges ✓ ✓ ✓ Chemical/fuel storage ✓ ✓ ✓ ✓ ✓ Waste management ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ Employment, procurement ✓ ✓ Major accidents ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ Closure ✓ Removal of structures ✓ ✓ ✓ ✓ ✓ ✓ ✓ Rehabilitation and restoration ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓

Watercourses and Groundwater A network of rivers and streams run through the Project area (Jagnjilo, Bigar, Valja streži, Tisnca, Crna reka, Vrkaluca, Valja Saka, Ogešu Krloši, Valja Mare) with the catchments of the Zlotska Reka, Mlava, Veliki Timok and rivers (Figure 20-2). Watercourses in this area drain to River Danube, which is internationally protected. Serbia is a signatory to the International Commission for Protection of the Danube River and projects which may affect water quality of the Danube could trigger the need for transboundary engagement between the Serbian and neighbouring governments (Romania, Bulgaria).

CSA Global Report Nº R302.2019 198 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

The waste rock dumps, the heap leach pad and to a lesser extent, the mine pits themselves, will be sited within the uppermost catchment valleys of streams in the area. These will require diversionary channels to redirect surface water flow around these facilities. Potential risks to surface water and groundwater quality include sedimentation, seepage of process chemicals or hydrocarbons, leachate or runoff from acid generating rock, planned discharges and pollution during flood events, which typically take place during April snowmelt (see hydrograph in Figure 20-3). Designing the Project to avoid or mitigate these impacts will be important, taking account of the fact that rivers in this area are heavily channelised in steep valleys, which will constrain locations available for mitigation measures such as sedimentation ponds. Parts of the site are underlain by karstic geology which poses a risk to the Project as infiltration and spread of pollutants to groundwater is typically greater through karst. The transportation and use of cyanide in the heap leach process presents potential risks to surface and groundwater quality. DPM is a signatory to the International Cyanide Management Code2, which provides standards of practice for protection of communities and the environment during transportation of cyanide and specific usage requirements on handling, storage, operation, disposal and decommissioning. The leach pad has been sited away from the karstic limestone zone, where infiltration to groundwater could more readily occur, and leach pad design includes a robust liner system that collects the gold solution and also prevents infiltration to groundwater. The Project site lies outside existing water supply protection zones. There are, however, wells and springs in the Project area (Figure 20-2), some of which may supply households and agriculture with water. Abstraction of groundwater to supply the Project and the consequent dewatering may affect community water use, agriculture and habitats. The water balance being developed for the Project will help identify potential impacts and further engagement with the local community is needed to fully understand potential impacts to domestic wells. The Bigar Hill Spring, to the south west of Bigar Hill, has been identified by local authorities as a potential source of municipal drinking water. Designation as a municipal drinking water source could result in additional project constraints. DPM is working with the authorities to find alternative sources.

2 https://www.cyanidecode.org

CSA Global Report Nº R302.2019 199 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Figure 20-2: Watercourses and wells in Project area Source:ERM GIS 2019

Figure 20-3: Bigar hydrograph Source ERM unpublished report

CSA Global Report Nº R302.2019 200 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Biodiversity The Project area is dominated by beech woodland, interspersed with pasture and agricultural land. Previous studies including interviews with local stakeholders have identified the potential presence species of conservation interest, including three large carnivore species protected at national and European level (brown bear (Ursus arctos), wolf (Canis lupus) and lynx (Lynx lynx)), a high abundance and diversity of nationally and European protected bat species, protected bird species (including golden eagle (Aquila chrysaetos), peregrine falcon (Falco peregrinus), grey partridge (Perdix perdix)), plant species of conservation concern, including the green-winged orchid (Orchis morio L.) and nationally and European protected aquatic species including stone crayfish (Austropotamobius torrentium). The presence of European protected and endemic species may trigger criteria for EBRD priority biodiversity habitat and features, requiring further investigation, evaluation and management. DPM has planned further baseline study to further understand the presence and distribution of species and impacts will be assessed as part of the environmental impact assessment. It is likely that all species present within the Project area have a wider distribution than that of the Project footprint alone.

Noise, Vibration and Nuisance Dust No air quality baseline surveys or noise surveys have yet to take place. Historically, the main source of industrial pollution in area is the Bor smelter complex, some 20 km to the southeast. The baseline environment is also known to have naturally elevated concentrations of arsenic and cadmium, which can be mobilised through dust. The Project will generate noise and dust from construction and operational activities and in closure. This may affect the people living permanently or seasonally in the area, as well as the flora and fauna. A full assessment of noise, vibration and dust will be required as part of the environmental impact assessment which will also need to take account of the existing background levels of arsenic, cadmium and other historical pollution.

Soils and Geology Preliminary soil surveys were undertaken in 2007 and further surveys are planned to align with Serbian regulations. The mineralogy of the mineralised material indicates that acid generation is unlikely to be a significant issue.

Cultural Heritage There are no protected areas for cultural heritage or known archaeological features within the project footprint. The region is well-known to be home to some of the earliest metallurgical technology in Europe and it is possible that there are buried remains in the project area. Serbian antiquities law and procedure relies on the principle of “prior protection” and buried antiquities uncovered during development have the potential to stop work. For this reason, the Project has planned further baseline studies to understand this potential risk. The potential for Palaeolithic remains in the Korkan Cave to the north of the Korkan pit (Figure 20-4) is currently being investigated. Some of the stone farm buildings and mills in the Project area have archaeological value. Project infrastructure should aim to avoid these areas. The Project area is important to the Vlach community, an ethnic community in Serbia with its own language, dress and culture. Religious customs are connected with the land, including beliefs in sprits of the woods (e.g. fairies) and celebration of inscription trees. In addition, transhumance is a fundamental element of Vlach culture, with grazing on higher ground, including that in the Project area, in the summer. The environmental impact assessment will address potential impacts on intangible heritage, such as these.

CSA Global Report Nº R302.2019 201 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Figure 20-4: Cultural heritage features Source: ERM GIS 2019

Land Acquisition and Livelihoods Land acquisition is required for the development of the Project, access to the site and the construction of associated facilities such as substations. Gaining access to this land requires purchasing land from private and public owners. Proof of land ownership is a prerequisite to obtaining key permits such as the exploitation of Mineral Resources and Reserves. The Project plans to acquire land from local landowners on a willing-buyer willing-seller basis. At present, the Project footprint largely includes summer grazing pastures and does not require resettlement of households. A 2012 survey confirmed that there are a large number of small land parcels within the Project area, of which the majority are privately owned. DPM has longstanding relationships with landowners and has previously successfully acquired a number of small parcels of land on a trial basis. Land acquisition will take farmland out of use. Though farming may not be the primary source of income for local people, this aspect of their livelihoods needs to be considered by the Project. A census and socio- economic survey of landowners is needed to identify landowners eligible for assistance and compensation and inform a livelihood restoration plan (or resettlement action plan if needed), in line with EBRD and good international practice. If it is not possible to purchase land using a willing-buyer willing-seller approach, expropriation through compulsory purchase may be possible. The Law on Expropriation allows government the right to acquire immovable property for projects that are demonstrated to be in the public interest. The Law on Mining and Geological Explorations allows holders of rights to mineral explorations to be beneficiaries of an expropriation process. Expropriation for the development of mining projects is widely understood to be an application of the ‘public interest’ requirement, though the process has yet to be tested for private sector mineral projects

CSA Global Report Nº R302.2019 202 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

in Serbia. Expropriation would result in cost and delay as applications are made, government entities review whether the project is in the public interest, compensation is confirmed and as any appeals are addressed and remains an option of last resort for the project. Additional actions would be taken by DPM to bring any land acquisition through expropriation into line with international good practice for such processes.

Other Issues There are also risks associated with health, safety and workers’ rights, at the Project and within the supply chain, landscape and visual impact, greenhouse gas emissions, handling hazardous materials and cumulative impacts as a result of several mining sites being progressed in the same location. The development of the Project design should consider alternatives to minimise and avoid these impacts where possible.

20.4 Social and Community Engagement

20.4.1 Stakeholder Engagement The Project will be subject to scrutiny by regulatory authorities and other stakeholders during the permitting process. There are six formal hearing and consultation periods included within the spatial plan, strategic environmental assessment and environmental impact assessment processes and a range of other points where the public and other interested parties could comment. The Project has worked to establish good relationships with the local community since 2007 and communications are managed through a stakeholder engagement plan, communications plan and grievance mechanism. DPM has expanded its resources and conducted training in 2019 to facilitate transparent and meaningful community engagement.

20.5 Mineral Wastes Mineral waste will be in the form of waste rock from the excavation of the pits and from some construction activities, stored at waste dumps adjacent to each of the pits (Figure 16-14). There will be no separate tailings storage facility as the tailings from the flotation plant will be added to the lined heap leach pad to maximise gold recovery. Information on the local soil and bedrock geochemistry indicates that acid rock drainage is unlikely to be a significant issue, and this will be further investigated at the PFS stage. The approach to mine waste management will be set out in the statutory project mine waste management plan, developed in line with the Serbian regulations on characterisation, classification and reporting on mining waste (ORGS 53/17) and good international practice (e.g. EU Extractive Waste Directive3 and International Finance Corporation (IFC) Health and Safety Guidelines on Mining).

20.5.1 Non-Mineral Wastes Non-mineral wastes will include non-hazardous and hazardous materials such as packaging, used oil, batteries, food, medical waste and sewage. The Project will develop a waste management inventory as part of the design process and a strategy for disposal of each waste stream, following the waste hierarchy (reduce, reuse, recycle, treat, dispose) and in line with Serbian regulations and international good practice. Suitable third-party waste carriers and treatment/disposal sites will be identified and the details of the approach for storage, transportation, treatment and disposal of each waste stream will be set out in the Project waste management plan.

3 EU Directive 2006/21/EC on the management of waste from the extractive industries.

CSA Global Report Nº R302.2019 203 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

20.6 Mine Closure and Aftercare The Project’s approach to closure will be to rehabilitate the mine site so that it is physically and chemically stable and compatible with the intended future land use, which has yet to be determined. The aim will be to minimise or eliminate long term active aftercare such as water treatment requirements. The site will be monitored post closure for physical and chemical stability. The closure approach will include: • Securing open pits with rock berms, rendering them safe and stable and allowing them to flood. • Restoring waste rock dumps and installing adequate drainage to prevent erosion and enhance physical stability. • Dismantling site buildings – some buildings may be retained as a part of a designated end use of the site. • Decontaminating machinery and equipment. • Scarifying, grading and contouring roads. • Removing chemicals, waste and explosives from the site by a licensed waste carrier and recycling and disposal in line with Serbian regulation and good international practice. • Testing, excavating and removing any contaminated soils. • Draining surface ponds. • Revegetating using species compatible with local habitats. The heap leach will remain after site closure as a low, unsaturated tabular landform. Appropriate stabilisation and covering options will be investigated during the PFS and tested during the operational phase. Stakeholders will be included in the closure planning process, covering information on likely future uses, changes in working patterns and closure schedule. Details of mine closure and aftercare will be set out in the mine closure plan, which will be developed in line with Serbian requirements (ORGS 27/974) and international good practice (e.g. EU Extractive Waste Directive5, EU Best Available Techniques6, IFC Health and Safety Guidelines on Mining7, International Council on Mining and Metals (ICMM) Good Practice Guide8) and submitted as part of the main mine design. Details of closure and aftercare of the waste rock dump and heap leach will set out in the mine waste management plan (in line with ORGS 53/179), submitted as part of the application for the mine waste management permit. A closure cost estimate of $10 million has been included in the economic cost model, spread over three years at the end of the mine life. This cost covers final site remediation. The Project will conduct a full closure planning and costing exercise at the PFS stage.

