Improved road network design models with the consideration of various link patterns and road design elements

Abstract

The success of an automatic road network layout over steep terrain mainly depends on the model design. Most previous models have used a grid representation that considers only eight adjacent cells when evaluating feasible road links. Here, we present improved models and alignment constraints mapped on a mathematical graph for better designs that are more applicable under field conditions. We have refined the link pattern by considering up to 48 neighbouring cells and have introduced 16 directional classes per grid cell. Optimization techniques, such as shortest path, minimum spanning tree, and Steiner minimum tree algorithms, are used on the graph to derive a road network that is optimal in terms of its construction costs. These improved models have been applied to different mountainous project areas. Our results show that, by considering various link patterns and alignment constraints, one can determine more appropriate and cost-effective locations for road networks, especially in steep terrain.