Dynamic risk analysis of construction delays using fuzzy-failure mode effects analysis

Considering the tremendous losses in the worldwide economy caused by construction delays, it is essential to invest in minimizing the risks of delays. In order to make this happen, two measures should be taken: 1) The roots and fundamental causes of delay should be identified and strategies to mitigate their risks be developed (General remedy). 2) The most significant potential causes of delay in each project should be identified and these causes should be given priority to control (Project-Specific Remedy). The current research invests in both of the measures. To provide the general remedy, causes of delay in the construction industry of the United States is investigated through a national survey responded by the 224 construction experts with an average experience of over 27 years. The results of this study rank the criticality of the thirty main causes of construction delay in the U.S. construction industry. The focus of the research is on the project-specific remedy. The research aims at designing a tool, which can prioritize different causes based on their criticality. This is crucial as there is often a large number of potential causes and investing in prevention of all of them is not practical. The designed tool is capable of identifying the most critical causes by assessing its status of the potential causes of delay in three elements of criticality which are: 1) The likelihood of occurrence of the cause, 2) the severity of the cause in creating delays (in case it happens), and 3) the resolvability or likelihood of handling the potential cause before it creates a delay, in case it happens. The three elements of assessment are inserted in a designed tool in Matlab®, which uses a fuzzy logic system to generate a "risk priority number". This number is a representative of the riskiness of each potential cause. The next contribution of the research is a model that is capable of predicting the percentage of delay based on the "fuzzy risk priority number". This model uses the output of the aforementioned fuzzy inference system to make a prediction about the percentage of delay. The model was tested by comparing its predictions with actual data (the delay that has actually happened) and has been able to predict the amount of delay with an error of less than 20%.