A Novel Intelligent Technique to Assess the Reservoir Model by Using the Pressure Derivative Data

Several techniques are used for reservoir model recognition purpose during the well test analysis.These techniques are categorized into conventional straight line analysis and type curvematching method. At the first method, the behavior of each mathematical reservoir model isturned into a form that leads to a straight line of pressure-time at Cartesian, semi-logarithmic orlogarithmic plot. The deficiency of this method is that it is not possible to see the signature ofthe all flow regimes at the same plot. Hence, pressure and pressure derivative type curves weredeveloped to deal with this problem (Bourdet et al., 1983).Nowadays, pressure derivative type curves are used frequently at the oil industry to recognizethe reservoir model and estimate its parameters simultaneously. At the derivative type curveanalysis, the derivative of the pressure data are calculated numerically and plotted against thewell test time at the logarithmic plot. Then, it is matched with the derivative type curvesbelonging to the variety of reservoir model, thereby assessing type of the reservoir model andestimate the corresponding model s parameters. Nonlinear regression well testing is a processwhich automatically matches the well test data with some synthetic type curves by employing aLaplace inverse and an optimization algorithm. However, usual nonlinear regression analysisrequires manual modification of the reservoir parameters and monitoring the resulting match forthe best possible match. Also, type of the reservoir model should be selected prior to theregression analysis and user should carry out the nonlinear analysis for each reservoir model andeach set of model parameters in a series manner. Therefore, aim of this study is to propose amethod based on the meta-heuristic optimization to increases the speed of the model recognitionand parameter estimation by parallelizing the calculations.