Layer | Fill | Outline |
---|
Map layers
Theme | Visible | Selectable | Appearance | Zoom Range (now: 0) |
---|
Fill | Stroke |
---|---|
Collaborating Authors
Results
Abstract Calibrating complex subsurface geological models against dynamic well observations yields to a challenging inverse problem which is known as history matching in oil and gas literature. The highly nonlinear nature of interactions and relationships between reservoir model parameters and well responses demand automated, robust and geologically consistent inversion techniques. The ensemble of calibrated and history matched models quality determines the reliability of production uncertainty assessment. Reliable production forecasting and uncertainty assessment are essential steps toward reservoir management and field development. The Bayesian framework is a widely accepted approach to incorporate dynamic production data to the prior probability distribution of reservoir models and obtain the posterior distribution of reservoir parameters. Uncertainly assessment is performed by sampling the posterior probability distribution which is a computationally challenging task. Markov-Chain Monte Carlo (MCMC) algorithm has shown successful application in reservoir model calibration and uncertainty quantification is recent years. MCMC can efficiently sample the high-dimensional and complex posterior probability distribution of reservoir parameters and generate history matched reservoir models that consequently can be used for production forecasting uncertainty assessment. MCMC method is a gradient-free approach which makes is favorable when gradient information is not available through reservoir simulation. In MCMC method normally to march to next iteration the new sample is independent of the previous sample and the proposal distribution is rather random. To improve the sampling procedure and make MCMC process more efficient we propose an approach based on locally varying mean (LVM) Kriging to base the new sample generation on the previous iteration sample. In this method, the previous sample is used as the varying mean map in the geostatistical simulation approach to generate the new proposal for the next iteration. Using LVM Kriging to relate the new sample to previous iteration sample, make the chain of samples in MCMC more correlated and geologically consistent. Also this new proposal distribution makes the sampling procedure more efficient and avoids random and arbitrary movements is the parameter space. We applied MCMC with LVM Kriging to a suite of 2D and 3D reservoir models and obtained the calibrated model. We observed that the application of the new proposal distribution based on LVM Kriging along with MCMC improved the quality of the samples and resulted in promising uncertainty quantification. We also observed meaningful improvement in calibrated reservoir models quality and uncertainty interval while utilizing LVM comparing to random proposal or transition distribution in MCMC. MCMC with LVM Kriging as proposal distribution results in improved uncertainty assessment through enhancing the quality of the generated samples from posterior probability distribution of reservoir model parameters. Traditional random or independent proposal distribution does not represent the dependency of the samples through MCMC chain and iterations while this challenge is addressed by combining MCMC with LVM.
- North America > United States (0.47)
- Europe > Austria (0.28)
- Reservoir Description and Dynamics > Reservoir Simulation > History matching (1.00)
- Reservoir Description and Dynamics > Reservoir Simulation > Evaluation of uncertainties (1.00)
- Information Technology > Artificial Intelligence > Machine Learning > Statistical Learning (1.00)
- Information Technology > Artificial Intelligence > Representation & Reasoning > Uncertainty > Bayesian Inference (0.69)
- Information Technology > Artificial Intelligence > Machine Learning > Learning Graphical Models > Undirected Networks > Markov Models (0.62)
- Information Technology > Artificial Intelligence > Machine Learning > Learning Graphical Models > Directed Networks > Bayesian Learning (0.54)
Abstract Adjusting geological properties of reservoirs to match production data obtained from dynamic well observations is a nontrivial task and thus results in a challenging inverse problem which is known as history matching in oil and gas literature. Of special interest in reservoir engineering is to develop and establish automated, robust and geologically consistent inversion approaches such that the highly nonlinear and underdetermined nature of interactions and relationships between reservoir model parameters and well responses can accurately be modeled. Furthermore, a reliable production data uncertainty assessment can be gained by the quality of the ensemble of calibrated and history matched models. Reliable production forecasting and uncertainty assessment are fundamental steps toward reservoir management and field development. The Bayesian framework is a widely accepted approach to incorporate dynamic production data to the prior probability distribution of reservoir models and obtain the posterior distribution of reservoir parameters. Uncertainly