4 Rulebook on the content of mining designs (ORGS 27/97) 5 EU Directive 2006/21/EC on the management of waste from the extractive industries. 6 EU (2009) Reference Document on Best Available Techniques for Management of Tailings and Waste-Rock in Mining Activities 7 IFC (2007) Environmental, Health and Safety Guidelines for Mining 8 International Council on Mining and Metals (2019) Integrated Mine Closure: A Good Practice Guide 9 Ordinance on the Conditions and Procedures for Issuing Licence for Waste Management and Criteria, Characterisation, Classification and Reporting of Mining Waste (ORGS 53/17)

CSA Global Report Nº R302.2019 204 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

21 Capital and Operating Costs

Readers are cautioned that the PEA is preliminary in nature. It includes Inferred Mineral Resources considered too speculative geologically to have the economic considerations applied to them that would enable them to be categorised as Mineral Reserves, and there is no certainty the PEA will be realised. Mineral Resources that are not Mineral Reserves do not have demonstrated economic viability. The estimate of Mineral Resources may be materially affected by environmental, permitting, legal, title, socio-political, marketing or other relevant issues.

21.1 Capital Cost Summary The initial and LOM capital cost estimates for the Timok Project are summarised in Table 21-1. All costs are expressed in US$ unless otherwise stated and are based on Q1 2019 pricing with a deemed overall accuracy of ±40%. Table 21-1: Capital cost estimate Capital category Pre-production ($‘000) Sulphide Processing ($‘000) Sustaining ($‘000) Total ($‘000) Mining 35,364 4,823 40,187 Processing (heap leach) 33,482 16,172 49,654 Flotation plant - 30,327 30,327 Infrastructure 34,157 3,520 37,677 Environmental - 10,000 10,000 Indirects 17,962 17,513 35,475 Contingency 15,138 13,140 28,278 Total 136,103 30,327 65,168 231,598 Note: Mining includes $2.206 million in pre-production stripping. Infrastructure includes $10.7 million in land acquisition cost. The Indirect and Contingencies are based on various percentages that are outlined in Table 21-2 and Table 21-3. Within the Indirects is a sum of $8.3 million for owner costs in the pre-production period and the remaining $4.4 million in sustaining. Indirects with the Infrastructure category are included in the base cost estimate. Table 21-2: Indirect costs and percentages Capital category Indirect cost ($‘000) Indirect cost (%) Open pit mining 380 1 Process plant 22,395 28 Infrastructure - - Environmental - - Owner costs 12,700 -

Table 21-3: Contingency costs and percentages Capital category Contingency cost ($‘000) Contingency cost (%) Open pit mining 1,899 5 Process plant 19,995 25 Infrastructure 5,383 20 Environmental 1,000 10 Owner costs - -

CSA Global Report Nº R302.2019 205 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

21.2 Mine Capital Costs The mine capital costs are estimated from quotations from local vendors and information in AGP’s database of recent projects. The distribution of the capital cost is completed using the units required within a period. If new or replacement units are needed, that number of units, by unit cost, determines the capital cost for that period. There is no allowance for escalation in any of these costs. Timing of major capital equipment costs is one year in advance of the need for that piece of equipment. The major mining equipment is shown in Table 21-4. Table 21-4: Major mining equipment – capital cost and estimated equipment life Equipment Units Capacity Equipment life Unit capital cost ($‘000) LOM quantity Rotary production drill mm 200 30,000 hrs 3,182 3 Production loader m3 13 35,000 hrs 1,783 2 Hydraulic excavator m3 6.7 35,000 hrs 1,188 2 Haulage truck t 63 35,000 hrs 870 12 Haulage truck t 40 35,000 hrs 600 Tracked dozer kW 337 35,000 hrs 664 12 Grader kW 163 20,000 hrs 487 7

Replacement times for the equipment are average values from AGP’s experience. Options around rebuilds and recertification of equipment such as track dozers is not considered, nor is used equipment, although that should be considered during the purchase of the mine fleet. The balancing of equipment units based on operating hours is completed for each major piece of mine equipment. The smaller equipment was based on number of units required, based on operational experience. This includes such things as pickup trucks (dependent on the field crews), lighting plants, mechanics trucks, etc. The most significant pieces of major mine equipment are the haulage trucks. At the peak of mining, 12 units are necessary to maintain mine production. The maximum hours per truck/per year are set at 6,000. There are periods where the maximum hours per unit are below the maximum possible capacity. In those situations, increasing the maximum on the number of trucks still leaves residual hours required to complete the material movement; therefore, the number of total trucks is unchanged. In these cases, the hours required are distributed evenly across the number of trucks. The other major mine equipment is determined in the same manner. Therefore, in some instances the loaders have a longer period of life (same number of hours between replacements) due to the sharing of hours with the other units in the fleet. The support equipment is usually replaced on a number of year’s basis. For example, pickup trucks are replaced every three years, with the older units possibly being passed down to other departments on the mine site, but for capital cost estimating new units are considered for mine operations, engineering, and geology. The mining capital cost estimate is shown in Table 21-5.

CSA Global Report Nº R302.2019 206 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Table 21-5: Mining capital cost estimate Equipment Pre-production ($‘000) Sustaining ($‘000) Total cost ($‘000) Rotary production drill 9,546 - 9,546 Production loader 3,566 - 3,566 Hydraulic excavator 1,188 1,188 2,376 Haulage truck (63-t) 8,700 1,740 10,440 Haulage truck (40-t) 1,200 600 1,800 Tracked dozer 2,657 - 2,657 Grader 974 - 974 Support equipment 5,327 1,295 6,622 Total equipment 33,158 4,823 37,981 Mine pre-stripping 2,206 - 2,206 TOTAL MINE CAPITAL 35,364 4,823 40,187

21.2.1 Pre-Production Stripping The mine is scheduled to initiate stripping activities in Year -1. Waste material on top of feed material must be moved with some of the waste used to build the HLF and roads. Oxide material that is encountered would be stockpiled near the crushing plant. A total of 1 Mt will be mined in this time period and the costs for this are being attributed to pre-production capital costs.

21.3 Process Capital Costs The 2.5 Mt/a HLF and 0.5 Mt/a mill/flotation process plant design described in Section 17 has been used as the basis for the PEA-level plant capital cost estimates.

21.3.1 Direct Capital Costs Capital costs for the HLF includes the crushing plant, heap leach pad, ponds and ADR plant. Whereas, capital costs for the mill/flotation plant includes the crushing plant, milling, flotation and dewatering circuits. Table 21-6: Summary of process plant direct capital costs Plant cost category (installed) HLF cost ($’000) Mill/Flotation cost ($’000) Civil and earthworks 22,111 3,304 Mechanical equipment – installed 11,190 13,214 Concrete - 991 Structural steel 1,134 3,402 Platework 1,051 2,266 Piping 2,841 2,169 Electrical and instrumentation 7,944 3,710 Buildings 2,040 589 Mobile equipment 64 682 Capital and insurance spares 1,279 - Total 49,654 30,327

Basis for Capital Estimates Heap Leach Facility Capital costs for the crushing plant were estimated by obtaining budget cost estimates from suppliers and using standard industry multipliers for installation.

CSA Global Report Nº R302.2019 207 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Heap leach pad capital costs were prepared from first principles by calculating Bill of Quantities for earthworks, materials etc. and applying unit rates. Mill/Flotation Plant The capital cost estimate for the 0.5 Mt/a mill/flotation plant is has been scaled using orders of magnitude calculations from recent PFS level studies completed by CSA Global. The capital cost estimate for the 0.5 Mt/a mill/flotation plant was calculated by applying the 6/10th engineering factor to the total capital cost of the 1.2 Mt/a concentrator. Additional capital was added for equipment required to dry stack tailings from processing the sulphide mineralisation. Capital costs for the Mechanical Equipment Supply (MES) were generated by sending out requests to various vendors in order to obtain budget quotes for each of the main equipment items. The total MES cost was subsequently used as the basis for applying standard industry multipliers for civils, earthworks, structural, mechanical, and electrical/instrumentation to generate an overall installed capital cost. Various indirect costs and contingencies are estimated and added to this, and these are described in Sections 21.6 and 21.7 below.

21.4 Infrastructure Capital Costs The infrastructure capital cost is attributable principally in the pre-production capital. Lifts for the HLF are included in the process capital cost estimate in Section 21.3. This includes the initial pad preparation as well as annual lifts/liner extension. Major infrastructure capital costs include the shop facilities, power line to site, surface power distribution, transformers, and an explosives plant. A breakdown of the major costs associated with infrastructure are shown in Table 21-7. Table 21-7: Infrastructure capital Equipment Pre-production ($‘000) Sustaining ($‘000) Total cost ($‘000) Power line 3,410 - 3,410 Surface power/transformers 3,350 - 3,350 Mine maintenance shop/dry/tools 4,450 - 4,450 Explosives plant/pad/earthmoving 3,580 - 3,580 Assay laboratory 1,500 - 1,500 Reagent storage 150 - 150 Admin building/gate houses 750 - 750 Freshwater/potable water/fire suppression 1,500 - 1,500 Mobile site equipment 1,250 - 1,250 Waste dump preparation, diversion ditches, ponds 600 400 980 Site roads 2,600 - 2,600 Communications 500 - 500 Engineering office equipment 750 - 750 Dewatering system 420 1,000 1,400 Land acquisition cost 10,770 - 10,770 Miscellaneous infrastructure costs 740 - 760 Total 36,320 1,400 37,700

CSA Global Report Nº R302.2019 208 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

21.5 Environmental Capital Costs An allowance of $10 million has been included in the capital cost estimate at the end of the mine life to complete reclamation activities. The mine operating costs include concurrent reclamation of the waste piles.

21.6 Indirect Capital Costs Indirect costs for the process plant are calculated with the main subcomponents being the engineering, procurement and construction management (EPCM) contract, first-fill consumables and initial warehouse stocking. The owner costs have also been included in the indirect costs. Indirect costs and percentages are shown in Table 21-8. Table 21-8: Indirect capital costs Capital category Indirect cost ($‘000) Indirect cost (%) Open pit mining 380 1 Process plant 22,395 28 Infrastructure - - Environmental - - Owner costs 12,700 -

21.7 Contingency Contingency allowances have been estimated using various percentages by category and applied to the direct capital cost of a particular category. These are shown in Table 21-9. Table 21-9: Contingency cost by category and percentages Capital category Contingency cost ($‘000) Contingency cost (%) Open pit mining 1,899 5 Process plant 19,995 25 Infrastructure 5,383 20 Environmental 1,000 10 Owner costs - -

21.8 Operating Costs Summary Operating costs were developed for a 2.5 Mt/a heap leach operation and the 0.5 Mt/a sulphide operation with a nine-year mine life. Total LOM operating costs have been summarised in Table 21-10. Table 21-10: Operating cost summary $/tonne LOM operating cost $/tonne $/tonne Operating category Oxide & Transitional ($’000) Sulphide Feed Total Processed Feed Open pit mining 162,026 8.58 8.58 8.58 Process plant 138,025 5.24 16.46 7.31 (blended) G&A 31,132 1.65 1.65 1.65 Total 331,182 15.47 26.69 17.54

All prices in the PEA study are quoted in US$ as of the Effective Date of this report unless otherwise noted. Diesel fuel pricing is estimated at $1.51/L. This estimate was derived from a price quotation for off-road diesel

CSA Global Report Nº R302.2019 209 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

fuel delivered to site with applicable taxes considered. The price for electrical power was set at $70/MWh, based on current industrial pricing in Serbia.

