Download as pdf or txt
Download as pdf or txt
You are on page 1of 19

Engineering Structures 227 (2021) 111398

Contents lists available at ScienceDirect

Engineering Structures
journal homepage: www.elsevier.com/locate/engstruct

Shear response of deep precast/prestressed concrete hollow core slabs


subjected to fire
Hang T.N. Nguyen a, b, Kang-Hai Tan a, *
a
School of Civil and Environmental Engineering, Nanyang Technological University, 50 Nanyang Avenue, Singapore 639798, Singapore
b
National University of Civil Engineering, 55 Giai Phong Street, Hanoi, Vietnam

A R T I C L E I N F O A B S T R A C T

Keywords: This paper presents experimental and numerical investigations on shear behavior of deep precast/prestressed
PCHC slabs concrete hollow core (PCHC) slabs under fire conditions. Five fire tests including one heat transfer (subjected to
Web-shear resistance ISO 834 fire) and four structural shear tests (subjected to a non-standard fire curve) were conducted. Data ob­
Shear behavior
tained from the experiments were used to investigate shear behavior of deep PCHC slabs in the event of fire and
Fire conditions
Finite element model
to verify two sequentially-coupled finite element (FE) models simulating thermal and structural responses of
Heat transfer deep PCHC slabs subjected to shear and under fire conditions. The verified FE models were then employed to
extend the existing experimental database. Fire resistance of the four PCHC specimens subjected to ISO 834 fire
was numerically obtained. It was shown that although all the specimens were designed to sustain 2 h fire
resistance following the prescriptive method in Eurocode 2, they failed at early stages and did not meet their
designed fire resistance. In addition, web-shear resistance of the hollow-core units decreased as slab depth
increased. Moreover, from the test results, it was evident that temperature-induced tensile stresses, rather than
fire-induced degradation in strength and stiffness of concrete and strands, governed the web-shear resistance of
deep PCHC units. Furthermore, load level has a critical effect on web-shear resistance of deep PCHC slabs. Last
but not least, data from the experimental and numerical studies were used to validate the formula to calculate
shear capacity of PCHC slabs under fire conditions in EN 1168:2005+A3:2011.

1. Introduction 318 or Eurocode 2 [1–4]. Under fire conditions, it is generally accepted


that the web-shear capacity significantly decreases due to a
Precast/prestressed concrete hollow core (PCHC) slabs have been the temperature-induced loss in strength of concrete and prestressing steels
favorite products for wall or floor systems since the 1970s. Notwith­ and tensile stresses in concrete webs caused by nonlinear thermal
standing this, there are still some concerns regarding the web-shear gradient, which is more pronounced for deeper hollow-core slabs
capacity of PCHC slabs. The reason for this lies in the fact that the exceeding 300 mm. However, most fire shear tests in the literature were
extrusion method, which is the most popular method to fabricate PCHC conducted on shallow PCHC slabs (with circular voids and depths
slabs, does not allow any stirrups or shear reinforcement to be placed in smaller than 300 mm) [5–7]. To date, there were only four fire shear
the slabs. Therefore, the web-shear capacity of PCHC slabs solely relies tests performed on deep PCHC slabs (with a depth of 400 mm), in which
on tensile strength of concrete webs which usually consist of thin three were reduced-scale tests, carried out on double-web specimens
sections. sawn from a standard hollow-core unit of 1.2 m wide. Furthermore,
With a reduction in cross section due to presence of longitudinal although there has been more than 40 shear tests on PCHC slabs under
hollow cores, relatively low tensile strength of concrete, and most fire conditions found in the literature [8], many of them were performed
importantly, the absence of stirrups, PCHC slabs are naturally vulner­ to serve the commercial goal of verifying fire resistance instead of
able to web-shear failure. This concern becomes more critical for the so- studying shear mechanisms of PCHC exposed to fire [7]. As a result,
called deep PCHC slabs, which have non-circular voids and with flat these fire tests did not provide adequate information for an in-depth
webs and depths exceeding 300 mm. These slabs could fail in web-shear study on shear response of PCHC slabs subjected to fire. Consequently,
mode at loads smaller than those predicted by design codes such as ACI the way fire-exposed PCHC slabs fail in web-shear has not been well

* Corresponding author.
E-mail address: CKHTAN@ntu.edu.sg (K.-H. Tan).

https://doi.org/10.1016/j.engstruct.2020.111398
Received 21 February 2020; Received in revised form 13 August 2020; Accepted 4 October 2020
Available online 2 November 2020
0141-0296/© 2020 Elsevier Ltd. All rights reserved.
H.T.N. Nguyen and K.-H. Tan Engineering Structures 227 (2021) 111398

Table 1
Summary of fire tests.
Slab ID. Depth (mm)* Axis distance (mm)** Shear span (mm)*** Ambient capacity**** (kN) Applied load (kN) (load level (%)) Type of test

H.HC.400 400 70 n.a n.a 0 (0) HT


F.HC1.320 320 40 800 186 74 (40) ST
F.HC2.400 400 45 800 511 204 (40) ST
F.HC3.500 500 52.5 800 1160 464 (40) ST
F.HC4.500 500 52.5 800 1160 290 (25) ST
*
Depth: distance from the bottom soffit to the top soffit of slabs.
**
Axis distance: distance from the bottom soffit to the center of prestressing strands.
***
Shear span: distance from center line of bearing support to center of line load (shown in Fig. 5).
****
Defined by tests by Nguyen et al. [14]; HT = Heat transfer; ST = Structural test.

Table 2 shear formula in EN 1168 for PCHC slabs in the event of fire. Details of
Geometric properties of H.HC.400. the experimental program, test results, numerical studies and code
comparison are presented and discussed in this paper.

