Grane Case Study: From Data and Model Ensembles to Actionable Insights Using a Cloud-Based Innovation Toolkit

Sequential assimilation of 4D seismic data for reservoir description using the ensemble Kalman filter

Assisted history matching methods are beginning to offer the possibility to use 4D seismic data in quantitative ways for reservoir characterization. We use the waveform data without any explicit inversion or interpretation step directly in an ensemble-based assisted history matching scheme with a 3D synthetic reservoir model. For this purpose a seismic simulator was developed to simulate the full waveform response to reservoir changes due to production. We discuss the iterative ensemble methodology, including additional steps that were found to be helpful in obtaining an ensemble that is properly conditioned to the initial base survey and maintains sufficient and reliable spread during the model updating process. The strength of combining the information contained in the waveform data with the prior information contained in the initial ensemble is visualized by means of ensemble covariances. The results of two history matching experiments show that more detailed features of the true permeability field can be reconstructed when production data are combined with time-lapse seismic data.

Time-lapse seismic changes can be compared with pressure and/or saturation changes derived from a flow simulator by using petro-elastic model. As the number of seismic data is huge and it has a 3D character, it is expected to be a great help to condition the reservoir model. An (automated) history matching loop comprises the following four stages: 1) parameterization, 2) objective function definition, 3) optimization and 4) loop-stopping criterion. Focusing the objective function, the brief explanation of some element technology is done. With the introduction of time-lapse seismic as an additional constraint, history matching is performed to construct a more reliable reservoir model, which can be used to predict a future oil production.The impact of data integration on production forecasting uncertainty is analyzed. The synthetic model used representative for a gravity fault system, such as East flank of Statfjord field. By changing the constraining data, i. e. time-lapse seismic and production data, its impact on the probability density function (pdf) of the future cumulative oil production (FCOP) can be quantified. The variance of FCOP-pdf is reduced and the maximum pdf value is increased by adding extra information to the reservoir model. For this specific synthetic model, the seismic data gives a reservoir model with a higher probability than the production data. The most probable models are obtained when conditioning the reservoir model with both types of data.

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

The amount of information available for field development planning is limited, forcing the Production Strategy to be designed under great amounts of uncertainty. During its implementation, new information allows the adaptation of the strategy for economic gain. This work reproduces the field development process under geological uncertainty in case study UNISIM-I-D (benchmark case based on Namorado Field in Brazil). The main objectives are to evaluate the process and observe the evolution of risk curves, all in a controlled environment with real field features. The methodology generates new geostatistical images based on new well logs, assimilates production data using an ensemble-based method and re-optimizes the production strategy using a hybrid algorithm. The field development is carried out by repeatedly applying this framework under human supervision. Each step is customized using algorithms to simplify the implementation and to reduce computational effort, making this methodology more appealing for practical use. New data are collected from a high-resolution reference model that does not belong to the ensemble of models. The process starts with a production strategy, previously optimized under the uncertainties of the case study, which yields the real economic outcome within the original uncertain range. Results show high quality history matching that excessively reduced the risk range and the variability of the updated model sets. Optimizations on the production strategy, based on the updated ensembles, consistently increased the expected monetary value of the project without guaranteeing an increment in the real NPV. Applying the methodology repeatedly throughout the field development increased the EMV by 29% (from 1.532 to 1.975 billion USD) while the real NPV decreased 2% (from 1.346 to 1.319 billion USD), falling out of the expected range and revealing that the model sets did not fully represent the real field. The lack of good representation is aggravated by heterogeneities inherent to the unknown reservoir, which are difficult to identify using only well logs and production data. The results from the application of a closed-loop reservoir development process in a controlled environment warn against similar hidden mechanisms happening on a real field development under similar circumstances. They reveal intrinsic pitfalls in reservoir modeling that may contribute to production forecast problems and call for a reflection on how reservoir uncertainty assessment is performed. We prove that large sets of models do not guarantee coverage of geologic uncertainties as the field development process naturally changes the risk curves.

