The U.S. commercial transportation network is heavily congested, and is rapidly approaching gridlock. There are limited resources available to alleviate congestion. There are a number of possible solutions to the problem of gridlock such as designing new aircraft, and expanding airport capacity. However, it is unlikely that any single action will be sufficient to solve the entire problem. One possible solution is to design and build a high-speed rail (HSR) infrastructure to alleviate the demand on the existing commercial air and ground transportation networks. Regardless of the infrastructure venue, decisions such as whether to invest in HSR or air transportation capacity expansions must be made under uncertainty in the future passenger transportation demand and true behavior. Problems of this type are multi-stage stochastic mixed integer nonlinear programming problems (MINLP) for which no scalable solution strategy exists. As such, the research objective of this proposal is to develop a novel decomposition method for multi-stage stochastic MINLP and apply this methodology to multi-modal intercity commercial transportation design and routing to include commercial air, HSR, and ground transportation.
This research project will purse the following objectives: (1) system analysis to include defining link performance, passenger mode choice, user equilibrium, and aircraft design and allocation, (2) surrogate modeling usage to approximate the optimal solution surface space using two-stage stochastic programming, (3) surrogate modeling usage to approximate the optimal solution surface space using approximate dynamic programming, (4) expand the theoretical contributions of objectives two and three and apply them to the design and usage of commercial passenger transportation in the California Corridor.