Water Content Profile On Fire Spalling
Water Content Profile On Fire Spalling
Water Content Profile On Fire Spalling
net/publication/364352618
CITATIONS READS
0 168
5 authors, including:
All content following this page was uploaded by Bruno Fernandes on 18 October 2022.
ABSTRACT
INTRODUCTION
In any case, water content and water content profile play a vital role in the spalling risk
assessment. This paper aims to analyse the influence of different water content profiles on
spalling. To this, fire tests were carried out on concrete made with natural aggregate (NA).
Before fire tests, initially saturated concrete prisms were dried at 80°C for different drying
periods (from 12 hours to 7 days) to induce different water content profiles throughout the
sample thickness, especially along the first centimeters of the exposed surface. After the
drying period, concrete prisms were subjected to the ISO 834-1 [11] fire curve and a constant
uniaxial compressive stress. Small concrete cylinders underwent the same drying procedure
to determine the water content at different depths of the samples. For postfire assessment,
spalling volume and depths were evaluated through digital photogrammetry. Lastly, spalling
was examined in light of the different water content profiles.
EXPERIMENTAL PROGRAM
Concrete mixes
One mix was produced: concrete made with natural aggregates (NA). Cement was CEM
II/A-L 42.5R by Eqiom, limestone filler was Betocarb HP-SC by Omya, and superplasticiser
was ViscoCrete Tempo-483 by SIKA. Fine aggregate was alluvial sand with a fineness
modulus of 3.10. The NA was made from diorite (Genouillac, France). Table 1 lists the
material properties.
Concrete was designed to meet NF EN 206/CN:2014 [12] durability requirements for XD3
exposure class. Table 2 presents the mixture proportions for all mixes. For the mixing
procedure, first aggregates and cement were mixed for 2 minutes. Then, water and SP were
added and mixed for two minutes.
Fresh and hardened properties are listed in Table 2. For each mix, cylindrical specimens (11
x 22 cm) were cast to determine the compressive strength at 28 days (NF EN 12390-3 [13]).
For the fire tests, prisms with dimensions of 20 x 20 x 10 cm (height x length x width) were
cast. In addition, prisms of 15 x 15 x 10 cm (height x length x width) were cast for the water
content study. Samples were then kept submerged in water between 70 and 90 days, then
subjected to the drying process (later specified).
Table 2 Mix design (in kg/m3) and properties.
Material Mix
Cement 350
Filler 60
Sand 804.3
NA 4/10 331.7
NA 10/20 711.1
Water 175
SP (% of cement) 0.9
Slump (mm) 195
fc28 (MPa) 47.0 (± 3.2)
A specific workflow was proposed to estimate the water content profile. From the saturated
prisms of 15 x 15 x 10 cm, small concrete cylinders of ø 1 x 10 cm were cored (Figure 1a).
Then, the curved surface (lateral) of the cylinders was wrapped with aluminium tape and
placed in the oven for drying (Figure 1b). They were dried at 80°C and 3% RH for different
drying periods (saturated, 12 hours, 1 day, 2 days, 3 days, 4 days and 7 days). Six cylinders
per drying condition were evaluated. After this period, cylinders were removed from the oven.
Then, they were sliced into small pieces of approximately 1 cm (Figure 1c) using dried
sawing.
Each small piece was measured and then dried (80°C and 3% RH) until mass stabilisation
(mass variation less than 0.05%) to determine the water content in each one of the slices.
The position of each slice was calculated based on piece thickness and the total height of
the small cylinders. This method was done to take into account the thickness of the saw cut.
With these two measures, the water content profile for each mix/condition was estimated.
Since drying was symmetrical, half of the cylinder was mirrored to obtain a denser cloud of
data, and better estimate the profile. Then, a regression analysis was done for each case to
estimate the water profile curve.
Figure 2 presents three examples of the water content profile estimation. It shows the water
content profile after 1, 4 and 7 days of drying. The scatter values of each piece were plotted
together with a regression line. For this regression, a second-degree polynomial was used
for the fit. Some variations from the individual values were observed, but they were expected
given the small size of the cylindrical sample. In any case, the method seems to be an
appropriate estimation of the water content profile. The same procedure was done for all the
studied conditions.
Figure 2 Water content profile after drying.
Fire tests
Spalling was evaluated by testing 21 concrete prisms subjected to a standard fire and
uniaxial mechanical loading. Three prisms were evaluated for each drying condition.
Saturated samples were removed from the water curing and sealed with aluminium foil tape
in the four lateral sides (20 x 10 cm). Then, they were placed in the drying oven at the same
conditions (80°C and 3% RH) and exposure days as the small cylinders (saturated, 12 hours,
1 day, 2 days, 3 days, 4 days and 7 days). After the referred drying period, samples were
removed from the oven and went to spalling tests. The samples were also insulated with 12
cm of rock wool from both sides.
The occurrence of every spalling event was recorded during each test. Figure 4 presents the
time-temperature curves from all tests (average of the three thermocouples). The x marks
indicate the time of the first spalling event for each sample. After the first spalling event, the
temperature recorded distances from the fire curve. The main criterion adopted for test
validation was whether the curve followed the tolerances up to the first spalling event. Hence,
only one mix (sample 2 from NA-1D) was excluded from further analysis.
