A practical and efficient iterative history matching workflow for shale gas well coupling multiple objective functions, multiple proxy-based MCMC and EDFM

Tripoppoom, Sutthaporn; Yu, Wei; Huang, Hao; Sepehrnoori, Kamy; Song, Wenjing; Dachanuwattana, Silpakorn

doi:10.1016/j.petrol.2019.01.080

Cited by 25 publications

(13 citation statements)

References 22 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The model performance is defined according to the root mean squared error (RMSE) metric, widely applied to assess models predicting numeric outcomes. 43,46,52 Moreover, multistep-ahead predictions are used to compute all performances since the final application of the model is to run this type of prediction. It should be highlighted that each model performance is defined using multiple multistep-ahead predictions, one for each set of geomechanical parameters.…”

Section: Hyperparameter Tuningmentioning

confidence: 99%

“…In addition, the longitudinal fluid flow inside the fracture channels follows the cubic law of parallel plates considering an initial fracture The proxy model of the KGD problem was built using several hydraulic fracturing numerical models based on the governing equations presented in Section 2. The Latin hypercube method generated 1000 sample points combining the geomechanical parameters investigated in the inverse analysis: Young's modulus 𝐸, tensile strength 𝜎 𝑡 , in situ compressive stress 𝜎 𝑜 , and rock permeability 𝑘. Tripoppoom et al [43][44][45][46] and Kohler et al 60 also employed this technique to build proxy models.…”

Section: Numerical Model and Dataset Generationmentioning

confidence: 99%

“…The model performance is defined according to the root mean squared error (RMSE) metric, widely applied to assess models predicting numeric outcomes 43,46,52 . Moreover, multistep‐ahead predictions are used to compute all performances since the final application of the model is to run this type of prediction.…”

Section: Artificial Neural Network Modelingmentioning

confidence: 99%

“…The Latin hypercube method generated 1000 sample points combining the geomechanical parameters investigated in the inverse analysis: Young's modulus E , tensile strength

{\sigma _t}

, in situ compressive stress

{\sigma _o}

, and rock permeability k . Tripoppoom et al 43–46 . and Kohler et al 60 .…”

Section: Parameter Identification Applicationmentioning

confidence: 99%

“…Several history matching procedures have considered uncertainty quantification of oil and gas applications using machine learning modeling to develop proxy models. Production data of hydraulic-fractured shale wells were matched using Markov chain Monte Carlo and k-nearest neighbors algorithm, 43 artificial neural networks, 44,45 or both. 46 However, these authors used the proxy model to perform predictions in some specified measured times.…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

Inverse analysis of hydraulic fracturing tests based on artificial intelligence techniques

Abreu

Mejia

Roehl

2022

Num Anal Meth Geomechanics

View full text Add to dashboard Cite

In the last decades, the study of fluid flow through discontinuities such as fractures has received significant attention from the petroleum industry owing to the discovery of naturally fractured reservoirs. Predicting the behavior of this type of reservoir is challenging and includes, as a crucial step, determining its geomechanical parameters. Several in situ tests and empirical relationships have been established to estimate these parameters. Nevertheless, these procedures have important limitations. This work proposes an artificial intelligence-based methodology to identify geomechanical parameters from borehole injection pressure curves obtained during hydraulic fracturing tests. This methodology applies a genetic algorithm to minimize the sum of squared differences between the observed and predicted borehole pressure curves. Furthermore, an artificial neural network is adopted to build a proxy model that substitutes hydraulic fracturing numerical simulations, substantially reducing computational time. A proper recursive strategy is adopted to predict the borehole pressure curves.Additionally, a hyperparameter tuning technique selects appropriate neural network hyperparameters based on the multistep-ahead prediction error. Finally, the framework is applied in a KGD problem that models hydraulic fracture propagation. The benchmark confirms the capability of the proposed methodology to identify geomechanical parameters from fracturing tests.

show abstract

Section: Hyperparameter Tuningmentioning

confidence: 99%

Section: Numerical Model and Dataset Generationmentioning

confidence: 99%

Section: Artificial Neural Network Modelingmentioning

confidence: 99%

“…The Latin hypercube method generated 1000 sample points combining the geomechanical parameters investigated in the inverse analysis: Young's modulus E , tensile strength

{\sigma _t}

, in situ compressive stress

{\sigma _o}

, and rock permeability k . Tripoppoom et al 43–46 . and Kohler et al 60 .…”

Section: Parameter Identification Applicationmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Inverse analysis of hydraulic fracturing tests based on artificial intelligence techniques

Abreu

Mejia

Roehl

2022

Num Anal Meth Geomechanics

View full text Add to dashboard Cite

show abstract

Parameter identification of minifrac numerical tests using a gradient boosting‐based proxy model and genetic algorithm

Abreu,

Mejia,

Roehl

et al. 2023

Num Anal Meth Geomechanics

View full text Add to dashboard Cite

In recent years, the petroleum industry has devoted considerable attention to studying fluid flow inside fracture channels due to the discovery of naturally fractured reservoirs. The behavior prediction of these reservoirs is a well‐known challenging task, in which the initial stage consists of identifying reservoir hydromechanical parameters. This work proposes an artificial intelligence‐based approach to identify hydromechanical parameters from borehole injection pressure curves acquired through minifrac tests. This approach combines proxy modeling with a stochastic optimization algorithm to match observed and predicted borehole pressure curves. Therefore, a gradient boosting‐based proxy model is built to predict borehole pressure curves, considering a proper strategy to develop time series modeling. Moreover, a Bayesian optimization algorithm is applied to compute the gradient boosting hyperparameters. In this optimization scenario, this paper proposes an appropriate objective function established from the assumed time series prediction strategy and the k‐fold cross‐validation. Finally, a genetic algorithm is adopted to identify unknown hydromechanical parameters, solving an inverse problem. Based on the proposed workflow, a study of the importance of the hydromechanical parameters is developed. To assess the methodology applicability, the approach is employed to identify parameters in synthetic and field minifrac tests. The results present how this approach can adequately identify hydromechanical parameters of hydraulic fracturing problems.

show abstract

Development and Application of Rock Fracture Propagation Numerical Simulation System

et al. 2022

View full text Add to dashboard Cite

A practical and efficient iterative history matching workflow for shale gas well coupling multiple objective functions, multiple proxy-based MCMC and EDFM

Cited by 25 publications

References 22 publications

Inverse analysis of hydraulic fracturing tests based on artificial intelligence techniques

Inverse analysis of hydraulic fracturing tests based on artificial intelligence techniques

Parameter identification of minifrac numerical tests using a gradient boosting‐based proxy model and genetic algorithm

Development and Application of Rock Fracture Propagation Numerical Simulation System

Contact Info

Product

Resources

About