Professional Documents
Culture Documents
ReliabilityanalysisofdirectshearandflexuralfailuremodesofRCslabsunderexplosiveloading PDF
ReliabilityanalysisofdirectshearandflexuralfailuremodesofRCslabsunderexplosiveloading PDF
www.elsevier.com/locate/engstruct
Received 12 February 2001; received in revised form 3 September 2001; accepted 3 September 2001
Abstract
Two loosely coupled SDOF systems are used to model the flexural and direct shear responses of one-way reinforced concrete
slabs subjected to explosive loading. Incorporating the effects of random variations of the structural and blast loading properties,
as well as the strain rate effect caused by rapid load application, failure probabilities of the two failure modes are analyzed. The
model is capable of predicting the failure probability of the slab with random material and geometrical parameters and subjected
to random blast loading. Considering the random variations of structural properties and blast loading, the failure probabilities of
one-way RC slab designed according to BS 8110 (Stuctural use of concrete, parts 1 and 2 (1985)) are calculated. The effect of
span length of the slab on its failure probability to blast loading is also investigated. Based on numerical results, a semi analytical
boundary that separates the slab’s flexural and shear failure modes is derived as a function of peak reflected pressure and duration
of blasting wave. 2002 Elsevier Science Ltd. All rights reserved.
0141-0296/02/$ - see front matter 2002 Elsevier Science Ltd. All rights reserved.
PII: S 0 1 4 1 - 0 2 9 6 ( 0 1 ) 0 0 0 8 7 - 6
190 H.Y. Low, H. Hao / Engineering Structures 24 (2002) 189–198
also used to model the direct shear failure by Ross [8], 2.1. Concrete strength
Karthaus and Leussink [2] and by Krauthammer et al.
[9]. However, the simplest approximation is by a SDOF Random variations of concrete properties have been
simplification. As commented in conwep [10], problems studied by many researchers. It was found that the con-
that involve non-oscillatory loads such as blast load, crete cube strength is normally distributed with coef-
when only the peak response is required, many structural ficient of variation (COV) of 0.07 if construction quality
systems may be sufficiently analyzed using only the first is well controlled [18]. Analyzing data obtained from a
mode. Works in this area include the references [4,5,11– number of published sources, the same reference pro-
15]. The behavioral prediction by the more advanced posed the following equation for the COV of concrete’s
Timoshenko beam theory has been shown by Krau- in-situ strength
thammer et al. [16] to be exactly the same as those COV(in⫺situ strength)⫽[COV(cylinder strength)2 (1)
derived by the SDOF approach. In all these studies, the
two failure modes of a slab to blast load, namely the ⫹0.0084] 1/2
冉冊
specific values of the cross-sectional shape and dimen-
fyd ė
sions, the position of reinforcing bars, ties and stirrups, ⫽l log10 ⫹1 (7)
the horizontality and verticality of the concrete lines, and fys ės
the alignment of columns and beams. Most researchers in which the parameter l is 0.03 and ės for steel is
recommended the use of normal distribution to model approximately 10⫺2/s.
the statistical variation of dimensions of structural mem- Coupled with the strain rate effect, the statistical vari-
bers [25,26]. Mirza and MacGregor [26] also analyzed ations of the basic random variables considered are tabu-
the variations of dimensions of cast in situ slabs. Based lated in Table 1. From these basic random variables of
on the above information on variability in structures’ material strengths, the flexural and direct shear strengths
dimensions, a COV of 0.03 is adopted for all dimensions can be obtained from the following models.
(slab height, h; effective depth, d; length, L; breadth, b),
and the designed dimensions are taken as the mean
values and their variations are assumed following nor-
4. Structural resistances
mal distributions.