21.9 Mining Operating Costs The mine operating costs are estimated from base principles. Each of the pit areas was examined and base operating costs were determined from a reference elevation and incremental costs above and below that level calculated for both waste and heap/mill feed. The costs are shown in Table 21-11. Table 21-11: Reference mining costs Operating category Units Bigar Hill Korkan Korkan West Reference elevation masl 710 710 620 Waste – base rate $/t moved 2.14 2.07 2.19 Waste – incremental cost (per 5 m bench) $/t moved 0.02 0.00 0.03 Heap/Mill Feed – base rate $/t moved 2.41 2.67 2.79 Heap/Mill Feed – incremental cost (per 5 m bench) $/t moved 0.01 0.01 0.01

Each of the phases in the schedule had these costs applied to them so that a weighted average mining cost was developed on an annual basis for use in the cash flow calculations. In this manner, mining at depth incurred a great mining cost and was reflected properly in the Project cash flow. Key inputs to the mine cost are fuel, labour and repair and maintenance parts. The fuel cost is estimated from the local price quotation delivered to site and applied to the various mine equipment. This was $1.51/L. Fuel consumption rates are provided by the equipment vendors and AGP’s database. The mine fleet is entirely diesel powered. Labour costs for the various job functions were discussed with DPM based on their experience in Serbia. Labour was estimated for staff and hourly on a 12-hour shift basis. Mine positions and salaries are shown in Table 21-12. Table 21-12: Mine staffing requirement and annual salaries Staff position Employees Annual salary ($/a) Mine Maintenance Maintenance Shift Foreman 4 27,000 Maintenance Planner/Contract Admin 1 22,950 Clerk/Secretary 1 8,640 Mine Operations Technical Superintendent – Expat 1 250,000 Mine General Foreman 1 54,000 Mine Shift Foremen 4 27,000 Road Crew/Services Foreman 1 27,000 Clerk/Secretary 1 8,640 Mine Engineering Chief Engineer – Expat 1 200,000 Senior Engineer 1 33,750 Open Pit Planning Engineer 2 30,375 Geotech Engineer 1 30,375 Surveyor/Mine Technician 3 13,500 Surveyor Helper/Mine Technician Helper 3 10,800 Clerk/Secretary 1 8,640

CSA Global Report Nº R302.2019 210 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Geology Chief Geologist 1 40,500 Senior Geologist 1 33,750 Grade Control/ Modeller Geologist 2 26,055 Sampling/Geology Technician 4 10,800 Clerk/Secretary 1 8,640 TOTAL MINE STAFF 35

The mine staff labour remains consistent for the entire mine life. Hourly labour force levels in the mine operations and maintenance departments fluctuate with production requirements. The annual salaries for mine staff labour are shown in Table 21-13. Table 21-13: Average annual salaries Hourly position Annual salary ($/a) Mine Maintenance Light Duty Mechanic 20,769 Tyre Man 20,769 Lube Truck Driver 13,856 Heavy Duty Mechanic 20,769 Welder 20,769 Electrician 20,769 Apprentice 13,856 Mine Operations Drill Operator 18,465 Excavator/Loader Operator 25,377 Haulage Truck Driver 14,993 Dozer Operator 17,297 Grader Operator 17,297 General Mine Equipment Operators 14,993 Road Crew/Pump Crew 10,384 General Mine Labourer 9,217 Trainee 13,856

Labour costs are based on an owner-operated scenario with blasting functions handled by DPM, but with delivery to the hole by the explosives supplier. A capital allocation has been made to construct an explosive plant and associated infrastructure for the magazine on site, which will be staffed by contractors provided by the explosives supplier. Drilling labour is based on one operator per drill, per crew while operating. Shovel, loader and truck operators are calculated in the same manner. Maintenance factors are used to determine the number of heavy-duty mechanics, welders and electricians. The factors for each maintenance staff type are outlined in Table 21-14. The factors in the table are simply multiplied by the number of operators in each area of the mine to arrive at the number of maintenance staff required (these are rounded up to whole numbers). Table 21-14: Maintenance labour factors (maintenance per operator) Maintenance job class Drilling Loading Hauling Mine operations support Heavy Duty Mechanic 0.25 0.25 0.25 0.25 Welder 0.25 0.25 0.25 0.25 Electrician 0.05 0.01 - -

CSA Global Report Nº R302.2019 211 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Apprentice - - - 0.25

The number of loader, truck, and support equipment operators is estimated using the projected equipment operating hours. The maximum number of employees is four per unit to match the mine crews. The vendors provided some repair and maintenance (R&M) costs for each piece of equipment with the missing values from AGP’s database. These came in the quotations for the capital cost. Fuel consumption rates are also estimated for the conditions expected at Timok and are used in the detailed costs for the mine equipment. The costs for the R&M are expressed in a $/hr form. The various suppliers provided the costs for different tire sizes that will be used during the Project. Estimates of the tire life are based on AGP’s experience and conversations with mine operators. The operating cost of the tires is expressed in a $/hr form. The life of the haulage truck tires is estimated to be 5,000 hours per tire assuming proper rotation from front to back. On the haulage trucks, each tire costs $9,000, so the cost per hour for tires is $10.80/hr for a truck using six tires, based on the calculation. Ground-engaging tool (GET) costing is estimated from other projects and conversations with personnel at other operations. This is an area of cost that is expected to be fine-tuned during mine operations. Drill consumables were estimated as a complete drill string using the parts list and component lives provided by the vendor. Drill productivity for both mill feed and waste is estimated at 26.5 m/hr. Equipment costs used in the estimate are shown in Table 21-15. Table 21-15: Major equipment operating costs – no labour ($/hr) Equipment Fuel Lube/Oil Tires R&M GET/Consumables Total Rotary production drill 166.10 16.61 - 134.43 55.86 373.00 Production loader 113.25 11.33 24.56 69.93 10.00 229.07 Hydraulic excavator 83.05 8.31 - 64.69 10.00 166.05 Haulage truck (63-t) 75.50 7.55 10.80 14.20 3.00 111.05 Track dozer 75.50 7.55 - 66.31 5.00 154.36 Grader 45.30 4.53 2.92 34.30 5.00 92.05

Drilling in the open pit will be performed using conventional rotary blasthole rigs with 200 mm bits. The pattern size is the same for heap/mill feed and waste. The drill pattern parameters are shown in Table 21-16. Table 21-16: Drill pattern specifications Specification Unit Heap/Mill feed Waste Bench height m 5 5 Sub-drill m 1.7 1.7 Blasthole diameter mm 200 200 Pattern spacing – staggered m 6.60 6.60 Pattern burden – staggered m 5.70 5.70 Hole depth m 6.70 6.70

The sub-drill was included to allow for caving of the holes in the weaker zones, avoiding re-drilling of the holes or short holes that would affect bench floor conditions and thereby increasing tyre and overall maintenance costs. Table 21-17 outlines the parameters used to estimate the drill productivity. Table 21-17: Drill productivity criteria Specification Unit Heap/Mill feed Waste Pure penetration rate m/minute 0.60 0.60

CSA Global Report Nº R302.2019 212 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Specification Unit Heap/Mill feed Waste Hole depth m 6.70 6.70 Drill time minute 11.17 11.17 Move, spot, and collar hole minute 3.00 3.00 Level drill minute 0.50 0.50 Add steel minute 0.00 0.00 Pull drill rods minute 0.50 0.50 Total setup/breakdown time minute 4.00 4.00 Total drill time per hole minute 15.20 15.20 Drill productivity m/hr 26.5 26.5

An emulsion product will be used for blasting to simplify the blasting plant operations. Explosives supply and blast hole loading in the pit will be provided by a contractor. Cost of the emulsion is quoted at $143.00/100kg. The supplier will also provide the explosives magazine and their labour for a monthly fee of $33,500 per month. The powder factor for heap/mill feed and waste were both 0.20 kg/t or 0.54 kg/m3. Loading of broken material will be completed by the hydraulic excavators and production loaders. They will split duties in loading heap/mill feed and waste. The excavator has a capability of 4.2 Mt/a while the loaders are each capable of 5.8 Mt/a of production. Haulage profiles were developed from the surface of each pit to the plant and waste dump locations and also from the bottom of the pits. Incremental costs were determined for the haulage that are used in the mine schedule to determine a weighted average mining cost. The mine support equipment costing is based on factors of various operating parameters. The factors are shown in Table 21-18. Table 21-18: Support equipment operating factors Mine equipment Factor Factor units Track dozer 25% of haulage hours to a maximum of 4 dozers Grader 10% of haulage hours to a maximum of 2 graders Water truck 15% of truck hours Tire manipulator 1 hours/day Lube/Fuel truck 8 hours/day Mechanics truck 12 hours/day Welding truck 12 hours/day Integrated tool carrier 2 hours/day Compactor 1 hours/day Lighting plants 12 hours/day Pickups 14 hours/day Pump truck 4 hours/day

These percentages resulted in the need for four track dozers, and two graders. Their tasks include clean-up of the loader faces, roads, dumps, and blast patterns. The graders will maintain the heap/mill feed and waste haul routes. In addition, water trucks have the responsibility for patrolling the haul roads and controlling fugitive dust for safety and environmental reasons. These hours are applied to the individual operating costs for each piece of equipment. Many of these units are support equipment so no direct labour force is allocated to them due to their function.

CSA Global Report Nº R302.2019 213 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Grade control will be accomplished with blasthole cuttings and RC drilling. The cost for the RC drilling and grade control has been included in the processing cost. This was $0.70/t heap/mill feed. The total LOM operating cost per tonne of material moved is $2.40/t moved. This was determined with the base operating costs by pit area and incremental costs as shown in Table 21-19. Table 21-19: Open pit mine base operating costs by pit area ($/tonne total material moved) Bigar Hill Korkan Korkan West Cost category Feed Waste Feed Waste Feed Waste General mine and engineering 0.19 0.19 0.19 0.19 0.19 0.19 Drilling 0.44 0.44 0.44 0.44 0.44 0.44 Blasting 0.42 0.42 0.42 0.42 0.42 0.42 Loading 0.27 0.27 0.27 0.27 0.27 0.27 Hauling 0.60 0.33 0.85 0.25 0.88 0.28 Support 0.49 0.49 0.50 0.50 0.59 0.59 Total 2.41 2.14 2.67 2.07 2.79 2.19

21.10 Processing Operating Costs

21.10.1 Heap Leach Facility The process plant operating cost is estimated at $10.2 million per annum, or $5.24/t crushed material stacked on the heap leach, which is made up of $4.10/t for processing, $0.44/t for haulage and placement of the feed on the pad and $0.70/t for grade control costs. A full breakdown of costs by type is provided in Table 21-20. Table 21-20: Operating cost estimate summary breakdown Cost Area $‘000 per annum $ per tonne Labour 3,997 1.60 Consumables 2,341 0.94 Power 1,389 0.56 Plant maintenance 2,500 1.00 Subtotal – Processing cost 10,228 4.10 Trucking – crushed feed to heap pad 1,100 0.44 Grade control 1,750 0.70 TOTAL – Processing cost to the heap pad 13,078 5.24

Labour Labour costs were calculated using typical plant staffing levels for a plant of this size. Pay scales were based on recent database rates, with adjustments for Serbian salaries. The plant shift schedule assumes a complement of four shifts with two shifts at site and two shifts off at all times. Table 21-21 shows the estimated labour breakdowns. Table 21-21: Labour cost summary Area Total persons Total cost ($/annum) Plant Operations 55 2,637,732 Maintenance 24 1,048,564 Laboratory 7 310,727 Total 86 3,997,023

CSA Global Report Nº R302.2019 214 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

The HLF is estimated to require a total complement of 89 persons at an annual cost of roughly $4 million, or $1.60/t.

