Mohammad Mahdi Moshir Farahi, Mohammad Ahmadi, B. Dabir
{"title":"地质经济不确定条件下基于模型的多目标粒子群方法生产优化","authors":"Mohammad Mahdi Moshir Farahi, Mohammad Ahmadi, B. Dabir","doi":"10.2516/ogst/2021039","DOIUrl":null,"url":null,"abstract":"Optimization of the water-flooding process in the oilfields is inherently subject to several uncertainties arising from the imperfect reservoir subsurface model and inadequate data. On the other hand, the uncertainty of economic conditions due to oil price fluctuations puts the decision-making process at risk. It is essential to handle optimization problems under both geological and economic uncertainties. In this study, a Pareto-based Multi-Objective Particle Swarm Optimization (MOPSO) method has been utilized to maximize the short-term and long-term production goals, robust to uncertainties. Some modifications, including applying a variable in the procedure of leader determination, namely crowding distance, a corrected archive controller, and a changing boundary exploration, are performed on the MOPSO algorithm. These corrections led to a complete Pareto front with enough diversity on the investigated model, covering the entire solution space. Net Present Value (NPV) is considered the first goal that represents the long-term gains, while a highly discounted NPV (with a discount rate of 25%) has been considered short-term gains since economic uncertainty risk grows with time. The proposed optimization method has been used to optimize water flooding on the Egg benchmark model. Geological uncertainty is represented with ensembles, including 100 model realizations. The k-means clustering method is utilized to reduce the realizations to 10 to reduce the computing cost. The Pareto front is obtained from Robust Optimization (RO) by maximizing average NPV over the ensembles, as the conservative production plan. Results show that optimization over the ensemble of a reduced number of realizations by the k-means technique is consistent with all realizations’ ensembles results, comparing their cumulative density functions. Furthermore, 10 oil price functions have been considered to form the economic uncertainty space. When SNPV and LNPV are optimized, considering uncertainty in oil price scenarios, the Pareto front’s production scenarios are robust to oil price fluctuations. Using the robust Pareto front of LNPV versus SNPV in both cases, one can optimize production strategy conservatively and update it according to the current reservoir and economic conditions. This approach can help a decision-maker to handle unexpected situations in reservoir management.","PeriodicalId":19424,"journal":{"name":"Oil & Gas Science and Technology – Revue d’IFP Energies nouvelles","volume":"17 1","pages":""},"PeriodicalIF":1.8000,"publicationDate":"2021-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"2","resultStr":"{\"title\":\"Model-based production optimization under geological and economic uncertainties using multi-objective particle swarm method\",\"authors\":\"Mohammad Mahdi Moshir Farahi, Mohammad Ahmadi, B. Dabir\",\"doi\":\"10.2516/ogst/2021039\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"Optimization of the water-flooding process in the oilfields is inherently subject to several uncertainties arising from the imperfect reservoir subsurface model and inadequate data. On the other hand, the uncertainty of economic conditions due to oil price fluctuations puts the decision-making process at risk. It is essential to handle optimization problems under both geological and economic uncertainties. In this study, a Pareto-based Multi-Objective Particle Swarm Optimization (MOPSO) method has been utilized to maximize the short-term and long-term production goals, robust to uncertainties. Some modifications, including applying a variable in the procedure of leader determination, namely crowding distance, a corrected archive controller, and a changing boundary exploration, are performed on the MOPSO algorithm. These corrections led to a complete Pareto front with enough diversity on the investigated model, covering the entire solution space. Net Present Value (NPV) is considered the first goal that represents the long-term gains, while a highly discounted NPV (with a discount rate of 25%) has been considered short-term gains since economic uncertainty risk grows with time. The proposed optimization method has been used to optimize water flooding on the Egg benchmark model. Geological uncertainty is represented with ensembles, including 100 model realizations. The k-means clustering method is utilized to reduce the realizations to 10 to reduce the computing cost. The Pareto front is obtained from Robust Optimization (RO) by maximizing average NPV over the ensembles, as the conservative production plan. Results show that optimization over the ensemble of a reduced number of realizations by the k-means technique is consistent with all realizations’ ensembles results, comparing their cumulative density functions. Furthermore, 10 oil price functions have been considered to form the economic uncertainty space. When SNPV and LNPV are optimized, considering uncertainty in oil price scenarios, the Pareto front’s production scenarios are robust to oil price fluctuations. Using the robust Pareto front of LNPV versus SNPV in both cases, one can optimize production strategy conservatively and update it according to the current reservoir and economic conditions. This approach can help a decision-maker to handle unexpected situations in reservoir management.