You are on page 1of 20

Engineering Structures 255 (2022) 113958

Contents lists available at ScienceDirect

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

Investigation of ultra-high performance concrete slabs under contact


explosions with a calibrated K&C model
Jian Liu a, Jun Li b, Jianguang Fang b, Kai Liu a, *, Yu Su c, Chengqing Wu b, *
a
Protective Structures Centre, School of Civil Engineering, Guangzhou University, Guangzhou 510006, China
b
School of Civil and Environmental Engineering, University of Technology Sydney, Ultimo, NSW 2007, Australia
c
Hubei RockTek Limited Company, Daye 435100, China

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

Keywords: Karagozian and Case (K&C) concrete model is extensively adopted in the numerical simulations of ultra-high
K&C concrete model performance concrete (UHPC) structural members subjected to impulsive loads such as impact and blast. In
Contact explosion this study, a calibration of the K&C concrete model was conducted for UHPC in terms of three strength surfaces,
Ultra-high performance concrete
equation of state, shear dilatancy, damage evolution and strain rate effect to offer simple and general guidelines
Numerical simulation
Empirical equation
on the determination of key model parameters for this new class of concrete. With the calibrated concrete model,
a single element method was adopted to verify its accuracy through a comparison to the results from the static
tests of the uniaxial compression, direct tension and triaxial compression. Furthermore, the numerical simula­
tions of contact explosion tests on the UHPC slabs with the incorporation of the strain rate effect were performed
and the numerical results exhibited good predictions regarding the failure mode, crater and scabbing damage as
compared to the test results. More importantly, this proposed numerical model and simulation methodology are
reasonable to be generally used for structural members constructed of UHPC materials under contact explosions
when lacking sufficient static and dynamic test data. Using the calibrated and validated K&C concrete model,
parametric studies were conducted to derive a new empirical equation for predicting the local damage mode of
UHPC slabs under contact explosions.

1. Introduction extremely fast loading and the damage evolution of the UHPC structural
members. With the rapid development of computational tools and me­
When concrete structural members are subjected to contact explo­ chanics, different commercial finite element software such as LS-DYNA
sions, spall phenomena on both front and rear surfaces frequently occur offers an alternative approach to study the dynamic response and
due to the propagation of the initial compressive stress waves and the damage of UHPC under blast and high-velocity impact loads. Reported
reflected tensile waves [1,2]. Normal strength concrete (NSC) and high from the open literature, a number of constitutive concrete models, e.g.
strength concrete (HSC) structures with brittle nature generally respond Elastic-Plastic Hydrodynamic (EPH) model, Karagozian and Case Con­
with highly localized damage under contact explosions resulted from the crete (K&C) model, Winfrith Concrete (WC) model, Brittle Damage
weak dynamic strength inferior to the stress waves. Known for its su­ model, Holmquist Johnson Concrete (HJC) model, Continuous Smooth
perior mechanical strength and ductility, ultra-high performance con­ Cap (CSC) model, Riedel Hiermaier Thomas (RHT) model, etc. have
crete (UHPC) has been widely developed in these decades [3–7] and its been successfully used in LS-DYNA to reproduce the dynamic response
outstanding performance was confirmed in resisting hazardous loads of concrete under blast and high-velocity impact loads. Compared to
such as blast and projectile impact by the previous physical tests [8–14]. other concrete models as mentioned above, the K&C concrete model
Physical testing data offer the credible index to demonstrate the exhibits more accurate basic behaviour with thorough considerations of
damage phenomena of UHPC subjected to hazardous loads, whereas the confinement effects, three-invariant of the deviatoric stress, shear
physical tests are frequently hindered by the safety concerns, high cost dilatancy, strain rate effects and tensile fracture [15–18]. Another
of facilities and test configurations, data acquisition difficulties as well advantage of the K&C concrete model is that a majority of the model
as non-repeatability in some cases especially attempting to capture the parameters along with the equation of state (EOS) can be automatically

* Corresponding authors at: School of Civil and Environmental Engineering, University of Technology Sydney, Ultimo, NSW 2007, Australia (C. Wu).
E-mail addresses: lkeason@sina.com (K. Liu), Chengqing.Wu@uts.edu.au (C. Wu).

https://doi.org/10.1016/j.engstruct.2022.113958
Received 13 July 2021; Received in revised form 10 January 2022; Accepted 28 January 2022
Available online 7 February 2022
0141-0296/© 2022 Elsevier Ltd. All rights reserved.
J. Liu et al. Engineering Structures 255 (2022) 113958

Fig. 1. Setup of the single element model with an element size of 5 mm (a) uniaxial compression (b) uniaxial tension.

Fig. 2. Stress–strain curves of UHPC using the K&C model with original model parameters (a) uniaxial compression (b) uniaxial tension.

generated after inputting simple parameters such as uniaxial compres­ determination of the key model parameters for more generic UHPC
sive strength, material density and Poisson’s ratio. However, these materials, Zhang et al. [26] presented an improved calibration approach
automatically generated model parameters are likely to be only in line and the applicability of the proposed approach was verified via a large
with the empirical relations based on a large number of experimental number of high-velocity impact tests with projectiles of various di­
data for NSC [15]. mensions and striking velocities, whereas the key model parameters that
In recent studies, calibrated versions of the K&C concrete model have controlled the softening behaviour of UHPC, e.g. b1 and b2, were still
been extensively adopted by many researchers to simulate the UHPC case-independent and the EOS parameters for UHPC were not well
structural members subjected to low-velocity impact, high-velocity calibrated due to the lack of relevant experimental data. Besides, the
impact and blast loads. Xu and Wille [19] modified the three strength proposed approach still needs to be further investigated for other dy­
surfaces, damage function and EOS to fit for UHPC, and then the validity namic loadings such as blast.
was confirmed through a comparison with the uniaxial and triaxial In this study, according to the similar approach as conducted by
compressive tests. With the same calibration method, the modified Zhang et al [26], a calibration of the K&C concrete model in a more
model was further validated via Split Hopkinson Pressure Bar (SHPB) comprehensive manner, especially establishing an empirical equation
tests as conducted by Xu and Wille [20] and Ren et al. [21]. Moreover, with respect to element size to determine the model parameters that
Mao et al. [22], Zhang et al. [23] and Lin [24] adopted the default pa­ govern the compressive and tensile softening behaviour, was conducted
rameters of the three strength surfaces but modified some of the other based on the static experimental data for a benchmark UHPC (as known
parameters such as the damage function, shear dilatancy, compressive as Ductal@ UHPC) to provide simple and general guidelines on the
and tensile softening behaviour, strain rate effect, localization width and determination of the key model parameters for generic UHPCs. Addi­
EOS, and then the applicability of the calibrated model was fairly tionally, EOS parameters were determined so as to be used for generic
demonstrated under the blast scenarios. Xu et al. [25] and Su et al. [12] UHPCs. Using the calibrated K&C model, full-scale numerical simula­
systematically calibrated the model parameters via a single element tions of the contact explosion tests were performed on the UHPC slabs
model based on a series of the uniaxial compression, direct tension and with various steel rebar reinforcements, slab thicknesses and TNT
triaxial compression tests. Moreover, the updated versions of the model explosive charge weights (as reported in Refs [9,27]) to demonstrate the
were respectively validated through the low-velocity impact and validity of the proposed method through a comparison to the experi­
medium-range blast tests. However, the calibrated parameters may not mental results. Moreover, parametric studies were performed to derive a
be valid for more generic cases owing to various compositions in UHPC new empirical equation to predict the local damage mode of the UHPC
and loading scenarios. In order to provide general guidelines on the slabs under contact explosions.

