A comparative analysis of artificial neural network and rock physics for the estimation of shear-wave velocity in a highly heterogeneous reservoir

Verifiable and accurate estimation of shear wave velocity (VS) in a well with no previous exposure of velocity profile is the bottom line for any technique to declare the VS estimation capabilities. So far all the available models and techniques for VS estimation have been attempted on well logs and laboratory based core measurements. This study aims to compare the role of virtual measurement using Artificial Neural Network (ANN) and rock physics analysis using Gassmann equation for the estimation of VS in a highly heterogeneous reservoir. The techniques were applied initially to the well containing specialized logging and laboratory analysis data and prediction of VS models were developed. The developed model was then applied to the offset well whose data was not included in the model development and the results were compared with measured VS data. The results show that although rock physics model provides acceptable results that show consistency in following the actual trend in shear and compressional wave velocities estimation it lacks a consistent generalization power, meaning that it does not work as well with the data of unstable zones. On the other hand, ANN provides more realistic results because of its discriminatory power and observe the actual trend in VS estimation even at unstable zones. Presentation Date: Wednesday, October 17, 2018 Start Time: 9:20:00 AM Location: Poster Station 1 Presentation Type: Poster