A Robust Downscaling Method for Integration of Static and Dynamic Models

Integration of dynamic data typically requires the solution of an inverse problem that can be computationally intensive and practically infeasible for fine scale reservoir models. In this paper we present a new methodology to directly update fine scale geostatistically-based reservoir models by combining gradual deformation parameterization for the fine scale geostatistical model and an upscaling technique for the coarse scale flow simulation model. The proposed methodology includes: Perturbation of the fine scale geostatistical model using the gradual deformation parameterization. Gradual deformation ensures the preservation of the overall geostatistical properties of the fine model. Generation of the coarse scale flow simulation model by upscaling the fine scale geostatistical model. Sensitivity computation of the flow simulation results with respect to the fine scale parameterization. This sensitivity computation is analytical and takes into account the upscaling process. Direct updating of the fine scale geostatistical model using classical optimization process. Direct updating ensures consistency between the fine and coarse scale models. The accuracy of the proposed methodology was improved by calibrating the flow simulation model. The objective of this calibration is to reduce the error introduced by the upscaling step during the flow simulation. We applied successfully our methodology for fine scale reservoir description by integrating permanent down-hole gauge measurements directly into a three-dimensional geostatistical model containing about two million grid blocks. This test is designed to highlight several key issues of the proposed methodology: Efficiency of the upscaling step coupled with gradient-based optimization to speed up the history matching process. Usefulness of the calibration step for a correct integration of upscaling techniques in history matching. Capability of the methodology for maintaining consistency and coherency between fine scale and coarse scale models. Improvement of the reservoir characterization by integrating dynamic data at the fine geostatistical scale. We conclude that the proposed methodology can be used effectively and efficiently for reservoir characterization purposes.

The association between invasive and native species varies across spatial scales and is affected by phylogenetic relatedness, but these issues have rarely been addressed in aquatic ecosystems. In this study, we used a non-native, highly invasive species of Poaceae (tropical signalgrass) to test the hypotheses that (i) tropical signalgrass success correlates negatively with success of most native species of macrophytes at fine spatial scales, but its success correlates positively or at random with natives at coarse spatial scales, and that (ii) tropical signalgrass is less associated with native species belonging to the family Poaceae than with species belonging to other families (Darwin’s naturalization hypothesis). We used a dataset obtained at fine (0.25 m2) and coarse (ca. 1,000 m2) scales. The presence/absence of all species was recorded at both scales, and their biomass was also measured at the fine scale. We tested the association between tropical signalgrass biomass and individual native species with logistic regressions at the fine scale, and using the T-score index between tropical signalgrass and each native species at both scales. The likelihood of the occurrence of six species (submersed and free-floating) was negatively affected by tropical signalgrass biomass at the fine scale. T-scores showed that three species were less associated with tropical signalgrass than expected by chance, but 22 species co-occurred more than expected by chance at the coarse scale. Associations between species of Poaceae and tropical signalgrass were null at the fine scale, but were positive or null at the coarse scale. In addition to showing that spatial scale affects the patterns of association among the non-native and individual native species, our results indicate that phylogeny did not explain associations between the invasive and native macrophytes, at both scales.

An efficient space-angle subgrid scale discretisation of the neutron transport equation

The field is located in the Persian Gulf and has been producing for the last 30 years with a strong natural aquifer support. The clastic reservoir exhibits highly heterogeneous permeability combined with shale streaks and therefore presents complex flow behavior. This paper describes the iterative seismic to simulation workflow followed to create a fine scale reservoir static and dynamic simulation model consistent with all available engineering, geologic and geophysical data. The process involved integration of static and dynamic modelling workflows. History matching the production data indicates locations with incorrect information in the static model, which can be corrected and re-exported for the dynamic model in very short time. The integration of static and dynamic modelling is seen as essential for the further commercial development of the field. A comprehensive integrated study was conducted starting from petrophysical log evaluation and resulting in fine scale reservoir models on a geo-cellular grid. To reduce the uncertainty, a model was created using a geostatistical inversion technique which honoured both geologic and seismic information. The use of high resolution geostatistical inversion provided good and reliable estimates of porosity and lithology away from the wells. The porosity and lithology models were further tested on history matching 92 producing wells for 30 years. The quick match resulted in less uncertainty of porosity and higher confidence in the prediction models. The history matched model is predicted further to define different potential development scenarios. It is now 3 years since all the data used in the study was acquired and the current field production matches with the model prediction. The models created using geostatistical inversion proved to be robust and predictive for the field development.

