Deep Conditional Generative Adversarial Network for History Matching with Complex Geologies

Multi-source information fused generative adversarial network model and data assimilation based history matching for reservoir with complex geologies

Developing automatic history matching (AHM) methods to replace the traditional manual history matching (MHM) approach in adjusting the permeability distribution of the reservoir simulation model has been studied by many authors. Because permeability values need to be evaluated at hundreds of thousands of grid cells in a typical reservoir simulation model, it is necessary to apply a reparameterization technique to allow the optimization algorithms to be implemented with fewer variables. In basic reparameterization techniques including zonation and pilot point methods, the calibrations are usually based solely on the production data with no systematic link to the geological and geophysical data, and therefore, the obtained permeability distribution may be not geologically consistent. Several other reparameterization techniques have attempted to preserve geological consistency by incorporating 4D seismic data; however, these techniques cannot be applied to our fractured basement reservoirs (FBRs) as they do not have 4D seismic data. Taking into account these challenges, in this study, an AHM methodology and workflow have been developed using a new reparameterization technique. This approach attempts to minimize the potential for geological nonconsistency of the calibrated results by linking the permeability to geophysical data. The proposed methodology can be applied to fields with only traditional geophysical data (3D seismic and conventional well logs). In the proposed workflow, the spatial distributions of seismic attributes and geomechanical properties were calculated and estimated from 3D seismic data and well logs, respectively. After that, a feed-forward artificial neural network (ANN) model trained by the back-propagation algorithm of the relationship between initial permeability with seismic attributes and geomechanical properties of their grid cell values is developed. Then, the calibration of the permeability distribution is performed by adjustment of the ANN model. Modification of the ANN model is performed using the simultaneous perturbation stochastic approximation (SPSA) algorithm to calibrate transmission coefficients in the ANN model to minimize the discrepancy between the simulated results and observed data. The developed methodology is applied to calibrate the permeability distribution of a simulation model of Bach Ho FBR in Vietnam. The effectiveness of the methodology is evident by comparing the historical matches with an available manually history-matched simulation model. The application shows that the proposed methodology could be considered as a suitable practical approach for adjusting the permeability distribution for FBR reservoir simulation models.

To improve the current industry standard one-parameter-at-a-time sensitivity analysis method, we propose a new sensitivity analysis framework that utilizes Plackett-Burman design and Random Forests (a well-known data mining method). The new framework significantly reduces the number of required simulation runs (i.e. samples), and at the same time compellingly improves the automatic history matching error. The proposed sensitivity analysis framework starts with generating samples/simulations using Plackett-Burman design. Here, each simulation is executed based on different combinations of the parameters' input values. Once the samples are ready, the parameters' input values and the target vector (i.e. the history matching error vector) are used to construct a Random Forests model. This model is used to rank the importance of each history matching parameter. Hence, parameters with low impact on the history matching error are discarded, and the remaining are used for Genetic Algorithm-based automatic history matching. The impact of an internal Random Forests parameter (number of decision trees) on the history matching error is also observed. The highly faulted reservoir with water injection is used for history matching and 10 parameters are defined uncertain consisting of fault transmissibilities, permeabilities, and connate water saturation. The aim is to match the bottom hole pressure and watercut for several oil producing wells. The one-parameter-at-a-time method requires 21 samples, and the selected top 4 parameters from this method are mainly fault transmissibilities. The resulting automatic history matching results gave a final error of 463.338. With the new framework, which requires only 12 samples instead of 21, the final error was 628.041, given that 100 trees are utilized. As we increased the number of trees to 500, the final error is significantly better (i.e. 36% improvement) than the results obtained from the one-parameter-at-a-time method. We can see that using the new framework, the majority of the top 4 parameters are related to permeabilities. Here, there is no significant change in terms of computation time as we increase the number of trees from 100 to 500; total computation time is less than 10 seconds. For this case study, not only this new framework requires significantly less number of initial simulation samples, it significantly improved the history matching error (as compared to the industry standard one-parameter-at-time sensitivity analysis method). Therefore, practicing engineers can utilize this framework to save time and simultaneously improve accuracy of the history matching parameter ranking. Parameter ranking computation with Random Forests takes less than 10 seconds.