2. Experimental program

2.1. Test program

The original test plan included fire shear tests on four hollow-core
specimens to investigate shear failure mechanisms of deep PCHC slabs
subjected to fire and to obtain experimental data for numerical verifi­
cations. However, as hollow-core slabs commonly fail at an early stage
H.HC.400 in fire shear tests (within 60 min of fire exposure) as shown in the
Cross-section area, A 0.265 (m2) literature [5,7,8], there was no temperature data more than 2 h to verify
Cross-section height, h 400 (mm) the developed FE heat transfer model. Thus, an additional and inde­
Cross-section width, b 1.2 (m) pendent heat transfer test was conducted on a 400 mm thick slab to
Specimen length, L 2.2 (m)
obtain temperature profiles up to 3 h to verify the FE heat transfer
Total web width, bw 350 (5 webs of 70 mm) (mm)
Top flange min. thickness 60 (mm) model. Therefore, two types of fire test were carried out in the experi­
Bottom flange min. thickness 60 (mm) mental program, i.e. heat transfer and structural fire tests. Table 1
summarizes the key information of the five tests.
All the specimens were cast by a local precast company using the
extrusion method. The specimens were designated as H.HC.400 for heat
established. transfer test or named as F.HC1.320 to F.HC4.500 for structural fire tests
Due to limitations of test data, there have been many finite element (Table 1), of which 320, 400, and 500 denoted slab thickness in mm. It
based studies on structural responses of PCHC slabs under fire conditions should be noted that the cross section of H.HC.400 was different to that
[9–13]. Nonetheless, due to the paucity of experimental results on deep of F.HC2.400 although they had the same depth (Table 2 vs. Fig. 3(b)).
PCHC slabs, most of the developed FE models in these numerical studies This is because H.HC.400 was designed with an axis distance (the dis­
were validated using fire tests on shallow PCHC slabs (depths smaller tance from the bottom soffit to the center of prestressing strands) of 70
than 300 mm). However, behavior of the two can be greatly different mm to achieve 4 h fire rating (FR) based on the prescriptive method in
even under identical loading and heating schemes. For instance, with the Eurocode 2 [15]. On the other hand, F.HC1.320, F.HC2.400, F.HC3.500,
same shear span of 1000 mm, a 400 mm thick slab can fail by web-shear and F.HC4.500 were designed with axis distances ranging from 40 to
while a 200 mm thick slab can collapse by flexure. Studies on deep PCHC 52.5 mm to sustain 2 h FR.
slabs cannot rely solely on FE models which were verified by fire tests on Since the objective of the heat transfer test was to obtain temperature
shallow units. Therefore, an independent numerical model validated profiles subjected to the ISO 834 fire curve, no imposed load was applied
using deep PCHC slabs is needed to accurately predict response of the on H.HC.400 (Table 1). The heat transfer test was conducted following
slabs under fire. the standard fire ISO 834 while a non-standard fire curve with tem­
To bridge the gap in knowledge, an experimental program consisting perature up to 1000 ◦ C was used in the structural fire tests. This was due
of five PCHC slabs with depth ranging from 320 to 500 mm tested under to the large dimensions (3 m wide × 3 m long × 0.75 m deep) of an
fire conditions was conducted at Nanyang Technological University electrical heating furnace used in the full-scale structural fire tests and
(NTU), Singapore. The effect of slab thickness and load level on web- limitations of power supply in NTU Annex Lab. Thus, the ISO 834 fire
shear behavior of deep PCHC slabs were investigated. Data obtained could not be generated in the structural fire tests. Therefore, results from
from the fire tests were analyzed to shed light on the web-shear failure the structural fire tests were used to examine shear failure mechanism of
mechanism of deep PCHC units. In addition, they were used to verify PCHC slabs, rather than to obtain their fire resistance under the standard
two FE models that simulated thermal and structural responses of deep fire curve. Numerical simulations were then conducted to check if the
PCHC slabs under fire conditions. The FE models, in turn, were used to four specimens in the structural tests could withstand the ISO 834 fire
extend the existing database of deep PCHC slabs subjected to fire. curve for 2 h.
Furthermore, the test data were also used to evaluate the validity of the In contrast, a smaller furnace (0.7 m wide × 1.7 m long × 0.75 m

2
H.T.N. Nguyen and K.-H. Tan Engineering Structures 227 (2021) 111398

a) Section A and Section B in longitudinal direction

SECTION A
6

5a

4a

3a

2a 2b
1a 1b
70

SECTION B

5b

4b

Strand Thermocouples

b) Thermocouples in vertical direction


Fig. 1. Thermocouple arrangement.

a) Test setup b) Fire scenario


Fig. 2. Heat transfer test setup & fire scenario.

deep) which could generate ISO 834 fire was used to conduct the heat test under ISO 834 fire for H.HC.400 could then be used to verify the
transfer test. No loads were applied to H.HC.400. The dimensions of H. developed FE heat transfer model for PCHC slabs under ISO 834 fire
HC.400 were 1.2 m wide by 2.2 m long so that it could fit into a smaller curve.
heating furnace. The temperature profile acquired from the heat transfer

3
H.T.N. Nguyen and K.-H. Tan Engineering Structures 227 (2021) 111398

SF.HC1.320

9.6 mm diameter strand (area of 55 mm2)


50
12.9 mm diameter strand (area of 99 mm2)

320
40

SF.HC3.500 and SF.HC4.500


a)
SF.HC2.400

60

500
400

55
45 52.5

1200 1200

b) c)

Fig. 3. Specimen cross sections: (a) F.HC1.320; (b) F.HC2.400; (c) F.HC3.500 and F.HC4.500.

Table 3
supported at the two ends by two A-frames. No axial restraint was
Geometric properties of the PCHC specimens.
applied to the slab during the test. The specimen was heated for three
Slab ID. F.HC1.320 F.HC2.400 F.HC3.500 F.HC4.500 hours following the ISO-834 fire curve. Fig. 2(b) shows the standard fire
Cross-section area, A (m2) 0.187 0.213 0.259 0.259 curve ISO-834 together with the measured temperature inside the
Cross-section depth, h 320 400 500 500 furnace. It illustrates that the electrical furnace could be controlled to
(mm)
follow the standard fire condition for a duration of up to three hours.
Cross-section width, b (m) 1.2 1.2 1.2 1.2
Specimen length, L (m) 4.15 4.15 4.15 4.15
Total web width, bw (mm) 240 266 292 292
2.3. Structural fire shear tests
2nd moment of area, I 2.47 × 4.42 × 8.22 × 8.22 ×
(m4) 10− 3 10− 3 10− 3 10− 3
1st moment of area top 1.53 × 2.19 × 3.26 × 3.26 × 2.3.1. Specimens
part, St (m3) 10− 2 10− 2 10− 2 10− 2 Full-scale tests were conducted on four hollow-core specimens with
Self weight of slab, w 3.78 4.31 5.23 5.23 depth ranging from 320 to 500 mm as shown in Table 1. All the speci­
(kN/m2)
Axis distance (mm) 40 45 52.5 52.5
mens were 4.15 m long by 1.2 m wide, comprising 4 non-circular voids.
Prestressing steel ratio 0.147 0.470 0.382 0.382 Fig. 3 shows cross-sectional configurations while Table 3 presents the
(%) detailed geometric properties of the four specimens.
Top flange min. thickness 40 40 45 45 The hollow-core specimens were prestressed using seven-wire, low
(mm)
relaxation strands with an ultimate strength of 1860 MPa. As shown in
Bottom flange min. 40 40 45 45
thickness (mm) Fig. 3, two types of prestressing strand were used: 9.6 mm or 12.9 mm
diameter with a cross sectional-sectional area of 55 or 99 mm2,
respectively. These strands were initially stressed at 62% of their ulti­
2.2. Heat transfer test mate capacity, resulting in a jacking stress of 1150 MPa.
As spalling was not included in the scope of this study, normal-
2.2.1. Specimen properties strength concrete with a characteristic compressive value of 40 MPa
Table 2 presents the geometric properties of H.HC.400. The average was employed. The target mean strength should be greater than 48 MPa.
concrete compressive strength measured on three concrete cubes of 150 Table 4 presents concrete properties of the four specimens determined
mm × 150 mm at 28 days was 65.3 MPa, which corresponded to 54 MPa from cylinders (∅50 mm × 100 mm long) cored out from tested hollow-
mean cylinder strength based on Table 3.1 – Eurocode 2 [16]. The core specimens of ambient shear tests [14]. The ambient shear tests were
specimen was considered as normal-strength concrete and spalling of conducted to obtain respective shear capacity at ambient condition of
concrete was not expected. The heat transfer test was conducted after these specimens.
248 days of casting. As a result, the moisture content of the specimen
was at a stable level. 2.3.2. Test setup & instrumentation
The test setup for structural fire tests is shown in Fig. 4. All the
2.2.2. Instrumentation specimens were placed on top of an electric furnace whose interior di­
The instrumentation included type K thermocouples installed over mensions were 3 m wide, 3 m long, and 0.75 m deep, and they were
the slab depth through drilled holes from the slab soffit. Thermocouples simply supported at their two ends by two A-frames without any axial
were embedded at two different sections along the specimen length, restraint. The slabs were loaded by two line loads that were symmetri­
namely, Section A, and Section B. Section A was at the center line of the cally placed about the slab center. The shear spans were 800 mm in all of
slab while Section B was 600 mm from one edge of the specimen. Fig. 1 the tests. In addition, since the width of all the hollow-core specimens
shows the thermocouple arrangement in the heat transfer test. was only 1.2 m while the width of the furnace was 3 m, steel panels and
thermal wool were placed upon the open parts of the furnace to prevent
2.2.3. Test setup & fire scenario heat loss to surrounding environment.
Fig. 2(a) shows the overall setup for the heat transfer test. The The instrumentation in the structural fire tests included load cells,
specimen was placed on top of the ISO electrical furnace and was simply linear variable displacement transducers (LVDTs) and type K

