Multidisciplinary Design Optimization

Reduced Order Modeling

Complex engineering design and analysis problems involving accurate computational models and algorithms for their solutions are limited by the capabilities of computing systems and the associated demand for enormous computational resources. Any modeling effort that leads to a reduction in the computational resources requirements without compromising accuracy is an invaluable asset to the industry. Reduced Order Modeling (ROM) or surrogate modeling strives to meet these goals. A very practical application of ROM is structural health management and prognosis where fast and inexpensive data is necessary for onboard information management.

Aeroelastic analysis involving nonlinear luid dynamic phenomena is a complex task. The coupling of high fidelity Computational Fluid Dynamics (CFD) codes and Computational Structural Dynamics (CSD) tools to address aeroelastic problems has received considerable interest in the recent years. However, direct incorporation of a CFD code into a CSD tool to solve aeroelastic problems leads to high computational cost. Ignoring nonlinear behavior of the structure or fluid in order to reduce the computational effort will usually lead to inaccurate predictions of aeroelastic response with the associated impact on the design parameters. The solution to this problem is the development of aerodynamic reduced order models (ROMs) with (i) only relevant states; (ii) meaningful description, and (iii) computational efficiency.

A methodology has been developed for identification of computationally efficient ROM for aeroelastic problems using Volterra based kernels. Some of the benefits offered by the developed ROM identification technique: (i) computational efficiency (about four orders of magnitude faster); (ii) can model the type of nonlinearities commonly present in an aeroelastic analysis; (iii) has a format suitable for multidisciplinary applications; (iv) converges in a single iteration. Validations conducted on the AGARD wing identify Mach 0.96 to be the most unstable transonic regime. Research sponsored by NASA Langley Research Center.


Analysis and Simulation of Turbulent Flows

Numerical simulation, analysis, and modeling of single- and multi- phase turbulent flows have acquired an important role in a number of engineering and scientific disciplines. Numerical simulation is being used as a tool, where appropriate, to understand fundamental aspects of fluid systems and to develop and assess methods for practical applications. Specifically, research is being pursued on multi-phase (gas-solid) flows, turbulent boundary layers, heat transfer in gas turbines, and massively separated turbulent lows. Several hierarchies of numerical simulations of the Navier-Stokes equations are being performed, including Direct Numerical Simulation (DNS), Large Eddy Simulation (LES), Detached-Eddy Simulation (DES) and Reynolds-averaged Navier- Stokes (RANS) computations.