Schemes for automatic history matching of reservoir modeling: A case of Nelson oilfield in UK

Technology Focus Production forecasting is one of the main tasks of engineers involved in reservoir management. To perform it, they resort to models, which vary from simple decline curves to complex numerical simulations. Regardless of the model’s level of complexity, uncertainty will always exist in the forecasts. Therefore, reservoir engineers devote significant efforts trying to incorporate data in these models (history matching), aiming to mitigate uncertainty and improve the reliability of the predictions. The reservoir-history-matching literature is quite extensive. The first works on automatic history matching were published in the 1960s. Nowadays, there is a wealth of history-matching methods. As a result, we observe that it is becoming a common industrial practice to generate multiple history-matched models, even for fairly complex problems with thousands of uncertainty variables and a large amount of data. However, we are still far from a definitive solution. Actually, it seems that there will be no single best history-matching technique. Instead, what we may have is the best method for the specific problem. Moreover, there are many challenges that are not completely addressed. In particular, most of the approaches seem overly simplistic to handle complex geology and structural uncertainties. Also, the computational demand for uncertainty quantification remains an issue. Therefore, industry and academia still have much work to do. Another point I want to mention is that, back in 2009, SPE promoted a forum meeting titled “Getting to Robust Production Forecasts.” This meeting became the startup of a Global Integrated Workshop Series on production forecasting. To date, there have been six meetings; the latest was held this year in Malaysia. These workshops have the ambitious goal of developing a production-forecasting guideline. I think we will hear more about this in the coming years. The papers summarized in this feature and the ones indicated in the additional-reading list are good examples of recent developments and field applications of history matching and forecasting. I hope you enjoy them. JPT Recommended additional reading at OnePetro: www.onepetro.org. SPE 163582 An Ensemble-Based Nonlinear Orthogonal Matching Pursuit Algorithm for Sparse History Matching of Reservoir Models by Ahmed H. Elsheikh, The University of Texas at Austin, et al. SPE 170622 Integrated Subsurface Uncertainty Study and Application—A Case Study of a Chalk Reservoir, Greater Ekofisk Area, North Sea by H.J. Yu, ConocoPhillips, et al. SPE 170945 Practical Considerations for Forecasting Production Data in Unconventional Reservoirs—Variable- Pressure Drop Case by P.W. Collins, Texas A&M University, et al.

The history matching, a key step in the reservoir numerical simulation, is performed to obtain a set of appropriate reservoir parameters to match the modeling results with the oilfield historical production data. The history matching problem is solved conventionally with Gauss-Newton, L-M, BFGS, and LBFGS methods, which require calculation of the gradient of the objective function, and is very slow in solving large-scale problems. In recent years, modern optimization algorithms, such as the genetic algorithm, particle swarm optimization, and differential evolution algorithm, have been applied in the reservoir history matching. In this paper, the hybrid particle swarm optimization and gravitational search algorithm (PSOGSA) is used to carry out the research on the automatic reservoir history matching. First, the basic concepts of the particle swarm optimization (PSO) and the gravity search algorithm (GSA) were introduced. Then, the concept and algorithm of PSOGSA were described. Moreover, the actual oil field data were used to establish a geological model and an objective function, and the PSOGSA was used to carry out the automatic history matching. The PSOGSA was compared with other algorithms. Finally, the effects of parameters on the PSOGSA were analyzed. The results show that the PSOGSA has a good performance in the automatic reservoir history matching, increases the diversity of particles, improves the global search capability, and accelerates the convergence. The PSOGSA is superior to the PSO or the GSA.

This paper presents a procedure to estimate relative permeability curves together with gridblock porosities and permeabilities by automatic history matching three-phase flow production data using the LBFGS algorithm for optimization and the adjoint method for generating sensitivity coefficients. Both power law and B-splines are considered for the representation of relative permeability curves. Power law is a simple representation; however, B-splines have the advantage of being able to accurately represent any set of relative permeability curves. In the B-spline representation, we provide a simple procedure based on a transformation of parameters (control points) defining the B-splines to ensure that monotonic curves are obtained. Without this transformation, the automatic history-matching process fails. If endpoint saturations are included as parameters, we show how to modify the adjoint procedure to account for the fact that initial conditions are sensitive to endpoint saturations. While the history-matching process is inherently non-unique, we show that reasonable estimates of relative permeability curves and porosity/ permeability fields can be obtained for a synthetic reservoir example. We also show the uncertainty quantification of the model parameters and reservoir performance prediction using the Randomized Maximum Likelihood (RML) method. Introduction Relative permeabilities are important parameters in reservoir performance calculations. Usually the relative permeability curves are obtained from labs through core flood tests. They can also be incorporated into the history matching procedure as model parameters. ARCHER and WONG (1973), and YANG and WATSON (1991) were among the early researchers to consider the estimation of relative permeability curves by history matching laboratory core flood data. A detailed literature review is given in REYNOLDS et al. (2004). REYNOLDS et al. (2004) also discussed the estimation of three-phase relative permeabilities by history matching roduction data. The current work is a natural extension of REYNOLDS et al. (2004). In this paper, we provide an alternative method for relative permeability estimation based on B-spline approximation. Compared to power law representation, Bsplines ave the advantage of being able to accurately represent any set of relative permeability curves. In the B-spline representation, we provide a simple procedure based on logtransformation of parameters (control points) defining the Bsplines to ensure the B-spline curves are monotonic or convex as used in CHEN et al. (2005). Since a reservoir simulator requires monotonic relative permeability curves, without this transformation, the automatic history-matching process may generate non-monotonic relative permeability curves and result in failed reservoir simulation run. A prior model for porosity and absolute permeability and relative permeability parameters is assumed to provide regularization, i.e., Bayesian estimation is applied to generate estimates. Model parameters, which are estimated by automatic history matching of production data, consist of gridblock porosties, gridblock absolute logpermeabilities and the parameters defining the relative permeability curves. We do not assume the endpoint saturations (i.e. irreducible water saturation, critical gas saturation and residual oil saturation) to be known and consider them as independent model parameters in history matching procedure.