Field X located in offshore South East Asia, is a deepwater, turbidite natural gas greenfield currently being developed using a subsea tieback production system. It is part of a group of fields anticipated to be developed together as a cluster. Due to the nature of this development, several key challenges were foreseen: i) subsurface uncertainty ii) production network impact on system deliverability and flow assurance iii) efficient use of high frequency data in managing production. The objective of this study was to demonstrate a flexible and robust methodology to address these challenges by integrating multiple realizations of the reservoir model with surface network models and showing how this could be link to "live" production data in the future. This paper describes the development and deployment of the solutions to overcome those challenges. Furthermore, the paper describes the results and key observations for further recommendation in moving forward to field digitalization. The process started with a quality check of the base case dynamic reservoir model to improve performance and enable multiple realization runs in a reasonable timeframe. This was followed by sensitivity and uncertainty analysis to obtain 10 realizations of the subsurface model which were integrated with the steady-state surface network model. Optimization under uncertainty was then performed on the integrated model to evaluate three illustrative development scenarios. To demonstrate extensibility, two additional candidate reservoirs for future development were also tied in to the system and modelled as a single integrated asset model to meet the anticipated gas delivery targets. Next, the subsurface model was integrated with a multiphase transient network model to show how it can be used to evaluate the risk of hydrate formation along the pipeline during planned production start-up. As a final step, in-built application programming interface (API) in the integration software was used to perform automation, enabling the integrated model to be activated and run automatically while being updated with sample "live" production data. At the conclusion of the study, the reservoir simulation performance was improved, reducing runtime by a factor of four without significant change in base case results. The results of the coupled reservoir to steady-state network simulation and optimization showed that the network could constrain reservoir deliverability by up to 4% in all realizations due to back pressure, and the most optimum development scenario was to delay first gas production and operate with shorter duration at high separator pressure. With the additional reservoirs in the integrated model, the production plateau could be extended up to 15 years beyond the base case without exceeding the specified water handling limit. For hydrates risk analysis, the differences between hydrate formation and fluid temperature indicated there was a potential risk of hydrate formation, which could be reduced by increasing inhibitor concentration. Finally, the automation process was successfully tested with sample data to generate updated production forecast profiles as the "new" production data was fed into the database, enabling immediate analysis. This study demonstrated an approach to improve forecasting and scenario evaluation by using multiple realizations of the reservoir model coupled to a surface network. The study also demonstrated that this integrated model can be carried forward to improve management of the field in the future when combined with "live" data and automation logic to create a foundation for a digital field deployment.

As carbon dioxide (CO{sub 2}) is sequestered from the bottom of a brine reservoir and allowed to migrate upward, the effects of the relative permeability hysteresis due to capillary trapping and buoyancy driven migration tend to make the reservoir patchy saturated with different fluid phases over time. Seismically, such a patchy saturated reservoir induces an effective anisotropic behavior whose properties are primarily dictated by the nature of the saturation of different fluid phases in the pores and the elastic properties of the rock matrix. By combining reservoir flow simulation and modeling with seismic modeling, it is possible to derive these effective anisotropic properties, which, in turn, could be related to the saturation of CO{sub 2} within the reservoir volume any time during the post-injection scenario. Therefore, if time-lapse seismic data are available and could be inverted for the effective anisotropic properties of the reservoir, they, in combination with reservoir simulation could potentially predict the CO{sub 2} saturation directly from the time-lapse seismic data. It is therefore concluded that the time-lapse seismic data could be used to monitor the carbon sequestrated saline reservoirs. But for its successful implementation, seismic modeling and inversion methods must be integrated with the reservoir simulations. In addition, because CO{sub 2} sequestration induces an effective anisotropy in the sequestered reservoir and anisotropy is best detected using multicomponent seismic data compared to single component (P-wave) data, acquisition, processing, and analysis is multicomponent seismic data is recommended for these time-lapse studies. Finally, a successful implementation of using time-lapse seismic data for monitoring the carbon sequestrated saline reservoirs will require development of a robust methodology for inverting multicomponent seismic data for subsurface anisotropic properties.