assessment is performed by sampling the posterior probability distribution which is a computationally challenging task. A very common and well-stablished technique towards reservoir model calibration and uncertainty quantification is Markov-Chain Monte Carlo (MCMC) algorithm that has recently attracted many due to its powerful and successful performance. The high-dimensional and complex posterior probability distribution of reservoir parameters can efficiently be sampled and generate history matched reservoir models by employing MCMC and thus can be utilized for production forecasting uncertainty assessment. Once gradient information is not available, e.g., in many reservoir simulation problems, MCMC approach is the method of choice due to its gradient-free procedure. In MCMC method normally to march to next iteration the new sample is independent of the previous sample and the proposal distribution is rather random. To improve the sampling procedure and make MCMC process more efficient we propose an approach based on locally varying mean (LVM) Kriging to base the new sample generation on the previous iteration sample. In this method, the previous sample is used as the varying mean map in the geostatistical simulation approach to generate the new proposal for the next iteration. Using LVM Kriging to relate the new sample to previous iteration sample, make the chain of samples in MCMC more correlated and geologically consistent. Also this new proposal distribution makes the sampling procedure more efficient and avoids random and arbitrary movements is the parameter space. We applied MCMC with LVM Kriging to a suite of 2D and 3D reservoir models and obtained the calibrated model. We observed that the application of the new proposal distribution based on LVM Kriging along with MCMC improved the quality of the samples and resulted in promising uncertainty quantification. We also observed meaningful improvement in calibrated reservoir models quality and uncertainty interval while utilizing LVM comparing to random proposal or transition distribution in MCMC.
- Europe (0.47)
- Asia > Middle East > Saudi Arabia (0.47)
- North America > United States (0.47)
- Reservoir Description and Dynamics > Reservoir Simulation > History matching (1.00)
- Reservoir Description and Dynamics > Reservoir Simulation > Evaluation of uncertainties (1.00)
- Information Technology > Artificial Intelligence > Machine Learning > Statistical Learning (1.00)
- Information Technology > Artificial Intelligence > Representation & Reasoning > Uncertainty > Bayesian Inference (0.69)
- Information Technology > Artificial Intelligence > Machine Learning > Learning Graphical Models > Directed Networks > Bayesian Learning (0.54)
Summary Advancements in horizontal-well drilling and multistage hydraulic fracturing have enabled economically viable gas production from tight formations. Reservoir-simulation models play an important role in the production forecasting and field-development planning. To enhance their predictive capabilities and to capture the uncertainties in model parameters, one should calibrate stochastic reservoir models to both geologic and flow observations. In this paper, a novel approach to characterization and history matching of hydrocarbon production from a hydraulic-fractured shale is presented. This new methodology includes generating multiple discrete-fracture-network (DFN) models, upscaling the models for numerical multiphase-flow simulation, and updating the DFN-model parameters with dynamic-flow responses. First, measurements from hydraulic-fracture treatment, petrophysical interpretation, and in-situ stress data are used to estimate the initial probability distribution of hydraulic-fracture and induced-microfracture parameters, and multiple initial DFN models are generated. Next, the DFN models are upscaled into an equivalent continuum dual-porosity model with analytical techniques. The upscaled models are subjected to the flow simulation, and their production performances are compared with the actual responses. Finally, an assisted-history-matching algorithm is implemented to assess the uncertainties of the DFN-model parameters. Hydraulic-fracture parameters including half-length and transmissivity are updated, and the length, transmissivity, intensity, and spatial distribution of the induced fractures are also estimated. The proposed methodology is applied to facilitate characterization of fracture parameters of a multifractured shale-gas well in the Horn River basin. Fracture parameters and stimulated reservoir volume (SRV) derived from the updated DFN models are in agreement with estimates from microseismic interpretation and rate-transient analysis. The key advantage of this integrated assisted-history-matching approach is that uncertainties in fracture parameters are represented by the multiple equally probable DFN models and their upscaled flow-simulation models, which honor the hard data and match the dynamic production history. This work highlights the significance of uncertainties in SRV and hydraulic-fracture parameters. It also provides insight into the value of microseismic data when integrated into a rigorous production-history-matching work flow.
