Using artificial neural network to predict power plant turbine hall key cost drivers

Abstract

The wave of sudden electricity shortage owing to the economic booms worldwide recently had resulted the tremendous time cut in power plant project development. The usual steps in project life cycle, like bidding time in the procurement process is one of them that have not been spared. Despite it has been recognised that the current traditional practice in cost estimation of power plant project is reliable but it is also very time consuming. As such, it is clearly imperative need to find alternate approach in preparation of bids, to meet the odds against time pressure. The study has been formulated to address such issue. The main aim of the study is to use Artificial Neural Network (ANN) as the faster alternative method in predicting key quantities for power plant project. However, the study will only focus on the construction of turbine hall section only. These key quantities normally will be priced by vendors in supply chain, subsequently compiled as latest price at bidding time. The 15 years old historical databases of photographs, drawings, as-built bill of quantities, and bids bill of quantities, from renown power plant constructors, have been used to enable the identification of key cost drivers and key parameters in estimating turbine hall and used to train ANN models. As a validation process, the results from the ANN model has been compared with the statistical method of Multi Level Regression (MLR). The result of the study has determined the ANN regression model is reliable and expected can be used by the contractor in the estimating process of turbine hall construction