**Peer Reviewed**

**Source**

**SPE Disciplines**

**Conference**

- EUROPEC 2015 (1)
- Offshore Technology Conference (2)
- Offshore Technology Conference Brasil (2)
- SPE Annual Technical Conference and Exhibition (2)
- SPE Europec featured at 78th EAGE Conference and Exhibition (2)
- SPE Europec featured at 79th EAGE Conference and Exhibition (1)
- SPE Europec featured at 80th EAGE Conference and Exhibition (1)
- SPE Europec featured at 81st EAGE Conference and Exhibition (2)
- SPE Europec featured at 82nd EAGE Conference and Exhibition (2)
- SPE Latin America and Caribbean Mature Fields Symposium (1)
- SPE Latin American and Caribbean Petroleum Engineering Conference (1)
- SPE Reservoir Simulation Symposium (1)

**Theme**

**Author**

- Almeida, Forlan La Rosa (2)
- Anyzewski, Alessandra Silva (1)
- Avansi, Guilherme Daniel (3)
- Botechia, Vinicius Eduardo (2)
- Caiado, Camila (3)
- Correia, Gil Gomes (1)
- Correia, Manuel Gomes (5)
- Câmara, Paulo Sérgio (1)
- Davolio, Alessandra (5)
- De Lima, Alexandre (1)
- de Souza Rios, Victor (1)
- de Souza Rios, Vinicius (1)
- Dos Santos, Marcos Sebastião (1)
- Drumond, Paulo Soares (1)
- Ferreira, Carla Janaina (2)
- Formentin, Helena Nandi (2)
- Fourno, André (1)
- Ghasemi, Mohammadreza (1)
- Gildin, Eduardo (2)
- Goldstein, Michael (3)
- Gomes, Carlos Eduardo de Aguiar Nogueira (1)
- Loomba, Ashish Kumar (2)
- Maschio, Celio (1)
- Maschio, Célio (1)
- Morosov, André Luís (2)
- Nandi Formentin, Helena (1)
- Noetinger, Benoit (1)
- Paes, Vinicius Costa Lopes (1)
- Pinto, Marcio Augusto Sampaio (2)
- Rosa Almeida, Forlan La (1)
- Santos, Antonio Alberto de Souza dos (2)
- Santos, Antonio Alberto Souza (2)
- Santos, Daniel Rodrigues dos (1)
- Santos, Letícia Siqueira dos (1)
- Santos, Susana Margarida da Graça (1)
**Schiozer, Denis José (25)**- Schmall dos Santos, Luiz Otávio (1)
- Silva, Celmar Guimarães da (1)
- Silva, Luis Otávio Mendes da (1)
- Silva, Marcia Ida Oliveira (1)
- Sorek, Nadav (1)
- Vernon, Ian (3)
- von Hohendorff Filho, João Carlos (2)

