Integrated Modelling and Simulation of Wellbore Heat Transfer Processes through High-level Programming, Sensitivity Analysis and Initial Approach with Machine Learning Predictive Models

Abstract

During designing of downhole systems and selecting equipment and materials, engineers must consider in-situ conditions before taking decisions in order to be able to handle any operation in a safe and adequate manner. The well temperature profile is mainly imposed by the formation temperature; however, this can vary significantly during operations in different ways. Several properties of fluids and pipes would take part of the heat transfer process such as flow rates, specific heat capacities, thermal conductivities, densities and viscosities. This work gathers and implements various complementary models to simulate the heat transfer across the wellbore during drilling, production and injection. The entire wellbore is discretized, and the models are solved numerically by applying the Crank-Nicholson finite differences method for two dimensions. All the calculations are programmed in python and are released as an open source repository. Besides, a sensitivity analysis is performed for the three main operations (drilling, production and injection), describing individual effects of the parameters on the temperature variation. In addition, prediction models are developed from true and simulated data. These are presented in detail from the data acquisition up to the model assessment. Thus, their performances are compared with the physics-based models, where accuracy, simplicity and computing time play a key role within the engineering tasks; specially when analyzing numerous wells and/or conditions.