*3.2. Mechanical Tests*

The cubic specimens with side a = 150 mm were used for compressive strength determination, with a diameter (d) of 100 mm and height (h) of 200 mm for the cylindrical samples for the splitting tensile strength tests. Three samples were used to test unheated concrete and two were used to test heated concrete. The modulus of elasticity was determined from the stress–strain (σ-ε) using one cylindrical sample (d = 100 mm; h = 200 mm). All E values were expressed in GPa and calculated from σ-ε curves as the stress to strain and strain ratio in the range of 10% to 40% of the ultimate stresses. For all properties six temperature levels were studied: T = 20, 200, 400, 600, 800, and 1000 ◦C. The compressive strength test procedures applied were presented in EN 12390-3 [32], and the splitting Brazilian tests were done according to EN 12390-6 [33].

#### **4. Test Results and Discussion**

### *4.1. Initial Properties*

For B CEMI, B CEMIII, RB CEMI, and RB CEMIII concretes, the initial physical properties of bulk density ρo20◦<sup>C</sup> and permeability k, and the mechanical properties of compressive strength fc20◦<sup>C</sup> tensile strength ft20◦<sup>C</sup> and modulus of elasticity E20◦<sup>C</sup> were determined after 90 days. The initial measurements, obtained for non-heated concrete properties, are presented in Table 5 and marked with the symbol 20 ◦C.


#### *4.2. Evolution of Bulk Density with Temperature*

The progressive increase of the temperature resulted in free water evaporation and progressive dehydration of the material. The C-S-H, as well as portlandite and calcium carbonate decomposition, were progressive in higher temperatures. As a result, weight loss was observed and the progressive density changes were recorded. The bulk density of B CEMI, B CEMIII, RB CEMI, and RB CEMIII concretes decreased as a function of the temperature. The mean values of bulk density are presented in Figure 2.

**Figure 2.** Bulk density of riverbed aggregates (RB) and basalt (B) concretes made with CEMI and CEMIII concretes; mean value of three samples.

In Figure 2 the bulk densities of the test concretes are presented. The values are mainly related to the type of aggregate: basalt or riverbed. The density of basalt CEMI concrete was 2558.8 kg/m3 and the B CEMIII 2533.2 kg/m3. The RB CEMI and RB CEMIII concrete were 2300.7 and 2315.6 kg/m3, respectively. Apart from the initial values of density observed in the non-heated pristine concrete, the changes of the density with the temperature were quite similar for both cement types.

#### *4.3. Evolution of Compressive Strength and Splitting Tensile Strength with Temperature Exposure*

Figure 3 depicts the average and individual values of compressive strength. From the figure it can be concluded that the compressive strength of unheated concrete was higher for both CEMIII concretes made with basalt and riverbed aggregates. This tendency is maintained at 200 ◦C. When the temperature is higher than 400 ◦C, there are few differences in strength between B CEMI and B CEMIII, as well as between RB CEMI and RB CEMIII concretes. They all presented almost the same strength of 60 MPa.

**Figure 3.** The compressive strength evolution CEMI and CEMIII concretes on basalt and riverbed aggregate.

In Figure 4, the average and individual values of ftT are presented. Heating resulted in a progressive reduction in strength, nevertheless, the differences between CEMI and CEMIII concretes over a whole range of temperatures may be considered insignificant, in the scope of measurement error, or the scatter of results for this mechanical property.

**Figure 4.** The changes in the splitting tensile strength of heated CEMI and CEMIII concretes on basalt and riverbed aggregate.

As has already been shown in previous research, an important aspect in the high temperature behavior of concrete is the thermal stability of aggregates at high temperatures. This can be evaluated by thermo-gravimetric and differential thermal analysis, which indicate the physical or chemical transformation of aggregates. As has already been reported [10], basalt is thermally stable up to 1000 ◦C; above this temperature melting is observed at 1050 ◦C and expansion and gas release both occur.

#### *4.4. Relationship between Stress and Strain, and the Modulus of Elasticity Evaluation*

The stress–strain relationships for the tested concretes are presented in Figure 5. Along with the temperature increase, a change of concrete stiffness was observed, as represented by the slope of the stress–strain curve. For the specimens heated to 600 ◦C and above, the stress–strain curve presents nonlinear behavior in compression due to the presence of cracks, which are closing partially when a compressive load is applied during the test. The similar stress–strain behavior of concrete in compression was observed for hot tested and tested after cooling down [8,15], an important cracking of samples was observed, especially for concretes with siliceous aggregates, heated without loading. The cracking of unloaded concrete was confirmed by the thermal strain evolution observation during heating [6].

The static modulus of elasticity values (ET) of heated B CEMI and B CEMIII, as well as RB CEMI and RB CEMIII, are shown in Figure 6. The pristine non-heated concretes' modulus of elasticity (E20◦C) were 44.4 and 48.9 GPa, respectively, for B CEMI and B CEMIII. For riverbed aggregate RB CEMI and RB CEMIII they were 30.6 and 29.7 GPa. These results show clearly that for concretes with the same volume of cement paste, the modulus of elasticity is related to the nature of the aggregate and is strongly related to concrete density. Higher values of ET were observed for both CEMIII concretes with RB and B aggregates.

A quasi linear decrease in the ET value over the whole range of heating temperatures was observed. The slope of ET decrease is most pronounced in the range of temperatures between 400 and 1000 ◦C (Figure 5). This sharp decrease of stiffness was attributed to crack development due to a mismatch of the strains between the cement paste and aggregates that is observed in this range of temperatures, and an increase in thermal strains resulting from cracking [6,7].