Different postfire analyses were done. First, mass measurements were made before and
after the fire tests. Digital photogrammetry was used to quantify the spalling (volume and
depth). This approach involves image acquisition and processing, 3D reconstruction, mesh
generation, and post-processing [19]. Samples were placed in a rotating table to orientate
3D reconstruction successfully. The camera (Panasonic DMC-GX80) was fixed on a tripod,
and between each shot, the rotating table was turned 10°. A total of 36 images were obtained
for each sample. The image processing and 3D reconstruction were done with Meshroom
[20] software. The 3D model of each model was then processed (cleaned and scaled) using
CloudCompare software [21]. A binary mesh of the 3D model was imported in a Cast3m [22]
algorithm, which post processed the model to obtain the spalling volume and depths. The
error between the dimensions measured by photogrammetry analysis and the caliper range
from 1.06% to 2.07%, as show in Fernandes et al. [18].
RESULTS
The water content profiles estimated from the measurements of the small pieces are
presented in Figure 5. As expected, it shows the progressively drying of the sample. For
example, after 2 days of drying, the NA samples showed 1.6% water content at 1cm, while
after 7 days of drying, the water content at 1 cm is lower, close to 0.7%.
Figure 5 Water content profiles after different drying processes - curves are a regression
(second-degree polynomial) of measured points.
From photogrammetry, two indicators were chosen to evaluate spalling: the mean spalling
depth and the spalled volume. Figure 6a shows the variation of spalled volume with drying
period. First, at the saturated state, concrete showed a high spalling volume. It is noteworthy
that the results are close to the ones obtained in previous works with similar conditions and
the same test setup [18].
However, drying changed the observed behaviour. Samples presented higher spalled
volume after 12 hours. It went from 2.4% saturation to 4.7% after 12 hours of drying. Further
drying led to reduced spalling after 1 and 2 days of drying. The latter already shown relevant
variations in spalling risk, given that, out of four samples tested at this drying condition, two
did not spall, one presented small located spalling (0.1%), and one had high spalled volume
(6.8%). Starting from 3 days of drying, no spalling was registered for all samples. Similar
observations can be drawn from the evaluation of mean spalling depth (Figure 6b). Again,
similar values from previous works were obtained at a saturated state [18] - 5.5 mm for
concrete made with NA. Then, even after 1 day, drying significantly reduces the mean
spalling depth.
To further evaluate the effect of the water content on the spalling behaviour, the water
content was calculated at different depths using the regression curves. Figure 7 shows mean
spalling depth versus water content (Wc) at 1, 1.5 and 2cm from the exposed surface. The
trends for all depths indicate the same: an increase of the mean spalling depth with the rise
in water content. These graphs also give an idea of the minimal water content needed in the
2 first centimetres for possible trigger spalling. For NA, spalling was observed when water
content was higher than 1.3% and 2.1% at 1 cm and 2 cm, respectively.
Figure 7 Mean spalling depth versus water content at different depths.
The results herein showed the influence of the water content profile in the spalling risk
evaluation. However, evaluating the water content or the amount of water needed to trigger
spalling is complex. When analysing the first centimetres, part of the water evaporates, and
another moves towards the centre to form the so-called 'moisture clog'. To further investigate
the triggering water content, three different graphs are presented in Figure 8.
First, Figure 8a plots the mean spalling depth versus the water content calculated at this
mean depth. For this first case, except for one spalled sample (out of 14), water content at
the mean spalling depth was higher than 1.5% for NA. Figure 8b plots the mean spalling
depth versus the mean of water content between the surface and the mean spalling depth.
As expected, this average water content is lower. In this case, the critical water content was
around 1%. In the last case, Figure 8c, the mean spalling depth is plotted against the mean
water content in all the first 5 centimetres (half cylinder). In this case, the value is dependent
on the sample size. For this case, spalling happened with a water content of 2.6%.
It is noteworthy that this analysis shows a difference between the 'triggering' water content.
The first two approaches are more conservative. These water contents agree with the ones
obtained by Maier et al. [5] They observed that concrete spalling is likely to occur for more
than 1% of water content. The last approach indicates a higher water content. It is noteworthy
that the value is close to the Eurocode 2 [23] recommendations, which states that explosive
spalling is unlikely to happen if the moisture content is lower than 3% by weight.
CONCLUSIONS
This paper presented an experimental investigation on the effect of different water content
profiles on the spalling sensitivity of concrete. Based on the results, the following conclusions
were drawn:
• The water content measurement through a small cylinder showed potential to correctly
estimate the water content profile along the thickness of samples before fire test.
• After 7 days of drying, concrete made with NA showed a low water content in the first
centimetres (0.7% at 1 cm depth, for example).
• Drying significantly reduced the spalling risk of concrete. After only 3 days of drying at
80°C, no spalling was registered.
• Most of the spalled samples showed a water content higher than 1.5% at the mean
spalling depth. This value was even lower in the case of average water content at the
first centimetres - between surface and spalling depth. This possible threshold is closer
to previous works. At the same time, when evaluating the mean water content through
all depth, the critical value is close to 2.6% - value closer to what is currently specified in
Eurocode 2 [23].
REFERENCES