Download as pdf or txt
Download as pdf or txt
You are on page 1of 10

Cement and Concrete Composites 99 (2019) 62–71

Contents lists available at ScienceDirect

Cement and Concrete Composites


journal homepage: www.elsevier.com/locate/cemconcomp

Effect of aggregate size and inclusion of polypropylene and steel fibers on T


explosive spalling and pore pressure in ultra-high-performance concrete
(UHPC) at elevated temperature
Ye Lia, Pierre Pimientab, Nicolas Pinoteaub, Kang Hai Tana,∗
a
School of Civil and Environmental Engineering, Nanyang Technological University, Singapore, 639798, Singapore
b
CSTB, Centre Scientifique et Technique du Bâtiment, Marne la Vallée, France

A R T I C LE I N FO A B S T R A C T

Keywords: This paper investigates the individual and combined effects of polypropylene (PP) fibers, steel fibers, and ag-
Explosive spalling gregate size on spalling behavior and pore pressure build-up of ultra-high-performance concrete (UHPC) exposed
Ultra-high-performance concrete to elevated temperature. Simultaneous measurements of pore pressure and temperature were conducted at
Pore pressure different depths in UHPC specimens under one-sided heating with a heating rate of 2 °C/min. Compressive,
Elevated temperature
tensile, and permeability tests were performed to analyze spalling behavior. Addition of PP fibers fully prevented
Polypropylene fiber
Steel fiber
spalling and they are much more effective in increasing permeability than steel fibers and larger aggregates. The
Large size aggregate combined use of PP and steel fibers, and PP fibers and larger aggregates showed strong synergistic effect on
increasing permeability. The higher the permeability, the lower was the maximum pore pressure measured in the
samples. Two plateaus were observed from the temperature history due to vaporization of liquid water (between
115 and 125 °C inside the specimens) and release of water vapor (starting from 180 °C), respectively. The second
plateau was identified as the functional temperature of PP fibers. Maximum pore pressures in spalled specimens
were much lower than their tensile strengths, which could imply the contribution of hydraulic pressure in the
region of moisture clog on spalling.

1. Introduction phenomenon is not a material intrinsic property. Spalling is influenced


by many factors such as mechanical and moisture and vapor transport
Fire is one of the most serious risks for tunnels, high-rise buildings, properties of concrete, admixtures in concrete, heating rate, dimensions
and underground structures [1]. Explosive spalling of concrete at ele- of samples, and detailing of structural components, etc. Currently, there
vated temperature is a complicated phenomenon that causes cracking is no widely-held mechanism that can predict explosive spalling accu-
of concrete, spalling of concrete cover, and subsequent exposure of steel rately.
reinforcement to fire, which compromise the load-carrying capacity of Addition of polypropylene (PP) fibers is the mostly accepted method
affected members [2–5]. to mitigate explosive spalling [16–19]. Aggregate size was also found to
Ultra-high-performance concrete (UHPC) has high strength have significant influence on spalling behavior of concrete [1,20], and
(f'c ≥ 150 MPa), high ductility, and enhanced durability [6,7]. It has it has strong interaction with PP fibers [21]. It is generally believed that
been increasingly utilized in improving structural resistance of build- larger aggregates introduce porous interface transition zones (ITZ) in
ings and bridges [8–12]. However, it is particularly vulnerable to ex- normal and high strength concretes. It was proposed by Dale Bentz [22]
plosive spalling under fire condition due to its densely packed micro- that the ITZ can be linked by PP fiber tunnels and create pathways for
structure and very low permeability compared to normal strength release of water vapor at elevated temperature close to the melting
concrete [13,14]. Therefore, it is important to study governing factors point of the PP fibers. In UHPC, however, no significant porous ITZ is
on explosive spalling of UHPC. For example, the French standard NF P observed due to the use of silica fume and low water-to-binder ratio
18–710 [15] states that spalling control is generally based on the in- [7,23]. It was found that inclusion of steel fibers could reduce explosive
troduction of polypropylene fibers in the mixes. However, spalling spalling and enhance fire endurance of HSC columns [3,19]. However,


Corresponding author.
E-mail addresses: LIYE0006@e.ntu.edu.sg (Y. Li), pierre.pimienta@cstb.fr (P. Pimienta), Nicolas.PINOTEAU@cstb.fr (N. Pinoteau),
CKHTAN@ntu.edu.sg (K.H. Tan).

https://doi.org/10.1016/j.cemconcomp.2019.02.016
Received 24 June 2018; Received in revised form 31 January 2019; Accepted 20 February 2019
Available online 23 February 2019
0958-9465/ © 2019 Published by Elsevier Ltd.
Y. Li, et al. Cement and Concrete Composites 99 (2019) 62–71

