Comparison of Drilling Rate Penetration Prediction Based on Deep Learning Approaches, A Volve Dataset Use Case

Abstract

Abstract Efficient drilling operations in the oil and gas field is an important area that can lead to major cost and hazard reduction. One of the key parameters for drilling optimization is predicting the rate of penetration. The penetration rate depends on the physical process which contains variables or features that will affect the values. Using these features, it is possible to predict the penetration rate more accurately during the drilling operation. In this study, we propose comparison of deep learning models between models based on deep recurrent neural network and transformer to predict penetration rate. The result shows that the transformer model outperforms the other models.