Layered Pseudo-Steady-State Models for tight commingled gas reservoirs

Abstract

Analysis of commingled reservoirs from limited data can be a challenge to most conventional reservoir engineering tools. The purpose of this research is to find an effective and easy technique that can be used to estimate the OGIP and predict the reservoir performance for tight commingled gas reservoirs. A two-dimensional, finite difference, gas reservoir simulator was used to generate rate-time data for hypothetical layers. These data were added together in several combinations to simulate commingled reservoirs' rate performance. A Layered PseudoSteady-State (PSS) Model for gas reservoirs was developed and used to analyze the ratetime data for the hypothetical commingled reservoirs. The model requires only an estimate for the initial reservoir pressure and the BHFP-Each layer within the commingled model is defined by two parameters (OGIP and the gas flow coefficient, Jg). This model couples the material balance (M. B.) equation with the PSS flow equation for gas and can be effectively used to model any number of layers in a commingled system. The analysis is done through history-matching the production rate with the model. An optimization routine was coupled to the model to aid in the matching process and help determining the best parameters that describe the performance of each layer. The Layered PSS Model gave satisfactory results and the match with the simulation data was excellent for cases that are in the PSS period. The effect of transient data on the model answers was also investigated. It was found that the model always gives conservative OGIP and conservative forecast if any transient data were included. It was also concluded that the Layered PSS Model also under-estimates the total OGIP for a commingled system if the number of layers in the commingled reservoirs are larger than the number of layers used in the analysis. The effect of Non-Darcy flow was also studied. It was found that the non-darcy flow does not have a significant effect on the model answers for tight layers. It was shown that the analysis of selective intervals rather than the whole data is useful in identifying transients and commingling. Also, an analysis methodology for any rate-time data was suggested.