® Originally appeared in World Oil JUNE 2020 issue, pgs 29-32. Posted with permission.

PRODUCTION OPTIMIZATION Developing a smart and intervention strategy, using automation and advanced data analytics

Selecting wells for workover make more informed decisions faster, re- frame, resulting in missed opportunities. or intervention is time- sulting in economic production gains. Without being able to easily access data consuming, resulting in fewer or automate low-value tasks for screening well reviews per year and INDUSTRY CHALLENGE wells, production engineering teams are longer intervals between the With oil and gas companies operating heavily constrained in implementing well under razor-thin margins, production en- interventions or that deliver pre-planning and execution gineering teams are under increased pres- the maximum economic benefit to their stages. This article presents a sure to squeeze production from their organizations. solution that uses automation existing assets, as efficiently and econom- and advanced data analytics ically as possible. CURRENT STRATEGIES to turn months-long manual A strategic well intervention and During the pre-planning or candidate well selection into a minutes- workover program can help operators screening stage of a traditional well inter- achieve this objective. However, the pre- vention or workover program, produc- long process that supports planning well review process for such tion engineering teams use a multitude of economic production gains. jobs, which involves screening wells, se- applications to collate data, and manual lecting the best intervention or workover engineering techniques to conduct their method, and estimating the post-job well screening and analysis. These analy- ŝ CIO CIO MARIO, RINI SAPUTRA, SERA value of the well against the cost, is often ses are repetitive in nature and limit the SAIFUDDIN and TRIYONO TRIYONO, PT Saka characterized by inefficiency. extent of study when managing high-well- Energi Indonesia; SABRINA METRA, PRAPRUT Time-consuming manual tasks, such count assets. Teams traditionally adopt a SONGCHITRUKSA, RAJEEV RANJAN SINHA as gathering historical and offset well reactive approach to managing the health Figure 1 and KARTHICK VADIVEL, data, and running standard of their wells. illustrates the stan- engineering techniques across multiple dard methods applied for the candidate wells, prevents production engineers selection stage. from spending more time working on Depending on the scope of the proj- Well workovers and interventions are higher-value engineering workflows for ect, the realization of the workover or essential strategies for operators to sus- planning and execution of a workover or intervention opportunity can take a very tain or increase their production. How- intervention. This considerable time in- long time, sometimes even up to a year. ever, the well review process for these vestment also limits the volume of wells The well screening process alone can take jobs is time-consuming, involving numer- that can be evaluated over a given time- more than three months. ous manual and repetitive tasks to gather data and analyze results, resulting in proj- Fig. 1. Manual analysis approaches for candidate recognition. (Courtesy of Schlumberger) ects taking weeks or even months before getting to the planning and execution stage. In addition, best practices and les- sons learned from past interventions are rarely captured, preventing a systematic improvement of the well selection meth- odology and choice of interventions for future well reviews. Leveraging automation and advanced data analytics, Schlumberger, working with PT Saka Energi Indonesia (PGN Saka), developed a well optimization so- lution that not only automated many of the manual data gathering tasks but also the engineering and economic workflows that are used during the pre-planning stage for a well intervention or workover. This has helped PGN Saka’s engineers

