We were honored to have Shobeir Pirayeh Gar from Halliburton Company present at the SIMULIA Americas Users Conference, May 1-2, 2024 in Novi, Michigan.

Abstract

Engineering design optimizations have been traditionally done using physics-based simulations, such as finite element analyses, where response of a system is determined through a set of partial differential equations satisfying the laws of physics. As the system becomes more complex with nonlinear behavior and highly coupled response parameters, exploring the entire design space by physics-based simulations may not be computationally cost effective. In this case, data-driven models can be added to the simulation workflow to enhance the efficiency and robustness of the optimization analysis.
This paper presents the optimization analysis of packers as one of the major downhole tools, where finite element simulations are conducted to generate discrete physics-based data points in the design space following Latin Hypercube Sampling (LHS) method. Neural network analysis is performed using the physics-based training data to build a surrogate model covering the feasible design space. A multi-objective performance function is established based on critical response parameters of the packer system for optimization. Numerical optimization analysis techniques such as genetic algorithm are used to find the optimum design. 
The results show the predictions of the surrogate model are found within 5% proximity of the finite element analysis. Examples of successful design and deployment of packers using the above method are discussed. Abaqus and Isight were respectively used for physics-based and data-driven simulations as part of the SIMULIA products.

Presenter Bio

Dr. Shobeir Pirayeh Gar is a Technical Advisor at the Halliburton Technology Center. His areas of expertise lie in structural and computational mechanics, engineering simulation and data science, and scientific machine learning.
He earned his PhD in Structural Engineering from Texas A&M University and is currently a Professional Licensed Engineer in the state of Texas. For the past 12 years, he has been involved in analyzing complex structural and mechanical systems in the Oil and Gas industry ranging from deep-water floating platforms to downhole tools.
Throughout his career, Dr. Gar has championed the use of simulation-guided detailed design engineering. In recent years, he has focused on leveraging the combined power of physics-based and data-driven approaches for robust design optimization and product developments. He is the holder of several US Patents.