Application of Genetic Programing technique for predicting Uniaxial Compressive Strength using reservoir formation properties

• Published in: Journal of petroleum science & engineering Vol. 159; pp. 35 - 48
• Main Authors: Koolivand-Salooki, M., Esfandyari, M., Rabbani, E., Koulivand, M., Azarmehr, A.
• Format: Journal Article
• Language: English
• Published: Elsevier B.V 01.11.2017
• Subjects: Density; Genetic Programming (GP); Porosity; Uniaxial Compressive Strength (UCS); Water saturation; Water saturation; Uniaxial Compressive Strength (UCS); Genetic Programming (GP); Porosity; Density
• ISSN: 0920-4105; 1873-4715
• DOI: 10.1016/j.petrol.2017.09.032

In this study, we developed a relationship for Uniaxial Compressive Strength (UCS) based on total formation porosity, bulk density and water saturation using Genetic Programming (GP). The numerical values of these parameters, which offered rock UCS, were obtained by analyzing various logs including sonic, neutron, gamma ray and electric logs. The elastic moduli were calculated by analyzing compressional and shear sonic logs and using mathematical correlations. The rock UCS was then analytically calculated using the empirical Wang and Plumb correlations (Plumb R.A. 1994). In order to predict UCS of the formation rock using the GP technique, approximately 5000 data points associated with 3 different wells in one of the Iranian oil fields were collected and analyzed. The data points associated with one of the wells were used to structure the GP model after being calibrated with some UCS data derived from core analysis while the other two sets of the data points were employed to test the accuracy of model's predictions. The predicted UCS values using GP model were consistent and in very good agreement with the corresponding values calculated analytically using well log data. •We developed a relationship for UCS based on total formation porosity, bulk density and water saturation using GP.•Prediction of UCS of the formation rock using the GP technique.•The predicted UCS values using GP model were consistent and in very good agreement with the corresponding values calculated analytically using well log data.•A well-trained and tested GP model was developed based on the data acquired from an Iranian oilfield.