Power Power is a significant element of the total operating cost, accounting for roughly 15% of the overall total. A summary of the total cost calculation for the heap leach operations is presented in Table 21-22. Table 21-22: Plant power cost estimate Item Unit Value Total connected power kW 3,250 Load factor % 82 Estimated power consumed kW 2,673 Annual running time (crusher) hours 6,570 Annual running time (ADR) hours 8,585 Annual consumption MWh 19,853 Cost per KWh $ 0.07 Total cost $‘000 1,390

The total connected power is taken from the mechanical equipment list, with load factors applied for each piece of equipment. A power supply rate of $0.07/kWh has been estimated for site power. This equates to an annual operating cost of $0.56/t of mill feed treated.

Consumables Consumable costs were estimated using unit costs from vendors and consumption rates from the laboratory testwork. The total consumable costs amount to $0.94/t of mill feed. A summary of the consumable costs by area is presented in Table 21-23. Table 21-23: Summary of estimated consumable operating costs Process area Annual cost (‘000 $) Crushing and screening 488 Heap leach 1,529 ADR 132 Gold-room 43 Laboratory 150 Total 2,342

A unit cost of $0.44/t was added for transportation of the crushed material to the heap with trucks.

Maintenance and Supplies An allowance for plant maintenance was factored from the mechanical supply cost for each area. The cost includes transportation but labour costs for replacement, installation, and maintenance are included in the labour allowances shown earlier in this section. Similar amounts were factored from the overall mechanical supply cost to cover electrical/instrumentation maintenance. The budget for maintenance and supplies totals $2,500,000 per annum, or $1.00/t milled.

Trucking – Crushed Feed to Heap Pad An analysis was completed on the best method to place material on the heap leach pad. This was a trade-off between traditional grasshopper conveyors and small truck placement. When the analysis was complete the

CSA Global Report Nº R302.2019 215 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

smaller trucks provided the lowest capital cost (initial and sustaining) as well as lower operating cost. For the PEA, the use of the smaller trucks allows easier development of cells within the leach to maximise recovery by material type. This trade-off should be examined in future studies to determine if compaction from the trucks will present problems but appears at this level of study to provide a viable option. For the PEA, this cost was estimated at $0.44/t of feed.

Grade Control Feed segregation by material type and grade will be key to maximising the recovery potential of the Timok deposits. Separation into oxide, transitional and sulphide in the field will assist the process team in daily operation. Grade control in the pit using dedicated RC drill rigs has been captured as a cost in this category. It is estimated that this will cost $0.70/t of feed including the assays.

21.10.2 Mill/Flotation Plant The process plant operating cost is estimated at $8.2 million per annum, or $16.5/t sulphide mineralisation processed. A full breakdown of costs by type is provided in Table 21-24. Table 21-24: Operating cost estimate summary breakdown Cost Area $‘000 per annum $ per tonne Labour 3,012 6.02 Power 1,160 2.32 Consumables 3,558 7.12 Plant maintenance 500 1.00 Total 8,230 16.46

Labour Labour costs were calculated using typical plant staffing levels for a plant of this size. Pay scales were based on recent database rates, with adjustments for Serbian salaries. The plant shift schedule assumes a complement of four shifts with two shifts at site and two shifts off at all times. Table 21-25 shows the estimated labour breakdowns. The mill is estimated to require a total complement of 69 persons at an annual cost of roughly $3 million, or $6.02/t. Table 21-25: Plant labour cost summary Area Total persons Total cost ($/annum) Plant Operations 40 1,712,139 Maintenance 18 824,134 Laboratory 8 351,699 Total 66 2,887,972

Power Power is a significant element of the total operating cost, accounting for roughly 14% of the overall total. A summary of the total cost calculation is presented in Table 21-26.

CSA Global Report Nº R302.2019 216 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Table 21-26: Mill / Flotation Plant power cost estimate Item Unit Value Total connected power kW 2,076 Load factor % 88 Estimated power consumed kW 2,097 Annual running time (crusher) hours 6,570 Annual running time (concentrator) hours 8,059 Annual consumption MWh 16,704 Cost per kWh $ 0.07 Total $’000 1,160

The total connected power is taken from the mechanical equipment list, with load factors applied for each piece of equipment. A power supply rate of $0.07/kWh has been estimated for site power. This equates to an annual operating cost of $2.32/t of mill feed treated.

Consumables Consumable costs were estimated using unit costs from vendors and consumption rates from the laboratory testwork. A summary of the consumable costs by area is presented in Table 21-27. Table 21-27: Summary of estimated consumable operating costs Reagent consumption Annual cost ($‘000) Mineralised feed re-handling/crushing 510 Grinding and classification 1,945 Flotation and concentrate regrind 356 Thickening and filtration 272 Laboratory 225 Miscellaneous 250 Total 3,558

The total consumable cost amounts to $7.12/t of mill feed.

Maintenance and Supplies An allowance for plant maintenance was factored from the mechanical supply cost for each area. The cost includes transportation but labour costs for replacement, installation, and maintenance are included in the labour allowances shown earlier in this section. Similar amounts were factored from the overall mechanical supply cost to cover electrical/instrumentation maintenance. The budget for maintenance and supplies totals $500,000 per annum, or $1.0/t milled.

21.11 Tailings Management Operating Costs The tailings management cost is included in the process operating cost. The sulphide circuit will be generating material that will be filtered and placed on the heap leach pad for long-term disposal.

21.12 General and Administrative Operating Costs The G&A costs were based on $1.60/t processed or $4.8 million per year. This cost covers expected costs associated with the General Manager and other Administration personnel are included. Offices in the local area and their costs are also part of the G&A cost calculation.

CSA Global Report Nº R302.2019 217 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

22 Economic Analysis

22.1 Caution to the Reader Readers are cautioned that the PEA is preliminary in nature. It includes Inferred Mineral Resources considered too speculative geologically to have the economic considerations applied to them that would enable them to be categorised as Mineral Reserves, and there is no certainty the PEA will be realised. Mineral Resources that are not Mineral Reserves do not have demonstrated economic viability. The estimate of Mineral Resources may be materially affected by environmental, permitting, legal, title, socio-political, marketing or other relevant issues.

22.2 Model Assumptions To analyse the economic potential of the Timok deposit, CSA Global created a discounted cash flow (DCF) model for the Project based on the mining inventory developed during this PEA as described in Section 16. The DCF model was developed in Microsoft Excel to analyse the economic potential, including: • Total revenue. • Operating and capital costs. • Mining royalties and corporate taxes. • Doré bar transportation, gold refining and TC/TR charges. • The internal rate of return (IRR). • Pre and post-tax cash flow. • Net present values (NPVs) at various discount rates. In addition, the model calculates: • The period required to repay the initial capital investment. • The operating cost per ounce of gold sold. • The all-in sustaining cost. • The all-in cash cost, IRRs, cash flows (pre-tax and post-tax) and NPVs at higher and lower gold prices and operating and capital costs. The underlying assumptions and parameters for the DCF include: • All units of measurement are metric unless otherwise stated. • All dollars are US$ unless otherwise stated. • No inflation or escalation is assumed (i.e. all dollars are real 2019 US$). • The gold price is aligned with the three-year average of the London Gold Market closing gold price. CSA Global has developed two mining schedules for the Timok open pit project. There are the 2.5 Mt of oxide and transitional mineralised material per annum (2.5 Mt/a) for the heap leach and the 500,000 t per annum sulphide flotation concentrate producing case. The DCF model for the PEA is based on these two mining options combined. Table 22-1 summarises the project metrics for these PEA open pit mining scenarios.

CSA Global Report Nº R302.2019 218 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Table 22-1: PEA results summary Assumptions Units Gold price $/oz 1,250 Production Profile Total tonnes of mineralized material mined and processed Million tonnes 18.9 Total tonnes waste mined Million tonnes 49.7 Strip ratio waste:feed 2.6:1 Head grade g/t Au 1.36 Peak tonnes per day mineralized material mined Tonnes 8,219 Average gold recovery % 81.5 Total gold ounces mined Oz 826,000 Total gold ounces recovered Oz 673,000 Average annual gold production Oz 75,000 Peak annual gold production Oz 132,000 Mine life Years 9 Unit Operating Costs LOM average cash cost $/oz Au 618 AISC(1) $/oz Au 717 Project Economics Royalties % 5.0 Average annual EBITDA $M 47 Pre-tax NPV 5% / After-tax NPV 5% $M 108 / 105 Pre-tax NPV 7.5% / After-tax NPV 7.5% $M 78 / 75 Pre-tax IRR / After-tax IRR % 18.9 / 18.6 Undiscounted operating pre-tax cash flow / after-tax cash flow $M 195 / 191 After-tax payback period Years 4.1 (1) All-in sustaining cost per ounce of gold represents mining, processing and site general and administrative costs, royalty, offsite costs and sustaining capital expenditures, divided by payable gold of 661,000 ounces. The DCF model allows for a three-year pre-production period in which to construct the processing plant, complete other required surface infrastructure including land acquisition and the initial waste stripping of the open pit. Once mining commences, the Project has a nine-year mine life. CSA Global has estimated metal recovery of gold based on metallurgical testwork available to date. Once commissioned, the leaching plant will produce gold doré bar and the sulphide flotation plant will produce a gold concentrate which can be treated at a nearby smelter/refinery plant to produced fine gold bullion which can be marketed on LME.

22.3 Metal Pricing At the end of January 2019, the spot gold price was $1,320 per troy ounce, the unadjusted three-year average gold price was $1,259 per troy ounce and after adjusting for inflation, the three-year trailing average gold price was $1,288 per troy ounce. Figure 22-1 and Figure 22-2 show the historical monthly gold price and the three-year moving average on both an uninflated basis and in 2019 US$.

