Thermal Stability of Fibers and Cement Pastes

One part of the proposed theory for the performance enhancement of polypropylene fibers in HPC exposed to fire is that the fibers burn out, providing a convenient escape pathway for the water vapor released during the thermal decomposition of the hydrated cement paste present in the concrete. Here, thermogravimetric analysis (TGA) will be applied to investigate this hypothesis more closely. Both the isotactic and atactic versions of polypropylene thermally decompose in the temperature range of 250 ºC to 450 ºC [41]. The decomposition products are reported to be a variety of hydrocarbons, with the major components being propylene, and pentene and heptene derivatives. Figure 7 provides the results of a high resolution TGA analysis (employing a nominal scan rate of 20 ºC/min) of fibers from two commercial manufacturers. The results are consistent with the literature, with substantial mass loss occurring at about 250 ºC and thermal decomposition being completed around 400 ºC.

Figure 7: Thermogravimetric analysis of fibers from two manufacturers.

These TGA curves can be contrasted against those obtained for the mortar components of an ordinary concrete (w/c = 0.5, no silica fume) and a high performance concrete (w/c = 0.23, 10 % condensed silica fume (CSF)), shown in Figure 8. For both concretes, a significant mass loss is observed in the temperature range of 100 ^oC to 250 ^oC, corresponding to the loss of water from the C-S-H gel and aluminate hydration products such as ettringite [7]. For the w/c = 0.5 concrete, a second significant mass loss occurs in the temperature range of 450 ºC to 550 ºC, corresponding to the loss of water from the calcium hydroxide (CH). This loss is not observed in the w/c = 0.23, 10 % CSF concrete, because the CH formed in this system has reacted with the silica fume to form pozzolanic C-S-H, which loses its water in the same temperature range as the primary C-S-H [7]. These results, taken together with those in Figure 7, suggest that when most of the water is released in a high-performance concrete, the polypropylene fibers would still be locally present in the system and the most probable escape pathway would be through the percolated ITZ regions (including those surrounding the fibers). However, two further points must be kept in mind. The first is that during exposure to a fire, a fairly sharp temperature gradient exists through the thickness of the concrete element [2]. Thus, when pressure is building at a depth of several centimeters within the concrete, the surface layer of the concrete is certainly at a temperature where the polypropylene fibers would be completely degraded. It would be of interest to ºquenchº an ongoing fire test and examine the distribution of fibers as a function of depth within the concrete. Second, the polypropylene fibers do soften at around 150 ºC [42], so that the possibility also exists that they are flowable and can be "pushed" out of the concrete or at least locally out of the "fiber channel" by the exiting steam. This suggests that within a fiber-reinforced concrete exposed to a fire, there are at least three different property gradients of relevance: temperature, moisture, and fiber content.

Figure 8: Thermogravimetric analysis of two concrete mixtures.

To further investigate the fiber softening, the viscosity of the polymer melt has been measured at a temperature of 225 ºC, using equipment conventionally employed for measuring the flow properties of asphalts. For a shear rate of about 1.2 s^-1, the measured viscosity of the polymer melt was on the order of 1000 Pa$\cdot$s. This is in reasonable agreement with values presented in the literature for linear low-density polyethylene [43], where a viscosity of 4000 Pa$\cdot$s was measured at a temperature of 200 ºC and a shear rate of 1 s^-1. Using this viscosity and the Hagen-Pouiselle equation for flow in a tube, one can estimate the time required for the polymer melt to flow through the fiber path (approximately cylindrical) as a function of the applied pressure. The time, t, to empty a fiber is given by:

$$t=\frac{V_{fiber}}{Q} = \frac{8 \mu L^2}{\Delta P r^2}$$ (1)

where V_fiber is the volume of the fiber, Q is the volumetric flow rate, µ is the polymer melt viscosity, L is the fiber length, r its radius, and $\Delta P$ is the pressure drop across the fiber length.

Computer modelling [5,6] and experimental measurements [6] have suggested that pressures on the order of 1 MPa to 4 MPa can be created at the depths where spalling typically occurs in a high-performance concrete exposed to a fire test. Substituting this range of values, along with the appropriate fiber geometrical parameters (L = 20 mm and r = 0.125 mm), into equation 1, one can calculate that between 50 seconds and 200 seconds of time would be required for the fiber to empty. If the fiber diameter were reduced from 0.25 mm to 0.1 mm, this time would increase by a factor of four, according to equation 1. This "removal" time would apply for a fiber attached to the surface of the concrete. For an interior fiber, the polymer melt would also need to flow through the capillary pore system (since the fibers themselves are not percolated). Using a pore diameter of 10 µm, a conservative estimate for this calculation, one arrives at a time of 2.25 hours to 9 hours for the polymer melt to flow a distance of 10 mm. These approximate calculations suggest that only those fibers in contact with the exterior surface of the concrete would be able to flow out of the concrete during a fire exposure. Since spalling often occurs at depths between 5 mm and 40 mm [5], the removal of these surface fibers, due either to complete burn out or flow of the melted fibers, could substantially reduce the spalling susceptibility of an HPC exposed to fire. Conversely, those fibers whose flow path for egress includes a portion of the capillary pore network will likely remain in place, but once melted could flow into and be absorbed by the surrounding cement paste matrix.

To further investigate this absorption process, mortar specimens (sand/cement = 2.02, w/c = 0.36) containing 0.33 mass percent fibers per gram of cement were prepared and were subjected to constant temperature heat treatments of two hours at either 150 ºC or 200 ºC. For the specimen heated to 150 ºC, the fibers basically remained intact and were readily observed in the interior of the specimens. However, for the specimen heated to 200 ºC, extremely few fibers were present intact after the heat treatment. The empty fiber channels, however, could be readily observed on fracture surfaces created by breaking the specimen. These empty channels, in cooperation with the ITZs, should provide a pathway for the exit of the saturated water vapor generated during the fire exposure. Since the maximum internal pressures generated within an HPC are typically characterized by a local temperature in the range of 200 ºC to 250 ºC, the "disappearance" of the fibers at a temperature near 200 ºC is quite fortuitous. This observation may provide a possible explanation for the superior performance of polypropylene fibers relative to comparable additions of steel fibers (which do not "burn out") concerning fire performance [5].