Beruflich Dokumente
Kultur Dokumente
h i g h l i g h t s
A finite element based numerical model is proposed for tracing fire response of load bearing concrete walls.
Validity of the proposed model is established.
Results from model are utilized to characterize three distinct stages in fire response of walls.
a r t i c l e i n f o a b s t r a c t
Article history: A generic three-dimensional (3D) finite element (FE) based numerical model is presented for predicting
Received 30 April 2017 thermo-mechanical behavior of load bearing reinforced concrete (RC) walls exposed to fire. The proposed
Received in revised form 1 August 2017 model is capable of accounting for critical parameters governing fire resistance of RC walls including wall
Accepted 3 August 2017
slenderness ratio, support restrains, and temperature dependent properties of reinforcement and con-
Available online 11 August 2017
crete. The model is validated by comparing predicted thermal and structural response parameters with
the experimental data on three full scale load bearing RC walls tested under fire exposure. The compar-
Keywords:
isons show good correlation between model predictions and measured data, indicating that the proposed
Concrete walls
Fire exposure
model can predict the thermo-mechanical behavior of RC walls from pre-loading to collapse stage under
Fire resistance fire exposure. The validated model can be applied to undertake parametric studies aimed at quantifying
Numerical model critical factors governing fire performance of load bearing RC walls under fire.
Ó 2017 Elsevier Ltd. All rights reserved.
1. Introduction Typically, walls are to be designed for fire safety assuming one
side fire exposure, and such a condition can develop steep thermal
Reinforced concrete (RC) walls are widely used as vertical sep- gradients within wall cross section. These thermal gradients cause
arations in buildings due to their cost effectiveness, architectural uneven material degradation across wall cross section at elevated
aesthetics, better space utilization, and low maintenance costs. temperatures, which in turn imparts eccentricity to gravity load-
These walls provide significantly higher axial and out-of-plane ing. Thus, fire exposure has strong potential for inducing out-of-
strength as compared to masonry infill walls, which results in plane instability in RC walls. However, studies pertaining to fire
small size requirements for the framing members; thus, leading behavior of walls, especially under eccentric loading, are rather
to larger space and economical construction. Consequently, the scarce in the literature.
use of load bearing RC walls in high rise building construction In case of fire, RC walls are expected to satisfy three criteria
has gained popularity in the recent decades. In these buildings, (failure limit states); which include ability to carry applied load
RC walls are subjected to both in-plane (gravity loads from upper under fire exposure (stability criterion), ability to keep tempera-
floors) and out-of-plane loads (arising from the eccentricity in ture on unexposed face below ignition temperature (insulation cri-
gravity loading and/or wind loading); and play a key role in devel- terion), and ability to provide fire compartmentation by preventing
oping load transfer mechanisms in the building. cracks and fissures (integrity criterion). Currently, these three cri-
teria are assumed to be satisfied if prescribed member dimensions
(for e.g. minimum thickness of the wall, concrete cover to main
⇑ Corresponding author. reinforcement etc.) are provided as per prescriptive code based
E-mail addresses: kumarpu2@msu.edu (P. Kumar), kodur@egr.msu.edu specifications. Such prescriptive approaches are developed under
(V.K.R. Kodur).
http://dx.doi.org/10.1016/j.conbuildmat.2017.08.010
0950-0618/Ó 2017 Elsevier Ltd. All rights reserved.
994 P. Kumar, V.K.R. Kodur / Construction and Building Materials 154 (2017) 993–1003
standard fire conditions and do not take into consideration many eccentric loading. Flexural cracking allows dissipation of pore
critical factors governing the fire behavior of RC walls such as load water pressure within wall under fire exposure, which reduces
level, wall slenderness, load eccentricity etc. the risk of explosive spalling. Therefore, walls with relatively low
Moreover, there is lack of validated numerical models for pre- flexural cracking are more prone to explosive spalling, which is
dicting realistic behavior of RC walls under fire exposure. Most of reflected by results of the study.
