STUDY OF MODELS FOR THE PREDICTION OF AVERAGE ROAD LOAD OF IN-USE DELIVERY TRUCKS

Coast-down studies of Class 6 trucks were conducted at near-zero road grade and wind conditions. Speed-versus-time data were recorded, smoothed to a quadratic equation to calculate deceleration rates, and reduced by the least-square method using two-, three-, and five-parameter linear models. The estimated parameters were analyzed statistically and compared to the U. S. Environmental Protection Agency (EPA) recommended procedure. The averages of the individual truck road loads for the three models were consistent within 4% when reference conditions of 40 mi/h, 15,000 lb weight, and 65 sq ft projected frontal area were used. These reference conditions represent the median values of experiment conditions that were varied during coast-down tests. At these conditions, road load variations among the trucks averaged 10% and 16% for the two- and five-parameter models, respectively. The coefficients of the five-parameter model displayed instability during sequential estimation, showing multicollinearity. Because of this instability and the significance of the truck-to-truck variations, the two-parameter model seemed most appropriate for predicting the truck road load. At reference conditions, the two-parameter model road load was 10% greater than that from the EPA recommended procedure.