The verification and validation of a probabilistic risk analysis method for road tunnels

Abstract: In this thesis, a new probabilistic risk assessment tool for road tunnels has been verified and validated. The verification and validation were performed along with a literature review to find different key variables that affect the risk in a road tunnel and find out how these variables can be handled. Examples of variables are heat release, ventilation mode and factors within evacuation modelling such as response time, recognition time, walking speed and detection systems. The proposed values can inform the selection of default values and help future users which design values to pick. Uncertainties needs to be handled in a risk assessment and recommendations for this are presented. To perform the validation of the 1D fluid-dynamics representation within the tool, five full-scale experimental data sets from experiments made in the Second Benelux tunnel in year 2000-2001 was used as a benchmark. Simulations with the Fire Dynamics Simulator (FDS) was used to facilitate the validation. To perform the verification of different sub-models within evacuation modelling and of the probabilistic risk analysis, hand calculations of different ideal cases were compared to the results from the tool. The verification of sub-models within the evacuation modelling has shown that the tool gives reasonable, most often conservative results with a margin of error within the order of -13 % and +22 %. The 1D fluid dynamics part gives results that are conservative in four out of five cases. The verification of the probabilistic risk analysis gives results in line with the expected calculated values, with a margin of error within +2 %. Overall, this report has concluded that ARTU provides conservative results for risk analyses in road tunnels. In order to confirm this further, future validation studies could be conducted with different experiments.