2
J. Liu et al. Engineering Structures 255 (2022) 113958

Fig. 3. Procedures for the calibration of K&C material model used in static tests and contact explosion tests.

Fig. 4. Failure surfaces of Ductal@ UHPC (a) 188 MPa UHPC (b) 250 MPa UHPC.

2. A brief overview of K&C model in LS-DYNA ⎧


⎪ P fc

⎪ a0m + , whenP ≥

⎪ a1m + a2m P 3
The K&C model is a plasticity-based concrete model and requires ⎪



three independent strength surfaces labelled as the maximum, yield and Δσ m =
3 fc
(P + ft ), when0 ≤ P ≤ orλ ≤ λm and − ft ≤ P ≤
fc
(1a)
residual surfaces. This model also accounts for damage, volumetric and ⎪


2ψ 3 3
⎪ ( )
deviatoric behaviour as well as the strain rate effect of concrete mate­ ⎪


⎪ P
⎩ 3 + ft , whenP ≤ 0andλ > λm
rials. This section presents the formulations adopted to realize the K&C η
model as reported in Ref [15].
For the principal stress difference in the yield strength surface (Δσ y ):
The compressive meridians of the three independent strength sur­
faces are formulated by: ⎧
⎪ P fyc

For the principal stress difference in the maximum strength surface ⎪


a0y +
a + a P
, whenP ≥
3
⎪ 1y 2y
(Δσm ): ⎪
⎨ ( )
Δσ y = 1.35ft fyc (1b)

⎪ 1.35ft + 3P 1 − , when0 ≤ P ≤

⎪ fyc 3




1.35(P + ft ), whenP ≤ 0

3
J. Liu et al. Engineering Structures 255 (2022) 113958

⎧ ∫ εp
⎪ dεP

⎪ ( )b , whenP ≥ 0

⎪ P 1

⎪ 0
r 1 +

⎨ f
rf ft
λ= ∫ (3)

⎪ ε p
dεP

⎪ , whenP < 0

⎪ ( )b 2

⎩ 0 rf 1 + P

rf ft
√̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
where dεP = 23 dεPij dεPij is a differential quantity of the effective
plastic strain and dεPij is a differential quantity of the plastic strain; b1 is
the damage scaling parameter in compression; b2 is the damage scaling
parameter in tension; rf is the factor of the strain rate enhancement re­
flected by dynamic increase factor (DIF). To capture the damage incre­
ment in a hydrostatic tension failure scenario, a volumetric damage
increment is defined in the deviatoric damage of λ, which is given by:
( )
Δλ = b3 fd kd εv − εv,yield (4)
Fig. 5. The derived EOS for UHPC.
where b3 is the damage evolution parameter that governs the triaxial
tension softening behaviour; kd is the internal scalar multiplier; εv and
For the principal stress difference in the residual strength surface εv,yield denote the volumetric strain and the volumetric strain at yield,
(Δσr ): respectively; fd is the factor for the increment of the damage to mitigate
P the effects of this change to the paths close to the triaxial tensile path,
Δσ r = (1c)
a1r + a2r P which is defined as:
⎧ ⃒√̅̅̅̅̅̅̅ ⃒
where aij (i = 0, 1, 2; j = m, y, r) are the corresponding strength ⎪
⎪ ⃒ 3J2 ⃒
⎪ ⃒ ⃒ ⃒√̅̅̅̅̅̅̅ ⃒〈
surface parameters obtained from triaxial compression tests under a ⎪

⎪ ⃒ P ⃒ ⃒ 3J2 ⃒
⎨1 − , when0 ≤ ⃒⃒ ⃒ 0.1
wide array of confinements; P = σ1 +σ32 +σ3 is the hydrostatic pressure and fd = 0.1 P ⃒ (5)
σ i (i = 1, 2, 3) is the principal stress; fc, ft and fyc = 0.45fc denote the ⎪
⎪ ⃒√̅̅̅̅̅̅̅ ⃒

⎪ ⃒ 3J ⃒

uniaxial compressive strength, uniaxial tensile strength and yield
2⃒

⎩ 0, when⃒⃒ ≥ 0.1
P ⃒
strength, respectively; ψ is the tensile-to-compressive meridian ratio,
which is defined as:


⎪ 0.5, whenP ≤ 0



⎪ ft fc

⎪ 0.5 + 1.5 , whenP =



⎪ fc 3


⎨ αfc 2αfc
ψ= 2αfc
, whenP = , α ≈ 1.15 (2)

⎪ 3

⎪ a0m +

⎪ 3a1m + 2a2m αfc



⎪ 0.753, whenP = 3fc





1, whenP ≥ 8.45fc

η ranging from 0 to 1 represents a failure surface interpolation


function with respect to the modified effective strain or the internal
damage parameter (λ). λm is the maximum magnitude as η = 1. λ is
determined from an equation including several state variables as well as
the strain rate effect, which is given by:

Fig. 6. Default and modified η-λ relationship.

Table 1
EOS parameters for UHPC.
n 1 2 3 4 5 6 7 8 9 10

μn 0 − P2/K1 − 0.0101 − 0.0305 − 0.0513 − 0.0726 − 0.0943 − 0.12 − 0.174 − 0.208


Pn (MPa) 0 200 300 901 1514 2141 2780 3537 5164 6973
Kn (GPa) Ec/3(1-2ν) Ec/3(1-2ν) 30.13 35.97 40.89 45.94 51.07 57.16 69 69

Table 2
Modified parameters in the η-λ function.
n 1 2 3 4 5 6 7 8 9 10 11 12 13

λn 0 8E-6 2.4E-5 4E-5 5.6E-5 7.2E-5 8.8E-5 2.5E-4 4E-4 5.5E-4 7.5E-4 1E-3 1E10
ηn 0 0.85 0.97 0.99 1 0.99 0.97 0.65 0.38 0.19 0.05 0 0

4
J. Liu et al. Engineering Structures 255 (2022) 113958

Fig. 7. Determination of b1 with different sizes of single element models under uniaxial compression (a) 2.5 mm (b) 5 mm (c) 8 mm (d) 12 mm (e) 15 mm.

√̅̅
where J2 is the second invariant of the deviatoric stress tensor. where θ can be determined based on cosθ = 3 √s1̅̅̅̅ or cos3θ =
2 J2
Combing the three strength surfaces, the current failure or plasticity √̅̅
J3
surface is defined considering three stress invariants (I1, J2 and J3),
3 3
2
̅; s1 = σ1 − P is the first principal deviatoric stress.
√̅̅̅̅̅̅̅
J23/2

which is given by: In addition to the deviatoric aspects of concrete behaviour as intro­
{ ’ ( )
r [η Δσm − Δσy + Δσ y , whenλ ≤ λm (strainhardening) duced above, a companion EOS, i.e. EOS_Tabulated_Compaction, is
Δσ = ’ (6) required to completely describe its volumetric properties. The pressure P
r [η(Δσ m − Δσr ) + Δσ r , whenλ > λm (strainsoftening)
is defined by:
It shall be noted that the failure surfaces for concrete in the devia­
toric plane generally convert from a triangular shape at low pressures to P = C’ (εv ) + γT(εv )E (8)
a circular shape at high pressures. Attempting to achieve this, a where C (εv ) and γT(εv )E represent the tabulated mechanical pres­

dependence on the Lode angle θ as proposed by William and Warnke sure and the thermally induced pressure, respectively; γ is the ratio of
[28] is included to offer a smooth and convex triangular surface the specific heat; T(εv ) is the tabulated temperature parameter depen­
generated by elliptical segments through a scale factor r’ , which is dent on the volumetric strain; E is the internal energy per initial volume.
defined by:
√̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
2(1 − ψ 2 )cosθ + (2ψ − 1) 4(1 − ψ 2 )cos2 θ + 5ψ 2 − 4ψ
r’ = (7)
4(1 − ψ 2 )cos2 θ + (1 − 2ψ )2

