Artificial neural network model for forecasting energy consumption in hot mix asphalt (HMA) production

This paper provides an overview of the artificial neural network (ANN) model development with the objective of successfully predicting natural gas consumption in the process of hot mix asphalt production. For the purposes of testing, data was collected pertaining to the production of hot mix asphalt during 2014 (155 production days). In total, 77,893 tons of hot mix asphalt was produced during the observed period. The total production for the modeling process is divided into 4 groups depending on the layer (type) of road construction in which the subject mixtures are incorporated: Base, Surf, Bin and SMA. By dividing total production into several groups, an attempt was made to take into account the influence of production and composition of asphalt on the efficiency of predicting natural gas consumption. The following independent variables are used in the modeling process: moisture content, hourly capacity, type of produced asphalt mix and the temperature of produced asphalt. From the obtained modeling results it can be concluded that it is possible to successfully use ANN in the process of predicting natural gas consumption in the production of hot mix asphalt, whereby the composition of asphalt and the specificity of the production itself should be considered.