World Oil® / JUNE 2020 29 PRODUCTION OPTIMIZATION

Due to this lengthy timeframe, reviews the actions can be used to enrich and im- neers, developed a well portfolio optimi- are generally only performed two to four prove the system. zation (WPO) solution termed Saka well times per year, and sometimes less, de- opportunity register, define and selection pending on the operator. This results in CASE STUDY (SWORDS) for the Ujung Pangkah as- significant time lags between the time PGN Saka operates the Ujung Pang- set. SWORDS is a decision-support ad- from review to execution in the field. The kah assets, an offshore oil and gas field in visory system that incorporates decades delay adversely affects the realization of Indonesia. Ujung Pangkah field is under- of proven analysis production potential. going a major development that involves methods and best practices, combined well intervention, workover and surface with PGN Saka’s business logic, to auto- SOLUTION optimization. To boost overall perfor- matically and autonomously: The adoption of various analytics and mance across the asset, PGN Saka wanted • identify underperforming wells machine learning tools into engineering to uncover opportunities to streamline its through their performance workflows, such as regression and clas- well review process. signatures sification models, make it possible to The main reservoir in Ujung Pangkah • determine root causes automate and improve many of the man- field is Kujung 1 carbonate (K1Z2-3), an • recommend the most appropriate ual processes that are used in a traditional early Miocene platform margin carbonate remedial interventions well workover or intervention programs. reef buildup. It is a multi-layered carbonate • probabilistically quantify post- Using these tools, Schlumberger has reservoir, which produces through multi- workover production developed a knowledge-based system lateral wells. Because it is a mature field, • evaluate risked economic viability (Fig. 2) that provides a standardized, au- the process of selecting wells for produc- of workover candidates tomated approach for the well opportunity tion enhancement projects can be tedious • estimate the chance of success for maturation process (OMP). The system is and challenging. PGN Saka was using a each intervention. based on a hybrid and automated decision- manual process to conduct its well reviews The SWORDS solution runs continu- support system anchored in a knowledge- and relied on each engineer’s individual ously in the background of PGN Saka’s based framework. Hybrid refers to the experience to assess the wells, as most operations, requiring only minimal manu- integration of the petroleum engineering of the best practices and lessons learned al interaction. Thus, PGN Saka’s engineers analysis methods and the operator’s busi- from previous projects weren’t captured. receive constantly updated insights on the ness logic with advanced machine learning In 2018, a workshop was conducted intervention and workover opportunities algorithms, to autonomously identify well to uncover solutions that would increase across the asset. Figure 3 describes the key performance signature and opportunities PGN Saka’s hydrocarbon production and elements of the advisory system. that are best aligned with the operator’s reserves in a cost-effective manner, and The key objectives of the SWORDS economic considerations. accelerate recovery and decision-making solution were to substantially decrease This system can rapidly screen and time. The workshop found that just re- time and effort for well reviews; automate rank high-well-count assets (hundreds viewing the results of the workover can- a manual well OMP, in particular the initial or thousands of wells) in a fraction of didate program took PGN Saka manage- candidate screening; increase decision ac- the time, compared with traditional ap- ment around 200 days. Engineers spent curacy and quality; and enable PGN Saka proaches. Such a solution enables the another 90 to 120 days conducting a field- to go from reactive “firefighting” mode to proactive management of existing wells wide well review to plan annual workover a more proactive approach, and shift engi- by keeping the production enhancement and intervention campaigns. Because of neering focus to higher cognitive engineer- opportunity pipeline full and expediting this significant time investment, PGN ing efforts required for the planning and potential candidates. The system detects, Saka only conducted reviews one to two execution of an intervention or a workover. diagnoses and recommends the appropri- times per year. The SWORDS design introduced new ate actions required to ensure that wells To address this issue, Schlumberger, automation and data analytics technolo- remain healthy. The results captured from working closely with PGN Saka engi- gies and a standard approach to well port- folio optimization for PGN Saka. Working closely with PGN Saka engineers, the fol- Fig. 2. Knowledge-based system for well portfolio optimization. (Courtesy of Schlumberger) lowing components of the new workflow, shown in Fig. 3, were evaluated to see if they could be automated or improved: 1. knowledge base 2. technical screening 3. ranking (production analytics ranking) 4. problem analysis 5. opportunity identification 6. technical potential or gain estimation 7. risk assessment 8. historical success 9. economic analysis.