**Concept Tag**

- Acceptance Range (3)
- algorithm (2)
- application (3)
- Artificial Intelligence (22)
- asset and portfolio management (4)
- assimilation (2)
- background facies (2)
- bottom-up analysis (2)
- cycle 1 (2)
- cycle 2 (2)
- cycle 3 (2)
- dataset (2)
- Decision analysis (2)
- deviation (3)
- discrepancy (2)
- dual-medium flow model (2)
- edkm (2)
- emulation (2)
- emulator (3)
- EMV (4)
- Engineering (2)
- ensemble (2)
- estimation (2)
- evaluation (3)
- evolution (2)
- field development (2)
- field development optimization and planning (4)
- flow in porous media (7)
- flow model (2)
- flow simulation (2)
- Fluid Dynamics (7)
- fracture (4)
- fracture system (2)
- frequency distribution (2)
- geologic modeling (2)
- geological modeling (2)
- geostatistical realization (2)
- Goldstein (2)
- grid (2)
- grid domain (2)
- History (3)
- history matching (6)
- information (7)
- informative output (2)
- input space (2)
- iteration (4)
- karst feature (2)
- karst reservoir (2)
- LoF (2)
- machine learning (7)
- matrix block (2)
- Modeling & Simulation (11)
- NPV (3)
- objective function (6)
- optimization (3)
- optimization problem (6)
- optimization process (3)
- PDP method (2)
- permeability (5)
- permeability curve (2)
- permeable thin layer (2)
- petroleum field development (2)
- porosity map (2)
- procedure (10)
- prod 10 (2)
- prod 11 (2)
- prod 5 (2)
- prod 9 (2)
- producer (4)
- production data (3)
- production strategy (4)
- realization (4)
- reduction (3)
- reference model (7)
- representation (2)
- representative model (4)
- Reservoir Characterization (8)
- Reservoir Management (2)
- reservoir model (4)
- reservoir property (2)
- reservoir simulation (22)
- resolution (2)
- risk curve (4)
- risk management (2)
- saturation (4)
- scaling method (4)
- Scenario (10)
- Schiozer (12)
- seismic data (3)
- selection (3)
- simulation model (6)
- simulation scenario (2)
- simulator (4)
- society of petroleum engineers (4)
- SPE Reservoir Evaluation (6)
- special connection (2)
- super-k layer (2)
- transmissibility (2)
- uncertainty reduction (4)
- Upstream Oil & Gas (25)

**Country**

**Industry**

**Oilfield Places**

**Technology**

Goto

Layer | Fill | Outline |
---|

Theme | Visible | Selectable | Appearance | Zoom Range (now: 0) |
---|

Fill | Stroke |
---|---|

Santos, Letícia Siqueira dos (UNICAMP) | Santos, Susana Margarida da Graça (UNICAMP) | Santos, Antonio Alberto de Souza dos (UNICAMP) | Schiozer, Denis José (UNICAMP) | Silva, Luis Otávio Mendes da (UNICAMP)

The Expected Value of Information (EVoI) is a criterion to analyze the feasibility of acquiring new information to deal with uncertainties and improve decisions at any stage of an oil field. Here, we evaluate the influence of the use of representative models (RM) on the EVoI estimation and on the decision to develop the petroleum field. These RM are used to represent a large set of models that honor production data (FM), considering uncertainties in reservoir, fluid and economic parameters, enabling the following processes: (1) optimize production strategies (specialized for each RM and robust to all RM), (2) risk analysis, (3) select the strategy to develop the field based on risk analysis, and (4) estimate the EVoI. We evaluated the influence of the number of RM on these processes, observing the impacts on the results of reducing computational costs. For the EVoI, we applied a Complete (EVoI_FM) and a Simplified (EVoI_RM) methodology, where EVoI_FM was evaluated with all models (FM) while EVoI_RM used different groups with different numbers of RM (GR1, GR2 and GR3, ranging from 9 to 150 models in each group). To assess the quality of the results, we used the complete estimate (EVoI_FM) as a reference. The study was conducted on UNISIM-I-D, a benchmark oil reservoir in the development phase, taking an appraisal well as a source of information to clarify a structural uncertainty. Using RM to optimize specialized production strategies proved useful, since optimizing strategies for all FM would require high computational costs. Moreover, the RM could be used to represent risk curves and select production strategies under uncertainty, but less precisely, affecting directly the results of the EVoI (which is the difference between the expected values of the two curves). The precision of EVoI_RM results varied according to the number and group of RM employed, also varying the best strategies selected for field development. The choice of using simplifications or not will depend on the accuracy required or available resources. Variations in EVoI_RM may be tolerable when compared to the time saved, being the decision maker free for choosing the best estimation method.

