In the technological sector of energy and technical chemistry the gas mixture is of hydrogen (H2) and carbon monoxide (CO) are of high importance, e.g. the syngas from reformation or gasification. Due to the mixtures wide explosive range in composition, the safety has to be guaranteed at any time. Especially in the case of a sever accident in nuclear power reactors, the formation of this mixture is of great concern.

The oxidation of the concrete's reinforcement steel (Molten Core Concrete Interaction - MCCI) as well as the oxidation of the fuel rod hull by CO2 and steam, respectively, builds up an explosive mixture. Due to leakage in the containment the gas mixture may enter the reactor containment. Buoyancy effects lead to the formation of a stratified, semi confined gas layer under the roof of the building [1]. Because of its wide ignition limits an ignition by hot surfaces or sparks is highly probable. Flame acceleration leads to fast deflagrative flame propagation. Under certain circumstances the flame transits from a deflagration to a detonation (DDT), resulting in high pressures and loads on the structure, causing severe damage [2]. Thereby the outer barrier between the nuclear inventory and the environment might be destroyed (e.g. Fukushima 2011) [1]. This course of events should be avoided at any costs.

In order to judge about the risks and control such events in the future precise knowledge about the combustion behavior of stratified H2-CO-Air-mixtures with partial containment is crucial. The presented research project aims to contribute knowledge in this area in terms of the KEK program (Competence preservation in Nuclear Technology at GRS). Today’s existing empirical criteria do not satisfy the prediction of DDT in complex geometries. Therefore, the Chair of Thermodynamics is developing a CFD code, based on OpenFOAM, for the prediction stratified H2-CO-Air detonations with partial containment. The numerical work is accompanied by experimental investigations for the code's verification. The research project includes experimental investigations on small and large scale test rigs as well as numerical simulations.

The test rig used in order to investigate the combustion behavior of H2-CO-Air-Mixtures was constructed and used at the Chair of Thermodynamics for previous studies focused on H2-Air-Mixtures [3]. The GraVent test rig (Figure 2) consists of a rectangular channel at a total length of six meters. The design allows for the installation of obstacles in order to vary the blockage ratio and enhance flame acceleration. A venting volume beneath the main flow channel can be used to investigate the effect of venting on the flame.

An optical segment can be installed at different positions. The optical segment features two silica fused windows in order to investigate the flame propagation by optical measurement systems. For flame visualisation a 20 kHz OH-PLIF system is used. Shadow or schlieren techniques are used in order to analyse flow and shock phenomena (Figure 1). Furthermore conventional measurement techniques such as photodiodes in order to measure the flame velocity and pressure transducers are installed in each segment [2].

The goal of the experimental part is to provide data for the numerical code's burning law validation. Furthermore experimental investigations are aimed to determine in which configurations a DDT can occur in order to reduce the test matrix at the large scale experiments at the project partners at KIT. Therefore experiments at different fuel concentrations and blockage ratios are conducted. In a second step the influence of vertical fuel concentration gradients is investigated [4]. Finally the venting volume is used in order to investigate partially contained flames.

Previous projects allowed the development Abbildung 2: CFD Simulation: Detonation in the RUT-facility of an OpenFOAM based CFD code for the DDT prediction of H2-air mixtures at the Chair of Thermodynamics. The code for under resolved prediction of the onset of DDT relies on modeling most of the important mechanisms instead of resolving them. It, thus, provides the possibility of calculating large scale objects like nuclear power plant containments [5].

In order to conduct simulations of H2-CO-Air- Mixtures the existing code is extended by the species CO. The extension requires an adjustment of the combustion parameters, like the mixture composition of the equivalence state, laminar ame speed, flame instability parameters and ignition delay times. Thus, reaction mechanisms including CO as a species are required. The extended code is validated with the previously mentioned experiments in a closed as well as in a partially open channel (GraVent & ProScience).

For further improvement of industry scale simulations, a porosity model is to be added into the code. Thereby, the influence of small scale installations on the flame propagation is implemented. The applied numerical methods, such as adaptive mesh refinement and the advective Riemann-Solver HLLC-Scheme, focus on grid independence. The goal is a good prediction of the DDT onset and pressure loads, rather then the prediction of precise flame structure.

Because of the facility's large scale and the possibility of trend prediction from existing H2-Air experiments, generic cases in the large scale RUT-facility (Figure 3) are calculated in the end. Finally the structural loads on a typical reactor housings is evaluated.