- North America > United States (1.00)
- North America > Canada > British Columbia (1.00)
- North America > Canada > Alberta (1.00)
- Geology > Geological Subdiscipline > Geomechanics (1.00)
- Geology > Rock Type > Sedimentary Rock > Clastic Rock > Mudrock > Shale (0.82)
- North America > Canada > Alberta > Western Canada Sedimentary Basin > Alberta Basin > Deep Basin (0.99)
- North America > Canada > British Columbia > Western Canada Sedimentary Basin > Horn River Basin > Otter Park Formation (0.94)
- North America > Canada > British Columbia > Western Canada Sedimentary Basin > Horn River Basin > Muskwa Field > Muskwa Formation (0.94)
- Well Completion > Hydraulic Fracturing (1.00)
- Reservoir Description and Dynamics > Unconventional and Complex Reservoirs > Shale gas (1.00)
- Reservoir Description and Dynamics > Reservoir Simulation > History matching (1.00)
- Reservoir Description and Dynamics > Fluid Characterization > Fluid modeling, equations of state (1.00)
- Information Technology > Modeling & Simulation (1.00)
- Information Technology > Artificial Intelligence > Machine Learning > Statistical Learning (0.93)
- Information Technology > Artificial Intelligence > Representation & Reasoning > Uncertainty > Bayesian Inference (0.46)
- Information Technology > Artificial Intelligence > Machine Learning > Learning Graphical Models > Directed Networks > Bayesian Learning (0.46)
Summary In this work, we develop and apply a general methodology for optimal closed-loop field development (CLFD) under geological uncertainty. CLFD involves three major steps: optimizing the field-development plan on the basis of current geological knowledge; drilling new wells, and collecting hard data and production data; and updating multiple geological models on the basis of all the available data. In the optimization step, the number, type, locations, and controls for new wells (and future controls for existing wells) are optimized with a hybrid particle swarm optimization-mesh adaptive direct search algorithm. The objective here is to maximize expected (over multiple realizations) net present value (NPV) of the overall project. History matching is accomplished with an adjoint-gradient-based โrandomized maximum likelihoodโ procedure. Because the CLFD history-matching component is fast relative to the optimization component, we generate a relatively large number of history-matched models. Optimization is then performed with a set of โrepresentativeโ realizations selected from the full set of history-matched models. We introduce a systematic optimization with sample validation (OSV) procedure, in which the number of realizations used for optimization is increased if an appropriate validation criterion is not satisfied. The CLFD methodology is applied to 2D and 3D example cases. Results show that the use of CLFD increases the NPV for the โtrueโ (synthetic) model by 10 to 70% relative to that achieved by optimizing over a large number of prior realizations. We also compare the results for CLFD with OSV to results that use a fixed number of geological realizations. These comparisons show that the use of too few realizations in the CLFD optimization step can result in lower true-model NPVs, whereas OSV provides a systematic approach for determining the proper number of realizations.
- Europe (0.92)
- North America > United States > Texas (0.46)
- Research Report > New Finding (0.34)
- Research Report > Experimental Study (0.34)
- Information Technology > Artificial Intelligence > Representation & Reasoning > Optimization (1.00)
- Information Technology > Artificial Intelligence > Representation & Reasoning > Agents (1.00)
- Information Technology > Artificial Intelligence > Machine Learning > Statistical Learning (1.00)
- (3 more...)
