2020 CCPS Canada TSC Meeting Slides

5th Annual CCPS Canadian Regional Meeting Hosted by Aon Opening – Fred Henselwood

• Welcome • Our 5th Session!

• Meeting Format Today • Zoom Meeting – Only Presenters will be able to Speak (meeting host will mute attendees) • Use the Zoom Chat feature if you have meeting issues • We will be using Slido.com for managing questions for our Speakers • Slido Event Code is #58659

• Please be patient as we work with this format • We may need to adjust as we go

Slido #58659 Slido Information

• Navigate to Slido.com from your browser • OR: Download the Slido app on your smart device • Enter the event code: 58659 • Please enter questions and poll responses for panelists into Slido (we will not be using the Zoom Q&A or polling features)

3 Morning Session (9:30 – 12:00 Mountain)

• Block 1 – Industry Group Updates 09:30 Opening / Industry Updates Fred Henselwood

09:35 CCPS Update Anil Gokhale / Louisa Nara 10:00 CSA Z767 Update/Opportunity for Input Graeme Norval 10:25 CIAC PSN Update Gilles Laurin 10:35 CSChE PSMD Update Francois Roche 10:45 Break • Block 2 – COVID 19 11:00 COVID -19 Frank Verbeek

11:05 Aon - Energy Engineering and Business Interruption Risk in the Face of COVID-19 Robert Scherer / Emma Whitworth / Robert Colver 11:20 Parkland Refining - Business Update Related to COVID-19 Nick Middleton 11:25 NOVA Chemical - COVID-19 Approach Sorin Dan 11:30 COVID-19 Questions and Answers Session Worked using Slido.com 12:00 Lunch Break Slido #58659 Afternoon Session (1:00 – 3:30 Mountain)

• Block 3 – Insurer Perspective on Process Safety 13:00 Insurer Perspective Frank Verbeek

13:05 Aon - Introduction to PSM and Insurance Frank Verbeek 13:10 Aon - Improving Process Safety Reporting from an Insurance Perspective Nigel Cairns 13:35 Aon - Industry Incidents and Losses Jeff Yuill 14:00 Break • Block 4 – University of Alberta Program Update

14:15 University of Alberta Fred Henselwood

14:15 Regional Promotion of Process Safety through Joint Government/Industry Management Yewei (Janet) Ni 14:30 Using Artificial Intelligence/Machine Learning to Analyze Incident Reports Daniel Kurian 14:45 Application of Artificial Intelligence in Reducing Incident Rates for Oil and Gas Fereshteh Sattari Companies 15:00 Examining Human Factor risks associated with implementing Enhanced Train Control Mona Rad 15:10 U of A Undergraduate Program Overview Christopher Coles 15:25 Closing Comments Fred Henselwood Industry Group Updates Moderated by Fred Henselwood

09:30 Opening / Industry Updates Fred Henselwood

09:35 CCPS Update Anil Gokhale / Louisa Nara 10:00 CSA Z767 Update/Opportunity for Input Graeme Norval 10:25 CIAC PSN Update Gilles Laurin 10:35 CSChE PSMD Update Francois Roche 10:45 Break

Slido #58659 CCPS Update Anil Gokhale and Louisa Nara 9:35 – 10:00 • Project and Business Updates from the Center for Chemical Process Safety

Slido #58659 CSA Z767 Update Graeme Norval 10:00 – 10:25 • Updates on CSA Standard Z767 • Slido Event Code is #58659

Slido #58659 CIAC PSN Update Gilles Laurin 10:25 – 10:35 • Updates from the Chemistry Industry Association of Canada – Process Safety Group

Slido #58659 CSChE PSMD Update Francois Roche 10:35 – 10:45 • Updates from the Canadian Society of Chemical Engineers – Process Safety Management Division

Slido #58659 Morning Break

10:45 – 11:00 • Please rejoin at 11:00 am Mountain

• If you have any questions during the break, please use the Zoom Chat features • You can also e-mail me at [email protected]

Slido #58659 COVID-19 Impacts Moderated by Frank Verbeek

11:00 COVID-19 Impacts Frank Verbeek

Slido #58659 Aon Robert Scherer, Emma Whitworth, Robert Colver 11:00 – 11:15 • Energy Engineering and Business Interruption Risk in the Face of COVID-19

Slido #58659 Parkland Refining Nick Middleton 11:15 – 11:20 • Business Update Related to COVID-19

Slido #58659 NOVA Chemicals Sorin Dan 11:20 – 11:25 • Process Safety Approach to COVID-19

Slido #58659 COVID-19 Question and Answer Session Moderated by Frank Verbeek 11:25 – 12:00 • Robert Scherer, Emma Whitworth, Robert Colver • Nick Middleton, Sorin Dan

• Please enter your questions using Slido.com • Event Code – 58659 • If you wish to speak, please send a note through Slido.com asking to speak

• We will try and get through all of the Questions • We may let this session run into the lunch break • Please be patient as we work with this format

Slido #58659 Lunch Break

12:00 – 1:00 • Please rejoin at 1:00 pm Mountain

• If you have any questions during the break, please use the Zoom Chat features • You can also e-mail me at [email protected]

• If the afternoon presenters could re-join at 12:30, we will test audio connections at that time

Slido #58659 Insurer Perspective on Process Safety Moderated by Frank Verbeek

13:00 Insurer Perspective Frank Verbeek

Slido #58659 Aon - Introduction to PSM and Insurance Frank Verbeek 1:00 – 1:10

Slido #58659 Aon – Improving Process Safety Reporting Nigel Cairns 1:10 – 1:35 • Improving Process Safety Reporting from an Insurance Perspective

Slido #58659 Aon – Industry Incidents and Losses Jeff Yuill 1:35 – 2:00 • Updates on Recent Incidents

Slido #58659 Afternoon Break

2:00 – 2:15 • Please rejoin at 2:15 pm Mountain

• If you have any questions during the break, please use the Zoom Chat features • You can also e-mail me at [email protected]

Slido #58659 University of Alberta Process Safety Program Moderated by Fred Henselwood

14:15 University of Alberta Fred Henselwood

Slido #58659 University of Alberta Yewei (Janet) Ni 2:15 – 2:30 • Regional Promotion of Process Safety through Joint Government/Industry Management • Yewei Ni, Dr. Fereshteh Sattari, Dr. Lianne Lefsrud & Modusser Tufail

Slido #58659 University of Alberta Daniel Kurian 2:30 – 2:45 • Using Artificial Intelligence/Machine Learning to Analyze Incident Reports • Daniel Kurian, Dr. Lianne Lefsrud & Dr. Fereshteh Sattari

Slido #58659 University of Alberta Fereshteh Sattari 2:45 – 3:00 • Application of Artificial Intelligence in Reducing Incident Rates for Oil and Gas Companies • Dr. Fereshteh Sattari, Dr. Lianne Lefsrud, Dr. Renato Macciotta, & Daniel Kurian

Slido #58659 University of Alberta Lianne Lefsrud, Mona Rad 3:00 – 3:10 • Examining Human Factor risks associated with implementing Enhanced Train Control • Dr. Mona A. Rad, Dr. Lianne Lefsrud & Dr. Michael Hendry

Slido #58659 University of Alberta Christopher Coles 3:10 – 3:25 • University of Alberta, Engineering Safety and Risk Management Program Overview

Slido #58659 Closing Fred Henselwood, Frank Verbeek 3:25 – 3:30 • Closing Comments and Thank Yous • Slideo – Event Code: 58659 • 2021 Program Ideas, what you like more of/less of

• Please send any follow-up questions or comments to • [email protected] • [email protected] • [email protected] • [email protected]

Slido #58659 CCPS Updates 10 September, 2020

Dr. Anil Gokhale, P.E. Director, CCPS Projects [email protected]

Louisa Nara, CCPSC Global Technical Director, CCPS [email protected] Antitrust Rules for Member Activities

No CCPS member (“Member”) nor any other participant (“Participant”) attending activities shall:

• Attempt to bring about anticompetitive or illegal understanding or agreement with regard to prices, terms or conditions of sale, distribution, volume of production, sales territories or customer or suppliers.

• Communicate, or engage in other information exchange with regard to prices or pricing methods.

• Discuss, or exchange information about restrictions on production or sales, or sales territories or customers or suppliers unless such action is strictly necessary to respond to an emergency and narrowly limited in duration and scope to the emergency circumstances, and every effort should be made to have counsel present.

• Participate in discussion of costs, or exchange of cost information, unless counsel is present. Antitrust Rules for Member Activities

• No Member or Participant in connection with CCPS activities shall engage in any discussion to prevent any person or business entity from gaining access to any market or customer for goods or services, or to prevent any business entity from obtaining a supply of goods or otherwise purchasing goods or services freely in the market.

• CCPS shall not make any effort to bring about the standardization of any product to prevent the manufacture or sale of any product not conforming to a specified standard.

• CCPS shall not require as a condition for participation that a Participant shall not associate or do business with a non-member.

• All members and participants have an obligation to comply with the antitrust laws in connection with their activity, and will advise an appropriate CCPS official of any known or suspected violation of these Guidelines. CCPS Vision

“To protect people, property and the environment by bringing the best process safety knowledge and practices to industry, academia, the governments and the public around the world through collective wisdom, tools, training and expertise.”

Slido #58659 CCPS Mission Eliminate process incidents in all industries globally by:

• ADVANCING global PS technologies, culture, and management practices; • SERVING as premier worldwide resource of Process Safety; • FOSTERING knowledge and understanding of Process Safety; • PROMOTING Process Safety as key societal value and expectation; and, • ESTABLISHING Process Safety as foundation for responsible operation.

Slido #58659 A Summary Since 2019 52 Thursdays ago Vision 20/20 - Mission 2030

Strategy Execution

Mission 2030 Revitalizing Vision 20/20 Establishing and using the right metrics for success. Focus on achievement and Sharing sustainable Best Excellence. Practices and working virtually Basis Moving Forward

• Where does CCPS want to go in the future? o Expand CCPS strategy o Get more people involved in the conversation (new members / different sectors.) o Expand competency in industries

Slido #58659 Chemical Manufacturing Oil/Gas Companies Food/ Pharmaceutical Companies by Size by Size Companies by Size

20% 27% 35% 42% 51% 13% 67% 31% 14%

Large Medium Small Large Medium Small Large Medium Small . How to help struggling/smaller companies attain process safety improvements/excellence (with fewer staff)

Small and medium companies are being a greater % of our membership.

