| In this dissertation we study logistics scheduling and network design problems. In a logistics problem, decision makers need to take cost, service level and environmental effects into consideration. We study the following problems from these aspects.;We first consider transportation consolidation problems on a single lane with service time window constraints, where cost savings can be achieved by adopting consolidation strategies. We formulate the problem as integer programs and show that it is NP-hard. For the constant truck capacity case, we derive the convex hull of the solutions. In addition, if the per-truck per-day price is constant, we present an optimal policy and show the worst case ratio of 2 for the send-when-full policy, which is commonly used in practice. We also conduct an industrial case study, which shows that the send-when-full policy can be efficient in practical situations. Next we study the two-mode transportation problem, where an accelerated transportation mode is available by paying a higher cost. We show several special cases of the two-mode consolidation problem is polynomially solvable.;In the second part of the thesis, we study a logistic network design problem where both cost and emissions need to be controlled. Logistics to remote pristine locations such as within the arctic circle require low emission operations. We focus on the long term economic assessment of research sites in Greenland and study the strategic supply chain design problem as a bi-objective optimization problem, where both cost and carbon footprint must be controlled. We model the problem as a time-spaced multi-commodity network flow problem with inventory tracking. Using forecasted demands as input, we solve the problem based on an integer programming formulation.;To solve the multi-year model efficiently, we decompose the problem using an approximate dynamic programming (ADP) algorithm based on a linear approximation of the value function. To establish the efficient frontier between cost and emissions, we propose an ADP based two phase algorithm. We compare our ADP based algorithms with traditional integer programming based methods. Computational results show that the cost-minimizing ADP is more efficient than directly applying the IP solver, and the ADP based two phase algorithm is more efficient and robust in determining the efficient frontier than IP based methods. We also show how these algorithms can provide guidance for the tactic level operations of the logistics network. |
