Concrete spalling sensitivity versus microstructure : Preliminary results on the effect of polypropylene fibers

The phisyco-mechanical processes triggering concrete explosive spalling are related to the heat-induced microand meso-structural changes. To have new information on concrete properties at the microstructural level, as well as on how concrete spalling sensitivity is affected by polypropylene and steel fibers, and by aggregate type, ordinary and high-performance concretes are investigated in this research project, after being heated to different temperatures. The focus is on the relationship among porosity, vapor permeability, pore pressure and microcracking inside the cementitious matrix. Polypropylene fibers are shown to increase the total porosity, to favor microcracking and to reduce significantly pore pressure, to the advantage of concrete resistance to explosive spalling, whose risk is markedly reduced – or even zeroed.


INTRODUCTION
Explosive spalling is a critical and still highly-debated issue, especially in high-performance concrete (HPC), which is particularly sensitive to spalling because of its denser matrix with respect to ordinary or normal-strength concrete (NSC), because of HPC's lower water/binder ratio, that is one of the conditions required to increase concrete compressive strength [1].Spalling ensues from different physico-mechanical phenomena interacting during the heating process: (a) pore-pressure increase due to water vaporization; (b) self-stresses induced by the thermal gradients; and (c) load-induced stresses.While points (b) and (c) are strictly related to geometrical and structural configuration, point (a) has to do with concrete microstructure (and more specifically, to pore structure).The high spalling sensitivity of HPC, in fact, derives mostly from mass transport properties of its matrix, which is denser and finer than that of NSC.Consequently, water and vapor expulsion is obstructed, and pore-pressure release is impeded [2].Polypropylene-pp fibers are commonly added to HPC mixes to decrease spalling sensitivity by reducing pore pressure.This is possible because pp fibers melt and degrade above 165 • C [3], and generate not only new voids (corresponding to their volume), but create a network of micropores connecting the already-existing pores.In such a way, continuous channels are formed, that allow the vapor to migrate inside the material, towards cooler and lower-pressure zones, and the risk of explosive spalling is reduced [4][5][6].On the other hand, it is well known that adding steel fibers (sf) increases concrete toughness in the post-peak range, and reduces spalling sensitivity too, even if less efficiently than pp fibers [7].(Steel fibers, however, prevent the pieces from breaking off the concrete mass).
In this study, the attention is focused on the microstructural changes induced by high temperature in HPC.A number of different-grade mixes is considered, with/without pp or steel fibers, that come in different types and amounts.The aim is to link material's microstructure to the physical and mechanical properties in order to shed new light on the interaction between concrete behaviors at the micro-and macro-scale, with particular reference to pore pressure and spalling.In this way, mix optimization to  limit the risk of concrete spalling becomes viable, and the designer will eventually be enabled to control such a complex phenomenon as spalling.

Materials and concrete mix designs
This experimental project includes 10 concrete mixes, differing for their compressive strength, type and content of fibers, as well as for the aggregate type.The cement type and class were selected in order to keep the volumetric fraction of the cementitious matrix as constant as possible.CEM I, 42.5 R was used in combination with Ground Granulated Blast Furnace Slag (GGBS), as shown in Table 1.Three concrete grades are considered with f cm,cube ≥ 45, 70, 95 MPa (determined at 28 days on 100 mm-side cubes), whose mix design includes silico-calcareous (SC) aggregate (Mixes 45-S, 70-S, 95-S).For the intermediate strength class (f cm, cube ≥ 70 MPa), pp and steel fibers have been added.Polypropylene fibers are either monofilament fibers (mnf, L = 12 mm; eq = 20 m, extruded straight fibers treated with a surfactant agent), designed specifically to prevent concrete spalling [5], or fibrillated fibers (fbr, L = 12 mm, eq = 48 m, straight fibers obtained by longitudinally cutting pp stripes) generally used to limit early-age shrinkage [8].Monofilament fibers come in three contents: 0. The aggregate-type role was evaluated by comparing Mixes 70-S and 70-C, with silico-calcareous and calcareous aggregates, respectively; both mixes had no fibers.The ten mixes and their properties at the fresh state, as well as their compressive strength at 28 days, are reported in Table 1 (see [9][10][11] for the testing procedures).In the mix designs, the binder is a combination of CEM I and GGBS, except in Mix 45-S, where GGBS is replaced with a calcareous filler.The surfactant agent covering the monofilament pp fibers is responsible for the decrease of the density at increasing values of the fiber content (see Table 1).Cubes with 100 mm-side were cast to investigate the microstructure and to IWCS 2013 measure pore pressure, while cylinders (Ø = 150 mm, h = 300 or 50 mm) were used to evaluate the compressive strength at 28 days or to measure vapor permeability.
All specimens were cured in water for 28 days according to [12].The cubes were later kept at 10 • C for further 28 days, in order to slow down the hydration process.