Slido #58659 Slido #58659 Free Tools and Resources

 CCPS website tour & login creation  Free Resources Categorized via CCPS’ 20 Elements of Risk Based Process Safety: https://www.aiche.org/ccps/resources/rbps https://www.aiche.org/ccps/resources/publications/process-safety-summaries  Process Safety Beacon: https://www.aiche.org/ccps/resources/process-safety-beacon  https://www.aiche.org/ccps/resources/process-safety-beacon/archives  Process Safety Glossary: https://www.aiche.org/ccps/resources/glossary  CSB Videos with Chinese Subtitles: https://www.aiche.org/ccps/resources/overview/ccps-videos/chinese-subtitles  Process Safety Videos: https://www.aiche.org/ccps/resources/overview/ccps-videos/videos-english  CCPS Annual Reports: https://www.aiche.org/ccps/ccps-annual-report  Free Tools – Chemical Reactivity Worksheet, Process Safety Metrics…: https://www.aiche.org/ccps/resources/tools  Downloadable as mobile App  PSIE  Glossary

Your Feedback Please

. Q1: Did you attend the Global Congress Virtual Meeting? . Yes / No . If you attended, please answer the next question . Q2: Please Rate your experience on Overall Experience: 1 to 9 . 1: Disappointed . 3: About the best for the time being . 5: As good as Face to Face meeting . 7: Some aspects better than before . 9: Overall Much Better than Face to Face meeting

Slido #58659 Slido #58659 16th GCPS - Other Feedback

GCPS Technical Program Metrics What we heard . >380 Abstracts Submitted + Great team effort to pull it together + Exceeded expectations . 55 Technical Sessions + Overall very positive feedback . >150 Technical Presentations + All live events were good . >100 Poster Presentation + Recorded sessions offered more flexibility ∆ Q&As format Spring & GCPS Attendees ∆ Networking . >1800 ∆ Exhibitors’ experience ∆ Registration cost 2020 New Members data as of 8.24.20

. United States . Other Regions Amgen Abu Dhabi National Oil Co (ADNOC) Dyno Nobel Adama Ingenero Beijing Strong International System Company. Ltd. Western Midstream Eneos HMC Polymers Jay Chemicals Mangalam Organics Ltd. . Canada Peñoles Procesadora De Gas Parinas Resolver Consultoria Rio Tinto Saudi Aramco Total Refining & Petrochemical Trans-Northern Pipelines, Inc. Saudi International Petrochemical Co Tata Steel Limited 228 Member Companies 8.24.20 September 2019 onwards…. . In 2019, CCPS published 5 Books – 3 in the last quarter . A Monograph on ‘Planning for Natural Hazards’ . Launched 8 new SAChE courses . Held 3 additional Student Boot-camps . Held the Global TSC meeting in Houston Nov 2019 . Focus on Early Career Professionals: (2 initiatives progressing) . The training program curriculum development . CCPSF recognition . Called off the Pipeline project in Canada (Insufficient Volunteers)

Then COVID-19 spread Happened

45 46

Select CCPS Programs UPSLI - Progress & Impact

• 32 new SAChE 118,724 Safety Certificates awardedto courses modules students worldwide • 25 Educating the 36,598 Educators Students completed SAChE modules worldwide workshops 520 • 27 Process Safety New Faculty Members educated Students 429 Workshops Universities participating

Slido #58659 Progress on UPSLI

1. Completed 5 Student Boot Camps – in person . Two more are planned – Virtual format 2. Completed 2 faculty workshops – Virtual format . Sharper focus on training the faculty . Global Attendance – About 45 each . Just one from Canada . One more workshop planned: 7-8 Oct; 9 AM  1 PM EDT 3. Launching 6 additional SAChE modules and upgrading 2

Slido #58659 UPSLI Next – ECP Training Program Why this effort? . Most early career professionals have not received much training in Process Safety . Scattered approach to learning vs Systematic Training . Training Objective: To become Independently Productive . Prevent stumbling around to gather knowledge from disparate sources . Not for becoming an expert

Jennifer Brittain, AdvanSix - Chair Brittney Corley Johnson, ExxonMobil – Vice Chair Nathan Thompson, ecolab – Vice Chair

49 Credentialing CCPSC CCPSF . Program Started in April 2016 . Concept Developed 2020 . ~230 Certified [CCPSC] . Planned Launch 2021 Individuals to-date . Focus: Recognition for PS Professionals . Now: Strong interest from the – ECP, Experts in Narrow areas, … . Why Pursue: Career advancement, global community New Opportunities, …. . Significant growth driven by non-native English speakers . Requirements: A curriculum built on . Three CCPSC Exams offered online SAChE modules . No final exam every year . Preliminary Feedback: Strong Interest

A Premier Recognition as an all A Premier Badge of Recognition as around Expert in Process Safety Process Safety Trained CHEF Webinar Series Chemical Hazards Engineering Fundamentals . A series of 10 Webinars Delivered - 2 Hours each . May 10 through July 22; Once a week . In collaboration with EPSC . 35 Worked examples, 2 detailed Case Studies

. Consistently strong interest from Around the World (350 to 425 reg)

51 Addressing COVID-19 . Working virtual since 12 March 2020 . Managing Risk workshop @ Global TSC meeting 18 June 2020 . Virtual Engagement with various regions . Africa region on 11 June 2020 - In collaboration with AON 1. Monograph -1 “Managing Process Safety despite COVID-19” 2. Monograph - 2 “ Reflection from Global Process Safety Leaders during and following Pandemic(s)” 3. Safe Restart Brochure (Collaborated with Int. Council of Chemical Associations) 4. Bow Tie for Understanding COVID-19 (Collaborated with Energy Institute)

https://www.aiche.org/ccps/publications/process-safety-monographs

Slido #58659 Upcoming Books

Slido #58659 Other Notable Activities • PSID Revamp – New Generation Design • Book of Beacons • Golden Rules of Process Safety for Specific Technologies • Human Factors for Process Plants Operations - A handbook • G/L for PHA Revalidation & Update 2nd Ed. • G/L for Managing Abnormal Situations • G/L for Process Knowledge Management • G/L for Managing CyberSecurity – a Risk Based approach • G/L for Chemical Reactivity Evaluation

Slido #58659 Join the Experts – Volunteers needed

Project Expertise needed Volunteers Peer Reviewers 283 Guidelines for Updating & Revalidating PHAs 2nd Ed. Latest trends & best practices in PHA No Yes 289 Golden Rules of Process Safety for specific Ammonia, Hydrogen Sulfide, Ethylene Oxide, FCCU, Yes Yes Technologies Hydrotreating, Alkylation 294 Managing Cybersecurity – A Risk Based approach Strong understanding of the overlap of Process Safety Yes Yes building on the Process Safety framework and Cyber Security 297 Training for Early Career Professionals Early Career Professionals who can relate to the No Yes challenges of being independently productive in Process Safety roles 299 Introduction to Process Safety for Undergraduates Specific experience in delivering undergraduate Yes Yes 2nd Ed. education in Process Safety 300 Book of Beacon Thorough familiarity with the Beacon, Writing clear and Yes Yes concise lessons learned 301 Interactive online process safety training using Bow Process Safety expertise and fluency in Portuguese Yes Yes Tie technique and/or Spanish 303 Guidelines for Chemical Reactivity Evaluation & R&D professionals, Process development & Reaction Yes Yes Application to Process Design 2nd Ed. safety experts along with engineers & designers with chemical reactivity expertise

Slido #58659 In Summary Very Busy Year, Covid-19 Not withstanding . YOUR THOUGHTS & SUGGESTIONS?

[email protected]

56 CAN/CSA-767-17 Status Update

Graeme Norval, TC Chair • Published in February 2017 • Revision due in 2022 • Have created a Docket system to • Need to adjust the TC matrix to manage proposals for revisions better align with potential members • Outreach to other standard technical committees (pipelines, • Not all are ready to require formal propane) reviews and documentation • Fall and winter goal to begin • On our third Project Manager in revision acceptance process 15 months • First reference in regulations (Ontario Operating Engineers) • Outreach to India

Slido #58659 2020 CCPS Canada TSC Meeting September 10, 2020

Chemistry Industry Association of Canada Process Safety Network Gilles Laurin Director, Responsible Care Process Safety Management Division (PSMD)

CE DOCUMENT EST PUBLIC CE DOCUMENT EST PUBLIC I AIR LIQUIDE, LE LEADER MONDIAL DES GAZ, TECHNOLOGIES ET SERVICES POUR L’INDUSTRIE ET LA SANTE CCPS Canada,date Calgary • September 10, 2020 Air Liquide, Large Industries, Canada François RocheFrançois ing., P.Eng Roche •ing.,∙ Air P.Eng Liquide, Large Industries, Canada Title PSMD Mission

1 to further and promote the interests of all of the members of the Division involved in the pursuit of and understanding of process safety management in accordance with the Society’s purposes, its Articles, By-laws, policies and this Charter.

1 to promote awareness, understanding and use of Process Safety Management (PSM) tools, services and techniques within Canadian facilities including manufacturing and distribution operations, universities, research facilities and laboratories.

1 to influence and encourage a public policy framework that incorporates sound PSM principles

1 to further the advancement of and development of new PSM ideas, theorems, tools, services and techniques.

1 to foster PSM in chemical and related engineering and science education.

1 to monitor the degree of implementation of PSM to identify gaps in knowledge or application and to facilitate and encourage appropriate corrective action.

CE DOCUMENT EST PUBLIC I AIR LIQUIDE, LE LEADER MONDIAL DES GAZ, TECHNOLOGIES ET SERVICES POUR L’INDUSTRIE ET LA SANTE date Air Liquide, Large Industries, Canada Slido #58659 François Roche ing.,∙ P.Eng Title Activities

Annual CCEC (Canadian Chemical Engineering Conference)

● 2020 Ottawa-Virtual October 26-30 ● 2021 Montreal-... October 24-27 ● 2022 Vancouver CSA Z767 Awareness and Promotion

● In the industry and universities ● Other industry outreach ● In other standards Technical Presentations

Photo: Air Liquide Canada

Environmental Risk Assessment Guideline

Risk Assessment Guideline

Also support to member initiatives in

● Training ● Communication

Joint Canadian PSM meetings and projects are essential

Photo: Air Liquide Canada

CE DOCUMENT EST PUBLIC I AIR LIQUIDE, LE LEADER MONDIAL DES GAZ, TECHNOLOGIES ET SERVICES POUR L’INDUSTRIE ET LA SANTE date Air Liquide, Large Industries, Canada François Roche ing.,∙ P.Eng Title Break

66 Proprietary & Confidential | April 2019

Copyright Aon UK Limited. All rights reserved. Aon UK Limited is authorisedSeptember and regulated by the Financial 2020 Conduct Authority. Robert Scherer Energy Team Leader Calgary • Over 15 years of experience in the Energy insurance product lines and risk solutions. • Provides strategic account management, consulting, insurance program design and coordination of client services for intermediate and large energy companies • Drives innovation to understand and addresses evolving client needs

Proprietary & Confidential | April 2019

Which theme is causing the most challenge during the COVID19 pandemic?