4
H.T.N. Nguyen and K.-H. Tan Engineering Structures 227 (2021) 111398

thermocouples (T). Three load cells (LC1 to LC3) were arranged on top

Average tensile strength/SD* (MPa)


of a loading beam and two spreader beams to distribute loads from a
100-ton hydraulic jack to the specimen. Eight LVDTs (LVDTs 1 to 8)
were placed on top of each specimen to monitor vertical displacements
at the mid-span, the locations of applied loads and the two supports.
Additionally, four LVDTs (LVDTs 9 to 12) were used to measure longi­
tudinal expansion at the two ends of the specimens during heating. In
addition, thermocouples were embedded in the specimens to track
temperature development during the fire tests. Fig. 5 shows the load

3.6/0.36
4.2/0.45
4.1/0.14
4.1/0.14
cells and LVDT arrangement whereas locations of the thermocouples are
shown in Fig. 6.

2.3.3. Testing procedure and heating program


The structural fire tests in this study followed a transient-state
No. of cylinder specimens (50 × 100 mm)

regime, in which vertical loads were introduced first and heat was
subsequently applied. To study the effect of load level on shear behavior,
two different load levels were applied. While F.HC1.320, F.HC2.400,
and F.HC3.500 were loaded up to 40% of their ambient capacity
(defined by tests, given in Nguyen, Tan [14] and shown in Table 1), F.
HC4.500 was only loaded up to 25%. The four hollow-core slabs were
Tensile splitting tests

then heated up to failure while the imposed loads were maintained


constant during the tests. All four of the specimens were subjected to a
non-standard fire curve which increased linearly from ambient to
1000 ◦ C at 60 min and remained at this temperature until the end of the
test as mentioned in Section 2.1. Fig. 7 shows the electric heating curve
2
2
2
2

used in this study, together with the measured furnace temperature with
time. It should be noted that the structural fire tests began with F.
HC2.400 and due to a problem of the furnace, the specimen was exposed
to a lower heating curve compared to the expected one (Fig. 7). Sub­
Elastic modulus (GPa)

sequently, the problem was fixed, and the latter three fire tests followed
the intended heating regime.

3. Test results and discussions


34.9
35.2
35.5
35.5

3.1. Heat transfer test (H.HC.400)

Temperature profiles recorded by the thermocouples (Fig. 1(b)) are


plotted in Fig. 8. It should be noted that for cross sections with more than
Average compressive strength/SD* (MPa)

one thermocouple, an average value was taken. Data from thermocou­


ples showed reasonable results since the cross-sectional temperatures in
the concrete decreased as the distance from the fire-exposed soffit
increased, due to low conductivity and high specific heat capacity of
concrete that delayed the heat transfer over the cross section.
Cracks observed in the heat transfer test are shown in Fig. 9. Basi­
cally, during the first 25 min of fire exposure, no cracks were observed.
However, when the furnace temperature reached about 830 ◦ C (at 28
min of fire exposure), longitudinal cracks started occurring at the top
and the bottom of the hollow cores, emanating from the two slab edges.
57/n.a
54/6

56/4
56/4

As the temperature increased, these cracks propagated towards the slab


center, resulting in longitudinal cracks along the longitudinal voids as
shown in Fig. 9. The reason for the occurrence of longitudinal cracks will
be further discussed in Section 3.2.3. It is noteworthy that apart from the
longitudinal cracks, spalling was not observed in the heat transfer test.
No. of cylinder specimens
Material properties of the PCHC specimens.

3.2. Structural fire shear tests – Test results


Compressive tests

Table 5 summarizes results from the structural fire tests. Discussions


of test results will be focused on thermal response, structural behavior,
SD = Standard deviation

crack pattern and failure mode.


3
1
2
2

3.2.1. Thermal response


Temperature profiles over the cross section of the four slabs
measured by the embedded thermocouples (recorded up to failure) are
F.HC1.320
F.HC2.400
F.HC3.500
F.HC4.500

plotted in Fig. 10. In the figure, x refers to the vertical distance from the
Slab ID.
Table 4

bottom soffit (the fire-exposed surface) to the measured section in mm.


Similar to thermocouple readings from the heat transfer test H.HC.400,
*

5
H.T.N. Nguyen and K.-H. Tan Engineering Structures 227 (2021) 111398

Fig. 4. Setup of the fire shear tests.

Load transferred from


LC: Load cell
hydraulic jack
Loading beam LVDT: Linear variable displacement transducer
LC1 4150

400
LVDT 5 LVDT 1 LVDT 3
LC3 LVDTs 5 & 6 LVDTs 3 & 4
LVDT 8 LVDTs 1 & 2 LC2 LVDT 7

1200
400
LVDT 8 LVDT 7

400
LVDT 6 LVDT 2 LVDT 4
75 800 800 75
Spreader beams
75 800 1200

Hollow-core specimen
b) LVDT arrangement - top view
Electrical Furnace

3000
LVDT 11 LVDT 9

h
LVDT 12 LVDT 10

4150
4000

a) Elevation c) LVDT arrangement - elevation

Fig. 5. Load cell and LVDT arrangement.

thermocouple data from the structural fire tests also showed an expected failure criteria specified in Eurocode 2 Part 1–2 [15], i.e. integrity
pattern; cross-sectional temperature reduced with increasing × from the (criterion E), insulation (criterion I), and load-bearing capacity (crite­
heated soffit. In addition, it was observed that at a certain distance × rion R). Criterion R is considered through deflection limits in the flexural
from the slab soffit, there was a sudden increase in temperature from 40 fire tests [18]. However, as fire shear tests were conducted, failure was
to 50 min of fire exposure (as evident in Fig. 10(a) to (d)). Such a sudden defined at the temperature or time when there was a significant drop in
increase in temperature was probably attributed to the formation of load-carrying capacity caused by the formation of shear cracks. At this
longitudinal cracks at the bottom slab along the longitudinal voids, moment, the specimens were unable to resist the imposed loads from the
causing an increase in temperature in regions nearby. hydraulic jack.
The temperature histories obtained from thermocouples also indi­ Time histories of the imposed load and the mid-span displacement
cated that temperature gradually increased from ambient to 100 ◦ C. from the four fire tests are plotted in Fig. 11. It is noteworthy that de­
After this, a temperature plateau was reached, which started from flections shown in Fig. 11 were average values from LVDTs 1 and 2
around 42 min of fire exposure. This plateau was caused by the migra­ (Fig. 5). It is shown that all four of the specimens lost their bearing ca­
tion of free water inside the concrete, in the upward direction from the pacity before the furnace temperature reached 1000 ◦ C. F.HC1.320, F.
fire-exposed soffit to the slab top surface. As the temperature exceeded HC2.400, F.HC3.500, and F.HC4.500 failed at 944, 803, 851, and
100 ◦ C, water was transformed from liquid to vapor, resulting in a 881 ◦ C, respectively. As shear cracks occurred near the supports, the
moisture clog that inhibited further temperature increase until the layers specimens could no longer sustain the imposed loads.
had eventually dried out [17].
3.2.3. Cracks and failure modes
3.2.2. Structural responses Failure modes of the four specimens are given in Table 5. While F.
It should be noted that for slabs subjected to fire, there are three HC2.400, F.HC3.500, and F.HC4.500 failed in web-shear with formation

6
H.T.N. Nguyen and K.-H. Tan Engineering Structures 227 (2021) 111398

A B

h
1100 A B

2075

4150

a) Section A and B in longitudinal direction (where thermocouples were placed)

b) Thermocouples in vertical direction


Fig. 6. Thermocouple arrangement.