the previously reported numerical studies on simulating behavior Go et al. [9] studied the effect of aggregate density (normal
of RC walls under fire are either two-dimensional (2D) idealiza- weight, and light weight concrete), vertical reinforcement spacing
tions or do not account for all critical factors governing fire behav- (10 mm, 20 mm, and 30 mm), and width of wall (1000 mm and
ior of RC walls [1–5]. In order to overcome current drawbacks, a 1500 mm) on post-fire in-plane structural response of five RC walls
generic 3D finite element (FE) based numerical model is developed (1500 mm 75 mm). The concrete strength for all wall specimen
in ANSYS [6]. Comprehensive details of the model development varied between 20 and 22 MPa, and they were subjected to one
along with its validation against fire test data are presented in this sided standard fire exposure. The post-fire in-plane structural
paper. Special consideration is given to incorporate thermal bow- response of walls was compared with that of walls tested at room
ing and reverse bowing in evaluating structural performance under temperature, and it was concluded that lightweight concrete walls
fire exposure, which play a key role in characterizing stability of retained relatively higher in-plane strength, stiffness, and ductility
concrete walls under fire exposure. as compared to normal weight concrete walls.
Lee et al. [10] studied the fire behavior of eight RC walls by
exposing both faces of wall to standard fire exposure. The key vari-
2. Behavior of RC walls under fire
ables of this study include: thickness of wall (100 mm–200 mm),
load ratio (0–6%), amount of reinforcement (0.3% and 0.6% with
There are limited numerical [1–5] and experimental [7–14]
equal vertical and horizontal reinforcement ratio), concrete
studies on behavior of load-bearing RC walls under fire. Some of
strength (10 MPa–40 MPa), and curing duration of concrete (38
the main findings in these studies are reported below.
and 120 days). The results of the study show that thin wall speci-
mens with shorter curing duration and high strength concrete
2.1. Experimental studies are susceptible to premature failure under fire exposure. Kang et.
al. [11] studied the effect of wall thickness (150 mm–250 mm)
Crozier and Sanjayan [7] tested a total of eighteen load bearing and moisture content (5–7%) on development of temperature pro-
slender RC walls (3600 mm 1200 mm) by exposing them to stan- file within RC wall cross-section, and results of this study illustrate
dard fire exposure [15] on the tension face of walls. The varied the influence of moisture content on thermal response of RC walls.
parameters included wall thickness (75 mm, 100 mm, and Ngo et al. [12] studied the effect of eccentric loading (10 mm
150 mm), clear cover to rebars (30 mm–110 mm), slenderness eccentricity), concrete strength (36 MPa–89 MPa), and polypropy-
(24, 36, and 48 height to thickness ratio), concrete strength lene fibers on ten full-scale (2400 mm 1000 mm 150 mm) RC
(44 MPa–70 MPa), and varying levels of eccentric in-plane loads. walls subjected to standard and hydrocarbon fire exposures. The
The in-plane load carrying capacity was evaluated through tests results of the study show that all normal strength concrete (NSC)
on eight wall specimens, supported on two short edges and sub- walls withstood fire exposure of 120 min with little or moderate
jected to eccentric in-plane and lateral loads –resulting from self- spalling under standard and hydrocarbon fire exposure. However,
weight of the wall in horizontal position. Also, to investigate the walls fabricated with high strength concrete (HSC) experienced
effect of spalling and thermal bowing, eight additional walls were severe spalling in case of hydrocarbon fire exposure, which led to
tested under the combined effect of gravity loading and standard ultimate collapse of the wall at 31 min. The authors inferred that
fire exposure. Based on the results, the study concluded that (a) this spalling problem in HSC can be remedied to a certain extent
thermal bowing degrades the in-plane load carrying capacity of by adding polypropylene fibers into HSC mix. This was shown
walls significantly; (b) strength of concrete had little influence on through testing a wall by adding polypropylene fibers, which
the fire behavior of RC walls; and (c) walls with vertical reinforce- enhanced the fire resistance of modified wall to 65 min (from
ment at mid-thickness performed better than doubly reinforced 31 min) under hydrocarbon fire exposure.