1. Business continuity planning 2. Organisation management 3. Projects 4. Operations 5. Supply Chain 6. Maintenance 7. Inspection 8. Process Safety 9. Personal safety 10.Emergency response 11.Municipal aid

Proprietary & Confidential | April 2019

83 Refinery Highlights

. Location: Burnaby, BC . Nameplate capacity: 55,000 bpd . Two crude units, resp. 25,000 and 30,000 bpd . Product slate: gasolines and distillates . Feedstock: Synthetic Canadian crude and light sweet crude . Operations: 300 employees and 100-350 contractors depending on workload and Turnarounds.

Slido #58659 84 Business update related to Covid-19 A quick and prudent response to the ongoing pandemic

Health and Safety Priority is to protect the health and safety of our employees and customers as we continue to provide an essential service in the communities we serve.

Providing an essential service We are proud to remain operational through this period and are committed to safely meeting our customer’s energy and convenience needs.

See End notes for further information 85 Initial COVID19 response during 5 yr FCC Major Turnaround

. Refinery designated as critical infrastructure by the BC Government . Business Continuity Plan exercised (Corp Crisis Team - daily, Local Team – communication/contingency planning) . Medical Team in place supporting Refinery (Iridia Medical – daily meetings at beginning of pandemic with Pandemic expert – industry best practice sharing, etc) . Procurement – Hand Sanitizer, respirators, faceshields . Daily Self-Check Protocol for all workers onsite . Physical Distancing Job Aids, posters, plexiglass, hand sanitizer stations, extra cleaning schedule/staff, extra lunch and washrooms, no group meals provided (box only), reduced meeting size (virtual where possible – TA DRB as example) . Non-essential staff all working from home . Essential refinery staff working modified ‘minimum crew’ schedules (Extra Operators on crew staying home – available for call in). Refinery Staff already split between days and nights – office space was not a huge issue. Staggered start/stop and break times. . Contingency staff workforce for Lab, Maint, Ops and Procurement

Slido #58659 86 COVID19 response ongoing

. Business Continuity Plan exercised (Corp Crisis Team - weekly, Local Team – communication/contingency planning) . Medical Team in place supporting Refinery (Iridia Medical – weekly meetings – industry best practice sharing, etc) . Non-essential staff all working from home still in effect . Essential refinery staff working modified ‘minimum crew’ schedules post Turnaround (7 day per week maintenance, Extra Operators on crew staying home – available for call in) . Refinery Visitor protocol (screening done by Medical team) . Cloth masks made available for use to all Employees in indoor settings . Spring 2021 TA moved to Fall 2021 . Non-essential projects cut

Slido #58659 87 88 PRELIMINARY PROCESS RISK ASSESSMENT ON THE EFFECTS OF THE COVID-19 CRISIS MANAGEMENT ON THE MW-Joffre SITE PROCESS RISKS

-Nova Chemicals approach- April 2020- Summary

• This -essential personnel only- operating mode, has resulted in reduced maintenance activities and most leadership and technical support has been provided remotely.

• With the reduced-personnel operation mode at the Joffre Site, a preliminary risk assessment was conducted to confirm the health of critical layers of protection currently in place and that the risks remained acceptable to the organization

• Bow Tie Analysis (BTA) was used to provide an easy to understand and simple visualization of the relationships between the causes that lead to loss of control, potential consequences, barriers preventing those events from occurring and mitigation measures in place to limit these consequences

Slido #58659 The objective: The goal was to provide the Joffre Site and Corporate leadership with a qualitative process risk assessment that will allow an overall view of the health of our critical layers of protection

• The scope- this assessment has been structured around layers of protection, which included: • a. Process Chemistry • b. Process Equipment • c. Basic Process Control and Monitoring Systems • d. Critical Alarms and Operator Intervention • e. Safety Instrumented Systems • f. Physical Protection Systems • g. Post Release Protection Systems • h. Plant Emergency Response • i. Community Emergency Response

The information collection process was centered on the current operating mode with limited consideration given to the long-term ability to sustain of this operating mode

Slido #58659 Process Equipment/Pipeline Operated outside of Safe Operating Conditions

(Ex: Process Excursions,Overpressure, Fire /Explosion High Level, High/Low Temperature , Plant Equipment Damage Upset, Plant Trip) Financial Impact

Safety Basic Process Inadequate Instrumentated Pressure Control and Inspection,Testing Functions and Relief Post Release Plant Monitoring Systems Inadequate Plant and PM Interlocks Devices Post Release Protection Systems Emergency PSVs/PSE Protection System (Gas Detection, Response Emergency Response Regular PM Technical (Ignition Control) ESDVs, etc.) scheduled in SAP Support/Resources Training Technical Support/Resources

Emergency shutdown valve fails to isolate (Any) Process DCS / Control Failure Operator Training to Regular Maintenance Technical respond to and PMs Support/Resources Emergencies

Inadequate Safety HAZARDS Inspection, Testing Instrumentated Physical Processing pressurized/flammable Inadequate Functions Protection System Ignition Control and PM (Pressure Relief) hydrocarbons or toxic material with reduced support personnel onsite Regular PM Technical Area Classification scheduled in SAP Support/Resources JSWP, WP and Operator Training Available Toxic Material: H2S, Cl2

Fire /Explosions Personnel Injury/fatalities Process Equipment/Pipeline (Any) Mechanical Failure Safety Impact (Ex: Internal or External mechanical failure, leak,seal failure, pipe leak/rupture, internal or external corrosion)

Plant Inadequate Plant Emergency Emergency EVENT Post Release Post Release Protection Systems Response Response Inadequate Repair Process Equipment Protection System (Gas Detection, or Maintenance Proccess (Ignition Control) ESDVs, etc.) Training Technical Control and Loss of Support/Resources Monitoring Regular Inspections, Technical PMs Maint. Support/Resources Process Available Containment Emergency shutdown valve fails to isolate Inadequate Process Control and Operator Training to Regular Maintenance Technical Monitoring respond to and PMs Support/Resources Emergencies, PPE

Technical Online monitoring Support/Resources available Available Inadequate Ignition Control Leyend Area Classification JSWP, WP and Improper Commissioning, Start-up, T/O Operator Training (Ex: Unit/Capital projects, commissioning Threats or failure modes or re-commissioning after maintenance, Layer of Protection (mitigating or preventing Degradation Factor Toxic Releases Management System Personnel Injury/fatalities Safety Impact Commissioning/Start-Up Effective barrier indicator Process Equipment procedure not followed Potentially weakened/degraded barrier indicator Ineffective barrier indicator Plant Technical Training/Aw arness Supervision FCM, PD&I, JSWP Inadequate Plant Support/Resources Support/Resources Emergency Available Available Post Release Response Emergency Response Protection Systems (Gas Detection, Training Technical ESDVs, etc.) Support/Resources

Impairements, Work Permit/Life Critical Procedures - not followed

Emergency shutdown valve fails to isolate

Safety Instrumented Operator Training to Regular Maintenance Technical Ops./Maint. not Systems respond to and PMs Support/Resources following Impairment Emergencies, PPE procedures Physical Protection Systems Supervision Training. Support/Resources Ops./Maint Follow Impairment Available Procedures Toxic Release

Environmental Impact

Plant Inadequate Plant Emergency Response (LP) Emergency Response

Training Technical Post Release Protection Support/Resources Systems (Gas Detection, ESDVs, etc.)

Emergency shutdown valve fails to isolate

Slido #58659 Operator Training to Regular Maintenance Technical respond toToxic Releases, and PMs Support/Resources LOPC Emergencies COVID-19 Group Discussion

93 Lunch Break

94 Aon’s Introduction to PSM and Insurance Frank Verbeek Aon’s Improving Process Safety Reporting from an Insurance Perspective

Nigel Cairns

Proprietary & Confidential | April 2019

Prepared by Aon Guiding Principles

. “Those who cannot remember the past are condemned to repeat it” (George Santayana) . “The first duty of business is to survive, and the guiding principle of business economics is not maximizing profit, it is the avoidance of loss” (Peter Drucker)

Aon | Risk Control Services Slido #58659 Proprietary and Confidential | September 2020 98 Agenda

Introduction Overview of major losses in the industry (2017-2019) Loss 1 Fire in a Cooling Tower Loss 2 An $8 Million Roll of Paper Towels Loss 3 Fires in vessel internals Loss 4 Large Tank Fire

Aon | Risk Control Services Slido #58659 Proprietary and Confidential | September 2020 99 Industry Operational Losses 2017

2017

Type of Loss Location Approximate Loss (USD)

Refinery Explosion UAE 2,000,000,000 Refinery Explosion Ivory Coast 210,000,000 Petrochem Plant Fire USA 67,000,000 Petrochem Plant Fire UAE 35,000,000 Petrochem Germany 280,000,000 Refinery Storage Tanks Fire Japan 84,500,000 Pipeline Contamination Canada 26,600,000 Petrochem Plant Fire Finland 578,000,000 Refinery Fire Italy 14,253,445 Refinery Fire Nigeria 10,000,000 Pipelines Construction Trenching Loss Canada 19,060,750 Petrochem Plant Canada 28,895,000 Petrochem Plant Mechanical Failure USA 45,000,000 Oilsands Pipeline Fire Canada 755,000,000 Fertilizer Plant Explosion Norway 46,000,000 Gas Plant Explosion Canada 15,460,000 Gas Plant Pipelines Explosion USA 30,000,000 2017 total 4,244,769,195

Downstream Energy market premium US$1.75 billion in 2017

Aon | Risk Control Services Slido #58659 Proprietary and Confidential | September 2020 100 Industry Operational Losses 2018

2018

Type of Loss Location Approximate Loss (USD)

