Professional Documents
Culture Documents
Engineering Structures
journal homepage: www.elsevier.com/locate/engstruct
a r t i c l e i n f o a b s t r a c t
Article history: A total of nine full-scale in-situ tests were carried out to investigate the performances of autoclaved aer-
Received 5 December 2016 ated concrete (AAC) masonry walls subjected to vented gas explosions. The testing data including over-
Revised 5 March 2017 pressure time histories of vented gas explosions, displacement time histories, and damage characteristics
Accepted 16 March 2017
of AAC masonry walls in each test were recorded and analysed. It was found that the responses of
Available online 2 April 2017
masonry walls mainly depend on the peak value of overpressure and couple with the time history of
gas explosion loads. Typical one-way or two-way flexural mode dominates the failure of AAC walls under
Keywords:
vented gas explosions. A detailed micro model for masonry walls was developed in LS-DYNA, incorporat-
Autoclaved aerated concrete (AAC)
Masonry wall
ing material parameters that were obtained from material tests. The accuracy of numerical model in pre-
Vented gas explosion dicting the responses of masonry walls was verified with the testing data. Parametric studies were
Field test conducted to explore the influences of block strength, boundary condition and wall thickness on the per-
Numerical simulation formances of masonry walls. The results reveal both wall thickness and boundary condition have signif-
Equivalent SDOF method icant influences on the response of the masonry wall while block strength has limited effect on its
performance. The testing data were compared with the analytical predictions by using design code
UFC 3-340-2 and equivalent single-degree-of-freedom (SDOF) methods developed respectively by
Biggs and Morison. The results indicate that these predictions on one-way specimens agree well with
the testing data, while the performance of two-way specimens is overestimated by using these three
methods.
Ó 2017 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.engstruct.2017.03.033
0141-0296/Ó 2017 Elsevier Ltd. All rights reserved.
Z. Li et al. / Engineering Structures 141 (2017) 444–460 445
observed in many blast tests, this oversimplification of gas explo- sound insulation properties, but leads to the strength reduction
sion loads may not lead to reliable prediction of structural and heterogeneity of the material. In addition, AAC is of less den-
responses. sity than traditional concrete or clay. Although the seismic perfor-
Although the existing study of masonry walls under gas explo- mance and fire resistance capacity of AAC masonry structures have
sion was limited [12], the performance of masonry walls under been well investigated [28,29], the study on the blast resistance
blast loads generated by high explosives have been intensively capacity of AAC masonry walls is very limited. Muszynski and Pur-
studied owing to the increasing hazards from terrorist attacks, con- cell [30] conducted field tests on the air-entrained concrete (AEC)
ventional weapons and accidental explosions [13–17]. The masonry walls subjected to high explosive detonations. Their study
dynamic responses, failure modes, debris distributions and retro- focused on investigating the effectiveness of using composite
fitting techniques with different retrofitting materials were stud- materials for retrofitting the masonry wall therefore the structural
ied by conducting field blasting tests. Some of the above testing behaviour of was not emphasized in the study. Yankelevsky and
results were used to verify numerical models. Three modelling Avnon [31] carried out the tests to study the localized responses
methods, i.e. detailed micro model, simplified micro model and of AAC under contact explosion and the effectiveness of surface
macro model were developed [18]. The damage level and the safety treatment on the wall performances. Test results showed that the
scaled distance of masonry walls derived from numerical simula- surface treatment enhanced the tensile strength and ductility of
tions were recommended for engineering practice [15,19,20]. AAC blocks, and contributed to higher resistance to tensile wave
Arching effect of masonry walls was also considered in the theoret- spalling. No study of AAC masonry wall subjected to gas explosion
ical analysis [21–23]. In addition, the equivalent SDOF method was can be found in the open literature yet. Because the blast loads
validated and used to develop P-I diagrams for engineering appli- generated by gas explosions have very different characteristics
cation [24]. Owing to its simplicity, the design approaches were such as lower amplitude, longer rise time, longer duration and pos-
based largely on the equivalent SDOF method in engineering sibly multiple peaks. As compared to those by high explosives, the
design [25–27]. It should be noted that most previous studies were response mechanism and damage mode of ACC masonry walls sub-
limited to the reinforced or unreinforced concrete masonry unit jected to blast loads from gas and high-explosive explosions could
(CMU) and clay brick masonry walls. be very different. Therefore, it is necessary to study the ACC
Autoclaved aerated concrete (AAC), as an ‘ultra-lightweight’ masonry wall responses to gas explosion for reliable predictions
concrete material with micro cellular structures, has been used of the masonry wall damages and better design of protection mea-
as an alternative to conventional normal-weight or lightweight sures for such walls to resist accidental gas explosions.
concrete products [3,4]. The existence of entrained air bubbles In this study, an experimental program was designed to per-
enhances the physical properties, such as excellent thermal and form a series of full-scale field tests on AAC masonry walls sub-
Watet tank
D1
P1
P2
RC chamber D2
P3
D3
Igniting pill
Window
Testing
specimen
D4
Sliding door
P4
D5
Probe
CH4
High Voltage Power Source
jected to vented gas explosions. A detailed micro model was also instruments and gas introducing/mixing/igniting instruments as
developed to predict the responses of AAC masonry walls by using shown in Fig. 1.
