Gas permeability of unsaturated cement-based materials: application of a multi-scale network model

A multi-scale network model based on the percolation theory and using a renormalization method was applied to describe the porous space of cement-based materials and to determine their gas transport properties. Gas permeability was measured on hardened cement pastes prepared with French OPC (CEM I) and BFS–PFA (CEM V) cements currently used in the nuclear waste industry. These experiments were carried out to study the influence of the degree of water saturation and of the microstructure on materials gas transport properties. Between 10 and 100% relative humidity, the cement pastes' average gas permeability varies over five orders of magnitude (10⁻¹⁶ –10⁻²¹ m²). Results show that in the range of water saturation considered, there is no linear relationship between the gas permeability and the degree of water saturation in percentage terms. Permeability change was well observed for cement pastes characterized by w/c ratios of 0·30, 0·40 and 0·50, in the low water saturation domain (<60%). CEM V pastes are globally more permeable to gas than CEM I pastes in the range of one order of magnitude. For the modelling aspect, the application of the XDQ multi-scale model made it possible to model a porous network equivalent to the one obtained by the mercury intrusion porosimetry test. Each class of pores is associated with an elementary cubic random network. The whole porous system is rebuilt by a recurrent process involving superposition and rescaling of all the elementary networks. Then, an equivalent conductivity of the multi-scale network is computed and extended to gas permeability. Modelling and experimental results are in agreement and show that mercury intrusion is a relevant method to characterize a porous space in order to estimate the gas permeability of cement-based materials with a conceptual model.