Saturation-based localization within EnKF approach for history matching in SAGD processes

One of the challenges in automatic history matching is to ensure the preservation of the realistic geological features of reservoir models. Realism of the history matched model is vital because of its impact on the flow response and the reliability of the model predictions. The issue is that models with unrealistic geological features can still match history but would fail to deliver reliable predictions. The work tackles the following problems in history matching: (1) identify only realistic geometry of geological bodies in history matching using non-uniform intelligent prior distributions for the uncertain geological parameters; (2) assess uncertainty in the connectivity between permeable geobodies and automatically select the geomodel realisations best matching the production data. The intelligent geological priors and the connectivity measures are incorporated in an automated history matching framework, where adaptive stochastic sampling is used to find multiple history matched models to quantify uncertainty of the reservoir predictions. Use of intelligent priors, also, improves the history matching efficiency by restricting the search only to the geologically realistic models. Connectivity control improves flexibly of the model definition and allows handling uncertainty from multiple stochastic realisations in a more realistic way.

High resolution sequence stratigraphical methods have been used in the detailed geological modeling of an element of the Ness formation of a North Sea field. This updated geological description has been used as input to the stochastic geological model MOHERES. 14 realizations of facies architecture and petrophysical properties have been generated. The stochastically generated realizations have been scaled up to a refined element of an existing deterministic, history matched reservoir simulation model for the full Upper Brent reservoir. The upscaled versions of the realizations have been connected to the full field model, and reservoir simulations have been performed to compare twelve years of production history with simulated results for well production, RFT- and PLT-data. Six of the realizations gave good to very good results when compared to measured production data. These realizations were used for simulation of the future production performance to the year 2010, making it possible to estimate its uncertainty.

It is becoming common to use automatic or assisted techniques for history matching of oil and gas reservoirs. These techniques are complicated by the need to update the properties of the model that are linked to the geological representation of the reservoir but also affect the misfit between observations and prediction. In this work we compare updating of regional properties for the former while updating local properties to satisfy the latter. We apply an automatic history matching approach to the Nelson field, a channelized turbidite reservoir, in order to better match the production rates of oil and water by well. We also integrate time lapse seismic data and we compare the result with production history matching output. We control the shale distribution within and between the major channel complexes by updating on the basis of geological environment variables which are regionally distributed. We compare this approach with more local updating by using pilot points with Kriginig. By updating the reservoir properties by environment variables we reduce the total production history match misfit by 22% for six years of data. We also improve the forecast of a further three years by 20%. By updating locally after by environment, this improvement is 60% and 30% in the matching and forecasting periods respectively. By integrating time lapse seismic data in history matching to update by environment, the production misfit is reduced by an equivalent amount while forecasting is improved by 19%. Seismically we reduce the misfit by 7%. Appropriate updating of the reservoir model is very important for reservoir management to reduce risk in development and to select appropriate models for forecasting. By updating the reservoir based on honouring the geological information or by using the geostatistical tools we can obtain a better representative of the reservoir with stronger prediction capability.