LS-DYNA 971. The parameters of the material model for AAC blocks
and mortar were obtained from material testing data. The predic- 2.1.1. Specimen preparation
tions from the numerical simulations are compared with the test- Three wall specimens with dimensions of 2 m by 3 m were
ing data to calibrate the numerical model. The verified numerical made of AAC blocks with dimensions of
model was then used to study the effects of block strength, bound- 590 mm 120 mm 240 mm and 5 mm-thick mortar, as shown
ary condition and thickness of the walls on the response of in Fig. 2. Two types of boundary conditions were considered.
masonry walls subjected to vented gas explosions. In addition, One-way specimens were constructed with the right and left
the equivalent single-degree-of-freedom (SDOF) method was also boundaries separated from the RC frames by the plastic film, as
used to analyse the response of AAC masonry walls under gas shown in Fig. 3(a). The right and left boundaries of two-way spec-
explosion loads. The predictions of displacement histories based imens were bolted into RC frames by the steel bars at an equal
on the approaches specified respectively by Biggs, Morison and space of 25 cm as shown in Fig. 3(b).
UFC 3-340-2 are compared with testing data. The accuracies of
these three approaches in predicting the responses of AAC masonry 2.1.2. Data acquisition and measurement devices
walls subjected to gas explosions are discussed. The testing data including gas explosion loads, displacements
and failure modes of test walls were recorded by pressure sensors,
2. Experimental investigations displacement transducers and high-speed camera, respectively.
Four piezoresistive water-cooling pressure sensors with a measur-
2.1. Experimental program ing range of -20 to 150 kPa (CYG 1409, Kunshan Shuangqiao,
China) were used to record the overpressure. Pressure sensors des-
A specially designed testing system was prepared for the vented ignated as P1 to P3 were mounted on the front wall at the height of
gas explosion tests. The testing system consisted of reinforced con- 0.2 m, 0.8 m and 2.5 m along the midline of the wall, respectively.
crete (RC) chamber, masonry wall specimens, data-recording Pressure sensor P4 was mounted at 0.9 m off the center line with
2m
Block
590mm
3m
Mortar 120mm
240mm
(a) Front view of wall specimen (b) Autoclaved aerated concrete (AAC) block
Fig. 2. Details of wall specimen.
Steel bar
Plastic film
the height of 2.5 m. The installation of pressure sensors is shown in imen in front of the RC chamber, install the displacement transduc-
Fig. 4(a). The water tank, a key part of water-cooling system, was ers and seal the gap between the RC chamber and RC frame by
located on the top the RC chamber. Designated as D1 to D5, five using foam sealant to avoid gas leakage during the course of filling
displacement transducers with a measuring range of 0–300 mm the gas. Next, the ignition pill was mounted at the center of RC
(WYJL, Xian Xinmin, China) were fixed along the midline of wall chamber and the vent cover was installed on the vent window.
specimens at an equal space of 75 cm to acquire the displacement Then, the methane was piped from the cylinder gas tank to the
time histories, as shown in Fig. 4(b). A high speed camera (FAS- RC chamber until the target concentration was reached. Mean-
TCAM SA-Z, Photron Co., Ltd., Japan) was placed behind the test while, the explosion-proof fan was turned on to evenly mix the
wall about 15 meters from one side of the wall to record the failure methane with air. After that, the steel sliding door was closed
process of wall specimens. The signals from the pressure sensors and the water-cooling system for pressure sensors was turned
and displacement transducers were logged by the data acquisition on. The interval between finishing the mixing and ignition was
instrument (DongHua 5927, Donghua Testing Technology Co., Ltd., kept at least 30 s to achieve a low and consistent initial turbulent
China) at a sampling rate of 30 kHz. intensity. At last, the gas mixture was ignited by high voltage
power source. Both data acquisition system and high-speed cam-
2.1.3. Gas introducing/mixing/igniting instruments era were triggered to record data.
As a main component of natural gas, the methane was chosen as
the flammable gas in the tests and piped into the RC chamber from 2.2. Testing scheme
the cylinder gas tanks. An infrared gas analyzer (QGS-08C, Nanjing
Xinfen, China), with a measuring range of 0–15 vol% and the accu- Nine tests were conducted to investigate the performance of
racy of 0.1 vol% was employed to monitor the methane concentra- three AAC walls under gas explosion loads, as listed in Table 1.
tion. An explosion-proof fan (CBF-300, Zhejiang Dafeng Blowers, Three wall specimens were constructed with two boundary condi-
China) was installed in the chamber to ensure the homogeneity tions. Different gas concentrations and vent covers were used to
of gas mixture. Triggered by the high voltage power source, the achieve the desired gas explosion loads.
igniting pill with energy of 100 mJ fired the methane/air mixture
at the center of the RC chamber. The high voltage power source 2.3. Testing results and discussions
also triggered the data acquisition instrument and the video cam-
era at the same time of ignition. 2.3.1. Gas explosion loading
In the previous studies [7,32], the primary characteristics of gas
2.1.4. Test procedure explosion loads have been analysed and discussed. The overpres-
All specimens were tested under the vented gas explosion sure with high frequency oscillation were acquired from four pres-
ignited at the center of the RC chamber. Each wall was tested by sure sensors and compared. The representative pressure-time
the following procedures. The first step was to place the wall spec- profiles recorded by the pressure sensor P1 are presented in this
Pressure sensor
Pressure sensor
Displacement
transducer
Displacement
transducer
Table 1 paper, and the pressure time profiles are filtered to remove high
Testing scheme. frequency pressure oscillations.