Microstructural characterization in residual conditions
In order to characterize the mixes from the microstructural point of view and to investigate heat-induced effects, several experimental techniques were used: Scanning (Ø = 20 mm, h = 100 mm) suitable for microstructural analysis were cored from a number of concrete cubes.The tests were performed in residual conditions considering five target temperatures: 20 (reference temperature), 105, 250, 500 and 750 • C. The specimens were heated inside an electric furnace (heating rate = 1 • C/min); once the target temperature was reached, that temperature was kept constant for 2 hours, followed by the cooling phase (coolingrate = −0.25 • C/min down to 200 • C and then natural cooling inside the closed furnace).So far, the tests have been performed on Mixes 45-S, 70-S, 70-S-Pm1, 70-S-Pm2 and 95-S; as for the remaining mixes, the analyses are in progress.

Physical and mechanical characterization
Concrete samples were characterized from the physico-mechanical point of view by measuring pore pressure and permeability on virgin specimens (20 • C, at 28 days, according to [13]).Pore pressure was measured during the heating process (see Section 3.3), in order to identify a possible relationship between the spalling sensitivity (that is related to the maximum pore pressure) and either the concrete porosity or the concrete vapor permeability.A special experimental set-up was designed and built [14].

Mercury intrusion porosimetry (Mixes 2, 6, 7)
The diagrams of the cumulative pore volume and of the relative volume obtained by means of MIP are plotted in Fig. 1 for three concrete mixes, as a function of pore radius.For the sake of comparison, in Fig. 1 reference is made to Mix 70-S with and without fibers (Mixes 70-S, 70-S-Pm 1 and 70-S-Pm 2).The porosity of the specimens treated at 105 • C is considered as a reference, since the evaporable water has been expelled; the porosity at 20 • C refers to room-conditioned specimens.Figure 1 refers to 70-S series and shows that increasing temperature brings in increasing values of the cumulative pore volume.At the same time, adding pp fibers brings in higher values for the cumulative pore volume, at any temperature.In Fig. 1b,d,f, the diagrams of pore-size distribution, for different temperatures, move towards larger-radius pores at increasing temperatures, because of the progressive expulsion of both free and bound water, with consequent increase in pore size.The most significant results concern the pore distribution in the mixes containing pp fibers: microporosities (whose size is of the same order of magnitude as the diameter of monofilament pp fibers) appear in Mixes 70-S-Pm1 and 70-S-Pm2, after being heated to 250 and 500 • C as a consequence of the melting/degradation of the fibers.The higher the pp-fiber content, the higher the relative volume peaks and the porosity generated by the thermal treatment, as discussed in Section 2.1.Figure 2 shows the values of the total porosity as a function of the temperature for the different mixes.As expected, higher-grade concretes have lower porosity at any temperature.Polypropylene fibers induce a significant increase of the porosity above 250

Scanning electron microscopy (Mixes 1, 2, 3, 6, 7)
The images obtained by means of SEM after concrete heating are shown in Fig. 3.In the case of polished sections the technique based on Back Scattered Electrons -BSE was used, while fractured surfaces were examined by means of Secondary Electrons -SE.Figures 3a-d show that microcracking occurs after heating to 500 • C. In addition, cracking in Mix 45-S (Fig. 3a) is hardly distributed, being mainly located at the aggregates-matrix interface.
In Mixes 70-S and 95-S (Figs.3b,c), cracking appears more distributed and extended to the cementitious matrix.Adding fibers seems to make cracks more distributed (Fig. 3d).Moreover, observing Mix 70-S-Pm2 after heating to 250 • C (Fig. 3f), the spaces originally occupied by pp fibers appear empty, because of the melting and subsequent disappearance of the fibers, and a few new cracks propagate through the channels left free by the fibers.