5
J. Liu et al. Engineering Structures 255 (2022) 113958

with the compressive strength of 223.6 MPa [7]. The equations of


strength surface parameters are shown as follows:
a0 = 0.3444fc (9a)

a1 = 0.4463 (9b)

a2 = 0.1847/fc (9c)

a1f = 0.4417 (9d)


/
a2f = 0.1737 fc (9e)

a0y = 0.2182fc (9f)

a1y = 0.625 (9g)


/
a2y = 0.5433 fc (9h)
Fig. 8. Fitting curve for the determination of b1 versus element size.
To further prove the applicability of the proposed equations, the
3. Determination of parameters in the K&C model derived strength surfaces from Eq. (9a) and Eq. (9b) for UHPC are
plotted in Fig. 4. The experimental data of Ductal@ UHPC with two
Most of the parameters in the K&C concrete model coupled with uniaxial compressive strengths of 188 MPa [31] and 250 MPa [32] are
EOS_Tabulated_Compaction can be automatically generated after extracted and plotted in Fig. 4. Observed from the comparison, it was
inputting some simple parameters including uniaxial compressive again demonstrated that the proposed equations were applicable to the
strength (fc), material density (ρ) and Poisson’s ratio (ν). To evaluate the prediction of strength surface parameters for generic UHPCs when
suitability of the K&C model for UHPC, two forms of analytic modelling, relevant triaxial compressive test data were insufficient or unavailable.
i.e. single element model and material system model, can be considered
to compare with the results obtained from standard material tests, in
which the single element model adopts a cubic element to compute the 3.3. Equation of state
state resulting from various loads and the material system model mimics
the response of the actual specimen adopted in a specific material test In EOS, three stages including linear elasticity, plastic compaction
[15]. Liu et al. [29] adopted these two models to mimic the uniaxial and non-porous compaction are involved. At the initial stage, UHPC
compressive behaviour of UHPC and similar stress–strain curves under exhibits a linear elastic behaviour until reaching an elastic limit state.
uniaxial compression were achieved. Therefore, to eliminate unwanted Afterwards, with the continuous increment in the pressure, UHPC starts
structural effect, the single element model was used for the evaluation to transfer from the plastic compaction to the fully compressed stage at
and calibration of the K&C model in this study. which all the air voids inside the material are crushed out. During the
non-porous compaction stage, the volumetric strain of UHPC no longer
3.1. Static behaviour of UHPC predicted by the K&C model with default increases in spite of the increased pressure.
parameters A series of hydrostatic pressure tests were performed to capture the
elastic limit pressure of UHPC. Vankirk [33] conducted hydrostatic
Using the single element model with an element size of 5 mm (as compression tests on UHPC with the uniaxial compressive strength of
shown in Fig. 1), the numerical stress–strain curves of Ductal@ UHPC 141 MPa to the target pressure of 400 MPa. It was observed that UHPC
under uniaxial compression and tension were obtained, which is illus­ started to behave inelastically at a hydrostatic pressure of 180 MPa.
trated in Fig. 2. The numerical results were compared with the experi­ Hydrostatic compression tests were also carried out on UHPCs by Wang
mental uniaxial compressive and tensile stress–strain curves of Ductal@ et al. [34] (uniaxial compressive strength > 200 MPa, target fluid
UHPC as provided by VSL in their Melbourne laboratory [30], from pressure = 400 MPa), Williams et al. [7] (uniaxial compressive strength
which it was observed that the original K&C concrete model exhibited = 190–244 MPa, target fluid pressure = 500 MPa) and Smith et al. [35]
greater elastic modulus and more brittle softening behaviour in both (uniaxial compressive strength = ~250 MPa, target fluid pressure =
compression and tension than the test results. In the following sections, 500 MPa). Although various mix ingredients and fibre reinforcements in
using the Ductal@ UHPC [30] as a benchmark, a systematic procedure these UHPCs may lead to distinct hydrostatic stress-volumetric strain
(as shown in Fig. 3) to calibrate the key model parameters was elabo­ responses, the elastic limit was in the range between 200 MPa and 300
rated for the applicability of UHPC materials. More importantly, this MP, thus 200 MPa was considered as the elastic limit pressure in EOS in
methodology and the corresponding guidelines aimed to be commonly this study. As the Young’s modulus of UHPC is internally generated in
adopted for structural members constructed of generic UHPC materials the K&C concrete model based on the elastic bulk modulus in EOS, the
subjected to contact explosions especially when relevant test data were elastic bulk modulus Ke at the first stage has to be calibrated according to
limited. the following equation:
Ec
Ke = (10)
3.2. Strength surfaces 3(1 − 2ν)
where Ec and ν are the Young’s modulus and Poisson’s ratio of UHPC,
The relevant parameters in the initial yield, maximum and residual
respectively. Moreover, if Ec and ν cannot be achieved from physical
strength surfaces are generally determined based on the triaxial
tests, some empirical equations and recommendations, e.
compressive test data under a wide array of confinement levels. As re­ ( )0.5 ( )0.5
ported in Zhang et al.’s study [26], the strength surface parameters, i.e. g. Ec = 3.84 fC (126 MPa ≤ fc ≤ 193 MPa) [36], Ec = 4.069 fC
( )0.36
a0, a1, a2, a0y, a1y, a2y, a1f and a2f, were respectively fitted by regressing a (97 MPa ≤ fc ≤ 179 MPa) [37], Ec = 8.1 fC (0 ≤ fc ≤ 250 MPa)
function with respect to the compressive strength of generic UHPCs or a [38], ν = 0.2 [39], etc. are provided for the determination of Ec and ν,
constant based on the triaxial compressive test data of Cor-Tuf UHPC in which f c is in the unit of MPa and Ec is in the unit of GPa.

6
J. Liu et al. Engineering Structures 255 (2022) 113958

Fig. 9. Determination of b2 with different sizes of single element models under direct tension (a) 2.5 mm (b) 5 mm (c) 8 mm (d) 12 mm (e) 15 mm.

In order to obtain EOS parameters under higher pressures, full-scale 3.4. Shear dilatancy
detonation or planar impact tests which possess very short duration high
pressures are required [40,41]. Based on a collection of the available Values of the shear dilatancy parameter, OMEGA, which governs the
data from the existing planar impact tests, Su et al. [12] proposed a volume expansion and depends on the confinement pressure is expected
tabulated EOS input data of UHPC (the pressure-volumetric strain curve to range from 0 to 1, wherein 0 means no volume expansion or change
is shown in Fig. 5 and the EOS parameters are given in Table 1). How­ occurring under plastic shear flow and 1 represents the situation when
ever, some experimental data adopted for curve fitting were obtained the model is completely associated under plastic shear flow. According
from NSC and HSC. To dispel the doubts of the applicability of the to the study by Wu et al. [43], the recommended values of OMEGA for
derived EOS for UHPC in this study, it was compared with the planar the scenarios with and without confinement are 0.9 and 0.75, respec­
impact test data of UHPC as reported in Refs [31,42], which is shown in tively. Considering the lack of relevant experimental data, OMEGA =
Fig. 5. A close match from the observation demonstrated that the EOS 0.75 had been used in other numerical studies on high-velocity projec­
parameters as proposed by Su et al. [12] can be used for generic UHPCs tile penetration of UHPC [11,26,44–46], and OMEGA = 0.75 was
in the present study. applied for all cases in the present study.