walls (i.e. rebars at two outer surfaces). Mueller et al. [13] tested two load bearing RC walls
It should be noted that testing of these walls in horizontal posi- (380 mm 1020 mm 3050 mm) with varying lateral restrains
tion led to development of significant flexural cracks on the tension under standard fire exposure. Both walls had similar reinforcement
side of walls, which led to degradation in fire resistance of walls. detailing with an average test day concrete strength of 51 MPa and
However, such flexural cracking is contrary to the service load con- 47 MPa, respectively. The walls were subjected to an axial load of
ditions of walls, as walls are predominantly subjected to axial 2400 kN, representing constant dead load from superstructure,
loads. This stimulated researchers to focus on providing realistic with fixed boundary conditions at the bottom. For one wall, the lat-
service loading conditions during fire tests, hence, most of the eral displacement was constrained using an actuator at about
experimental studies after Crozier and Sanjayan [7] pursued test- 230 mm from the top of wall; whereas, another was subjected to
ing of RC walls in vertical position along with appropriate support a step wise incremental lateral load, pushing wall towards the fur-
conditions [8–14]. nace at the same location. The thermo-mechanical behavior was
Guerrieri and Fragomeni [8] tested four slender monitored using conventional strain gauges, digital image correla-
(1300 mm 1300 mm 50 mm) concrete walls, by exposing one tion, and infrared thermography. The results of this study revealed
face to standard fire exposure, to gauge fire induced spalling in that constraining thermal bowing of RC walls can result in devel-
walls. All walls were tested after a duration of six months from opment of significant out-of-plane loads. Therefore, boundary con-
casting, with reported concrete strength to be approximately ditions play a key role in characterizing thermo-mechanical
40 MPa. Two of these walls were tested under the effect of self- behavior of RC walls.
weight alone, whereas, other two were subjected to an eccentric Same experimental setup was further utilized by Mueller and
axial loading. In the tests, explosive spalling was observed in walls Kurama [14] to evaluate the effect of concrete strength (47 MPa–
subjected to self-weight alone, and walls under eccentric loading 123 MPa average test day strength), eccentric loading, wall thick-
did not experience any explosive spalling. This is explained on ness (203 mm–380 mm), reinforcement detailing, end restrains,
the account of relatively low flexural cracking in walls without and lateral loading on thermo-mechanical behavior of load bearing
P. Kumar, V.K.R. Kodur / Construction and Building Materials 154 (2017) 993–1003 995
Fig. 1. Developed FE model for RC wall, (a) symmetric FE model, (b) reinforcement detailing.
F
S 0; ð3Þ
fc
In which, F is a function of principal stress state; and S is a con- f cb ¼ 1:2f c ; f 1 ¼ 1:45f c ; and f 2 ¼ 1:725f c ; ð4Þ
tinuous failure surface with five temperature dependent input
parameters as: uniaxial tensile strength, ft, uniaxial compressive Once the failure criterion for crushing of concrete is satisfied,
strength, fc, biaxial compressive strength, fcb, and ultimate com- the stiffness of the corresponding SOLID65 element is reduced to
pressive strengths for a biaxial compression superimposed on a negligible value thus making it crushed. Whereas, in case of
hydrostatic stress state i.e. f1 and f2. The relative magnitudes of cracking, a plane of weakness is introduced in a direction perpen-
principal stresses on the octahedral plane are defined using angle dicular to that of the tensile principal stress which satisfies the fail-
of similarity, g, and parameters r1 and r2 (see Fig. 2). The resulting ure criterion. To implement this reduction in stiffness due to
three-dimensional failure envelope in the principal stress-state is tensile cracks additional parameters, open (bt) and closed crack
shown in the Fig. 2, and detailed mathematical formulation of (bc) factors, are utilized by SOLID65 elements. The value of these
the same can be found elsewhere [21]. factors range between 0 and 1, with 0 representing complete loss
In case of absence of experimental data, failure surface (S) can of shear transfer (smooth crack) and 1 representing no loss of loss
be represented using just two fundamental strength characteristics of shear transfer (rough crack). Based on the preliminary studies, bt
of concrete as ft and fc [6]. In this case, other input parameters are and bc are assigned a value of 0.53 and 0.98 in the developed
assigned a default value as: model, respectively.