4.1. Flexural resistance
3. Strain rate effect on strengths The flexural resistance of the structure analyzed is
obtained first by computing the moment–curvature
As mentioned in Bischoff and Perry [27], properties relation of the section. Then, by considering the support
of the materials used in reinforced concrete structures conditions, the deflection of the structure under uni-
are almost all strain rate dependent. It is also pointed out formly distributed loading is calculated from the onset
in the same reference that the expected magnitude of of loading to failure. The incremental procedure for
strain rate for blast loading ranges from 100 to 1000/s. obtaining the static load-deflection relationship of the
However, owing to the difficulty in carrying out concrete structural element is shown in Fig. 1.
material tests at high loading rate, knowledge on con- In the analysis, it was assumed that strains had a linear
crete material property enhancement is limited to strain distribution over the beam cross section and that tensile
rates of 100/s. Moreover, during the blast loading pro- behavior of concrete located below the neutral axis is
cess, a slab will experience varying strain rates. For these neglected. An iterative process is utilized to satisfy the
reasons, in the present study, a constant strain rate of force equilibrium condition of the section. The entire
100/s is used in the analysis. This is acceptable as shown procedure is repeated for incremental values of strain in
by Krauthammer and his co-authors [28] that a constant tensile reinforcement until failure of the cross section is
strain rate with a reasonable order of magnitude is suf- reached. Here, since ultimate failure is of concern, it
ficient to yield good results. implies a severe deformation of the steel bars (when
reinforcement strain reached esu where esu=10 esy) or
3.1. Concrete crushing of the concrete (when concrete strain reaches
ecu). The section analyzed is designed according to BS
The strain rate enhancement formulae for concrete’s 8110 [1] without considering the severe blast load. The
uniaxial compressive strength (s), Young’s modulus (E) elastic limit of the moment–curvature relation thus corre-
and critical axial strain (e0) from the CEB [29] rec- sponds to the point of the first yielding of the reinforce-
ommendation shown below are adopted in the present ment.
study. The stress–strain relationship employed for the con-
192 H.Y. Low, H. Hao / Engineering Structures 24 (2002) 189–198
Table 1
Basic structural variables adopted (dynamically enhanced)
dynamic analysis (Fig. 2). The simplification is based on 6. Equivalent SDOF systems
the justification that as long as the area under the stress–
displacement curve remains constant, the energy Two coupled SDOF systems with bilinear resistance
absorbed by the system would be the same, and thus the functions are used to represent the shear and flexural
displacement calculated would be the same as well [32]. response mode of the slab. The first system is used for
The yielding and the maximum allowable shear slip are modelling the flexural response at the point of maximum
taken as 0.1 mm and 0.6 mm, respectively as shown in displacement along the span of the structure (ie. at mid-
Fig. 2. An enhancement factor of 2 is multiplied to the span as the loading and structure are symmetrical), and
direct shear strength, as done in Krauthammer et al. [5] the second is for monitoring the direct shear response at
to account for the rapid loading rate. This factor is the support.
indeed obtained from tests by Chung [33]. In a series of The SDOF system for modelling the flexural response
push-off tests for investigating the shear strength is based on Biggs [32] where the deflected shape of the
enhancement under dynamic load, Chung [33] reported structure is assumed to be the same as that resulting from
dynamic increase factors for shear strength from 1.80 to the static application of the dynamic load. The equival-
2.02, under a stressing rate of 10,000 to 12,800 N/mm2/s. ent mass and stiffness parameters are derived based on
the mass density, Young’s modulus, moment inertia,
span length and boundary condition of the slab, as well
as the deflection shape of the slab under distributed static
5. Blast loading load. More details can be found in Biggs [32]. The accu-
racy of using a SDOF system to model the slab response
The blast load pressure time history on a structure is to blast loads was proven by Krauthammer and his co-
usually simplified to an exponential or triangular shape authors [5]. This simplification for flexural response has
as been widely accepted and recommended, for instance,
冉 冊
by the US Air Force Manual AFM 88–22 [35] and US
t Department of Army [36].
P(t)⫽Prmaxe−at or Prmax 1⫺ (8)
to The second SDOF system is used to model the direct
shear response of the slab. Since the direct shear mode
where Prmax is the peak reflected pressure of the blast is expected to occur within a very short duration after
wave, a is the decaying rate for exponential represen- the initiation of the explosive loading, the structure
tation and to the duration for triangular loading simplifi- would not have any significant deformation at that time,
cation. The peak reflected pressure Prmax can be easily and because the failure plane occurs very near to the
estimated by Mills [34] in the following equation. support, the phenomenon is very much like a sudden
2Pmax(710+4Pmax) collapse of the entire beam. This implies that the shape
Prmax⫽ KPa (9) function of the structure can be taken as unity with negli-
710+Pmax gible deflection. The transformation factors for the shear
in which Pmax is the peak pressure of the blast wave in mass, direct shear stiffness as well as the loading are
free air. taken as unity too [11].
