Reservoir simulation models that represent Saudi Arabia’s unique large reservoirs require significant use of High-Performance Computing resources. Several solutions aim to reduce the load of an individual simulation such as up-scaling to consequently reduce the number of grid cells of the model, cropping the model which loses the flux boundary conditions and running the studies independently, or dividing the model to sectors which preserves the flux boundary condition and running these models independently. Sector modeling could potentially result in the least erroneous solution, especially if the sectors are defined on the regions of least connectivity.
This paper aims to propose an automated, intelligent method to divide the reservoir model into an arbitrary number of least-connected smaller sectors. Streamline simulation output is used as a representation of reservoir connectivity. A graph is built using cells as graph vertices, and the edge weight is calculated based on the time of flight of oil and water between cell pairs.
By presenting the problem in this manner, a graph is built where the non-water cells have equal weight and the stronger the connection between two cells, represented by the lower time of flight, the larger the edge weight. A graph partitioning tool is used with the purpose of minimizing the total weight of edges cut while keeping the number of vertices in each sub-graph balanced up to a specified tolerance. Partitioning of a graph specified this way is equivalent to splitting the reservoir model into sub-models while avoiding cutting of strongly connected parts, hence minimizing the boundary flux. Applying this method to sector modeling allows splitting a model into a number of smaller sub-models that can be used independently as the interactions between them are minimized, as demonstrated by minimizing flux between them.
The proposed novel method has been tested on several real-field as well as synthetic reservoir models. The method has shown to result in sub-models of loosely connected reservoirs. The advantage of our proposed method is especially seen for strongly connected models where it is difficult to identify the least erroneous partitioning for sector generation manually. However, with the use of our method, it is guaranteed to automatically find the least connection, while minimizing the error that sector division produces as evident by the low flux between the sector models.
Electrical utilities are subject to voltage sags, poor power factor and even voltage instability as long as it suffers from shortage of its reactive power sources. The only mitigation to this problem is to have proper and fast acting reactive power control on the grid. This in turn will overcome the major concerns of network voltage instability, especially during the transient and sub-transients conditions such as major switching operation, and electrical faults. In many cases, the traditional solutions of switching capacitors is too coarse and slow to stabilize a weak network. The most advanced solution to compensate reactive power is to incorporate a STATCOM (or Static Synchronous Compensator) as a variable source of reactive power. STATCOM is an advanced power electronic system for fast capacitive/inductive reactive power supply providing reactive power compensation and, steady state and transient voltage stability, etc… These systems offer advantages compared to standard reactive power compensation solutions in demanding applications. In order to mitigate utility network power quality problems, ADNOC Gas Processing has recently introduced a STATCOM solution in two running plants. This paper deals with the design process of customized Power Quality solutions involving STATCOM Technology to resolve network quality issue up to full commissioning of the system, elaborate on challenges and difficulties faced, and highlight how it was resolved.
The Mishrif formation in Abu Dhabi comprises progradational shelf margin facies. The western platform sediments are characterized by stacked clinoforms of clean high energy carbonates, with generally good reservoir properties at the top deteriorating gradually toward the west flank. In contrast in the east the formation is thicker and characterized by more differentiated, coarsening, shoaling- upwards sequences. High quality reservoir facies occur only near the prograding shelf edges.
The reservoir in this study had finely layered pillar grid models of the field using onlap and offlap layering to capture the vertical property heterogeneity and layering within the clinoforms implied by the depositional environment. However, this grid structure posed challenges for flow simulation as there were unphysical barriers and connections across the clinoform boundaries caused by the pinching out layers at boundaries between clinoform units.
By construction, in a depogrid each clinoform may be independently gridded with coordinate lines that do not need to be continuous through the vertical extent of the reservoir. Layers within a sequence can truncate arbitrarily against bounding discontinuities since the cells are polyhedral and the grid globally unstructured. Therefore, an evaluation of the depogrid cut-cell grid was undertaken for the reservoir.
A volume-based model was constructed using horizon surfaces and fault surfaces extracted from the pillar grid model, and the depospace transform calculated. The depogrid was created using the same areal resolution as the pillar grid. The layer parameters were set to give approximately the same number of layers in each zone as the pillar grid. The rock type base properties from the pillar grid were upscaled onto the depogrid, in the interest of time, to populate the static depogrid model. Reservoir fluid properties, relative permeability, and capillary pressure curves were taken from existing simulation models of the field. A high-resolution reservoir simulator was then run on the depogrid model. Run time comparisons of the pillar and depogrid simulations were made
The evaluation concluded that the cut-cell stratigraphically layered depogrid provided a more geologically consistent representation of the complex stacked clinoform structural elements and required significantly fewer grid layers to achieve the required vertical resolution. The depogrid simulations could represent the expected connectivity for flow. Improved run times were observed
Reservoir simulation results currently provide the basis for important reservoir engineering decisions; grid complexity and non-linearity of these models demand high computational time and memory. The physics-based simulation process must be repeated to increase model prediction accuracy or to perform history matching; consequently, the simulation process is often time-consuming. This paper describes a new methodology based on a deep neural network (DNN) technique, the graph convolutional neural network (G-CNN). G-CNN increases the modeling prediction speed and efficiency by compressing the computational time and memory usage of the reservoir simulation. A G-CNN model was used to perform the reservoir simulations described.