1200 800
x: distance from the bottom soffit (mm)
End of the tests 700 x = 25 (T1a & T1b)
1000 due to failure x = 50 (T2a & T2b)
x = 110 (T3a)
600
x = 180 (T4a & T4b)
Temperature (oC)

800 x = 280 (T5a & T5b)


Temperature (oC)

500 x = 400 (T6)

600 Low temperatures 400


due to a problem of
the furnace
400 300
Programmed curve
F.HC1.320
200
200 F.HC2.400
F.HC3.500
F.HC4.500 100
0
0 20 40 60 80 100 120 140 160 180 0
Time (min) 0 50 100 150
Time (min)
Fig. 7. Electric heating curve and measured furnace temperatures.
Fig. 8. Development of temperature over the cross-section of H.HC.400.

of diagonal tensile cracks in the concrete webs near the support, F.


HC1.320 lost its bearing capacity due to a combination of web-shear and concrete webs exceeding concrete tensile strength. Fig. 12 shows the
anchorage failure. The web-shear cracks occurred in a region between failure modes in the four tests whereas Fig. 13 presents typical cracks
the support and applied-load locations, due to a combination of thermal observed in the tests. Four types of cracks were identified, viz. splitting,
stress and stresses due to prestressing force and applied load in the longitudinal, thermal and web-shear cracks. Other than these, no fire-

7
H.T.N. Nguyen and K.-H. Tan Engineering Structures 227 (2021) 111398

Fig. 9. Longitudinal cracks at the top and bottom of H.HC.400.

Table 5
Test results.
Slab ID. a/d* Applied load (kN) Furnace temperature at failure (oC) Failure time (min) Maximum mid-span deflection (mm) Failure mode

F.HC1.320 2.86 74 944 57 57.8 WS & A**


F.HC2.400 2.25 204 803 118 34.4 Web-shear
F.HC3.500 1.79 464 851 48 14.2 Web-shear
F.HC4.500 1.79 290 881 52 16.9 Web-shear
*
a/d: Shear span-to-effective depth ratio;
**
WS & A = Web-shear and Anchorage

induced spalling was observed in any of the four tests during the heating transverse direction. These cracks were first generated at the two sup­
time. porting edges. Due to elevated temperatures and an absence of steel
Localized splitting cracks (Fig. 13(a)) occurred between the concrete meshes at the top and the bottom flanges of the hollow-core slabs, these
and the steel strand when the furnace temperature reached about cracks propagated towards the slab center, resulting in full-length lon­
700 ◦ C. These cracks occurred due to a high-temperature gradient within gitudinal cracks as shown in Fig. 13(b). Furthermore, vertical thermal
the lower part of the section, resulting in incompatible thermal expan­ cracks occurred due to a temperature gradient across the section of the
sion between the two materials. The verified FE heat transfer model specimen, inducing tensile stresses in web regions and forming cracks
presented in Section 4.1 showed that at the furnace temperature of along the concrete webs. Finally, web-shear cracks developed in regions
700 ◦ C, average temperature of concrete below the strand level (taken between the bearing and applied load. Once occurred, these cracks
using average temperature of all concrete nodes below the strand level immediately propagated towards the applied load and the support lo­
in one vertical section) and temperature of the strand were 190 and cations, forming diagonal web-shear cracks, resulting in a significant
98 ◦ C, respectively. Thermal expansion of concrete and strand could drop in load-bearing capacity and causing a brittle failure. To illustrate
accordingly be calculated using formulas for thermal strain in Eurocode damages and web-shear failure of the tested slabs, Fig. 15 shows crack
2 Part 1–2 [15] as follows: patterns of F.HC2.400 observed after testing.
For concrete (using siliceous aggregates):
εc (θ) = − 1.8 × 10− 4 + 9 × 10− 6 θ + 2.3 × 10− 11 θ3 for 20 ◦ C ≤ θ ≤ 700 ◦ C 4. Numerical analyses
Thermal strain of prestressing steel:
εp (θ) = − 2.016 × 10− 4 + 10− 5 θ + 0.4 × 10− 8 θ2 for 20 ◦ C ≤ θ ≤ 1200 ◦ C
In this study, two 3-D FE models were sequentially developed, viz.
heat transfer and mechanical models to respectively investigate thermal
and structural responses of PCHC slabs subjected to fire. First, the heat
For a temperature of 190 ◦ C for concrete and 98 ◦ C for strand, thermal transfer model using ABAQUS/Heat-transfer [19] was built to obtain
elongation of concrete and strand at one end was 3.5 and 1.7 mm, temperature evolution inside PCHC slabs when exposed to fire. The
respectively. Such a discrepancy in thermal expansion between the temperature field determined from the heat transfer model was subse­
concrete and the strand induced shear stresses between them, promoting quently mapped onto the mechanical model which was developed using
splitting cracks to occur. The difference in thermal expansion of the two ABAQUS/Explicit [19] for structural analysis. The heat transfer model
was evident from the observed slippage of the strands at the early stage was verified using temperature data under both ISO and non-ISO fire
of fire exposure as shown in Fig. 14. In general, the localized cracks were obtained in the five tests. Additionally, the mechanical FE model was
initiated at the two slab ends and then propagated into longitudinal verified against the test data in terms of load and deflection histories, as
cracks along the strand length as shown in Fig. 13(a). well as damage and failure modes from the four structural fire tests.
On the other hand, longitudinal cracks (Fig. 13(b)) occurred at the Subsequently, the verified mechanical model was used to determine the
top and bottom surfaces of hollow cores when the furnace temperature fire resistance of the PCHC slabs subjected to the ISO-834 fire, obviating
reached about 800 ◦ C due to thermal expansion of concrete in the

8
H.T.N. Nguyen and K.-H. Tan Engineering Structures 227 (2021) 111398

120 300
x = 45 x: distance from the bottom soffit (mm)
x: distance from the bottom soffit (mm)
x = 55
250 x = 70
x = 122
90 x = 60 Sudden increase in x = 200
x = 75

Temperature (oC)
temperature 200 x = 400
Tempeature (oC)

x = 107
x = 160
Temperature plauteu
x = 320 150
60
Sudden increase in
100
temperature

30
50

0 0
0 10 20 30 40 50 60 0 20 40 60 80 100 120 140

Time (min) Time (min)

a) F.HC1.320 b) F.HC2.400

120 120
x: distance from the bottom soffit (mm) x: distance from the bottom soffit (mm)

100 Sudden increase in


100 Sudden increase in
temperature
temperature
x = 40
Temperature (oC)

Temperature (oC)
80 x = 43
x = 55 80
x = 50
x = 125
x = 90
60 x = 250
x = 130
x = 500 60
x = 250

40
40

20
20
0
0 10 20 30 40 50 60 0 10 20 30 40 50 60
Time (min) Time (min)
c) F.HC3.500 d) F.HC4.500
Fig. 10. Temperature development over the cross sections.