Test No. Wall No. Gas Concentration Boundary Vent cover The overpressure-time histories acquired from nine tests are
(%) condition shown in Fig. 5 and summarized in Table 2. Weak vent cover (i.e.
1 W1 12.50% One-way 1.5 lm film 1.5 lm film and 6 lm film) and high methane concentration (i.e.
2 12.50% 6 lm film 12.5%) were applied in most of the tests to obtain low gas explo-
3 12.50% 4 mm glass panel sion loads as desired. As shown, the number of peaks and peak
4 W2 12.50% One-way 1.5 lm film value of gas explosion loads are different in different tests and
5 12.50% 6 lm film are sensitive to the test conditions (i.e. gas concentration level
6 9.50% 12 mm glass
panel
and vent cover). Three peaks, namely DP1 , DP 2 and DP 3 are promi-
nent in the load profiles of five tests 1/4/5/7/8, while the recorded
7 W3 12.50% Two-way 6 lm film
8 12.50% 4 mm glass panel
overpressure time histories in the other tests do not have three
9 9.50% 1.5 lm film prominent peaks. The effect of gas concentration on gas explosion
loads can be observed by comparing the data of Test 3 and Test 6.
Table 2
Vented gas explosion loads and mid-span displacements of wall specimens.
Because the vent covers in both the Test 3 (i.e. 4 mm glass panel) peak pressure of 13.25 kPa, indicating that the two-way wall W3
and Test 6 (i.e. 12 mm glass panel) survived the gas explosions, can withstand the load with the peak overpressure more than
there was only one prominent peak in the recorded overpressure 2.5 times of the one-way wall W1.
profile since failure of wall specimens results in the relief of the
high pressure gas products inside the RC chamber. When the con-
centration changes from 12.5% (Test 3) to 9.5% (Test 6), The over- 2.3.3. Damage characteristics of AAC walls
pressure rising rate increases from 26.2 kPa/s to 145 kPa/s, and The crack/damage patterns of two wall specimens (i.e. W1 and
the peak pressure of DP1 increases from 5.15 kPa to 13.61 kPa. This W3) after Test 2 and Test 8 are shown in Fig. 6. As shown in Fig. 6
is because 9.5% stoichiometric concentration of methane results in (a), the wall specimen W1 with free boundary condition at left and
the highest laminar flame speed, which is associated with the right side experiences typical damage mode of one-way bending
burning rate and energy release of gas explosions. The most devas- and crack damage at the mid-height area of the wall. The two-
tating load is generated in Test 9, which have the vent cover of way wall specimen W3 experiences an X pattern crack damage
1.5 lm film and the methane concentration of 9.5%, as shown in as shown in Fig. 6(b). The permanent displacement of 3.8 mm is
Fig. 5(e). The load profile has two peaks, DP 1 and DP 4 . Owing to recorded at the center of the specimen after Test 8. Figs. 7–9 show
the weak vent cover of 1.5 lm film that leads to the quick pressure the failure process of wall specimens W1-W3 under devastating
relief, the value of DP1 is only 1.23 kPa, which is much lower than gas explosions, respectively. For the typical one-way wall speci-
DP4 (i.e. 85.88 kPa). Obviously, the large DP 4 causes the collapse of mens W1 and W2, a through crack similar to the typical yield line
wall specimen in Test 9. of one-way panels occurs at the horizontal middle area at the early
stage of failure process. With the rise of pressure inside the RC
chamber, the crack develops and divides the wall specimen into
2.3.2. Displacements of wall specimens
two sections. Then the wall is blown away and falls on the ground
The mid-span displacement-time histories of wall specimens
due to the gravity. In contrast, the wall specimen W3 experiences
are shown in Fig. 5. It should be noted that the displacement was
the damage mode of X pattern crack, which divides the wall spec-
not measured in the last destructive test of wall specimens (W1,
imen into four major sections. The different damage patterns are
W2 and W3) to avoid damaging the displacement transducers.
due to different boundary conditions.
The curves show similar trend with the corresponding
The effect of gas explosion loads on the failure of masonry walls
overpressure-time history profiles. Non-linear responses of wall
is investigated by comparing the failure modes of W1 and W2. As
specimens are summarized in Table 2. The ratio of peak overpres-
shown in Fig. 7, the specimen W1 is divided into two sections at
sure to maximum displacement, i.e. P/D, is also shown in the table.
the mid-span area in Test 3, under the gas explosion load with peak
As can be noticed, the plastic deformation of masonry wall
increases more significantly as compared to the peak overpressure
when the vent cover is stronger or the relief pressure DP1 is larger.
For example, for W3 in Test 7 the P/D ratio is 1.31, while in Test 8 it
decreases to 0.41 with the peak pressure increasing from 2.91 kPa
to 13.25 kPa. Similar phenomenon is found in the test data of W1
and W2.
The effect of boundary condition on the maximum displace-
ment of wall specimens is studied by comparing the test data of
Test 5 and Test 7. As given in Table 2, under the same testing con-
ditions (i.e. concentration of 12.5% and vent cover of 6 lm film),
the maximum displacement of the one-way specimen W2 is
7.8 mm in Test 5 and the two-way specimen W3 experiences the
maximum displacement of 2.22 mm in Test 7. This indicates that
the change of boundary conditions from one-way to two-way
results in about a 70% decrease in deflection. Similar conclusion
can be drawn by comparing the results from Test 3 and Test 8 with
the same testing conditions (i.e. concentration of 12.5% and vent
cover of 4 mm glass panel). In Test 3, the one-way specimen W1
collapses under the load with the peak pressure of 5.15 kPa, which (a) wall specimen (b) wall specimen
is unable to break the vent cover of 4 mm glass panel before the W1 in Test 2 W3 in Test 8
failure of masonry wall. However, in Test 8, the vent cover of
4 mm glass panel is broken before the two-way wall W3 at the Fig. 6. Damage of wall specimens.