P. Kumar, V.K.R. Kodur / Construction and Building Materials 154 (2017) 993–1003 997
3.4. Numerical convergence criteria experimental results in thermal and structural domains to gauge
the efficacy of the model in predicting behavior of RC walls under
In the above analysis procedure, transient thermal analysis uti- fire.
lizes an explicit solver, and a Newton-Raphson iterative solver is
used in the structural analysis. The convergence of the FE solution
is defined using stringent vector norms in both analysis domains, 4.1. Selection and description of RC walls utilized for validation
ensuring physical behavior of the FE model. However, temperature
dependent cracking and crushing of concrete, high temperature Some of the major challenges in utilizing previously discussed
constitutive models, and material and geometric non-linearities experimental studies [7–11] for validating numerical models are:
can often lead to an ill-conditioned stiffness matrix in the struc- (a) impractical member dimensions -arising from testing equip-
tural analysis. This imparts several convergence hurdles to the FE ment limitations (for example: 50, and 75 mm wall thickness;
analysis and often leads to divergence from reaching a solution, and 1300, and 1500 mm height), (b) lack/no reporting of compre-
inhibiting further progress of FE analysis. To overcome such con- hensive structural response which is vital for validating structural
vergence hurdles, the load is incremented gradually using very response, and (c) use of low load levels in tests, representing typ-
small load increments which allows the appropriate plastic defor- ical non-load bearing walls. These limitations are well accounted
mation of the cracked and crushed concrete and continuity of the for in the experimental studies of Ngo et al. [12], Mueller et al.
FE analysis. However, it should be noted that using very small load [13], and Mueller and Kurama [14]; hence, these studies have been
increments enhances the computational efforts exponentially. utilized for the validation of developed FE model. Comprehensive
Therefore, smaller load increments are only implemented at con- details of the selected RC walls from these studies are provided
vergence hurdle points and the solution is allowed to follow nor- in Table 1.
mal load increments otherwise. Wall-W1 represent a typical load bearing wall in low rise con-
struction, and was tested by Ngo et al. [12] under one face standard
3.5. Failure limit states fire exposure. The wall was subjected to a gravity/axial load of
485 kN at an eccentricity of 10 mm –towards fire exposed face
RC wall can attain failure under fire exposure through reaching from the central axis of wall. This eccentricity in axial loading is
insulation, or integrity, or stability limit states (as discussed in Sec- representative of typical accidental eccentricity (eccentricity due
tion 1). Therefore, to evaluate fire resistance, failure of wall should to construction imperfections, geometry etc.) in the service loading
be checked against these failure limit states at each time step of of walls, hence, provides an ideal case to gauge its effect on thermal
analysis. The failure of wall in insulation limit state occurs when bowing of wall under fire exposure. The wall was tested under sim-
temperature rise on the unexposed face exceeds 139 °C above its ply supported conditions at top and bottom, and withstood fire
initial temperature [23]. The integrity criterion is assessed as per exposure for a duration of two hours with very small amount of
the recommendations of Eurocode 2 [20], according to which spalling.
requirements for integrity criterion are considered satisfied where Wall-W2 and Wall-W3 represent load bearing walls in high rise
the minimum thickness of wall and concrete cover are in accor- buildings with high gravity loads from the superstructure. These
dance with prescribed values. As per stability criterion, the failure walls were tested by Mueller et al. [13], and Mueller and Kurama
occurs when the capacity at a critical section falls below the [14] using identical experimental setup. These walls represent
applied loading effects. In addition to these failure limit states, two typical configurations of reinforcement in load bearing walls,
ISO834 [24] specifies deflection or rate of deflection as a failure illustrated in Fig. 3. Therefore, validation using these studies allow
limit state for walls. According to ISO834 [24] compression mem- the numerical model to account for typical reinforcement configu-
ber (wall) undergoes failure when the axial deformation exceeds rations in load bearing walls. Both walls are subjected to an axial
h 3h load of 2400 kN, which is kept constant until the failure of wall.
