Enhancing Reservoir Simulation History Matching Using SHAP Value for Parameter Range Selection; Energies; Vol. 19, iss. 10

מידע ביבליוגרפי
Parent link:Energies.— .— Basel: MDPI AG
Vol. 19, iss. 10.— 2026.— Article number 2285, 36 p.
מחברים אחרים: Magdeev A. M. Adel Maratovich, Al-Shargabi M. A. T. S. Mokhammed Abdulsalam Takha Sallam, Matveev I. V. Ivan Vasiljevich, Smirnov I. E. Ivan Evgenjevich, Shishaev G. Yu. Gleb Yurievich, Davoodi Sh. Shadfar, Eremyan G. A. Grachik Araikovich
סיכום:Title screen
Accurate reservoir history matching is indispensable for reliable subsurface forecasting and optimal field development; however, it remains a formidable challenge due to the high dimensionality of geological parameter spaces, strong nonlinear dynamics, and the competing demands of computational efficiency and data fidelity. Conventional automated history matching workflows frequently rely on Tornado sensitivity analysis, which evaluates parameters in isolation and fails to capture critical interdependencies that govern reservoir response, leading to inefficient exploration and suboptimal convergence. This study, therefore, introduces a novel Shapley Additive Explanations (SHAP)-assisted adaptive history matching framework that integrates SHAP with a CatBoost (v1.0.0) regressor to enable quantitative, model-agnostic assessment of parameter contributions while explicitly resolving non-linear feature interactions unattainable with conventional methods. The workflow establishes a systematic methodology for refining parameter uncertainty ranges prior to optimization through SHAP dependence plots, which identify subspaces where the objective function decreases, yielding 25–68% range reductions across all cases without compromising geological plausibility. Both SHAP-assisted and Tornado-based workflows, coupled with identical Particle Swarm Optimization settings, are rigorously validated on two synthetic reservoirs (SRM-6, Egg) and a real Siberian field case. Results demonstrate that the SHAP-assisted approach achieves equivalent or superior history match quality while reducing required optimization cycles by 6–60%, depending on model complexity, delivering lower objective function values and improved alignment with observed data. Computational efficiency is further enhanced by reusing the Latin Hypercube Sampling dataset for surrogate training, sensitivity analysis, and range refinement, contrasting sharply with Tornado’s requirement for 2N + 1 dedicated simulation runs with limited reusability. This framework advances interpretable, data-driven history matching by providing engineers with a systematic method to prioritize calibration efforts, reduce computational burden, and improve forecast reliability in complex reservoir systems
Текстовый файл
שפה:אנגלית
יצא לאור: 2026
נושאים:
גישה מקוונת:https://doi.org/10.3390/en19102285
פורמט: אלקטרוני Book Chapter
KOHA link:https://koha.lib.tpu.ru/cgi-bin/koha/opac-detail.pl?biblionumber=687172

MARC

LEADER 00000naa0a2200000 4500
001 687172
005 20260630113256.0
090 |a 687172 
100 |a 20260630d2026 k||y0rusy50 ba 
101 0 |a eng 
102 |a CH 
135 |a drcn ---uucaa 
181 0 |a i   |b  e  
182 0 |a b 
183 0 |a cr  |2 RDAcarrier 
200 1 |a Enhancing Reservoir Simulation History Matching Using SHAP Value for Parameter Range Selection  |f Grachik Eremyan, Adel M. Magdeev, Mohammed Al-Shargabi [et al.] 
203 |a Текст  |b визуальный  |c электронный 
283 |a online_resource  |2 RDAcarrier 
300 |a Title screen 
320 |a References: 70 tit 
330 |a Accurate reservoir history matching is indispensable for reliable subsurface forecasting and optimal field development; however, it remains a formidable challenge due to the high dimensionality of geological parameter spaces, strong nonlinear dynamics, and the competing demands of computational efficiency and data fidelity. Conventional automated history matching workflows frequently rely on Tornado sensitivity analysis, which evaluates parameters in isolation and fails to capture critical interdependencies that govern reservoir response, leading to inefficient exploration and suboptimal convergence. This study, therefore, introduces a novel Shapley Additive Explanations (SHAP)-assisted adaptive history matching framework that integrates SHAP with a CatBoost (v1.0.0) regressor to enable quantitative, model-agnostic assessment of parameter contributions while explicitly resolving non-linear feature interactions unattainable with conventional methods. The workflow establishes a systematic methodology for refining parameter uncertainty ranges prior to optimization through SHAP dependence plots, which identify subspaces where the objective function decreases, yielding 25–68% range reductions across all cases without compromising geological plausibility. Both SHAP-assisted and Tornado-based workflows, coupled with identical Particle Swarm Optimization settings, are rigorously validated on two synthetic reservoirs (SRM-6, Egg) and a real Siberian field case. Results demonstrate that the SHAP-assisted approach achieves equivalent or superior history match quality while reducing required optimization cycles by 6–60%, depending on model complexity, delivering lower objective function values and improved alignment with observed data. Computational efficiency is further enhanced by reusing the Latin Hypercube Sampling dataset for surrogate training, sensitivity analysis, and range refinement, contrasting sharply with Tornado’s requirement for 2N + 1 dedicated simulation runs with limited reusability. This framework advances interpretable, data-driven history matching by providing engineers with a systematic method to prioritize calibration efforts, reduce computational burden, and improve forecast reliability in complex reservoir systems 
336 |a Текстовый файл 
461 1 |t Energies  |c Basel  |n MDPI AG 
463 1 |t Vol. 19, iss. 10  |v Article number 2285, 36 p.  |d 2026 
610 1 |a электронный ресурс 
610 1 |a труды учёных ТПУ 
610 1 |a Shapley Additive Explanations analysis 
610 1 |a CatBoost regression 
610 1 |a history matching 
610 1 |a reservoir simulation 
610 1 |a uncertainty quantification 
701 1 |a Magdeev  |b A. M.  |g Adel Maratovich 
701 1 |a Al-Shargabi  |b M. A. T. S.  |c specialist in the field of petroleum engineering  |c Engineer of Tomsk Polytechnic University  |f 1993-  |g Mokhammed Abdulsalam Takha Sallam  |9 22768 
701 1 |a Matveev  |b I. V.  |c specialist in the field of petroleum engineering  |c engineer of Tomsk Polytechnic University, Candidate of physical and mathematical sciences  |f 1986-  |g Ivan Vasiljevich  |9 22147 
701 1 |a Smirnov  |b I. E.  |g Ivan Evgenjevich 
701 1 |a Shishaev  |b G. Yu.  |c Mathematician  |c Engineer of Tomsk Polytechnic University  |f 1984-  |g Gleb Yurievich  |9 21456 
701 1 |a Davoodi  |b Sh.  |c specialist in the field of petroleum engineering  |c Research Engineer of Tomsk Polytechnic University  |f 1990-  |g Shadfar  |9 22200 
701 1 |a Eremyan  |b G. A.  |c specialist in the field of petroleum engineering  |c Research Engineer, Tomsk Polytechnic University  |f 1989-  |g Grachik Araikovich  |9 22149 
801 0 |a RU  |b 63413507  |c 20260626 
850 |a 63413507 
856 4 0 |u https://doi.org/10.3390/en19102285  |z https://doi.org/10.3390/en19102285 
942 |c CF