450 Z. Li et al. / Engineering Structures 141 (2017) 444–460
value of 5.15 kPa. The wall specimen W2 with the same boundary masonry walls. In this study, the anisotropic material model
⁄
condition and thickness as the specimen W1 experiences a similar Mat_96 (MAT BRITTLE DAMAGE) in LS-DYNA was used to model
damage at the early stage but more severe damage at the later the AAC blocks and mortar [33,34]. In this model, compressive fail-
stage of Test 6 owing to higher pressure, as shown in Fig. 8. This ure is governed by a simplified J2 flow correction, which can be dis-
phenomenon can also be observed in Fig. 9. The wall W3 bursts abled if unwanted. Progressive degradations of tensile and shear
into pieces in Test 9 under the load with higher peak overpressure strengths across smeared cracks are initiated under tensile loads.
and overpressure rising rate, which is different from damage The degradation is implemented by reducing the material’s modu-
modes observed on other specimens. It is because the AAC blocks lus normal to the smeared crack plane according to a maximum
of W3 fails by the high intensity stress wave due to gas explosion dissipation law that incorporates exponential softening.
loads with high peak pressure and rising rate in Test 9.
/1 ¼ S1 : r f n þ kn f n ð1 exp½HaÞ 6 0 ð1Þ
along with n; r is the stress tensor; f n is the critical tensile traction limit, fracture toughness and shear retention factor are acquired
that can be held across the smeared crack field; f s is the critical from references [29,34,37,38]. The parameters used in numerical
shear traction that can be held across the smeared crack field; kn simulations are listed in Table 3. The element erosion was used
and ks are coupling constants; H is the softening modulus; a is an to simulate the failure of material. The maximum principal strain
internal variable. at failure of block and mortar is 0.01 and 0.005, respectively, which
In order to determine the parameters of material models for the is calibrated with testing data by trial calculations.
AAC block and mortar, a series of laboratory tests, as shown in
Fig. 10, were conducted according to the testing standards i.e. 3.2. Geometric model
GB/T 11971-1997 and JGJ/T70-2009 [35,36]. The mean uniaxial
compressive strength of AAC blocks and mortar was 3.07 MPa For the detailed micro model, the AAC blocks and mortar are
and 7.53 MPa, respectively. The density and Young’s modulus of modelled separately and the nodes of the mortar and AAC blocks
blocks and mortar are acquired from the material tests directly, along their interfaces are merged. The dimensions of wall models
as summarized in Table 3. The Poisson’s rate, tensile limit, shear with thickness of 120 mm, 3.0 m height and 2.0 m width, as in
Table 3
Material parameters for AAC block and mortar [29,34,37,38].
Density Young’s Modulus Poisson’s Tensile limit Shear limit Compressive yield Fracture toughness Shear retention
(kg/m3) (MPa) Ratio (MPa) (MPa) stress (MPa) (N/m) factor
AAC block 625 530 0.2 0.7 1.0 3 80 0.03
Mortar 2100 2211 0.25 0.75 1.1 7.5 140 0.03
Fig. 13. Comparison of mid-span displacement between numerical predictions and testing data.
Z. Li et al. / Engineering Structures 141 (2017) 444–460 453
boundary condition was used to simulate the two-way wall, as ment in Test 7 is 1.77 mm, which is about 25% less than the test
shown in Fig. 12. In this case the RC frame was included in the data of 2.21 mm. In Test 8, the prediction of 31.2 mm in the numer-
model and the interface between the RC frame and the masonry ical model is only 4% less than the test data of 32.6 mm. The failure
wall was simulated as a tie-break contact. In LS-DYNA 971, the modes of both one-way and two-way masonry walls simulated
tie-break contact is governed by the stress-based failure criterion from the numerical model are compared with the testing results,
as follows [33]. as shown in Fig. 14 and Fig. 15, respectively. It is found that, the
2 2 predicted failure mode of one-way specimen agrees well with
jrn j jss j the test observation when the specimen is subjected to gas explo-
þ P1 ð4Þ
fn fs sion loads. As for the two-way masonry wall that failed at higher
peak pressure and faster rising rate as discussed above, the numer-
where rn is the normal stress on the contact surface; ss is the shear
ical simulation is still able to capture the damage characteristics of
stress on the contact surface; f n is the normal failure stress and f s is
wall specimen at the early stage, but cannot well predict the frac-
the shear failure stress. This tie-break contact algorithm has been
ture damage of AAC blocks into pieces, which is deemed as short-
widely used to simulate the bond-slip behaviour of structure com-
coming of this numerical model. It is attributed to the following
ponents. The static and dynamic coefficients of friction for the top
reasons: 1) The material model used in this study is not precise
and bottom edges are 0.4 and the normal failure stress f n and shear
enough to predict the damage of AAC block subjected to high
failure stress f s are zero. As to the left and right edges, the static and intensity stress wave; 2) There is a difference between the simu-
dynamic coefficients of friction are 0.4. The normal failure stress f n lated boundary conditions and the real boundary conditions.