100
mm or the rate of deformation reaches 1000 (mm/min) during
Also, these walls are subjected to a varying lateral load at a dis-
fire exposure, where h denotes the height of wall in mm.
tance of 230 mm from the top of wall –representing typical
restraint from the floor slab assembly. Bottom of the walls is fixed
4. Model validation to the laboratory strong floor using a concrete base, whereas sim-
ply supported conditions are maintained at the top. Once the axial
The above developed FE model is validated against data from and initial lateral loading stabilize, wall is subjected to one face
fire tests on full-scale load bearing RC walls [12–14] (see Table 1). standard fire exposure over half height of the wall using a custom
Predictions from the developed FE model are compared with furnace. Comprehensive thermo-mechanical response, in terms of
Table 1
Description of RC walls selected for validation.
Fig. 5. Comparison of predicted and measured cross sectional temperatures at various depths for Wall-W1.
values. However, the final predicted and measured exposed face face of the wall. This variation in temperature at exposed face is
temperatures are very close to each other i.e. 845 °C and 865 °C, explained by possible dislocation of thermocouples at exposed face
respectively. For all other depths of consideration, model predic- of wall during the casting process [14]. Moreover, capability of the
tions are very close to the experimental results in beginning, and developed FE model in predicting temperatures near exposed face
then slightly high temperatures are predicted; making tempera- can be clearly observed from Fig. 5. Therefore, the developed FE
ture predictions slightly conservative. It should be noted that as model is capable of predicting temperatures at any depth within
temperatures predicted at various depths throughout the fire test RC walls.
are in good agreement with measured values, therefore, thermal
gradients are also predicted with similar accuracy throughout the
fire test. This manifest the trust in FE model to predict accurate 4.4. Structural response
thermal gradients, which play a key role in thermal bowing and
spalling of concrete. To illustrate validity of the FE model in predicting thermo-
In case of Wall-W2 and Wall-W3, both walls share similar mechanical response of RC walls under fire exposure, the out-of-
geometry, thermal properties, and fire exposure which results in plane displacement predictions are compared with the fire test
a similar thermal response. The temperature history is reported data in Figs. 7–9. In general, the out-of-plane displacement
at depths near the exposed face and unexposed face (25 mm from response of RC walls under one face exposure can be categorized
the face of wall) for both walls. Therefore, the temperature data into three main stages, based on their thermo-mechanical response
from these studies is not as suitable as Wall-W1 data to analyze (see Figs. 7, 8(a), and 9(a)). In stage 1, high thermal inertia of con-
the efficacy of model in predicting temperatures at any depth crete leads to development of steep thermal gradients within con-
within wall. A comparison between predicted and measured tem- crete cross section, while keeping temperatures within wall cross
peratures for Wall-W2 is illustrated in Fig. 6. It can be observed section relatively below 500 °C. This is illustrated by large differ-
from Fig. 6 that temperature predictions of model near unexposed ence in the temperature rise near exposed and unexposed face
face of wall are in good agreement with the experimental data.
However, the model over predicts temperatures near fire exposed
Fig. 6. Comparison of predicted and measured cross sectional temperatures in Fig. 7. Comparison of predicted and measured out-of-plane displacements for
Wall-W2. Wall-W1.
1000 P. Kumar, V.K.R. Kodur / Construction and Building Materials 154 (2017) 993–1003
Fig. 8. Comparison between out-of-plane displacement predictions and Wall-W2 test data at wall height of (a) 2.29 m, (b) 2.79 m and 0.76 m, (c) 1.78 m, and (d) 1.27 m.
(see Figs. 5 and 6). This results in relatively high thermal expansion Beyond stage 2, the effect of shift in neutral axis (resulting from
near exposed face, as compared to near unexposed face, and causes material degradation) dominates the thermo-mechanical response
movement of wall away from exposed face (thermal bowing). of wall, and is marked as stage 3. Initial eccentricity in axial loading
Therefore, this stage is marked by rapid increase in the out-of- and restraints from floor-slab assembly play a key role in charac-
plane displacement, where the effect of material degradation is terizing rate of displacement in this stage. The total duration of
not significant as temperatures within wall (away from exposed stage 3 depends on wall material properties, geometry, loading
face) are considerably below 500 °C. and boundary conditions, and duration of fire exposure. Comple-
Once the temperatures near exposed face exceed 500 °C, signif- tion of this stage marks onset of failure in strength domain at
icant material degradation occurs in the strength and stiffness which major failure modes can be observed in terms of excessive
properties of concrete and steel [20]. This causes a shift in the neu- deformation, buckling etc.
