Estimation of Rock Compressibility for Indonesian Limestones by Artificial Neural Network

Compressibility is one of the important physical properties of rocks. Rock compressibility greatly influences the calculation of the volume of oil and gas in a reservoir based on pressure drop data. In addition, compressibility plays a role in the process of pushing oil and gas through rock pores towards production wells. Compressibility is influenced by overburden and pore pressure. In this study, three ANN models were applied to estimate rock compressibility. The optimum number of neurons and number of hidden layers were determined for the three ANN models based on the highest correlation coefficient. The ANN 2 model with porosity as an input parameter is the most accurate model because it has the lowest mean absolute relative error value and the highest correlation coefficient value. Based on the correlation coefficient value, the optimum number of neurons and number of hidden layers for the ANN 2 model are 8 and 3 respectively. In general, compared to conventional models, all ANN models have better accuracy. Among conventional models, the use of the Horne equation shows the best accuracy in predicting the compressibility of limestone in Indonesia.