Performance of successively complex rules for locational decision-making

The relative performance of five simple rules for selecting facility locations are compared with two heuristic location-allocation algorithms in the context of a real region. It is hypothesized that the performance of a rule is likely to be systematically related to the spatial structure of the region in which it is used. In the Iowa settlement system, known to have predictable structural regularity, some rules performed over 95 percent of the level achieved by the algorithms (the greedy add and the modified vertex exchange algorithms) on a range of commonly used criteria for evaluating locational efficiency. Under certain circumstances, these rules may be recommended as alternatives to more formal algorithms.