tral axis of axial loading towards unexposed face, and imparts It can be clearly observed from Figs. 7–9 that the model predic-
additional eccentricity to axial loading. The moment caused by this tions are in good agreement with test data for all three walls, and it
eccentricity counteracts the movement of wall away from exposed correctly predicts occurrence and trends in structural response of
face, which can be clearly observed by the decreased rate of out-of- walls for all three stages. In case of Wall-W1, the model predictions
plane displacement for all three walls after stage 1 (see Figs. 7–9). are in good agreement with measured values in stage 1 and 3.
Once the moment caused by shift of neutral axis exceeds the effect However, the model slightly over predicts out-of-plane displace-
of displacement caused due to thermal gradients, wall starts to ments (by approximately 3 mm) in stage 2. This is primarily due
move towards exposed face. This is illustrated by onset of recovery to the perfect bond assumption between steel reinforcement and
in out-of-plane displacement, and can be clearly observed from concrete, which is implemented in model to save computational
Figs. 7–9. This change in movement of wall often leads to reversal effort. However, in reality, there is some level of slippage at
in the initial deflected shape of RC wall, and is termed as reverse rebar-concrete interface at elevated temperatures. Due to this
bowing. The onset of recovery of out-of-plane displacement, when assumption, the thermal strains of steel rebars (with relatively
wall starts to progress towards reverse bowing, is marked as com- higher thermal expansion [20]) are completely transferred to con-
pletion of stage 2. crete elements, and cause wall to deform beyond the measured
P. Kumar, V.K.R. Kodur / Construction and Building Materials 154 (2017) 993–1003 1001
Fig. 9. Comparison between out-of-plane displacement predictions and Wall-W3 test data at wall height of (a) 2.79 m, (b) 2.29 m and 0.25 m, (c) 1.78 m, and (d) 1.27 m.
value. This effect becomes more pronounced in the stage 2 as 4.5. Failure modes and thermal bowing
material degradation at elevated temperatures (which dominates
this stage) has significant impact on bond between rebars and con- One of the governing failure mechanism in RC walls subjected
crete. However, this effect is not captured currently by the model, to one face fire exposure, is through buckling during reverse bow-
due to lack of reliable high temperature bond properties. Hence, ing under the influence of lateral restrains and shift in neutral axis
the model slightly over predicts in stage 2. at elevated temperatures. This has been observed in various exper-
A similar effect of bond on structural response in stage 2 can be imental studies [12–14]. Mueller et al. [13] and Mueller and Kur-
observed for Wall-W2 and Wall-W3 as well (see Figs. 8 and 9). For ama [14] demonstrated that thermal bowing and reverse bowing
both walls, the effect of bond becomes significant towards end of played a key role in the failure of RC walls under fire exposure.
stage 1, and attains its maximum effect in stage 2. However, the Also, they showed that restraining even small displacements in
effect of bond is more pronounced in Wall-W2 as compared to RC wall could lead to significant additional lateral forces in the wall
Wall-W3, which is mainly due to relatively high reinforcement for which it may not be designed. Therefore, thermal bowing and
ratio in Wall-W2. Beyond stage 2, the effect of bond is slightly reverse bowing phenomenon plays a key role in determining the
reduced as shift in neutral axis and lateral restrains become more failure of RC walls under fire exposure, and can be utilized as an
dominating factors for structural response of walls. At the comple- important tool to investigate the same.