An Efficient and Practical Workflow for Probabilistic Forecasting of Brown Fields Constrained by Historical Data
Yang, Chaodong (Computer Modelling Group Ltd) | Nghiem, Long (Computer Modelling Group Ltd) | Erdle, Jim (Computer Modelling Group Ltd) | Moinfar, Ali (Computer Modelling Group Ltd) | Fedutenko, Eugene (Computer Modelling Group Ltd) | Li, Heng (Computer Modelling Group Ltd) | Mirzabozorg, Arash (Computer Modelling Group Ltd) | Card, Colin (Computer Modelling Group Ltd)
Abstract Brown fields are fields with significant production history. Probabilistic forecasting for brown fields requires multiple history-matched models that are conditioned to available field production data. This paper presents a systematic and practical workflow to generate an ensemble of simulation models that is able to capture uncertainties in forecasts, while honoring the observed production data. The proposed workflow employs the Bayes theorem to define a posterior Probability Density Function (PDF) that represents model forecast uncertainty by incorporating the misfit between simulation results and measured production data. Previous workflows use the Markov Chain Monte Carlo (MCMC) sampling method, which requires an extremely large number (thousands) of simulation runs. To alleviate this drawback, a Proxy-based Acceptance-Rejection (PAR) sampling method is developed in this study to generate representative simulation models that characterize the posterior PDF using hundreds of simulation runs. The proposed workflow can be summarized in five key steps: Run an initial set of reservoir simulations by simultaneously varying multiple uncertain parameters using an experimental design method. Construct a proxy function using a Radial-Basis Function (RBF) neural network to approximate the posterior PDF calculated from the initial set of simulation results. Sample the posterior PDF using the PAR sampling method and run an ensemble of new simulation models using the sample values. The new simulation results obtained in step 3 are added to the training datasets generated in step 1 to improve the accuracy of the proxy function. Steps 3 and 4 are repeated until a predefined stop criterion is satisfied. Filter all simulation runs generated in step 3 using appropriate tolerance criteria for various history-match quality indicators. The selected filtered cases constitute the final ensemble of simulation models that can be further used for uncertainty quantification in forecasting. The proposed workflow is demonstrated on two reservoir simulation models: Synthetic case. The forecast uncertainty of the ninth SPE Comparative Solution Project (SPE9) simulation model is investigated with a synthetic production history and 32 uncertain parameters. Unconventional oil field case. The workflow is applied to an Eagle Ford shale oil well to determine P90 (conservative), P50 (most likely), and P10 (optimistic) cases of estimated ultimate recovery (EUR). Results of the workflow are compared to those obtained using the Metropolis-Hasting MCMC sampling method. The comparison shows that the proposed workflow only requires 800 simulation runs to obtain results as accurate as the MCMC method with 8000 simulation runs. This translates into a 10- times speedup, which makes the proposed workflow practical for many real reservoir simulation studies.
- North America > United States > Texas > Terry County (0.81)
- North America > United States > Texas > Gaines County (0.81)
- Europe > United Kingdom > North Sea > Southern North Sea (0.81)
- Geology > Rock Type > Sedimentary Rock > Clastic Rock > Mudrock > Shale (0.34)
- Geology > Petroleum Play Type > Unconventional Play (0.34)
- North America > United States > Texas > West Gulf Coast Tertiary Basin > Eagle Ford Shale Formation (0.99)
- North America > United States > Texas > Sabinas - Rio Grande Basin > Eagle Ford Shale Formation (0.99)
- North America > United States > Texas > Maverick Basin > Eagle Ford Shale Formation (0.99)
- (2 more...)
- Information Technology > Modeling & Simulation (1.00)
- Information Technology > Artificial Intelligence > Machine Learning > Statistical Learning (1.00)
- Information Technology > Artificial Intelligence > Representation & Reasoning > Uncertainty > Bayesian Inference (0.48)
- Information Technology > Artificial Intelligence > Machine Learning > Learning Graphical Models > Directed Networks > Bayesian Learning (0.48)
Abstract Uncertainty quantification is an important task in reservoir simulation studies used for decision making. There have been many techniques proposed in the SPE literature for quantifying uncertainty, such as Markov chain Monte Carlo (MCMC). MCMC is statistical method for sampling from an arbitrary probability distribution to quantifying uncertainty in reservoir simulation. The major difficulty in applying MCMC methods is high computational cost. The purpose of this paper is to demonstrate the performance of a new technique โ Multilevel Markov Chain Monte Carlo (MLMCMC) โ for quantifying uncertainty in reservoir simulation with less computional cost compared to Standard MCMC. MLMCMC algorithm is based on decomposing the desired results into a set of components calculated with different level of coarsening level. This technique demonstrated a speed up and provided a forecast with no significant loss in accuracy compared to Standard MCMC. It makes Monte Carlo estimation a feasible technique for uncertainty quanti?cation in reservoir simulation applications. There are only a few applications of MLMCMC in the petroleum industry as it is a new technique. We show results for two fields. The first is Teal South in the Gulf of Mexico and the second is Scapa in a North Sea.
- North America > United States (1.00)
- Europe > United Kingdom > North Sea > Central North Sea (0.29)
- Europe > United Kingdom > North Sea > Central North Sea > Central Graben > West Central Graben > Block 21/25 > Anasuria Cluster > Teal South Field > Skagerrak Formation (0.99)
- Europe > United Kingdom > North Sea > Central North Sea > Central Graben > West Central Graben > Block 21/25 > Anasuria Cluster > Teal South Field > Heather Formation (0.99)
- Europe > United Kingdom > North Sea > Central North Sea > Central Graben > West Central Graben > Block 21/25 > Anasuria Cluster > Teal South Field > Fulmar Formation (0.99)
- (2 more...)
