Assessment of risk allocation in construction projects

The objective of this study was to provide owners with a systematic analytical approach to identify, analyze, and manage activities associated with risks in construction projects. In developing this approach, attention was given to the predicting stage utilizing neural network technology. Predicting the risk events is expected to reduce the expenses associated with risky activities. Risk assessment is a difficult process in construction projects. The risk assessment model suggested by the researcher provides the owner with an effective and systematic framework for quantitatively identifying, evaluating, and responding to risk in construction. The prototype designed to conduct construction risk assessment used the NeuroShell Predictor software. Contractor default is one of the most costly activities that are associated with risk. A prototype was developed to predict the likelihood of contractor default and the expected amount of loss in case of default. The level of success for this model depends on the availability of raw data concerning defaulted contractors, the projects where the defaults occurred and the environmental (economic, regional, social, etc. ) conditions. Two sets of data were used in this prototype. One of them was extracted from the files of a public owner in Saudi Arabia, where the surety bond is not used. The other set of data was extracted from the files of a major Surety Company in United States. A neural network was trained using these two sets of data. Then, the network was tested to demonstrate its capability to predict contractor default. At the end of the study, a recommendation was drawn from the results and a risk allocation flowchart was presented.