This paper discusses an integrated reservoir study utilizing structured and novel machine learning and data analytics approach for history matching a giant mature multilayered oil field in Mahakam Delta of Indonesia. The unique reservoir modeling challenges and novelty of the data science methods will be presented, along with preliminary results and lessons learnt. One of the most important elements in reservoir characterization and history matching process is integration between static and dynamic modeling. With numerous layers as study perimeter, a large number of uncertain parameters is unavoidable, from dynamic uncertainties such as hydrocarbon contacts and communication between regions to static properties like porosity, water saturation, etc. These in turn will create hundreds of possible scenarios during History Matching. Using python scripting embedded in the reservoir simulator and the agile reservoir modeling (ARM) approach, these uncertainties can be handled quickly, and each ensemble can be analyzed easily with Data Analytics and Machine Learning based proxy approaches. The case study presented here is a giant mature oilfield with more than 50 zones and 100 contact regions. With more than 45 years of production and injection history, the conventional reservoir modeling where each zone is modeled individually and independently, assuming no communication between the regions, has been deemed as taking too much time and effort. The integrated approach bypasses this challenge by allowing simultaneous reservoir modeling as well as quick sensitivity analysis and history matching quality checking. Uncertainties were managed early on; for instance, the porosity model was generated through available algorithms and hydrocarbon contacts with Latin-hypercube sampling method. Preliminary results showed that overall time required to perform the modeling has been reduced significantly, while also establishing communication between the regions. Analytical aquifer modeling and communication between the regions were observed as the most sensitive parameters especially when matching the pressure behavior. Moreover, embedded python script and Data Analytics Dashboard have made it possible to perform fast and systematic analysis, thus more effort and time can be allocated to plan the way forward. The Machine Learning results will be further finalized at the next gate review, considering the project was initially proposed into several gates. Static modeling using Machine Learning, coupled with dynamic modeling workflow and data analytics, has created a complete loop of reservoir study and characterization. All of these are conducted in a structured cloud-based platform, ensuring time-efficient process and repeatability while at the same time enabling hybrid approach by combining conventional method and advanced data driven approach

This work shows the realization of a 3D static and dynamic model with feasibility analysis and conceptual planning in the field of "La Itala" in Los Perales. Incorporate in one unique project feasibility and conceptual model is the principal benefits of this method because until now we haven't this kind of project in this all field. To perform this project, it was used some basic data of the wells, like SP and resistivity and a petrophysical model. To do the feasibility analysis we use wells basics curves envelops. Combining with core analysis a static facies model was generated. Using averages of rock and fluid parameters along with the history of production and injection, a dynamic model was initialized. This model permits to do a "History Match" at field level. This allowed visualizing the evolution in time of displacement of fluids, product of the water injection. The conclusions of this model define continuity with the conceptual model using a complete petrophysical study (VCL-Phie and Sw). Combination of the results of SP and short resistivity envelopes yielded to a first approximation of a Vclay. From there a binary log was generated. Settle the same curve with a vertical proportion curve; reference levels used to separate the model in zones were defined. The study of cores fostered a relationship between facies and resistivity. To make a reliable 3D structural model the control was made by some surfaces created on well tops by correlation. In this static model were charged all wells data available (perforations, facilities, production and injection). Along with rock properties and fluid average, the model was initialized. At this stage of visualization, a quick historical setting, at field level, injection and production served to understand the behavior of fluids in the reservoir. This understanding in the changes of water saturation turns to be a very important input for the next phase of conceptualization. Having a static and dynamic visualization model purchase in both ways the conceptualization phase, whether or not pass to the next stage. Everything done in the first stage is the starting point of the next (Front End Loading - FEL). The FEL methodology is deeply rooted in the DNA of YPF. In this case, with software who allowed enhancing this work process, creating a unique project where all available parameters are incorporated, with the possibility of being used in the different phases of the study. At the same time, each analysis on the project, adds value to the next step. This work allowed to reservoir engineer of this field to improve the oil recuperation factor.