- Reservoir Description and Dynamics > Reservoir Simulation > History matching (1.00)
- Reservoir Description and Dynamics > Reservoir Simulation > Evaluation of uncertainties (1.00)
- Information Technology > Artificial Intelligence > Machine Learning > Statistical Learning (1.00)
- Information Technology > Artificial Intelligence > Machine Learning > Learning Graphical Models > Undirected Networks > Markov Models (0.85)
- Information Technology > Artificial Intelligence > Representation & Reasoning > Uncertainty > Bayesian Inference (0.68)
- Information Technology > Artificial Intelligence > Machine Learning > Learning Graphical Models > Directed Networks > Bayesian Learning (0.68)
Assisted History Matching of Channelized Models Using Pluri-Principal Component Analysis
Chen, C.. (Shell International Exploration and Production) | Gao, G.. (Shell Global Solutions US Inc.) | Ramirez, B.A.. A. (Shell International Exploration and Production) | Vink, J.C.. C. (Shell Global Solutions US Inc.) | Girardi, A.M.. M. (Shell Brasil Exploration and Production)
Abstract Assisted history matching (AHM) a channelized model is a very challenging task because it is very difficult to gradually deform the discrete facies in an automated fashion, while preserving geological realism. Recently, different Karhunen-Loeve or Principal Component Analysis (PCA) methods have been introduced to map gridblock-based reservoir properties from the spatial domain to a reduced component space. AHM is then conducted by tuning the coefficients of principal components to match production data. However, reservoir models obtained by these methods may violate or alter prior geological or geostatistical descriptions. Improved methodologies based on truncation or CDF-based mapping have been proposed, but their application is limited to facies with natural ordering. In this paper, a Pluri-PCA method, which supports PCA with a Pluri-Gaussian model, is proposed to reconstruct geological and reservoir models with multiple facies. PCA extracts the major geological features from a large collection of training channelized models and generates gridblock-based properties and real-valued (i.e., non-integer valued) facies. The real-valued facies are mapped to discrete facies indicators according to pre-defined Rock-Type-Rules (RTRs) that determine the fraction of each facies and neighboring-connections between different facies. Like the Pluri-Gaussian method, changing the order of facies indicators (e.g., from 0/1/2 to 0/2/1 for shale/levee/sand facies) does not impact the facies models reconstructed by Pluri-PCA. Pluri-PCA also preserves both geological and geostatistical characteristics of the prior models. As part of this new method, we show how an ensemble of training realizations can be used to automatically build the RTRs. The proposed AHM workflow is developed by integrating Pluri-PCA with a derivative-free optimization algorithm. Furthermore, this workflow is validated on a real field case channelized model with 0.1 million gridblocks and three facies types. The models generated by Pluri-PCA satisfy the geological/geostatistical descriptions and therefore can be used to quantify the uncertainty in the forecasting results before and after production data assimilation. This workflow has great potential for practical applications in large-scale history-matching and uncertainty-quantification. Pluri-PCA is able to reparameterize the original non-Gaussian gridblock-based properties to a new group of Gaussian random parameters (i.e., PCA coefficients). Because Bayesian-based uncertainty quantification methods such as the Ensemble-Smoother and Randomized-Maximum-Likelihood are formulated by assuming a Gaussian prior, Pluri-PCA can effectively render these methods applicable for AHM in reservoir models that originally have a non-Gaussian prior uncertainty description.
