VERIFICATION AND COMPARISON OF FOUR DIFFERENT PREDICTION METHODS FOR RUTTING --BITUMEN, FLEXIBLE AND DURABLE. 3RD EUROBITUME SYMPOSIUM 1985, THE HAGUE, NETHERLANDS, SEPTEMBER 11-13, 1985

Three methods, based on the laboratory study of the mechanical properties of bituminous materials were used to forecast the rutting behaviour of two experimental sections. For the long term behaviour an empirical prediction method is used as reference. Both test sections are overlays on state road no 28, one with a normal and the other with a long design life. For the theoretical prediction the following methods, based on static, semi-dynamic and dynamic creep testing, are used: the shell method, a method developed by bolk of the road engineering division of de rijkswaterstaat (wbd) and the esso road design technology (erdt) method. For the long term in situ behaviour an extrapolation is made by the empirical method developed by buiter of the wbd. This method is based on measured rutting and has proved its reliability for use in a pavement management system. The rutting prediction according to the shell and the bolk method is based on wbd research and the erdt prediction on esso research, all combined with environmental and traffic data. For its investigations the wbd has made use of road specimens, while esso used laboratory prepared specimens. The calculated values are compared with rutting measurement results since 1978 and with the long term behaviour according to the empirical model. The calculated rutting level according to the shell method and the bolk method is too high in the initial stage. However, in general the bolk method shows a good agreement with the expected rutting behaviour in the long term. For the shell method the gradient of the calculated deformation curve is too small after the initial stage. The theoretical methods, except for the erdt method, are predicting more rutting for the long life section than for the normal life section. The difference is between 1.8 and 4.0 mm after 15 years. The erdt prediction for the normal life section shows an excellent similarity with the empirical model in both the initial stage and the linear stage. However, the calculated deformation pattern for the long life section is obviously too low. The fact of having chosen full friction interface conditions between layers may be an explanation for the low level of rutting computed (overestimation of lateral stresses) and the use of laboratory prepared specimen might be also a possible cause. Special attention is paid to the rutting contribution of the individual asphalt layers, including the old structure. In general the results of the bolk and the erdt method show a more realistic picture than the shell method. For the covering abstract of the symposium see IRRD 815173.