Neural-Risk Assessment System for Construction Projects

Abstract

Risk management is not a new topic in construction industry. Its main problem relies in assessing the risk impacts and as well in forecasting the possible costs of these risks. Transforming the risk impact into money terms certainly is not an easy thing to do. Traditionally within construction companies, risk management has been adopted; nevertheless, the work has been concentrated mainly in risk analysis. For doing this risk analysis, a large number of mainly probabilistic techniques (for example Monte Carlo Simulation, Latin-Hypercube and Sensitivity Analysis) for determining the behaviour of the risks have been employed; nevertheless, the quantification of the risks is money terms is still lacking. Nowadays, having a risk management system integrated into the company’s organisational structure is a necessity that needs to be satisfied; this system should deals with all the project risks. The main goal of this research-work is to quantify in terms of money the cost of the risks involved in construction projects. For that reason, a list of Risk-Factors (RF) has been created in order to model the behaviour of the risks for the chosen construction project. This is made with the help of the Neural-Risk Assessment System (NRAS). The core parts of the system are represented by Artificial Neural Networks (ANN) and risk management. The objectives of the research are: to identify the most common risks in infrastructure projects in Germany, to quantify the risks in terms of money, to analyze and evaluate their impact to the Contractor’s profit, to proof the reliability of using neural networks in the management of risks, to offer to the contractor an alternative tool to forecast the possible cost of risks.