This new methodology combines physics-based and data-driven models in reservoir simulation. The workflow generates training datasets, enabling intelligent sampling of the reservoir production data in the G-CNN training process. Bottomhole pressure constraints were set for all simulations. The production data generated by the reservoir model, with the mesh connectivity information, is used to generate the G-CNN model. This approach can be viewed as hybrid data-driven, retaining the underlying physics of the reservoir simulator. The resulting G-CNN model can perform reservoir simulations for any computational grid and production time horizon. The method uses convolutional neural network and mesh connections in a fully differentiable scheme to compress the simulation state size and learns the reservoir dynamics on this compressed form.
G-CNN analysis was performed on an Eagle Ford-type reservoir model. The transmissibility, pore volume, pressure, and saturation at an initial time state were used as input features to predict final pressure and saturations. Prediction accuracy of 95% was obtained by hyperparameter tuning off the G-CNN architecture. By compressing the simulation state size and learning the time-dependent reservoir dynamics on this compressed representation, reservoir simulations can be emulated by a graph neural network which uses significantly less computation and memory. The G-CNN model can be used on any computational grid because it preserves the structure of the physics. The G-CNN model is trained by mapping an initial time state to a future prediction time state; consequently, the model can be used for generalizing predictions in any grid sizes and time steps while maintaining accuracy. The result is a computationally and memory efficient neural network that can be iterated and queried to reproduce a reservoir simulation.
This novel methodology combines field-scale physics-based reservoir modeling and DNN for reservoir simulations. The new reservoir simulation workflow, based on the G-CNN model, maps the time state predictions in a resampled grid, reducing computational time and memory. The new methodology presents a general method for compressing reservoir simulations, assisting in fast and accurate production forecasting.
Chen, Zhiming (State Key Laboratory of Petroleum Resources and Prospecting, China University of Petroleum at Beijing) | Liu, Hui (State Key Laboratory of Petroleum Resources and Prospecting, China University of Petroleum at Beijing) | Liao, Xinwei (State Key Laboratory of Petroleum Resources and Prospecting, China University of Petroleum at Beijing) | Zhao, Xiaoliang (State Key Laboratory of Petroleum Resources and Prospecting, China University of Petroleum at Beijing) | Tang, Xuefeng (State Key Laboratory of Petroleum Resources and Prospecting, China University of Petroleum at Beijing) | Meng, Meiling (State Key Laboratory of Petroleum Resources and Prospecting, China University of Petroleum at Beijing)
Due to the complexity of shale reservoir geology, hydraulic and micro-fractures can be coupled into an extremely complex symmetrical or asymmetrical fracture network around vertically fractured wells (VFW) after fracturing. The important and useful work is to analyze the transient pressure response of the VFW, to more accurately predict the productivity of VFW. In this paper, a numerical method to accurately simulate the complex fracture network geometry and analyze the transient pressure responses of the VFW, due to the complexity of the fracture network geometry. The results show a longer fracture length on one side causes a smaller pressure depletion, a shorter bilinear flow, and a deeper and longer the degree of "dip". The more fractures on one side can lead to a greater degree of "dip" and a smaller pressure depletion. With the fracture conductivity on the right side increases, while that in other side remains constant value, it results in a shorter bilinear flow, a deeper and longer the degree of "dip", a smaller pressure depletion, and a weaker bi-radial flow (BRF). In addition, it is found that flow regimes affected by magnitude of fracture networks are mainly bi-linear flow (BLF), "dip" and BRF. The pressure behaviors between asymmetrical fracture networks and symmetrical fracture networks are mainly in the periods of BLF, "dip", and BRF. Through analyzing the transient pressure responses of the VFW, the parameters of the complex fracture network can be well predicted, so that the productivity of the VFW can be estimated more accurately.
