NUMERICAL VS EXPERIMENTAL INVESTIGATION ON FUEL GAS LEAK DISPERSION IN A REAL SCALE GT ENCLOSURE

Fuel gas leak is a possible scenario that could occur inside a gas turbine (GT) enclosure due to the large number of connections in the fuel gas supply line. Forced ventilation and gas detection systems are used as main mitigations to guarantee an acceptable level of safety. While the ventilation air promotes the dilution below the flammable range of any fuel gas release, the gas detectors monitor the concentrations levels and trigger corrective actions anticipating hazard scenarios. The combined design of these two systems is based on the use of CFD analyses as predictive tool according to guidelines given by ISO 21789. The introduction of hydrogen as fuel for gas turbines makes this design more critical for all its characteristics such as reactiveness and large range of flammability. This leads to an intrinsic higher risk of ignition in case of leakage respect to other conventional fuels like natural gas. In this context, it is fundamental to build reliable and robust computational models for the prediction of fuel gas dispersion inside a complex geometry with forced ventilation flow like a gas turbine package. For this purpose, experimental data of a fuel gas leakage inside a real scale GT enclosure would be needed to tune and validate such numerical models. Since up to now this kind of data is not available in literature, an experimental campaign on NovaLT16 100% H2 package was executed to create a real reference for CFD simulations. Hydrogen is rarely utilized in full scale tests for obvious safety concerns, thus methane was chosen to be released reproducing a real operating scenario. The same set up was replicated with CFD analyses and the results were finally compared.