7
J. Liu et al. Engineering Structures 255 (2022) 113958

Owing to the addition of fibres, UHPC exhibits a much more ductile


softening response than conventional NSC, thus the default parameters
of the η-λ function which are initially established for NSC may no longer
fit for UHPC. The determination of the relevant parameters in the η-λ
function can be achieved through the trial-and-error method until the
estimated uniaxial compressive and tensile stress–strain curves show fair
agreement with the experimental results [47]. In this study, combing the
determination of other two plastic strain accumulation parameters, i.e.
b1 and b2, (which will be introduced afterwards), a set of modified pa­
rameters of the η-λ function was proposed for all the scenarios through a
large number of trials of uniaxial compressive and tensile tests. The
modified parameters in the η-λ function are given in Table 2, and the
default and modified η-λ relationships are presented in Fig. 6.
b1 is a mesh-dependent parameter that governs the softening region
in the uniaxial compressive stress–strain curve of UHPC. According to
the previous study by Wu et al. [43], an empirical equation of b1 con­
cerning the element size h (b1 = 0.34 h + 0.79, h is in the unit of mm) was
established to regularize the compressive softening behaviour of NSC.
Fig. 10. Fitting curve for the determination of b2 versus element size. To derive an appropriate relationship between b1 and element size for
UHPC as the characteristic element size alters in the numerical studies, a
3.5. Damage evolution series of single element tests under uniaxial compression with the
element size of 2.5 mm, 5 mm, 8 mm, 12 mm and 15 mm were per­
A modification to the damage evolution parameters including η-λ formed to find the best fitting value of b1 under various element sizes for
function, b1 and b2 that dominate the softening behaviour of UHPC was the experimental uniaxial compressive stress–strain curves. The com­
elaborated by using the single element model (as shown in Fig. 1) and parison of the experimental and numerical stress–strain curves for UHPC
the experimental stress–strain curves of Ductal@ UHPC [30]. under uniaxial compression is shown in Fig. 7. Based on the fitting

Fig. 11. Effect of wlz on the uniaxial compressive and tensile behaviour of UHPC.

Fig. 12. DIF of UHPC (a) compressive strength (b) tensile strength.

8
J. Liu et al. Engineering Structures 255 (2022) 113958

Fig. 13. Setup of triaxial compression simulations on UHPC using the single element model.

Fig. 14. Comparison of triaxial compressive behaviour of UHPC from experimental and numerical results (a) 5 MPa confinement (b) 10 MPa confinement (a) 20 MPa
confinement (a) 60 MPa confinement.

values of b1 as shown in Fig. 8, an empirical equation to predict b1 was introduces a user-defined parameter (localization width, wlz) to offer a
proposed as follows: possibility of handling softening to spread (in the band width direction)
over multiple elements [48]. When h is smaller than wlz, the stress–strain
b1 = 0.54 × ln(h) − 0.25 (11)
relation is accommodated to expect that the fracture energy is dissipated
where h is the characteristic element length in the unit of mm. within the localization width instead of the single element length. When
b2 is also a mesh-dependent parameter to control the softening h is larger than wlz, the standard treatment prevails that the band width
branch in the uniaxial tensile stress–strain curve of UHPC. One way to should be equal to the element size [48]. In the original K&C concrete
eliminate the mesh dependency of b2 in this local damage model is to model, the determination of wlz is affected by the maximum aggregate
remain the tensile softening strain energy (determined from the uniaxial size, e.g. wlz is generally one to six times the maximum aggregate size
tensile stress–strain curve) identical to the material tensile fracture en­ [49] and is recommended to be taken as three times the maximum
ergy divided by the characteristic band width. The K&C concrete model aggregate diameter in the LS-DYNA manual [50]. However, in generic

9
J. Liu et al. Engineering Structures 255 (2022) 113958

Table 3 was not considered unless otherwise stated, and the objective was to
Contact explosion test programs. offer a simple empirical equation to predict b2 so as to be directly used in
Slab Depth TNT Steel Steel rebar the numerical simulations especially when the relevant experimental
No. (mm) charge fibres reinforcement data were insufficient or unavailable.
weight (%) In order to investigate the effect of wlz on the compressive and tensile
(kg)
softening behaviour of UHPC, a single model with the size of 2.5 mm was
Blast UHPC- 100 1.0 2.0 No steel rebar taken as an example for analysis, which is shown in Fig. 11. It was
event 1 reinforcement observed that when wlz was smaller than h, the compressive and tensile
1 [27]
Blast UHPC- 120 0.1 2.5 9 longitudinal
behaviour of UHPC were not affected, which demonstrated that to apply
event 2 rebars each side Eq. (11) and Eq. (12), the localization width should be set to no more
2 [9] UHPC- 120 1.0 2.5 9 longitudinal than the element size.
3 rebars each side
UHPC- 150 1.0 2.5 9 longitudinal
3.6. Strain rate effect
4 rebars each side
UHPC- 100 1.0 2.5 9 longitudinal
5 rebars each side UHPC is less sensitive than conventional NSC under high strain rates
UHPC- 120 1.0 2.5 5 longitudinal owing to the addition of fibres, lower water to cement ratio, denser
6 rebars each side microstructure, etc. [51,52], thus some empirical equations [53] to
predict DIF that were initially derived for NSC may not be acceptable for
UHPC. In recent studies, based on the experimental data, many empir­
Table 4 ical equations have been proposed to evaluate DIF values for compres­
Parameters for UHPCs in the calibrated K&C model. sive and tensile strengths of UHPC [54–59]. Although the DIF values of
Triaxial compression test Blast event 1 Blast event 2 UHPC may vary from material composition, strength, fibre type, volu­
[25] [27] [9] metric content, shape and geometry, etc. [26,55,57], the influence was
ρ (kg/m3) 2500 2425 2425
neglected in the current study so as to provide simple and efficient
fc (MPa) 156 128.9 [8] 145 empirical DIF equations to reflect the strain rate effect in the numerical
ft (MPa) 9.5 8.0 [31] 9.0 [31] simulation. In this study, a set of empirical DIF equations as proposed by
E (GPa) 40.0 51.5 52.0 Fujikake et al. [60] was adopted for UHPC, which is given by:
0.2 0.2 0.2
For compressive strength:
ν
a0 (MPa) 53.66 44.31 49.88
a1 0.4463 0.4463 0.4463 ⎧ ⎡ ⎛ ⎞⎤0.951
a2 (MPa¡1) 1.184E-3 1.434E-3 1.274E-3 ⎪

a1f 0.4417 0.4417 0.4417 ⎨ ( )0.0055⎣log⎝ε̇scε̇ ⎠⎦

a0y (MPa) 34.04 28.1 31.64 DIFc = ε̇
forε̇ ≥ ε̇sc (13a)

⎪ ε̇
a1y 0.625 0.625 0.625 ⎪
⎩ sc
a2f 1.113E-3 1.349E-3 1.198E-3 1forε̇ < ε̇sc
(MPa¡1)
a2y 3.483E-3 4.218E-3 3.747E-3
(MPa¡1) ε̇sc = 1.2 × 10− 5 s− 1

b1 1.0 1.0 1.0


b2 − 1.0 − 1.0 − 1.0
For tensile strength:
b3 1.15 1.15 1.15 ⎧ ⎡ ⎛ ⎞⎤1.95
Omega 0.75 0.75 0.75 ⎪

wlz (mm) ≤10 ≤10 ≤10 ⎨ ( )0.0013⎣log⎝ε̇ε̇st ⎠⎦

η-λ Table 2 Table 2 Table 2 DIFt = ε̇
forε̇ ≥ ε̇st (13b)