Artificial Intelligence, asset and portfolio management, Bayesian Inference, east block, EMV, equation of state, estimation, EVOI, field development optimization and planning, fluid modeling, information, machine learning, optimization problem, petroleum field development, probability, production strategy, representative model, reservoir simulation, Risk analysis, risk and uncertainty assessment, risk assessment, risk curve, risk management, RMO, RMT, robust production strategy, Scenario, Schiozer, selection, specialized strategy, Upstream Oil & Gas

Oilfield Places:

- South America > Brazil > Rio de Janeiro > South Atlantic Ocean > Campos Basin > Namorado Field (0.99)
- South America > Brazil > Campos Basin (0.99)

SPE Disciplines:

Technology:

- Information Technology > Artificial Intelligence > Representation & Reasoning > Optimization (0.69)
- 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)

Numerical reservoir simulation often requires upscaling of fine-scale detailed models and coarse-scale models are necessary to reduce computational time for dynamic evaluations. However, these simplifications may degenerate results due to loss of resolution of the small-scale phenomena, averaging of sub-grid heterogeneity and numerical dispersion, especially in oil fields where miscible gas is injected.

Most of the existing upscaling techniques focus on reproducing the results of a specific geological realization, in a deterministic approach. Nowadays, however, reservoir simulation studies commonly include uncertainty quantifications, which is performed by simulating multiple geological realizations. For that, the use of fine-scale models can be computationally prohibitive and this requires a proper procedure to upscale the coarse-scale simulation models in multiple realizations environment.

In this work, we propose and test an ensemble-level upscaling technique for compositional systems with miscible gas injection. The new approach considers the classical Koval factor, calculated for the fine-scale models, as a guide for selecting representative fine-scale models to train pseudo-functions for the coarse-models. Only a few fine models are simulated (about 1%), and the uncertainty quantification process with coarse-scale models can be significantly improved.

The proposed workflow is guided by ranking the fine-scale models in increasing order of their Koval Factor. We selected representative models and applied a two-step methodology to improve upscaled coarse-scale results for these models. We then propose a consistent procedure to expand the fitted pseudo-functions to all the coarse models, providing an effective ensemble-level upscaling.

The correlation between Koval factor and oil recovery is a useful guide to extrapolate the pseudo-functions obtained for each selected representative model, enabling better coarse-scale simulation results when multiple realizations are considered. This procedure can be applied for continuous miscible gas injection and can be adapted for WAG scheme.

This work was motivated by the lack of practical procedures to improve coarse-scale results at the ensemble-level. With our approach, we can better represent uncertainty quantification using coarse-scale models with reduced computational cost and requiring only a few fine-scale simulation runs.

application, Artificial Intelligence, coarse model, coefficient, correlation, fine-scale model, Koval factor, machine learning, miscible gas injection, Modeling & Simulation, oil recovery factor uncertainty curve, procedure, realization, representative model, reservoir simulation, scaling method, uncertainty curve, Upstream Oil & Gas, workflow

Technology:

Ferreira, Carla Janaina (Durham University and University of Campinas) | Vernon, Ian (Durham University) | Caiado, Camila (Durham University) | Formentin, Helena Nandi (Durham University and University of Campinas) | Avansi, Guilherme Daniel (University of Campinas) | Goldstein, Michael (Durham University) | Schiozer, Denis José (University of Campinas)

