Application of Critical Drawdown Pressure Prediction in Completion Design to Minimize Sanding in a Clastic Gas Reservoir in Saudi Arabia

Abstract

A clastic gas reservoir with unconsolidated sandstone layers present great challenges for an effective development, because the tendency of these layers to produce sand. The objective of this paper is to present and highlight the applications of geomechanics in predicting critical drawdown pressure during the completion design and flowback test design with the ultimate purpose of minimizing the sand production. This paper will evaluate the perforation strategy for wells that may be prone to produce sand as part of the completion design optimization. A geomechanical approach was implemented to evaluate the interaction between stress field and the mechanical properties of rocks. A 1-D Mechanical Earth Model (MEM) was built and calibrated with offset wells in the nearby area. The overburden pressure was calculated by integrating density logs from the offset wells. The minimum horizontal stress was calibrated using closure pressure derived from the offset fracture analysis carried out in the offset wells. The rock elastic properties were calibrated with lab test data from an offset well locate ~2 km away from Well_A. Two case studies will be presented in this paper. The first case is Well_A drilled in a Devonian clastic reservoir, this vertical gas well was perforated with 60 degrees phasing guns. The well had a good performance during the flow back but the production was short-lived due to significant sand production. The second case is a blind test to validate the robustness of the methodology used in the first case study. An integrated approach was used to determine the most optimum way to perforate similar wells that has a potential to produce sand. Modified Lade failure criterion was used to predict the critical drawdown pressure because it takes into account the intermediate stress along with other geomechanical properties.

Predicting Onset of Sand Production in Oil Wells using Machine Learning

“…It is the difference between the minimum well bottom hole flowing pressure and reservoir pressure. To forecast the CTD, some studies employed analytical models such as Mohr-Coulomb and modified Lade; however, the models contain certain assumptions, such as the mechanical properties of the rock formation are homogeneous and isotropic, making them lack accuracy [7,8]. Other models such as artificial neural networks (ANNs), feed-forward backpropagation networks (BPN), generalized regression neural networks, multiple linear regression (MLR), and the genetic algorithm MLP (GA-MLR) have been applied to predict CTD, but these models are proven to lack accuracy [4].…”