Parallel allocation-mission optimization of a 128-route network

In aircraft design, a simultaneous optimization of the airline allocation, mission profile, and the airframe-propulsion design can be advantageous because it can capture how design changes might affect the aircraft utilization at the airline level. As a first step towards this goal, previous research solved the allocation-mission optimization problem by taking a modular adjoint approach, but with only a 3-route network to simplify the problem. This paper extends the previous work to solve a 128-route allocation-mission optimization problem using a parallel computational framework. The computation of the constraint gradients can be parallelized by extending the framework to support linear algebra with multiple right-hand sides. The 128-route optimization problem contains roughly 6,000 design variables and 23,000 constraints, and it converges about 3 orders of magnitude in optimality and feasibility in approximately 8 hours on 128 processors.