Search

In the deep geological repository for radioactive waste in France, the Intermediate Level Long Lived Waste (ILLW) is planned to be disposed in ILLW emplacement drifts that are several hundred meters long. Inside the drift, primary nuclear waste packages received from producer is packed in concrete disposal containers which are stacked. Each disposal container contains one or several primary nuclear waste packages. Both the disposal containers and the primary nuclear waste packages contain various components that can react chemically or radiologically as e.g., by radiolysis, corrosion. These processes may consume or form oxygen and hydrogen gas. The mixture of both of those gas phase components poses the risk for an oxyhydrogen explosion (ATEX). The behavior of oxygen and hydrogen is thus a major concern for the operational safety of the repository and reversible phases with an envisaged duration of roughly one hundred years. Free oxygen is introduced to the system through ventilation of the access galleries and open emplacement drifts while at the same time, radiolysis of organic waste produces hydrogen and corrosion of concrete reinforcement, disposal container envelopes, and metallic elements in the waste may consume oxygen or produce hydrogen. The purpose of the present study is to determine the gas migration path and concentrations in ILLW disposal containers in particular with respect to hydrogen and oxygen concentrations and the induced risk for explosion. In order to characterize the relevant two-phase hydraulic conditions, two kinds of work have been undertaken. On the one hand, a numerical model was constructed and simulations were performed. On the other hand, gas release experiments have been conducted on a full scale demonstrator of an ILLW disposal container. The joint analysis of numerical simulation and experimental results improved the understanding of the gas transport processes and will allow for the optimisation of the design to avoid ATEX. For the numerical model, a two-phase multi-component thermo-hydraulic model was developed. The model geometry is based on the CAD model of the disposal containers and consists in a 3-dimensional representation of the disposal container and the waste packages (Figure 1). Interstices at e.g., container's lid and body are modelled explicitly. The components of the liquid phase are water, dissolved oxygen, dissolved hydrogen, and dissolved nitrogen. Equivalently, the gas phase consists of water vapour, nitrogen, oxygen, and hydrogen. Simulations are non-isothermal, since the emplacement tunnel and the waste container are not initially at thermal equilibrium. The emplacement conditions are considered by appropriate boundary conditions. During the exploitation phase, ventilation is taken into account in order to represent exchanges with the environment as e.g., the drying of the concrete as accurately as possible. The hydrogen formed by radiolysis, is modelled by source terms located inside the primary waste packages. The model was implemented with a custom made version of TOUGH2- MP, allowing for a material-dependant generalised Millington-Quirk representation of diffusion and its dependence on saturation and porosity. Reference calculations as well as numerous sensitivity studies were performed. The results On the experimental side, the setup is the CS4 (Andra nomenclature) concrete disposal container, which has been equipped with a gas injection line source at the location of the primary waste packages where either helium or hydrogen can be injected. Some experimental gas injection sequences have been defined, ranging from several days to several months. Those sequences have been applied successively with two different types of contacts between the disposal container's body and lid. First, the container was closed with screws and, in a second phase, the container's lid was sealed with concrete. Specific dynamic chambers [1] have been installed at several locations of the container in order to measure hydrogen or helium outflowing gas rates and to deduce concentrations by the molar mass balance method. Moreover, the environmental conditions (temperature, humidity, atmospheric pressure) are carefully measured. The measurements performed during first experiments with a screwed lid showed that the body-lid interstice is by far the most preferential escape flow path of the injected gas. This was confirmed by further tests where this interstice was temporarily confined. Thanks to the numerous measurement points, experiments also allowed to establish a cartography of the gas outflow through the container walls, distinguishing zones in function of the local effective thickness of the concrete. Following the first experimental results, the numerical model was adapted to better correspond to the experimental setup and sequences. In particular, the representation of the interstices and the concrete with its two-phase properties had to be reconsidered. Also, an evaluation of the impact of the dynamic chambers on the system behaviour was conducted and the boundary conditions were adapted to correspond to the experimental setup. This interplay between numerical modelling and experiments has improved the interpretation of the gathered experimental data and has improved the overall systems understanding of hydrogen and oxygen flow and transport processes in ILLW containers designed for a clay repository.