EARLY PROJECT DURATION ESTIMATION FOR ROAD PROJECT USING ARTIFICIAL NEURAL NETWORK IN CASE OF SOUTH DISTRICT

Abstract

Construction industry is an important industry worldwide and data-intensive industry as well as relatively complicated project process. Successful projects relies on the accuracy of its project duration estimates. The application of an artificial neural network (ANN) to predict the construction duration of road at the early stage was described in this study. Determining the early realistic duration for new road construction projects represents a problem for construction professionals. The purpose of this study is to develop an artificial neural network (ANN) model for determining the early duration for bituminous surfaced road projects. According to the review of literature eighteen factors were identified and listed. After conducted questionnaire survey and interviewing experts who deal with design and supervision team, nine factors were identified and selected consisting of: project length, project width, number of crossover includes bridge and culverts, volume of earth work to cut and to fill and cut to waste, masonry stone work for ditch pitch, culvert head, retaining wall and bridge side soil protection, volume of paving work, volume of concrete work, type of terrain classification, and location of projects were ranked as first, second, third, fourth, fifth, sixth, seventh, eighth and ninth respectively based on their calculated relative importance index value and decided with severity from moderate to very high impact on early duration estimation of road projects from other factors. Based on the selected factors data for 43 completed road projects from the Ethiopian Road Administration (ERA) were collected and analyzed using the ANN techniques. The mean squared error (MSE) obtained show that construction professionals can use the developed ANN model for prediction of early duration for road projects. The study shows that the best neural network is the multi-layer perceptron with a structure 9-7-1 based on a back propagation feed forward algorithm. The developed network produces good results with an MES of 0.00 percent or an average accuracy of 100 percent. Apart from the fact that the sample size was small, the developed model does not incorporate the implications of other likely factors that may affect contract duration. The outcome of this study is to help construction experts to fix realistic contract duration for road construction projects.