CSA Global Report Nº R302.2019 219 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Figure 22-1 Monthly gold price and three-year moving average before inflation Source CSA Global 2019 - based on data sourced from London Bullion Market Association

Figure 22-2 Monthly gold price and three-year moving average in 2019 US$ Source: CSA Global 2019 - based on data sourced from London Bullion Market Association

22.4 Mine Production Summary The mining schedule consists of three open pit operations, Korkan, West Korkan and Bigar Hill. In addition, each mining operations contains three mineral types: oxide, transitional and sulphide. Table 22-2 summarises

CSA Global Report Nº R302.2019 220 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

the proposed mining schedules. CSA Global has estimated gold recovery at 66.8% based on metallurgical testwork available to date. The gold recovery varies by mineral type with an average of 88.1% in the oxide material, 69.3% in the transitional material and 75.0% in the sulphide material. The Timok Project will produce a gold product that will be send to a refinery for final processing. Table 22-2: Mineral Resources in PEA Mining Scenario Indicated Inferred Total Mineralised Material Grade Material Grade Material Grade Design Area Contained Recovered material type mined Mined mined Mined mined mined gold (oz) gold (oz) (ktonnes) (g/t) (ktonnes) (g/t) (ktonnes) (g/t) Oxide 3,098 0.81 103 0.57 3,201 0.80 82,500 75,500 Transitional 1,139 1.00 8 0.57 1,147 1.00 36,900 25,600 Korkan Sulphide 1,067 2.09 1,067 2.09 71,800 53,900 Subtotal 5,304 1.11 111 0.57 5,415 1.10 191,200 155,000 Oxide 1,881 1.18 364 0.93 2,245 1.14 82,300 59,900 Transitional 196 0.88 26 0.76 222 0.87 6,200 4,300 West Korkan Sulphide 10 1.43 10 1.43 400 300 Subtotal 2,087 1.16 390 0.92 2,477 1.12 88,900 64,500 Oxide 6,714 1.35 275 0.70 6,989 1.33 299,500 273,500 Transitional 1,569 1.49 44 0.78 1,613 1.48 76,500 53,000 Bigar Hill Sulphide 2,398 2.20 3 3.00 2,401 2.20 169,900 127,400 Subtotal 10,681 1.56 322 0.73 11,003 1.54 545,900 453,900 Oxide 11,693 1.18 742 0.79 12,435 1.16 464,300 408,900 Transitional 2,904 1.26 78 0.75 2,983 1.25 119,600 82,900 Total Sulphide 3,475 2.16 3 3.00 3,477 2.17 242,100 181,600 TOTAL 18,072 1.38 823 0.80 18,895 1.36 826,000 673,400

22.5 Operating and Capital Cost Summary Operating and capital costs have been estimated by CSA Global based on variety of sources such as benchmark rates, data from CSA Global databases of similar projects, indicative quotes and on first principles. The results of this work are summarised in the Table 22-3 and Table 22-4 and are reflected in the DCF model. Table 22-3: Project operating costs $/tonne of LOM $/oz Au $/tonne of Operating Costs(1) Oxide & ($ million) Recovered Sulphide Feed Transitional Feed Mining costs 162 245 9 9 Processing costs 138 209 5 16 G&A costs 31 47 2 2 Cash Costs 331 501 15 27 Royalty (5% NSR to Serbian Gov’t) 40 60 2 3 Offsite costs (Treatment and Refining Charges) 38 57 0 10 Total Cash Costs 409 618 18 39 Sustaining capital 65 99 3 3 AISC(2) 474 717 21 43 (1) Due to rounding, some columns may not total exactly as shown. (2) All-in sustaining cost per ounce of gold represents mining, processing and site general and administrative costs, royalty, offsite costs and sustaining capital expenditures, divided by payable gold of 661,000 ounces.

CSA Global Report Nº R302.2019 221 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Table 22-4: Project total capital costs Initial Sulphide Sustaining Capital LOM Capital Costs ($million) Processing ($million) ($million) ($ million) Mining 35 – 5 40 Processing 33 30 16 80 Infrastructure 34 – 4 38 Total Direct Costs 103 30 25 158 Indirect & Owner's Costs 18 – 18 35 Total Indirect Costs 18 – 18 35 Contingency 15 – 13 28 Reclamation – – 10 10 Total Capital 136 30 65 232 Note: • Mining includes $2.206 million in pre-production stripping. Infrastructure includes $10.7 million in land acquisition cost. • The model calculates sustaining capital as 1% of initial capital costs. • Reclamation is $10 million and is expended in the last two years of the mine life and the year after the mine is closed. • Processing capital has been split into oxide processing plant initial capital and flotation facility sustaining capital. • The model includes 12.5% overall contingency allowance for unknown or omitted costs.

22.6 Gold Doré and Concentrate Transport, Refining, Insurance and Treatment Charges The Microsoft Excel model assumes that all Timok gold will be both doré and concentrate produced on site (minimum 30 g/t) which will be subsequently shipped to a nearby facility for treatment, refining and preparation for bullion market in London. CSA Global has assumed the following charges based on industry norms with the following terms: • Transportation to the smelter – $10 per wet metric tonne (DMT). • Concentrate smelting – $120 per dry metric tonne (DMT). • Refining losses – 1% per oz contained in concentrate. • Gold doré treatment and refining cost – $14.00 per oz of payable gold. • Gold concentrate payment – deduct 4% per DMT and pay for 100% of the remaining gold at the current London Bullion PM closing price.

22.7 Royalties and Taxation The Government of Serbia imposes a 5% royalty on the NSR. The royalty is deductible from taxable income. The corporate tax rate in Serbia is 15% of taxable income. However, the government also provide a series of incentives to try and attract foreign capital. Corporate taxes are calculated as EBITDA, less: • Accelerated depreciation using the declining balance method (15%). • Taxable income losses may be carried forward up to five years. Tax losses may not be carried back. • The Government of Serbia provides a 10-year tax credit for companies that invests over 1 billion Serbian Dinars (approximately US$9.6 million) and employs an additional 100 new employees. CSA Global has been advised that the Timok Gold Project could receive this credit in the form of a reduction of income tax owing in the range of 90% to 100%. A tax credit equivalent to a 91% tax reduction has been used in this model.

CSA Global Report Nº R302.2019 222 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Serbia imposes a 20% value-added tax (VAT). The model assumes the gold production will be exempt from VAT and that the Project will pay VAT on half of its operating costs. The model further assumes the VAT payments will be refunded within four weeks.

22.8 Working Capital and Cash Flow Treatment Working capital represents the money required to fund the operations until the revenues generated by the gold sales are received. The model calculates Working Capital as equal to Gold Inventory plus Accounts Receivable, Spare Parts and Supplies, and VAT Receivable less Accounts Payable. Working Capital is recaptured at the end of the mine life. The parameters used in calculating working capital are: • Gold inventory – two weeks of bullion at the sales price.

• Accounts Receivable – six weeks of gold revenue. This allows for the time in transit as well as the delay in payment from the gold refinery. • Spare parts and supplies – 2% of initial capital or approximately US$2.8 million are budgeted annually. • VAT refund receivable. As per the taxation law VAT should be refunded by the Government of Serbia in four weeks. • Accounts payable – is assumed to be four weeks of one half of operating costs. The post-tax cash flow is calculated as gold revenue less: • Operating costs. • Serbian Mining Royalty. • Corporate taxes. • Capital expenditures. • Changes in working capital.

22.9 Results Based on the schedule and the items outlined in the preceding sections, the DCF model calculates the IRR, and the NPV of the cash flow at 5%, 7.5%, 10% and 15% discount rates. All NPVs are discounted to the mid- year. The model also calculates the “Payback Period” (the time required for the Project repay the initial capital) with the same discount rates. The model also calculates the cash operating cost per ounce and per tonne of material mined/processed and the All-In Sustaining Cost. The model calculates these variables on: • A pre-tax basis. • A post-tax basis Note: CSA Global has used the World Gold Council definitions of Operating Costs, All-in-Sustaining Costs and All-in Costs. In the current project, Operating Costs include all operating costs. All-In-Sustaining Costs include operating costs plus sustaining capital. Finally, All-In Costs include operating costs, initial capital, and sustaining capital. Table 22-5, Table 22-6 and Table 22-7 provide the key metrics for the Project from the DCF model.

CSA Global Report Nº R302.2019 223 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Table 22-5: EBITDA and net profit to Project with tax credit

Economic results Total ($ M)

EBITDA $426.65 Less: Book depreciation $231.60 Serbia corporate taxes $3.99 Net profit after taxes $191.06

Table 22-6: Project pre-tax results summary

Pre-tax Total ($ M)

IRR 18.9% Undiscounted pre-tax cashflow $195.06 NPV at 5% $108.37 NPV at 7.5% $77.92 NPV at 10% $53.62 NPV at 15% $18.56

Table 22-7: Project post-tax results with tax credit summary

Post-tax with tax credit Total ($ M)

IRR 18.6% Undiscounted post tax $191.06 cashflow NPV at 5% $105.49 NPV at 7.5% $75.46 NPV at 10% $51.50 NPV at 15% $16.96

A DCF model was completed for the Project based on the various inputs and costs outlined in this document. The cash flow model summary is presented in Table 22-8.

CSA Global Report Nº R302.2019 224 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Table 22-8: Summary DCF model

CSA Global Report Nº R302.2019 225 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

22.10 Sensitivity Analysis To better understand the economic viability of the Project, CSA Global has undertaken a sensitivity analysis of the Timok Project. Figure 22-3 to Figure 22-6 below chart the sensitivity of the Project’s pre-tax and post- tax IRR and NPV discounted at 5% to changes in gold prices and capital and operating costs. Gold prices and costs were varied from -30% to +30%.

Figure 22-3: Pre-tax sensitivity of Project IRR to changes in gold price and capital and operating costs Source CSA Global 2019

Figure 22-4: Pre-tax sensitivity of Project NPV discounted at 5% to changes in gold price and capital and operating costs (US$ M) Source CSA Global 2019

CSA Global Report Nº R302.2019 226 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Figure 22-5: Post-tax sensitivity of Project IRR to changes in gold price and capital and operating costs Source CSA Global 2019

Figure 22-6: Post-tax sensitivity of Project NPV discounted at 5% to changes in gold price and capital and operating costs (US$M) Source CSA Global 2019 As would be expected, the Project is most sensitive to metal prices, followed by operating costs and finally capital costs. The Timok Project is robust at a base gold price of $1,250/oz. Even a 20% reduction in metal prices produces a positive post tax cashflow of $36.1 million (Table 22-9).

CSA Global Report Nº R302.2019 227 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Table 22-9: Sensitivity of Project at a base gold price of $1,250/oz ±30% post-tax with a 5% Serbian gold royalty Project (post tax) Change Base Value IRR Cash Flow ($M) NPV (@5%) -30% -30% -5.3% -$41.41 -$61.48 -25% -0.3% -$2.66 -$33.65 -20% -20% 4.1% $36.08 -$5.82 -15% 8.2% $74.83 $22.01 -10% -10% 11.9% $113.57 $49.84 -5% 15.3% $152.32 $77.66 Base +100% 18.6% $191.06 $105.49 +5% 21.7% $229.81 $133.32 +10% +10% 24.6% $268.55 $161.14 +15% 27.4% $307.29 $188.97 +20% +20% 30.1% $346.04 $216.80 +25% 32.6% $384.78 $244.63 +30% +30% 35.1% $423.53 $272.45

22.11 Operating Costs per Ounce of Gold Finally, the model calculates the Operating Cost per ounce of gold using the World Gold Council definitions of Operating Cost. CSA Global has used the World Gold Council definitions of Operating Costs, All-in-Sustaining Costs and All-in Costs. In the current project, Operating Costs include all operating costs. All-in-Sustaining Costs include operating costs plus sustaining capital. Finally, All-in Costs include operating costs, initial capital, and sustaining capital less by-product credits. Table 22-10: Operating cost per ounce of gold (World Gold Council definition) Cost area Total $/oz Au Operating costs Gold ounces production 661,218 Operating costs (mining etc.) $331,182,346 $500.87 Royalties and production taxes $39,886,128 $60.32 Smelting and refining costs $37,883,657 $57.29 Total Operating Costs $408,952,131 $618.48 All-in Sustaining Costs (includes Sustaining Capital but not Initial Capital) Operating costs $408,952,131 $618.48 Plus: Sustaining capital $65,167,941 $98.56 All-in Sustaining Costs $474,120,072 $717.04

CSA Global Report Nº R302.2019 228 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

23 Adjacent Properties

Whilst other companies are actively exploring in the area to develop potential projects, there are no adjacent properties that are considered material to the Project.