the limitation of the furnace employed in the experimental program. web thickness as illustrated in Fig. 16. It was assumed that heat from the
bottom core surface was transported to the top core surface by the up­
ward movement of hot air inside the core via convection and radiation
4.1. Heat transfer model [20,21]. In addition, the bottom core surface was also affected by the
cooler temperature from the top core by radiation. This phenomenon
In the heat transfer model, concrete was simulated using 8-node involved absorption, reflection or transmission as the top and bottom
linear heat transfer brick elements. The material and geometric prop­ core surfaces could radiate back and forth to each other. Inside the
erties of H.HC.400 were adopted to develop the model. A mesh hollow cores, the emissivity factor was also taken as 0.7 and a convective
convergence study was performed with three mesh sizes of 20, 30, and factor of 25 W/m2K produced the best-fit results.
40 mm. It was shown that all the three mesh sizes were suitable to
simulate thermal response of H.HC.400. Thus, the mesh size of 40 mm 4.1.2. Material properties
was chosen to reduce computational time. In the FE heat-transfer model, temperature-dependent thermal
properties of concrete including conductivity, specific heat and density,
4.1.1. Thermal interaction had to be defined. Such thermal properties are not constant but vary
The slab was modeled in a way that it interacted with the fire at the with temperature. In this study, thermal properties of concrete were
bottom surface and with the natural environment at the top surface. The assumed following the values in Eurocode 2 [15], as shown in Fig. 17.
convection factors for the fire exposed and unexposed surfaces were 25 Thermal properties of longitudinal steel could be ignored since they
W/m2K and 10 W/m2K, respectively. The emissivity factor related to the had very little effect on temperature development in the PCHC slabs.
concrete surface was taken as 0.7. Thermal boundary conditions within Moreover, the prestressing steel ratios in the PCHC slabs were very low,
the hollow cores were taken into account. The hollow core was divided varying from 0.15% (F.HC1.320) to 0.47% (F.HC2.400). Therefore,
into two parts, i.e., the top part and the bottom part that were separated strand temperature was assumed to be equal to the concrete temperature
by a horizontal plane cutting through the intersections at the smallest

9
H.T.N. Nguyen and K.-H. Tan Engineering Structures 227 (2021) 111398

100 60 250 50

50
80 200 40

Mid-span deflection (mm)


Mid-span deflection (mm)

Applied load (kN)


Applied load (kN)

40
60 150 Failure at 803 oC 30
o
LC1 Failure at 944 C
LC1
LC2 + LC3 30
LC2 + LC3
Deflection 100 Deflection 20
40
20

20 50 10
10

0 0 0 0
0 10 20 30 40 50 60 0 20 40 60 80 100 120 140

Time (min) Time (min)

a) F.HC1.320 b) F.HC2.400

c) F.HC3.500 d) F.HC4.500
Fig. 11. Applied load and mid-span deflection vs. time.

at the same level from the slab bottom soffit [13,21,22]. 4.2. FE mechanical model

4.1.3. Model validation 4.2.1. Modeling technique


The heat transfer model was also developed for the other four The FE mechanical model using ABAQUS/Explicit [19] was devel­
specimens in the structural fire tests using the same technique. The oped for structural analyses of PCHC slabs subjected to fire, with an
model was verified by comparing its predicted temperature and the emphasis on shear behavior. The concrete mesh discretization in the
measured temperature developed in the five specimens from the fire mechanical model had to be identical to that used in the heat transfer
tests (shown in Figs. 8 and 10). While Fig. 18 presents the predicted model, to allow the temperature at each node from the heat transfer
temperature contour of H.HC4.400 after 3 h of fire exposure, Fig. 19 model to be mapped onto those for the mechanical model. Concrete was
representatively shows temperature comparisons between the actual fire modeled using 8-node linear brick elements with a reduced integration
tests and the FE models for H.HC.400 and F.HC2.400, which had the scheme. Steel strands were simulated using T3D2 truss elements (2-node
longest heating duration and the highest recorded temperatures (up to 3 linear displacement) and embedded into the concrete assuming perfect
h and 2 h, respectively). Fig. 19 shows good correlation between the two bonding. Prestressing loss and transmission length of the strands
sets of temperature, illustrating that the developed heat transfer model (calculated using Eurocode 2 [16] and shown in Table 6) were adopted
was capable of tracing the temperature development in deep PCHC slabs in the numerical model. The strands were simulated to be stress-free at
exposed to both ISO and NON-ISO heating. Therefore, temperature their two ends and to attain their full effective stress (strand stress after
profiles from the FE heat-transfer model were reliable and could be prestressing loss) at the end of their transmission length, following a
applied to the mechanical model for subsequent structure-fire-load linear-stress-distribution assumption. A mesh size of 10 mm was applied
analysis. for the strand while a concrete mesh size of 40 mm was used as

10
H.T.N. Nguyen and K.-H. Tan Engineering Structures 227 (2021) 111398

a) Web-shear and anchorage failure – F.HC1.320

b) Web-shear failure – F.HC2.400 c) Web-shear failure – F.HC4.500

d) Web-shear failure – F.HC3.500

Fig. 12. Failure modes observed in the tests.

mentioned in Section 4.1. Fig. 21(a) and (b) respectively show concrete constitutive model for F.
HC1.320 in compression and in tension. Similar constitutive models
4.2.2. Material model were developed for F.HC2.400, F.HC3.500 and F.HC4.500 based on
Constitutive model for prestressing strands under elevated temper­ their properties defined by material tests as shown in Table 4. It is
atures specified in EN-1992-1-2 [15] shown in Fig. 20 was adopted to noteworthy that for the CDP model, it was only necessary to define the
represent temperature-dependent stress–strain behavior. Concrete dam­ stress-inelastic strain/crack opening constituent.
age plasticity model (the CDP model) in Abaqus/Explicit (2014) was The curves in Fig. 21 describe the behavior of concrete in a uniaxial
selected to simulate nonlinear behavior of concrete. stress state. The behavior of concrete in bi-axial and tri-axial stress states
In the CDP model, concrete response under uniaxial stress state must in the CDP model is defined by four plastic parameters, i.e. the dilation
be defined. The stress–strain curve of concrete in compression with a angle, the ratio of initial biaxial compressive yield strength to initial
linear descending branch and strength reduction factors accounting for uniaxial compressive yield strength, the eccentricity and the ratio of
temperature effects complied with EN-1992-1-2 [15]. The uniaxial second stress invariant on tensile meridian. While the values of the last
tensile behavior of concrete, on the other hand, was described with a two parameters were taken as 0.1 and 0.667, respectively, as recom­
stress-crack opening relation to avoid difficulties in convergence due to mended by ABAQUS [19], the dilation angle that controls the change in
mesh sensitivity [23]. The stress-crack opening function proposed by plastic volume of concrete under pressure is not only a material property
Hordijk [24] and the Eurocode concrete tensile strength reduction fac­ but is also affected by structural behavior [14]. Dilation angle of 38ο,
tors due to temperature were adopted in this study. It should be noted within the range of 15ο to 56ο as used in previous studies [14,21,23,27],
that concrete fracture energy in tension (the area below the stress-crack reasonably captured behavior of PCHC slabs exposed to fire and was
opening curve) was determined using the Model Code [25] and assumed adopted in this investigation. Lastly, the ratio of initial biaxial
to be constant during heating. Such an assumption was made as there are compressive yield strength to initial uniaxial compressive yield strength
limited test data on the fracture energy of concrete at elevated tem­ was defined to bilinearly increase from 1.16 at ambient temperature to
peratures. In addition, existing test data did not show a clear trend 1.3 at 300 ◦ C and finally to 1.7 at 700 ◦ C. These values were based on
describing the effect of temperature on concrete fracture energy [26]. experimental data from Kordina et al. [28].