and shear failure stress f s are 0.5 MPa. The above parameters are Although there are some shortcomings in predicting the failure
acquired from trial calculations and validated by the test data. mode of masonry walls subjected to the loads at the higher peak
overpressure value and rising rate, the model successfully predicts
3.4. Validation of numerical model the masonry wall failure. Therefore, it can be concluded that the
numerical model developed in this study can give reasonably good
Fig. 13(a and b) present the comparisons of numerical predic- prediction of the structural response of AAC masonry walls against
tions and test data of one-way wall specimens W1 and W2. It vented gas explosions.
shows that the predicted period of vibration and the peak value
of mid-span displacement of one-way specimens agree well with
the test data in both elastic and plastic phases. As for the two- 3.5. Parametric study
way specimen W3, shown in Fig. 13(c and d), it is found that the
predicted peak value and shape of displacement histories are also In this section, parametric studies are carried out to investigate
close to the testing data. For instance, the predicted peak displace- the effects of AAC block strength, wall thickness and boundary con-
dition on the performance of AAC masonry walls under vented gas represents three DOFs of UX, UY and UZ are constrained. Table 5
explosion. The structural responses including maximum displace- summarizes the conditions of seven models analysed to examine
ment are compared herein. the effects of AAC block strength, wall thickness and boundary con-
dition on the performances of masonry walls.
3.5.1. Brief description of numerical model for parametric studies 3.5.2. Effect of block strength
A single-peak overpressure wave with a triangular shape was The effect of block strength on the response of masonry walls is
adopted for the parametric study. This simplified method of gas examined under three gas explosion loads with three peak pres-
explosion loads was also used by Louca et al. and Parisi et al. sures of 2 kPa, 4 kPa and 10 kPa, respectively. The maximum dis-
[10,11] in their studies on blast walls of offshore platforms and tuff placement time histories of masonry walls with three block
stone masonry walls, respectively. The rising rate of the pressure strength of A3.5, A5.0 and A7.5 are shown in Fig. 18. It is found that
wave was set as 400 kPa/s and the decreasing rate was set to be when the peak pressures are 2 kPa and 4 kPa, the profiles of dis-
1000 kPa/s based on the testing data. placement time histories match well with each other. When the
The masonry wall used for parametric study was 3.0 m in height peak pressure increases to 10 kPa, the deflections of masonry walls
and 2.0 m in width. The material parameters for three types of AAC with strength grade of A3.5 are the largest followed by A5.0 and
blocks, i.e. A3.5, A5.0 and A7.5 are summarized in Table 4. Three A7.5. However, the differences between the displacements are
wall thicknesses, i.e. 120 mm, 180 mm and 240 mm were consid- quite limited and the whole trend of their profiles is the same. It
ered. It should be noted that the variations of wall thickness (i.e. therefore can be concluded that the block strength has insignifi-
120 mm, 180 mm and 240 mm) in engineering practice are cant influence on the maximum displacement of the wall, espe-
achieved by changing the thickness of AAC blocks as shown in cially when the gas explosion load is of low peak pressure. This
Fig. 16. The other dimension of AAC blocks was 590 mm 240 mm phenomenon could be explained as follows: 1) when the peak
and the thickness of the mortar was 5 mm. Three types of bound- pressure of loads is low, the response of masonry walls is within
ary conditions of masonry walls (i.e. Type A/B/C) were also consid- elastic range. The structural response of masonry wall is mainly
ered in this study, as shown in Fig. 17. ‘‘Fixed” discussed here determined by the structural stiffness of masonry wall rather than
the AAC block strength; 2) the arching action and the effect of
Table 4
Material parameters of AAC blocks for different strength classes [38]. block strength on the arching action are insignificant owing to
the high slenderness ratio (wall height over thickness) of masonry
Strength Compressive yield Tensile limit Shear limit Density
walls.
class stress (MPa) (MPa) (MPa) (kg/m3)
A3.5 3.5 0.38 0.53 600
3.5.3. Effect of wall thickness
A5.0 5.0 0.46 0.75
A7.5 7.5 0.58 1.1 Three masonry walls with the thicknesses of 120 mm, 180 mm
and 240 mm are compared in this section. As shown in Fig. 19, the
Z. Li et al. / Engineering Structures 141 (2017) 444–460 455
3.0 m
fixed
fixed
fixed
fixed
fixed
fixed
Table 5 placements are 22.8 mm and 9.6 mm, respectively. When the peak
Case conditions in parametric studies pressure increases to 15 kPa, the walls with the thickness of
Case Strength Block density Wall thickness Boundary 120 mm and 180 mm collapse, but the wall with the thickness of
No. class (kg/m3) (mm) condition 240 mm still survives and the maximum displacement increases
1 A3.5 600 120 Type A to 33.2 mm. It is found that with the increase of wall thickness,
2 A5.0 600 120 Type A the maximum displacement of walls reduces and the resistance
3 A7.5 600 120 Type A capacity of walls increases significantly. The results could be
4 A3.5 600 180 Type A
attributed to three reasons: 1) the structural stiffness of wall is
5 A3.5 600 240 Type A
6 A3.5 600 120 Type B enhanced with the increase of wall thickness, 2) the cracking pres-
7 A3.5 600 120 Type C sure of masonry wall increases with the increasing wall thickness,
and 3) with the increase of wall thickness, the arching action of
masonry walls is improved, which has also been reported in [22].
