energy equation is employed [7], with the elevation difference between the gutter water level and the point of discharge being equated to the head losses in the system. However, steady-state design methods are not truly applicable when a siphonic system is exposed to a rainfall event below the design criteria, or an event with varying rainfall intensity. Furthermore, current design methods cannot account for commonly occurring operational problems, such as the blockage of outlets and the submergence of the system exit due to down- stream sewer ooding. 3. Description of research As mentioned previously, the main aim of the work detailed herein was to develop a numerical model, capable of accurately simulating the operational perfor- mance of roof drainage systems. As such, it was envisaged that the model would have to be able to route rainfall from roof surfaces down to ground level. To achieve this, it was necessary to undertake both experimental and numerical work. Experimental data concerning the performance of conventional systems during extreme rainfall events are very scarce. It was therefore necessary to undertake laboratory-based experimental work to determine the operational characteristics of conventional systems under such conditions. In contrast, the performance of siphonic systems under all types of rainfall events is relatively well understood, primarily as a result of a research programme initiated at Heriot-Watt University in 1996 [8]. Consequently, no further siphonic experi- mental work was required. In terms of numerical modelling, it was necessary to develop new sub-routines to simulate the ow condi- tions on roof surfaces, along gutters and within conventional system pipework. With respect to siphonic system pipework, much of the developmental work had already been completed during previous research projects [8]. However, the complexity of the existing numerical sub-routines has been found to often lead to numerical instabilities and result in extended computa- tional run times; both of these drawbacks restrict the general applicability of the model. Consequently, it was considered necessary to simplify the modelling techni- ques associated with siphonic systems. 4. Laboratory investigations As the capacity of conventional roof drainage systems is controlled primarily by the capacity of the gutter outlets, the main purpose of the laboratory work was to investigate the ow conditions in the vicinity of the outlets, and their dependence on gutter cross-sectional characteristics