tion of stage 3, failure due to excessive deformation is observed for Moreover, deflected shape of wall can be utilized by several
Wall-W2 and Wall-W3. This is attributed to excessive spalling on simplified numerical models for the validation of their predicted
the exposed face which exposes the rebars to direct fire exposure, response [1,2,4,5]. Therefore, deflection predictions from model
as temperatures increase well beyond 800 °C. Currently, the model can be utilized to develop simplified two dimensional (2D) models
does not account for the spalling of concrete, hence, the model for RC walls. Such models can provide significant computational
under predicts structural response beyond this stage. It should be benefits over the three-dimensional (3D) models in terms of gener-
noted that even with these two limitations in terms of bond and ating large databases. Hence, in order to assess the capability of
spalling, the model predictions are in good agreement with struc- developed FE model in capturing failure modes of RC walls under
tural response. Therefore, the numerical model can be utilized to fire, a comparison between predicted and measured deflected
capture the thermo-mechanical response of RC walls. shapes is shown in Figs. 10 and 11, respectively.
1002 P. Kumar, V.K.R. Kodur / Construction and Building Materials 154 (2017) 993–1003
Fig. 10. Comparison between predicted and measured out-of-plane deformation Fig. 11. Comparison between predicted and measured out-of-plane deformation
profile for Wall-W2 at (a) initial loading, (b) 1 h, (c) 2.5 h, (d) 3.5 h, (e) 5 h, (f) 7.5 h, profile for Wall-W3 at (a) initial loading, (b) 1 h, (c) 2.5 h, (d) 3.5 h, (e) 5 h, (f) 7.5 h,
and (g) 10.2 h fire exposure. and (g) 9.7 h fire exposure.
P. Kumar, V.K.R. Kodur / Construction and Building Materials 154 (2017) 993–1003 1003
It should be noted that the accuracy in prediction of RC wall [2] A.J. O’Meagher, I.D. Bennetts, Modelling of concrete walls in fire, Fire Safety J.
17 (4) (1991) 315–335.
deflected shape is directly dependent on the capability of model
[3] S. Lee, C. Lee, Fire resistance of reinforced concrete bearing walls subjected to
to predict out-of-plane displacements along various heights of all-sided fire exposure, Mater. Struct. 46 (6) (2013) 943–957.
the wall. Therefore, similar to the displacement response discussed [4] S.F. El-Fitiany, M.A. Youssef, Assessing the flexural and axial behaviour of
in previous section, the model predicts deflected shape with rea- reinforced concrete members at elevated temperatures using sectional
analysis, Fire Safety J. 44 (5) (2009) 691–703.
sonable accuracy in all three stages. Further, it can be clearly [5] D.T. Pham, P. de Buhan, C. Florence, J.V. Heck, H.H. Nguyen, Yield design-based
observed from Figs. 10 and 11 that model correctly predicts the analysis of high rise concrete walls subjected to fire loading conditions, Eng.
thermal bowing and reverse bowing trends for both walls. Hence, Struct. 87 (2015) 153–161.
[6] ANSYS, Mechanical APDL version 15, Canonsberg (PA): ANSYS Inc. (2013).
the model can be utilized to generate deflected profiles of the RC [7] D.A. Crozier, J.G. Sanjayan, Tests of load-bearing slender reinforced concrete
walls under fire exposure which can be further utilized to develop walls in fire, ACI Struct. J. 97 (2) (2000) 243–251.
simplified 2D numerical models. [8] M. Guerrieri, S. Fragomeni, Spalling of normal strength concrete walls in fire,
in: 6th International Conference on Structures in Fire, East Lansing, USA (2010)
301–311.
[9] C.G. Go, J.R. Tang, J.H. Chi, C.T. Chen, Y.L. Huang, Fire-resistance property of
5. Conclusions reinforced lightweight aggregate concrete wall, Construction Building Mater.
30 (2012) 725–733.
A FE based numerical model is proposed to predict the fire [10] C. Lee, S. Lee, H. Kim, Experimental observations on reinforced concrete
bearing walls subjected to all-sided fire exposure, Magazine Concrete Res. 65
response of load bearing RC walls. Predictions from the numerical (2) (2013) 82–92.
model are compared against fire test data in structural and thermal [11] J. Kang, H. Yoon, W. Kim, V. Kodur, Y. Shin, H. Kim, Effect of wall thickness on
domains to illustrate the efficacy of proposed model in predicting thermal behaviors of RC walls under fire conditions, Int. J. Concrete Struct.