they may not be very effective in preventing spalling of UHPC with very cylindrical (∅150 mm × 48 mm), and cuboid (300 × 300 × 120 mm3)
dense microstructure [24,25]. The combined use of PP and steel fibers molds for compressive strength, uniaxial tensile, permeability, and
to prevent spalling of fiber-reinforced self-consolidating concrete spalling tests, respectively. Pore pressure gages and thermocouples
(FRSCC) and HSC was investigated [16,17], but their effect on moisture were located on a steel support and inserted into the specimens during
migration and connectivity at micro-scale in UHPC has not been well casting for all the samples. The depths of the sensors were 2, 5, 10, 20,
studied. 30, and 50 mm from the bottom surface of the specimens. During
It is noteworthy that pore pressure is a critical factor leading to casting, the molds were compacted with a mechanical vibrator. After
spalling because it can generate large tensile stresses in heated samples, gradually pouring UHPC into the molds, vibration was continued for
which may exceed tensile strength of concrete [26]. Build-up of pore 1–2 min. No segregation of mixture was observed during casting.
pressure is directly related to migration of moisture, which in turn is Specimens were covered with a plastic sheet at room temperature for
controlled by permeability of concrete [2,3,20,27]. However, quanti- the first 24 h. Afterwards, they were demolded and stored in lime-sa-
tative analysis of build-up of pore pressure is one of the difficulties in turated water at room temperature for 2 weeks. Then the specimens
investigation of spalling. Monitoring of pore pressure in UHPC speci- were stored in air for another 2 weeks (28 days after casting) for
mens during fire exposure is even more challenging. This paper studies compressive and tensile tests and 2.5 months (3 months after casting)
the individual and combined effects of PP fibers, steel fibers, and ag- for spalling tests. Spalling tests have been carried out in CSTB in France.
gregate size on explosive spalling and pore pressure build-up of UHPC Samples were transported between 4 weeks and 8 weeks after casting.
at elevated temperature. Six UHPC mix designs with the size of
300 × 300 × 120 mm3 were prepared and subjected to a unidirectional 2.2. Temperature and pore pressure measurement
heat source to measure temperature and pore pressure simultaneously.
Compressive strength, tensile strength, and residual permeability at As a joint study, the experimental device for measuring pore pres-
ambient temperature were measured to analyze spalling and pore sure and temperature was developed by Kalifa et al. [28] and modified
pressure of the UHPC mixes. by the authors in CSTB, France. The mechanical and permeability tests
of UHPC were conducted at NTU, Singapore. Fig. 1a and b show the
2. Experimental program schematic experimental setup. The specimens were placed at the
opening of the electric furnace. Five gages which could measure pore
2.1. Mix proportions and specimen preparation pressure and temperature simultaneously were placed in all the speci-
mens at the central area of the specimens at 5, 10, 20, 30, and 50 mm
Portland cement (ASIA@ CEM I 52.5 N), natural river sands sieved depths from the heated surface, respectively. A tube without a metal
to 600 μm (reference mixes) or to 5 mm (AG mixes), micro silica sands cup was located at 2 mm from the heated surface for measuring tem-
(median particle size of 130 μm), a highly reactive silica fume (Grade perature. Fig. 1c shows the configuration of the pore pressure gages.
940 from Elkem Microsilica@), the 3rd generation polycarboxylate- The gages were made of round sintered metal plates (∅12 × 1 mm),
based superplasticizer (Sika@ ViscoCrete@-2044), steel fibers of 13 mm which were encapsulated into metal cups and were welded to thin
length and 0.22 mm diameter supplied from Dramix@, and monofila- metal tubes with 1.6 mm inner diameter. The sintered metal with
ment cylindrical PP fibers of 12 mm length and 30 μm diameter sup- evenly distributed micro pores can collect moisture vapor in an evenly
plied by DFL were used to prepare the UHPC mixes. Properties of ce- manner, which leads to stable pressure measurements [14]. The tubes
ment, silica fume, and fibers are provided in Tables 1 and 2. Water-to- emerged from the rear face of the specimens. During testing, connectors
binder ratio was 0.22 to achieve dense packing microstructure. were placed at the free end of the tubes. The first exit of the connectors
Six mix proportions were prepared as shown in Table 3. A plain connected the gages to a piezoelectric pressure transducer through
UHPC mix was used as a control mix design. Water-to-binder ratio (W/ flexible tubes partially filled with silicone oil. Thermocouples
B), silica fume, silica sand, and superplasticizer content were kept (∅1.5 mm) were inserted from the second exit of the connectors to the
constant for all the six mix designs. PP fibers, steel fibers, and aggregate metal plate and were sealed up. As the inner tube diameter is 1.6 mm,
size were the only variables to be investigated. The mix designs were the accessible volume to the fluids is very low. It is to be emphasized
named as Control, PP (PP fibers), ST (steel fibers), AG (larger ag- that as the sintered metal are not filled with oil, these sensors allow to
gregate), PPST (PP + steel fibers), and PPAG (PP fibers + larger ag- measure gas pressures but not liquid pressures. Liquid water may ex-
gregates). PP mix contains 3.0 kg/m3 (0.33% by volume) of PP fibers pand and vaporized when it contacts with the gage head. Two tests
while ST mix has 196.3 kg/m3 (2.5% by volume) of steel fibers. The were carried out for six UHPC mixes (Table 3) except only one test was
maximum aggregate size of AG was 5.0 mm. PPST samples contained conducted on the PP mix design due to an accident. Thermal load was
combined PP and steel fibers, while PPAG consisted of PP fibers and applied on one face (300 × 300 mm2) of the UHPC specimens at a
larger aggregates of maximum size of 5 mm. The other mixes only had heating rate of 2 °C/min. The target temperature was 600 °C.
600 μm of aggregates.
Raw materials were mixed in a Hobart@ planetary mixer. Binders 2.3. Permeability measurement
(cement and silica fume) and fillers (silica sand and sieved river sand)
were dry-mixed for 2–3 min to ensure good dispersion during mixing. After curing, cylindrical samples with diameter of ∅150 mm were
Thereafter, premixed water and superplasticizer were added and mixed grinded from two ends to achieve a disc thickness of 40 mm. Surfaces of
for another 3–5 min until the fresh mortar was homogenous and con- the discs were finely polished to ensure good contact with the perme-
sistent. Fibers were gradually added in 3–5 min and mixed for another ability measuring device. Apparent gas permeability was determined by
2 min. The fresh UHPC mixtures were then cast into cubical means of RILEM-CEMBUREAU method [29]. The apparent permeability
(50 × 50 × 50 mm3), dog-bone shaped (36 × 18 mm2 cross section), was determined based on Darcy's law which was later modified by the
Hagen-Poiseuille relationship [30]:
Table 1
Chemical composition of cement and silica fume. Q 2μLpatm
ka =
A (pi2 − patm
2
) (1)
Compositions (wt.%) CaO SiO2 Al2O3 Fe2O3 SO3 LOI
where ka [m2] is the apparent gas permeability, Q [m3/s] is the gas flow
Cement 67.17 22.14 3.12 2.51 2.13 1.68
Silica fume > 90.0 rate, A [m2] is the cross-sectional area of the specimen, L [m] is the
thickness of the specimen, μ [Pa·s] is the dynamic viscosity of air, pi

