Motivation

Analysis of combustion phenomena plays an important role in safety considerations of a nuclear power plant. In a severe accident scenario, e.g. a reactor core meltdown, large quantities of hydrogen are typically released into the surrounding structure.

In case of molten core interaction with concrete, carbon monoxide is released as well. The resulting mixture of flammable gases with air introduces a high risk of combustion due to its wide flammability limit and available ignition sources such as hot surfaces or core particles. Flame propagation inside of the reactor containment poses a danger for its structural integrity, through pressure loads that develop in case of a strong explosion.

In order to understand how detonation conditions occur, the entire evolution of the flame has to be considered in the model. After ignition, a slow laminar deflagration starts to propagate. Then the flame accelerates, intensified by the effects of turbulence, and transitions into a fast flame regime, surpassing the speed of sound. Under right conditions, this process can lead to a detonation. Challenges in modeling the described combustion phenomena appear in numerous different flame regimes and flow conditions occurring at heterogeneous scales. Additional difficulties arise from generally complex, large computational domains, required for an industry-scale analysis. Therefore, numerical methodology development particularly focuses on computational efficiency.

Flammable mixture formation is an important stage to consider in an accident event chain. However, gas mixing process is a challenge for CFD codes. The reason is that the process lasts multiple hours which causes prohibitive computational cost due to the fact that CFD methodology requires many orders of magnitude smaller time-steps to achieve stability and accuracy. Lumped parameter calculation approach, on the other hand, is computationally much less expensive. It is a well established and validated strategy for the gas mixing phase and it includes many corresponding sub-models.

Objectives

Approach and Ongoing Work

The starting point of the project is an existing OpenFOAM-based CFD solver capable of efficient analysis of fast combustion regimes that was developed at the Chair of Thermodynamics at the Technical University of Munich [1].

Current state of the art in CFD modeling consists of choosing a solver algorithm with respect to the anticipated flow-speed regime. Pressure based solution algorithms dominate incompressible flow regimes, while density-based methods are superior for transonic and supersonic flows. Considering the nature of combustion phenomena in an accident scenario, such a choice is not possible a priori. A method capable of accurate prediction across flow regimes with spontaneously accelerating flames and fluid flow is required.

Lean hydrogen mixture flames are characterized by high molecular diffusion, causing a distinct flame structure. Efficient combustion models used for reactor safety analysis typically don't account for this, resulting in difficulties in the prediction of flame speeds. A modeling correction based on previous experimental work at the Chair [2] is planned in the course of this work to address this shortcoming.

After validating the CFD method, a coupling interface between lumped parameter and CFD components will be developed. The efficiency and predictive capability of the overall workflow (Fig. 1) is to be assessed on a generic case of a nuclear power plant containment structure.

References

[1] Barfuss, C.; Heilbronn, D.; Sattelmayer, T.: Simulation of Deflagration-to-Detonation Transition of Lean H2-CO-Air Mixtures in Obsructed Channels, Proceedings of 8th International Conference on Hydrogen Safety (ICHS2019), International Conference on Hydrogen Safety (ICHS2019), 2019

[2] Katzy, P.; Hasslberger, J.; Boeck, L.; Sattelmayer, T.: The Effect of Intrinsic Instabilities on Effective Flame Speeds in Under-Resolved Simulations of Lean Hydrogen-Air Flames. Journal of Nuclear Engineering and Radiation Science 3 (4), 2017, pp. 041015

Acknowledgements