NCS Drilling Data Based ROP Modelling and its Application

Abstract

Predicting the ROP (rate of penetration) before drilling is essential to improve the overall drilling efficiency and thereby reduce the non-productive time and costs. For this, good ROP model is required. Several ROP models are available in the industry which is derived based on both mechanistic and empirical methods. However, each model has its strengths and shortcomings. In this thesis, five ROP modelling workflows are used to model field data. These are the methods of multiple regression, least squares, MSE, D-exponent and the Warren model. The applicability and the limitations of the models are tested on nearby, distant and very distant field data in the North Sea. Six wells are used for the analysis located in the Alvheim, Kvitebjørn and Valhall fields. During modelling, two hypotheses were tested. One common observation from the result is that the model application is limited within the same block, where the lateral geology expected to be similar. Based on the assessment of the results, applicability and limitations of the five modelling techniques are summarized. In addition to the overall study, an optimization procedure is also developed.