63
Y. Li, et al. Cement and Concrete Composites 99 (2019) 62–71

Table 2
Properties of PP and steel fibers.
Diameter (μm) Length (mm) Tensile strength (MPa) Elastic modulus (GPa) Density (kg/m3)

PP fiber 30 12 > 500 >5 910


Steel fiber 220 13 > 2800 200 7850

Table 3 PP fibers addition and change of permeability on migration of water


Mixture proportion of the specimens. vapor. It has been reported that characterization of permeability of
Mix Design PP fiber Steel fiber Relative weight ratio to cement
concrete subjected to 200 °C is pivotal to understanding spalling me-
(kg/m3) (kg/m3) chanism of concrete materials [33] since permeability of UHPC with PP
C AG SF SSΔ SP W/B fibers changed much more significantly below 200 °C and in pore
pressure measurement, evaporation of water vapor mainly happened
C 0.0 0.0 1.0 1.1ˆ 0.25 0.25 0.04 0.22
PP 3.0 0.0 1.0 1.1ˆ 0.25 0.25 0.04 0.22
below 200 °C. Therefore, to perform the permeability test, the speci-
AG 0.0 0.0 1.0 1.1* 0.25 0.25 0.04 0.22 mens were first heated to 200 °C. They were kept for 48 h to remove
PPAG 3.0 0.0 1.0 1.1* 0.25 0.25 0.04 0.22 free water from the specimens. The heating rate was lower than 1 °C/
ST 0.0 196.3 1.0 1.1ˆ 0.25 0.25 0.04 0.22 min to prevent undesirable temperature gradients within the speci-
PPST 3.0 196.3 1.0 1.1ˆ 0.25 0.25 0.04 0.22
mens. After heating, the specimens were cooled down naturally within
C: cement, AG: aggregates (sieved from river sand), SF: silica fume, SS: silica the furnace. Three pressure levels were applied, i.e. 2, 3, and 5 bars
sand, SP: superplasticizer, W/B: water-to-binder ratio. ˆ denotes aggregates (1 bar = 1 atmospheric pressure) to determine intrinsic permeability.
with maximum size of 600 μm, * denotes aggregates with maximum size of Fig. 2 shows the experimental setup for measuring permeability
5 mm, Δ denotes median particle size of 130 μm. [34]. The specimen was secured in a pipe segment sandwiched between
two symmetric lids. Silicone was applied on the lateral surface of the
specimen and in between the sealing rings and the specimen. The two
lids were clamped by twelve bolts around the flange. Compressed air
was supplied into the upper chamber by a compressor and the pressure
was adjusted via a gas pressure regulator varying between 2 and 5 bars.
The air flow rate through the specimen was measured from the outlet.

2.4. Compressive and tensile tests

Three cubes (50 × 50 × 50 mm3) and three dog-bone shaped sam-


ples (36 × 18 mm2 cross section) were prepared to conduct compres-
sive and tensile tests for each mix design, respectively. According to An
et al. [35], cubes with 50 mm length, may have around 10% higher
compressive strength than cubes with 100 mm length. Smaller sample
will also result in higher tensile strength because fibers may be aligned
by smaller mold [36]. Compressive strength measurements were con-
ducted at ambient temperature following ASTM C109/C109M-11 [37].
The hydraulic compression machine had a capacity of 3000 kN. A
constant loading rate of 100 kN/min was adopted and the maximum
force was recorded automatically. Uniaxial tensile strength tests were
carried out by means of an electro-mechanical testing machine with a
tensile capacity of 30 kN. A personal computer was connected to the
testing machine to acquire the test data. The tests were displacement-
controlled at a rate of 0.2 mm/min.
Fig. 1. (a) schematic experimental setup (b) locations of gages (c) configuration
of gages (d) photo of experimental setup. 3. Results and discussions

[Pa] is the inlet pressure, and patm [Pa] is the outlet pressure which is at 3.1. Mechanical properties and permeability of UHPC
atmospheric pressure.
In case of UHPC with a very dense microstructure, the apparent gas To study the influence of PP fiber, steel fiber, and aggregate size on
permeability is influenced by the applied inlet pressure due to slip flow mechanical properties of UHPC, compressive strength and uniaxial
(Knudsen diffusion) [31]. Intrinsic gas permeability was adopted be- tensile strength tests were conducted on three samples, respectively. As
cause it is independent from the inlet pressure to correct the apparent shown in Table 4, the control mix had a compressive strength of
gas permeability [32]. 149.6 MPa and a tensile strength of 8.9 MPa. These values were used as
a comparison basis for the other mix designs. Adding PP fibers only had
b ⎞ marginal effect on compressive strength (159.7 MPa versus 149.6 MPa)
ka = k int ⎜⎛1 + ⎟

⎝ pm⎠ (2) and tensile strength (8.5 MPa versus 8.9 MPa). This is due to the frac-
tion of PP fibers (0.3% by volume) which is much lower compared to
2
where kint [m ] is the intrinsic permeability (equal to ka when mean that of normal fiber reinforced concrete (1% to several percent by vo-
pressure pm approaches infinity), b [Pa] is the Klinkenberg constant, lume) [38]. AG samples showed a slight reduction on compressive and
and pm = (pi + patm )/2 [Pa] is the average value of the inlet and outlet tensile strength because larger aggregates may introduce defects into
pressures on two faces of the heated specimen. UHPC. However, the reduction is not significant due to the river sands
One of the purposes of the present study is to unravel the effects of used as larger aggregates were strong and no distinct porosity can be

64
Y. Li, et al. Cement and Concrete Composites 99 (2019) 62–71

Fig. 2. Device used for permeability tests in NTU.