Many empirical formulae and charts are available to
estimate Pmax and duration to. From a statistical analysis
of eight available publications carried out by the author 7. Performance function formulation
[17], it was found that the mean value of the peak
reflected pressure at various scaled distances Z=R/w1/3 The exponentially decaying function in Eq. (8) is
(R and w are the stand-off distance in metres and equiv- selected to represent the explosive pressure’s time his-
alent TNT charge weight in kg) can be estimated by tory. Upon transforming the structural slab into its equiv-
alent flexural SDOF system, the post-yielding response
154.67 617.19 3069.3
Prmax⫽ ⫹ 2 ⫹ 3 ⫺1.2024 kPa (10) of the equivalent system is given by
Z Z Z
zy(Ky−Kn)
and the average COV at each Z is 0.3227. The statistical z̈⫹w2yz⫽Ae−at⫹ (11)
my
analysis also revealed that the average COV of loading
duration is about 0.13. in which z=mid-span displacement; wy=post-yielding
From the principle of conservation of impulse, the frequency; A=equivalent explosive force per unit mass;
decaying rate a is found to be related to duration to zy=yield displacement; Kn, Ky=pre- and post-yielding
through the relation a=2/to. Since the decay rate and flexural stiffness, and my=post-yielding mass.
loading duration are inversely related, it is reasonable to The motion equation of the equivalent direct shear
assume that they share the same COV. SDOF system is
194 H.Y. Low, H. Hao / Engineering Structures 24 (2002) 189–198
where y=shear slip at support; w1=pre-yielding direct To the authors’ knowledge, there is no definition yet
shear stiffness; L,b=length and breadth of slab; KL=load in the available literature to define the conditions under
factor, and ms=shear mass. On the right hand side is the which the flexural or direct shear failure occurs. Ross
applied shear force at the support, where the first term [8] has come up with failure curves that divide the peak
is contributed by the pressure loading and the second pressure–rise time diagram into two parts; for loading
term from the inertia force of the structure. The two conditions above the failure curve, direct shear failure is
SDOF systems are loosely coupled through the acceler- expected, and for the rest of the region, flexural failure
ation of the flexural response that appears in the direct or no failure can be anticipated. He compared the
shear equation as inertia force. demand and supply of flexural and direct shear strengths
Their maximum responses can be derived as follows. of the slab in each time increment. The mode in which
the strength demand reaches its capacity first is the fail-
z(tmax,f)⫽C1sinwytmax,f⫹C2coswytmax,f (13) ure mode. However, no information is given on the prob-
ability of such damage to the structure.
A −t a zy(Ky−Kn)
⫹ e max,f ⫹ Krauthammer et al. [5,9,16] used SDOF and Timosh-
a +w2y
2
w2ymy enko beam theory to model the structural behavior of
slabs under blast. Although their numerical results com-
y(tmax,ds)⫽H1sinw2tmax,ds⫹H2cosw2tmax,ds (14) pare well with the test data, no prediction is given as to
0.5 PrmaxbL −t 0.5KLz̈(tmax,ds) y1(K2−K1) when one failure mode happens instead of the other.
⫹ 2 2
e max,dsa⫹ ⫹ Fig. 3(a)–(c) show the failure probabilities of the two
ms(a +w2) w22 w22ms
failure modes of a one-way simply supported RC slab,
where z(tmax,f) and y(tmax,ds) are the maximum responses of dimensions 1*3*0.17 m3 with 1% tension reinforce-
of the flexural and direct shear modes respectively where ment and nominal top reinforcement. Detailed infor-
w2=post-yielding direct shear frequency, C1, C2, H1 and mation of the slab used is given in Table 2. The peak
H2 are constants dependent on initial conditions. reflected pressure is 1.5 and 2 MPa, with varying loading
duration on the horizontal axis. The results are calculated
The performance functions of both the flexural and
by using calrel [38] with the above two performance
direct shear modes relate the maximum responses to the
functions and the statistical properties of structural para-
ultimate limits. This is done by dividing the maximum
meters defined in Table 1.
displacement reached under the blast loading by the ulti-
From the figure, a few important characteristics can
mate displacement; if the value is greater than one, fail-
be observed:
ure is initiated. Therefore, the performance functions of
the two modes are shown as follows.
