In this paper the data collection, analyses and results of three studies in Southern Africa on the effect of road geometry on accident rates are described. These studies used data from different periods for roads in the provinces of the Cape and Natal in South Africa and for roads in Malawi. The studies covered a combined total of 27,000 km of roads and included about 43,000 accidents. All the studies considered cross-sectional elements such as number of lanes, lane width, shoulder width and shoulder surface. In the first study the alignment was defined by indices such as hilliness and bendiness which are determined from the sum of vertical rise plus fall in m/km and the sum of horizontal deflections in degrees/km respectively. In the other two studies the alignment was described in terms of the terrain types of flat, rolling and mountainous. The results of the studies are compared to those of similar studies in the USA. The major difference was found to be in the effect of paved shoulders on the accident rate. In South Africa roads with such shoulders did not yield the safety benefit that was found overseas. This can be ascribed to the fact that these shoulders are used by both drivers and pedestrians - the former to allow other vehicles to overtake them.
Abstract