Abstract:
Analysis of well performance data can deliver decision-making solutions regarding field development, production optimization, and reserves evaluation. Well performance analysis involves the study of the measured response of a system, the reservoir in our case, in the form of production rates and flowing pressures. The Eagle Ford shale in South Texas is one of the most prolific shale plays in the United States. However, the ultra-low permeability of the shale combined with its limited production history makes predicting ultimate recovery very difficult, especially in the early life of a well. Use of Rate Transient Analysis makes the analysis of early production data possible, which involves solving an inverse problem. Unlike the traditional decline analysis, Rate Transient Analysis requires measured production rates and flowing pressures, which were provided by an operator based in the Eagle Ford. This study is divided into two objectives. The first objective is to analyze well performance data from Eagle Ford shale gas wells provided by an operator. This analysis adopts the use of probabilistic rate transient analysis to help quantify uncertainty. With this approach, it is possible to systematically investigate the allowable parameter space based on acceptable ranges of inputs such as fracture length, matrix permeability, conductivity and well spacing. Since well spacing and reservoir boundaries were unknown, a base case with a reservoir width of 1500 feet was assumed. This analysis presents a workflow that integrates probabilistic and analytical modeling for shale gas wells in an unconventional reservoir. To validate the results between probabilistic and analytical modeling, a percent difference of less than 15% was assumed as an acceptable range for the ultimate recoverable forecasts. Understanding the effect of existing completion on the cumulative production is of great value to operators. This information helps them plan and optimize future completion designs while reducing operational costs. This study addresses the secondary objective by generating an Artificial Neural Network model. Using database from existing wells, a neural network model was successfully generated and completion effectiveness and optimization analysis was conducted. A good agreement between the predicted model output values and actual values (R² = 0.99) validated the applicability of this model. A completion optimization study showed that wells drilled in condensate-rich zones required higher proppant and liquid volumes, whereas wells in gas-rich zones required closer cluster spacing. Analysis results helped to identify wells which were either under-stimulated or over-stimulated and appropriate recommendations were made.
