Wear-Factor Prediction Based on Data-Driven Inversion Technique for Casing Wear Estimation

Abstract

Abstract Wear factor is an important parameter for estimating casing wear, yet the industry lacks a sufficient data-driven wear-factor prediction model based on previous data. Inversion technique is a data-driven method for evaluating model parameters for a setting wherein the input and output values for the physical model/equation are known. For this case, the physical equation to calculate wear volume has wear factor, side force, RPM, tool-joint diameter, and time for a particular operation (i.e., rotating on bottom, rotating off bottom, sliding, back reaming, etc.) as inputs. Except for wear factor, these values are either available or can be calculated using another physical model (wear-volume output is available from the drilling log). Wear factor is considered the model parameter and is estimated using the inversion technique method. The preceding analysis was performed using soft-string and stiff-string models for side-force calculations and by considering linear and nonlinear wear-factor models. An iterative approach was necessary for the nonlinear wear-factor model because of its complexity. Log data provide the remaining thickness of the casing, which was converted into wear volume using standard geometric calculations. A paper [1] was presented in OMC 2019 discussing a method for bridging the gap. A study was conducted in this paper for a real well based on the new method, and successful results were discussed. The current paper extends that study to another real well casing wear prediction with this novel approach. Some methods discussed are already included in the mentioned paper.