ASSESSMENT OF CFD CODES CAPABILITIES TO PREDICT THE RISK OF FLAMMABILITY OR EXPLOSIVITY IN THE EVENT OF A HYDROGEN LEAK IN NUCLEAR FACILITIES
Résumé
In the context of nuclear safety, a 4 m3 experimental ventilated enclosure called CARDAMOMETTE has been implemented at IRSN to study the risk of explosion in the event of a hydrogen leakage from a duct in a nuclear facility. Different configurations of hydrogen leakage have been studied allowing to identify those that could potentially lead to explosivity conditions. For safety reason, helium was considered to simulate the behavior of hydrogen. Thanks to high level instrumentation (PIV, He mass spectrometry) and a well-equipped facility allowing local measurements inside the enclosure, a lot of data has been acquired, ensuring a very accurate validation of the CFD code ANSYS CFX. The objective of this validation is to evaluate the capabilities of the CFD code to predict the potential risk of explosivity depending on gas leakage and ventilation configurations.
For this purpose, an experimental and numerical program has been launched to study the influence of ventilation (location of air inlets, renewal rates), gas leakage configurations (location and flowrate, impinging jet) and space clutter (cylindrical container, tubes network, suspended ceiling) on helium dispersion inside the experimental bench and to highlight those leading to hazardous situations.
First, code-experiment comparisons of airflows inside the enclosure were led to ensure the capability of the CFD code to reproduce experimental airflows for some configurations. PIV velocity fields and experimental air renewal curves have been compared to those obtained with CFD calculations, showing a satisfactory agreement. Thanks to this first step, optimal numerical parameters (turbulence model, mesh, boundary conditions) have been chosen.
Secondly, studies of helium dispersion were carried out according to the different configurations presented before. In this paper, only results for free helium jet and impinging helium jet on the wall are presented. Experimental and numerical results of local concentrations were compared, showing a very good agreement and hence the capability of the code to highlight the high concentration areas. Sensitivity studies about turbulent Schmidt number were also led, allowing to define the best numerical dataset depending on the helium injection configurations.
Other experimental and numerical comparisons are currently in progress, especially for the configuration of an impinging helium jet on a cylindrical container.