The Australia Pacific LNG Project is scheduled to come online in 2015 and consists of the extensive development of substantial coal seam gas resources across the Surat Basin. A variety of unique and complex challenges are presented when undertaking the geological and dynamic modelling and performance prediction of the laterally discontinuous and thin coal seams within the Walloons package. These challenges require a targeted approach and evolving solutions.Specific challenges that have been assessed and addressed in the characterisation and modelling workflow of the complex Walloons coal seams include:– Complex coal packages exhibiting thin coal seams with high degrees of lateral and vertical heterogeneity – Appropriate prediction of uncertainty – Coal body distribution and connectivity – Application of learning from fine scale sector modelling to larger regional coarse scale models – Significant reservoir dynamics across large distances within Surat acreage – Production performance variation over time resulting from stress impacting the matrix – Vast scale of the Surat Walloons development combined with fast-cycle decision making – Adopting appropriate and unique workflows for regions of different production maturityThe developed workflow addresses coal seam gas modelling challenges both within the history matching phase and in the subsequent probabilistic forecasting. The workflow identifies uncertain reservoir properties and their expected maximum ranges based on available data and previous studies. An uncertainty analysis is conducted with statistical approaches including Tornado Chart and Latin Hypercube (LHC) to identify the most influential reservoir parameters and fine-tune their ranges in order to optimize the probabilistic history matching process. Subsequently, assisted history matching and optimisation techniques follow a stochastic algorithm of experiments to reduce the mismatch and create convergence with the production history. A number of representative, non-unique history matched models are identified which satisfy matching criteria and capture the uncertainty of the subsurface.The selected acceptable history matched models, together with forecasting parameters and their ranges, including operational variables, are used in forecasting. Random sampling of uncertainties by LHC is used with assisted history matched cases resulting in thousands of forecasts based on which P10-P50-P90 probabilistic forecasts are selected.This paper presents solutions to address uncertainty, assisted history matching and performance prediction of the Walloons coal measures for rigorous forecasting, and incorporation of learnings from localized fine scale models into larger coarse scale modelling for time efficient, regional performance prediction.

FGIIP (Field Gas Initially in Place) is one of the most essential elements in building dependable static and Integrated Asset Model (IAM). A good estimation of FGIIP will improve history matching and generate reliable forecast. The mature gas field producing under depletion mode is an ideal example where P/Z technique can fit well to re-estimate the FGIIP. Even more, the estimation is also important to narrow down FGIIP uncertainties that initially existed in static model. Reliable FGIIP estimation is achieved by performing multiple P/Z calculations. The process involves dividing reservoir into key areas and each area is represented by individual P/Z prior to summing-up all P/Z to get the total FGIIP. Several scenarios are developed by defining key areas based on permeability variation, areal distribution and PVT behavior. The best FGIIP estimation is then fed back into the static model to generate numerous realizations considering the static uncertainties to produce the same FGIIP. Static models with realistic distribution of properties and good history match are used in the IAM model to generate forecast. The giant onshore gas field is highly heterogeneous having permeability, lateral composition variation and dynamic interaction between wells. To ensure that the heterogeneity observed in the field is honored, multiple key areas are defined by making areal sectorization and lateral PVT variation when estimating FGIIP with P/Z approach. Communication between areas was evidenced from the sectoral P/Z. The field history matching was improved after applying the new estimated FGIIP. It was observed that the sectoral history matching both for production and pressure matches from some selected realizations built in static model resulted in better matches. Succinctly the re-evaluation of static derived FGIIP with P/Z method for the mature gas field was able to reduce the uncertainty range that initially existed. Incorporating the correct estimation of FGIIP in IAM has helped to yield reliable forecast and has enabled the asset to plan proper work programs for further field development. Analytical material balance with P/Z is very applicable for mature gas reservoirs producing under depletion mode. The approach is worth doing to narrow down the uncertainty range that was previously calculated. Moreover, the integration of analytical P/Z with static and dynamic model (IAM) has achieved more reliable forecasting of the mature gas field to proceed with further development plan.