Chemical Plant Ice/Snow/Freeze USA 92,500,000 Refinery Fire & Explosion Israel 11,250,000 Petrochemical Ice/Snow Freeze USA 12,000,000 Gas Plant Mechanical Failure USA 20,000,000 Gas Pipeline Subsidence/Landslide Peru 32,000,000 Petrochemical Mechanical Failure Venezuela 97,500,000 Gas Plant Earthquake Papua New Guinea 25,000,000 Chemical Plant Mechanical Failure Australia 21,700,000 Refinery Fire USA 82,000,000 Chemical Plant Fire Austria 155,000,000 Petrochemical Fire India 11,000,000 Fertilizer Plant Explosion Canada 114,500,000 Refinery Fire & Explosion/VCE USA 650,000,000 Petrochemical Plant Supply Nigeria 19,000,000 Pipeline Subsidence/Landslide USA 26,000,000 Petrochemical Plant Fire USA 30,000,000 Refinery Explosion Germany 680,697,000 Onshore Pipelines Fire & Explosion USA 12,000,000 Petrochemical Fire Saudi Arabia 600,000,000 Pipeline Explosion Canada 100,000,000 Refinery Explosion Canada 300,000,000 Chemical Plant Fire & Explosion Russia 106,000,000 2018 total 3,198,147,000 Downstream Energy market premium US$1.75 billion in 2018

Aon | Risk Control Services Proprietary and Confidential | September 2020 101 Industry Operational Losses 2019

2019 Approximate Loss (USD) Type of Loss Location Low Estimate High Estimate Refinery Qatar 80,000,000 80,000,000 Chemicals US 80,000,000 125,000,000 Refinery Vietnam 17,000,000 17,000,000 Refinery Algeria Unknown Unknown Storage US 150,000,000 150,000,000 Refinery US 500,000,000 1,000,000,000 Petrochemoical US 500,000,000 600,000,000 Chemicals Egypt 60,000,000 60,000,000 Refinery Russia 80,000,000 100,000,000 Refinery Cameroon 400,000,000 400,000,000 Petrochemical Kuwait 30,000,000 30,000,000 Storage US 50,000,000 50,000,000 Refinery US Should be below deductible Should be below deductible Refinery US Unknown Unknown Refinery France 240,000,000 240,000,000 Refinery US Unknown Unknown Total 2,187,000,000 2,852,000,000

Aon | Risk Control Services Slido #58659 Proprietary and Confidential | September 2020 102 2016-19 Downstream Premium versus Losses

Underwriting year 2016 2017 2018 2019 Estimated Global Downstream/Midstream Premium ($) 2,000,000,000 1,750,000,000 1,750,000,000 2,150,000,000 Estimated Total Loss Reserves (above $10m) 1,712,900,000 4,244,769,195 3,198,147,000 2,852,000,000 Number of significant losses (above $10m) 12 17 22 16

Aon | Risk Control Services Proprietary and Confidential | September 2020 103 Fire in a Cooling Tower

. Texas, USA July 1997

Aon | Risk Control Services Proprietary and Confidential | September 2020 104 Cooling Tower − Background

. Ammonia and Urea plant . Originally constructed in 1968 by M.W. Kellogg − 1000-tons/day ammonia . Capacity expanded in 1989 − 1450-tons/day ammonia − 240-tons/day urea

Aon | Risk Control Services Slido #58659 Proprietary and Confidential | September 2020 105 Cooling Tower − Background

. Cooling Tower – 5-cell – Induced draft – Cross flow – Constructed with redwood and steel supports – Fill consisted of fiberglass – Circulated 50,000-USgpm

Aon | Risk Control Services Slido #58659 Proprietary and Confidential | September 2020 106 Cooling Tower − Background

Aon | Risk Control Services Proprietary and Confidential | September 2020 107 Cooling Tower − Background

. Heat exchanger 124C. – Horizontal, shell and tube heat exchanger – Built in 1989 – Tubes − contained 977 seamless u-tubes designed to MAWP of 2276 psig, 3/4” in diameter, 0.065” thick – Shell − designed to MAWP of 250 psig, measured 4’-6” in diameter and 15’- 4” long

Aon | Risk Control Services Slido #58659 Proprietary and Confidential | September 2020 108 Cooling Tower − Background

Aon | Risk Control Services Proprietary and Confidential | September 2020 109 Cooling Tower − Events

. July 1997(All times local) . Early morning − plant was running smoothly . 4:30 a.m. − A common trouble alarm rings for the #4 cooling tower fan − Operator proceeds outdoors to investigate and discovers fire − Calls fire department and shuts down plant . Fire extinguished within two hours

Aon | Risk Control Services Slido #58659 Proprietary and Confidential | September 2020 110 Cooling Tower − Outcome

. Fire was limited to the cooling tower only – All pumps and metal piping were unaffected – All fans, fan shrouds, top decking and fiberglass piping were destroyed – The fiberglass fill and wooden struts were also destroyed . Source of fuel determined by running water through heat exchangers

Aon | Risk Control Services Slido #58659 Proprietary and Confidential | September 2020 111 Cooling Tower − Outcome

Aon | Risk Control Services Proprietary and Confidential | September 2020 112 Cooling Tower − Outcome

Aon | Risk Control Services Proprietary and Confidential | September 2020 113 Cooling Tower − Outcome

Aon | Risk Control Services Slido #58659 Proprietary and Confidential | September 2020 114 Cooling Tower − Cause

. The source of the fuel was a leak in heat exchanger 124C. − Fretting corrosion • The exact cause of the ignition could not be determined however possibilities are: − static electricity − Electrical equipment (non-classified in tower) − Spark created from separation of riser from tower

Aon | Risk Control Services Slido #58659 Proprietary and Confidential | September 2020 115 Cooling Tower − Cause

Aon | Risk Control Services Proprietary and Confidential | September 2020 116 Cooling Tower − Lessons Learned

. The heat exchanger should have been analyzed via a hazard analysis model to determine if increased throughputs would affect performance

Aon | Risk Control Services Slido #58659 Proprietary and Confidential | September 2020 117 Cooling Tower − Damages

. All amounts in US dollars (1997) . Amount of loss − $1,550,000 physical damage − $502,000 loss of profits − $2,052,000 Total . Plant was back to 100% output within 55 days

Aon | Risk Control Services Slido #58659 Proprietary and Confidential | September 2020 118 Fires in vessel internals

. Various metal fires

Aon | Risk Control Services Proprietary and Confidential | September 2020 119 Internal vessel fire

. Metal structured packing presents unique hazards to refineries and distillation equipment . When packed columns are opened to the atmosphere, metal structured packing has the potential to ignite if certain conditions are present. . The presence of pyrophoric material, flammables residue and hot work inside the column all increase the likelihood of metal structured packing fires.

Well known mechanism, but these incidents still happen!

Aon | Risk Control Services Slido #58659 Proprietary and Confidential | September 2020 120 Internal vessel fire

. On February 11th, 2001 a distillation column on the styrene monomer production unit collapsed at the Chevron Phillips Chemical Company St James Refinery. . New internal supports were being welded during a column revamp where the stainless-steel packing was to be replaced, it is believed sparks ignited the stainless-steel packing resulting in the fire. . No injuries occurred as a result of the tower collapse, but the unit remained out of commission for approximately 7 months while repairs were completed.

Aon | Risk Control Services Slido #58659 Proprietary and Confidential | September 2020 121 Chevron Phillips Chemical Company St James Refinery

Aon | Risk Control Services Proprietary and Confidential | September 2020 122 Internal vessel fire

. On April 3rd, 2019 a tower located at the Imperial refinery in Sarnia Ontario, Canada. The tower, which was about to under-go a turnaround was taken out of service to conduct inspection and maintenance activities. . It was determined that the cause of the fire was the result of pyrophoric scale deposited on the inside of the column. When the column was opened to atmosphere the pyrophoric material ignited subsequently igniting the metal packing in the column. . No injuries were reported as a result of the tower collapse in addition to no releases to air or water being reported. A new column was built locally and installed in December 2019, an approximate downtime of 9 months. Imperial Oil is part of ExxonMobil and therefor is self-insured and there was no exposure to the insurance industry. (Sarnia Observer)

Aon | Risk Control Services Slido #58659 Proprietary and Confidential | September 2020 123 Imperial Sarnia Tower Collapse

Aon | Risk Control Services Proprietary and Confidential | September 2020 124 Causes of metal packing fires

Presence of combustible materials: . Pyrophoric Material . Hydrocarbon Residues . Coking/Polymers in Column Internals . Highly Reactive Metals in Packing Material

Ignition sources: . Hot work . Spontaneous Ignition . Static Electricity

Aon | Risk Control Services Slido #58659 Proprietary and Confidential | September 2020 125 Prevention techniques

Shutdown: . Cooling of the tower/vessel to ambient temperature before opening . Washing of the internals to remove residual product (steam) . Purge with inert gas . Vessel should be properly blinded

Maintenance: . Temperature monitoring . Fire watch . When hot work needed, physical barriers should be in place . Staff should be prepared to extinguish a potential fire

Aon | Risk Control Services Slido #58659 Proprietary and Confidential | September 2020 126 An $8 million roll of paper towels

. Foreign Object Damage (FOD) to a Gas Turbine 2015

Aon | Risk Control Services Proprietary and Confidential | September 2020 127 Gas Turbine − Background

. 500-MW combined cycle plant located in Texas . Site had two GE 7F Combustion Turbines . During an outage in June 2015, manual cleaning of the Variable Inlet Guide Vanes (VIGV) was performed using paper towels

Aon | Risk Control Services Slido #58659 Proprietary and Confidential | September 2020 128 Gas turbine − Events

. A roll of paper towels was inadvertently left in the top of the bell-mouth (inlet) casing . Upon start-up the roll of paper towels where sucked into and stuck in between VIGV. . The blockage caused an area of low pressure which unloaded the compressor blades for a fraction of a second on each rotation. . After 23 hours, a first stage blade failed due to fatigue and traveled downstream.

