Collaborative optimization: an architecture for large-scale distributed design

Collaborative optimization is a design architecture specifically created for large-scale distributed-analysis applications. In this approach, a problem is decomposed along domain-specific boundaries into a user-defined number of subspaces which are driven towards interdisciplinary compatibility and the appropriate solution by a system-level coordination process. This design approach allows domain-specific issues to be accommodated by disciplinary analysts, while requiring interdisciplinary decisions to be reached by consensus. In a large-scale practical design environment, this scheme has several advantages over traditional solution strategies. These advantageous features include reducing the amount of information transferred between disciplines, the removal of large iteration-loops, allowing the use of customized optimizers within the domain-specific subspace analyses, a solution framework that is easily parallelized and operable on heterogeneous equipment, and a structural framework that is well-suited for conventional disciplinary organizations. In this dissertation, the fundamental concepts leading to the development of the collaborative architecture are presented and the architecture's mathematical foundation is discussed. The design architecture is shown to be applicable to any set of arbitrarily-connected analyses, regardless of the interdisciplinary coupling structure. Example applications in trajectory optimization and launch vehicle design illustrate the the architecture's potential for use in large-scale design applications. Applied in a multidisciplinary design environment, numerous operational advantages of this optimization scheme are demonstrated. These advantageous features are a direct result of empowering the subspaces in the domain-specific decision process, thereby distributing design authority as well as analysis responsibility. While applicable to any problem, the characteristics of the collaborative optimization architecture are shown to be best suited for large-scale, highly-constrained, distributed-analysis applications. Because such problems are common in practical design settings, the collaborative optimization architecture should provide design teams with an intriguing alternative to current practices.