Optimizing Maintenance Operations for Multimodal Transportation on the Inland Waterway System

Abstract

The U.S. inland waterway system has more than 25,000 miles of the maintained navigation channel, which carries a significant percentage of the national freight total (17% of all intercity freight by volume[1]). Maintenance operations, including dredging and lock and dam maintenance & repair, are important to ensuring the effective and efficient operation of the inland marine transportation system. Each year, there is a budget [2] for the maintenance of lock/dam and navigable waterway dredging, etc. This dissertation specifically deals with maintenance fund allocation to these projects. The goal of project selection is to maximize the total network benefits, such as the network shipping capacity. In literature, there are models dealing with the maintenance project selection without considering the resulting shoaling effect. Deep dredging, although costly, brings large benefits. However, these benefits are subject to more severe shoaling effect over the years after. Selecting dredging projects by consideration of the resulting shoaling effect can cause the budget to be used in a much better-informed way and therefore more efficiently. In this dissertation, the project selection is optimized on the network by explicitly considering the stochastic nature of the shoaling. The consideration is conducted through alternative mathematical models. Those alternative models, each described in the following chapters, progress to be increasingly realistic and complex. In Chapter III, we develop a multimodal formulation that minimizes the total cost of shipping commodities in a network after the lock/dam and dredging maintenance. This formulation looks at the stochastic nature of the shoaling in a deterministic way. The findings show the difference in budget allocation towards locks and dams versus dredging operations and compare their effects on the network. In Chapter IV, we compare the results from the formulation in Chapter III with industry practices to gain managerial insights. These practices include the benefit/cost analysis and the through tonnage method. The managerial insights help managers improve their empirical decisions in light of the results from the optimization models. We also show sensitivity analysis results on the dam maintenance costs to see if the results would significantly change. In Chapter V, we analyze the shoaling data along the Ohio River to find a shoaling distribution in different reaches. The purpose of this analysis is to characterize the random shoaling distribution in order to accurately model its effect in a formal stochastic formulation next. We also run some statistical analysis to see which distributions fit better on the data. Chapter V prepares the probability distribution of shoaling for the stochastic model next. In Chapter VI, we propose a two-stage formulation that tries to allocate dredging budgets by considering two consecutive time periods (e.g. year). The budget available to allocate is the budget available in the current period, and the formulation tries to maximize the benefit realized in both the first year and the year after when shoaling has taken place as a result of the year one dredging. Multiple scenarios are developed to capture the stochasticity of the shoaling. The result shows that the dredging decisions are changed significantly by having the knowledge of shoaling rate in the future. In Chapter VII, multiple solution algorithms for the stochastic formulation are summarized. The solution algorithms decompose the formulation into multiple sub-problems and solve them. Benders and L-Shape methods are explained for general cases and for integer formulation.