X-ray diffraction and thermogravimetric analysis (Mixes 1, 2, 3, 6, 7)
The XRD curves pertaining to 5 mixes in virgin conditions (no heating, T = 20 • C) are plotted in Fig. 4a, while in Fig. 4b the curves pertaining to Mix 70-S are grouped together from 20 to 750 • C. In Fig. 4a, the crystalline phases of the aggregate can be recognized, and specifically quartz, dolomite and calcite.The portlandite content decreases for increasing values of GGBS content; at the same time, the compressive strength increases (as should have been expected).It is worth noting that no fiber influence has been detected on cement hydrated phases.As shown in Fig. 4b, increasing the temperature leads to the transformation of portlandite into lime (CaO), due to the dehydration process taking place between 400 and 500 • C; lime comes also from the decarbonation of calcite above 650 • C [15].XRD analysis shows no -quartz (above 750 • C) because of the slow cooling of the concrete to room temperature, that turns -quartz into -quartz (due to process reversibility).The transformations resulting from the crystalline phases can also be observed in TGA curves.In dehydration; after 700 • C, a more significant mass loss occurs, because of the decarbonation of calcite, as already shown by XRD analysis.
In Fig. 5b it is worth noting that the mass-loss rate has a maximum between 100 and 200 • C (due to the expulsion of both the free and the bound water).The same occurs for the mass loss due to portlandite dehydration and calcite decarbonation, that tend to grow weaker above 500 • C and 750 • C, respectively.

Porosity, vapor permeability and pore pressure
The experimental results concerning the vapor permeability of the ten mixes show that the lower the compressive strength, the higher the vapor permeability, in agreement with the porosity trend (see Fig. 6).Adding pp fibers does not significantly affect the vapor permeability with respect to the plain mix (Mix 70-S).
Pore pressure was measured on 100 mm-side cubes with the heating and measuring system developed in [14].Four mixes were investigated (Mixes 70-S, 70-S-Pm 1, 70-S-Pm 2 and 95-S).The pore pressure was measured during the heating process until the maximum value was detected.In order to minimize thermal stresses, the heating rate was rather small (0.5 • C/min).The maximum pressure occurred in a narrow thermal range (180-220 • C, [14]).
Since the evolution of pore pressure during the heating process is controlled by vapor permeability (according to Darcy's law [16]) and vapor diffusion (according to Fick's law [16]), both mass transport processes being related to concrete porosity, it is rather interesting to plot the maximum values of pore pressure as a function of total porosity (at 200 • C, see Fig. 6a) and vapor permeability (at 20 • C, see Fig. 6b).In Fig. 6a the experimental values can be easily fitted by means of a hyperbolic curve with appreciable statistical significance; this confirms the major role played by the total porosity in moisture transport and vaporization, as well as on concrete strength [17].(For the mathematical formulation of the curve and the R 2 value, see the inserts inside Fig. 6a). Figure 6b shows that the decrease of pore pressure is consistent with the increase of vapor permeability at 20 • C only for plain mixes (Mixes 70-S and 95-S), while for fiber-reinforced mixes (Mixes 70-Pm1 and 70-Pm2) the vapor permeability at 20 • C does not seem to be a proper parameter to describe the decrease of the peak pressure in the pores as a function of the concrete type.This fact can be ascribed to the much more significant increase of the permeability with the temperature in fiber-reinforced concrete compared to plain concrete (due to the melting and degradation of the fibers -above 165 • C -and to the formation of new microcracks starting from the voids left by the fibers [4]).Unfortunately, vapor permeability in heated specimens has not been measured so far in this project, but tests have been planned and will be performed shortly.

Figure 2 .Figure 3 .
Figure 2. Total porosity as a function of the temperature.

11 Figure 6 .
Figure 6.Peak values of the pore pressure as function of concrete porosity at 200 • C (a); and of vapor permeability at room temperature (b).
Electron Microscopy (SEM): both polished sections and fractured surfaces were examined by means of a Zeiss EVO MA15 (thermoionic source in Lanthanum hexaboride LaB 6 ); Mercury Intrusion Porosimetry (MIP) by means of Pascal 140 and 240 porosimeters (Mercury intrusion rate = 5 and extrusion rate = 6); X-Ray Diffraction (XRD): the analyses were performed by using a Bruker D8 Advance Diffractometer; data acquisition was performed for different values of the diffraction angle 2 comprised between 5 • a 70 • with increments of 0.02 • and a time steps of 0.4 seconds; Thermogravimetric Analysis (TGA): thermal ramp between 25 • C and 975 • C with heating rate 10 • C/min, in nitrogen atmosphere; test set-up: Mettler Toledo TGA/SDTA851 e .First of all, small cylinders