11
H.T.N. Nguyen and K.-H. Tan Engineering Structures 227 (2021) 111398

Fig. 13. Four types of cracks observed in the fire tests. (the sequence of formation: (a) → (b) → (c) → (d)).

Fig. 14. Strands slipped inside concrete at early stage of heating.

4.2.3. Model validation that in the experiment, F.HC2.400 in the numerical analysis failed after
The proposed mechanical model was verified using the experimental the furnace temperature reached 854 ◦ C, similar to the failure temper­
results from the four structural fire tests. Displacement-time histories, ature of 803 ◦ C in the test. Moreover, web-shear failure mode of F.
cracks, failure modes and furnace temperature at failure were compared HC2.400 in the test was well predicted in the FE model (Fig. 24).
between the structural fire tests and the FE models. Fig. 22 shows a Additionally, as can be seen in Fig. 23, the formations of longitudi­
comparison between the recorded and predicted mid-span displacement nal, thermal and web-shear cracks observed in the tests were also
against time, whereas comparisons of furnace temperature at failure and captured in the FE models in terms of plastic strain (PE). Moreover, the
failure mode between the tests and the FE models are presented in sudden loss of load-bearing resistance resulting in failure due to prop­
Table 7. Good agreement between predictions and measurements in agation of diagonal web-shear cracks could be reproduced in the FE
terms of deflection development, failure point (furnace temperature at models (Figs. 23(a) and 24). The comparisons clearly demonstrated that
failure) and failure mode, was obtained for F.HC1.320, F.HC3.500, F. the FE models could predict shear behavior of the PCHC slabs subjected
HC4.500 (Fig. 22(a), (c), (d) and Table 7). For F.HC2.400, it is shown to fire. However, it is noteworthy that splitting cracks that occurred in
that the slab in the FE model failed earlier compared to the tested slab. the tests could not be captured in the FE models. In addition, anchorage
However, it is worth mentioning that in the test of F.HC2.400, furnace failure in F.HC1.320 could not be simulated. Such failure was caused by
temperature decreased after reaching 854οC at 65 min due to a problem a gradual loss in bond strength between the concrete and the strands.
of the electrical furnace as mentioned in Section 2.3.3. Therefore, even However, in the FE models, perfect bond condition was assumed be­
fire endurance of F.HC2.400 in the FE model was shorter compared to tween them. Consequently, the splitting cracks between the concrete

12
H.T.N. Nguyen and K.-H. Tan Engineering Structures 227 (2021) 111398

Fig. 15. Cracks and web-shear failure mode of F.HC2.400.

Fig. 16. Thermal interaction.

2.5 1600
Heat conductivity (W/moK)

Specific heat (J/kgoK)

2 The upper limit in EC2 Part 1-2 1400 Moisture content = 1.5%
1.5
1200
1
1000
0.5
0 800
0 200 400 600 800 1000 1200 0 200 400 600 800 1000 1200

Temperature (oC) Temperature (oC)


a) Thermal conductivity b) Specific heat
Fig. 17. Thermal properties of concrete based on Eurocode 2 [15].

and the strands, as well as anchorage failure, could not be simulated in resistance of identical four specimens in the structural tests (Table 3) if
the FE models. These are limitations of the developed numerical models. they are subjected to ISO 834 fire. The actual geometric and material
properties, as well as the loading schemes of F.HC1.320, F.HC2.400, and
4.3. Fire resistance of the PCHC slabs under ISO-834 fire curve F.HC3.500, and F.HC4.500 were simulated. FE results showed that
behavior of web-shear failure mode in which the slabs lost their bearing
The verified FE models were employed to numerically obtain fire capacity due to formation of web-shear cracks could be observed in all

13
H.T.N. Nguyen and K.-H. Tan Engineering Structures 227 (2021) 111398

Fig. 18. Temperature distribution at 3 h of fire exposure – H.HC.400.

a) H.HC.400 b) F.HC2.400
Fig. 19. Temperature comparison between the test and the FE heat transfer model.

Table 6
Calculated transmission length and prestressing losses based on Eurocode 2.
Slab ID. Strand diameter (mm) Transmission length Prestressing loss (%) Effective stress (MPa)

F.HC1.320 9.6 313 8.6 1051


F.HC2.400 9.6/12.9 302/406 11.4 1019
F.HC3.500 12.9 410 12.4 1007
F.HC4.500 9.6 313 12.4 1007

four of the slabs as shown in Fig. 25. The furnace temperature at failure
2.0E+09 and the fire resistance of the four slabs are summarized in Table 8.
1.8E+09 It is noteworthy that the axis distances of all the strands in the four
Prestressing strand stress (N/m2)

slabs were greater than 40 mm. Therefore, based on the prescriptive


1.6E+09 20oC
method in Eurocode 2 [15], the hollow-core specimens should sustain at
100oC
1.4E+09 least 2 h FR. However, numerical predictions from Table 8 show that all
200oC
1.2E+09 300oC the slabs subjected to a load level of 40% of ambient capacity would fail
400oC within 90 min of the standard fire curve, displaying premature failure.
1.0E+09
500oC These numerical findings are in good agreement with experimental re­
8.0E+08 600oC sults from Fellinger [7] who conducted 25 fire shear tests on PCHC units,
700oC of which only four were carried out on 400 mm deep hollow core
6.0E+08
800oC specimens.
4.0E+08 Comparing the furnace temperature at failure between the tests using
2.0E+08 a non-ISO fire curve and the FE models using the ISO-834 fire curve
(Table 8), it is suggested that fire tests using non-ISO curves can be used
0.0E+00
0.00 0.02 0.04 0.06 0.08 0.10 0.12 0.14 to estimate the web-shear resistance of PCHC slabs exposed to fire, since
regardless of the difference in the fire curves, the slabs still failed at very
Strain
comparable furnace temperatures as shown in Table 8. Moreover,
Fig. 20. Constitutive model for prestressing strands (adapted from EN-1992-1- considering the results of F.HC1.320, F.HC2.400 and F.HC3.500, which
2 [15]). were loaded at the same level of 40% of their ambient capacities, it is
evident that fire resistance decreased as overall slab thickness increased.
This behavior was expected, since under fire conditions nonlinear

14
H.T.N. Nguyen and K.-H. Tan Engineering Structures 227 (2021) 111398

6E+07 A3:2011 provides an empirical equation to determine the shear strength


Concrete compressive strength (N/m2)

20 oC
of PCHC slabs subjected to fire. It should be noted that EN 1168 is a
100 oC harmonized European standard, specifically drafted for the design of
5E+07
200 oC PCHC slabs. The shear capacity of fire-exposed PCHC slabs according to
300 oC EN 1168 is given as follows:
4E+07 400 oC
VRd,c,fi = [Cθ,1 + αk × Cθ,2 ] × bw × d (Sec.G.1.3)
500 oC
600 oC
3E+07 where
700 oC
800 oC FR,a,fi,p
2E+07 900 oC Cθ,1 = 0.15min(kp (θp )σ cp,20o C ; Ac ) is a coefficient which accounts
1000 oC for concrete stress under fire conditions;
√̅̅̅̅̅̅
1E+07 αk is a size factor = 1 + 200 d ⩽2 where d is in mm;
√̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
FR,a,fi
Cθ,2 = 3 0.58 × fyk ×b w ×d
fc,fi,m is a coefficient that accounts for the
0E+00
anchored longitudinal reinforcement;
0.00 0.01 0.02 0.03 0.04
bw is the total web thickness (mm);
Strain
d is the effective depth of the hollow core slab (mm).
a) Compressive behavior
To apply this formula, the strand and concrete temperatures at the
4.0E+06 web mid-height must be known. However, strand-temperature data
were not available in the tests since thermocouples were embedded into
3.5E+06
the specimens through drilled holes as mentioned in Section 2.3.2.
Concrete tensile strength (N/m 2)

