SBIR-STTR Award

The main objective of this project is to develop an efficient stochastic-optimization approach that can be applied to large-scale high-complexity Army military vehicle applications in a high-speed distributed computer hardware environment. Specifically, the proposed work will integrate the refined stochastic response models (stochastic field expansion approximations) developed by GP Technologies Inc. with the RBDO Probabilistic Measure Approach (PMA) developed by the University of Iowa. The research project team will explore and refine the stochastic response analysis part of the UOI RBDO approach based on state-of-the-art stochastic field/network modeling and artificial intelligence concepts. The proposed development will also include fast Dynamic Monte Carlo simulation algorithms, plus Bayesian inference for optimal model fitting and selection. Phase I will also provide a probabilistic-design methodology for incorporating the data and modeling uncertainties in the RBDO problem and will explore the application of other non-deterministic approaches to RBDO including multi-objective problems. In Phase I the proposed RBDO methodology will be applied to the Army M1A1 tank road arm component via ANSYS FEA code. The RBDO approach will operate in the distributed environment of hardware available at the U.S. Army. The parallel programming task will be conceptually initiated in Phase I and completed in Phase II.

Benefits:
The benefits for Army are: (1) reliability estimation of failure modes, (2) design optimization and validation through a non-deterministic analysis including probability/ possibility/ evidence analysis and (3) design improvement with respect to reliability and cost. It is highly expected that the proposed RBDO approach will have a great demand for automotive, aerospace and defense industries.

Keywords:
reliability, stochastic fields, optimization, MCMC, reliability-based design, RBDO, vehicles

With the current tight monetary constraints placed on the US military there is a continuous need to reduce the cost of ownership of the Army ground vehicles. The consideration of uncertainty and design optimization is critical for designing high-performance, reliable, cost-effective Army combat vehicles. The proposed work develops an adequate computational tool for accurately performing military vehicle reliability-based design optimization (RBDO) using large-scale finite element models in distributed computing environments. To approach vigorously the RBDO problem for the Army vehicles, GP Technologies has teamed with University of Iowa CCAD and University of California at Berkeley. Under this project a new, powerful computational stochastic mechanics technology is developed and then is integrated with the RBDO software developed by University of Iowa CCAD. The proposed computational stochastic finite element technology includes two key components: (i) an accurate stochastic approximation of highly complex physical behavior using high-order stochastic field models and (ii) a fast physics-based stochastic simulation using a parallel multigrid solver for unstructured mesh problems. The new computational stochastic tools will be demonstrated on critical components of the Army High-Mobility Trailer and the Stryker. The resulted RBDO technologies will enable TACOM to design more cost-effective high-performance, durable & affordable military vehicles.

Keywords:
Reliability, Optimization Under Uncertainty, Stochastic Fields, Finite Elements, Stochastic Approximation, Stochastic Simulation, Design Optimization