DescriptionWe study the operations scheduling problem with delivery deadlines over capacitated multi-echelon shipping networks. Our main results consist of new mathematical models, structural analysis, and solution methodologies for this type of operations scheduling problems, which are in general computationally difficult due to their inherent combinatorial nature. Part I of the dissertation investigates three polynomial-time solvable cases, including case 1) identical order sizes; case 2) designated suppliers; and case 3) divisible order sizes. For the first case, we prove that the original problem can be decomposed into two sub-problems: the transportation problem and a specially structured mixed integer programming model that is totally unimodular. For the second case, we show that the original problem can be solved by the Minimal Spanning Tree algorithm that runs in polynomial time. The third case is shown to be solvable in polynomial time by extending the literature results for a special case of the well- known bin packing problem. Part II of this dissertation analyzes the structure properties of the network scheduling problem with a single processing center (PC) between the suppliers and customers. A dynamic programming-based search algorithm that correctly identifies the optimal subset of customer orders to be fulfilled under each given utilization level of the PC capacity is proposed. We also prove that the resulting search algorithm converges to the optimal solution within pseudo-polynomial time. Part III of the dissertation focuses on the methodology of solving the general operations scheduling problems with customer delivery deadlines. We propose a linear programming relaxation-based algorithm. With this algorithm, a given network scheduling problem is solved through an iterative process. During each iteration, a threshold parameter is used to select the relaxed linear variables to be binary variables for the next iteration, while a subset of binary variables is still relaxed to bounded linear variables. The iteration continues until the values of all the binary variables are determined. This partial relaxation allows us to avoid dealing with the generalized knapsack problem, a difficult NP-hard problem, in the solution process.