EOS Table 1 Table 1 Table 1 ⎪
⎪ ε̇st

1forε̇ < ε̇st
UHPCs, coarse aggregates are generally replaced by ultra-high reactive
ε̇st = 1.0 × 10− 6 s− 1
particles to strengthen the underlying material homogeneity, which
means the maximum aggregate size is sufficiently small and will not be As also noted, a cut-off value to the empirical DIF equations was
appropriate as a major basis to determine wlz. In this study, it was suggested in the LS-DYNA manual [50] and the study by Zhang et al.
assumed that the fracture energy was realized over a single element, i.e. [26], with the purpose of avoiding an overestimation of DIF values at
wlz = h, to avoid unnecessary disputes, and this assumption was also high strain rates caused by the inertia effect. In this study, according to
employed in other studies for UHPC subjected to dynamic loads [25,29]. the proposed equation to calculate DIF values of NSC [53], the recom­
Wu et al. [43] proposed an empirical equation with the variables of mended cut-off values of DIF in compression and tension were set at the
wlz and fc to calculate b2 for NSC. Since wlz was taken as the same value strain rate of 300 s− 1 and 160 s− 1, respectively. In order to evaluate the
of h and it was expected that the compressive strength of UHPC had a suitability of the Fujikake et al.’s empirical DIF models, DIF values for
slight effect on the softening response, the same strategy as the deter­ compressive strength [59,61–69] and tensile strength [54,56,66,70–80]
mination of b1 was taken to find the optimal fitting value of b2 under of UHPC under a wide array of strain rates were collected from existing
various sizes of single element models (2.5 mm, 5 mm, 8 mm, 12 mm open literature and then compared to the empirical results, which is
and 15 mm) for the experimental uniaxial tensile stress–strain curve, as shown in Fig. 12. Through the comparison, the Fujikake et al.’s empir­
shown in Fig. 9. Based on the fitting values of b2 as shown in Fig. 10, an ical DIF model for the compressive strength was in good agreement with
empirical equation to predict b2 was given by: the experimental data, while the empirical DIF model for the tensile
strength overestimated the experimental data. In this study, based on the
b2 = − 4.15 × ln(h) + 8.6 (12)
Fujikake et al.’s empirical DIF models, an updated version was derived
Fibre properties such as type, shape, geometry and volumetric con­ for UHPC, which is given by:
tent, etc. may affect the softening behaviour of UHPC, thus Eq. (12) may For compressive strength:
not be accurate and needs further calibration for other UHPCs. However,
in this study, the influence of fibre properties for the determination of b2

10
J. Liu et al. Engineering Structures 255 (2022) 113958

Fig. 15. Schematic diagram of UHPC slabs and TNT explosives (a) blast event 1 (b) blast event 2.

Fig. 16. Reinforcement of UHPC slabs in the blast event 2 of contact explosion tests (a) UHPC-2 to UHPC-5 (b) UHPC-6.

11
J. Liu et al. Engineering Structures 255 (2022) 113958

Fig. 17. Supporting systems in contact explosion tests (a) blast event 1 (b) blast event 2.

Fig. 18. Numerical models in contact explosion tests (a) air, TNT explosive and slab (b) UHPC slab without reinforcement of steel rebars (c) 9 longitudinal steel
rebars reinforced UHPC slabs (d) 5 longitudinal steel rebars reinforced UHPC slabs.

⎧ 1.87forε̇ ≥ ε̇cut− off 4. Validation of the calibrated K&C model



⎪ ⎡ ⎛ ⎞⎤1.88




⎨ After the systematic calibration of the K&C concrete model based on
( )0.0013⎣log⎝ε̇scε̇ ⎠⎦ (14a)
DIFc =
⎪ ε̇ the uniaxial compressive and tensile behaviour of the reference Ductal@

⎪ forε̇sc ≤ ε̇ < ε̇cut−


⎪ ε̇sc off
UHPC as introduced in Section 3, its applicability was verified under

1forε̇ < ε̇sc triaxial stress states and contact explosions for generic UHPCs using the
commercial finite element software LS-DYNA.
For tensile strength:
⎧ 3.61forε̇ ≥ ε̇cut−
4.1. Triaxial compression tests and single element analysis
off

⎪ ⎡ ⎛ ⎞⎤1.95





DIFt = ( )0.0013⎣log⎝ε̇ε̇st ⎠⎦ (14b) Triaxial compression tests on UHPC were simulated using the single
⎪ ε̇


⎪ forε̇ ≥ ε̇st element model as shown in Fig. 13. The lateral confining pressures
⎪ ε̇

⎩ st acting on the model were 5 MPa, 10 MPa, 20 MPa and 60 MPa, and
1forε̇ < ε̇st Table 4 lists all the relevant model parameters of UHPC referring to the
where ε̇sc = 1.2 × 10− 5 s− 1 ; ε̇st = 1.0 × 10− 6 s− 1 ; ε̇cut− off = 300s− 1
for guidelines in Section 3. The numerical results were compared to the
compressive strength; ε̇cut− off = 160s− 1 for tensile strength. experimental results for UHPC (uniaxial compressive strength = 156
MPa, direct tensile strength = 9.5 MPa and Young’s modulus = 40 GPa)
under the consistent confinement levels as conducted by Xu et al. [25]. It
can be indicated that the numerical results for the calibrated K&C
concrete model agreed fairly with the experimental axial stress–strain

12
J. Liu et al. Engineering Structures 255 (2022) 113958

Table 5 curves under various confinement levels from 5 MPa to 60 MPa.Fig. 14


Model parameters for TNT explosive, air and steel rebars.
Material Material model Parameter Value 4.2. Contact explosion tests
3
TNT Mat_High_Explosive_Burn ρ (kg/m ) 1650
explosive The calibrated K&C concrete model was further verified by two
Detonation velocity 6900
events of contact explosion tests on the UHPC slabs, wherein the test
(m/s)
C-J pressure (Pa) 2.1E + 11 programs are presented in Table 3.
EOS_JWL A 3.71E + In the first event of the contact explosion tests, one 2% steel fibre
11 reinforced Ductal@ UHPC slab with dimensions of 2000 mm × 1000 mm
B 3.29E + 9 × 150 mm was tested. The uniaxial compressive strength and Young’s
R1 4.15
R2 0.95
modulus of UHPC were 128.9 MPa and 51.5 GPa, respectively [8]. 1 kg
ω 0.3 cylindrical TNT explosive with a diameter to length ratio of 1.0 was
E0 (J/m3) 7E + 9 placed on the upper surface of the slab. Fig. 15(a) shows the schematic
Air Mat_Null ρ (kg/m3) 1.29 diagram of the UHPC slab and the TNT explosive.
EOS_Linear_Polynomial C4 0.4 In the second event of the contact explosion tests, five UHPC slabs
C5 0.4
reinforced with 2.5% steel fibres and steel rebars were tested. The di­
E0 2.5E5
V0 1 mensions of the UHPC slabs were 2000 mm × 800 mm, and the thickness
Steel rebars Mat_Plastic_Kinematic ρ (kg/m3) 7850 varies between 100 mm and 150 mm. The uniaxial compressive strength
E (GPa) 210 and Young’s modulus for UHPC were 145 MPa and 52 GPa, respectively.
ν 0.3 The diameters of the longitudinal rebar and stirrup rebar were 12 mm
Yield stress (MPa) 360
Tangent modulus (GPa) 2.1
and 8 mm, respectively, and they shared the same yield stress of 360
C (s− 1) 6844 MPa. 0.1 kg cylindrical TNT explosive (diameter = 40 mm and depth =
p 3.91 45 mm) and 1 kg cylindrical TNT explosive (diameter = 120 mm and
Failure strain 0.12 depth = 60 mm) were placed on the upper surface centre of the slab.
Fig. 15(b) presents the schematic diagram of the UHPC slabs and the
TNT explosives. Fig. 16 shows the steel rebar reinforcement configura­
tions in the UHPC slabs.
TNT explosives in both events of contact explosion tests were acti­
vated by an electrical detonator bonded in the core of the explosive.