30 JUNE 2020 / WorldOil.com PRODUCTION OPTIMIZATION

Adding a new, robust and integrated An automated decline curve analy- interventions. The provided data were database as the knowledge base, which sis (DCA) workflow was developed to conditioned according to the opportuni- holds all the variables and values needed screen outliers, detect segmentation and ties being evaluated in SWORDS. A crite- by the analysis, was critical to ensuring forecast remaining reserves, based on his- rion was defined to determine how a job result quality. torical production data. The forecasting could be considered successful. Steps two through six, which are the workflow embedded within this solution Once the technical analysis results technical analysis workflow components was used as a foundation for automating were reviewed, the final step was the shown in Fig. 3, were modified with the the forecasting process. economic analysis. SWORDS economic approaches described in Fig. 4. These The next step was to assess the risk analysis adopted the Indonesian PSC steps were key to evaluating the quality of factor, which was determined from an fiscal tax model, which is represented in candidate wells. exercise based on 12 sub-categories. This Fig. 5, to rank the final list of candidates A performance-based initial screen- differentiated the candidate wells, based with the opportunity’s profitability. This ing was developed to group wells as un- on the amount of assessed risk they pos- was achieved through a set of two fac- der- and over-performers. This screening sessed, both from subsurface and opera- tors: The first was the expected value and incorporated multi-level operator KPIs tional perspectives. Some of the examples cost per equivalent barrel of hydrocarbon (well, aerial and field level) to compute of risks being identified included: High gain, and the second was the cost per bar- candidate scores, using multi-criteria de- water production; pressure depletion; liq- rel, which helped aid the selection pro- cision-making analysis techniques. uid loading; poor cement; tubing accessi- cess, if two opportunities had a similar At the problem analysis stage, the goal bility; supply line availability; scale expected value. was to identify whether any constraints and sand production; formation damage; The expected value accounts for tech- existed in the well at any level, including and well interference. nical potential, costs, fiscal model, risk at the completion or the wellbore level. For the historical success stage, his- analysis and intervention history analysis, This analysis was heavily dependent on torical workover and well service jobs and is derived with the equation below. deploying analytics to diagnose the well from Ujung Pangkah field were used as an Incremental NPV × problems. For example, Chan analysis input to SWORDS for determining the Expected Value = (1 – Risk Factor) × was deployed to determine the root cause historical success rate of recommended Historical Success of high water production. At the opportunity identification stage, Fig. 3. SWORDS WPO solution components. (Courtesy of Schlumberger) the goal was to identify the workover or intervention opportunities that could be implemented in the candidate well to solve the highlighted problems and im- prove the production. This was based on a problem-opportunity logic set driven by WPO solution components like reserves and analytics signatures. For the technical potential or gain esti- mation stage, data-based models estimat- ed how much production could be im- proved in the candidate well, and, when applicable, found the deliverability of new zones that could be opened in a well. Fig. 4. SWORDS WPO technical analysis section. (Courtesy of Schlumberger) Using the data from producing comple- tions, a decision random forest model was deployed to determine the potential of vir- gin reservoirs and recommend new com- pletions. The unproduced zones were iden- tified as “behind-casing opportunities.” The technical potential was based on automated forecasting. There were ways to improve current practice by creating automated forecasting and ac- cessing completion level analysis. Prior to SWORDS, forecasts were done at the string level. While string-level fore- casts are not uncommon, it doesn’t take zonal performance into account. It was, therefore, recommended to forecast at the completion level using production allocations.

