Advancing CFD fire modelling in mechanically ventilated nuclear facilities

The main goal of this PhD thesis is to advance knowledge on some physical phenomena and improve existing numerical modelling capabilities for fire behaviour in mechanically ventilated compartments. To achieve this goal, the following different studies were undertaken in the project: 1. Implementation of a Mechanical Ventilation Model: This study improves the existing numerical capabilities of the Computational Fluid Dynamics (CFD) code FireFOAM by implementing a model based on a simplified Bernoulli Equation in order to predict the pressure variations in the compartment and flow rates in the ventilation system. 2. Conjugate Heat Transfer (CHT) Study: Heat transferred between combustion gases from the fire and the compartment's walls, was studied using CHT. The work improves the numerical modelling capability of the CFD solver by combining a combustion solver and a CHT solver. 3. Modelling Liquid Fuel Evaporation: This study advances the modelling of the physical phenomena involved in liquid fuel evaporation with the development of an evaporation model and its numerical implementation in the CFD solver FireFOAM. The evaporation model was validated by investigating its accuracy in predicting the fuel Mass Loss Rate (MLR). Spills of flammable liquid leading to fires are typical scenarios encountered in nuclear power plants. 4. Vertical Smoke Propagation (VSP) Study: To further validate the predictive capability of the evaporation model and provide more insights to physical phenomena, smoke propagation through a horizontal opening between two superposed compartments was also studied. The study of VSP was included in this thesis because of the nature of most nuclear power plants compartments that have multiple rooms usually superposed that are connected through a horizontal vent or opening. 5. In-depth Convective Heat Transfer Modelling: This study contributes to advancing the modelling of physical phenomena, by modifying the developed evaporation model with emphasis on heat transfer within the fuel. This was accomplished by implementing a novel method of predicting convective currents within the fuel. It involved deriving the effective thermal conductivity of the liquid fuel during combustion. 6. Combustion Modelling: This study involved improving the prediction of the fuel MLR by implementing a general combustion model. The combustion model was formulated by iii combining the already existing Eddy Dissipation Model (EDM) with the finite rate model. The inclusion of the finite rate model would generate laminar flow that allows for fuel vaporization without combustion at the initial stage. 7. Convective Heat Transfer modelling: Improvements were made in the predictive capability of the evaporation model by making use of an average flame temperature value for calculating the convective heat flux at the fuel surface. The different studies summarized above include a mixture of phenomena that are well understood and those that are still in early stages of development. The contribution from well understood phenomena is the advancement of the predictive capability of the CFD code FireFOAM which will allow the fire community to explore some new problems. The well understood phenomena include the mechanical ventilation, CHT, combustion modelling and convective heat transfer. The thesis makes some significant and novel contributions on research areas that include modelling liquid fuel evaporation and in-depth convection heat transfer. These research areas are in infancy and under-development particularly in the context of mechanically ventilated fires. There have been previous works conducted on evaporation modelling, but further knowledge continues to be developed due to the increasing complexities surrounding fuel combustion i.e., open atmosphere, single and multiple mechanically compartments. Phenomenon such as VSP are not well understood in the nuclear fire safety, and in developing modelling capabilities in the CFD FireFOAM, the thesis makes an important contribution. Different experiments from the literature were used for verification and validation purposes in the studies mentioned above. However, the main experiments used in this thesis were conducted in the OECD PRISME Projects (PRISME 1 &2). The PRISME Projects consisted of experiments conducted in open atmosphere and in a mechanically ventilated compartment representative of a typical nuclear power plant. PRISM data were mostly used for the validation of the models in a mechanically ventilated compartment. Additionally, the experiments conducted by Vali, Nobes and Kostiuk (2014), and Kang, Lu and Chen (2010) were used for further validation of some of the models in open atmosphere. PRISME projects concerned nuclear power plants that constitute a major risk due to the potential leak of radioactive materials during a fire incident. The use of well validated fire models offers a better alternative to the use of traditional prescriptive fire safety regulations. Nuclear power plants make use of mechanical ventilation which provides dynamic iv confinement for nuclear materials by maintaining the required pressure. The occurrence of fires could potentially result in pressure variations within power plants. Although dynamic confinement along with other safety measures are in place to prevent fires, there is a continuous need to assess fire safety measures and reduce the risk of fire propagation with the use of fire simulation codes. Nuclear power plants are usually superposed with a horizontal vent or opening connecting both rooms. Therefore, VSP during