The aim of this work is to present the effectiveness of a fully integrated approach for ensemble-based history matching on a complex real field application. We show that the predictive ability of the ensemble of models is greatly enhanced through an integrated workflow promoting multidisciplinary collaboration between all subsurface disciplines. Consistent integration of geological and engineering understanding within the dynamic data conditioning phase of a reservoir study is a challenging task. The ensemble-based approach offers an efficient solution to this challenge, especially when it is tied in with an appropriate reservoir modeling approach, suitable parameterization and repeatable workflows. All these are key factors to ensure geological consistency while improving the match quality and reliability of the generated reservoir models. One key feature of the ensemble-based method is that it overcomes the typical limitation of the traditional approaches where the number of uncertainty parameters often has to be reduced resulting from practical or algorithm constraints. This is especially important on complex reservoirs, where the subsurface uncertainties cannot be represented in a handful of scalar multipliers while honoring the static and dynamic data measurements. We organize the proposed methodology in four main steps: firstly, an initial ensemble of models able to capture the model uncertainties in all parts of the modeling processes are generated. Next, the match level for different observed data specifying the likelihood error is assigned. Then the uncertainty parameters that should be modified in the history matching process are identified. Finally, we perform a computational intensive step on large-scale computing facilities, where a state-of-art iterative ensemble algorithm, tightly connected with standard geo-modeling tools, is used to consistently update the full set of models. The result is an ensemble of reservoir models where we can quantify both the uncertainty in the production forecast and in the reservoir model parameters. The methodology has been applied on a challenging History Matching problem for a turbiditic channel complex reservoir. We demonstrate the efficiency of the method, by generating an ensemble of reservoir models that offer geologically consistent explanations of the currently measured static and dynamic data. The model forecasts reliability is confirmed through a validation process using dynamic data that have been left out of the history matching process. More importantly, however, is that the proposed workflow, and resulting models provides a good starting point for a more efficient update of the reservoir models as new static and dynamic data become available. With the proposed approach is possible to move from sporadic updates of reservoir models towards frequent model updates when the new data arrive. Utilizing all available data in a consistent manner - when the data are collected - is the key to making more robust reservoir management decisions, especially for complex reservoirs.

Time-lapse seismic data has been widely used for the detailed reservoir characterization, and data assimilation algorithms are commonly used for petroleum reservoir history matching of production data. However, hardly any seismic data has been integrated into the reservoir inverse modeling workflow, due to the large data size, finer gridding, and especially the scarcity in time of the seismic data. Popular ensemble-based reservoir inverse modeling methods such as ensemble Kalman filter (EnKF) also face the problems of high computational cost due to the storage of the intermediate variables, restarting the reservoir simulating process and the inconsistency of the full-step and step-wise simulations. The intrinsic sequential data assimilations characteristics of the EnKF set obstacles to assimilate 4D seismic data. The alternative method can be the ensemble smoother (ES), which is an ensemble based method for data assimilation. The ES is based on a Bayesian updating scheme of the reservoir model to match the production history and improve the production forecast. To improve the algorithm convergence, a multiple data assimilation (MDA) method was proposed by Emerick and Reynolds (2012a). The algorithm we used is called ensemble smoother with multiple data assimilation (ESMDA), we modified the ESMDA to integrate the geophysical data, and created the new workflow to history match both the production and geophysical data. The available production data is well measurements, including oil production rate, well water cut and bottom-hole pressure, as well as time-lapse geophysical data, including P-wave impedance. In this paper, we first formulated the mathematical descriptions of ESMDA, we then down-scaled the seismic data, and modified the data assimilation algorithms, illustrated the history matching workflow for both production and geophysical data, finally, we showed how it works with a case study on a water flooding operation in a synthetic reservoir. In comparison, we also showed the history matching only on the production data, which yields an inferior results than the one matches both production and seismic data.

Summary Inversion of synthetic seismic data for a viscoelastic reservoir model produces estimates of changes in effective pressure (Pe) and water saturation (Sw). Application is to analysis of time-lapse seismic data for a gas sandstone reservoir. To generate synthetic time-lapse viscoelastic seismic data, reservoir properties [porosity (�-), clay content, permeability (k), Pe and Sw] are used to define the corresponding viscoelastic seismic properties [compressional and shear velocities (Vp and Vs), density (�! ) and compressional and shear quality factors (Qp and Qs)] using petro-seismic modeling. It is assumed that porosity, clay content and permeability do not change during production and are known. Full wavefield inversion is performed for different production scenarios, on synthetic viscoelastic time-lapse seismic data, using a grid search refined by a conjugate gradient solution. Estimation of unique values of Pe and Sw from two independent data (such as Vp/Vs and Qs/Qp) is feasible, if the other reservoir properties are known. Noisy data produces a range of solution values. The composite effect of ®Pe and ®Sw is approximately equal to the sum of the two separate effects. Seismic time-lapse changes are less sensitive to the pressure change than to changes in saturation. For the reservoir properties (�- , clay content, k), and the dominant seismic frequency used in the example, �! and Qs/Qp are most sensitive to Sw, Vp is most sensitive to Pe, and Vs, Qp and Qs and Vp/Vs are sensitive to both Sw and Pe.

