Efficient Trade Space Exploration

Two of the principal challenges in efficient trade space exploration are (1) quickly evaluating options, and (2) quickly convincing stakeholders to accept the results of the evaluation. This paper describes the process and tools that have led to a factor of nine improvement in the efficiency of trade space exploration of space systems in Team-X at the Jet Propulsion Laboratory. The principal method that has enabled this increase in efficiency is the separation of the exploration figures of merit into two distinct types, which are then addressed in an efficient order. The figures of merit in a trade space exploration of N subsystems either scale with the number of interactions between subsystems, O(N boolean AND 2-N), or with the number of interactions between the subsystems and the external constraints, O(N). It is more computationally efficient to filter against O(N) figures of merit first, and only proceed to filtering against O(N boolean AND 2-N) figures of merit if warranted, than the other way around. However, the latter is the typical bottoms up design approach in space systems engineering organizations. In addition to the computational efficiencies gained through this pre-filtering of infeasible or non-viable configurations from further work, cognitive efficiencies are also gained by this partitioning of the figures of merit. By partitioning the O(N) external constraints on the system along the border of an N-squared diagram-based dashboard, and the O(N boolean AND 2-N) of interfaces between subsystems within the dashboard, stakeholders gain several insights: (1) how the external boundary conditions inform the what the optimal internal subsystem choices are, (2) how the internal subsystem choices affect the selectability of the options, and perhaps most importantly, (3) how their own requirements - derived by them from the O(N) external constraints - affect the selectability of the system options.