Pattern Search Algorithm for Blackboard-Based Multidisciplinary Design Optimization Frameworks

Abstract

Preliminary aircraft design necessitates the use of a range of analysis tools, which are often scattered among many departments in an organisation and require regular tuning from skilled operators. For this reason, a distributed Multidisciplinary Design Optimisation approach that permits individual organisational domains to use their preferred analysis and optimisation tools would be most suitable. This paper revisits a Blackboard framework, which uses simple heuristics to automatically guide organisational design domains to a single optimum by narrowing the bounds on the shared design variables. The authors present a newly developed rule base for this legacy framework, which has been given the title "Multidisciplinary Pattern Search". Two examples, one of which is for conceptual transonic wing design, demonstrate the merit of the newly developed rule base, database and visualisation modules. They also serve as a means for comparisons with two established Multidisciplinary Design Optimisation architectures. The results indicate that the Blackboard performs better than the distributed Collaborative Optimisation approach, albeit worse than the monolithic Simultaneous Analysis and Design method that tends to be organisationally disruptive to implement.