When performing classic uncertainty reduction according to dynamic data, a large number of reservoir simulations need to be evaluated at high computational cost. As an alternative, we construct Bayesian emulators that mimic the dominant behavior of the reservoir simulator, and which are several orders of magnitude faster to evaluate. We combine these emulators within an iterative procedure that involves substantial but appropriate dimensional reduction of the output space (which represents the reservoir physical behavior, such as production data), enabling a more effective and efficient uncertainty reduction on the input space (representing uncertain reservoir parameters) than traditional methods, and with a more comprehensive understanding of the associated uncertainties. This study uses the emulation-based Bayesian history-matching (BHM) uncertainty analysis for the uncertainty reduction of complex models, which is designed to address problems with a high number of both input and output parameters. We detail how to efficiently choose sets of outputs that are suitable for emulation and that are highly informative to reduce the input-parameter space and investigate different classes of outputs and objective functions. We use output emulators and implausibility analysis iteratively to perform uncertainty reduction in the input-parameter space, and we discuss the strengths and weaknesses of certain popular classes of objective functions in this context. We demonstrate our approach through an application to a benchmark synthetic model (built using public data from a Brazilian offshore field) in an early stage of development using 4 years of historical data and four producers. This study investigates traditional simulation outputs (e.g., production data) and also novel classes of outputs, such as misfit indices and summaries of outputs. We show that despite there being a large number (2,136) of possible outputs, only very few (16) were sufficient to represent the available information; these informative outputs were used using fast and efficient emulators at each iteration (or wave) of the history match to perform the uncertainty-reduction procedure successfully. Using this small set of outputs, we were able to substantially reduce the input space by removing 99.8% of the original volume. We found that a small set of physically meaningful individual production outputs were the most informative at early waves, which once emulated, resulted in the highest uncertainty reduction in the input-parameter space, while more complex but popular objective functions that combine several outputs were only modestly useful at later waves. The latter point is because objective functions such as misfit indices have complex surfaces that can lead to low-quality emulators and hence result in noninformative outputs. We present an iterative emulator-based Bayesian uncertainty-reduction process in which all possible input-parameter configurations that lead to statistically acceptable matches between the simulated and observed data are identified. This methodology presents four central characteristics: incorporation of a powerful dimension reduction on the output space, resulting in significantly increased efficiency; effective reduction of the input space; computational efficiency, and provision of a better understanding of the complex geometry of the input and output spaces.

Artificial Intelligence, discrepancy, emulate, emulation, emulator, evaluation, Goldstein, history matching, informative output, input space, input-parameter space, machine learning, objective function, reservoir model, reservoir simulation, Scenario, space cutout, uncertainty analysis, uncertainty reduction, Upstream Oil & Gas, vernon, water rate

Country:

- Europe (1.00)
- North America > United States (0.93)

Oilfield Places:

- Europe > Norway > Norwegian Sea > Halten Terrace > Block 6608/10 > Norne Field > Tofte Formation (0.99)
- Europe > Norway > Norwegian Sea > Halten Terrace > Block 6608/10 > Norne Field > Not Formation (0.99)
- Europe > Norway > Norwegian Sea > Halten Terrace > Block 6608/10 > Norne Field > Ile Formation (0.99)
- (3 more...)

SPE Disciplines:

Technology: Information Technology > Artificial Intelligence > Machine Learning > Statistical Learning (1.00)

Closed-loop field development (CLFD) is a feedback-based approach to optimize production strategy by utilizing new information iteratively during field development. Positive and negative results have been presented in previous work using CLFD, so this paper presents a bottom-up assessment of CLFD workflow to exhibit the impact of individual steps on the workflow, highlighting associated potential problems and, finally, proposing a methodology to tackle these problems.

Our CLFD workflow consists of updating static information, assimilating dynamic data to select an appropriate subset of models and, finally, selecting representative models to optimize the production strategy under uncertainties. We performed four activities to expand our understanding of CLFD by applying the workflow on UNISIM-I, a benchmark case study: (1) applying the aforementioned CLFD workflow, (2) applying the workflow after modifying objective function of optimization process, emphasizing the likelihood of success of optimized production strategy over the ensemble of simulation models, (3) applying the workflow after modifying the objective function, as well as the criteria for selecting the appropriate subset of models, post dynamic data assimilation, and (4) reanalyzing the results of activity 1 after using flexibility of drilling (FoD) - an approach proposed to partially imitate the real-time decision-making process to ensure that the heel of a well is not drilled in a non-reservoir zone, by utilizing the available information.

The application of the initial CLFD methodology (first activity) led to an increase in expected monetary value (EMV) based on simulation models, but we observed a decrease in EMV when we implemented this optimized strategy in the reference case UNISIM-I-R (a very refined model that emulates a "true field"). This negative result formed the basis for our motivation to perform the bottom-up analysis. During the second activity, our attempt to improve the optimization process using a new objective-function, led to a significant improvement in EMV for the reference case, compared to the first activity. Re-applying the CLFD workflow using this newly tested objective function, while using stricter criteria for selecting approved models (third activity), provided an optimized production strategy for the reference case. Last activity provided a deeper insight into the CLFD workflow. Application of FoD, during the fourth activity, revealed that the poorer results during the first activity can be segregated into two separate components: (a) relatively poorer CLFD workflow and, (b) ignoring FoD to ineptly drill the wells in non-reservoir zones.

