Reservoir simulation is routinely used as a reservoir management tool. The static model that is used as the basis for simulation is the result of an integrated effort that usually includes the latest geological, geophysical and petro-physical measurements and interpretations. As such, it is inherently a model with some uncertainty. Analysis of these uncertainties and quantification of their effects on oil production and water cut using a new and efficient technique is the subject of this paper. Typical uncertainty analysis techniques require many realizations and runs of the reservoir model. In the day and age that reservoir models are getting larger and more complicated, making hundreds or sometimes thousands of simulation runs can put considerable strain on the resources of an asset team. This paper summarizes the results of uncertainty analysis on a giant oil field in the Middle East using a new technique that incorporates a Surrogate Reservoir Model (SRM). A Surrogate Reservoir Model that runs and provides results in real-time is developed to mimic the capabilities of a full field model that includes about one million grid blocks and takes 10 hours to run on a cluster of twelve 3.2 GHz CPUs. This Surrogate Reservoir Model is used as the objective function of a Monte Carlo Simulation to study the impact of the uncertainties associated with several parameters on the model outcome, i.e. oil production and water cut is analyzed. The analysis can be performed individually on each of the 165 horizontal wells. During the analyses of uncertainty, the Surrogate Reservoir Model will serve as an objective function for the Monte Carlo Simulation. In this study, uncertainties associated with several reservoir parameterts and their quantitative effect on cumulative oil production and stantaneous water cut are examined.

https://shahab.pe.wvu.edu/Publications/Pdfs/102492.pdf

Quantifying Uncertainties Associated With Reservoir Simulation Studies Using a Surrogate Reservoir Model

Waterflooding is a significantly important process in the life of an oil field to sweep previously unrecovered oil between injection and production wells and maintain reservoir pressure at levels above the bubble-point pressure to prevent gas evolution from the oil phase. This is a critical reservoir management practice for optimum recovery from oil reservoirs. Optimizing water injection volumes and optimizing well locations are both critical reservoir engineering problems to address since water injection capacities may be limited depending on the geographic location and facility limits. Characterization of the reservoir connectivity between injection and production wells can greatly contribute to the optimization process. In this study, it is proposed to use computationally efficient methods to have a better understanding of reservoir flow dynamics in a waterflooding operation by characterizing the reservoir connectivity between injection and production wells. First, as an important class of artificial intelligence methods, artificial neural networks are used as a fully data-driven modeling approach. As an additional powerful method that draws analogy between source/sink terms in oil reservoirs and electrical conductors, capacitance---resistance models are also used as a reduced-physics-driven modeling approach. After understanding each method's applicability to characterize the interwell connectivity, a comparative study is carried out to determine strengths and weaknesses of each approach in terms of accuracy, data requirements, expertise requirements, training algorithm and processing times.

/pdf/characterizing-interwell-connectivity-in-waterflooded-44ttkxtiyl.pdf

Characterizing interwell connectivity in waterflooded reservoirs using data-driven and reduced-physics models: a comparative study

A holistic review on artificial intelligence techniques for well placement optimization problem

Development of a data-driven forecasting tool for hydraulically fractured, horizontal wells in tight-gas sands

Water Alternating CO2 injection (WAG CO2) is one of the most promising EOR techniques for further recovery improvement A successful implementation of this method depends mainly on its optimal design parameters, which are determined using a significant number of routinely numerical simulations However, these latter are very time-consuming and their analysis can put a considerable strain on the flexibility of application The aim of this study is to develop a new method to optimize WAG processes in presence of multiple conflicting criteria and time-depending constraints To this end, a hybrid model based on multilayer perceptron (MLP), which was used as a replica of the simulator in mimicking the outputs, and Non-Dominated Sorting Genetic Algorithm version II (NSGA-II) was applied Three different MLP models were built by implementing LMA, BR and SCG algorithms during the MLP training phase The results showed that MLP-LMA model is the most accurate proxy In addition, the proposed MLP-LMA emulates the outputs in real-time (dynamic proxy) with a high accuracy and a tiny running-time The hybridization NSGA II-proxy ensures the approximation of the Pareto front to the formulated WAG problem The generated Pareto front provided many WAG scenarios yielding to practical decision-making capabilities The findings of this review can help for better understanding and studying multi-objective based WAG optimization problems in presence of time-depending constraints

Cyrus Abdallah Modavi

Papers

Development of Surrogate Reservoir Models (SRM) For Fast Track Analysis of Complex Reservoirs