This study aims to apply machine learning (ML) to make history matching (HM) process easier, faster, more accurate, and more reliable by determining whether Local Grid Refinement (LGR) with transmissibility multiplier is needed to history match gas-condensate wells producing from geologically complex reservoirs and determining how LGR should be set up to successfully history match those production wells. The main challenges for HM gas-condensate production from Hai Thach wells are large effect of condensate banking (condensate blockage), flow baffles by the sub-seismic fault network, complex reservoir distribution and connectivity, highly uncertain HIIP, and lack of PVT information for most reservoirs. In this study, ML was applied to analyze production data using synthetic samples generated by a very large number of compositional sector models so that the need for LGR could be identified before the HM process and the required LGR setup could also be determined. The proposed method helped provide better models in a much shorter time, and improved the efficiency and reliability of the dynamic modeling process. 500+ synthetic samples were generated using compositional sector models and divided into training and test sets. Supervised classification algorithms including logistic regression, Gaussian, Bernoulli, and multinomial Naïve Bayes, linear discriminant analysis, support vector machine, K-nearest neighbors, and Decision Tree as well as ANN were applied to the data sets to determine the need for using LGR in HM. The best algorithm was found to be the Decision Tree classifier, with 100% and 99% accuracy on the training and the test sets, respectively. The size of the LGR area could also be determined reasonably well at 89% and 87% accuracy on the training and the test sets, respectively. The range of the transmissibility multiplier could also be determined reasonably well at 97% and 91% accuracy on the training and the test sets, respectively. Moreover, the ML model was validated using actual production and HM data. A new method of applying ML in dynamic modeling and HM of challenging gas-condensate wells in geologically complex reservoirs has been successfully applied to the high-pressure high-temperature Hai Thach field offshore Vietnam. The proposed method helped reduce many trial and error simulation runs and provide better and more reliable dynamic models.

The prediction of drug-drug interactions (DDIs) from in vitro data usually utilizes an average dosing interval estimate of inhibitor concentration in an equation-based static model. Simcyp®, a population-based ADME simulator, is becoming widely used for the prediction of DDIs and has the ability to incorporate the time-course of inhibitor concentration and hence generate a temporal profile of the inhibition process within a dynamic model. Prediction of DDIs for 35 clinical studies incorporating a representative range of drug-drug interactions, with multiple studies across different inhibitors and victim drugs. Assessment of whether the inclusion of the time course of inhibition in the dynamic model improves prediction in comparison with the static model. Investigation of the impact of different inhibitor and victim drug parameters on DDI prediction accuracy including dosing time and the inclusion of active metabolites. Assessment of ability of the dynamic model to predict inter-individual variability in the DDI magnitude. Static and dynamic models (incorporating the time course of the inhibitor) were assessed for their ability to predict drug-drug interactions (DDIs) using a population-based ADME simulator (Simcyp®V8). The impact of active metabolites, dosing time and the ability to predict inter-individual variability in DDI magnitude were investigated using the dynamic model. Thirty-five in vivo DDIs involving azole inhibitors and benzodiazepines were predicted using the static and dynamic model; both models were employed within Simcyp for consistency in parameters. Simulations comprised of 10 trials with matching population demographics and dosage regimen to the in vivo studies. Predictive utility of the static and dynamic model was assessed relative to the inhibitor or victim drug investigated. Use of the dynamic and static models resulted in comparable prediction success, with 71 and 77% of DDIs predicted within two-fold, respectively. Over 40% of strong DDIs (>five-fold AUC increase) were under-predicted by both models. Incorporation of the itraconazole metabolite into the dynamic model resulted in increased prediction accuracy of strong DDIs (80% within two-fold). Bias and imprecision in prediction of triazolam DDIs were higher in comparison with midazolam and alprazolam; >50% of triazolam DDIs were under-predicted regardless of the model used. Predicted inter-individual variability in the AUC ratio (coefficient of variation of 45%) was consistent with the observed variability (50%). High prediction accuracy was observed using both the Simcyp dynamic and static models. The differences observed with the dose staggering and the incorporation of active metabolite highlight the importance of these variables in DDI prediction.