World Oil® / JUNE 2020 31 PRODUCTION OPTIMIZATION

There were workovers and interven- standardized, domain-driven, fully auto- ACKNOWLEDGEMENT This article contains elements from paper SPE-196326-MS, tions performed following the recom- mated and able to capture best practices presented at the SPE/IATMI Asia Pacific Oil & Gas Conference & mendation from SWORDS. The results and lessons learned from past interven- Exhibition (APOGCE), held in Bali, Indonesia, Oct. 29–31, 2019. are shown in Table 1. tions, operators can maximize the poten- Almost all SWORDS recommenda- tial of their assets. CIO CIO MARIO is base management team tions resulted in additional oil and gas Solutions, such as SWORDS, are in- leader for PT Saka Energi Indonesia. Prior to production, and only one job (B-04) did dicative of our industry’s transition to his current role, Mr. Mario worked for Chevron Pacific Indonesia and Brunei Shell Petroleum not meet the target. From the investiga- more intelligent operations. SWORDS as a senior production engineer/production tion, the unsuccessful result of B-04 was enables the screening of candidates, technologist. He has more than 20 years of experience in the oil and gas industry, and holds due to an inaccurate estimation of the res- portfolio-wide, to identify economically bachelor’s and master’s degrees in petroleum ervoir pressure. viable workovers or interventions with engineering, both from Bandung Institute of PGN Saka reported that by using the greater accuracy and certainty. Technology, Indonesia. SWORDS solution, they performed full Furthermore, the automation of the RINI SAPUTRA recently joined Semeru Energy well optimization, workover and inter- manual tasks opens more opportunities as subsurface manager, the same role she held vention across 196 completions weekly, for engineering innovation during the in the past three years with PGN Saka. Prior to this role, she worked as a reservoir engineering compared to yearly on the previous man- planning and execution stages of a well specialist in HCML, CNOOC and , working ual review. intervention. In the current operating en- in a great variety of upstream activities in PGN Saka saved 89% of time spent vironment, enabling engineering talent to various asset locations in Indonesia, Spain, Algeria, Brazil and Bolivia. Her recent project in both performing and reviewing candi- focus on the tasks that are the highest and was data analytics and automation in workover dates for workovers, and 88% cost-savings best use of their time is more important to opportunity identification. She holds a bachelor’s resulted from eliminating routine work operators than ever. degree in petroleum engineering from Trisakti University, and a master’s degree in reservoir and reducing time—worth an estimated In summary, a comprehensive reason- geophysics from University of Indonesia. $160,190 per year. In addition, PGN Saka ing system, such as the SWORDS solution, is expecting a 48% increase in ROI over maximizes ROI by identifying an evergreen SERA SAIFUDDIN is a production engineer for PT Saka Energi Indonesia, mainly for the Ujung the next five years. list of workover and intervention opportu- Pangkah assets since 2010. Prior to her current nities that are rigorously screened against role, Ms. Saifuddin worked for Hess Indonesia CONCLUSION Pangkah Limited as a facilities engineer and operational and physical constraints. This production engineer. She has 10 years of With the creation of a digitally driven has brought significant business value to experience in the oil and gas industry, and holds comprehensive reasoning system that is PGN SAKA and other operators. bachelor’s and master’s degrees in chemical engineering, both from University of Indonesia.

Fig. 5. Economic analysis in the SWORDS workflow. TRIYONO TRIYONO is a senior reservoir engineer for PT Saka Energi Indonesia, mainly for non-operating assets. Prior to his current role, Mr. Triyono worked for Saka Indonesia Pangkah Limited as base management team leader and has more than 10 years of experience in the oil and gas industry. He holds bachelor’s and master’s degrees in petroleum engineering, both from Bandung Institute of Technology, Indonesia.

SABRINA METRA is a senior production engineer and production analytics lead at Schlumberger in Indonesia. Prior to her current role, she implemented various production simulation models and workflows for the digital oilfield in Far East Asia. She is also a certified trainer for production software and has been involved in database management projects across Indonesia. Ms. Metra holds a bachelor’s degree in petroleum engineering from Bandung Institute of Technology.

PRAPRUT SONGCHITRUKSA is a senior data Table 1. Recommendations realized from SWORDS. scientist with the production analytics team at Actual gain Schlumberger in Houston. He holds a doctorate (with allocation) in civil engineering from Purdue University and Well Target zone Intervention type Oil, BOPD, gas, Mscfd is a licensed professional engineer in Texas. B-10 K2_K3 Re-access zone 90 - RAJEEV RANJAN SINHA is a senior reservoir K1 3747 engineer with the production analytics team at B-17 LTC Perfo BCO and stimulation 974 Schlumberger in Houston. He holds a bachelor’s UTC 279 degree in petroleum engineering from Indian School of Mines. B-04 UTC Perfo BCO and stimulation 0 B-01 LTC, K1, K2, K2_K3 Flowline acid 10 KARTHICK VADIVEL is a data scientist with B-02 K2_K3 Flowline acid 10 the production analytics team at Schlumberger B-11 K2_K3 GLVCO 30 in Houston. He holds a bachelor’s degree in B-12 K2_K3 Gas lift optimization 40 electronics and telecommunication engineering from Solapur University.

Article copyright © 2020 by Gulf Publishing Company. All rights reserved. 32 JUNE 2020 / WorldOil.com Not to be distributed in electronic or printed form, or posted on a website, without express written permission of copyright holder.