(c) 10 kPa
Fig. 18. Displacement time histories of masonry walls with different block strengths.
(c) 15 kPa
Fig. 19. Displacement time histories of masonry walls with different wall thicknesses.
Z. Li et al. / Engineering Structures 141 (2017) 444–460 457
(c) 10 kPa
Fig. 20. Displacement time histories of masonry walls with different boundary conditions.
survives and experiences a maximum displacement of 28.1 mm. It two-way walls, the transformation factors are obtained by the fol-
can be concluded that boundary conditions have a significant influ- lowing equations,
ence on the response of masonry walls. Z Z
1
KL ¼ KR ¼ /ðx; yÞdA ð5Þ
A
4. Simplified analytical method
Z Z
1
In this section, testing results of masonry walls are compared KM ¼ /2 ðx; yÞdA ð6Þ
A
with the analytical predictions by adopting design code of UFC 3-
340-2 and the equivalent SDOF approaches by Biggs and Morison Because of the uniformly distributed gas explosion loads, the
[25–27]. shape functions /ðx; yÞ are mainly determined by the boundary
conditions and work stages (elastic or plastic) of masonry walls.
4.1. Equivalent SDOF model The transformation factors suggested by Biggs, UFC 3-340-2 and
Morison are summarized in Table 6 [25–27]. Biggs proposed the
The equivalent single-degree-of-freedom (SDOF) approach has factors and tables, which were based on the classical plate theory
been widely applied in the design of structures against accidental for deflections in the elastic range and the yield line theory in
explosions [39]. The masonry wall under gas explosion loading the plastic range. The suggested factors for two-way slabs by Mor-
was simplified into an equivalent SDOF system as shown in ison were based on the results of transient finite element analysis.
Fig. 21. According to the Newton’s laws of motion and the principle The charts, tables and formulae defining transformation factors,
of energy conservation, the real system is transformed into an resistance-deflection function presented in UFC 3-340-2 were
equivalent SDOF system by using equivalent load factor K L , equiv- derived from the equilibrium method of yield-line analysis and
alent mass factor K M and resistance factor K R . The distributions of elastic theory.
mass and load along the walls are uniform in this study. For the
4.2. Resistance-deflection function
Table 6
Parameters in the SDOF analysis [25–27].
7 7 7
1:74 10 N=m 1:75 10 N=m 1:77 10 N=m 2:83 106 N=m
Re 26:53Mt 7:52M ta þ 19:38M tb Mt a 12M n
0:07570b b
47,754 N 48,420 N 35,664 N 14,400 N
Xe Re Re Re Re
Ke Ke Ke Ke
2.74 mm 2.76 mm 1.99 mm 5.10 mm
K ep 199:8EI 199:6EI EIa 384EI
0:00247604b ð1t2 Þ
a2 a2 3 3
5b
Note: a is the short span dimension of walls; b is the long span dimension of walls; E is Young’s modulus; I is the second moment of cross section per unit width; M t is the
moment resistance per unit length; Mfa , Msa , Msb , Mfb are moment resistance per unit length at fixed edges and spans; M n , M p are the moment resistance at fixed edges and
spans K LMe , K LMep , K 0LMep , K LMp are the load-mass transformation coefficients of different work stages. K e , K ep , K 0ep are the stiffness of different work stages; X e , X ep , X p are the
deflection limitation of different work stages; Re , Rep , Ru are the resistance limitation of different work stages.
tensile membrane resistance. However, the tensile stage of models developed in this study can be used to predict the deflec-
resistance-deflection function can be neglected for the unrein- tions of masonry walls for the cases with limited deflection. The
forced masonry walls owing to its low ductility of structures. The arching action, which affects the resistance-deflection function
resistance-deflection functions suggested by Biggs and Morison after cracking, was not considered.
were simplified as two step elastic-plastic system as shown in
Fig. 23(a). However, in UFC 3-340-2, one-way masonry wall and 4.3. Comparison between testing data and SDOF predictions
two-way masonry wall were proposed to have the resistance-
deflection functions of two-step elastic-plastic and three-step In order to evaluate the accuracy of the SDOF models provided
elastic-plastic, which are shown in Fig. 23(a and b), respectively. by Biggs, UFC 3-340-2 and Morison, the SDOF predictions of the
The appearance of each plastic hinge represents the change of stiff- maximum displacement for one-way specimen W1 and two-way
ness of the equivalent system. According to the recommendations specimen W3 were calculated and compared with testing data.
by Biggs, UFC 3-340-2 and Morison [25–27], the resistance- The input parameters of masonry materials for the SDOF models
deflection functions were calculated and the relevant parameters were the same as those described above which were acquired from
are listed in Table 6. It should be noted that the equivalent SDOF material tests, and the recorded load in each test was used to
derive the equivalent load in SDOF analysis. Average acceleration
method was used to integrate the SDOF equations of motion [27].