Table 4 induced cracking during the curing stage of UHPC specimens [43,44]. A
Mechanical properties and permeability of UHPCs. 1040-fold increase in permeability was observed in PPST mix with
Mix Design fc (MPa) ft (MPa) Intrinsic Normalized to combined use of PP fibers and steel fibers. From the ongoing research,
Permeability permeability of the very high permeability of PPAG and PPST is attributed to enhanced
(10−18 m2) C connectivity of PP fiber tunnels by multiple microcracks generated from
thermal expansion and strain incompatibility between steel fibers or
C 149.6 ± 4.8 8.9 ± 0.7 0.733 ± 0.493 1
PP 159.7 ± 5.7 8.5 ± 0.7 13.8 ± 7.0 18.8
larger aggregates with the UHPC matrix.
AG 145.0 ± 4.1 8.0 ± 0.9 1.60 ± 0.37 2.18
PPAG 147.7 ± 1.5 8.2 ± 0.6 1090 ± 70 1490
3.2. Observations of spalling behavior
ST 172.1 ± 3.7 12.4 ± 0.3 0.152 ± 0.073 0.207
PPST 154.8 ± 2.3 11.1 ± 0.5 758 ± 173 1040
Fig. 3 shows the heated faces of UHPC specimens after the spalling
fc: Compressive strength; ft: Uniaxial tensile strength; Permeability: Intrinsic tests. It can be seen that C, AG, and ST specimens with very low per-
permeability after preheating exposure at 200 °C. meability (Table 4) spalled severely, but PP, PPST, and PPAG specimens
with PP fibers did not spall at all. This clearly showed the effect of PP
found in the interfacial transition zone (ITZ). A homogenous mor- fibers in preventing spalling. For C and AG samples, intermittent loud
phology of the paste structure and the compact ITZ structure lead to explosive sounds of spalling were heard during the tests accompanied
high-strength mortar [23]. However, due to the introduced hetero- with intermittent popcorn-cooking like sounds. The debris were thin
geneity, tensile strength actually decreased by 10%. PP fibers and larger disk-shaped, less than 5 mm thick and diameter ranging from 10 to
aggregates did not change the strength of UHPC independently; thus, 150 mm. This should imply that high stresses, moisture accumulated
their combined effect on compressive strength of UHPC is not sig- and/or high pore pressure happened at very shallow depth. Moreover,
nificant as well. Addition of steel fibers significantly increased com- even after the initial spalling, progressive spalling (layer by layer)
pressive strength by 15% (172.1 MPa) and tensile strength by 41% propagated into the deeper, cooler regions of the specimen. Steel fibers
(12.4 MPa) because steel fibers restrict internal material deterioration helped to improve mechanical properties of UHPC, but spalling could
and crack propagation by absorbing developed stresses at the fiber's tip not be prevented. Unlike C and AG specimens, only several loud sounds
and consequently enhanced compressive strength of the samples [39]. were heard when testing ST specimens. The spalled areas were non-
Combined PP and steel fibers increased both compressive and tensile uniform, partly due to non-homogeneous distribution and random or-
strength of UHPC, but the enhancement was not as much as adding steel ientation of steel fibers in the concrete. Due to enhanced mechanical
fibers only. One plausible reason is that more air voids trapped by properties, ST specimens spalled after a longer heating duration and at
higher fraction of mixed fibers reduced the mechanical properties of higher temperature. This will be discussed in detail in Sections 3.3 and
UHPC matrix, thereby weakening the bonding between the matrix and 3.4.
the fibers [40].
For permeability, C showed a very low intrinsic permeability and a 3.3. Results of temperature and pore pressure measurements
relatively large standard deviation due to the limitation of the experi-
mental set up. Sole addition of 3 kg/m3 PP fibers significantly increased Fig. 4 presents temperature and pore pressure history of UHPC mix
the residual permeability of UHPC from 0.733 × 10−18 to designs. Two samples were tested for each mix design, but due to the
13.8 × 10−18 m2, i.e. a 19-fold increase, due to creation of microcracks limited space, only one result of each UHPC mix is presented here. The
caused by thermal mismatch between the PP fibers and the matrix and maximum pressure and temperature of all the samples will be presented
melting of PP fibers [41,42]. Increasing aggregate size only increased in the following sections. In the figures, T and P stand for temperature
permeability of UHPC by about 2-fold because thermal expansion and vapor pressure, respectively. The numbers after them show the
coefficient between aggregates and UHPC matrix are not so significant locations of the sensors. From the temperature history, plateaus be-
and cracks caused by aggregates were finer compared to those created tween 115 °C and 125 °C can be observed in all the UHPC mix designs
by PP fibers. Combined use of PP and larger aggregates showed sy- due to phase change of liquid water which is an endothermic reaction.
nergistic effect and significantly increased permeability by about 1500- It is worth noting that with vaporizing of water, pore pressure kept
fold with respect to the control mix (1490 × 10−18 m2 vs. increasing, which shows that water vapor was not totally released from
0.73 × 10−18 m2). Addition of steel fibers alone slightly reduced the the specimens. Mindeguia et al. concluded that the higher vaporization
intrinsic permeability, which may be due to a reduction of shrinkage- temperature than 100 °C is due to the presence of high capillary

65
Y. Li, et al. Cement and Concrete Composites 99 (2019) 62–71

Fig. 3. Spalling behavior of UHPC specimens.

Fig. 4. Temperature and pore pressure as a function of heating time.