In general only a single 3D seismic data set is used for the characterization of the static properties of a reservoir, even though these static properties could be integrated with dynamic properties. But sometimes, more than one 3D seismic data volume has been acquired over the same field after the production starts. We propose to use these additional data not only to understand the dynamic properties of the reservoir using a time-lapse seismic analysis as presented last year, but to improve the accuracy of the initial reservoir description and study its sensitivity to the input data during the optimization process. Using two 3D seismic data sets acquired over a field in the Gulf of Mexico, this paper demonstrates a method to improve the porosity map obtained from the geostatistical characterization of the base 3D seismic data set by an analysis of the second 3D data volume. It is not possible using just the base data to differentiate factors such as fluid content and pressure from the geostatistical porosity characterization. But by using the flow information from the wells and the second 3D seismic data which was acquired after the production started, it is possible to decouple the porosity, fluid contents and even pressure. Consequently, it provides a more representative porosity characterization compared to using a single 3D seismic data set. The relative permeability curves are optimized by comparing actual time-lapse seismic and production data to synthetic ones generated with a Gassmann rock physics model. P. 527

Assimilating time-lapse seismic data in the presence of significant spatially correlated model errors

Numerical simulations of depressurization-induced gas production from gas hydrate reservoirs at the Walker Ridge 313 site, northern Gulf of Mexico

Effective reservoir characterization approaches aim to integrate all available and reliable sources of data to mitigate the effects of uncertainty of reservoir parameters and have a robust reservoir model(s). Reservoir permeability is one of the most critical, and usually, uncertain parameters in reservoir simulation models. The permeability model has a direct impact on production forecasts and consequently the economics of decisions made based on reservoir models. Having a calibrated permeability model is a challenging task. In most cases, the permeability model undergoes several modifications to achieve a history-matched reservoir model that honors the historical pressure and production data. The objective of this work is to show a workflow that integrates two different types of data sources to help calibrate the permeability of reservoir models. The proposed approach relies on adjusting the permeability of reservoir models by integrating seismic amplitude values and data from wells’ pressure transient analyses. The seismic amplitude data in the vicinity of a candidate well is correlated with permeability obtained from recorded pressure transient data in a systematic way. The generated correlation is then applied to the entire reservoir model in order to calibrate the permeability value of each grid block in the model. The application of the proposed approach on actual field data is also presented. As a result, the seismic and pressure transient data both can make valuable contributions towards calibrating geo-cellular 3D reservoir models at their respective and different scales.

In current practice history matching is often performed on upscaled reservoir models without updating the original geological model. Taking into account new production data in an existing model is often problematic because of the lack of appropriate techniques. This problem becomes crucial when using geostatistical models. In such a case, the integration of new static and dynamic constraints requires the modification of an initial realization without destroying the initial history match and the overall geostatistical properties. This paper presents a new methodology based on advanced history matching techniques for updating dynamically 3D stochastic reservoir models to account for new production data. The proposed approach is based on the gradual deformation method. It allows local or global smooth transformations of the model while conserving the overall statistical characteristics. History matching is performed by coupling this method with an optimization process which integrates geostatistical modeling, upscaling, and fluid flow simulation in the same loop. This approach provides great flexibility in history matching various types of production data. In addition, inversion parameters can be selected throughout the entire modeling loop, from the facies spatial distribution, to petrochemical parameters and fluid flow parameters. Finally, local transformations may be performed to account for new wells or to update the reservoir model with new dynamic information. A successful application to a 3D stochastic reservoir model inspired from a real field is presented. The first step was to perform a history match using the production history available from an initial group of production wells. Then, new dynamic information is acquired from infill wells and over one year of additional production data. The heterogeneity distribution is locally modified by gradual deformation to update the model with this new information. The benefit of the proposed approach for improving the model characterization and reducing uncertainty on production forecasts is demonstrated.