CSA Global Report Nº R302.2019 229 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

24 Other Relevant Data and Information

There is no other relevant information on the Timok Gold Project known to the authors that would make this report more understandable or if undisclosed would make this report misleading.

CSA Global Report Nº R302.2019 230 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

25 Interpretations and Conclusions

Readers are cautioned that the PEA is preliminary in nature. It includes Inferred Mineral Resources considered too speculative geologically to have the economic considerations applied to them that would enable them to be categorised as Mineral Reserves, and there is no certainty the PEA will be realised. Mineral Resources that are not Mineral Reserves do not have demonstrated economic viability. The estimate of Mineral Resources may be materially affected by environmental, permitting, legal, title, socio-political, marketing or other relevant issues.

25.1 Geology and Resources CSA Global has the following conclusions that are relevant to this study: • Avala Resources d.o.o., a wholly owned subsidiary of DPM, has conducted exploration and drilling since 2006 at the Timok Gold Project, which now comprises four exploration licences in the region of Bor, Serbia. • The Bigar Hill, Korkan, Korkan West and Kraku Pester sediment-hosted gold deposits have been defined as a result of a systematic sequence of exploration activities from soil sampling, trenching, and mapping, through geophysical evaluation and structural and stratigraphic interpretation, RC and diamond core drilling, metallurgical testwork and, finally, estimation of Mineral Resources. • CSA Global has reviewed procedures, visited the site, viewed core, verified the locations of several drill holes, conducted spot checks between hard copy data and digital data, reviewed QAQC results and had extensive discussions with site personnel as part of the data verification work. CSA Global has found the site to be extremely well run, with excellent procedures, a good understanding of the deposit geology and an emphasis on data quality that has contributed to a high degree of confidence in the data used in the MRE. • Drilling at Bigar Hill and Korkan have served to confirm the structural setting, the stratigraphy, and the geometric, spatial and lithological relationships of the gold mineralisation. The controls on the mineralisation at a local (sample interval) level remain less well understood, and this translates into uncertainties regarding the estimates of gold at the mining scale. This local uncertainty is unlikely to be material in an open pit mining scenario, with a relatively low level of mining selectivity. The level of uncertainty will likely increase under circumstances where cut-off grades are raised and where more selective mining regimes are applied. • This Technical Report includes the 2018 MRE update with an effective date of 15 May 2018 and first reported on 7 November 2018 (CSA Global, 2018). The 2018 MRE has been updated from the 2017 MRE on the basis of revised oxidation profiles informed by bottle roll testwork and revised metallurgical parameters based on this work. • The 2018 MRE update includes Inferred and Indicated Mineral Resources. Mineral Resources that are not Mineral Reserves do not have demonstrated economic viability. There are no Mineral Reserves defined over the Project. • Most of the Mineral Resources estimated at the Timok Gold Project are Indicated Mineral Resources supported by good geological knowledge, drill coverage, robust standard operating procedures and data quality, and have been classified under the guidelines of the CIM Standards on Mineral Resources and Reserves, Definitions and Guidelines prepared by the CIM Standing Committee on Reserve Definitions and adopted by CIM Council, and reported in accordance with the Canadian Securities Administrators NI 43-101.

CSA Global Report Nº R302.2019 231 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

25.2 Geotechnical A previous geotechnical assessment of the Project was completed in 2013. The wall angles proposed were based on pit wall heights of 230 m but the final pits for this study only had a small portion of the pit perimeter where the pit depth exceeded 200 m. Instances were found in Bigar Hill where poorer quality rock was overlain by more competent material. This could cause instability as the pit deepens and will need further study in order to verify the overall pit slope angles to be used in the PFS. Based on the previous review, the analysis showed that for the planned pit depths, and a factor of safety of 1.2, overall slope angles can be in the region of 45° to 52.5°. The oxidised zone varies by pit area and requires an inter-ramp slope of 45°. Additional geotechnical-focused drilling should be undertaken to provide a clearer picture of wall slopes by sector and deposit as well as hydrogeological information.

25.3 Mining The PEA is based solely on a scenario of open pit mining of three deposits: Bigar Hill, Korkan and Korkan West. Three pits designs were developed for the three areas. The Bigar Hill pit has three phases, Korkan has three distinct pit areas in close proximity and Korkan West is a single phase. These provide an estimated total of 18.9 Mt of run-of-mine (ROM) process feed material grading 1.36 g/t gold, and an estimated 49.7 Mt of waste, for a delivered strip ratio of 2.63:1 (waste:feed). The heap leach feed marginal cut-off grades varied by pit area due to metallurgical response and material type. The oxide material cut-off grades are 0.19 g/t for Bigar Hill and Korkan, and 0.24 g/t for Korkan West. The transitional material cut-off grades were 0.25 g/t for all deposits. The sulphide concentrator feed material used a cut-off grade of 0.69 g/t for all deposits. The estimated 18.9 Mt/a of process feed material over an estimated nine-year mine life is comprised of 15.4 Mt of oxide and transitional material grading 1.18 g/t delivered to a heap leach pad, plus 3.5 Mt of sulphide material grading 2.17 g/t delivered to a small flotation concentrator facility. The mining fleet is comprised of three 200 mm rotary drills, two 6.7 m3 hydraulic excavators, two 13 m3 production loaders and twelve 63-t haulage trucks. Typical mine support equipment have been included, such as track dozers, graders and water trucks. Three waste dump locations are envisioned to accommodate the mine waste: one near Bigar Hill, one at Korkan and one adjacent to Korkan West. These facilities are designed to accommodate the full amount of waste and total 35.6 Mm3. The LOM mining operating cost is estimated from first principles to be $2.40 per tonne of material moved. Pre-production stripping costs of $2.2 million have been included in the initial capital cost. Initial mine capital costs total $35.3 million and sustaining capital costs total $4.8 million.

25.4 Mineral Processing Coarse sample bottle roll and column leach tests have been carried out on representative oxide samples representing material from the Bigar Hill, Korkan and Korkan West deposits. Gold recoveries for oxide material ranged from 82.1% to 94.4%, whilst gold recoveries for transitional material ranged from 60.3% to 67.9% (uncorrected for full scale heap leach).

CSA Global Report Nº R302.2019 232 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Laboratory leach cycle times were 65 days for oxide material, and 93 days for transitional material. Cyanide and lime consumptions for oxide material were on average 1.03 kg/t and 0.29 kg/t. Average cyanide and lime consumptions for transitional material were 0.79 kg/t and 0.59 kg/t respectively. Results of column leach tests showed that the oxide and transitional material types are amenable to processing using heap leach technology, at a fairly coarse crush size (80% passing 12.5 mm). Gold leach extraction appears to be dependent on sulphide sulphur content, and independent of crush size up to 1 inch (25 mm). Coarse sample bottle roll tests carried out on representative sulphide material resulted in low gold extraction ranging from 6% to 7%, at a crush size of 80% passing 12.5 mm. These results indicate the refractory nature of the sulphide material. Flotation tests carried out indicate that gold recoveries to a bulk sulphide concentrate ranged from 68% to 80%, at a concentrate grade ranging from 30 g/t Au to 50 g/t Au. At these concentrate grades, the gold- bearing concentrate would be deemed to be saleable to smelters/roasters.

25.5 Infrastructure and Site Layout The site infrastructure includes major items for the Project, such as: • Process plant. • Waste rock storage facilities. • Mine shop facilities. • Raw water supply. • Power line tie-in. • Access roads. • Heap leach facility. The process plant is envisioned to be located to the east of Bigar Hill, on a flatter area of ground and west of the proposed heap leach facility. Mine shop facilities are located north of the process plant facilities in order to provide easy access to the pit. Water for the process plant will come from wells in the area and power will come from the main high voltage line 5 km to the east of the Project. Various roads need to be constructed for access to the Project. These include the main access road to the process plant and heap leach facility, as well as to the various pit areas. Korkan West will require the temporary placement of a culvert over the valley bottom to avoid disturbance to natural drainage. Filtered tailings from the 0.5 Mt/a sulphide plant will be stored on the heap leach facility. These tailings will be stored along the edges or in the back to avoid disrupting the oxide/transition leaching process.

25.6 Environmental, Permitting, Social and Community The permitting strategy for the Project will need to be flexible as the Serbian permitting system evolves to align with EU requirements. Potential environmental and social risks have been identified during this phase of the project, although none of these are considered likely to prevent the project progressing. Key risk mitigation measures are similar to those associated with other gold mining projects and include safeguarding rivers, groundwater and biodiversity and planning for land acquisition. Management of the use and transportation of cyanide will be particularly important, which DPM will do through the International Cyanide

CSA Global Report Nº R302.2019 233 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Management Code. Further assessment of environmental and social risks will be undertaken through the environmental impact assessment. Details of water management, mine closure and the potential for acid generating rock will be developed at the PFS stage.

25.7 Economic Analysis The LOM capital cost for the Project is estimated at US$231.6 million, with an initial capital expenditure of US$136.1 million, which includes US$35.4 million of capitalised pre-stripping. At a gold price of US$1,250/oz, the Project is estimated to have an after-tax IRR of 18.6% and an undiscounted payback period of 4.1 years from start of production. With a discount rate of 5%, the after-tax NPV is estimated at $105 million and an undiscounted payback period of 4.1 years from the start of production. The corporate income tax rate in Serbia is 15%; however, a 10-year tax credit is available to reduce corporate taxes by up to 100%. In practice, tax reductions in the 90% range are typical. The model assumes a tax reduction of 91% for the initial 10-year period in the DCF. Serbian corporate taxes are calculated as EBITDA, less: • Accelerated depreciation using the decline balance method with annual rates of 15% (i.e. nine years) for all capital investment. • All mining royalties are deductible. • Tax losses may be carried forward for five years, but tax losses may not be carried back to previous years. Serbia has a 20% VAT. The Project does not collect VAT on its production as all gold production is exported and is therefore VAT exempt. The Project however pays VAT on all supplies and services once production has begun. The model assumes that the Government of Serbia will refund all VAT expended by the Project within four weeks. Note that VAT is not refundable on capital expenditures. A significant item in the analysis are the royalties. The basic royalty rate for gold is 5% based on net payable ounces produced either as doré or concentrate net smelter payable. The Project is most sensitive to metal prices, followed by operating costs and, finally, capital costs.

