Good understanding of the mechanical behavior of reservoir rock is very important in reducing the problems related to wellbore stability, sand production and reservoir subsidence. To carry out any operation, a continuous profile of rock mechanical parameters is needed. Retrieving reservoir rock samples throughout the depth of the reservoir and performing laboratory tests are extremely expensive and time consuming. Therefore, these parameters are estimated from the sonic and compressional wave velocities obtained from well-logs. Parameters obtained from laboratory tests are termed as static parameters while those obtained from sonic logs are dynamic parameters. The former case represents closely the condition in the reservoir. Since the well-logs provide a continuous profile of parameters, they have to be calibrated with respect to the static parameters. This paper presents a rigorous empirical correlations based on the weights and biases of Artificial Neural Network to predict sonic logs (compressional and shear wave travel times), elastic parameters (static Young's modulus and Poisson's ratio) and failure parameter (Unconfined compressive strength).The testing of new correlations on real field data resulted in less error between actual and predicted values, suggesting that the proposed correlations are very robust and accurate, and can help geo-mechanical engineers to construct representative earth model.
|Original language||English (US)|
|Title of host publication||51st US Rock Mechanics / Geomechanics Symposium 2017|
|Publisher||American Rock Mechanics Association (ARMA)firstname.lastname@example.org|
|Number of pages||11|
|State||Published - Jan 1 2017|