Research Projects

This research thrust develops advanced frameworks for the probabilistic design of engineering systems under uncertainty, nonlinearity, and discontinuous behavior. At the PREDICT Laboratory, we have developed novel domain-decomposition strategies combining spectral clustering and support vector machines to enable accurate global sensitivity analysis of non-deterministic structural responses. His work integrates uncertainty quantification, surrogate modeling, active learning, and stochastic optimization to achieve robustness-driven design. The lab further leverages deep learning to build data-informed predictive models that accelerate reliability assessment and design of aerospace composites, lattice structures, and energy systems.

This research focuses on the uncertainty-aware design and experimental validation of Variable Angle Tow (VAT) composite laminates manufactured via Automated Fiber Placement (AFP). At the PREDICT Laboratory, we develop B-spline and NURBS-based fiber path parameterizations for smooth, manufacturable tow steering using automated robotic systems. The framework explicitly accounts for manufacturing-induced uncertainties and defects—including tow gaps, overlaps, fiber waviness, and thickness variations—and integrates probabilistic analysis with robust optimization. Experimental validation is conducted to quantify structural performance and reliability, bridging intelligent design, manufacturing variability, and next-generation aerospace composite applications.

This research thrust investigates architected lattice structures and meta-materials for lightweight, high-performance structural applications. We explore a range of lattice topologies, with particular emphasis on Triply Periodic Minimal Surface (TPMS) architectures, using a parametric and automated design framework that integrates MATLAB with ANSYS APDL for geometry generation, nonlinear finite element analysis, and high-throughput simulation. The study examines the influence of temperature-dependent material behavior—specifically for Inconel 718, Ti-6Al-4V, and AlSi10Mg—on compressive response, energy absorption capacity, and densification characteristics. In parallel, we develop functionally graded lattice configurations and employ homogenization-based approaches to estimate effective material properties, enabling computationally efficient optimization and reliability assessment under material and geometric uncertainties.

This research thrust focuses on the uncertainty-aware aero-structural optimization of wind turbine blades to enhance energy production, reliability, and structural robustness. We develop an integrated low-fidelity aero-structural framework combining XFOIL for airfoil analysis, Blade Element Momentum (BEM) theory for aerodynamic performance prediction, and finite element analysis using ANSYS APDL for structural response and stress evaluation. The methodology incorporates uncertainty quantification and stochastic optimization to account for variability in material properties, geometric parameters, and wind loading conditions. Using this framework, we have optimized the NREL 5MW reference blade to achieve improved energy production while maintaining structural integrity. Current efforts extend this approach to ducted wind turbine configurations, where coupled aerodynamic–structural interactions and design uncertainties are being systematically investigated for performance enhancement.