Marine reserves are widely used to manage for the potentially conflicting objectives of conserving biodiversity and improving fisheries. The fisheries and conservation benefits of alternative reserve designs would ideally be assessed using dynamic models, which consider spillover of fish and larvae to fished areas, and the displacement of fishers to unprotected sites. In practice, however, decisions about the location of marine reserves generally rely on cheaper and faster static models. Static models analyze only spatial patterns in habitats, and typically assume fisheries profits are reduced by the amount that was generated in areas designated as reserves. To help determine the benefits of developing dynamic fisheries models, we assessed how well static models estimate costs of reserve systems to fisheries and how outcomes from reserves designed using either static or dynamic models differ. We tested these questions in two case studies, the network of marine protected areas in southern California, USA and the proposed Tun Mustapha Marine Park in Malaysia. Static models could either under or over‐estimate the cost of reserve plans to fisheries, depending on the relative importance of fisher movement and larval dispersal dynamics. Despite the inaccuracy of static models for estimating costs, reserves designed using static models were similar to those designed with dynamic models if fisheries were well managed; or larval dispersal networks were simple. If larval networks were complex or there was overfishing, dynamic models generated substantially different reserve networks from static models, which improved conservation outcomes by up to 10% and fishing profits by up 20%. The time‐scale of management was also important, because only dynamic models accounted for larval dispersal, so could find reserves that maximized the long‐term benefits of larval spillover. Our case studies provide quantitative support for the assertion that static models can be useful for planning marine reserves for short‐term objectives in well managed fisheries, but are not reliable for evaluating the relative costs of reserves to fisheries.

This paper addresses proven best practice in modeling workflow including procedure and Qa/Qc criteria, which have to be applied during simulation models construction. The main issues discussed in this paper are as follows: –Static model with acceptable petrophysical parameters distribution including and honoring log and model Swi derived data per well as the basis for reliable dynamic model with realistic predictive mode.–Dynamic model as management tool with reasonable history match quality as assurance for reliable predictive mode of wells, areas and reservoir performance.–Define and quantify the volume of fluids-in-place, movable oil, residual oil and volumetric sweep efficiency to assess the reservoir potential, rate sustainability and economic ultimate recovery.–Assess the associated risks to development plans under selected development schemes with water/gas flood, WAG, artificial lift (ESP or Gas lift) and other EOR methods.–Model prediction mode quality and impact on strategic development decisions. As the oil industry has long experience in simulation techniques supported by availability of super computers and advanced software, it is observed that there are still major gaps that are not bridged yet. This paper will highlight some of those gaps and propose effective and practical solution based on best practice and lessons learnt in modeling studies to ensure reliable reservoir simulation predictive mode capabilities. This paper also includes the main criteria and assurance elements which were used to define modeling procedures that would participate in enhancing model reliability, and how they could impact development optimization process of selected production scheme towards achieving maximum recovery. Summary of these elements is as follows: Static to Dynamic Models Transition Phase–Well-per-well Swi match of log and model derived data. Acceptable level and trend match by using representative Pc's based on rock types & petrophysical data, MICP's, Height functions or combination.–Stability test to ensure good equilibrium condition with fluids distribution.–Well-per-well RFT/MDT field data and model derived data match.Dynamic Model History Match–Well-per-well acceptable trend match of observed data can be reached through a cycle of iterative process between geology, static and dynamic models to improve match.–Matching parameters and Qa/Qc criteria will be discussed later in details including; oil, gas and water rates and cumulative production, BHCIP, BHFP, WHFP, WCT and GOR.Prediction Mode of Development Plan–Well-per-well acceptable trend match (Rate, Pressures, WCT & GOR).–In case of abnormal predictive trend, consider the following remedial action:Review field measured data for accuracy, screen data as justified.Review imposed model constraints at well, group and field levels.Investigate solution with iterative process including static and dynamic models based on geology.