66
Y. Li, et al. Cement and Concrete Composites 99 (2019) 62–71

pressure induced by small pores at the interface between liquid water,


gas phase (vapor and dry air) and concrete matrix [1].
For the un-spalled specimens PP, PPST, and PPAG, a second plateau
between 180 °C and 220 °C can be observed (as marked in Fig. 4b). This
has not been reported in previous literature [1,16,17,27,28,33]. The
second plateaus are due to release of water vapor caused by expansion
and melting of PP fibers (with melting point of 167 °C). When the rate
of vapor release (due to outward vapor flow) exceeded the rate of vapor
flowing into the cooler regions (due to dehydration and transport), pore
pressure started to drop [1,14]. Therefore, the second plateau was ac-
companied by peaks of pore pressure, which were bell-shaped with
smooth transitions between the ascending and the descending branches.
As shown in Fig. 4a, c and e, for the spalled UHPC mixes (C, AG, and
ST), a sharp drop of temperature and pore pressure to zero indicates the
occurrence of spalling. This is due to severe breaking-off of concrete
from the heated surface exposing embedded pressure gages. Im-
mediately, compressed water vapor was suddenly released [28]. For C
and AG (Fig. 4a and c), the first spalling time was 204 and 210 min,
respectively. However, due to enhanced mechanical properties, ST
specimens withstood a higher temperature and spalling took place at Fig. 6. Maximum pore pressure of UHPC mix designs versus depth.
238 min. After spalling had occurred, the temperature curves increased
rapidly and showed perturbations since some of the thermocouples between the UHPC mix designs (Fig. 4).
were directly exposed to the heating elements in the furnace. The dif- Fig. 6 summarizes the mean maximum pore pressures of each mix
ference in temperature and pore pressure history between spalled and design at different locations. The maximum pore pressures of the
un-spalled UHPC mixes suggests that the migration of water inside spalled UHPC mixes (C, AG, and ST) concentrated in shallower regions
UHPC specimens is chiefly responsible for build-up of pore pressure. (less than 20 mm from the heated face). This is because the very low
permeability of these mixes (< 1.60 × 10−18 m2) hindered water vapor
3.4. Discussion on pore pressure and temperature results from migrating into cooler deeper regions of the specimens. Spalling
occurred in a shallow depth and created a new spalled surface, which
Fig. 5 shows the temperature gradients of UHPC mixes when the allowed release of trapped water vapor. The maximum pore pressure of
temperature at 2 mm deep reached 105 °C. Temperature at a later stage 5.2 MPa was observed in the ST specimen at 10 mm from the heated
could not be compared because spalling and vaporization of water surface, while the lowest maximum pore pressure of 1.4 MPa was ob-
strongly affected the temperature distribution in the spalled samples. It served in the AG specimen at the same location. This is probably due to
can be seen that, all the specimens had similar thermal gradients. This the inclusion of steel fibers in the ST sample (12.4 MPa in tension at
implies that thermal properties of UHPC mix designs were not sig- ambient temperature), which helped to resist a greater pore pressure.
nificantly influenced by the inclusion of PP fibers, steel fibers, and However, the AG mix had the lowest tensile strength of 8.0 MPa and the
larger aggregates. According to literature, thermal diffusivity of con- expansion of larger aggregates induced microcracks (Table 4). But this
crete is mainly controlled by the fraction and nature of aggregates [45], effect pales in comparison with PP samples.
which were maintained constant in the current study. Although steel The un-spalled UHPC mixes (PP, PPST, and PPAG) had higher
fibers have higher conductivity (around 50 W/m/°C) than that of con- permeability (> 13.8 × 10−18 m2), so moisture further penetrated into
crete (less than 2 W/m/°C), since the steel fibers are not percolated, the deeper regions and the maximum pore pressure took place in deeper
heat conductivity of specimens is governed by concrete matrix [46]. It layers (30 mm from the heated surface). It can be seen from Fig. 6,
is worth noting that, thermal stresses also depend on thermal dilation, among PP, PPST, and PPAG, the higher the permeability, the lower was
Young modulus and transient thermal strain, etc. Since the thermal the maximum pore pressure measured, since water vapor could be more
gradients are similar, it is believed that the different spalling behavior easily released to the outer surface. Among all the 6 UHPC mix designs,
can mainly be attributed to differences in measured pore pressures PPAG showed the lowest measured maximum pore pressure of 1.1 MPa,
in accordance with its highest permeability (Table 4). This highlights
the effect of permeability on thermo-hydral behavior, in particular, on
the build-up of pore pressure. It has been shown in the authors' ongoing
study and Heo et al.’s [21] research that PP fibers and larger aggregates
have synergistic effect on increasing permeability due to melting of PP
fibers and formation of microcracks. The combined use of PP and steel
fibers reduced the maximum pore pressure (2.8 MPa for PPST versus
3.3 MPa for PP). Based on Bangi and Horiguchi's study [17], more fibers
can introduce more discrete air bubbles during the mixing process,
which may act as discontinuous reservoirs for trapped vapor and con-
densed water. However, the new introduced pore volume may not be
large enough to affect vapor pressure significantly. Moreover, under
high pressure, water vapor can pass through pressure-induced tangen-
tial space (PITS) between the steel fibers and the UHPC matrix. Mi-
crocracks caused by expansion of steel fibers can also help to release
water vapor.
Fig. 7 plots temperature Tpmax, which corresponding to the max-
imum pore pressure measured (Pmax). For the spalled samples C, AG,
and ST, Tpmax indicates the temperature at which spalling occurred.
Fig. 5. Temperature gradients of UHPC mixes. Tpmax of C and ST samples occurred around 250 °C, while Tpmax of AG

67
Y. Li, et al. Cement and Concrete Composites 99 (2019) 62–71

elevated temperature. The fiber tunnels (due to melting of PP fibers)