Fig. 19. Development of SDF versus strain for UHPC in blast event 1 (a) uniaxial compression (b) uniaxial tension.

Fig. 20. Development of SDF versus strain for UHPC in blast event 2 (a) uniaxial compression (b) uniaxial tension.

13
J. Liu et al. Engineering Structures 255 (2022) 113958

Fig. 21. UHPC-1 (a) experimental results (b) numerical results with default model parameters (c) numerical results with calibrated model parameters.

system. Since contact explosions would cause highly localized damage


Table 6 to the slabs, the boundary conditions offered a negligible influence on
Comparison of experimental and numerical results.
the test results.
Experiment Numerical simulation

Crater (mm) Scabbing Crater (mm) Scabbing 4.3. Numerical modelling of contact explosions
diameter diameter
Diameter Depth Diameter Depth
(mm) (mm)
For the numerical simulation of contact explosions, three approaches
UHPC- 350 / 380 330 / 370 including multi-material arbitrary Lagrangian–Eulerian (ALE),
1
smoothed particle hydrodynamics (SPH) and particle blast method
UHPC- 90 27 / 100 20 /
2 (PBM) are generally implemented. However, considering the drawbacks
UHPC- 230 / 450 270 / 430 of the SPH-FE method (e.g. results dependent on the quantity of SPH
3 particles for the explosive and discretization of the blast load acted on
UHPC- 270 / 470 300 / 410
the specimen) and the PBM method (e.g. disability to define EOS for
4
UHPC- 220 / 410 260 / 390
explosives) [81], the ALE formulation was adopted in the current nu­
5 merical modelling.
UHPC- 250 / 480 290 / 450 Fig. 18 presents the three-dimensional numerical models of the
6 UHPC slab, steel rebar, TNT explosive and air. The concrete slab was
Note: numerical crater and scabbing diameters were recorded by measuring the modelled by Lagrangian solid elements with the mesh size of 10 mm and
maximum diameter of the damaged area in red. the steel rebar was meshed by Hughes-Liu beam elements with the mesh
size of 10 mm. Constrained_Lagrange_In_Solid was used to define the
Owing to the small net explosive quantity for each detonator (up to 1.0 fully bonding effect between the steel rebar and the slab. The TNT
g) as compared to the TNT charge weight, the effect of the detonator on explosive and air were modelled by Euler solid elements. The blast wave
the test results was neglected. Fig. 17 shows the steel rig supporting propagation and its interaction with the slab were activated by Con­
system in the contact explosion tests, in which the base steel frame was strained_Lagrange_In_Solid. The slab was engulfed within the air
fully fixed to the ground to remain the stabilization of the supporting domain, and the air domain was set as a non-reflecting boundary to
eliminate the blast wave reflection. It shall be noted that sufficiently

14
J. Liu et al. Engineering Structures 255 (2022) 113958

Fig. 22. UHPC-2.

Fig. 23. UHPC-3.

small Euler solid elements were always needed for the air to capture the distortions of UHPC under such high loading rates to avoid computa­
blast wave propagation and to transform the blast pressures into the tional overflow. On the other hand, the major concern of determining
slab, whereas large dimensions of the air domain would aggravate the the erosion criterion for UHPC was to prevent the massive deletion of
computational efficiency. After a mesh sensitivity test, ~11.25 mm elements and to ensure mass and momentum conservation. Therefore,
mesh size was adopted for the air to balance the accuracy and efficiency the erosion criterion needs to be determined with caution. In this study,
of the numerical modelling. as suggested by Li et al. [8], a principal tensile strain value of 0.1 was
taken as the erosion criterion for UHPCs.

4.4. Material models 4.4.2. TNT explosive, air and steel rebars
Mat_High_Explosive_Burn was adopted for TNT explosives. To
4.4.1. UHPC slabs reproduce the detonation process, a commonly used Jones–Wilkens–Lee
Following the systematic implementation procedures and guidelines (JWL) EOS model was also used.
as introduced in Section 3, model parameters of the K&C concrete Mat_Null coupled with EOS_Linear_Polynomial was used for the
model, as well as the DIF curves in compression and tension, for UHPCs ambient air.
under contact explosions were determined and listed in Table 4. Mat_Plastic_Kinematic was used for steel rebars. The strain rate effect
Numerical erosion algorithm, i.e. Mat_Add_Erosion, was adopted in on steel rebars was defined via the Cowper-Symonds model, which is
the numerical simulations of contact explosions to remove large element

15
J. Liu et al. Engineering Structures 255 (2022) 113958

Fig. 24. UHPC-4.

Fig. 25. UHPC-5.

derived as: maximum strength surface, and thereafter changes from 1 to 2 as the
( ε̇ )1/p concrete material moves from the maximum to the residual strength
σd
=1+ (15) surface [50]. Figs. 19 and 20 show the evolution of SDF versus axial
σy C
strain for UHPCs adopted in blast event 1 and blast event 2, respectively.
where σd denotes the dynamic yield stress; σy denotes the static yield It was observed that when the residual stress in the softening phase
stress; ε̇ is the strain rate; C and p are the coefficients in the Cowper- ranged from 0 to 30 percent of the peak value, the SDF value exhibited
Symonds model. The failure strain of steel rebars was set as 0.12. an extremely narrow range, which was from 1.98 to 2. Therefore, in the
Material models and the corresponding model parameters for the present study, the SDF value ranging from 1.98 to 2 was used to reflect
numerical modelling of air, TNT explosive and steel rebars are presented the crater and scabbing damage areas of the UHPC slabs subjected to
in Table 5. contact explosions.
Fig. 21(b) shows the numerical results of the local damage for the
Ductal@ UHPC slab without steel rebar reinforcement (UHPC-1) against
4.5. Comparison of the experimental and numerical results 1 kg contact explosion using the original K&C concrete model with the
simple input of material density, Poisson’s ratio, uniaxial compressive
In the K&C concrete model, scaled damage factor (SDF), which is and tensile strengths. As compared to the experimental results (Fig. 21

defined as δ = λ+λ m
, is always used to reflect the damage level. SDF varies (a)), the original K&C concrete model was too conservative to reproduce
from 0 to 1 as the concrete material migrates from the yield to the the local damage owing to the more brittle softening behaviour in

16
J. Liu et al. Engineering Structures 255 (2022) 113958

Fig. 26. UHPC-6.

Fig. 27. Typical local damage mode of concrete slabs induced by contact explosions.

