estimation of original hydrocarbon in place via neural network

accurate estimation of hydrocarbon volume in a reservoir is important due to future development and investment on that reservoir. estimation of oil and gas reservoirs continues from exploration to end of reservoir time life and is usual upstream engineer’s involvements. in this study we tried to make reservoir properties models (porosity and water saturation) and estimate reservoir volume hydrocarbon based on artificial neural network tools, petrophysical and geophysical data. so with gridding the reserve, separate it to same volume cells. based on porosity and lithology variation in wells, constructed petrophysical zonation in each well and by correlation these zones in wells reservoir has been zoned. porosity, water saturation and 3d seismic data have been averaged in cells and assigned one value for each cell. at final a three layer perceptron neural network by back propagation error algorithm has been designed and trained by using cells which had petrophysical data; then these parameters have been estimated in other cells and original hydrocarbon in place calculated and compared with results from mont carlo method.