PHX ModelCenter Framework Integrates AUV Development in Conceptual Design Stage
Challenge: Multidisciplinary Design Optimization for Autonomous Underwater Vehicles
Researchers at Virginia Tech recognized that current design techniques for Autonomous Underwater Vehicles (AUVs) are stovepiped into separate disciplinary processes1. Design decisions made in one discipline constrain choices in others, which prevents creating a globally optimal design. Why not integrate processes for all AUV disciplines at conceptual design, as is already done for aircraft and manned ships?
Solution: Integrate Design Models for Multiple Disciplines in ModelCenter
Several disciplinary modules were created for an AUV to be used for oceanographic studies. These largely consisted of low-fidelity MATLAB models that were quick to execute. An input module stored design variables and fixed parameters. An electronics module computed electrical power needs by summing individual component loads and generating a total maximum load. A hull geometry module computed an axisymmetric hull form, along with a stacking algorithm to arrange payload items within the vehicle’s volume. A resistance module calculated skin friction and the power required to move the vehicle. A feasibility module computed whether constraints satisfied the model. A cost module gathered costs of components and materials, largely as a function of weight or volume. A risk module determined overall risk by performance, cost, and schedule for each component in the model. An effectiveness module calculated the OMOE for all performance attributes based on US Navy UUV Master Plan guidelines. Weighting for risk and effectiveness were determined by expert opinion and an analytic hierarchical process for decision making across competing alternatives. All was integrated using ModelCenter and the Darwin optimizer based on a genetic algorithm. Discrete variables were configurations (groups of components that formed a subsystem). Continuous variables were metrics such as vehicle diameter, length to diameter ratio, and shape coefficients. Five continuous and six discrete variables were optimized. Prior to optimization runs, a set of design parameters fixed the vehicle’s general operating environment.
Benefit: An Integrated Environment for Optimizing Future AUV Designs
Results, presented in ModelCenter’s Data Visualizer, showed the Pareto optimal designs from cost, risk, and effectiveness views. A CAIV analysis revealed where a small increase in cost would deliver a large increase in effectiveness. This proof of concept showed how an AUV design process could be more effective when all disciplines are integrated upfront. The modular nature of this model and the transparency of its implementation make changes to it straightforward and the addition of new modules easy.
1 Martz, M., Neu, W.L., "Multi-Objective Optimization of an Autonomous Underwater Vehicle", Marine Technology Society Journal, Vol. 43, No. 2, Spring 2009, pp. 48-60.