Transportation network design involves deciding which roads to build to form the transportation infrastructure and which roads to upgrade to efficiently increase the capacities of the network. Traffic assignment models are needed as an essential part of network design procedures. Most network design models ignore the stochastic and dynamic aspects of traffic demand in the assignment part. These aspects are considered in this thesis.

A logit-based stochastic traffic assignment model is explored and new algorithms for solving stochastic static traffic assignment problems are developed. A new branch and bound approach for solving discrete network design problems with stochastic static traffic assignment is developed. This is the first time that stochastic traffic assignment is included in a network design model.

A new traffic assignment algorithm, using the method of successive averages, is developed for dynamic traffic assignment. Theoretical analysis of this algorithm in a simplified network is provided. A branch and bound approach for solving network design problems with dynamic traffic assignment is developed. By incorporating the dynamic traffic assignment model into a network design model, one is able to consider peak hour traffic situation in solving network design problems.

Traffic assignment and network design models with time-varying demands are extended to formulate a parts routing problem and a system design problem in a manufacturing system. The manufacturing problems are solved by the techniques developed in this transportation study.