25.8 Risks

25.8.1 General Environmental, permitting, legal, title, taxation, socio-economic, marketing, and political or other relevant issues could potentially materially affect access, title, or the right or ability to perform the work recommended in this report on the Project. However, at the time of this report, authors and Qualified Persons are unaware of any such potential issues affecting the Project and work programs recommended in this report.

25.8.2 Mineral Resource Estimate As noted in Section 14.2, the MRE could be affected by: • Future, yet unknown environmental, permitting, legal, title, taxation, socio-economic, marketing, political, or other relevant issues. • Metal price and valuation assumptions. • Changes to geological interpretation. • Changes to mining assumptions. • Changes to processing assumptions.

CSA Global Report Nº R302.2019 234 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

At a local level, many of the interpreted fault planes are inferred only and the extents of any faulted offsets is not clear at this stage. The local controls on mineralisation are less well understood, where there can be sharp changes in gold grade between samples, with no corresponding change in visual features. Given there are different processing methods for the oxide and sulphide mineralisation domains, any changes to the resolution of these domain boundaries will impact the Mineral Resource. In particular, the existing bottle roll assaying has been completed on 5m composites, adding to the uncertainty of the precise location of the oxide-sulphide boundaries. The interpreted fault structures were not explicitly incorporated into the mineralisation wireframes. Although faulting is implicitly expressed in some cases where both the lithological and mineralisation wireframes have been interpreted as flexing sharply downwards. Where drill spacing increases, the level of grade continuity is less certain. This has been accounted for in classification. The MRE has been completed using UC which results in an estimate of tonnage and grade about a set of cut- offs per 20 m x 20 m x 10 m panel. The model provided for further downstream work has been localised which results in a single grade per SMU (5 m x 5 m x 5 m). This is for ease of use in downstream processes. However, it is important to note that the estimates of SMUs can be considered reliable within a panel (taking classification into account), but the exact location of an individual SMU remains unknown until pre-production drilling.

25.8.3 Metallurgy/Mineral Processing • Selective mining of the different mineralisation types separately may not be fully achievable. • Adverse effect of degree of sulphidation on gold extraction. • Cold climate effect on leach kinetics. • Degree of clay content on percolation rates.

25.8.4 Mining • Possible flatter highwall angles especially where competent rock overlays weathered in Bigar Hill. • Skarn areas not previously identified under waste rock sites. • Additional recovery loss and dilution (due to slow results from sampling or analyses). • Elevated water tables requiring highwall dewatering.

25.8.5 Environment, Permitting, Social and Community Risks • Changes to the Serbian permitting system as part of alignment with EU requirements. • Additional environmental or social risks identified in PFS phase, as a result of further design development or baseline studies. • Stakeholder concerns slowing or stopping the permitting process.

CSA Global Report Nº R302.2019 235 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

25.9 Opportunities

25.9.1 Geology/Mineral Resource Estimate • A number of shallow gold oxide exploration targets within 5 km of the proposed plant have been identified. • With further infill drilling, there are opportunities to increase the confidence of Inferred material, with a view to converting to Indicated Mineral Resources. • If metal prices increase and support a deeper pit, there are opportunities to extend the resource inventory at depth. • Delays and costs associated with acquiring land, particularly if compulsory purchase is needed. Potential for acquisition to be unsuccessful.

25.9.2 Metallurgy/Mineral Processing The Project has potential to: • Reduce capital expenditure costs by adopting modular crushing plant vs static crushing plant. • Reduce operating expenditure costs by crushing to 1 inch (25 mm) with two-stage crushing. • Improve metallurgical assumptions for oxide and transitional mineralisation types by carrying out further column leach tests and populating the testwork results matrix. • Increase sulphide flotation recovery by further optimisation testing.

25.9.3 Mining The Project has potential to: • Lower drilling and blasting costs in weathered rock. • Further optimisation of pit shells to target additional sulphide material. • Improve delineation of leach pad feed and sulphide material for milling. • Detailed haulage profiles by year may provide lower costs in the initial years. • Possible use of used conveyors for heap leach placement. • Improve balance of mining schedule to process sulphide and oxide material concurrently to end of mine life.

25.9.4 Environment, Permitting, Social and Community • Micro-siting infrastructure to avoid karst-affected areas where possible. • Ongoing engagement with local community to maximise impact of employment and economic development. • Planning closure to provide positive ongoing legacy.

CSA Global Report Nº R302.2019 236 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

26 Recommendations

26.1 Introduction Based on the results of the PEA study, CSA Global recommends that DPM proceed forward with additional studies, including a PFS for its Timok Gold Project. Recommended work and associated budgets for this next phase of work are described in the sections below.

26.2 Geology and Resources Any high priority exploration targets that lie within 5 km of the proposed plant site be followed-up with exploration (e.g. the gold-silver porphyry-style mineralisation at Coka Rakita, sediment-hosted gold mineralisation at Bigar East, and the Korkan East-style mineralisation at Valja Saka). Relating to QAQC procedures, CSA Global recommends ongoing vigilance to ensure that standards and blanks are correctly identified and labelled. When drilling is taking place, it is recommended that a site visit be arranged to inspect drilling and sampling practices as they are occurring. On any future site visit, the drill sites at Korkan, Korkan West and Kraku Pester be inspected. It is recommended that sampling of core continue in 1 m increments but should break to honour geological boundaries to enable enhanced analysis and effective modelling of the contact, if sharp. Higher gold recoveries during the cyanide-gold leach bottle roll tests are mostly controlled by the oxidation degree of the arsenic and gold-rich pyrite. The usefulness of these data is limited by the fact that the composites are at 5 m intervals, which is in contrary to the logging data which is at 1 m intervals. Spatially, data is irregular and is not always available in the areas of interest. Future bottle roll testwork should be undertaken on shorter intervals (1 m or 2 m) with the selection of the intervals prioritised in mineralised areas. Undertake a geometallurgical study into the three main deposits to better define the oxide, transitional and sulphide material domains. Use these results of this work to build a geometallurgical model for the deposits and to update future MREs. This work should then be used to refine the mining schedule to maximise NPV. DPM’s planned 2019 exploration activities at the Timok Gold Project include infill soil sampling, geological mapping, trenching and up to 5,000 m of exploration drilling on near-resource targets, with the aim of increasing near-surface oxide Mineral Resources. In addition to these planned exploration activities, further drilling during 2019 of the existing Bigar Hill, Korkan and Korkan West Mineral Resources includes 7,000 m of infill drilling and 4,300 m of condemnation drilling. The author concurs with DPM’s exploration program and budget.

26.3 Geotechnical A comprehensive geotechnical and hydrogeological study of the Project will be required, consisting of: • Drilling focused on anticipated final wall positions. • Pit hydrogeological drilling of an estimated eight holes. These holes can be the same as some of the geotechnical holes estimated above. • Infrastructure geotechnical drilling in the area of the process plant, crusher, tailings facilities and waste dumps.

CSA Global Report Nº R302.2019 237 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

• DPM’s 2019 initial geotechnical drilling plan totals 2,200 m at a cost $550,000.

26.4 Mining Additional work is required to advance the mine plan to a PFS level: • An analysis of the impact of changes in the wall slope angles resulting from the geotechnical study noted above. This study is estimated to cost $150,000, in addition to the cost of the drilling, mentioned above. • A study of the impact of improved sulphide recoveries on overall pit shapes and sizes. The cost for this is included in the PFS estimate. • A study of the relative process throughput rates for heap leach and sulphide concentrator feed materials to optimise the mine life and reduce G&A costs. The cost for this is included in the PFS estimate. • A detailed analysis of pit dewatering requirements as part of site overall water balance. A budget of $200,000 is estimated for analysis, testing and reporting. • An analysis of the mineralised material to determine appropriate grade control procedures for classification of run-of-mine material as either waste, heap leach feed, or sulphide floatation feed materials. This analysis would include sample sizes, methodology of sample selection and assay procedures. This separate study is estimated to cost $75,000. • Equipment selection and sourcing. The cost for this is included in the PFS estimate. • Blasting analysis with the objective to reduce mining unit costs. The cost for this is included in the PFS estimate.

26.5 Mineral Processing and Metallurgy Based on the preliminary scoping tests carried out, further testing is required to establish: • The optimum crush size, and whether high gold extraction can be obtained at a coarse crush size of 1 inch (25 mm); for both oxide and transitional material. • Any variability in metallurgical performance associated with testing of different domain types. • The relationship between sulphide sulphur and gold leach extraction. Incorporate findings and results into the geometallurgical model. An estimate of the testwork cost is shown in Table 26-1. Table 26-1: Recommended metallurgical testwork budget Task Estimated cost (US$) Geometallurgical modelling 20,000 Variability testwork 350,000 Comminution 25,000 Tailings characterisation 5,000 Total 400,000

26.6 Infrastructure Additional studies are required with respect to: • Detailed surveys of the plant site locations, waste dumps, and road accesses. • The provision and cost for power to the site. • Waste storage locations.

CSA Global Report Nº R302.2019 238 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

• HLF detailed design. • Project water system. • Site-wide water balance. This work will also include incorporation of the geotechnical work discussed earlier into the designs. These studies, consultations and surveys are estimated to cost $400,000.

26.7 Environmental Continue with the program of baseline data collection. Progress with delivering the permitting plan. Keep environmental and social risk register up to date, develop/integrate with environmental and social management system. Continue regular engagement with stakeholders.

26.8 Estimated Budget A summary of the estimated costs per discipline of these recommendations for this next phase of work, leading up to the completion of a PFS, is shown in Table 26-2. Table 26-2: Summary of recommendation budgets for PFS Area Recommended budget (US$) Geology and drilling 3,500,000 Geotechnical 550,000 Mining 500,000 Metallurgy/Processing 400,000 Infrastructure 400,000 Environmental 250,000 PFS 1,000,000 Total $6,600,000

CSA Global Report Nº R302.2019 239 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

