1. Comparison of modeling approaches to atmospheric carbonation of concrete in the context of deep geological disposal of Intermediate Level Waste O. Bildstein (1), P. Thouvenot (1), L. Trotignon (1), S. Poyet (2), B. Cochepin (3) and I. Munier (3) (1)CEA, DEN, DTN/SMTM/LMTE, 13108 Saint-Paul-lez-Durance – France (2)CEA, DEN, DPC/SCCME/LECBA, 91191 Saclay – France (3) ANDRA - 92298 Châtenay-Malabry Cedex - France Context : long term behavior of materials in deep geological disposal of ILLW The present design for the intermediate-level, long-lived radioactive waste (ILLW) is based on concrete structures and reinforced concrete waste packages placed into the deep Callovo-Oxfordian clay-stone geological formation. During the disposal operating period (up to 100 years) these concrete components will be subjected to ventilation in order to (1) guarantee operating safety and (2) contribute to the evacuation of residual heat from the exothermic waste. The ventilated air, drawn from the surface, will generate a partial drying of the concrete components as well as the development of atmospheric carbonation. These processes may potentially lead to a progressive lowering of pH inside the cement paste and trigger corrosion of the steel reinforcement, contributing to a global deleterious effect on the concrete integrity. Conclusions 2 approaches are compared : a complete multiphase model (TOUGH2-EOS4) and a model based on Richards equation (TOUGH2-EOS9 and CAST3M) Model description (Andra, 2005) The carbonation process is difficult to simulate especially due to the overall tendency of models to underestimate the CSH transition to lower C/S values and to precipitate amorphous silica early in the simulation. Both codes indicate similar paragenesis but with a sharper front for TOUGHREACT. Some differences remain concerning the evolution of minerals such as monocarboaluminate and Fe(OH) 3 New simulations should take the effect of water saturation into account for the calculation of mineral reactivity The dynamics of the concrete drying process is captured by the different codes with different modelling approaches (multiphase vs. Richards). Special attention has to be paid to the choice of the water vapour diffusion coefficient because this process may significantly participate to the overall drying rate and the water saturation profile During drying of the concrete, atmospheric carbonation in unsaturated conditions involves intricate couplings between capillary flow, transport of both vapour and liquid water as well as aqueous and gaseous CO 2. Chemical reactions lead, in the same time, to the alteration of the cement hydrates (reactions with dissolved CO 2 ). In order to model the waste package carbonation, a 1D half section of the container (section = 0.11 m) is represented, with atmospheric carbonation occurring on the left face. The drying process The carbonation process ▲ Results obtained with EOS9 and CAST3M are in good agreement (2 codes with the same physics) 2 different approaches are compared : a complete multiphase reactive model (TOUGHREACT-EOS4) and a unsaturated (constant saturation distribution) reactive model (CRUNCH) ◄ A fairly good agreement is reached between the drying dynamics upon using a 10 times lower value for the diffusion coefficient in the gas phase in EOS4  depending on the diffusion coefficient, the drying process is not entirely controlled by water advection ▲ Results obtained with EOS4 and EOS9 are quite different: the diffusion of water in the gas phase (EOS4) contributes to a faster drying front from the surface towards the inner part of the waste container. This latter part also reaches lower water saturation ◄ Primary mineral phases : the overall behaviour ► of both models is similar, but results obtained with TOUGHREACT show a sharper front than in CRUNCH. The dissolution front is the same in both cases except for monocarboaluminate which dissolves much further ahead with CRUNCH ◄ Secondary mineral phases : the same sharp ► front observed with EOS4. The results obtained with CRUNCH show a much smoother front with a less complete portlandite to calcite transformation. Also, intermediate CSH phases are not stable in CRUNCH and are replaced by straetlingite at the front Waste disposal package Disposal cell Protection airlock (dual gate system) Access drift Waste disposal package Concrete lid Concrete overpack Primary waste package to 1.3 to 2.9 m 1.2 to 2.25 m ToughReact Crunch Millington Quirk : D = D 0  a+1 S l b