To provide services, industries and the government rely heavily on their ability to efficiently transport goods. In their simplest form, transportation problems can be solved efficiently using traditional optimization algorithms. However, practical applications often exhibit a variety of additional combinatorial requirements that necessitate involved modeling, often in the form of integer programs. This modeling is often seen as unnatural by practitioners and results in models that are difficult to solve with state-of-the-art commercial software. The research objective of this award is to develop a new modeling and solution paradigm for such transportation problems. In this paradigm, a few key constructs are identified that allow the user to succinctly impose often-occurring combinatorial constraints without the need to formulate them using integer programming. These constructs can then be exploited to design more efficient exact and heuristic solution methodologies that take advantage of the structure of the network and of the combinatorial requirements. The research will also seek to demonstrate the advantages of this paradigm on practical problems emanating from bulk delivery in railroads.
If successful, this research will result in an integrated modeling and solution paradigm that will provide faster solutions to larger transportation problems and that will bring within the range of tractability many practically relevant problems that are currently intractable. It will lead to solutions where transportation costs and miles are reduced and will therefore improve the efficiency of industries relying on transportation. Results will be disseminated timely via conferences, publications and a dedicated internet page that will contain relevant models, stylized applications and prototype implementations. Undergraduate and graduate students will also benefit from this award through the development of course material and involvement in research.