27 References

Abzalov, M., 2008. Quality control of assay data: A review of procedures for measuring and monitoring precision and accuracy. Exploration and Mining Geology, 17, 131-144. AMEC, 2013. Timok Gold Project, Serbia, Technical Report and Mineral Resources Estimates for Avala Resources Ltd. AMEC (UK) Report No. AMEC 413006, 14 October 2013. AMEC, 2013. Indicative Pit Slope Angles Assessment. AMEC Document Reference Proj. No. 7879130129; Doc. Ref. A129- 13-R2166. Avala Resources d.o.o. internal report. 4 December 2013. AMEC, 2014. Preliminary Economic Assessment and Updated Mineral Resource. May 1, 2014, filename: Technical_Report_Timok_PEA_SEDAR_20140715.pdf. CIM Standing Committee on Reserve Definitions. CIM Definition Standards for Mineral Resources and Mineral Reserves. May 10, 2014. Cline, J.S., Hofstra, A.H., Muntean, J.L., Tosdal, R.M., Hickey, K.A., 2005. Carlin-Type Gold Deposits in Nevada: Critical Geological Characteristics and Viable Models Critical Geologic Characteristics and Viable Models: Economic Geology, 100th Anniversary Volume, p. 468-471. CSA Global, 2017. NI 43-101 Technical Report. Timok Gold Project, Serbia, 31 March 2017, filename: 119.2017 DPMREV02 Dundee Timok NI 43-101.docx CSA Global, 2018, NI 43-101 Technical Report – Mineral Resource Estimate Update for the Timok Project, Serbia, dated 07 November 2018, effective 15 May 2018, p. 182.Knaak, M., Márton, I., Tosdal, R., van der Toorn, J., Davidović, D., Strmbanović, I., Zdravković, M., Živanović, J., Hasson, S., 2016, Geologic Setting and Tectonic Evolution of Porphyry Cu-Au, Polymetallic Replacement, and Sedimentary Rock-Hosted Au Deposits in the Northwestern Area of the Timok Magmatic Complex, Serbia: Economic Geology, Special Publication 19, p. 1-28. Maygar, L., 2018. Petrographic study of gold-rich oxide ore materials (and related rocks) from Timok NW drill cores and from the surface. DPM Internal Report, 76 p + 2 appendixes with SEM-EDX results. Ministry of Energy and Mines, Health, Safety and Reclamation Code for Mines in British Columbia, revised June 2017. Pačevski, A., 2012a. Mineralogical study of the 13 samples from the Bigar Hill area (drill-holes: BHDD029, BHDD044, BHDD060, BHDD068). Avala Resources Ltd internal report, p26. Pačevski, A., 2012b. Mineralogical study of the 19 samples from the Korkan area (drill-holes: KODD044, KODT075, KODD085, KODD096, KODD133). Avala Resources Ltd internal report, p67. Pačevski, A., 2013. Mineralization in the Korkan and Bigar Hill areas: final conclusions and discussion. Avala Resources Ltd internal report, p9. SGS UK, 2012a. An investigation by High Definition Mineralogy into the mineralogical characteristics of three samples from Avala Deposit. Avala Resources Ltd internal report, p60. SGS UK, 2012b. An investigation by High Definition Mineralogy into the mineralogical characteristics of eight metallurgical products from the Timok Gold Project, Republic of Serbia, Final Report. Avala Resources Ltd internal report, p121. SGS UK, 2013. An investigation by High Definition Mineralogy into the mineralogical characteristics of four metallurgical products from the Timok Gold Project, Republic of Serbia, Final Report. Avala Resources Ltd internal report, p121. Stanley, C.R., and Lawie, D., 2007. Average relative error in geochemical determinations: Clarification, calculations and a plea for consistency. Exploration and Mining Geology, 16, 267–275. Vasić, N., 2012. Sedimentology of Jurassic and Cretaceous deposits – Bigar Hill: Avala Resources Ltd., Unpublished report, p. 1-97

CSA Global Report Nº R302.2019 240 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

APPENDIX 1: Glossary of Technical Terms and Abbreviations

Abbreviation Description ° degrees °C degrees Celsius % percent $/a dollars per annum $/t dollars per tonne $/tr oz dollars per troy ounce µm microns 3D three-dimensional AA atomic absorption AAS atomic absorption spectroscopy ABCD Alpine-Balkan-Carpathian-Dinaride metallogenic-geodynamic province ABTSB Apuseni-Banat-Timok-Srednogorie Belt ADR adsorption-desorption-recovery Ag silver AGP AGP Mining Consultants Inc. AMEC AMEC Australia Pty Ltd ARD acid rock drainage As arsenic Au gold Avala Avala Resources Ltd azimuth Drill hole azimuth deviation (from north) BH Bigar Hill binary Digital file containing characteristics readable by computer only Ca calcium CIC carbon-in-column CIL carbon-in-leach CIM Canadian Institute of Mining, Metallurgy and Petroleum cm centimetre(s) CN cyanide collar Geographical coordinates of the collar of a drill hole or a working portal compositing In sampling and resource estimation, process designed to carry all samples to certain equal length core sampling In exploration, a sampling method of obtaining mineralised material or rock samples from a drill hole core for further assay COS crushed feed stockpile CRM certified reference material CSA Global CSA Global (UK) Ltd CSV Digital computer file containing comma-separated text data

CSA Global Report Nº R302.2019 241 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Abbreviation Description Cu copper cut-off grade The threshold value in exploration and geological resources estimation above which mineralised material is selectively processed or estimated CV coefficients-of-variation (In statistics, the normalised variation value in a sample population) DA dynamic anisotropy Datamine StudioRM™ Software product for resource estimation and the mining industry DD diamond drill DCF discounted cash flow de-clustering In geostatistics, a procedure allowing bounded grouping of samples within the octant sectors of a search ellipse digital terrain model (DTM) Three-dimensional wireframe surface model (e.g. topography DTM) dip Angle of drilling of a drill hole dmt dry metric tonne DPM Dundee Precious Metals Inc. D-SIMS dynamic secondary ions mass spectrometry DST Dundee Sustainable Technologies E (X) easting EBIT earnings before interest and taxes EBITDA earnings before interest, taxes, depreciation and amortisation EBRD European Bank for Reconstruction and Development Elaborat Survey or report (Serbian) EPCM engineering, procurement and construction management EU European Union EUR Euro FA fire assay Fe iron FEL front-end loader flagging Coding of cells of the digital model FROM Beginning of intersection FYRM Former Yugoslav Republic of Macedonia G&A General and Administration g gram(s) g/L grams per litre g/m³ grams per cubic metre g/t grams per tonne geochemical sampling In exploration, the main method of sampling for determination of presence of mineralisation. A geochemical sample usually unites fragments of rock chipped with a hammer from drill hole core at a specific interval geometric mean The antilog of the mean value of the logarithms of individual values. For a logarithmic distribution, the geometric mean is equal to the median. For a logarithmic distribution, the geometric mean is equal to the median GET ground-engaging tool GHG greenhouse gas

CSA Global Report Nº R302.2019 242 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Abbreviation Description GPS global positioning system hr hour(s) ha hectare(s) HARD half absolute relative difference HDPE high-density polyethylene histogram Diagrammatic representation of data distribution by calculating frequency of occurrence HLF heap leach facility HQ3 Size of diamond drill rod/bit/core ICP-AES inductively coupled plasma - atomic emission spectrometry ICMM International Council on Mining and Metals ICP-MS inductively coupled plasma - mass spectrometry IFC International Finance Corporation IRR internal rate of return ISO International Standards Organisation JLS Jurassic Limestone K potassium kg kilogram(s) kg/t kilogram(s) per tonne km kilometre(s) km2 square kilometres KLS Cretaceous Limestone Kriging Method of interpolating grade using variogram parameters associated with the samples’ spatial distribution. Kriging estimates grades in untested areas (blocks) such that the variogram parameters are used for optimum weighting of known grades. Kriging weights known grades such that variation of the estimation is minimised, and the standard deviation is equal to zero (based on the model) KE kriging efficiency KO Korkan KP Kraku Pester KW Korkan West kV kilovolt kW kilowatts L litre L/m2/hr litres per square metre per hour lag The chosen spacing for constructing a variogram LDPE low-density polyethylene lognormal Relates to the distribution of a variable value, where the logarithm of this variable is a normal distribution LOM life-of-mine LUC localised uniform conditioning M million(s) m metres m3/hr cubic metres per hour

CSA Global Report Nº R302.2019 243 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Abbreviation Description m3/s cubic metres per second m3/t cubic metres per tonne Ma million years macro A set of Datamine StudioRM™ commands written as a computer program for reading and handling data masl metres above sea level mean Arithmetic mean median Sample occupying the middle position in a database MES Mechanical Equipment Supply Mg magnesium MIBC metyl isobutyl carbinol ml millilitre(s) ML Mining Licence mm millimetre(s) Mm3 million cubic metres Mo molybdenum MoM&E Ministry of Mining and Energy MRE Mineral Resource estimate Mt million tonnes Mt/a million tonnes per annum MW megawatt N (Y) northing Na sodium NAG non-acid generating Ni nickel NI 43-101 National Instrument 43-101 Standards of Disclosure for Mineral Projects NPC net present cost NPV net present value NQ size of diamond drill rod/bit/core NQ2 size of diamond drill rod/bit/core NSR net smelter return OCC open cycle cleaner OK ordinary kriging PAX potassium amyl xanthate PCU particle collection unit PEA Preliminary Economic Assessment percentile In statistics, one one-hundredth of the data. It is generally used to break a database down into equal hundredths PFS prefeasibility study PLS pregnant leach solution population In geostatistics, a population formed from grades having identical or similar geostatistical characteristics. Ideally, one given population is characterised by a linear distribution

CSA Global Report Nº R302.2019 244 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Abbreviation Description POX pressure oxidation ppm parts per million probability curve Diagram showing cumulative frequency as a function of interval size on a logarithmic scale Project Timok Gold Project psi pounds per square inch PVC poly vinyl chloride QAQC quality assurance/quality control QC quality control QEMSEM Quantitative Evaluation of Minerals by scanning electron microscopy Q-Q quantile-quantile quantile In statistics, a discrete value of a variable for the purposes of comparing two populations after they have been sorted in ascending order. quantile plot Diagrammatic representation of the distribution of two variables. It is one of the control tools, e.g., when comparing grades of a model with sampling data. It is one of the control tools, e.g., for comparing model grades with sampling data R&M repairs and maintenance RAB rotary air blast range Same as Influence Zone; as the spacing between pairs increases, the value of corresponding variogram as a whole also increases. However, the value of the mean square difference between pairs of values does not change from the defined spacing value, and the variogram reaches its plateau. The horizontal spacing at which a variogram reaches its plateau is called the range. Above this spacing there is no correlation between samples. RC reverse circulation Mineral Reserves Mineable geological resources Mineral Resources Geological resources with the potential for economic extraction RF revenue factor RL Reconnaissance Licence RL (Z) Elevation of the collar of a drill hole, a trench or a pit bench above the sea level RMS root mean squared ROM run-of-mine RQD rock quality designation RSD Republic of Serbia Dinars S sulphur SAG semi-autogenous grinding sample Specimen with analytically determined grade values for the components being studied scatterplot Diagrammatic representation of measurement pairs about an orthogonal axis SD standard deviation SEA Strategic Environmental Assessment SEDAR System for Electronic Document Analysis and Retrieval SFR staged flotation reactor SG specific gravity Si silica SI International System of Units

CSA Global Report Nº R302.2019 245 DUNDEE PRECIOUS METALS INC. UPDATED PRELIMINARY ECONOMIC ASSESSMENT FOR THE TIMOK GOLD PROJECT, SERBIA

Abbreviation Description sill Variation value at which a variogram reaches a plateau SMD stirred media detritor SMU selective mining unit SPI SAG power index SQL Structured Query Language standard deviation Statistical value of data dispersion around the mean value string Series of 3D points connected in series by straight lines t tonne(s) TMC Timok Magmatic Complex t/m³ tonnes per cubic metre t/a tonnes per annum TO end of intersection TS total sulphur UC uniform conditioning UFG ultra-fine grinding US$ United States of America dollars UTM Universal Transverse Mercator variation In statistics, the measure of dispersion around the mean value of a data set variogram Graph showing variability of an element by increasing spacing between samples variography The process of constructing a variogram VAT value-added tax WGS World Geodetic System wireframe model 3D surface defined by triangles wmt wet metric tonne waste:feed Ratio of material handled as waste to material handled as feed for gold extraction processing Woodgrove Woodgrove Technologies Inc. X Coordinate of the longitude of a drill hole, a trench collar, or a pit bench Y Coordinate of the latitude of a drill hole, a trench collar, or a pit bench Z Coordinate of the elevation of a drill hole, a trench collar, or a pit bench

CSA Global Report Nº R302.2019 246