Evaluation of particle swarm optimization in synthesis of shear wave velocity using conventional well log data

Based on the extensive studies, undoubtedly, the shear wave velocity (Vs) plays a fundamental role in hydrocarbon reservoir evaluation. Using Vs often allows us to identify the seismic signatures of lithology, pore fluid type and pore pressure, efficiently. Unfortunately Vs data is not available in all reservoirs and it is necessary to predict it. To date, intelligent systems have been utilized as powerful and routine tools for this purpose. In this study, PSO algorithm that is one of the artificial intelligence system used for Vs estimation. The algorithm is utilized in linear and nonlinear ways by intelligently derived Equations. This study have a total 3190 data points from Asmari reservoir from two wells in one of the oilfields, SW Iran. All data points have porosity log and measurement Vs by Dipole Shear Sonic Imager. These data are divided into two parts, one part included 2090 data points are used for constructing model and the other part included 1100 data points are used for testing and validation model. Shear wave velocity is predicted by PSO algorithm and results compared with real Vs measurements by DSI. Regression between predicted Vs by Linear PSO algorithm and measured Vs is about 0.92, whereas it is about 0.95 by using nonlinear PSO algorithm. Results of this study and comparison with other studies confirm that particle swarm optimization is suitable for predicting Vs in other oilfields.