The bottom-up approach helped us systematically improve the CLFD workflow. Implementation of the improved workflow ensured that the optimized production strategy not only improved EMV for simulation models, but also the reference case.

SPE Disciplines:

Technology:

Closed-loop field development (CLFD) is a feedback-based approach to optimize production strategy by utilizing new information iteratively during field development. Positive and negative results have been presented in previous work using CLFD, so this paper presents a bottom-up assessment of CLFD workflow to exhibit the impact of individual steps on the workflow, highlighting associated potential problems and, finally, proposing a methodology to tackle these problems.

Our CLFD workflow consists of updating static information, assimilating dynamic data to select an appropriate subset of models and, finally, selecting representative models to optimize the production strategy under uncertainties. We performed four activities to expand our understanding of CLFD by applying the workflow on UNISIM-I, a benchmark case study: (1) applying the aforementioned CLFD workflow, (2) applying the workflow after modifying objective function of optimization process, emphasizing the likelihood of success of optimized production strategy over the ensemble of simulation models, (3) applying the workflow after modifying the objective function, as well as the criteria for selecting the appropriate subset of models, post dynamic data assimilation, and (4) reanalyzing the results of activity 1 after using flexibility of drilling (FoD) - an approach proposed to partially imitate the real-time decision-making process to ensure that the heel of a well is not drilled in a non-reservoir zone, by utilizing the available information.

The application of the initial CLFD methodology (first activity) led to an increase in expected monetary value (EMV) based on simulation models, but we observed a decrease in EMV when we implemented this optimized strategy in the reference case UNISIM-I-R (a very refined model that emulates a "true field"). This negative result formed the basis for our motivation to perform the bottom-up analysis. During the second activity, our attempt to improve the optimization process using a new objective-function, led to a significant improvement in EMV for the reference case, compared to the first activity. Re-applying the CLFD workflow using this newly tested objective function, while using stricter criteria for selecting approved models (third activity), provided an optimized production strategy for the reference case. Last activity provided a deeper insight into the CLFD workflow. Application of FoD, during the fourth activity, revealed that the poorer results during the first activity can be segregated into two separate components: (a) relatively poorer CLFD workflow and, (b) ignoring FoD to ineptly drill the wells in non-reservoir zones.

The bottom-up approach helped us systematically improve the CLFD workflow. Implementation of the improved workflow ensured that the optimized production strategy not only improved EMV for simulation models, but also the reference case.

Artificial Intelligence, bottom-up analysis, clfd workflow, ensemble, equation, evolutionary algorithm, field development, icm, information, machine learning, morosov, negative result, objective function, optimization problem, optimization process, production data, reference case, reservoir simulation, rms, Scenario, Schiozer, simulation model, Upstream Oil & Gas, workflow

Country:

- South America (1.00)
- North America > United States > Texas (0.93)

Oilfield Places:

- South America > Brazil > Rio de Janeiro > South Atlantic Ocean > Campos Basin > Namorado Field (0.99)
- South America > Brazil > Campos Basin (0.99)

Technology:

The significant oil reserves related to karst reservoirs in a Brazilian presalt field add new frontiers to the development of upscaling procedures to reduce time for numerical simulations. This work aims to represent karst reservoirs in reservoir simulators using special connections between matrix medium and karst medium, each modeled in different grid domains of a single-porosity flow model. This representation intends to provide a good relationship between accuracy and simulation time.

