Methods and Tools
Multidisciplinary Design Optimization (MDO)
Gradient-Based Aircraft Sizing Framework
Modular functional-form approach that enables it to design conventional and unconventional aircraft, fuel-based and novel electrified propulsion architectures.
A solver-independent field representation that makes it easy to integrate with other tools.
Automated gradient computation language named CSDL makes gradient-based optimization ideal when dealing with hundreds of design variables.
Select Publication
Sarojini, D., Ruh, M. L., Joshy, A. J., Yan, J., Ivanov, A. K., Scotzniovsky, L., ... & Hwang, J. T. (2023). Large-Scale Multidisciplinary Design Optimization of an eVTOL Aircraft using Comprehensive Analysis. In AIAA SCITECH 2023 Forum (p. 0146). [Download from ResearchGate]
Slides
evtol_optimization.mp4
Rapid Airframe Design Environment (RADE)
Rapid geometry manipulation of conventional and unconventional aircraft.
Modification of the outer mold line (OML)
Automated wingbox structural layout
API to interface with physics-based solvers
AVL for low-fidelity aerodynamics
Nastran for structural and aeroelastic analysis
Hypersizer for structural sizing
Select Publication
Sarojini, D., Solano, H. D., Corman, J. A., & Mavris, D. N. (2022). Parametric Wingbox Structural Weight Estimation of the CRM, PEGASUS and Truss-Braced Wing Concepts. In AIAA AVIATION 2022 Forum (p. 4054). [Download from ResearchGate]
Slides
Dynamic Environment for Loads Prediction and Handling Investigation (DELPHI)
6 degrees of freedom flight dynamics simulation tool
API allows for any aerodynamic, propulsion, and mass properties model to be provided
Define aircraft with an arbitrary number of propulsion devices and control surfaces
Closed-loop control system
Emphasis on prediction of regulation specified dynamic maneuvers and handling qualities
Select Publication
Sarojini, D., Harrison, E., & Mavris, D. N. (2021). Dynamic Environment for Loads Prediction and Handling Investigation (DELPHI). In AIAA Scitech 2021 Forum (p. 0326).[Download from ResearchGate]
Slides
SciTech 2021 DELPHI; Checked Pitch Manuever; Rudder Kick Maneuver; Roll Maneuver
Model Reduction for Structural Weight Estimation
Beam Models for Aircraft Wingbox Structures
The research involves exploring physics-based model reduction to create a simplified beam model representation of the aircraft wing primary structure.
Developed using the modular OpenMDAO platform and CasADi to automate derivative computations.
Consider arbitrary cross-sections, dynamic loads, both strength and buckling structural constraints, and can model multiple beam members connected by joints, as well as the addition of multiple masses and point loads.
The scientific contribution of the research lies in the development of a one-time correction using the higher-order Variational Asymtotic Method (VAM).
Weight predicted by the beam model compared to higher-fidelity shell models was reduced to < 5% difference with a 12x speed-up, even for complex aircraft configurations like the truss-braced wing.
Select Publication
Sarojini, D., & Mavris, D. (2022). Structural Analysis and Optimization of Wings Subjected to Dynamic Loads. AIAA Journal, 60(2), 1013-1023. [Download from ResearchGate]
Slides
Multi-Fidelity Reduced-Order Modeling (ROM)
The research applies a recently developed parametric, non-intrusive, and multi-fidelity ROM method on high-dimensional displacement and stress fields arising from structural analysis
Leverages manifold alignment to fuse inconsistent field outputs from high- and low-fidelity simulations by individually projecting their solution onto a common subspace.
Outputs from structural simulations using incompatible grids, or related yet different topologies, are easily combined into a single predictive model, thus eliminating the need for additional preprocessing of the data.
Achieves a relatively higher predictive accuracy at a lower computational cost when compared to a single-fidelity model.
Select Publication
Perron, C., Sarojini, D., Rajaram, D., Corman, J., & Mavris, D. (2022). Manifold alignment-based multi-fidelity reduced-order modeling applied to structural analysis. Structural and Multidisciplinary Optimization, 65(8), 236. [Download from ResearchGate]
Slides
Model-Based Systems Engineering (MBSE)
Requirements-Functional-Logical-Physical: RFLP Approach to Integrate MBSE and MDO
Extended RFLP: Extended the conventional RFLP framework to include two components: a product RFLP for the physical system and a novel process RFLP specifically designed for simulations, numerical models, and direct integration with MDO.
Unambiguous RFLP Rules: Created clear rules for mapping requirements, functions, and logical components to physical and process representations.
Clear Delineation: Separated MBSE development from MDO development, improving system model clarity and maintainability.
Select Publication
Swaminathan, R., Sarojini, D., & Hwang, J. T. (2023). Integrating MBSE and MDO through an Extended Requirements-Functional-Logical-Physical (RFLP) Framework. In AIAA AVIATION 2023 Forum (p. 3908). [Download from ResearchGate]
Slides
Graph-based Methods for MBSE and MDO
The research direction focuses on treating the MBSE model as a knowledge graph, similar to how MDO models are treated as computational graphs. This approach enables a graph-based link between the two models, leading to the following potential benefits:
Post-optimality sensitivity analysis: Analyze system response sensitivity to requirements after optimization. This allows for a deeper understanding of how changes in requirements impact the system's performance and enables informed decision-making.
Automated MDO model generation: By leveraging a library of discipline models, the proposed method aims to automate the generation of MDO models.