\",\"PeriodicalId\":19424,\"journal\":{\"name\":\"Oil & Gas Science and Technology – Revue d’IFP Energies nouvelles\",\"volume\":\"17 1\",\"pages\":\"\"},\"PeriodicalIF\":1.8000,\"publicationDate\":\"2021-01-01\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"2\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"Oil & Gas Science and Technology – Revue d’IFP Energies nouvelles\",\"FirstCategoryId\":\"5\",\"ListUrlMain\":\"https://doi.org/10.2516/ogst/2021039\",\"RegionNum\":4,\"RegionCategory\":\"工程技术\",\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"Q4\",\"JCRName\":\"ENERGY & FUELS\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"Oil & Gas Science and Technology – Revue d’IFP Energies nouvelles","FirstCategoryId":"5","ListUrlMain":"https://doi.org/10.2516/ogst/2021039","RegionNum":4,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q4","JCRName":"ENERGY & FUELS","Score":null,"Total":0}
Model-based production optimization under geological and economic uncertainties using multi-objective particle swarm method
Optimization of the water-flooding process in the oilfields is inherently subject to several uncertainties arising from the imperfect reservoir subsurface model and inadequate data. On the other hand, the uncertainty of economic conditions due to oil price fluctuations puts the decision-making process at risk. It is essential to handle optimization problems under both geological and economic uncertainties. In this study, a Pareto-based Multi-Objective Particle Swarm Optimization (MOPSO) method has been utilized to maximize the short-term and long-term production goals, robust to uncertainties. Some modifications, including applying a variable in the procedure of leader determination, namely crowding distance, a corrected archive controller, and a changing boundary exploration, are performed on the MOPSO algorithm. These corrections led to a complete Pareto front with enough diversity on the investigated model, covering the entire solution space. Net Present Value (NPV) is considered the first goal that represents the long-term gains, while a highly discounted NPV (with a discount rate of 25%) has been considered short-term gains since economic uncertainty risk grows with time. The proposed optimization method has been used to optimize water flooding on the Egg benchmark model. Geological uncertainty is represented with ensembles, including 100 model realizations. The k-means clustering method is utilized to reduce the realizations to 10 to reduce the computing cost. The Pareto front is obtained from Robust Optimization (RO) by maximizing average NPV over the ensembles, as the conservative production plan. Results show that optimization over the ensemble of a reduced number of realizations by the k-means technique is consistent with all realizations’ ensembles results, comparing their cumulative density functions. Furthermore, 10 oil price functions have been considered to form the economic uncertainty space. When SNPV and LNPV are optimized, considering uncertainty in oil price scenarios, the Pareto front’s production scenarios are robust to oil price fluctuations. Using the robust Pareto front of LNPV versus SNPV in both cases, one can optimize production strategy conservatively and update it according to the current reservoir and economic conditions. This approach can help a decision-maker to handle unexpected situations in reservoir management.
期刊介绍:
OGST - Revue d''IFP Energies nouvelles is a journal concerning all disciplines and fields relevant to exploration, production, refining, petrochemicals, and the use and economics of petroleum, natural gas, and other sources of energy, in particular alternative energies with in view of the energy transition.
OGST - Revue d''IFP Energies nouvelles has an Editorial Committee made up of 15 leading European personalities from universities and from industry, and is indexed in the major international bibliographical databases.
The journal publishes review articles, in English or in French, and topical issues, giving an overview of the contributions of complementary disciplines in tackling contemporary problems. Each article includes a detailed abstract in English. However, a French translation of the summaries can be provided to readers on request. Summaries of all papers published in the revue from 1974 can be consulted on this site. Over 1 000 papers that have been published since 1997 are freely available in full text form (as pdf files). Currently, over 10 000 downloads are recorded per month.
Researchers in the above fields are invited to submit an article. Rigorous selection of the articles is ensured by a review process that involves IFPEN and external experts as well as the members of the editorial committee. It is preferable to submit the articles in English, either as independent papers or in association with one of the upcoming topical issues.