The concept follows the embedded discrete-fracture model (EDFM) developed by Li and Lee (2008) and later extended by Moinfar et al. (2014); however, this work extends the approach for karst reservoirs [embedded discrete-karst model (EDKM)] by adding a representative volume through gridblocks to represent karst geometries and porosity. For the extension of the EDFM approach in a karst reservoir, we adapt the methodology to four stages: construction of a single-porosity model with two grid domains; geomodeling of karst and matrix properties, each for the corresponding grid domain; application of special connections through the conventional reservoir simulator to represent the transmissibility between the matrix and the karst medium; and calculation of transmissibility between the matrix and the karst medium.

For a proper verification, we applied the EDKM methodology in a carbonate reservoir with megakarst structures, which consists of nonwell-connected enlarged conduits and greater than 300 mm of aperture. The reference model was a refined grid with karst features explicitly combined with matrix facies, including coquinas interbedded with mudstones and shales. The gridblock of the reference model measures approximately 10 x 10 x 1 m. For the simulation model, the matrix-grid domain has a gridblock size of approximately 100 x 100 x 5 m. The karst-grid domain had the same block size as the refined grid. Flow in the individual karst-grid domain or matrix-grid domain is governed by Darcy’s equation, implicitly solved by a simulator. However, the transmissibility for the special connections between karst and matrix blocks is calculated as a function of open area to flow, matrix permeability, and block-center distance. The matrix properties were scaled up through conventional analytical methods. The results show that EDKM had a considerable performance regarding a dynamic matching response with the reference model, within a reduced simulation time, while maintaining a higher dynamic resolution in the karst-grid domain without using an unconstructed grid.

This work aims to contribute to the extension of the EDFM approach for karst reservoirs, which can be applied to commercial finite-difference reservoir simulators. This could be a solution to reduce simulation time without disregarding the explicit representation of karst features in structured grids.

Artificial Intelligence, edkm, flow in porous media, flow simulation, Fluid Dynamics, fracture, grid domain, karst feature, karst reservoir, matrix block, Modeling & Simulation, permeability, reference model, reservoir simulation, saturation, simulator, SPE Reservoir Evaluation, special connection, transmissibility, Upstream Oil & Gas

Country:

- South America (0.94)
- North America > United States (0.93)
- Asia (0.93)

Oilfield Places:

- South America > Brazil > Brazil > South Atlantic Ocean > Santos Basin (0.99)
- Asia > China > Xinjiang Uyghur Autonomous Region > Tarim Basin (0.99)

SPE Disciplines:

Technology:

History matching (HM) is a complex process that aims to increase the reliability of reservoir-simulation models. HM is an inverse problem with multiple solutions that call for probabilistic approaches. When observed data are integrated with sampling methods, uncertainty can be reduced by updating the uncertainty distribution of the reservoir properties. This work presents a practical methodology to systematically reduce uncertainties in a multiobjective assisted HM while dealing with multiple scenarios and assimilating well and 4D-seismic (4DS) data quantitatively. The frequency-distribution update goes through an iterative process. The distribution of the current iteration is combined with the histogram generated using the best-matched simulation scenarios from the current iteration to generate the updated distribution. To evaluate the matching quality, multiple local objective functions (LOFs) are independently evaluated, enabling the identification of LOFs that expose the need for reparameterization. This quantitative process was applied in two phases: Phase 1, in which only well data were used to constrain the scenarios, and Phase 2, when 4DS data were added. The methodology was successfully validated against a synthetic benchmark case of medium complexity, with the production-history data generated at a fine scale (reference model). Each iteration increased the number of matched scenarios, demonstrating good convergence. Most of the reservoir properties had uncertainty reduced gradually while avoiding the premature reduction of the uncertainty range (minimizing convergence to an incorrect solution). Local probabilistic perturbations were conducted on the petrophysical realizations in the regions around the wells that manifested LOFs, which hindered the match. The method efficiently achieved multiple matched simulation scenarios, with all (87) LOFs evaluated within the defined tolerance range. The 4DS data were included regionally with an acceptable increase in computation time. In both phases, the matched simulation scenarios presented production forecasts similar to the reference model. The quantitative assimilation of 4DS data generated scenarios that forecast production with less variability than did scenarios generated without 4DS data. This was expected for this study because the 4DS data provided do not present noise or artifacts.

