Exploring the significance of design variable choice on solution quality, convergence and robustness

In the design process of large engineering systems, the system has to be divided into various subsystems because of the multidisciplinary nature of the problem. There are various frameworks which model the communication and co-ordination between these subsystems. Usually, the resource sharing between various subsystems is represented as shared design parameters. In a Multidisciplinary Design Optimization (MDO) framework, the choice of system level design variables and the way in which a problem is subdivided into subsystems affects the overall design and synthesis of systems. In this research the convergence time, quality and robustness of the solutions are compared for three different system variable sets in a case study of MDO based satellite design.;The satellite design problem is simplified and divided into four basic subsystems-Payload, Orbital, Power and Propulsion. A deterministic robust design approach is applied to verify robustness of solutions. An improvement in the original robust design approach is suggested in order to overcome its shortcoming. The overall aim of these three case studies is to demonstrate the significance of system design variable choices on MDO solutions when robustness of the solutions is emphasized.