1. When the peak pressure increases, failure prob-
abilities of both modes increase at the same loading
zmax
gf⫽1⫺ (15) duration, as expected.
zu 2. For a low enough peak reflected pressure (Fig. 3(a)),
flexural failure mode dominates the whole range of
ymax loading duration considered.
gds⫽1⫺ (16)
y2 3. When the peak reflected pressure increases, the “inter-
mediate” behavior occurs, where both modes are
where zu and y2 are the ultimate displacement of the active within the probability of failure range of zero
structure at the mid-span and the ultimate direct shear and one. In this case, as demonstrated in Fig. 3(b),
slip at the support corresponding to the flexural and the direct shear mode dominates in the shorter loading
direct shear responses respectively. zu needs be calcu- duration, and the flexural mode is important for longer
lated iteratively as indicated in the flowchart given in loading duration.
Fig. 1. The description of the procedure for calculating 4. If the peak reflected pressure further increases, failure
zu is given in Section 4.1. The ultimate shear slip, y2, is is dominated solely by the direct shear mode in the
chosen as 0.6 mm as discussed in Section 4.2. loading duration range considered, as shown in Fig.
With the performance functions of both the flexural 3(c).
and direct shear modes defined, the corresponding failure
probabilities of them can be computed by the first-order If the dominating failure probabilities for some particular
reliability method (FORM). The standard reliability values are of concern, say 0.1, 0.5 and 0.8, then the peak
method is used in the present study [37]. The numerical reflected pressure and the corresponding loading dur-
calculation is performed by using a computer software ation that give these failure probabilities can be
calrel [38]. extracted, as shown in Fig. 4. In this figure, each contour
H.Y. Low, H. Hao / Engineering Structures 24 (2002) 189–198 195
Table 2
Standard values adopted in analysis
failure mode for the slab with a given peak reflected This is demonstrated in the contour of failure prob-
pressure and loading duration. ability of 0.1, corresponding to different span lengths in
Fig. 6 with the failure modes differentiated by symbol.
8.1. Effect of span length To present a clearer picture, only three span lengths stud-
ied are shown in this figure. It shows that the load resist-
A series of simulations is run for slabs with different ance of the slab in the direct shear mode is higher as
span lengths. It should be noted that in all the cases, the the span length L decreases. However, when the peak
two shorter sides of the slabs are simply supported, and pressure is small and the dominating failure mode is
the other two sides are free. To ensure that the total force flexural, the reverse is true. This can be seen from the
acting on the slabs under the same pressure loading is points corresponding to the peak reflected pressures of
approximately the same, the area of the slab is main- 0.5 and 0.7 MPa in the figure. It is clear that the order
tained at around 10 m2. Hence, the slab width b also swapped as the failure mode changed from direct shear
varies with its span length L. The actual dimensions of to flexural. The above observations indicate that as the
each slab used in the simulation are given in Table 3. slab approaches a square shape, it tends to fail in flexural
Other parameters of the slab such as the thickness, effec- mode. In other words, if a structure does not fail in direct
tive depth and reinforcement ratio, remain the same as shear mode during the earlier stage when subject to an
in Table 2. It should be noted that although the L/b ratio explosion, a rectangular slab stands a better chance of
in some cases is less than 2.0, the slabs are designed as surviving the flexural failure mode than a square one
one-way slabs and analyzed by the beam theory with the same surface area.
described above. Since the flexural rigidity of a beam Fig. 7 demonstrates the failure mode transitions for
depends on I/L3, in which I is the moment inertia of the slabs with different L, with their equations shown against
cross section, increasing L and reducing b results in a the transitions. As loading points above a transition is
smaller flexural rigidity and a larger allowable deflec- more likely to initiate a direct shear failure and the ones
tion. On the other hand, it reduces its strength capacity below in flexural failure, a slab with a smaller span
as the cross sectional area decreases. Hence, for slabs length has less chance of failing in direct shear mode.
with the same surface area, the larger the span length is,
the better is its capacity to resist blast loading in flexural
mode, but lower is its capacity in direct shear failure
mode.