Acceptance Range, Artificial Intelligence, deviation, frequency distribution, history matching, iteration, level 0, LoF, Modeling & Simulation, procedure, prod 10, prod 11, prod 5, prod 9, realization, reference model, reservoir property, reservoir simulation, Scenario, Schiozer, simulation scenario, SPE Reservoir Evaluation, Upstream Oil & Gas

Country:

- North America > United States (0.67)
- South America > Brazil (0.46)

SPE Disciplines: Reservoir Description and Dynamics > Reservoir Simulation > History matching (1.00)

Technology:

The significant oil reserves related to karst reservoirs in a Brazilian presalt field add new frontiers to the development of upscaling procedures to reduce time for numerical simulations. This work aims to represent karst reservoirs in reservoir simulators using special connections between matrix medium and karst medium, each modeled in different grid domains of a single-porosity flow model. This representation intends to provide a good relationship between accuracy and simulation time.

The concept follows the embedded discrete-fracture model (EDFM) developed by Li and Lee (2008) and later extended by Moinfar et al. (2014); however, this work extends the approach for karst reservoirs [embedded discrete-karst model (EDKM)] by adding a representative volume through gridblocks to represent karst geometries and porosity. For the extension of the EDFM approach in a karst reservoir, we adapt the methodology to four stages: construction of a single-porosity model with two grid domains; geomodeling of karst and matrix properties, each for the corresponding grid domain; application of special connections through the conventional reservoir simulator to represent the transmissibility between the matrix and the karst medium; and calculation of transmissibility between the matrix and the karst medium.

For a proper verification, we applied the EDKM methodology in a carbonate reservoir with megakarst structures, which consists of nonwell-connected enlarged conduits and greater than 300 mm of aperture. The reference model was a refined grid with karst features explicitly combined with matrix facies, including coquinas interbedded with mudstones and shales. The gridblock of the reference model measures approximately 10 x 10 x 1 m. For the simulation model, the matrix-grid domain has a gridblock size of approximately 100 x 100 x 5 m. The karst-grid domain had the same block size as the refined grid. Flow in the individual karst-grid domain or matrix-grid domain is governed by Darcy’s equation, implicitly solved by a simulator. However, the transmissibility for the special connections between karst and matrix blocks is calculated as a function of open area to flow, matrix permeability, and block-center distance. The matrix properties were scaled up through conventional analytical methods. The results show that EDKM had a considerable performance regarding a dynamic matching response with the reference model, within a reduced simulation time, while maintaining a higher dynamic resolution in the karst-grid domain without using an unconstructed grid.

This work aims to contribute to the extension of the EDFM approach for karst reservoirs, which can be applied to commercial finite-difference reservoir simulators. This could be a solution to reduce simulation time without disregarding the explicit representation of karst features in structured grids.

Artificial Intelligence, edkm, Engineering, flow in porous media, flow simulation, Fluid Dynamics, fracture, grid domain, karst feature, karst reservoir, matrix block, Modeling & Simulation, permeability, reference model, reservoir simulation, simulator, SPE Reservoir Evaluation, special connection, transmissibility, Upstream Oil & Gas

Country:

- South America (0.94)
- North America > United States (0.93)
- Asia (0.93)

Oilfield Places:

- South America > Brazil > Brazil > South Atlantic Ocean > Santos Basin (0.99)
- Asia > China > Xinjiang Uyghur Autonomous Region > Tarim Basin (0.99)

SPE Disciplines:

Technology:

Providing an overview of an ensemble of oil reservoir models could help users compare and analyze their characteristics. Approaches that show a single model at a time may hamper analysts’ understanding of the whole model set. In this paper, we propose two visualization approaches that show multiple reservoir models, simultaneously and on a single screen, with the goal of helping users to compare models and improve their understanding of ensemble characteristics. First, we calculate 2D models from the ensemble's 3D models. We then create two visualizations that represent ensembles of these 2D models. The

