A study of the impact of a major change in a campus street network began with the collection of link flows before the change. Despite the care with which the link counts were made, conservation of flow at each node was not satisfied. Therefore, the flows through the nodes had to be "balanced." This paper discusses the variety of techniques developed to balance the network. The techniques fell into twocategories: algorithms and mathematical programming formulations. Acoparison was made between these procedures and the maximum-likelihood method advocated in the literature. It became evident that the node-balancing solution depends on the criteria chosen to evaluate the solution, which in turn can offer guidance as to the specific method to choose or develop. This paper appears in transportation research record no. 1220, Forecasting.
Abstract