fires poses another major risk particularly if there are combustible or radioactive materials in the adjacent room. The flow at the horizontal vent and inside a mechanically ventilated compartment can undergo complex dynamics which would affect the behaviour of fire including liquid fuel evaporation. This flow behaviour can be determined by several factors and from prior experimental studies conducted, the major factors affecting the behaviour are the position of the fuel, the ventilation configuration in both rooms and fuel MLR (Pretrel and Vaux, 2019; Pretrel et al., 2014; Pretrel et al., 2017). These factors affecting the horizontal vent flow coupled with the superposed nature of nuclear rooms provide justifications for some of the studies undertaken above in this thesis. As mentioned above, the presence of complex flows in mechanically ventilated configurations could undoubtedly affect the validation study of the models developed in this thesis. In order to address this, additional fuel evaporation is studied in open atmosphere particularly to validate the evaporation model and in-depth convective heat transfer. The results obtained in open atmosphere allows the assessment of the accuracy of the models and provides the necessary confidence to perform the study in mechanically ventilated nuclear compartments. This thesis builds on existing research by making use of an emerging open-source computational fluid dynamics (CFD) fire simulation code known as FireFOAM, to predict fire behaviour in a mechanically ventilated nuclear compartment. An existing in-house modified version of FireFOAM developed by the authors' research group, is further modified in the present work to implement the different models developed in this thesis. This includes the Conjugate Heat Transfer (CHT) model to account for the heat transfer between combustion gases and solid boundaries. The CHT is validated using the wall temperatures and heat fluxes from PRISME experiments. Furthermore, a mechanical ventilation model has been developed and implemented into FireFOAM. The mechanical ventilation model was coupled initially with the experimental fuel MLR to predict the pressure variations in the nuclear compartment and v predict the flow rates in the ventilation network. Further step was taken in modifying FireFOAM by including an evaporation model used to predict the fuel MLR. The evaporation model was initially validated by comparing it to the experimental fuel MLR. Further validation of the evaporation model includes predictions of the flow behaviour at the horizontal vent and the predictions of the flow rate at the mechanical ventilation. Modifications were made to the evaporation model by the advancements of in-depth convective heat transfer, combustion modelling and convective heat transfer mentioned above. The fuel MLR and temperature values inside the liquid fuel were used to validate the In-depth convective heat transfer model. Fuel MLR was used to validate the combustion model and convective heat transfer. The different studies conducted in this thesis are highly important for fire risk safety assessment in nuclear facilities and any findings would help in future assessment. In nuclear power plants, pool fires may result from ignition of spilled lubricating oil from turbines and pumps, or from liquid fuel spilled from standby generators. The thesis work could also go beyond nuclear facility and into other industry like the petrochemical industry and building services industry that continuously face fire hazards. The overall work performed in this thesis is believed to have advanced CFD fire modelling in mechanically ventilated compartments. The summary of the individual work accomplished in this thesis are highlighted below: • Mechanical Ventilation: The mechanical ventilation model predicted the pressure variations and flow rates at the ventilation branch with a relatively good level of accuracy. The results from the validations show that the burning rate is controlled by both the mechanical ventilation and the flow from the vent. • Conjugate Heat Transfer (CHT): It was concluded that there is no significant difference between the use of CHT and wall laws in simulating the interaction between the wall and combustion gases. However, suggestions were made regarding complex configurations where the use of CHT would be much more beneficial than wall laws. • Evaporation Model: The evaporation model was able to predict the fuel MLR in open atmosphere and in a compartment, particularly the steady state stage of fuel MLR. The lack of a gas phase extinction model that would capture the effects of lack of oxygen, led to discrepancies in the predictions of fuel MLR particularly in the compartment. vi • Vertical Smoke Propagation: The direction of flow at the horizontal vent was captured for some of the tests. The absence of a gas phase extinction model led to discrepancies in predictions of flow at the vent. • In-depth Convective Heat Transfer: Heat transfer within the fuel was improved with the use of an effective thermal conductivity. It was shown that without the convective currents inside the fuel, the fuel MLR is over-predicted in open atmosphere, including temperatures inside the fuel. • Combustion Modelling: Improvement was made at the initial start of fuel combustion. The initial spikes of fuel MLR observed in previous studies was reduced with the general combustion model. • Convective Heat Transfer: The convective heat flux derived with grid cell temperature was found to be dependent on grid size, while the use of an average flame temperature was discovered not to be dependent on grid size.