 7R&

**Figure 6.** Modulus of elasticity change with the temperature of CEMI and CEMIII concretes.

From Figure 6 it can be concluded that the relative change of the modulus of elasticity is quasi identical for the concretes tested, and does not depend on cement type. The differences between the modulus of elasticity values of RB CEMI and RB CEMIII are not significant except for differences occurring at 20 ◦C.

#### *4.5. Heated Concrete Permeability Evolution*

For RB CEMI and RB CEMIII, the initial reference permeability, measured on non-heated concrete after exposure to 20 ◦C, reached values of 1.20 <sup>×</sup> <sup>10</sup>−<sup>17</sup> <sup>m</sup><sup>2</sup> and 1.00 <sup>×</sup> <sup>10</sup>−<sup>17</sup> m2, respectively. For B CEMI and B CEMIII this permeability was 0.70 <sup>×</sup> <sup>10</sup>−<sup>17</sup> m2 and 0.52 <sup>×</sup> <sup>10</sup>−<sup>17</sup> m2. With the increase of heating temperature residual permeability was increased. For the specimens heated to 1000 ◦C the permeability could not be measured due to crack development, and the gas flows could not be stabilized, so the permeability could not be measured with the Cembureau set-up. The results of the permeability measurements are presented in Figure 7. For B CEMIII and RB CEMIII concretes generally, lower values of permeability were observed. For the riverbed aggregate concrete RB CEMIII, permeability measured after exposure to high temperatures at 200, 400, 600, and 800 ◦C was systematically slightly lower than for RB CEMI. Basalt aggregate-based concretes provide lower permeability than riverbed ones. Nevertheless, these differences could not be considered significant. For all the concretes heated up to 1000 ◦C, the permeability could not be measured with the Cembureau method due to the significant damage to the concrete and crack development.

**Figure 7.** Effect of heating on the permeability of the test materials: RB CEMI and RB CEMIII, B CEMI and CEMIII. The reference permeability at 20 ◦C and permeability after heating to 200, 400, 600, and 800 ◦C.

#### *4.6. Permeability vs. High Temperature Damage Factor*

Previous studies [34,35] have indicated that the degradation of concrete at high temperatures, arising from a coupled hygro-thermal, chemical (dehydration) and mechanical interaction, can be modelled by means of the isotropic damage theory of Mazars [36]. Following Gawin et al. [9], the total damage *D* may be described by a multiplicative format of mechanical and thermo-chemical damage components, as shown in Equation (2):

$$D = 1 - \frac{E(T)}{E\_0(T\_0)} = 1 - \frac{E(T)}{E\_0(T)} \frac{E\_0(T)}{E\_0(T\_0)} = 1 - (1 - d) \times (1 - V),\tag{2}$$

where V corresponds to the thermo-chemical damage and *d* to the mechanical damage. The term (1 – d) corresponds to <sup>E</sup>(T) E0(T), and (1–V) to E0(T) E0(T0). In the above equation E0(T0) is the initial value of the static modulus of elasticity, E0(T) is the modulus for mechanically undamaged material expressed in a function of heating temperature, and E(T) represents the static modulus of elasticity of mechanically damaged heated concrete.

Following this approach, in Figure 8 the effect of temperature on the damage parameter for heated concretes is presented. The damage factor was calculated on the basis of the change in the modulus of elasticity with temperature (see Figure 6), leading to Equation (3), and this evaluates the deterioration of the stiffness of the heated concrete samples by comparing them with the parameters found in non-heated concrete:

$$D\_E = \mathbb{1} - E\_T \mathbb{E}\_{20} \circ\_{C\_{\mathcal{L}}} \tag{3}$$

where E20◦<sup>C</sup> is the static modulus of elasticity tested at 20 ◦C and ET is the value obtained for heated concrete.

The damage factor follows a comparable increasing change for all tested materials and almost reaches the value of 0.9, which means that 90% of the concrete has deteriorated. However, at 400 ◦C the damage value becomes much higher for the basalt aggregate concretes in comparison with the riverbed aggregate ones. Overall, the damage values for the CEMIII concretes appear to be slightly lower than for the CEMI concretes, especially for the basalt-based materials.

**Figure 8.** Damage factor (*DE*) as a function of temperature.

These changes may be qualitatively compared to the change of total damage with temperature of a high performance concrete [9]. However, the damage values obtained and cited in this study are higher (damage of 0.8 at 600 ◦C). The reason for this difference may be due to the heating conditions, and notably the heating rate, which was four times higher in the study by Gawin et al. [9] than in our procedure, and which may provide stronger thermal gradients and therefore greater degradation.

It has already been noted that the changes to the inner micro-structure and permeability of the concrete may be characterized using this mechanistic approach, using damage evaluation to describe the high temperature degradation and/or micro-cracking effects [9,26,37,38]. The results of such a correlation are presented in Figure 9 for all the test materials. One may observe that all the data follow a single master law, independent of cement type or aggregate type. The results follow an exponential relationship, except for the permeability values obtained at 800 ◦C (Equation (4)):

**Figure 9.** Juxtaposition of permeability (log scale) and damage to unheated concretes (20 ◦C) and concretes heated to 200, 400, 600, and 800 ◦C.

In Equation (4) k is the permeability of the heated material, k0 the initial reference permeability, DE the damage factor, and CDE is the material dependent parameter, here equal to 8, which confirms the value obtained for another high performance concreteat elevated temperatures, but based only on the CEMI cement [9]. The CDE value being equal to 8was obtainedfrom the regression curve with the coefficient of determination R<sup>2</sup> of 0.86 value. Therefore, the proposed regression curve is limited

in the range of temperature from 20 to 600 ◦C.Three points that do not follow the trend in Figure 9 correspond to permeability values obtained at 800 ◦C. At this temperature, important cracking occurs following the already mentioned nonlinear mechanical behavior.