Table 7
Local damage mode of Ductal@ UHPC slabs (2% steel fibres) under contact explosions.
Slab thickness (mm) TNT explosive (kg)

0.125 0.3 0.6 1.0 1.6 2.4

80 Crater only Perforation Perforation Perforation Perforation Perforation


100 Crater only Crater only Perforation Perforation Perforation Perforation
120 Crater only Crater only Crater and scabbing Perforation Perforation Perforation
150 Crater only Crater only Crater only Crater and scabbing Perforation Perforation
180 Crater only Crater only Crater only Crater only Crater and scabbing Crater and scabbing
200 Crater only Crater only Crater only Crater only Crater and scabbing Crater and scabbing
220 Crater only Crater only Crater only Crater only Crater only Crater and scabbing
250 Crater only Crater only Crater only Crater only Crater only Crater only

tension (Fig. 2(b)). Fig. 21(c) presents the numerical results of the local reinforcement and TNT explosive charge weight (0.1 kg and 1 kg) are
damage of UHPC-1 using the calibrated model, wherein the damage shown in Figs. 22-26, from which the numerical results in terms of the
mode, crater diameter and scabbing diameter agreed well with the damage mode, crater diameter and scabbing diameter exhibited a fair
experimental results (which can be seen in Table 6). More numerical match with the experimental results. This observation again demon­
simulation cases for generic UHPC slabs under contact explosions with strated that the calibrated method and the numerical models can be
the variables of slab thickness (100 mm-150 mm), steel rebar effectively adopted for the prediction of UHPC slabs under contact

17
J. Liu et al. Engineering Structures 255 (2022) 113958

UHPC slabs under contact explosions. In this study, to avoid unnecessary


agreements about the effect of steel rebar reinforcement on the perfor­
mance of UHPC slabs under contact explosions, UHPC-1 in blast event 1
was used as the reference UHPC slab for the parametric studies. Table 7
presents the parameters in the parametric studies of the contact explo­
sion simulation, where the slab thickness ranges from 80 mm to 250 mm
and the TNT explosive weight varies from 0.125 kg to 2.4 kg. In each
case, the cylindrical TNT explosive was in the shape of a cylinder with a
diameter to length ratio of 1.0, and the detonation position was in the
core of the explosive. The numerical local damage mode for each UHPC
slab is also shown in this figure.
Based on the numerical results from the parametric studies, the
relationship between the slab thickness and the cubic root of the TNT
explosive weight for the local damage mode of UHPC slabs under con­
tact explosions was obtained, which is shown in Fig. 28, and then a new
empirical equation was derived through curve fitting, as given by:
For the limit of crater only:

T/W 1/3 > 0.22 (17a)


Fig. 28. Limit of each local damage mode for UHPC slabs under con­
tact explosions. For the limit of crater and scabbing:

0.135 < T/W 1/3 ≤ 0.22 (17b)


explosions. As also noted, in some cases, the calibrated K&C concrete
model moderately underestimated the scabbing diameter of UHPC slabs, For the limit of perforation:
i.e. UHPC-4 and UHPC-6. This was probably attributed to two reasons.
T/W 1/3 ≤ 0.135 (17c)
Firstly, the proposed empirical equation to predict DIF values of UHPC
in tension at high strain rates might be overestimated. Secondly, the As also noted, the empirical equation was valid for 80 mm ≤ T ≤ 250
bonding effect between the steel rebars and the slab might not be perfect mm and 0.125 kg ≤ W ≤ 2.4 kg. Attempting to verify the derived
in real tests. empirical equation and broaden the range of the validity, more contact
explosion tests on UHPC slabs still needs further implementations.
5. Parametric studies
6. Concluding remarks
Empirical equations to predict the local damage of concrete struc­
tures under contact explosions are of great significance owing to the This study presents a systematic calibration of the K&C concrete
complexity of the phenomena. McVay [82] established an empirical model for UHPC in terms of three strength surfaces, equation of state,
equation to predict the local damage of concrete slabs under close-in shear dilatancy, damage evolution, and strain rate effect, with the
explosions based on 334 sets of field blast test data. In this equation, assistance of single element models. With the calibrated K&C concrete
two key parameters, i.e. scaled slab thickness (T/W1/3) and scaled model, numerical simulations of triaxial compression under a wide
standoff distance (R/W1/3), were included, where T and R are the range of confinement levels from 5 MPa to 60 MPa were fairly repro­
thickness of the slab and the standoff distance, respectively, and W is the duced. Further, local damage mode, crater and scabbing damage of
equivalent TNT mass. According to McVay’s empirical equation and UHPC slabs with the variables of UHPC mix design, slab thickness, steel
their own experimental results, Morishita et al. [83] classified three rebar reinforcement and TNT explosive charge weight were well simu­
categories of local damage modes for concrete slabs under contact ex­ lated to demonstrate the applicability of the calibrated K&C concrete
plosions, i.e. ‘crater only’, ‘crater and scabbing’ and ‘perforation’ (as model for UHPC under contact explosions. More importantly, the cali­
shown in Fig. 27), and then derived an empirical equation (the unit bration approach and the corresponding guidelines on the determina­
system is kg-m) for the limit of each local damage mode for NSC slabs, tion of key model parameters in this concrete model are rational to be
which is given by: generally applied for structural members constructed of generic UHPCs
For the limit of crater only: subjected to contact explosions even when lacking sufficient static and
dynamic test data. Using the calibrated and verified K&C concrete
T/W 1/3 > 0.36 (16a)
model, parametric numerical studies on the UHPC slabs under contact
For the limit of crater and scabbing: explosions were conducted and a new empirical equation to predict the
local damage mode was derived based on the numerical results.
0.2 < T/W 1/3 ≤ 0.36 (16b)
For the limit of perforation: CRediT authorship contribution statement

T/W 1/3 ≤ 0.2 (16c) Jian Liu: Conceptualization, Methodology, Software, Writing –
The validity of Morishita et al.’s equation for conventional steel rebar original draft, Writing – review & editing, Funding acquisition. Jun Li:
reinforced NSC slabs was also demonstrated by the previous studies Methodology, Supervision, Writing – review & editing. Jianguang
[9,84], whereas it underestimated the local damage mode for UHPC Fang: Methodology, Supervision, Writing – review & editing. Kai Liu:
slabs subjected to contact explosions [9]. Methodology, Project administration, Writing – review & editing. Yu
It has been acknowledged that empirical equations which are based Su: Methodology, Writing – review & editing. Chengqing Wu: Meth­
primarily on a large number of experimental data are costly to be ach­ odology, Resources, Project administration, Supervision, Writing – re­
ieved. To overcome this shortcoming, the calibrated and validated K&C view & editing.
concrete model could be adopted in the parametric studies with the
variables of slab thickness and TNT explosive weight, to propose a new
empirical equation to predict the limit of each local damage mode for