Table 3
Slabs of different span lengths
Prmax 冉
L2
A 冊
⫹16.3847 ⫽107.1571to (17)
References
national Conference on Structures Under Shock and Impact, Research Report Submitted to ACI Committee 117: Tolerances,
Cambridge, Massachusetts, USA, 1989. Auburn, Ala.: Auburn University, 1975.
[14] Kiger SA, Hyde DW, Guice LK. Flexural response of reinforced [26] Mirza SA, MacGregor JG. Variations in dimensions of reinforced
concrete structures to conventional weapons. Proceedings of the concrete members. J Struct Div 1979;105(ST4):751–66.
International Symposium on the Conventional Munitions With [27] Bischoff PH, Perry SH. Compressive behavior of concrete at high
Protective Structures, Federal Republic of Germany: Manheim, strain rates. Mat Struct/Matériaux et Construct 1991;24:425–50.
1987:N42–N53. [28] Krauthammer T. Workshop on structural concrete slabs under
[15] Baker WE, Spivey KH. BIGGS — Simplified elastic–plastic impulsive loads. Proceedings of the Third International Confer-
dynamic response. Proceedings of the First International Confer- ence on Structures under Shock and Impact, Spain: Madrid,
ence on Structures Under Shock and Impact, Cambridge, Massa- 1994:99–106.
chusetts, USA, 1989. [29] CEB. Concrete structures under impact and impulsive loading.
[16] Krauthammer T, Assadi-Lamouki A, Shanaa HM. Analysis of Synthesis Report, Bulletin d’Information No. 187. Lausanne:
impulsive loaded reinforced concrete structural elements. II: Comité Euro-International du Béton, 1988.
Implementation. Comput Struct 1993;48(5):861–71. [30] Liu GQ, Owen DRJ. Ultimate load behavior of reinforced con-
[17] Low HY. Reliability analysis of structures under explosive load- crete plates and shells under dynamic transient loading. Int J
ing. Masters thesis, Nanyang Technological University, 2000. Numer Meth Engng 1986;22:189–208.
[18] Mirza SA, Hatzinikolas M, MacGregor JG. Statistical descrip- [31] Hognestad E. A study of combined bending and axial load in
tions of strength of concrete. J Struct Div ASCE reinforced concrete members. University of Illinois Engineering
1979;105(6):1021–37. Experimental Station, Bulletin Series No. 399, 1951:128.
[19] ACI Committee 318 building code requirements for reinforced [32] Biggs JM. Introduction to structural dynamics. New York:
concrete and commentary. Detroit: American Concrete Insti- McGraw-Hill Book Company, 1964.
tute, 1995. [33] Chung HW. Shear strength of concrete joints under dynamic
[20] Australian Standard 3600. Concrete structures. North Sydney: loads. J Concrete Soc 1978;12(3):27–9.
Standards Association of Australia, 1988. [34] Mills CA. The design of concrete structures to resist explosions
[21] MacGregor JG, Mirza SA, Elingwood B. Statistical analysis of and weapon effects. Proceedings of 1st International conference
resistance of reinforced and prestressed concrete members. ACI for hazard protection, Edinburgh, 1987.
J 1983;80:167–76. [35] Structures to resist the effects of accidental explosion. Air Force
[22] Arafah AM. Statistics for concrete and steel quality in Saudi Ara- Manual AFM 88-22, UNCL, 1969.
bia. Mag Concrete Res 1997;49(180):185–93. [36] Fundamentals of protective design for conventional weapons.
[23] Val DV, Bljuger F, Yankelevsky DZ. Reliability assessment of Technical Manual TM5-855-1, US Department.
damaged RC framed structures. J Struct Engng [37] Melchers RE. Structural reliability analysis and prediction. New
1997;123(7):889–95. York: John Wiley and Sons, 1987.
[24] Mirza SA, MacGregor JG. Variability of mechanical properties [38] Liu PL, Lin HZ, Der Kiureghian A. calrel User Manual, Report
of reinforcing bars. J Struct Div ASCE 1979;105:921–37. No. UCB/SEMM-89/18, Department of Civil Engineering, Uni-
[25] Connolly JP. Construction tolerances in reinforced concrete. versity of California at Berkeley, 1989.