3.0E+06
20 oC During installation, to prevent potential damage, the drill bit used
100 oC should not come into contact with the strands. Therefore, these tem­
200 oC
2.5E+06 peratures were determined using the verified FE heat transfer model.
300 oC
Concrete temperature at the web mid-height, on the other hand, was
2.0E+06 400 oC
500 oC
defined using the experimental data. The strand and concrete temper­
600 oC atures with their respective strength reduction factors according to
1.5E+06
Eurocode 2 Part 1–2 [15] are presented in Table 9, whereas shear-
1.0E+06 strength comparisons between the tests and the EN 1168 predictions
are shown in Table 10. It should be noted that concrete cylinder mean
5.0E+05 strength (including both the compressive and tensile strengths) was used
in the calculations instead of characteristic strength. In addition, all
0.0E+00
0.0000 0.0001 0.0002 0.0003 0.0004
safety factors were taken as unity.
With Vtest/Vpredict > 1 in all the cases (Table 10), it is clearly shown
Crack opening (m)
that the shear formula in EN 1168 gives conservative predictions for
b) Tensile behavior shear strength of PCHC slabs similar to the ones tested in the experi­
Fig. 21. Constitutive model for concrete (F.HC1.320). mental program. However, it is worth mentioning that currently, design
of fire resistance of deep PCHC slabs is mostly based on the prescriptive
method specified in Table 5.8 - Eurocode 2 Part 1-2 [15]; the method has
been shown in Section 4.3 to be unsafe for deep PCHC slabs.
thermal gradient existed across the slab section; thermal gradient was
Additionally, it is shown that the two 500 mm thick specimens failed
even more pronounced for deeper units, producing greater tensile
at an early stage when the strand and the concrete temperatures had not
stresses at the concrete webs. Consequently, the web-shear resistance of
even reached 140 ◦ C (Table 9). Below this temperature, based on
PCHC slabs decreased as the slab depth increased.
Eurocode 2, concrete and strand strengths were almost the same as those
The effect of load level on web-shear resistance of PCHC slabs was
at ambient condition, suggesting that under fire conditions, temperature-
evaluated by comparing the fire resistance between F.HC3.500 and F.
induced stresses in concrete due to nonlinear thermal gradient instead of
HC4.500. Although the load level was decreased from 40 to 25%, the
temperature-induced degradation in strength governed the web-shear
500 mm thick slab could only resist the imposed load for additional 16
resistance of deep hollow-core slabs. This observation is supported by
min but still failed within 60 min, as shown in Table 8. Numerical in­
FE results shown in Fig. 26. The figure compares principal-stress dis­
vestigations were extended with two additional cases, i.e. the same 500
tribution inside F.HC3.500 at two different times: at t = 0 (before
mm thick slab was subjected to load levels of 20% and 10%. The results
heating) vs. t at failure. It is shown that at t = 0, F.HC3.500 was in
showed that the 500 mm thick slab could achieve 106 and 159 min of
compression due to prestressing force and there were some locations
fire exposure with these loads, respectively. However, in general, slabs
where the slab was in tension due to the applied load (Fig. 26(a)).
were not subjected to such a low load ratio. Therefore, it is advised to
However, when heat was applied, the entire web of F.HC3.500 was
enhance the web-shear capacity of PCHC slabs through using steel fibers.
subjected to tensile stresses (Fig. 26(b)), making the slab more suscep­
However, more research is needed to quantify this approach.
tible to shear failure under load-cum-heating effect. Failure occurred
when the maximum principal tensile stress in the concrete web reached
5. Shear strengths of the PCHC slabs according to EN 1168:2005 concrete tensile strength.
þ A3:2011 [29]

With regard to shear and anchorage failures, EN 1168:2005 +

15
H.T.N. Nguyen and K.-H. Tan Engineering Structures 227 (2021) 111398

0 10 20 30 40 50 60 70
0 0
5
10
200

Furnace temperature (oC)


15
Max deflection (mm)

20
25 400

30
35 600
40
Test
45
FEM 800
50 Furnace temperature
55
60 1000
0 10 20 30 40 50 60 70

Time (min)

a) F.HC1.320 b) F.HC2.400
0 10 20 30 40 50 60
0 10 20 30 40 50 60 0 0
0 0

5 200
5 200

Furnace temperature ( oC)


Furnace temperature ( C)

Max deflection (mm)


o
Max deflection (mm)

10 400 10 400

15 600 15 600
Test
FEM Test
Furnace temperature FEM
20 800 20 800
Furnace temperature

25 1000
25 1000
0 10 20 30 40 50 60
0 10 20 30 40 50 60
Time (min)
Time (min)
c) F.HC3.500 d) F.HC4.500
Fig. 22. Comparison of displacement development between the tests and the FE models.

Table 7
Comparison between the tests and the FE models.
Slab ID. a/d Applied load (kN) Furnace tempt at failure (◦ C) Failure mode

Test FEM Test/FEM Test FEM

F.HC1.320 2.86 74 945 917 1.03 WS &A WS


F.HC2.400 2.25 204 803 834 0.96 WS WS
F.HC3.500 1.79 464 851 828 1.03 WS WS
F.HC4.500 1.75 290 881 841 1.05 WS WS

*WS & A = Web-shear and Anchorage; WS = web-shear.

6. Conclusions cracks. The formation of web-shear cracks substantially reduced


the load-carrying capacity of the slabs, directly causing brittle
This paper presents experimental and numerical investigations on failure.
the web-shear behavior of deep precast/prestressed concrete hollow 2. The proposed FE models were capable of tracing temperature
core (PCHC) slabs subject to fire. Based on the test results and numerical development and predicting web-shear behavior of PCHC slabs
simulations presented, the following findings can be drawn: under fire conditions. The numerical analyses show that shear
behavior of PCHC slabs subjected to fire was sensitive to dilation
1. Four types of crack were sequentially observed in the structural fire angle. In addition, thermal, longitudinal and web-shear cracks were
tests, i.e. splitting, longitudinal, thermal, and diagonal web-shear well predicted by the FE models. However, the models assumed

16
H.T.N. Nguyen and K.-H. Tan Engineering Structures 227 (2021) 111398

a) Longitudinal crack and web-shear


b) Thermal cracks (F.HC1.320)
failure mode (F.HC3.500)
Fig. 23. Comparison of cracks and failure mode between the experiments and the FE models.

min of fire exposure, displaying premature failure. This finding


agrees well with the test results from Fellinger [7] on 400 mm-thick
hollow-core slabs exposed to the standard fire.
4. With the same load level of 40%, a clear trend of deeper slabs having
lower web-shear resistance was observed. Besides, it is evident that
temperature-induced thermal stress, instead of temperature-induced
degradation in strength, governed the web-shear behavior of deep
PCHC slabs.
5. Finally, the prescriptive method in Eurocode 2 provides uncon­
servative design for fire resistance of deep PCHC slabs subjected to
Fig. 24. Failure of F.HC2.400 due to the formation of web-shear cracks in the shear while the formula in EN 1168 gives safe predictions for shear
FE model. strength of PCHC slabs under elevated temperatures similar to the
ones tested in this paper.
perfect bonding between the concrete and the strand. Therefore,
splitting cracks and anchorage failure could not be simulated. These CRediT authorship contribution statement
are limitations of the developed FE models.
3. Results from verified numerical models showed that being subjected Hang T.N. Nguyen: Conceptualization, Methodology, Validation,
to ISO 834 fire, F.HC1.320 failed at 83 min of fire exposure while the Formal analysis, Investigation, Writing - original draft, Visualization.
other three slabs with depth equal or exceeding 400 failed within 60 Kang-Hai Tan: Supervision, Writing - review & editing, Funding
acquisition.

a) F.HC3.500 b) F.HC4.500
Fig. 25. Web-shear failure of PCHC slabs subjected to ISO-834 fire curve in the FE models.