- Reservoir Description and Dynamics > Reservoir Simulation > History matching (1.00)
- Reservoir Description and Dynamics > Reservoir Characterization > Geologic modeling (1.00)
- Data Science & Engineering Analytics > Information Management and Systems (1.00)
- Information Technology > Artificial Intelligence > Representation & Reasoning > Optimization (0.88)
- Information Technology > Artificial Intelligence > Representation & Reasoning > Uncertainty > Bayesian Inference (0.87)
- Information Technology > Artificial Intelligence > Machine Learning > Learning Graphical Models > Directed Networks > Bayesian Learning (0.66)
- Information Technology > Artificial Intelligence > Machine Learning > Statistical Learning > Principal Component Analysis (0.61)
Abstract The use of proxy models for optimisation of expensive functions has demonstrated its value since the 1990's in many industries. Within reservoir engineering, similar techniques have been used for over a decade for history matching, both in commercial tools and in-house software. In addition to efficient history matching, proxy models have a distinct advantage when performing uncertainty quantification of probabilistic forecasts. Markov Chain Monte Carlo (MCMC) methods cannot realistically be applied directly with reservoir simulations, and even fast proxy models can fail dramatically to adequately represent the range of uncertainty if implemented without due care. A pitfall of the use of proxy models is that they are considered โblack boxโ and their quality is difficult to measure. Engineers prefer to deal with deterministic simulation models which they can evaluate and understand. The main pitfall of simple random walk MCMC techniques, which have begun to appear within reservoir engineering workflows, is a focus on theoretical properties which are not observed in practical implementations. This gives rise to potential gross errors, which are not generally appreciated by practitioners. Advances in recent years within the field of Bayesian statistics have significantly improved this situation, but have not yet been disseminated within the oil and gas industry. This paper describes the limitations of random walk MCMC techniques which are currently used for reservoir prediction studies, and shows how Hamiltonian MCMC techniques, together with an efficient implementation of proxy models, can lead to a more reliable and validated probabilistic uncertainty quantification, whilst also generating a suitable ensemble of deterministic reservoir models. Scientific comparison studies are performed for both an analytical case and a realistic reservoir simulation case to demonstrate the validity of the approach. The benefit of this methodology is to allow asset teams to effectively manage reservoir decisions using a robust and validated understanding of uncertainty. It lays the scientific foundations for the next generation of uncertainty tools and workflows.
- Europe > United Kingdom (0.92)
- Africa (0.92)
- North America > United States > Texas (0.67)
- Overview (0.92)
- Instructional Material > Course Syllabus & Notes (0.45)
- Research Report > New Finding (0.45)
- North America > United States > Gulf of Mexico > Central GOM > East Gulf Coast Tertiary Basin > Mississippi Canyon > Block 582 > Medusa Field (0.99)
- Europe > Norway > North Sea > Central North Sea > Central Graben > Block 2/8 > Valhall Field > Tor Formation (0.99)
- Europe > Norway > North Sea > Central North Sea > Central Graben > Block 2/8 > Valhall Field > Hod Formation (0.99)
- (5 more...)
- Reservoir Description and Dynamics > Reservoir Simulation > History matching (1.00)
- Reservoir Description and Dynamics > Reservoir Simulation > Evaluation of uncertainties (1.00)
- Management > Risk Management and Decision-Making > Risk, uncertainty, and risk assessment (1.00)
- Data Science & Engineering Analytics > Information Management and Systems (1.00)
- Information Technology > Artificial Intelligence > Representation & Reasoning > Uncertainty > Bayesian Inference (1.00)
- Information Technology > Artificial Intelligence > Machine Learning > Learning Graphical Models > Directed Networks > Bayesian Learning (1.00)
- Information Technology > Artificial Intelligence > Machine Learning > Statistical Learning > Regression (0.67)
Integration of Principal-Component-Analysis and Streamline Information for the History Matching of Channelized Reservoirs
Chen, C.. (Shell International Exploration and Production) | Gao, G.. (Shell Global Solutions US Inc.) | Honorio, J.. (MIT) | Gelderblom, P.. (Shell Global Solutions International) | Jimenez, E.. (Qatar Shell GTL Limited) | Jaakkola, T.. (MIT)
Abstract Although Principal Component Analysis (PCA) has been widely applied to effectively reduce the number of parameters characterizing a reservoir, its disadvantages are well recognized by researchers. First, PCA may distort the probability distribution function (PDF) of the original model, especially for non-Gaussian properties such as facies indicator or permeability field of a fluvial reservoir. Second, it smears the boundaries between different facies. Therefore, the models reconstructed by PCA are generally unacceptable for geologists. A workflow is proposed to seamlessly integrate Cumulative-Distribution-Function-based PCA (CDF-PCA) and streamline information for assisted-HM on a two-facies channelized reservoir. The CDF-PCA is developed to reconstruct reservoir models using only a few hundred of principal components. It inherits the advantage of PCA to capture the main features or trends of spatial correlations among properties, and more importantly, it can properly correct the smoothing effect of PCA. Integer variables such as facies indicators are regenerated by truncating their corresponding PCA results with thresholds that honor the fraction of each facies at first, and then real variables such as permeability and porosity are regenerated by mapping their corresponding PCA results to new values according to the CDF curves of different properties in different facies. Therefore, the models reconstructed by CDF-PCA preserve both geological (facies fraction) and geostatistical (non-Gaussian distribution with multi-peaks) characteristics of their original or prior models. Our preliminary results indicate that the history-matched model using the CDF-PCA alone may not satisfy the requirement of geologists, e.g., some channels may become disconnected during history-matching. Therefore, we propose a method of combining CDF-PCA together with streamline information. Because velocity of the tracer in the streamline provides connectivity information between injectors and producers, it enhances channel connectivity without over-correction on cell-based permeability during the process of history matching. The CDF-PCA method is applied to a real-field case with three facies to quantify the quality of the models reconstructed. The history matching workflow is applied to a synthetic case. Our results show that the geological facies, reservoir properties, and production forecasts of models reconstructed with CDF-PCA are well consistent with those of the original models. The integrated HM workflow of CDF-PCA with streamline information generates reservoir models that honor production history with minimal compromise of geological realism.
