Analysis of the ambient condition in a IL-LL waste storage cell of a deep clay repository during the waiting closure period

Document available in extended abstract form only. In the ANDRA repository concept, Intermediate Level Long Life (IL-LL) wastes are stored in 0,5 km long modules. The transport of IL-LL wastes to the storage emplacement is done by machines moving underneath the bottom of the emplacement part of the cell. The large sections remaining open for the movement of these machines in the bottom of the cells offer an easy path to the ventilation air which must have a major effect on flow pattern, especially during the waiting period before closure when ventilation rate are reduced. This study aims at assessing the efficiency of air ventilation with this design in terms of thermal condition along the cell, in the air and in the infrastructures, to find out if reached levels of temperatures and thermal gradients could be damageable for the concrete structures and the waste packages. The effects upon temperatures of local closure of the 'machine sections', changes of the ventilation rate and variable heat release level are also analyzed. The issue is addressed by means of numerical simulation. The physical modeling has been designed to describe the distribution of the air flow all along the storage cell, from deci-metric scale to semi-kilometric scale. The fluid domain is a network of gaps between waste packages, walls and rows where head pressure losses due to inertia and friction forces have major effects on the flow because of the huge contrast between the hydraulic diameters and the length of the cell. The effects of inertia and friction forces are taken into account through head loss models instead of been modelled with complete Navier-Stokes equations, mainly because this last solution needs much higher mesh resolution and a complete turbulence model. Head pressure losses regarding splitting or gathering streams are applied into gap junctions by orthotropic matrix. Finally, the global model are re-calibrated and validated by comparison with results of 2D and 3D simulations using Navier-Stokes equations with turbulent k-ε model and standard wall functions. The physical modeling of thermal transfers includes (i) air/wall heat exchanges due to forced and natural convection, (ii) wall/wall radiant heat exchanges through air gaps, (iii) advection flux in the air, (iv) thermal storage and conduction flux into concrete structures and host rock and, (v) heat release in waste package. Thermal transfers and air flow are coupled through pressures and buoyancy forces. Thus, the results of the simulations are to be considered in terms of order of magnitudes but are relevant to determine whether a problem can occur in terms of temperature rise or of temperature gradient. A 325 m long cell containing 225 waste packages rows into a 80 m thick host rock layer is meshed with 358.517 elements. The flow problem has been discretized with a Mixed Hybrid Finite Element method and the thermal problem with a Finite Volume method. Physical models and algorithm coupling air flow and heat transfers have been set up with GIBIANE language. All calculations and meshing work have been achieved with Cast3M software. In the basic calculation case, the thermal transient is performed with 3 m3/s ventilation flow and 60 W of heat released per waste package (about 250 kW per cell). The air flow enters the first row through 4 different apertures : (i) 'machines sections', (ii) gaps between package piles (81 mm), (iii) wall and piles (95 mm) and (iv) packages and ceiling (170 mm). In all the apertures, air velocity at the entrance is assumed to be equal to 0.75 m/s. The flow distribution between gaps changes along the 30 first rows. The 'machine section' drains 78% of the ventilation flow because of higher hydraulic diameter and a chimney effect setting downstream: pressure forces pull fresh air downward to supply natural convection farther in the cell. Then, Air flow distribution remains quite the same until the 100. row. Beyond, air flow decreases into 'machine sections' and increases in other longitudinal gaps, especially in the ceiling one, while thermal stratification appears in all vertical gaps due to natural convection. The transversal gaps between rows participate a little to heat transfers with few centimetres per second convection flux and highest temperatures (38 C). Around the 175 row, the air temperature difference between the bottom and the ceiling of the cell rises up to 10 C. Along the last ten rows, the air flow distribution changes due to a substantial head loss downstream. At the entrance of the exit gallery, the flow in the 'machine section' lower to 14% of the ventilation rate. Two third of the heat release over the 4 years of the simulation have been stored in the rock and the concrete structure while a third has been removed by the ventilation. Heat released in the waste packages is mainly evacuated by radiant flux and forced advection, through sides in the front of walls, ceiling and ground. The range of the radiant exchange coefficients is 4.3-5.2 W/m2. Convective exchanges are about 30% less efficient in wall gaps and 15% in 'machine sections'. Temperature gradient in waste package reaches 10 C/m at the bottom of rows. The closure of 'machine sections' every 20 rows destroys the thermal stratification but has a small effect on heat removed by the ventilation and maximum of air temperatures. However, it leads to lower temperature gradient in waste package, because it increases the effective exchange area by forcing the air flow out of the 'machine sections' into transversal gaps and others longitudinal apertures. The air ventilation reduction by 10 increases temperature up to 52 C, thermal gradient up to 11 C/m in the upstream waste packages and induces a natural recirculation stretching over the first 150 rows. Upstream, the flow through 'machine sections' rises up to 150% of the ventilation rate