were well connected by multiple microcracks. The enhanced con-
nectivity of fiber tunnel by multiple microcracks significantly increased
permeability of UHPC. Through the interconnected network moisture
can be transferred to the embedded plates of pressure gages (Fig. 10c).
In addition, it can be observed in Fig. 7 that for all PP fiber-reinforced
mixes, most of the maximum pore pressure occurred at temperature
close to 200 °C. This can be attributed to the functional temperature of
PP fibers described just above.
Fig. 11 compares the maximum measured pore pressures to the
tensile strength of the UHPC mix designs. Based on Eurocode 2, a re-
duction factor of 0.8 for tensile strength at 200 °C was applied [47]. For
the spalled samples (C, AG, and ST), it is obvious that the pore pressures
were far lower than their respective tensile strength. Moreover, Felicetti
et al. found that only 0.24 times of pore pressure contributed to reduced
apparent tensile strength of HPC with 90 MPa compressive strength (i.e.
1.0 MPa pore pressure reduces 0.24 MPa of tensile strength) due to less
interconnected porosity of high grade concrete [48]. Therefore, it is
questionable if spalling could be triggered by pore pressure alone [48]
Fig. 7. Temperature at maximum pressure versus depth.
and if tensile strength is at all a reasonable failure criterion for ex-
plosive spalling. One possible explanation is that pore pressure is not
was lower than those of C and ST because the lower tensile strength the only factor leading to spalling since thermal stresses cannot be
caused by its heterogeneous microstructure (Table 4). For PP, PPAG, neglected. However, from Fig. 5, the specimens in this study had similar
and PPST, most of the maximum pore pressure occurred in a relatively thermal gradients. Another explanation is that spalling was caused by
narrow range of temperature between 180 and 220 °C. This can be build-up of hydraulic pressure in the saturated zone (moisture clog).
identified as the functional temperature of PP fibers. With an increase However, the used gages could not measure liquid pressure (see section
of temperature, permeability also increased due to formation of mi- 2.2). In fully saturated pores, thermal pressurization coefficient of the
crocracks and the PP fibers melting [41]. As a result, water vapor can be pore fluid is equal to 0.6 MPa/°C. Thus, discrepancy between thermal
released from the UHPC specimens. expansion of pore fluid and concrete matrix can lead to a very high
Fig. 8 presents the evolution of pore pressure as a function of build-up of hydraulic pressure [49]. The hydraulic pressure can lead to
temperature. Pore pressure profiles are compared to saturated vapor progressive breakdown of microstructure [50,51]. However, as shown
pressure line (Pvsat) to analyze the moisture state and contribution of in Fig. 10a, due to very low permeability of UHPC matrix, the high
water vapor to total pore pressure. For most spalled specimens (C, AG, pressure in the region of saturated water in the shallower regions might
and ST), pore pressure increased with elevated temperature, but the not be transferred to the pressure gage. This explanation is especially
measured pressures were lower than the Pvsat curve since the initial reasonable for UHPC with very dense microstructure. On the contrary,
moisture content of UHPC was lower than the saturated state. More- for PP, PPST, and PPAG, PP fibers created an interconnected network at
over, low permeability of C, AG, and ST specimens hindered migration elevated temperature. The pressure in the surrounding area can be
and accumulation of water vapor inside UHPC specimens. After spalling transferred to the gage (Fig. 10c). The measured pore pressure is ac-
took place, the pore pressure dropped to nearly 0 MPa accompanied tually the vapor pressure of the area in front of the pressure gage.
with a sudden decrease of temperature. With addition of PP fibers,
water vapor could migrate inward towards the cooler layers of the 4. Conclusion and outlook
specimens and reach the pressure gages. Therefore, the pore pressure of
PP (Fig. 8b) and PPST (Fig. 8f) at 20 mm depth caught up with the Pvsat Individual and combined effects of polypropylene (PP) fibers, steel
curve. The measured pressure for PP even exceeded the Pvsat at 30 mm. fibers, and aggregate size on development of pore pressure and tem-
This could be due to limited space inside a capillary pore, with a high perature in UHPC during heating were investigated in the current
degree of saturation, liquid water occupied a part of the space, so the study. Simultaneous measurements of pressure and temperature at the
free volume available for dry air to expand is lower. As a consequence, same location were conducted at different depths of specimens.
partial pressure of the dry air enclosed in the specimens contributed to Mechanical properties and permeability of UHPC mixes were measured.
total pressure [28]. Results showed that C, ST, and AG specimens had very low permeability
Fig. 9 summarizes the maximum vapor pressure of UHPC mix de- and consequently suffered severe explosive spalling due to their very
signs at different locations as a function of temperature. It can be seen compact microstructure. Addition of PP fibers significantly increased
that, most of the maximum pore pressure measurements of the spalled permeability and suppressed explosive spalling since thermal mismatch
specimens were much lower than the Pvsat (Fig. 9 solid blue circle). This created micro-cracks network and melting PP fibers that helped to re-
may be because moisture accumulated in the shallower regions of the lease trapped water vapor at elevated temperature. Combined use of PP
spalled specimens (C, ST, and AG), but their low permeability hindered and steel fibers, or PP fibers and larger aggregates synergistically in-
the moisture from reaching the pressure gages as shown in Fig. 10a. For creased permeability and completely prevented explosive spalling due
some locations of the ST specimens (Fig. 9 dotted blue circle), the to enhanced connectivity of PP fiber tunnels by micro-crack network
maximum pore pressures were very high and close to the Pvsat curve. caused by strain incompatibility between steel fibers or larger ag-
This can be explained by the formation of microcracks due to high gregates with the UHPC matrix.
vapor pressure and expansion of steel fibers helped to transfer moisture For pore pressure and temperature results, plateaus in the tem-
to the pressure gages (Fig. 10b). On the other hand, the un-spalled perature curve were observed in all the UHPC mixes between 115 °C
specimens with PP fibers had maximum pore pressures closer to the and 125 °C since vaporization of liquid water is an endothermic reac-
Pvsat curve (Fig. 9 red circle). From the authors’ previous study [41], PP tion. For PP, PPST, and PPAG, second plateau in the temperature curve
fibers melted at elevated temperature, thus water vapor can pass accompanied with maximum pore pressure was observed between 180
through the fiber tunnels. Moreover, microcracks will be formed in the and 220 °C due to release of trapped water vapor through inter-
radial direction of the fiber tunnel due to expansion of polypropylene at connecting micro-crack network and fiber tunnels.

68
Y. Li, et al. Cement and Concrete Composites 99 (2019) 62–71

Fig. 8. Pressures as a function of temperature plotted with the saturated vapor pressure curve.

Fig. 9. Maximum vapor pressure as a function of temperature.

69
Y. Li, et al. Cement and Concrete Composites 99 (2019) 62–71

Fig. 10. Possible mechanism around pressure gages during heating (a) without fibers (b) with steel fibers, and (c) with PP fibers.

Acknowledgements

This material is based on research/work supported by the Land and


Liveability National Innovation Challenge under L2 NIC Award No.
L2NICCFP1-2013-4. The authors thankfully acknowledge the support
received from Dr. Jihad MIAH, Dr. Romain MEGE, and Mr. Pierre-Jean
DEGIOVANNI during the experiments at CSTB.