Moreno Ortiz, Jaime Eduardo (Schlumberger) | Gossuin, Jean (Schlumberger) | Liu, Yunlong (Schlumberger) | Klemin, Denis (Schlumberger) | Gurpinar, Omer (Schlumberger) | Gheneim Herrera, Thaer (Schlumberger)
Challenges on EOR process upscaling have been discussed extensively in the industry and effects of diffusion, dispersion, heterogeneity, force balance and frontal velocity -among others, recognized and qualified, along with the importance of understanding the numerical model finite difference equations and modeling strategy. Augmenting the upscaling complexity is the often-limited understanding/data on the EOR displacement at different scales (from micro to full field), including the EOR agent/rock/fluid interactions that is often available at the early stages of the EOR process de-risking.
A common denominator for the EOR process characterization and upscaling (along with the discretization of the displacement) is the non-uniqueness nature of the problem. As the complexity of numerical representation of the EOR process increases (thus increasing data characterization requirements), so does the number of plausible solutions and challenges when dealing with an otherwise incomplete dataset. Digital rock has evolved as a strong alternative to complement laboratory corefloods, allowing for EOR agent optimization on a high-resolution digital representation of the pore structure, detailed digital fluid model of both reservoir fluids and EOR agents and physical rock-EOR agent-reservoir fluid interaction, thus providing several calibration points to ensure the finite-difference model calibration and upscaling preserve the process behavior.
This paper discusses the use of digital rock solutions on the EOR deployment, particularly on translating the results to numerical finite difference models, addressing the inherent laboratory measurement uncertainty and proposing a fit-for-purpose multi-scale upscaling strategy that addresses both effects of heterogeneity and EOR agent characterization during the upscale process.
This paper addresses the challenges of chemical flooding upscaling, particularly polymer by using a real-life polymer injection case where digital rock, corefloods and more importantly pilot results are available to test and validate our observations. Using a polymer coreflood and digital rock results as input, numerical finite difference simulation models were built and calibrated to effectively reproduce the displacement physics observed on both digital rock and corefloods, digital flood results were used to bridge the laboratory-to-numerical model step by providing effective upscaled polymer properties as well as intrinsic rock properties such as relative permeability and capillary pressures, which are then taken through a series of multi-scale finite difference models to identify, validate and quantify upscaling requirements, addressing polymer deformation through pore throats and effective simulation viscosity. Digital rock is used to rank and resolve ambiguity on the finite difference model calibration by providing an otherwise rare opportunity to visualize the displacement in the 3D space. The analysis shed a new light on fluid-fluid and fluid-rock interaction at pore scale and enabled us to improve on the finite difference model generation and polymer properties.
Wu, Yonghui (China University of Petroleum-Beijing, and Texas A&M University) | Cheng, Linsong (China University of Petroleum-Beijing) | Killough, John E. (Texas A&M University) | Huang, Shijun (China University of Petroleum-Beijing) | Fang, Sidong (Sinopec Exploration and Production Research Institute) | Jia, Pin (China University of Petroleum-Beijing) | Cao, Renyi (China University of Petroleum-Beijing) | Xue, Yongchao (China University of Petroleum-Beijing)
The large uncertainty in fracture characterization for shale gas reservoirs seriously affects the confidence in making forecasts, fracturing design, and taking recovery enhancement measures. This paper presents a workflow to characterize the complex fracture networks (CFNs) and reduce the uncertainty by integrating stochastic CFNs modeling constrained by core and microseismic data, reservoir simulation using a novel edge-based Green element method (eGEM), and assisted history matching based on Ensemble Kalman Filter (EnKF).
In this paper, the geometry of CFNs is generated stochastically constrained by the measurements of hydraulic fracturing treatment, core, and microseismic data. A stochastic parameterization model is used to generate an ensemble of initial realizations of the stress-dependent fracture conductivities of CFNs. To make the eGEM practicable for reservoir simulation, a steady-state fundamental solution is applied to the integral equation, and the technique of local grid refinement (LGR) is applied to refine the domain grids near the fractures. Finally, assisted-history-matching based on EnKF is implemented to calibrate the DFN models and further quantify the uncertainties in the fracture characterization.
The proposed technique is tested using a multi-stage fractured horizontal well from a shale gas field. After analyzing the history matching results, the proposed integrated workflow is shown to be efficient in characterizing fracture networks and reducing the uncertainties. The advantages are exhibited in several aspects. First, the eGEM-based Discrete-Fracture Model (DFM) is shown to be quite efficient in assisted history matching of large field applications because of eGEM’s high precision with coarse grids. This enables simulations of CFNs without upscaling the fractures using continuum approaches. In addition, CFNs geometry can be generated with the constraints of core and microseismic data, and a primary conductivity of CFNs can be generated using the hydraulic fracturing treatment data. Moreover, the uncertainties for CFNs characterization and EUR predictions can be further reduced with the application of EnKF in assimilating the production data.
This paper provides an efficient integrated workflow to characterize the fracture networks in fractured unconventional reservoirs. This workflow, which incorporated several efficient techniques including fracture network modeling, simulation and calibration, can be readily used in field applications. In addition, various data sources could be assimilated in this workflow to reduce the uncertainty in fracture characterization, including hydraulic fracturing treatment, core, microseismic and production data.