Fig. 24 shows the comparison of displacement time histories
between analytical predictions and test data. It can be found that
the predictions from the SDOF methods by Biggs and UFC 3-340-
Flexural resistance Tensile membrane resistance 2 agree well with the test data of Test 1, in which the response
of wall specimen W1 is elastic under a low gas explosion load, as
shown in Fig. 24(a). It is inferred that the stiffness used in the SDOF
A B model well reflects the stiffness of tested wall. As shown in Fig. 24
Resistance (R)
D
(b), the predicted first peak of displacement time history is in good
agreement with the testing data while there is notable difference
of permanent displacement between the SDOF prediction and the
results of Test 2. This may be caused by the following reasons: 1)
C
the predicted ultimate resistance is smaller than the actual value;
2) arching action of structures is not incorporated into the
resistance-deflection function in the SDOF model. From Fig. 24(c
and d), it is found that the predictions derived from three SDOF
Deflection (X)
methods by Biggs, Morison and UFC 3-340-2 are close to each other
because the input parameters have only slight differences. In Test
Fig. 22. Resistance-deflection function of reinforced masonry walls. 7, the predicted maximum displacement is only 50% of the testing
Z. Li et al. / Engineering Structures 141 (2017) 444–460 459
R
R
Ru Ru
Rep '
Kep
Resistance (R)
Resistance (R)
1
Kep
Kep
Re 1 Re 1
Ke Ke
1 1
Xe Xp Xm X Xe Xep Xp Xm X
Deflection (X) Deflection (X)
(a) Two step elastic-plastic (b) Three step elastic-plastic
Fig. 23. Resistance-deflection function of unreinforced masonry walls [26].
(a) Test 1 (one-way specimen W1) (b) Test 2 (one-way specimen W1)
(c) Test 7 (two-way specimen W3) (d) Test 8 (two-way specimen W3)
Fig. 24. Comparison of analytical predictions and testing data.
data and the predicted displacements are always smaller than the essarily yield reliable predictions of the response of two-way spec-
testing data throughout the entire response duration. This phe- imens accurately.
nomenon is also observed in the comparison of Test 8. In Test 8,
the predicted maximum displacement is only 25% of the testing 5. Conclusion
data. The poor performance of these three SDOF models in predict-
ing the masonry wall responses could be contributed to the follow- The responses of unreinforced AAC masonry walls under vented
ing factors: 1) The difference of moment resistance in different gas explosions were examined and analysed by using full-scale
directions was neglected and the homogenization simplification field tests, numerical simulations and simplified analytical method.
for the masonry materials was used in the SDOF models; 2) In The conclusions are summarized as follows:
the SDOF models, the parameters, which were derived from ideal
fixed boundary conditions, overestimate the stiffness of structures. (1) The dynamic responses of AAC masonry walls under vented
These results demonstrate the methods and parameters provided gas explosions depend mainly on the peak overpressure.
by Biggs, Morison and UFC 3-340-2, if directly applied, do not nec- Typical one-way or two-way flexural modes dominate the
460 Z. Li et al. / Engineering Structures 141 (2017) 444–460
failure of masonry walls under vented gas explosions. With [13] Dennis ST, Baylot JT, Woodson SC. Response of 1/4-scale concrete masonry
unit (CMU) walls to blast. J Eng Mech 2002;128:134–42.
the increase of peak pressure and rising rate of gas explosion
[14] Davidson JS, Fisher JW, Hammons MI, Porter JR, Dinan RJ. Failure mechanisms
loads, the failure mode of AAC masonry walls changes from of polymer-reinforced concrete masonry walls subjected to blast. J Struct Eng
flexural failure of the wall to combined flexural failure and 2005;131:1194–205.
fragmentation of AAC blocks. [15] Chen L, Fang Q, Jiang C, Fan J, Hao H. Response and damage level of confined
masonry walls to blast. Disaster Adv 2013;6:380–94.
(2) The numerical models are developed and validated by com- [16] Chen L, Fang Q, Fan J, Zhang Y, Hao H, Liu J. Responses of masonry infill walls
paring with the testing data of both one-way walls and two- retrofitted with CFRP, steel wire mesh and laminated bars to blast loadings.
way walls. The results from parametric studies indicate both Adv Struct Eng 2014;17:817–36.
[17] Hao H. Numerical modelling of masonry wall response to blast loads. Aust J
masonry wall thickness and boundary conditions greatly Struct Eng 2009;10:37–52.
affect the wall responses. The influence of block strength is [18] Lourenco PB. Computational strategies for masonry structures. Delft, The
however insignificant due to the weak arching action of thin Netherlands: Delft University Press; 1996.
[19] Wu C, Hao H. Safe scaled distance for masonry infilled RC frame structures
walls in this study. subjected to airblast loads. J Perform Constr Facil 2007;21:422–31.
(3) The equivalent SDOF models for masonry walls are estab- [20] Wang M, Hao H, Ding Y, Li Z-X. Prediction of fragment size and ejection
lished based on the approaches by Biggs, UFC 3-340-2 and distance of masonry wall under blast load using homogenized masonry
material properties. Int J Impact Eng 2009;36:808–20.
Morison. It is found that the predictions of Biggs and UFC [21] Zapata BJ, Weggel DC. Collapse study of an unreinforced masonry bearing wall
on the response of one-way specimens agree well with the building subjected to internal blast loading. J Perform Constr Facil
testing data. All the three SDOF models by Biggs, Morison 2008;22:92–100.
[22] Abou-Zeid BM, El-Dakhakhni WW, Razaqpur AG, Foo S. Time-response
and UFC on two-way walls underestimate the wall
analysis of arching unreinforced concrete block walls subjected to blast
responses. They should be used with caution in the analysis loads. J Struct Eng 2013;140:04013099.
of masonry wall responses to gas explosion loads. [23] Hrynyk TD, Myers JJ. Out-of-plane behavior of URM arching walls with modern
blast retrofits: experimental results and analytical model. J Struct Eng
2008;134:1589–97.