Where land surface air temperature data are not available, satellite land surface temperature are used. However, the coarse spatial resolution of satellite‐derived products may yield errors at the local scale. This work shows the differences between the MODIS Land Surface Temperature and Emissivity (MOD11A1) product and ground measurements at two different scales. We used data from 21 SNOTEL stations across the northern Front Range of Colorado to represent the coarse scale and 17 iButton temperature sensors across the Colorado State University Mountain Campus to represent the fine scale. We found significant differences in the temperature and its changes with elevation for the two spatial scales. At the fine scale, cold air drainage can induce an inversion of the temperature gradient with elevation. A higher correlation was found during the nighttime at the fine scale, while, at the coarse scale, higher correlations were observed during the daytime. On windy nights, temperatures do not cool as much as on calmer nights, and the coarse scale near‐surface temperature gradient with elevation persists, though the fine scale inversions do not develop. The near‐surface temperature gradients with elevation based on the MODIS pixels are similar to the ground‐based data at the coarse scale but not at the fine scale. Thus, one must be cautious in selecting the near‐surface temperature gradients with elevation for mountainous terrain when different scales are considered, and a proper validation of satellite products is necessary prior to their use to avoid the propagation of uncertainties.

In the coarse scale simulation of heterogeneous reservoirs, effective or upscaled flow functions, e.g., oil and water relative permeability and capillary pressure, can be used to represent heterogeneities at subgrid scales. The effective relative permeability is typically upscaled along with absolute permeability from a geostatistical model. However, the potentially important effects of smaller scale heterogeneities (on the centimeter to meter scale) in both capillarity and absolute permeability will not be captured by this approach. In this paper, we present a new two-stage upscaling procedure for two-phase flow. In the first stage, we upscale from the core (fine) scale to the geostatistical (intermediate) scale, while in the second stage we upscale from the geostatistical scale to the simulation (coarse) scale. The computational procedure includes numerical solution of the finite difference equations describing steady state flow over the local region to be upscaled, using either constant pressure or periodic boundary conditions. The two-stage method is applied to synthetic two-dimensional reservoir models with strong variation in capillarity on the fine scale. Results are presented in terms of both oil production rates and saturation fields. Accurate reproduction of the fine grid solutions (simulated on 500 × 500 grids) is achieved on coarse grids of 10 × 10 for different flow scenarios. It is shown that, although capillary forces are important on the fine scale, the assumption of capillary dominance in the first stage of upscaling is not always appropriate, and that the computation of rate dependent effective properties in the upscaling can significantly improve the accuracy of the coarse scale model. The assumption of viscous dominance in the second upscaling stage is found to be appropriate in all of the cases considered.

Traditionally, the history matching process is done only on the dynamic model, without any direct update to the geological (or static) model. As a result, geological uncertainties are not fully evaluated in the dynamic model. Non-integration of static and dynamic modelling results in either too much time being spent modelling detailed geological phenomena that have little impact on the dynamic behaviour of the reservoir, or, conversely, important geological and petrophysical parameters being misrepresented or missed out which may have significant impacts on the overall field development strategy. Ideally, if any updates to static parameters are required as result of history matching in the dynamic model, these changes should be reflected directly in the static reservoir model, thereby ensuring consistency between the static and dynamic models. In this paper, a workflow is presented where both the static and dynamic modelling software packages are integrated as part of the history matching process. This workflow involves input parameters being adjusted in the geological model directly. Uncertainty analysis tools are used to obtain multiple history-matched models, which results in an order of magnitude increase in speed compared to traditional history-matching processes. Not only will this methodology result in improved history-matched models with a wider range of production forecasts being captured, but more importantly, it will result in better understanding of the static and dynamic uncertainties and their interdependencies, leading to a more informed decision-making process with regards to overall field development. In addition, this methodology offers a platform where the subsurface professionals involved in reservoir model construction and simulation processes can focus their efforts on improving reservoir characterization and identify areas that require further data acquisition or improvement. This paper also describes how the workflow was successfully applied to a recently developed, producing and waterflooded oil field in South East Asia, and eventually delivering an optimized reservoir model for reservoir management and a probabilistic approach to production forecasting.