{"title":"利用代理模型和元启发式算法优化邻井设计:Bakken案例研究","authors":"Ahmed Merzoug, Vamegh Rasouli","doi":"10.3390/eng4020075","DOIUrl":null,"url":null,"abstract":"Fracture-driven interaction FDI (colloquially called “Frac-hit”) is the interference of fractures between two or more wells. This interference can have a significant impact on well production, depending on the unconventional play of interest (which can be positive or negative). In this work, the surrogate model was used along with metaheuristic optimization algorithms to optimize the completion design for a case study in the Bakken. A numerical model was built in a physics-based simulator that combines hydraulic fracturing, geomechanics, and reservoir numerical modeling as a continuous simulation. The stress was estimated using the anisotropic extended Eaton method. The fractures were calibrated using Microseismic Depletion Delineation (MDD) and microseismic events. The reservoir model was calibrated to 10 years of production data and bottom hole pressure by adjusting relative permeability curves. The stress changes due to depletion were calibrated using recorded pressure data from MDD and FDI. Once the model was calibrated, sensitivity analysis was run on the injected volumes, the number of clusters, the spacing between clusters, and the spacing between wells using Sobol and Latin Hypercube sampling. The results were used to build a surrogate model using an artificial neural network. The coefficient of correlation was in the order of 0.96 for both training and testing. The surrogate model was used to construct a net present value model for the whole system, which was then optimized using the Grey Wolf algorithm and the Particle Swarm Optimization algorithm, and the optimum design was reported. The optimum design is a combination of wider well spacing (1320 ft), tighter cluster spacing (22 ft), high injection volume (1950 STB/cluster), and a low cluster number per stage (seven clusters). This study suggests an optimum design for a horizontal well in the Bakken drilled next to a well that has been producing for ten years. The design can be deployed in new wells that are drilled next to depleted wells to optimize the system’s oil production.","PeriodicalId":10630,"journal":{"name":"Comput. Chem. Eng.","volume":"39 1","pages":"0"},"PeriodicalIF":0.0000,"publicationDate":"2023-04-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"3","resultStr":"{\"title\":\"Offset Well Design Optimization Using a Surrogate Model and Metaheuristic Algorithms: A Bakken Case Study\",\"authors\":\"Ahmed Merzoug, Vamegh Rasouli\",\"doi\":\"10.3390/eng4020075\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"Fracture-driven interaction FDI (colloquially called “Frac-hit”) is the interference of fractures between two or more wells. This interference can have a significant impact on well production, depending on the unconventional play of interest (which can be positive or negative). In this work, the surrogate model was used along with metaheuristic optimization algorithms to optimize the completion design for a case study in the Bakken. A numerical model was built in a physics-based simulator that combines hydraulic fracturing, geomechanics, and reservoir numerical modeling as a continuous simulation. The stress was estimated using the anisotropic extended Eaton method. The fractures were calibrated using Microseismic Depletion Delineation (MDD) and microseismic events. The reservoir model was calibrated to 10 years of production data and bottom hole pressure by adjusting relative permeability curves. The stress changes due to depletion were calibrated using recorded pressure data from MDD and FDI. Once the model was calibrated, sensitivity analysis was run on the injected volumes, the number of clusters, the spacing between clusters, and the spacing between wells using Sobol and Latin Hypercube sampling. The results were used to build a surrogate model using an artificial neural network. The coefficient of correlation was in the order of 0.96 for both training and testing. The surrogate model was used to construct a net present value model for the whole system, which was then optimized using the Grey Wolf algorithm and the Particle Swarm Optimization algorithm, and the optimum design was reported. The optimum design is a combination of wider well spacing (1320 ft), tighter cluster spacing (22 ft), high injection volume (1950 STB/cluster), and a low cluster number per stage (seven clusters). This study suggests an optimum design for a horizontal well in the Bakken drilled next to a well that has been producing for ten years. The design can be deployed in new wells that are drilled next to depleted wells to optimize the system’s oil production.\",\"PeriodicalId\":10630,\"journal\":{\"name\":\"Comput. Chem. Eng.\",\"volume\":\"39 1\",\"pages\":\"0\"},\"PeriodicalIF\":0.0000,\"publicationDate\":\"2023-04-23\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"3\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"Comput. Chem. Eng.\",\"FirstCategoryId\":\"1085\",\"ListUrlMain\":\"https://doi.org/10.3390/eng4020075\",\"RegionNum\":0,\"RegionCategory\":null,\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"\",\"JCRName\":\"\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"Comput. Chem. Eng.","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.3390/eng4020075","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
Offset Well Design Optimization Using a Surrogate Model and Metaheuristic Algorithms: A Bakken Case Study
Fracture-driven interaction FDI (colloquially called “Frac-hit”) is the interference of fractures between two or more wells. This interference can have a significant impact on well production, depending on the unconventional play of interest (which can be positive or negative). In this work, the surrogate model was used along with metaheuristic optimization algorithms to optimize the completion design for a case study in the Bakken. A numerical model was built in a physics-based simulator that combines hydraulic fracturing, geomechanics, and reservoir numerical modeling as a continuous simulation. The stress was estimated using the anisotropic extended Eaton method. The fractures were calibrated using Microseismic Depletion Delineation (MDD) and microseismic events. The reservoir model was calibrated to 10 years of production data and bottom hole pressure by adjusting relative permeability curves. The stress changes due to depletion were calibrated using recorded pressure data from MDD and FDI. Once the model was calibrated, sensitivity analysis was run on the injected volumes, the number of clusters, the spacing between clusters, and the spacing between wells using Sobol and Latin Hypercube sampling. The results were used to build a surrogate model using an artificial neural network. The coefficient of correlation was in the order of 0.96 for both training and testing. The surrogate model was used to construct a net present value model for the whole system, which was then optimized using the Grey Wolf algorithm and the Particle Swarm Optimization algorithm, and the optimum design was reported. The optimum design is a combination of wider well spacing (1320 ft), tighter cluster spacing (22 ft), high injection volume (1950 STB/cluster), and a low cluster number per stage (seven clusters). This study suggests an optimum design for a horizontal well in the Bakken drilled next to a well that has been producing for ten years. The design can be deployed in new wells that are drilled next to depleted wells to optimize the system’s oil production.