Table 8
Summary of numerical results.
Slab ID. Applied load (kN) Load level (%) Experiments FEM (exposed to ISO 834 fire)

Maximum furnace temperature (◦ C) Maximum furnace temperature (◦ C) Fire resistance (min)

F.HC1.320 74 40 945 993 83


F.HC2.400 204 40 845 856 33
F.HC3.500 464 40 851 815 24
F.HC4.500 290 25 881 891 40

17
H.T.N. Nguyen and K.-H. Tan Engineering Structures 227 (2021) 111398

Table 9
Strand and concrete temperatures and their reduction strength factors based on Eurocode 2.
Slab ID. Time at failure (mm) Strand temperature (oC) Reduction factor for strand strength Mid web temperature (◦ C)

F.HC1.320 57 226 0.83 100


F.HC2.400 118 269 0.77 99
F.HC3.500 48 118 0.97 34
F.HC4.500 52 135 0.95 43

Table 10
Shear-strength comparison.
Slab ID. a/d Time at failure Experimental failure load (kN)* EN 1168 failure load (kN) Vtest/Vpredict

F.HC1.320 2.86 57 104 70.5 1.48


F.HC2.400 2.25 118 245 144.5 1.70
F.HC3.500 1.79 48 500 166.6 3.00
F.HC4.500 1.79 52 326 163.8 1.99
*
Including weight of specimen and loading system.

a) t = 0 (before heating) b) t at failure


Fig. 26. Principal-stress comparison: Before heating vs. at failure (F.HC3.500).

Declaration of Competing Interest [5] Andersen N, Lauridsen D. Danish Institute of Fire Technology Technical Report X
52650 Part 2-Hollow core Concrete Slabs. Hvidovre: DIFT; 1999.
[6] Borgogno W, Fontana M. Structural behaviour of slim floor slabs with prestressed
The authors declare that they have no known competing financial hollow core elements at room temperature and in fire. Proceedings of the
interests or personal relationships that could have appeared to influence conference: Composite construction in steel and concrete IV. 2000.
the work reported in this paper. [7] Fellinger J. Shear and anchorage behaviour of fire exposed hollow core slabs. TU
Delft: Delft University of Technology; 2004.
[8] Jansze W, Acker AV, Klein-Holte R. Fire resistance of hollow core floors regarding
Acknowledgment shear and anchorage capacity. In: Structures in Fire. Zurich, Switzerland; 2012. p.
709–18.
[9] Dotreppe J-C, Franssen J-M. Precast hollow core slabs in fire: numerical
The collaboration of SPC Industries SDN BHD (Malaysia) is highly simulations and experimental tests. Proceedings of SiF’04: Third International
appreciated. The support of the National University of Civil Engineering, Structures in Fire workshop. IRC; 2004.
Vietnam where the first author is a lecturer for granting the first author a [10] Fellinger J. Shear and anchorage behavior of fire exposed hollow core slabs. Heron
2005;Vol. 50(No 4):279–301.
sabbatical leave to conduct this research at Nanyang Technological [11] Chang JJ, Moss PJ, Dhakal RP, Buchanan AH. Effect of aspect ratio on fire
University is also acknowledged. This material is based on research/ resistance of hollow core concrete floors. Fire Technol 2010;46(1):201–16.
work supported by the Singapore Ministry of National Development and [12] Ellobody E. Advanced analysis of prestressed hollow core concrete slabs exposed to
different fires. Adv Struct Eng 2014;17(9):1281–98.
National Research Foundation under L2 NIC Award No. L2NICCFP1- [13] Kodur VKR, Shakya AM. Factors governing the shear response of prestressed
2013-4. Any opinions, findings, and conclusions expressed in this ma­ concrete hollowcore slabs under fire conditions. Fire Saf J 2017;88:67–88.
terial are those of the authors and do not necessarily reflect the views of [14] Nguyen TNH, Tan K-H, Kanda T. Investigations on web-shear behavior of deep
precast, prestressed concrete hollow core slabs. Eng Struct 2019;183:579–93.
L2 NIC.
[15] EN-1992-1-2. Eurocode 2: Design of concrete structures, in Part 1-2: General rules -
Structural fire design. Brussels; 2004.
References [16] EN-1992-1-1. Eurocode 2: Design of concrete structures, in Part 1-1: General rules
and rules for buildings. Brussels; 2004.
[1] Hawkins NM, Ghosh S. Shear strength of hollow-core slabs. PCI J 2006;51(1): [17] Nguyen HTN, Tan KH. Experimental studies on shear behavior of deep prestressed
110–4. concrete hollow core slabs under fire conditions. In: 10th international conference
[2] Walraven JC, Mercx WPM. The bearing capacity of prestressed hollow core slabs. on structures in fire, Belfast, Northern Ireland, UK, 6-8 June, 2018, A. Nadjai, et al.,
Heron 1983;Vol. 28(No. 3). Editors. 2018. p. 51-59.
[3] Rahman MK, Baluch MH, Said MK, Shazali MA. Flexural and shear strength of [18] 1363-1, C.E., Fire Resistance Tests. Part 1: General Requirements, in Part 1:
prestressed precast hollow-core slabs. Arab J Sci Eng 2012;37(2):443–55. General Requirements. Brussels; 1999.
[4] Palmer KD, Schultz AE. Experimental investigation of the web-shear strength of [19] Simulia AV. 6.14 Documentation. Dassault systemes 2014.
deep hollow-core units. PCI J 2011:83–104. [20] Shakya AM, Kodur VKR. Response of precast prestressed concrete hollowcore slabs
under fire conditions. Eng Struct 2015;87:126–38.

18
H.T.N. Nguyen and K.-H. Tan Engineering Structures 227 (2021) 111398

[21] Aguado JV, Albero V, Espinos A, Hospitaler A, Romero ML. A 3D finite element [26] Gao WY, Dai J-G, Teng JG, Chen GM. Finite element modeling of reinforced
model for predicting the fire behavior of hollow-core slabs. Eng Struct 2016;108: concrete beams exposed to fire. Eng Struct 2013;52:488–501.
12–27. [27] Genikomsou K, Polak M. Effect of openings on punching shear strength of
[22] Shakya AM. Flexural and shear response of precast prestressed concrete hollocore reinforced concrete slabs—finite element investigation. ACI Struct J 2017;114:
slabs under fire conditions. Michigan State University; 2016. 1249–61.
[23] Demir A, Öztürk H, Dok G. 3D numerical modeling of RC deep beam behavior by [28] Kordina K, Ehm C, Schneider U. Effect of biaxial loading on the high temperature
nonlinear finite element analysis. Disaster Sci Eng 2016;2(1):13–8. behaviour of concrete. In: Proceeding of the first international symposium on fire
[24] Hordijk DA. Tensile and tensile fatigue behaviour of concrete; experiments, safety science, in fire safety science-proceedings of the first international
modelling and analyses. Heron 1992;37(1). symposium. Gaithersburg, MD; 1985. p. 281–90.
[25] FIB, fib model code for concrete structures 2010. 2010: Case Postale 88, CH-1015 [29] EN-1168:2005+A3, European standard, in precast concrete products - Hollow core
Lausanne, Switzerland. slabs. Brussels; 2011.

19

You might also like