Computational visual perception, also known as computer vision, is a field of artificial intelligence that enables computers to process digital images and videos in a similar way as biological vision does. It involves methods to be developed to replicate the capabilities of biological vision. The computer vision’s goal is to surpass the capabilities of biological vision in extracting useful information from visual data. The massive data generated today is one of the driving factors for the tremendous growth of computer vision. This survey incorporates an overview of existing applications of deep learning in computational visual perception. The survey explores various deep learning techniques adapted to solve computer vision problems using deep convolutional neural networks and deep generative adversarial networks. The pitfalls of deep learning and their solutions are briefly discussed. The solutions discussed were dropout and augmentation. The results show that there is a significant improvement in the accuracy using dropout and data augmentation. Deep convolutional neural networks’ applications, namely, image classification, localization and detection, document analysis, and speech recognition, are discussed in detail. In‐depth analysis of deep generative adversarial network applications, namely, image‐to‐image translation, image denoising, face aging, and facial attribute editing, is done. The deep generative adversarial network is unsupervised learning, but adding a certain number of labels in practical applications can improve its generating ability. However, it is challenging to acquire many data labels, but a small number of data labels can be acquired. Therefore, combining semisupervised learning and generative adversarial networks is one of the future directions. This article surveys the recent developments in this direction and provides a critical review of the related significant aspects, investigates the current opportunities and future challenges in all the emerging domains, and discusses the current opportunities in many emerging fields such as handwriting recognition, semantic mapping, webcam‐based eye trackers, lumen center detection, query‐by‐string word, intermittently closed and open lakes and lagoons, and landslides.

Limited by the computing capacity of current computer and difficulty in inversion of large quantity of reservoir parameters for large scale reservoir history matching, the current automatic history matching methods are still hard to be applied in real fields. A parameterization method based on matrix decomposition is proposed to reduce the computational cost and improve the efficiency of history matching without decreasing the precision by mapping the original optimization problem from high dimensional space to low dimensional space. Base on the updated reservoir model after history matching, a reservoir production optimization control model is established. Considering the complexity of gradient calculation for history matching and production optimization objective function, a new general upscaling stochastic approximation (GUSA) algorithm is given based on the approximation perturbation gradient algorithm. The GUSA algorithm aims to get maximum increment of an objective function in unit step by obtaining an optimal lower triangular matrix in each iterative step. Toy problem results show that the calculation efficiency is improved much compared with the simultaneous perturbation stochastic approximation (SPSA) algorithm. For application of a real reservoir, a good history matching result is obtained for historic production data that verifies the validity of parameterization method and GUSA algorithm. An optimal control for production optimization performs an increase of oil production and a decrease of water cut. Besides, the optimal production scheme is accordant with the practical characteristics of dynamic production, which also proves the availability of GUSA algorithm.KeywordsHistory matchingParameterization methodProduction optimal controlApproximate gradientOptimization algorithm

The use of gradient-based optimization methods in automatic history matching requires that the reservoir simulator provides the necessary derivative information in the most efficient way. The implementation of these techniques in existing complex simulators is a difficult task, however. This paper describes the implementation of the forward and adjoint methods for derivative calculation in a full-featured adaptive implicit black-oil simulator. To show how the derivative calculation methods can be used even outside the framework of standard optimization algorithms, a new algorithm for automatic history matching, based on the truncated singular value decomposition, is proposed. Application to simple synthetic cases shows promising results.

Most geostatistical methods for generating plausible facies models require prior knowledge of the facies proportions distribution to account for heterogeneity within the reservoir. This parameter affects the flow behavior, predictive performance and fluid volume distribution of the model. In practice, it is hardly known with certainty and it is important to account for this uncertainty in the modeling and history matching phases. The ensemble Kalman filter (EnKF), when modified appropriately, to account for non-linearity and non-Gaussianity, is quite robust for automatic history matching problems. However, because the problem of history matching applied to facies is complex, most attempts at solving this problem typically assume constant petrophysical properties and stationary facies proportions. However, realistic representation of facies distribution requires incorporating the variation in petrophysical properties in the geologic model, in addition to any associated uncertainty. This is critical to correctly defining the flow connectivity within the model and the variations in storativity within facies. In this paper, we investigate a number of practical issues in history matching applied to facies with heterogeneous petrophysical properties and non-stationary proportions: (1) Can we account for heterogeneity in facies petrophysical properties during history matching and subsequently reduce any associated uncertainty in them by sequential calibration to data? To the best of our knowledge, this is the first time that heterogeneity in facies petrophysical properties has ever been applied to problems of this kind; (2) What is the EnKF performance when the facies proportions are stationary but implicitly uncertain? (3) Can EnKF efficiently adjust the facies boundaries and distributions when the prior proportions are assumed known but non-stationary? To answer these questions we designed three different test problems. We show that when the facies proportion is non-stationary the EnKF is a robust in estimating the facies properties and distribution.