[24] Smith NL, Tait MJ, El-Dakhakhni WW, Mekky WF. Response analysis of
Acknowledgement reinforced concrete block infill panels under blast. J Perform Constr Facil
2016;04016059.
[25] Biggs JM. Introduction to structural dynamics. McGraw-Hill College; 1964.
The authors acknowledge the financial support from the [26] Morison CM. Dynamic response of walls and slabs by single-degree-of-
National Basic Research Program of China (973 Program, Grant freedom analysis—a critical review and revision. Int J Impact Eng
No. 2015CB058000), the National Natural Science Foundation of 2006;32:1214–47.
[27] Department-of-Defense, Structures to resist the effects of accidental
China (Grant Nos. 51238007, 51622812, 51427807, 51378016), explosions, United Facilities Criteria (UFC) 3–340-02, in, 2008.
and the Australian Research Council (Grant No. LP150100259). [28] Tanner JE, Varela JL, Klingner RE, Brightman MJ, Cancino U. Seismic testing of
The first author would also like to acknowledge the financial sup- autoclaved aerated concrete shearwalls: a comprehensive review. ACI Struct J
2005;102:374.
port from the China Scholarship Council. [29] Ferretti D, Michelini E, Rosati G. Mechanical characterization of autoclaved
aerated concrete masonry subjected to in-plane loading: experimental
References investigation and FE modeling. Constr Build Mater 2015;98:353–65.
[30] Muszynski LC, Purcell MR. Use of composite reinforcement to strengthen
concrete and air-entrained concrete masonry walls against air blast. J Compos
[1] Bjerketvedt D, Bakke JR, Van Wingerden K. Gas explosion handbook. J Hazard
Constr 2003;7:98–108.
Mater 1997;52:1–150.
[31] Yankelevsky DZ, Avnon I. Autoclaved aerated concrete behavior under
[2] Lees F. Lees’ Loss Prevention in the Process Industries: Hazard Identification,
explosive action. Constr Build Mater 1998;12:359–64.
Assessment and Control. Oxford, UK: Elsevier Butterworth-Heinemann; 2012.
[32] Bao Q, Fang Q, Yang S, Zhang Y, Xiang H, Chen L, et al. Experimental
[3] Wittmann FH, Balkema A. Advances in autoclaved aerated concrete. Citeseer;
investigation on the deflagration load under unconfined methane-air
1992.
explosions. Fuel 2016;185:565–76.
[4] Narayanan N, Ramamurthy K. Structure and properties of aerated concrete: a
[33] Halquist J. LS-DYNA Keyword User’s Manual Version 971. Livermore,
review. Cement Concr Compos 2000;22:321–9.
CA: Livermore Software Technology Corporation; 2007.
[5] Cooper M, Fairweather M, Tite J. On the mechanisms of pressure generation in
[34] Govindjee S, Kay GJ, Simo JC. Anisotropic modelling and numerical simulation
vented explosions. Combust Flame 1986;65:1–14.
of brittle damage in concrete. Int J Numer Meth Eng 1995;38:3611–33.
[6] Mercx W, Van Wingerden C, Pasman H. Venting of gaseous explosions. Process
[35] General-Administration-of-Quality-Supervision-Inspection-and-Quarantine-
Saf Prog 1993;12:40–6.
of-PR-China. Test Methods for Mechanical Property of Aerated Concrete (GB/T
[7] Bao Q, Fang Q, Zhang Y, Chen L, Yang S, Li Z. Effects of gas concentration and
11971-1997). Beijing, China: Standards Press of China; 1998.
venting pressure on overpressure transients during vented explosion of
[36] Ministry-of-Construction-of-the-PR-China. Standard for Test Method of
methane–air mixtures. Fuel 2016;175:40–8.
Performance on Building Mortar (JGJ/T70-2009). Beijing, China: China
[8] NFPA-68. Standard on Explosion Protection by Deflagration Venting. Quincy,
Architecture and Building Press; 2009.
MA: National Fire Protection Association; 2007.
[37] Serrano-Perez JC, Vaidya UK, Uddin N. Low velocity impact response of
[9] EN-14994. Gas Explosion Venting Protective Systems. Brussels: European
autoclaved aerated concrete/CFRP sandwich plates. Compos Struct
Committee for Standardization; 2007.
2007;80:621–30.
[10] Louca L, Pan Y, Harding J. Response of stiffened and unstiffened plates
[38] Argudo F, Fern J, Argudo O. Evaluation and Synthesis of Experimental Data for
subjected to blast loading. Eng Struct 1998;20:1079–86.
Autoclaved Aerated Concrete [M.S. thesis]. The University of Texas at Austin;
[11] Parisi F, Balestrieri C, Asprone D. Blast resistance of tuff stone masonry walls.
2003.
Eng Struct 2016;113:233–44.
[39] Hao H, Hao Y, Li J, Chen W. Review of the current practices in blast-resistant
[12] Li Z, Chen L, Fang Q, Hao H, Zhang Y, Xiang H, et al. Experimental and
analysis and design of concrete structures. Adv Struct Eng 2016;19:1193–223.
numerical study of unreinforced clay brick masonry walls subjected to vented
gas explosions. Int J Impact Eng 2017.