MATERIALS AND METHODS Neutron Tomography Tests The experiments presented in the current work were performed on the NeXT equipment at the Institut Laue Langevin (ILL) with a final spatial resolution of 160 μm and a time interval between tomographies of only 58.5 s. Fig. 3 presents the layout of the experiment further details can be found in Tengattini et al11. During the experiments an infrared radiator heater (Elstein HTS - High Temperature Heater) was placed on the top surface of the sample and its heating followed two distinct protocols: (i) the “slow heating” where the sample was heated with a 10 °C/min rate until reaching 500 °C and (ii) a “fast heating” of 158 °C/min rate until 500 °C (see Fig. 8). A dense CAC-bonded refractory castable was prepared (Table 1) as described in Luz et al.12. The samples were cast in cylindrical PVC molds or in a sintered alumina ceramic casing with an inner diameter of 33 mm and a height of 50 mm, depending on the use or not of a casing during the neutron tomography test. Table 1: 5CAC composition used in the neutron tomography tests based on the Andreasen’s particle packing model with q = 0.21. Raw materials Composition (wt.%) Tabular alumina AT (d 6) 74 Calcined and reactive alumina CL370 11 CT3000SG 10 Calcium aluminate cement Secar 71 5 Distilled Water 4.5 Fig. 3. (a) Layout of the neutron tomography test conducted in the NeXT equipment at Institut Laue Langevin (ILL) and (b) vertical slice of the reconstructed data. Post-processing of the Results The 3D representation of the samples was obtained using the Feldkamp filtered back projection and a commercial reconstruction software (X-act, by RX Solutions). The digital representation of the sample was composed of a set of 300 slices in the cylinder’s height. Even though several advanced calculations can be performed with the obtained data, the basic quantity used for the analysis in the current work was the relative difference, 𝜓𝜓(𝐱𝐱, 𝑡𝑡), calculated by Equation 1. 𝜓𝜓(𝐱𝐱, 𝑡𝑡) = �𝐼𝐼(𝐱𝐱, 𝑡𝑡) − 𝐼𝐼(𝐱𝐱, 𝑡𝑡0)� 𝐼𝐼(𝐱𝐱, 𝑡𝑡 0 ) × 100% (1) where 𝐼𝐼(𝐱𝐱, 𝑡𝑡) is the intensity of a voxel on position 𝐱𝐱 (𝐱𝐱 = (x, y, z)) at time 𝑡𝑡, and 𝐼𝐼(𝐱𝐱, 𝑡𝑡 0 ) is the voxel value at the reference state, which was defined to be the median of the first 10 tomographies, as described by Dauti et al.8. RESULTS AND DISCUSSION Casing Effect Analysis The first set of results describes the effect of a ceramic casing around the sample. This has several implications and can ultimately yield a truly unidirectional mass and heat transport inside the sample.