References

[1] J.-C. Mindeguia, et al., Temperature, pore pressure and mass variation of concrete
subjected to high temperature — experimental and numerical discussion on spalling
risk, Cement Concr. Res. 40 (3) (2010) 477–487.
[2] E.W.H. Klingsch, Explosive Spalling of Concrete in Fire, ETH, Zurich, 2014.
[3] V. Kodur, L. Phan, Critical factors governing the fire performance of high strength
concrete systems, Fire Saf. J. 42 (6) (2007) 482–488.
[4] Q. Ma, et al., Mechanical properties of concrete at high temperature—a review,
Constr. Build. Mater. 93 (2015) 371–383.
[5] D. Gawin, et al., Effect of damage on permeability and hygro-thermal behaviour of
HPCs at elevated temperatures: Part 1. Experimental results, Comput. Concr. 2 (3)
Fig. 11. Maximum pore pressure and tensile strength of UHPC mix designs. (2005) 189–202.
[6] D. Wang, et al., A review on ultra high performance concrete: Part II. Hydration,
microstructure and properties, Constr. Build. Mater. 96 (2015) 368–377.
[7] C. Shi, et al., A review on ultra high performance concrete: Part I. Raw materials
The maximum pore pressure of the spalled specimens was much
and mixture design, Constr. Build. Mater. 101 (2015) 741–751.
lower than the Pvsat curve because the initial moisture content of UHPC [8] S. Yang, et al., Influence of aggregate and curing regime on the mechanical prop-
was lower than the saturated state and the pressure gages cannot detect erties of ultra-high performance fibre reinforced concrete (UHPFRC), Constr. Build.
liquid pressure in the region of saturated zone. With addition of PP Mater. 23 (6) (2009) 2291–2298.
[9] M. Schmidt, E. Fehling, Ultra-high-performance concrete: research, development
fibers, moisture could penetrate into the deeper regions of the speci- and application in Europe, ACI Spec. Publ. 228 (2005) 51–78.
mens and accumulate. Therefore, the maximum pore pressure of PP, [10] J. Resplendino, First recommendations for ultra-high-performance concretes and
PPST, and PPAG occurred in deeper region of the specimens and caught examples of application, International Symposium on Ultra High Performance
Concrete, 2004.
up with the Pvsat curve. [11] P. Buitelaar, Heavy reinforced ultra high performance concrete, Proceedings of the
The maximum pore pressure measured was much lower than tensile Int. Symp. On UHPC, Kassel, Germany, 2004.
strength of corresponding UHPC mix designs, so that tensile strength [12] E. Denarié, E. Brühwiler, Structural rehabilitations with ultra high performance
fibre reinforced concretes, Int. J. Restor. Build. Monum. 12 (2006) 453–465 MCS-
may not be an adequate reasonable failure criterion for explosive ARTICLE-2007-011.
spalling. One possible reason is that spalling is caused by a build-up of [13] V. Kodur, Spalling in high strength concrete exposed to fire—concerns, causes,
hydraulic pressure in the region of moisture clog, or within closed critical parameters and cures, Proceedings, ASCE Structures Congress, 2000
(Philadelphia, PA).
pores. The UHPC matrix between the closed pores with super-heated
[14] M.R. Bangi, T. Horiguchi, Pore pressure development in hybrid fibre-reinforced
water and the open pores can be destroyed by a step pressure differ- high strength concrete at elevated temperatures, Cement Concr. Res. 41 (11) (2011)
ence, which may lead to progressive spalling of concrete layers. 1150–1156.
[15] NF P 18-710 Calcul des structures en béton – Règles spécifiques pour les bétons
The method used in the present study is the most widely accepted
fibrés à ultra-hautes performances (BFUP), AFNOR, Paris, 2016.
approach to measure pore pressure in heated concrete [16] Y. Ding, et al., Influence of different fibers on the change of pore pressure of self-
[1,14,16,17,28,33,48,52–55]. However, limitation of the current device consolidating concrete exposed to fire, Constr. Build. Mater. 113 (2016) 456–469.
is that it can only measure gas pressure in connected porous media, but [17] M.R. Bangi, T. Horiguchi, Effect of fibre type and geometry on maximum pore
pressures in fibre-reinforced high strength concrete at elevated temperatures,
hydraulic pressure caused by liquid water cannot be measured. More- Cement Concr. Res. 42 (2) (2012) 459–466.
over, the maximum pore pressure might happen between two pressure [18] G.-F. Peng, et al., Explosive spalling and residual mechanical properties of fiber-
gages. To better study explosive spalling, more accurate testing toughened high-performance concrete subjected to high temperatures, Cement
Concr. Res. 36 (4) (2006) 723–727.
methods are needed to detect liquid pore pressure during spalling test. [19] V. Kodur, et al., Effect of strength and fiber reinforcement on fire resistance of high-
strength concrete columns, J. Struct. Eng. 129 (2) (2003) 253–259.
[20] G.A. Khoury, Effect of fire on concrete and concrete structures, Prog. Struct. Eng.
Mater. 2 (4) (2000) 429–447.
Disclaimer [21] Y.-S. Heo, et al., Relationship between inter-aggregate spacing and the optimum
fiber length for spalling protection of concrete in fire, Cement Concr. Res. 42 (3)
Any opinions, findings, and conclusions or recommendations ex- (2012) 549–557.
[22] D.P. Bentz, Fibers, percolation, and spalling of high-performance concrete, ACI
pressed in this material are those of the author(s) and do not necessarily
Mater. J.-Am. Concr. Inst. 97 (3) (2000) 351–359.
reflect the views of the L2 NIC. [23] C. Wang, et al., Preparation of ultra-high performance concrete with common
technology and materials, Cement Concr. Compos. 34 (4) (2012) 538–544.