Artificial Intelligence, benchmark case, dataset, distance matrix, ensemble, feature vector, heatmap, layout, machine learning, oil saturation, pixelization, pixelization approach, Reservoir Characterization, reservoir model, reservoir simulation, saturation, similarity, small multiple approach, structural geology, Upstream Oil & Gas, visual mapping, Visualization

SPE Disciplines:

Technology: Information Technology > Artificial Intelligence > Machine Learning > Statistical Learning > Clustering (0.47)

Ferreira, Carla Janaina (Durham University / University of Campinas) | Vernon, Ian (Durham University) | Caiado, Camila (Durham University) | Formentin, Helena Nandi (Durham University / University of Campinas) | Avansi, Guilherme Daniel (University of Campinas) | Goldstein, Michael (Durham University) | Schiozer, Denis José (University of Campinas)

When performing classic uncertainty reduction based on dynamic data, a large number of reservoir simulations need to be evaluated at high computational cost. As an alternative, we construct Bayesian emulators that mimic the dominant behaviour of the reservoir simulator, and which are several orders of magnitude faster to evaluate. We combine these emulators within an iterative procedure that involves substantial but appropriate dimensional reduction of the output space, enabling a more effective and efficient uncertainty reduction on the input space than traditional methods, and with a more comprehensive understanding of the associated uncertainties. This study uses a Bayesian statistical approach for uncertainty reduction of complex models which is designed to address problems with high number of both input and output parameters. We detail how to efficiently choose sets of outputs that are suitable for emulation and that are highly informative to reduce the input parameter space and investigate different classes of outputs and objective functions. We use output emulators and implausibility analysis iteratively to perform input space reduction, and we discuss the strengths and weaknesses of certain popular classes of objective function in this context. We demonstrate our approach via an application to a benchmark synthetic model (built using public data from a Brazilian offshore field) in an early stage of development using four years of historical data and four producers. This study investigates traditional simulation outputs (e.g. production data) and also novel classes of outputs, such as misfit indexes and summaries of outputs. We show that despite there being a large number (2,136) of possible outputs, only a very small number (16) was sufficient to represent the available information; these informative outputs were utilized using fast and efficient emulators at each iteration (or wave) of the history match to perform the uncertainty reduction procedure successfully. Using this small set of outputs, we were able to substantially reduce the input space by removing 99.8% of the original volume. We found that a small set of physically meaningful individual production outputs were the most informative at early waves, which once emulated, resulted in the highest space reduction, while more complex but popular objective functions that combine several outputs were only modestly useful at later waves. The latter point is due to objective functions such as misfit indices having complex surfaces that can lead to low-quality emulators and hence result in non-informative outputs. We present an iterative emulator-based Bayesian uncertainty reduction process in which all possible input parameter configurations that lead to statistically acceptable matches between the simulated and observed data are identified. This methodology presents four central characteristics: (1) incorporation of a powerful dimension reduction on the output space, resulting in significantly increased efficiency, (2) effective reduction of the input space, (3) computational efficiency, and (4) provision of a better understanding of the complex geometry of the input and output spaces.

Artificial Intelligence, emulation, emulator, evaluation, Goldstein, history matching, informative output, input space, machine learning, model output, objective function, reduction, reservoir model, reservoir simulation, Scenario, selection, space cut, uncertainty analysis, uncertainty reduction, Upstream Oil & Gas, vernon, water rate

Oilfield Places:

- North America > United States > Texas > East Gulf Coast Tertiary Basin > Nome Field (0.99)
- Europe > Norway > Norwegian Sea > Halten Terrace > Block 6608/10 > Norne Field > Tofte Formation (0.99)
- Europe > Norway > Norwegian Sea > Halten Terrace > Block 6608/10 > Norne Field > Not Formation (0.99)
- (4 more...)

SPE Disciplines: Reservoir Description and Dynamics > Reservoir Simulation > History matching (1.00)

Technology:

Thank you!