Darshan Sarojini "Advancing Aircraft Design through Scientific Computing"

2:00 pm
Friday, February 2, 2024
310 Kelly Hall
Faculty Host:  Dr. Rakesh Kapania

Abstract:  Aircraft design is a complex process over a long period that involves iterations between the ‘outer loop’ and ‘inner loop’. The outer loop primarily focuses on the overarching concepts, such as selecting between different aircraft concepts, with or without winglets, or deciding upon the type of electric powertrain architecture. These are design activities involving variables that are usually not differentiable. The inner loop, on the other hand, deals often with differentiable variables. For instance, once the choice is made to have or remove the winglet, aeroelastic optimization is done to minimize the drag and structural weight. In this talk, I will present an overview of my research journey, focusing on key advancements and future directions in the realm of aircraft design. The presentation is structured around three main themes: 1. Innovations in Early-Stage Aircraft Design: Physics-based tools and modern scientific computing methods have become integral to aircraft design. However, their conventional application at later stages in the design process poses challenges. I will discuss gradient-based Multidisciplinary Design Optimization (MDO), showcasing recent advancements that enable large-scale MDO. Furthermore, I will explore diverse design studies utilizing gradient-based MDO as a sub-problem, ranging from wingbox structural topology design to certification-driven design of the aircraft empennage. 2. Overcoming Design Challenges using Graph-Based Methods: I will address a key limitation in MDO: obtaining derivatives automatically. Proposing to overcome this challenge using computational graphs, I will present preliminary studies of accelerating computations on modern high-performance computing platforms such as GPUs and TPUs. Additionally, I will highlight another limitation in MDO, emphasizing the lack of a systematic design process and traceability. I propose a novel approach utilizing RFLP-based Model-Based Systems Engineering (MBSE) formulated as  knowledge graphs. These graphs not only enhance traceability but also unlock features beneficial to aircraft design, including post-optimality sensitivity analysis, uncertainty quantification, and decision-making capabilities. These methods will integrate outer-inner-loop considerations with MDO. 3. Integrating Computational Approaches into Curriculum and Certification-Driven Design: My research vision at Virginia Tech extends to practical application and education. I aim to integrate the computational methods developed into the curriculum, particularly in my senior design class. Students will actively employ these tools to design aircraft, culminating in demonstrating compliance with certification regulations virtually in a flight simulator. Additionally, design-build-fly teams will construct and fly aircraft prototypes, serving as a feedback loop for refining computational models and addressing disparities between flight tests and simulations. This integrated approach not only fosters a new generation of aerospace engineers with a robust blend of computational expertise and practical experience but also marks a significant step towards certification-driven design in a virtual environment.

Bio:  Darshan Sarojini is a computational aircraft designer focusing his research on utilizing scientific computing methods for design. He holds a bachelor’s degree in mechanical engineering, a master's degree in computational science and engineering, and a Ph.D. in aerospace engineering. His current role is as a postdoctoral scholar at the Large-Scale Design Optimization Lab at the University of California San Diego. Dr. Sarojini’s key contributions are in multidisciplinary design optimization (MDO) and model-based systems engineering (MBSE). His work aims to lower aircraft design cycle time at the conceptual and preliminary design stages, particularly for sustainable aviation aircraft concepts, encompassing urban air mobility, regional air mobility, and long-range transport. Actively engaged in collaborative projects, Dr. Sarojini has contributed to funded initiatives involving both government entities such as NASA and industry partners including Aurora Flight Sciences, Airbus, Siemens, and Gulfstream. His service roles include membership in the AIAA Modeling and Simulation Technologies technical committee since 2022, committee leadership in the SuperComputing conference since 2021, and a reviewer for various journals in the aerospace engineering and scientific computing domains.