Aon | Risk Control Services Slido #58659 Proprietary and Confidential | September 2020 129 Gas turbine − Outcome

Aon | Risk Control Services Proprietary and Confidential | September 2020 130 Gas turbine − Outcome

Aon | Risk Control Services Proprietary and Confidential | September 2020 131 Gas turbine − Outcome

. Damage extended to the inlet casing, requiring rotor overhaul and a new compressor casing. . Damage US$8 million, outage time unknown. . A foreign material exclusion (FME) program would have prevented this incident by controlling material into and out of the unit

Aon | Risk Control Services Slido #58659 Proprietary and Confidential | September 2020 132 Gas turbine − Outcome

Aon | Risk Control Services Proprietary and Confidential | September 2020 133 Gas turbine − Outcome

Aon | Risk Control Services Proprietary and Confidential | September 2020 134 Large Tank Fire

. Thailand, December 1999

Aon | Risk Control Services Proprietary and Confidential | September 2020 135 Large Tank Fire (Thailand, December 1999)

Aon | Risk Control Services Proprietary and Confidential | September 2020 136 Large Tank Fire (Thailand, December 1999)

Aon | Risk Control Services Proprietary and Confidential | September 2020 137 Thailand − Background

. Oil refinery rated at 220,000 bbls/day . Originally built 1962 . Located 130 kilometers from Bangkok . Site occupies 150 hectares – Refinery process occupies SW portion of site – Tank farms occupy NE and NW portions – Admin, shops and stores occupy center of site

Aon | Risk Control Services Slido #58659 Proprietary and Confidential | September 2020 138 Thailand − Background

. Within the diked area- 9 gasoline tanks – tank #3004 − 12 million liters capacity (1970) – tank #3005 − 12 million liters capacity − 2.2 million liters at time of loss – tank #3006 − 12 million liters capacity − 12.2 million liters at time of loss – tank #3022 − empty at time but normally 2 million liters (scaffolding was erected) – tank #3003 − 3 million liter capacity − 1.8 million liters at time of loss

Aon | Risk Control Services Slido #58659 Proprietary and Confidential | September 2020 139 Thailand − Background

Aon | Risk Control Services Proprietary and Confidential | September 2020 140 Thailand − Events

. December 2 (All times are local) – 2053 hours − In line blending into tank #3005 begins – 2144 hours − Pump feeding tank #3005 trips and is reset at 2153 hours . – 2145 hours − Tank #3004 filling is completed by ‘C’ shift team. Gauge level of 1254 cm – 2200 hours − shift change – 2236 hours − high level alarms rings in offsite control room. This was apparently not heard. – 2255 hours − High-high audible alarm also rings. This too was reportedly not heard. – 2322 hours − two operators are dispatched to investigate why tank #3005 was not filling. – 2325 hours − explosion and subsequent fire

Aon | Risk Control Services Proprietary and Confidential | September 2020 141 Thailand − Events

. December 4 – Afternoon-fire is finally extinguished. Fire had rekindled twice during extinguishment due to loss of foam supplies. Foam was flown in from Singapore.

Aon | Risk Control Services Slido #58659 Proprietary and Confidential | September 2020 142 Thailand − Cause

. Over-topping of storage tank #3004 due to operator error (overfill of 203,000 liters) – During the inline blending operation a valve was incorrectly opened which filled tank #3004 instead of tank #3005. . Due to the blast damage it has be suggested that the vapors were ignited in the vicinity of the fire station

Aon | Risk Control Services Slido #58659 Proprietary and Confidential | September 2020 143 Thailand − Lesson Learned

. Location of fire station and administration buildings should be moved. . Alarm system should be revised and possible automatic trips added . Operating procedures should be revised/reviewed

Aon | Risk Control Services Slido #58659 Proprietary and Confidential | September 2020 144 Thailand − Fire Damages

. 7 staff/contractors were killed . Fire damage was largely confined to 5 of the 9 tanks within the dike (#3004, 3005, 3006,3003 & 3022)-all destroyed, along with 41 million liters of gasoline . Two other tanks within the dike were less seriously damaged (#3058 & 3036) . Fire damage to piperack traversing the diked area

Aon | Risk Control Services Slido #58659 Proprietary and Confidential | September 2020 145 Thailand − Explosion Damages

. Blast damage was more widespread: – Walls of several tanks facing the blast were rippled (tanks T113 to T126) – Main office/administration building – Warehouses #1 & #2 – Fire station and six fire trucks parked inside . No damage to process units however site was shutdown in a controlled manner

Aon | Risk Control Services Slido #58659 Proprietary and Confidential | September 2020 146 Thailand − Explosion Damages

Aon | Risk Control Services Proprietary and Confidential | September 2020 147 Thailand − Explosion Damages

Aon | Risk Control Services Proprietary and Confidential | September 2020 148 Questions/Thank you Important: This report contains proprietary and original material which, if released, could be harmful to the competitive position of Aon Reed Stenhouse Inc. Accordingly, this document may not be copied or released to third parties without Aon’s consent. Break

150 Regional Promotion of Process Safety Through Joint Government/Industry Management

Yewei Ni, Dr. Fereshteh Sattari, Dr. Lianne Lefsrud, & Modusser Tufail September 10, 2020 Methodology

Title: A Rising Tide Raises All Boats: Regional Promotion of Process Safety Through Joint Government / Industry Management

Goal: To enhance benefits of SCES Industrial Engagement Program by conducting a comparative study between SCES and two regional bodies:

• Contra Costa Health Services Hazardous Materials Programs (CCHSHMP)

• Technical Standards and Safety Authority (TSSA)

Slido #58659

1 SCES Industrial Engagement Program

Phase 1 (started) : document release incidents, complete Industrial Response Worksheet (IRW) and finish fire inspections

Phase 2 (in progress): PSM education and implementation

Phase 3 (to be determined): focus on the injury prevention

Slido #58659

2 Providing recommendations by focusing on:

1. Regulation and Guidance

1) PSM elements and PSM regulations

2) Hazard Assessments

3) Risk Tolerance and Land-use planning

2. PSM Engagement and Education

3. Annual Performance Indicators

4. Public Participation

Slido #58659

3 1. Regulation and Guidance 1.1 PSM Elements and PSM Regulations

SCES • CSA Z767-17 Process Safety Management • 4 fundamental pillars and 16 elements Contra Costa County • CalARP regulation / EPA’s Risk Management Program • 14 PSM elements TSSA • CSA Z767-17 Process Safety Management • 4 fundamental pillars and 16 elements

Slido #58659

5 PSM Elements

CSA Z767 - 17 CalARP Program / EPA RMP

Accountability Regulations, codes, and standards Process Safety Information Process Safety Leadership Process Safety Culture Employee participation Conduct of operations

Process knowledge and Process Safety Information/ documentation Operating procedures

Project review and design Understanding hazards and procedures risks Process risk assessment and Process Hazard Analysis / reduction Pre-Startup Safety Review

Human factors Process Hazard Analysis

Training and competency Training / Contractors

Management of Change Management of Change Risk management Process and equipment integrity Mechanical integrity

Emergency management planning Emergency response program

Investigation Incident Investigation Audit process Compliance Audits Review and improvement Enhancement of process safety Slido #58659 knowledge Key performance indicators Contractors Table 1. Comparison between SCES and CCCHSHMP on PSM elements 6 1.2 Hazard Assessments

SCES CCCHSHMP TSSA Worst credible scenarios Worst-case scenarios Worst-case scenarios Alternative scenarios Alternative scenarios No parameters provided Provide 7 types of Provide a few parameters parameters No calculation model Simple calculation Simple calculation models provided models are provided for are toxic release, vapor provided for vapor cloud cloud fires, vapor cloud explosion explosion, pool fires and BLEVEs

No mention of domino No mention of domino No mention of domino effects effects effects

Table 2. Comparison between SCES and CCCHSHMP on hazard assessment Slido #58659

7 1.3 Risk Tolerance and Land-use planning

Risk Tolerance • Similar risk tolerance criterion between SCES and TSSA

Figure 1. Risk acceptability criteria for land-use planning. Figure 2. TSSA’s risk sources

TSSA, Annual Safety Performance Report, 2019 https://www.tssa.org/en/about-tssa/resources/Annual-Safety-Performance-Report-2019.pdf Ertugrul, A. (2007). Risk Assessment and Process Safety Management. CSChE PSM Award Presentation 8 1.3 Risk Tolerance and Land-use planning

Land-use planning of SCES • Land buffer – 3.0 km

Figure 1. Risk acceptability criteria for land-use planning. Slido #58659

Ertugrul, A. (2007). Risk Assessment and Process Safety Management. CSChE PSM Award Presentation 9 1.3 Risk Tolerance and Land-use planning

Land-use planning of CCHSHMP • Hazard score

• Credit for reductions or Project to be closed: hazard score of eighty or more

Chapter 84-63 Land Use Permits for Development Projects Involving Hazardous Waste or Hazardous Material https://cchealth.org/hazmat/pdf/iso/land_use_ordinance.pdf 10 1.3 Risk Tolerance and Land-use planning

SCEC • Development proposal: traffic impact analysis and detailed site plan

CCHSHMP • A more defined methodology • For example, instead of using an exact probability of a truck incident, they have a different transportation risk score considering the traveling route and the percentage change of the planned total quantities of hazardous material

Slido #58659

11 2. PSM Engagement and Education PSM Engagement and Education

Goal: • To ensure the education of developed PSM regulation/program  PSM Review and Engagement

To have an effective review/engagement: • Planning o Organize review team, provide the protocol and establish the engagement schedule • Performance o Meetings, interviews, document spot checks, field spot checks and close out meetings • Follow-up o Fact checking, report finalization and resolution of findings

Slido #58659

Birkmire, J. C., Lay, J. R., & McMahon, M. C. (2007). Keys to effective third-party process safety audits. Journal of Hazardous Materials, 142(3), 574–581. https://doi.org/10.1016/j.jhazmat.2006.06.065 13 Auditing and Inspection (CCHSHMP)

Type: audit, compliance audit, inspection, etc. Increase frequency depends on: • Accident history • Accident history of other facilities in the same industry • Quantity of chemicals present; • Location of the facility with respect to public and environmental receptors • Presence of specific chemicals • Hazards identified in the RMP; and • Random selection

Slido #58659

California Accidental Release Prevention (Cal ARP) Program Administering Agency Guidance ,Jan 2005 14 3. Annual Performance Indicators Annual Performance Indicators – CCHSHMP

Figure 3. Major chemical accidents and releases from 1999 to 2019

Figure 4. Major chemical accidents and releases Slido #58659 weighted score for ISO and RISO

Industrial Safety Ordinance Annual Performance Review and Evaluation Report, February 27, 2020 https://cchealth.org/hazmat/pdf/iso/iso-report.pdf 16 Annual Performance Indicators - TSSA

Figure 5. State of public safety in Ontario in 2019

Slido #58659

Technical Standards and Safety Authority, Annual State of Public Safety report, 2019 Edition https://www.tssa.org/en/about-tssa/resources/Annual-Safety-Performance-Report-2019.pdf 17 Annual Performance Indicators

Figure 6. Process safety indicator pyramid Slido #58659

Z767-17 Process safety management (Vol. 33, Issue 4). Canadian Standards Association. https://doi.org/10.1002/prs.11678 18 4. Public Participation Public participation

Missions of advisory councils or commissions:

Broad purpose Specific purposes and activities Get public input in decisions made elsewhere 1. provide information to public 2. Fill information gaps 3. Information contestability 4. Problem solving and social learning Share decision making with public 1. Reflect democratic principles 2. Democracy in practice 3. Pluralist representation Alert distribution of power and structures of 1. Involve marginalized groups decision making 2. Shift the locus of decision making 3. Entrench marginalization Table 3. Purpose of public participation

