Performing Reservoir Simulation with Neural Network Enhanced Data

Abstract

Abstract The use of artificial neural networks (ANN) for reservoir analysis now makes it possible to predict important reservoir properties from combinations of data such as well logs, production data, seismic data, etc. In this work, an ANN was combined with a geostatistical linear estimation algorithm in a technique called the hybrid approach, which was used to enhance sparse data to include in a reservoir simulation model with the goal of reducing history matching time. The case study field, Fort Collins Field, is situated on the N-S anticline on the western edge of the Denver Basin in Colorado. The available porosity and permeability data needed to build a reservoir simulation model are old and sparse. Available well logs and cores were used as inputs to the hybrid model. This model was then used to predict porosity and permeability for the reservoir and these values were then included in a reservoir simulation model. To evaluate the hybrid approach, the reservoir simulation model was history matched with the case study historical production data and compared to a model with average data. The result confirms that the hybrid approach history matched better and faster compared to the simple averaging-technique. The history match results from both methods were compared based on the percentage error. This unique approach will benefit older fields with sparse data.