- North America > United States (0.93)
- Europe (0.93)
- Overview > Innovation (1.00)
- Research Report (0.88)
- Geology > Rock Type (0.49)
- Geology > Sedimentary Geology > Depositional Environment (0.48)
- Information Technology > Modeling & Simulation (1.00)
- Information Technology > Artificial Intelligence > Representation & Reasoning > Uncertainty > Bayesian Inference (0.93)
- Information Technology > Artificial Intelligence > Machine Learning > Learning Graphical Models > Directed Networks > Bayesian Learning (0.67)
- Information Technology > Artificial Intelligence > Machine Learning > Statistical Learning > Principal Component Analysis (0.61)
Strategic Scope of Alternative Optimization Methods in History Matching and Prediction Workflows
Schulze-Riegert, R.. (SPT Group) | Chataigner, F.. (SPT Group) | Kueck, N.. (SPT Group) | Pajonk, O.. (SPT Group) | Baffoe, J.. (Clausthal University of Technology) | Ajala, I.. (Clausthal University of Technology) | Awofodu, D.. (Clausthal University of Technology) | Almuallim, H.. (Firmsoft Inc.)
Abstract Reservoir simulation workflows from history to prediction are built on a number of alternative optimization and sampling techniques with different characteristics. Adjoint techniques derive analytical sensitivities directly from the flow equations of the simulator. In a model update step those sensitivities are used for property modifications on grid cell level. Derivative-free optimization techniques like evolutionary algorithms are flexible and deliver alternative history matched cases. The propagation of reservoir uncertainties from history to prediction is often investigated using ensemble-based approaches. Markov Chain Monte Carlo and EnKF techniques are applied to generate an approximate posterior distribution as a basis for estimating prediction uncertainties. In this work we investigate alternative optimization, sampling and data assimilation techniques with application to history matching workflows within one application framework for comparison. Validated simulation models are used for production forecast and field development planning. Methods are investigated with a focus on history matching efficiency, flexibility to integrate different types of model parameters, integration option for static and dynamic modeling workflows, capability of supporting uncertainty quantification workflows for estimating prediction uncertainties and the ability to leverage on distributed computing. Optimization workflows are applied to a reasonably complex benchmark problem for comparison. Well production data is used in the model calibration phase and history matched simulation models are carried forward to prediction. The impact of alternative optimization concepts on estimating prediction uncertainties is discussed. This work gives an overview on alternative optimization concepts and relates capabilities, strengths and weaknesses. The impact of workflow choices on estimating prediction uncertainties is discussed. Practical conclusions are drawn for real field applications scenarios.
- Europe (0.93)
- Asia (0.68)
- North America > United States > Texas (0.28)
- Reservoir Description and Dynamics > Reservoir Simulation > History matching (1.00)
- Reservoir Description and Dynamics > Reservoir Simulation > Evaluation of uncertainties (1.00)
- Information Technology > Modeling & Simulation (1.00)
- Information Technology > Artificial Intelligence > Representation & Reasoning > Optimization (1.00)
- Information Technology > Artificial Intelligence > Machine Learning > Statistical Learning (1.00)
- (2 more...)