Mater. 10 (3) (2016) 19–31.
realistic fire behavior of RC walls. Based on the results presented
[12] T. Ngo, S. Fragomeni, P. Mendis, B. Ta, Testing of normal-and high-strength
in this paper, the following conclusions can be drawn: concrete walls subjected to both standard and hydrocarbon fires, ACI Struct. J.
110 (3) (2013) 503–510.
[13] K.A. Mueller, Y.C. Kurama, M.J. McGinnis, Out-of-plane behavior of two
The proposed finite element model is capable of predicting the
reinforced concrete bearing walls under fire: a full-scale experimental
response of RC walls from initial loading to collapse stage under investigation, ACI Struct. J. 111 (5) (2014) 1–10.
combined effects of fire exposure and structural loading. [14] K.A. Mueller, Y.C. Kurama, Out-of-plane behavior and stability of five planar
High thermal inertia of concrete leads to development of steep reinforced concrete bearing wall specimens under fire, ACI Struct. J. 112 (6)
(2015) 701–712.
thermal gradients within a RC wall, which imparts out-of-plane [15] AS1530.4, Methods for fire tests on building materials, components and
instability to wall under fire loading. structures: part 4 fire resistance tests of elements of building construction,
Structural response of a RC wall under fire exposure occurs in Standards Association of Australia, 1997.
[16] ACI-216.1-07, Code requirements for determining fire resistance of concrete
three distinct stages. First stage is marked by rapid increase in and masonry construction assemblies, Reported by Joint ACI-TMS Committee
out-of-plane displacement, resulting from steep thermal gradi- 216, 2007.
ents. While second stage is mainly characterized by shift in the [17] M.B. Dwaikat, V.K.R. Kodur, Hydrothermal model for predicting fire-induced
spalling in concrete structural systems, Fire Safety J. 44 (3) (2009) 425–434.
neutral axis, resulting from material degradation at elevated [18] V. Kodur, Properties of concrete at elevated temperatures, ISRN Civil Eng. 2014
temperatures. The third stage is marked by onset of reverse (2014), http://dx.doi.org/10.1155/2014/468510.
bowing, representing dominance of shift in neutral axis effects [19] V.K.R. Kodur, M.M.S. Dwaikat, M.B. Dwaikat, High-temperature properties of
concrete for fire resistance modeling of structures, ACI Mater. J. 105 (5) (2008)
over thermal expansion due to thermal gradients.
517–527.
The proposed model can capture the thermal bowing and [20] Eurocode 2, Design of concrete structures, Part 1-2: General rules-structural
reverse bowing phenomenon, which plays a critical role in ini- fire design, CEN: European Committee for Standardization, UK, 2004.
[21] K. Willam, E. Warnke, Constitutive model for the triaxial behavior of concrete,
tiating failure mechanism in a load bearing RC wall under fire
Proceedings, Int. Assoc. Bridge Struct. Eng., ISMES, Bergamo, Italy 19 (1975) 1–
exposure. 30, http://dx.doi.org/10.5169/seals-17526.
Results from the proposed numerical model can be further uti- [22] P. Kumar, Characterization of in-plane and out-of-plane behavior of infill
lized to develop simplified 2D numerical models for evaluating panels subjected to thermal exposure Master’s Thesis, Indian Institute of
Technology, Gandhinagar, 2015. http://repository.iitgn.ac.in/handle/
fire resistance of load bearing RC walls. 123456789/1889.
[23] E119-16a, Standard test methods for fire tests of building construction and
materials, American Society for Testing and Materials, ASTM, West
Conshohocken, PA, 2016.
References [24] ISO, Fire resistance tests-elements of building construction, ISO 834-1,
International Organization for Standardization, Geneva, Switzerland (2012).
[1] A.H. Buchanan, V.R. Munukutla, Fire resistance of load-bearing reinforced [25] ACI 318-11 Building Code Requirements For Structural Concrete and
concrete walls, Fire Safety Sci. 3 (1991) 771–780. Commentary Reported by ACI Committee 318, 2011.