Slido #58659

O'Faircheallaigh, C. (2010). Public participation and environmental impact assessment: Purposes, implications, and lessons for public policy making. Environmental impact assessment review, 30(1), 19-27. 20 Conclusion Recommendations

1. Develop a comprehensive PSM regulation based on CSA Z767-17 with enforcement activities and noncompliance penalties

2. Require both worst-case scenarios, alternative scenarios and domino effect for consequence analysis and provide simple models and technologies for hazard assessment

3. Regular audits and inspections

• Planning, performance and follow-up

4. Using both lagging and leading performance indicators

5. Set advisory councils or commissions

Slido #58659

22 Thank you Reference

1. Baybutt, P. (2017). Process safety management. In Process Safety Progress (Vol. 33, Issue 4). https://doi.org/10.1002/prs.11678 2. Emergency Services Requirements for Heavy Industrial Developments. (2019). January, 1–14. 3. Brouillard, G. (2017). Adjusting existing company PSM standards to CSA Z767. Presentation. 4. Industrial Safety Ordinance Annual Performance Review and Evaluation Report, February 27, 2020 https://cchealth.org/hazmat/pdf/iso/iso-report.pdf 5. Chapter 84-63 Land Use Permits for Development Projects Involving Hazardous Waste or Hazardous Material https://cchealth.org/hazmat/pdf/iso/land_use_ordinance.pdf 6. Technical Standards and Safety Authority, Annual State of Public Safety report, 2019 Edition https://www.tssa.org/en/about-tssa/resources/Annual-Safety-Performance-Report-2019.pdf 7. Birkmire, J. C., Lay, J. R., & McMahon, M. C. (2007). Keys to effective third-party process safety audits. Journal of Hazardous Materials, 142(3), 574–581. https://doi.org/10.1016/j.jhazmat.2006.06.065 8. Tufail, M. (2019). Strathcona County's Industrial Engagement Program: Leading the Way Using the MIACC Model. Presentation, Hilton New Orleans Riverside. 9. Division Human Factors Impact Ltd. (2010). "Cumulative Risk Assessment Study" For Strathcona County FINAL REPORT. Edmonton. 10. O'Faircheallaigh, C. (2010). Public participation and environmental impact assessment: Purposes, implications, and lessons for public policy making. Environmental impact assessment review, 30(1), 19- 27. 11. California Accidental Release Prevention (Cal ARP) Program Administering Agency Guidance ,Jan 2005 12. Louvar, J. (2010). Guidance for safety performance indicators. Process Safety Progress, 29(4), 387- 388. 24 USING MACHINE LEARNING AND KEYWORD ANALYSIS TO ANALYZE INCIDENTS AND REDUCE RISK IN OIL SANDS OPERATIONS

PRESENTED BY: DANIEL KURIAN ([email protected])

SUPERVISED BY: DR. LIANNE LEFSRUD ([email protected]) DR. FERESHTEH SATTARI ([email protected])

CCPS SEPTEMBER 2020 BACKGROUND

 B. Sc. in Chemical Engineering (Co-op Program)  Cost Engineer for NOVA Chemicals and Enbridge  Quality Control Materials Technician for J.R. Paine  Engineering student for CAN-K  M.Sc. in Engineering Management (Risk Management)  Thesis: Using Machine Learning and Keyword Analysis to Analyze Incidents and Reduce Risk in Oil Sands Operations

SLIDO #58659

2 BACKGROUND

 Current Position  Online courses: Python, SQL  Research projects: Microbially Induced Corrosion (MIC), Knowledge Transfer  Statistics Lab Manager/Instructor for MBA students at the U of A

SLIDO #58659

3 INTRODUCTION  Why is the digitalization of incident reports important?  Incidents still occur and are costly:  In 2016, costs and claim payments cost Canadians ~$8.7 billion1  Production loss, absenteeism, medical costs, and compensation equate to 4% of the annual global gross domestic products2  Gaps in current methods of incident reporting:  Typos and misinformation  Inadequate follow-ups for incidents  Time consuming3  Bias (different interpretations)4

1. 2016 STATISTICS FROM CANADIAN CENTRE FOR OCCUPATIONAL HEALTH AND SAFETY, HTTPS://WWW.CCOHS.CA/EVENTS/MOURNING/. 2. TAKALA, J., HAMALAINEN, P., SAARELA, K.L., YUN, L.Y., MANICKAM, K., JIN, T.W., HENG, P., TJONG, C., KHENG, L.G., LIM, S., AND LIN, G.S. (2014). GLOBAL ESTIMATES OF THE BURDEN OF INJURY AND ILLNESS AT WORK IN 2012. JOURNAL OF OCCUPATIONAL AND ENVIRONMENTAL HYGIENE, 11: 326E337. 3. NIJS JAN DUIJM, “RECOMMENDATIONS ON THE USE AND DESIGN OF RISK MATRICES,” SAFETY SCIENCE, 76, PP. 21–31, 2015. 4. PHILIP THOMAS, J. ERIC BICKEL, AND REIDAR B. BRATVOLD, “THE RISK OF USING RISK MATRICES,” SPE ANNUAL TECHNICAL CONFERENCE AND EXHIBITION, 2013. 4 METHODOLOGY (OVERVIEW)

 Analyzed existing records using machine learning 4. Output Results

 Extracted useful features and generated predictions for Risk Matrix (likelihood, operations consequence, score)

2. Apply Machine Learning 1. Input Data Trend Analysis (past 12 and Keyword Analysis months) 3. User Input Data storage Customized Library Select Statements Prevention / Mitigation Strategy

 Flow chart is color-coordinated with succeeding methodology slides Leading Factors

SLIDO #58659 5 METHODOLOGY

1.1 Input data •Asset management system •Safety procedures and guidelines •Incident database

2.1. Manually classify data •Assign identifying labels to incident reports (survey subject matter experts to determine labels)

2.2. Prepare data for machine learning classification •Convert text (incident reports) to numerical vectors •Separate data into training and test data

2.3. Use classifiers from scikit-learn library to classify data •Adaboost •Decision tree •K-nearest neighbor •Logistic regression •Multi-layer perceptron •Multinomial Naive Bayes •Random forest •SVM (including Linear SVC)

2.4. Calculate metrics for each classifier •Confusion matrix •Precision, recall, F1-score, support •Accuracy

6 METHODOLOGY

1.1 Input data •Asset management system •Safety procedures and guidelines •Incident database 2.1. Manually classify data Manual classification labels: •Assign identifying labels to incident reports (survey subject matter experts to determine labels)  Communication 2.2. Prepare data for machine learning classification •Convert text (incident reports) to numerical vectors  Health/Safety •Separate data into training and test data

2.3. Use classifiers from scikit-learn library to classify data  Leak/Spill •Adaboost •Decision tree  Operation •K-nearest neighbor •Logistic regression •Multi-layer perceptron  Miscellaneous •Multinomial Naive Bayes •Random forest  Vehicle •SVM (including Linear SVC)

2.4. Calculate metrics for each classifier  (Uncategorized) •Confusion matrix •Precision, recall, F1-score, support •Accuracy

7 METHODOLOGY

1.1 Input data •Asset management system •Safety procedures and guidelines •Incident database  TfidfVectorizer5 2.1. Manually classify data •Assign identifying labels to incident reports (survey subject matter experts to o Tokenizing, text pre-processing, removing determine labels) stop words 2.2. Prepare data for machine learning classification •Convert text (incident reports) to numerical vectors o Builds a feature dictionary and transforms •Separate data into training and test data incident reports into feature vectors 2.3. Use classifiers from scikit-learn library to classify data •Adaboost •Decision tree o Calculate frequency of word occurrence •K-nearest neighbor •Logistic regression o Fits an estimator to the data and then •Multi-layer perceptron •Multinomial Naive Bayes transforms the count-matrix to a tf-idf •Random forest •SVM (including Linear SVC) representation 2.4. Calculate metrics for each classifier  •Confusion matrix Train a classifier •Precision, recall, F1-score, support •Accuracy

5. GARRETA, R., HAUCK, T., & HACKELING, G. (2017). SCIKIT-LEARN: MACHINE LEARNING SIMPLIFIED. BIRMINGHAM, UK: PACKT PUBLISHING. 8 METHODOLOGY

1.1 Input data •Asset management system •Safety procedures and guidelines •Incident database 2.1. Manually classify data Classification Method5 Accuracy •Assign identifying labels to incident reports (survey subject matter experts to determine labels) Support Vector Classifier (SVC) 56.98% 2.2. Prepare data for machine learning classification Linear SVC 88.48% •Convert text (incident reports) to numerical vectors •Separate data into training and test data MLP Classifier (Neural Network) 85.50% 2.3. Use classifiers from scikit-learn library to classify data NearestClassification Neighbors Metrics 73.56% •Adaboost •Decision tree Decision Tree 75.95% •K-nearest neighbor •Logistic regression •Multi-layer perceptron Random Forest 75.80% •Multinomial Naive Bayes •Random forest Adaboost 63.21% •SVM (including Linear SVC) Multinomial Naïve Bayes 66.76% 2.4. Calculate metrics for each classifier •Confusion matrix Logistic Regression 84.37% •Precision, recall, F1-score, support •Accuracy

5. GARRETA, R., HAUCK, T., & HACKELING, G. (2017). SCIKIT-LEARN: MACHINE LEARNING SIMPLIFIED. BIRMINGHAM, UK: PACKT PUBLISHING. 9 METHODOLOGY

2.5. Apply natural language processing

•Add identifying labels from machine learning classification •Lemmatize incident database •Identify and include the most commonly used words

2.6. Generate customized library

•Create statements that can be used to accurately describe risks •Match identifying labels (from machine learning classification) and most commonly used words (from keyword analysis) to statements used to analyze risk

3.1 User Input

•User inputs an incident report and selects statements that match the incident being reported

4.1. Analyze data / outputs

•Risk matrix •Trend analysis

4.2. Provide recommendations

•Prevention and mitigation strategies •Leading indicators

10 METHODOLOGY

2.5. Apply natural language processing

•Add identifying labels from machine learning classification •Lemmatize incident database •Identify and include the most commonly used words

2.6. Generate customized library

3.1 User Input statements to outputs

•User inputs an incident report and selects statements that match the incident being reported  Many variables can be modified

4.1. Analyze data / outputs

•Risk matrix •Trend analysis

4.2. Provide recommendations

•Prevention and mitigation strategies •Leading indicators

11 METHODOLOGY

2.5. Apply natural language processing

•Add identifying labels from machine learning classification •Lemmatize incident database •Identify and include the most commonly used words

2.6. Generate customized library

3.1 User Input

•User inputs an incident report and selects statements that match the incident being reported

