We have investigated whether a plaster/substrate system which has three layers with different pore sizes, can act as an accumulating system. The system consisted of a gypsum and lime-cement plaster layer on a Bentheimer sandstone (GPS system). In figure 1 the effect of a non-uniform initial salt distribution was measured for such a system. The external (gypsum) layer and the base (plaster) layer were initially saturated with pure water and the Bentheimer sandstone was saturated with a NaCl solution (c = 4 mol/l).
Initially the samples were saturated with a salt-solution of different concentrations. Due to the differences in pore structure between the sandstone and the calcium silicate brick totally different drying and crystallization behavior is observed.  In figure 1 the measured moisture distribution for plaster/Bentheimer sandstone as a function of time is given.

Figure 1: a) Moisture profiles in the GPB system for every 2 hours of the drying process.
The gypsum and plaster were initially saturated with pure water and the Bentheimer sand-stone
was saturated with an NaCl solution ( c = 4 mol/ l )  b) Water saturation vs. drying time in
the gypsum and plaster layers, the Bentheimer sandstone substrate, and the whole sample.

In this GPB system the gypsum layer is dry already after about 6 hours. During the same time the water saturation in the plaster layer increases. This results from water transport from the gypsum layer to the plaster layer. The calculated water velocity profiles are presented in figure 2. Negative velocities in this figure indicate that water moves towards the external surface (negative x axis direction). Inspection of this figure shows that at the start of the drying process the water velocity at both sides of the gypsum/plaster interface is positive, which implies that water is transported from this region to the region more close to the substrate.

Figure 2: Water velocity in a GPB sample versus position for several drying times.

In figure 3 the Na profiles are presented for several moments of the drying process. Initially, all salt is present in the Bentheimer sandstone substrate. During drying the amount of salt in the Bentheimer sandstone substrate decreases by about 60 %, while it increases in the plaster layer. Obviously salt is transported from the Bentheimer sandstone to the plaster layer. This is confirmed by the high value of the Peclet number calculated for the Bentheimer sandstone (Pe~18), which indicates that advection dominates during this drying stage. Since the gypsum layer is already dry, further transport of salt to the drying surface is disabled. Therefore, we expect that salt mainly accumulates in the plaster layer, as required for an accumulating system

Figure 3: Profiles of the dissolved Na in the GPB system for several moments of the drying process..

From all the measurements is can be concluded that it is difficult to make an accumulating system only based on differences in the pore-size distributions of the individual layers.