In evaluating the feasibility of alternative urban road networks a common problem faced by transportation planners is that in existing networks traffic signals are, to a greater or lesser extent, optimised, given the current traffic flows. However for alternative networks with conceivably quite different flow patterns present signal settings (stage times and offsets) may be not at all appropriate and, unless suitable settings can be determined, there may be a built-bias against these alternative systems. This paper investigates one specific aspect of this problem but one which has wider implications, that is the optimum choice of offsets. The method developed is incorporated with the SATURN (Simulation and Assignment of Traffic to Urban Road Networks) suite of programs and determines, for a fixed pattern of flow, how to co-ordinate signals so as to minimise delays. The method may of course be applied to existing networks and therefore has implications for current practice. An important aspect of this problem is the intrinsic feedback between offsets and flows; thus if we coordinate offsets along one major artery we may encourage more traffic to use it, potentially therefore affecting the optimum offsets. The closely associated problem of setting optimum green times is known to be very sensitive to such feedbacks. Quantitative tests on a small real-life network with 12 signalised junctions indicate that optimum network offsets can be determined by a straightforward iterative procedure and that these are stable with respect to changes in flow. It is also found that the optimised offsets increase travel speeds by 15% over the initial arbitrary offsets. (A)
Abstract