4.1. Analyze data / outputs

•Risk matrix •Trend analysis

4.2. Provide recommendations

•Prevention and mitigation strategies •Leading indicators

12 METHODOLOGY

2.5. Apply natural language processing

•Add identifying labels from machine learning classification •Lemmatize incident database •Identify and include the most commonly used words

2.6. Generate customized library

3.1 User Input

•User inputs an incident report and selects statements that match the incident being reported

4.1. Analyze data / outputs

•Risk matrix •Trend analysis

4.2. Provide recommendations

•Prevention and mitigation strategies •Leading indicators

13 METHODOLOGY

1.1 Input data 2.5. Apply natural language processing •Asset management system •Safety procedures and guidelines •Add identifying labels from machine learning classification •Incident database •Lemmatize incident database •Identify and include the most commonly used words 2.1. Manually classify data

•Assign identifying labels to incident reports (survey subject matter experts to 2.6. Generate customized library determine labels)

2.2. Prepare data for machine learning classification •Create statements that can be used to accurately describe risks •Match identifying labels (from machine learning classification) and most •Convert text (incident reports) to numerical vectors commonly used words (from keyword analysis) to statements used to analyze risk •Separate data into training and test data

2.3. Use classifiers from scikit-learn library to classify data 3.1 User Input

•Adaboost •User inputs an incident report and selects statements that match the incident being •Decision tree reported •K-nearest neighbor •Logistic regression •Multi-layer perceptron 4.1. Analyze data / outputs •Multinomial Naive Bayes •Random forest •Risk matrix •SVM (including Linear SVC) •Trend analysis

2.4. Calculate metrics for each classifier 4.2. Provide recommendations •Confusion matrix •Precision, recall, F1-score, support •Prevention and mitigation strategies •Accuracy •Leading indicators

14 RESULTS: CASE STUDY

 Highlight methodology  Demonstrate versatility  Display practicality

SLIDO #58659

15 RESULTS – CASE STUDY

 Incident: “Icy road conditions. Employee truck and 3rd party vehicle made driver side contact. Employee complained about minor whiplash.”  Selected Statements: Injury, Vehicle (Light Vehicle), Vehicle Collision (With Injury)

SLIDO #58659

16 RESULTS – CASE STUDY

 Selected Statements: Injury, Vehicle (Light Vehicle), Vehicle Collision (With Injury)

Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec

Risk matrix for Case Study Trend report for Case Study depicting number of incidents per month 17 RESULTS – CASE STUDY

 Selected Statements: Injury, Vehicle (Light Vehicle), Vehicle Collision (With Injury)  Prevention and Mitigation Strategies  Drive at a speed suitable to road conditions  Ensure that vehicle is properly equipped for winter weather (e.g. winter tires, first aid kid, etc.)  Pay attention to other vehicles on the road (e.g. make sure other drivers are not distracted, maintain safe following distance, check blind spots)  Make sure that the seat is properly adjusted to provide ample neck and lumbar support  Provide training for workers to drive in winter road conditions

SLIDO #58659

18 RESULTS – CASE STUDY

 Selected Statements: Injury, Vehicle (Light Vehicle), Vehicle Collision (With Injury)  Leading indicators  Winter weather  Poor traction

SLIDO #58659

19 CONCLUSION

 Used combination of supervised machine learning and keyword analysis to generate a customized library  Statements in customized library are tailored to inputted incident database, safety procedures, operating guidelines, and asset management systems

SLIDO #58659

20 APPLICATION OF DEVELOPED MODEL

 Identification of potential hazards and risks  Prevention/mitigation strategies and actions to take based on identification  Economic benefit

SLIDO #58659

21 Thank you! Application of Bayesian Network and Artificial Intelligence to Reduce Accident/Incident Rates in Oil & Gas Companies

Dr. Fereshteh Sattari, Dr. Lianne Lefsrud, Dr. Renato Macciotta, & Daniel Kurian September 10, 2020 Research Problems

• While companies keep incident data in 1000s of reports, rarely do they analyze these to learn and prevent future incidents.

• Further, related datasets (maintenance data, performance data, employee survey data) are rarely integrated to understand these as leading indicators. Objectives

• Identify which Process Safety Management(PSM) elements most influence the occurrence of an incident/accident rate, by using AI/ML (Bayesian Network Analysis, Cross-correlations, etc.).

• Provide practical recommendations to reduce/eliminate these latent causes.

• Support and empower critical thinking, enhanced communication, and learning to improve organizational safety.

• This enables follow-on integration with complementary datasets (i.e., maintenance, performance, survey data) and further explanatory analysis.

Slido #58659 Results

Automatically “clean” databases for Using Artificial intelligence and Machine Learning analysis to categorize incident types to Improve Safety Performance Library of latent causes, contributing Multiple users means conditions, and controls specific to unique standards and operation/site errors in data entry Consistent method for reporting incidents and evaluating risks by Databases Keyword assessing frequency and describing actual and potential consequences Data Analysis Incident report Cross Clean Validation Real-time feedback on the relative prevalence of incident type, given Standard operating Machine Artificial seasonality, and operational activities Statistics procedures Learning(ML) Intelligence(AI) and suggest controls as drawn from a company’s standard operating procedures Pattern Probability Performance data Recognition Determination Analyze the relative influence of and relationship between various leading Perception survey indicators that can be used to discover patterns, identify best leading Dr. Lefsrud Research Team indicators, and predict incidents Step 1: Categorizing Incidents into PSM Elements by Using Machine Learning and Key Word Analysis

PSM Element Keyword Compliance with Standards compliance, comply, regulatory Process Safety Information psi Hazard Identification & Risk hazard, risk Analysis Operating Procedures procedure Safe Work Practices safety, safe work Asset Integrity & Reliability equipment, asset Contractor Management contractor, third party, 3rd party Training training, inexperience, lack of experience Management Review & management Continuous Improvement

Downloadable AIChE CCPS reference material from: https://www.aiche.org/sites/default/files/docs/summaries/overview-of-risk-based-06-25-14.pdf Step 2: Determining the Relationships between PSM Elements and the Total Number of Incidents by Using AI and Bayesian Network

Group Variables: PSM Elements & Total Number of Incidents 1 Compliance with Standards 2 Process Safety Information 3 Hazard Identification & Risk Analysis 4 Operating Procedures 5 Safe Work Practices 6 Asset Integrity & Reliability 7 Contractor Management 8 Training 9 Management Review & Continuous Improvement 10 Total Number of Incidents Step 3: Calculating the Strength Value of all the Arcs

From Group To Group Arc Strength 6 (Asset Integrity & Reliability) 10 (Total Number of Incidents) -315.06 9 (Management Review & Continuous Improvement) 10(Total Number of Incidents) -102.35 3 (Hazard Identification & Risk Analysis) 10 (Total Number of Incidents) -78.86 7 (Contractor Management) 10 (Total Number of Incidents) -52.13 4 (Operating Procedures) 10 (Total Number of Incidents) -50.86 5 (Safe Work Practices) 10 (Total Number of Incidents) -30.81 8 (Training) 10 (Total Number of Incidents) -9.96 2 (Process Safety Information) 10 (Total Number of Incidents) -4.73 1 (Compliance with Standards) 10 (Total Number of Incidents) -3.49 3 (Hazard Identification & Risk Analysis) 4(Operating Procedures) -7.84 3 (Hazard Identification & Risk Analysis) 5(Safe Work Practices) -4.63 4 (Operating Procedures) 8((Training) -1.67 5 (Safe Work Practices) 7(Contractor Management) -1.07 Note: Total number of incidents (group 10) has maximum dependency with Asset Integrity & Reliability (group 6); by addressing this, we can reduce the total number of incidents by half. Step 4: Preforming Linear Cross Correlation Technique for Result Validation

Note: This technique confirms that group 6 and group 10 have the most correlated variables (R2=0.73), consistent with findings from BNA. Step 5: Providing Some Actionable Recommendations to Improve the Safety Practices Asset Integrity & Reliability

• Management of assets through the AIM. It is used to create the framework for an organization to evaluate its processes and assets in connection with its objectives.

• Implementation of cloud-based software to better link the different parties involved within a project: technicians, manufacturers, regulators, end users, etc.

Management Review and Continuous Improvement

• Harmonizing safety practices and regular operating procedures, standardizing safety culture, increasing the competency of the workforce, increasing the involvement of the workforce in decision-making, and motivating their employees to accept ownership of safety.

• Supporting risk-taking behavior and having tolerance for failure encourage individuals to learn and develop creativity.

• Promoting creativity within organizations that have high expectations towards conformity.

Note: To improve organizational safety, we need to support and empower critical thinking, enhanced communication, and learning Culture. Slido #58659 Conclusions

In analyzing a company’s tailings incident reports we found that: • Bayesian Network Analysis is a useful method to analyze incidents. • Asset Integrity & Reliability is the most influential element (50%), followed by Management Review & Continuous Improvement (17%). Thus, Asset Integrity And Reliability leading indicators would most predictive. We can extend this work by: • Examining the relationship between incidents, additional process safety management elements, and other databases (i.e., safety culture). • Illuminating how we can best enhance safety culture to empower critical thinking, enhanced communication, and learning. • Leveraging our method to other industries, like railroads.

Slido #58659 Future Step: Data Fusion • Data fusion methods have a long history of development starting with applications in robotics. Current application areas include mine detection, maintenance engineering, and weather prediction.

Proposed Methodology: Interview/ Incident • Collect information/data from multiple sources survey report of data. Data cleaning/ Data • Employ a combination technique for fusion and NLP cleaning concatenation of the data vectors from different sources into a single data vector for each data Scaled/ Scaled/ point. Normalized data Normalized data

Fused data

Slido #58659 Thank You for Listening! Examining Human Factor risks associated with implementing Enhanced Train Control

Drs. Mona A. Rad, Lianne Lefsrud & Michael Hendry

September 10, 2020 • High profile railway accidents in the US motivated the adoption of Positive Train Control (PTC) • After $14 billion industry investment, the system remains confounded by complexity, inter-operability issues, and ongoing costs (Popke, 2019)

Chatsworth train collision occurred at 4:22:23 p.m. PDT on Friday, September 12, 2008, when a Union Pacific freight train and a Metrolink commuter train collided head-on. The Metrolink engineer was texting while on duty. 25 people died and 135 were injured. • Transport Canada is considering the adoption of Enhanced Train Control (a variation of PTC in US).