70
Y. Li, et al. Cement and Concrete Composites 99 (2019) 62–71

[24] K. Hertz, Heat-induced Explosion of Dense Concretes, (1984). [41] Y. Li, K.H. Tan, E.-H. Yang, Influence of aggregate size and inclusion of poly-
[25] K.D. Hertz, Danish investigations on silica fume concretes at elevated temperatures, propylene and steel fibers on the hot permeability of ultra-high performance con-
ACI Mater. J. 89 (4) (1992). crete (UHPC) at elevated temperature, Constr. Build. Mater. 169 (2018) 629–637.
[26] G.A. Khoury, et al., Modelling of heated concrete, Mag. Concr. Res. 54 (2) (2002) [42] D. Zhang, A. Dasari, K.H. Tan, On the mechanism of prevention of explosive spalling
77–101. in ultra-high performance concrete with polymer fibers, Cement Concr. Res. 113
[27] M. Ozawa, H. Morimoto, Effects of various fibres on high-temperature spalling in (2018) 169–177.
high-performance concrete, Constr. Build. Mater. 71 (2014) 83–92. [43] H.S. Wong, et al., Influence of the interfacial transition zone and microcracking on
[28] P. Kalifa, F.-D. Menneteau, D. Quenard, Spalling and pore pressure in HPC at high the diffusivity, permeability and sorptivity of cement-based materials after drying,
temperatures, Cement Concr. Res. 30 (12) (2000) 1915–1927. Mag. Concr. Res. 61 (8) (2009) 571–589.
[29] J. Kollek, The determination of the permeability of concrete to oxygen by the [44] W. Sun, et al., The effect of hybrid fibers and expansive agent on the shrinkage and
Cembureau method—a recommendation, Mater. Struct. 22 (3) (1989) 225–230. permeability of high-performance concrete, Cement Concr. Res. 31 (4) (2001)
[30] C. Gallé, J. Sercombe, Permeability and pore structure evolution of silicocalcareous 595–601.
and hematite high-strength concretes submitted to high temperatures, Mater. [45] Z.P. Bažant, M.F. Kaplan, Z.P. Bazant, Concrete at High Temperatures: Material
Struct. 34 (10) (2001) 619–628. Properties and Mathematical Models, (1996).
[31] H. Sun, et al., Gas transport mode criteria in ultra-tight porous media, Int. J. Heat [46] S.D. Abyaneh, H.S. Wong, N.R. Buenfeld, Simulating the effect of microcracks on
Mass Transf. 83 (2015) 192–199. the diffusivity and permeability of concrete using a three-dimensional model,
[32] L. Klinkenberg, The permeability of porous media to liquids and gases, Drilling and Comput. Mater. Sci. 119 (2016) 130–143.
Production Practice, American Petroleum Institute, 1941. [47] 1992-1-2, E., Eurocode2: Design of Concrete Structures-Part 1-2: General Rules-
[33] P. Kalifa, G. Chene, C. Galle, High-temperature behaviour of HPC with poly- Structural Fire Design, (1995) ENV 1992-1-2.
propylene fibres: from spalling to microstructure, Cement Concr. Res. 31 (10) [48] R. Felicetti, F. Lo Monte, P. Pimienta, A new test method to study the influence of
(2001) 1487–1499. pore pressure on fracture behaviour of concrete during heating, Cement Concr. Res.
[34] E.W. Klingsch, et al., Tests on the hot and residual permeability of concrete at high 94 (2017) 13–23.
temperatures, in: F. Dehn (Ed.), 4th International Workshop on Concrete Spalling [49] S. Ghabezloo, J. Sulem, J. Saint-Marc, The effect of undrained heating on a fluid-
Due to Fire Exposure, MFPA Leipzig, Leipzig, 2015, pp. 65–75. saturated hardened cement paste, Cement Concr. Res. 39 (1) (2009) 54–64.
[35] M.-z. An, L.-j. Zhang, Q.-x. Yi, Size effect on compressive strength of reactive [50] V. Petrov-Denisov, L. Maslennikov, A. Pitckob, Heat-and moisture transport during
powder concrete, J. China Univ. Min. Technol. 18 (2) (2008) 279–282. drying and first heating of heat resistant concrete, Concr. Reinf. Concr. 2 (1972)
[36] D.-Y. Yoo, N. Banthia, Mechanical properties of ultra-high-performance fiber-re- 17–18.
inforced concrete: a review, Cement Concr. Compos. 73 (2016) 267–280. [51] K.D. Hertz, Limits of spalling of fire-exposed concrete, Fire Saf. J. 38 (2) (2003)
[37] A.S.T. Mater, ASTM C 109/C 109 M-11, Standard Test Method for Compressive 103–116.
Strength of Hydraulic Cement Mortars (Using 2-in. Or 50-mm Cube Specimens), [52] M. Ozawa, et al., Study of mechanisms of explosive spalling in high-strength con-
(2011). crete at high temperatures using acoustic emission, Constr. Build. Mater. 37 (2012)
[38] H.R. Pakravan, M. Latifi, M. Jamshidi, Hybrid short fiber reinforcement system in 621–628.
concrete: a review, Constr. Build. Mater. 142 (2017) 280–294. [53] M. Ozawa, et al., Behavior of ring-restrained high-performance concrete under
[39] S. Abbas, A.M. Soliman, M.L. Nehdi, Exploring mechanical and durability properties extreme heating and development of screening test, Constr. Build. Mater. 162
of ultra-high performance concrete incorporating various steel fiber lengths and (2018) 215–228.
dosages, Constr. Build. Mater. 75 (2015) 429–441. [54] J.-C. Mindeguia, et al., On the influence of aggregate nature on concrete behaviour
[40] K. Wille, et al., Ultra-high performance concrete and fiber reinforced concrete: at high temperature, Eur. J. Environ. Civ. Eng. 16 (2) (2012) 236–253.
achieving strength and ductility without heat curing, Mater. Struct. 45 (3) (2011) [55] J.C. Mindeguia, et al., Experimental discussion on the mechanisms behind the fire
309–324. spalling of concrete, Fire Mater. 39 (7) (2015) 619–635.

71

You might also like