The simulation of the In Situ Combustion (ISC) process is a very challenging process due to the complexity and nonlinear nature of the problem. In this work, we propose an efficient technique to simulate experimental procedures for the ISC process including heterogeneity. The effects of permeability on mass flow and heat transfer were studied through a series of numerical frameworks. Different approaches to model the reactions occurring during combustion were attempted and simulation results were validated using experimental results. We focus on two different key areas: the integration of chemical reaction kinetics obtained through kinetic cell experiments, and the analysis of efficient simulations of combustion tube experiments that account for the flow element. After establishing a robust framework that accurately matches lab-scale results, combustion tube simulation results using a commercial simulator were analyzed to corroborate conclusions. Through observing the propagation of the combustion front and the oil bank in heterogeneous zones, assessments around the effects of permeability on the ISC process were performed. This work provides valuable information that would be instrumental in understanding experimental behavior of in-situ combustion and upgrading results to field scale after matching numerical results with experimental data collected in our future work.
We develop a novel ensemble model-maturation method that is based on the Randomized Maximum Likelihood (RML) technique and adjoint-based computation of objective function gradients. The new approach is especially relevant for rich data sets with time-lapse information content. The inversion method that solves the model-maturation problem takes advantage of the adjoint-based computation of objective function gradients for a very large number of model parameters at the cost of a forward and a backward (adjoint) simulation. The inversion algorithm calibrates model parameters to arbitrary types of production data including time-lapse reservoir-pressure traces by use of a weighted and regularized objective function. We have also developed a new and effective multigrid preconditioning protocol for accelerated iterative linear solutions of the adjoint-simulation step for models with multiple levels of local grid refinement. The protocol is based on a geometric multigrid (GMG) preconditioning technique. Within the model-maturation workflow, a machine-learning technique is applied to establish links between the mesh-based inversion results (e.g., permeability-multiplier fields) and geologic modeling parameters inside a static model (e.g., object dimensions, etc.). Our workflow integrates the learnings from inversion back into the static model, and thereby, ensures the geologic consistency of the static model while improving the quality of ensuing dynamic model in terms of honoring production and time-lapse data, and reducing forecast uncertainty. This use of machine learning to post-process the model-maturation outcome effectively converts the conventional continuous-parameter history-matching result into a discrete tomographic inversion result constrained to geological rules encoded in training images.
We demonstrate the practical utilization of the adjoint-based model-maturation method on a large time-lapse reservoir-pressure data set using an ensemble of full-field models from a reservoir case study. The model-maturation technique effectively identifies the permeability modification zones that are consistent with alternative geological interpretations and proposes updates to the static model. Upon these updates, the model not only agrees better with the time-lapse reservoir-pressure data but also better honors the tubing-head pressure as well as production logging data. We also provide computational performance indicators that demonstrate the accelerated convergence characteristics of the new iterative linear solver for adjoint equations.
As data computing and big data driven analytics become more prevalent in a number of spatial industries, there is increasing need to quantify and communicate uncertainty with those data and resulting spatial analytical products. This has direct implication in oil & gas exploration and development where big data and data analytics continue to expand uses and applications of spatial and spatio-temporal data in the industry without providing for effective communication of spatial uncertainty. The result is that communications and inferences made using spatial data visuals lack crucial information about uncertainty and thus present a barrier to accurate and efficient decision making. With increasing cost awareness in oil & gas exploration and development, there is urgent need for methods and tools that help to objectively define and integrate uncertainty into business decisions.
To address this need, the Variable Grid Method (VGM) has been developed for simultaneous communication of both spatial patterns and trends and the uncertainty associated with data or their analyses. The VGM utilizes varying grid cell sizes to visually communicate and constrain the uncertainty, creating an integrated layer that can be used to visualize uncertainty associated with spatial, spatio-temporal data or data-driven products.
In this paper, we detail the VGM approach and demonstrate the utility of the VGM to intuitively quantify and provide cost-effective information about the relationship between uncertainty and spatial data. This allows trends of interest to be objectively investigated and target uncertainty criteria defined to drive optimal investment in improved subsurface definition. Examples are presented to show how the VGM can thus be used for efficient decision making in multiple applications including geological risk evaluation, as well as to optimize data acquisition in exploration and development.
Today, uncertainty, if it is provided at all, is generally communicated using multiple independent visuals, aggregated in final displays, or omitted altogether. The VGM provides a robust method for quantifying and representing uncertainty in spatial data analyses, offering key information about the analysis, but also associated risks, both of which are vital for making prudent business decisions in oil & gas exploration and development.