• Our research group is examining: 1. Human Error Modes (a.k.a. potential Human Factor Events (HFE) or Unsafe Actions (UA)) for railway tasks 2. Performance Shaping Factors (PSF) that contribute to the causes and consequences of errors (e.g., fatigue, poor human-machine interface) 3. Prioritization of ETC automation features by the potential to reduce human-induced risk and support High Reliability Operations (sharing and processing of info, collaborative communication, identifying costs and benefits, maintaining complexity, and empowering Subject Matter Experts)

Slido #58659 Human Reliability Analysis

Human factors & ETC Human- Accident System Analysis Interaction Methods

Slido #58659 Human Reliability Analysis

Human factors & ETC Human- Accident System Analysis Interaction Methods

Slido #58659 • HRA aims at systematically identifying and analysing the causes, consequences and contributions of human failures in socio-technical systems. • HRA involves the use of qualitative and quantitative methods to evaluate human contribution to risk.

• The origins of HRA lie in the early probabilistic risk assessments (PRA) performed as part of the US nuclear energy development programme in the 1960s.

• Probabilistic Risk Assessment (PRA): Probabilistic Safety Assessment (PSA), Quantitative Risk Assessment (QRA), Formal Risk Assessment (FRA) (as it may be referred to, depending on the industrial sector).

Slido #58659 The context factors: • Performance Shaping Factor (PSF); • Performance Influencing Factor (PIF); • Performance Affecting Factor (PAF), First generation • It focused on human error probabilities and human operational errors. • Error Producing Condition (EPC); and, (1970–1990) • THERP, HEART, SLIM • Common Performance Condition (CPC).

 Rail specific PSF taxonomies: Second • It focused on human performance-shaping generation factors (PSFs) and cognitive processes. (1990–2005) • ATHEANA, CREAM

Third • It focuses on human performance shaping generation factors, relations, and dependencies. (2005-present) • NARA, RARA, BN

Slido #58659 The complete R-PSFs taxonomy (Kyriakidis et al., 2018)

Slido #58659 Kyriakidis et al. (2018)

The network of interdependences between the categories of R-PSFs. In blue and orange are shown the dynamic and static R-PSFs categories respectively. Human Reliability Analysis

Human factors & ETC Human- Accident System Analysis Interaction Methods

Slido #58659 •Domino model (DM) •Fault Tree Analysis (FTA) sequential •Event Tree Analysis (ETA) •Failure Modes and Effects Analysis (FMEA) (simple linear) •Root Cause Analyses (RCA) •Five Whys method

epidemiological • Swiss Cheese model (complex linear) • Human Factors Analysis & Classification System (HFACS)

systemic • Accimap (complex non- • systems Theoretic Analysis Model and Processes model (STAMP) linear) • Functional Resonance Analysis Method (FRAM)

Slido #58659 Accident analysis models and methods Overall, various challenges arise in the analysis from the following factors: • Scale and complexity • Number and types of failure modes components can have • Dependencies among system components • Number of states the system can be in • Temporal behaviour of the systems • Eﬀect of organizational and human errors on system failure • Uncertainties in system behaviour and failure data • Supplementary methods:

Statistical analysis, BN, PN, MADM (ANP, AHP, DEMATEL), Fuzzy set theory, Machine learning

Slido #58659 Human Reliability Analysis

Human factors & ETC Human- Accident System Analysis Interaction Methods

Slido #58659 Human System Interaction (HSI) • Approaches to define the assignment of functions to people and automation in terms of a more integrated team approach are (Kaber and Endsley, 2004) : • Level of Automation (LOA) or ‘level of control’: static function assignments. • Adaptive Automation (AA) or Dynamic Function Allocation (DFA): dynamic control allocations (automated or manual, varying over time).

Table 3. Categorization of variables that influence human- automation interaction with respect to error management (McBride et al., 2014). • Researchers from all around the world including the UK, the US, Germany, Australia, and Canada have investigated the effects of train control automation on train drivers.

• The publications and their findings are summarized in our paper, currently under review.

Slido #58659 • Interesting preliminary findings from Phase 1 - HFA: – Distrust and suspicion of the new system (Monk et al., 2017) – With higher trust in the system, there is lower situational awareness and expectation bias (Giesemann, 2013) – Optimal workload – neither too much nor too little (Brandenburger et al., 2017, 2018, 2019) – Re-distribution of attention from out-of-cab to in-cab, and thus higher reaction time and risk (Hely et al, 2015; Naghiyev et al., 2017) – Change in train driving (Sebok et al., 2017); fatigue (Van der Weide et al., 2017); distraction (Zimmermann, 2015); complacency (Roth et al., 2013); Loss of team work and informal communication (Roth et al., 2013) Thank you! Questions?

Mona A. Rad, [email protected] Lianne Lefsrud, [email protected] Michael Hendry, [email protected] University of Alberta - Undergraduate Education Engineering Safety & Risk Management

Chris Coles – MEng, PEng, CSP Associate Director, Lynch School of Engineering Safety & Risk Management Presentation Overview

 U of A Faculty of Engineering

 David & Joan Lynch School of Engineering Safety & Risk Management

 ESRM Undergraduate Courses

Slido #58659 228 Faculty of Engineering

 Classes offered since 1908 when University first opened

 5 Departments • Chemical and Materials Engineering • Civil and Environmental (School of Mining & Petroleum) • Electrical and Computing Engineering • Mechanical Engineering • Biomedical Engineering (joint with Medicine)

 >200 Faculty members  4400 undergrads, 1500 graduate students  >$50M annual research grants  >1000 journal papers/year

Slido #58659 229 Undergraduate Engineering

 9 undergraduate programs • Co-op – 5 years • Traditional – 4 years • Chemical Engineering (options: process control, biomedical) • Civil Engineering (options: environmental, biomedical) • Computer Engineering (option: nanoscale systems, software) • Electrical Engineering (options: biomedical, nanoengineering) • Engineering Physics (option: nanoengineering) • Materials Engineering (options: biomedical, nanomaterials) • Mechanical Engineering (option: biomedical) • Mining Engineering • Petroleum Engineering  Accredited by the Canadian Engineering Accreditation Board (CEAB)

Slido #58659 230 David & Joan Lynch School of ESRM  Former Engineering Dean David Lynch Legacy • Stepped down in 2015 after more than two decades • He oversaw a number of milestones . Added five new engineering buildings . Student enrollment increased from 2900 to 6000 . Faculty increased from 93 to 205 . Entrenched ESRM in the formal engineering curriculum  The School was established in 2015 (expanding the reach of the successful ESRM program)  Unique in Canada – integration of risk management as a core competency into all the engineering curriculums at the U of A  School Vision: Taking Risk Management to the Next Level  School Mission: To provide organizations with leaders in engineering safety and risk management through exceptional teaching, research and advocacy

Slido #58659 231 Engineering Safety & Risk Management School

Professor Winkel, P.Eng., M.Sc.Eng.; Director, Chair

Professor Cocchio, Dr. Lefsrud, Dr. Macciotta, Dr. White, Professor Coles, P. Eng., MBA; PhD, P.Eng.; PhD, P.Eng.; PhD, P.Eng.; P.Eng., M.Eng.; Industrial Professor Assistant Professor Industrial Professor Industrial Professor Associate Director, Industrial Professor

 Two 400 level courses in Engineering Safety & Risk Management (ESRM) • ENGG404 (ESRM - Leadership in Risk Management) – taken by all students (enrollment of 800+ students/year) • ENGG406 (ESRM – Methodologies and Tools) – elective (most students are Chemical Engineers, ~ 20 students/year)

Slido #58659 232 Our School Engagement

Oil & Gas Industry

233 ENGG404 - Leadership in Risk Management Overview

• What is ESRM? Why Are They Important?, The Engineer’s Survival Guide, Hazard Identification, The Fundamentals Incident Pyramid, The Discovery Approach, Occupational Safety and Process (Technical) Safety, Risk as a Function of Likelihood and Consequence, Metrics, Health and Hygiene

RM Program • RM System Elements, Administrative Controls and Work Practices, Engineering Controls, The Elements Business Case for a RM Program

• How Do We Manage Risk?, The RM Work Process, Substandard Practices and Conditions, The RM Tools Hierarchy of Controls, Risk Matrices, FLRAs, JSAs, SQRAs, Checklists, Bow Tie Analysis, Job Observations, Planned Inspections, Audits, Prevention Through Design, What If Analysis

Incident • Why is II important?, PEAP, Cause and Effect Model, Root Cause Analysis, Incident and Near Miss Investigation Reporting, Follow Up, Student Team Project & Report

• Sunrise Propane, Hub Oil, USAF Czar B52, Imperial Metals Mount Polley, MMA Lac Megantic, UCIL Case Studies Bhopal, Nypro Works Flixborough, BP Macondo DWH, Boeing 737 Max 8, ALCOA

Leadership in • Why is Leadership Important?, Safety Culture, Personal Iceberg, ABC, and Training Models, Plan- RM Do-Check-Act, Application of the ESG, Risk Communication

Legal & Ethical • Occupational Health and Safety in Canada and Alberta, WCB in Alberta, Due Diligence for Aspects Engineers, Professionalism and Ethics

Industry & Gov’t • Guest Lectures from Magna IV Engineering, PCL Construction, AB Justice, WCB, Imperial Oil, BBA Engagement Engineering, Director of the School of ESRM

234 ENGG406 – ESRM Methods Overview

Hazards • Risk Based Process Safety; Hazard ID; Risk Management Concepts; Fires, Explosions & Releases

Inherent Safety • Inherently Safer Design; Bhopal Loss Incident

Design • PFD, P&ID, Hazardous Area Classification; Regulations; Occupational Health & Toxicology

• Pre-Startup Safety Review; Management of Change; Human Factors; Flixborough Incident; Phillips Operations 66 Polyethylene Plant incident (Pasadena, TX); Strathcona Refinery Tour (Imperial Oil)

• Independent Layers of Protection; Basic Process Control Systems; Safety-Instrumented Systems; Alarms Safety System Impairment; BP Texas City Incident

Safeguards • Overpressure protection; specialized safeguards; asset integrity/barrier analysis

• Chemical Reactivity Worksheet; What-if; HAZOP; FMEA; Event Tree; Fault Tree; Dispersion & Risk Assessment Consequence Analysis; LOPA; Dow FEI & CEI tools; BowTie; Safety Integrity Level; Piper Alpha & BP Macondo incidents; Calvert City incident

Misc Topics • Emergency Management; Key Performance Indicators; Audits; PSM frameworks

Guest Lecture • U of A Faculty of Medicine & Dentistry; Alberta Boilers Safety Association (ABSA); Syncrude; Organizations Imperial Oil; Chevron; Dow; NOVA Chemicals; Shell; C-FER

235 Questions

236 Close of Meeting

237