18
J. Liu et al. Engineering Structures 255 (2022) 113958

Declaration of Competing Interest [24] Lin X. Numerical simulation of blast responses of ultra-high performance fibre
reinforced concrete panels with strain-rate effect. Constr Build Mater 2018;176:
371–82.
The authors declare that they have no known competing financial [25] Xu S, Wu P, Wu C. Calibration of KCC concrete model for UHPC against low-
interests or personal relationships that could have appeared to influence velocity impact. Int J Impact Eng 2020;144:103648. https://doi.org/10.1016/j.
the work reported in this paper. ijimpeng.2020.103648.
[26] Zhang F, Shedbale AS, Zhong R, Poh LH, Zhang M-H. Ultra-high performance
concrete subjected to high-velocity projectile impact: implementation of K&C
Acknowledgements model with consideration of failure surfaces and dynamic increase factors. Int J
Impact Eng 2021;103907.
[27] Li J, Wu C, Hao H. Investigation of ultra-high performance concrete slab and
The authors greatly acknowledge the support from National Natural normal strength concrete slab under contact explosion. Eng Struct 2015;102:
Science Foundation of China under Grants No. 51908148 and No. 395–408.
51978186. [28] Willam KJ. Constitutive model for the triaxial behaviour of concrete. Proc Int Assoc
Bridge Structl Engrs. 1975;19:1–30.
[29] Liu J, Wu C, Li J, Liu Z, Xu S, Liu K, et al. Projectile impact resistance of fibre-
References reinforced geopolymer-based ultra-high performance concrete (G-UHPC). Constr
Build Mater 2021;290:123189. https://doi.org/10.1016/j.
[1] Li J, Hao H. Numerical study of concrete spall damage to blast loads. Int J Impact conbuildmat.2021.123189.
Eng 2014;68:41–55. [30] Wu C, Oehlers DJ, Rebentrost M, Leach J, Whittaker AS. Blast testing of ultra-high
[2] Liu J, Peng Y, Xu S, Yuan P, Qu K, Yu X, et al. Investigation of geopolymer-based performance fibre and FRP-retrofitted concrete slabs. Eng Struct 2009;31(9):
ultra-high performance concrete slabs against contact explosions. Constr Build 2060–9.
Mater 2022;315:125727. https://doi.org/10.1016/j.conbuildmat.2021.125727. [31] Pontiroli C, Erzar B, Buzaud E. Ultra high performance fiber reinforced concrete
[3] Graybeal BA. Material property characterization of ultra-high performance behavior under static and high velocity impact. Key Eng Mater: Trans Tech Publ
concrete. United States: Federal Highway Administration. Office of Infrastructure 2016: 171–8.
Research and Development; 2006. [32] Magallanes J, Wu Y, Morrill K, Crawford J. Feasibility studies of a plasticity-based
[4] Schmidt M, Fehling E. Ultra-high-performance concrete: research, development constitutive model for ultra high performance fiber-reinforced concrete for MABS.
and application in Europe. ACI Special Publ 2005;228:51–78. 2010;21:3–8.
[5] Su Yu, Wu C, Li J, Li Z-X, Li W. Development of novel ultra-high performance [33] Vankirk GH. Residual strength of a high-strength concrete subjected to triaxial pre-
concrete: From material to structure. Constr Build Mater 2017;135:517–28. stress. Mississippi State University; 2020.
[6] Wille K, Naaman AE, Parra-Montesinos GJ. Ultra-high performance concrete with [34] Wang Y-B, Liew J, Lee SC, Xiong D. Experimental study of ultra-high-strength
compressive strength exceeding 150 MPa (22 ksi): a simpler way. ACI Mater J concrete under triaxial compression. ACI Mater J 2016;113.
2011;108. [35] Smith J, Cusatis G, Pelessone D, Landis E, O’Daniel J, Baylot J. Discrete modeling
[7] Williams EM, Graham SS, Reed PA, Rushing TS. Laboratory characterization of of ultra-high-performance concrete with application to projectile penetration. Int J
Cor-Tuf concrete with and without steel fibers. Engineer research and development Impact Eng 2014;65:13–32.
center Vicksburg MS geotechnical and development center Vicksburg MS [36] Graybeal BA. Compressive behavior of ultra-high-performance fiber-reinforced
geotechnical and structures lab; 2009. concrete. ACI Mater J 2007;104:146.
[8] Li J, Wu C, Hao H. An experimental and numerical study of reinforced ultra-high [37] Graybeal BA, Stone B. Compression response of a rapid-strengthening ultra-high
performance concrete slabs under blast loads. Mater Des 2015;82:64–76. performance concrete formulation. United States: Federal Highway
[9] Li J, Wu C, Hao H, Wang Z, Su Yu. Experimental investigation of ultra-high Administration. Office of Infrastructure Research and Development; 2012.
performance concrete slabs under contact explosions. Int J Impact Eng 2016;93: [38] Alsalman A, Dang CN, Prinz GS, Hale WM. Evaluation of modulus of elasticity of
62–75. ultra-high performance concrete. Constr Build Mater 2017;153:918–28.
[10] Liu J, Li J, Fang J, Su Yu, Wu C. Ultra-high performance concrete targets against [39] Russell HG, Graybeal BA, Russell HG. Ultra-high performance concrete: a state-of-
high velocity projectile impact–a-state-of-the-art review. Int J Impact Eng 2022; the-art report for the bridge community. United States: Federal Highway
160:104080. https://doi.org/10.1016/j.ijimpeng.2021.104080. Administration. Office of Infrastructure Research and Development; 2013.
[11] Liu J, Wu C, Li J, Su Yu, Shao R, Liu Z, et al. Experimental and numerical study of [40] Yankelevsky DZ. Resistance of a concrete target to penetration of a rigid projectile-
reactive powder concrete reinforced with steel wire mesh against projectile revisited. Int J Impact Eng 2017;106:30–43.
penetration. Int J Impact Eng 2017;109:131–49. [41] Riedel W, Wicklein M, Thoma K. Shock properties of conventional and high
[12] Su Q, Wu H, Sun HS, Fang Q. Experimental and numerical studies on dynamic strength concrete: Experimental and mesomechanical analysis. Int J Impact Eng
behavior of reinforced UHPC panel under medium-range explosions. Int J Impact 2008;35(3):155–71.
Eng 2021;148:103761. https://doi.org/10.1016/j.ijimpeng.2020.103761. [42] Neel C. Compaction and spall of UHPC concrete under shock conditions. J Dyn
[13] Zhong R, Zhang F, Poh LH, Wang S, Le HTN, Zhang M-H. Assessing the Behav Mater 2018;4(4):505–28.
effectiveness of UHPFRC, FRHSC and ECC against high velocity projectile impact. [43] Wu Y, Crawford JE, Lan S, Magallanes JM. Validation studies for concrete
Cem Concr Compos 2021;120:104013. https://doi.org/10.1016/j. constitutive models with blast test data. In: 13th Int LS-DYNA® Users Conf, LSTC,
cemconcomp.2021.104013. Livermore, CA; 2014.
[14] Wu H, Fang Q, Chen XW, Gong ZM, Liu JZ. Projectile penetration of ultra-high [44] Shao R, Wu C, Su Yu, Liu Z, Liu J, Xu S. Numerical analysis on impact response of
performance cement based composites at 510–1320 m/s. Constr Build Mater 2015; ultra-high strength concrete protected with composite materials against steel
74:188–200. ogive-nosed projectile penetration. Compos Struct 2019;220:861–74.
[15] Crawford J, Wu Y, Choi H, Magallanes J, Lan S. Use and validation of the release III [45] Shao R, Wu C, Su Yu, Liu Z, Liu J, Chen G, et al. Experimental and numerical
K&C concrete material model in LS-DYNA. Karagozian & Case, Glendale; 2012. investigations of penetration resistance of ultra-high strength concrete protected
[16] Cui J, Hao H, Shi Y. Discussion on the suitability of concrete constitutive models with ceramic balls subjected to projectile impact. Ceram Int 2019;45(6):7961–75.
for high-rate response predictions of RC structures. Int J Impact Eng 2017;106: [46] Liu J, Wu C, Su Yu, Li J, Shao R, Chen G, et al. Experimental and numerical studies
202–16. of ultra-high performance concrete targets against high-velocity projectile impacts.
[17] Durant B, Oesterle M, Magallanes J, Headquarters N. Modeling reinforced concrete Eng Struct 2018;173:166–79.
protective construction for impact scenarios. 2018. [47] Kong X, Fang Q, Li Q, Wu H, Crawford JE. Modified K&C model for cratering and
[18] Li J, Zhang YX. Evaluation of constitutive models of hybrid-fibre engineered scabbing of concrete slabs under projectile impact. Int J Impact Eng 2017;108:
cementitious composites under dynamic loadings. Constr Build Mater 2012;30: 217–28.
149–60. [48] Xu J, Lu Y. Numerical modelling for reinforced concrete response to blast load:
[19] Xu M, Wille K. Calibration of K&C concrete model for UHPC in LS-DYNA. understanding the demands on material models. Special Publ 2016;306:3.1–3.22.
Advanced materials research: Trans Tech Publ 2015: 254–9. [49] Shasha W. Experimental and numerical studies on behavior of plain and fiber-
[20] Xu M, Wille K. Numerical investigation of the effects of pulse shaper, lateral inertia, reinforced high-strength concrete subjected to high strain rate loadings; 2011.
and friction on the calculated strain-rate sensitivity of UHP-FRC using a split [50] Hallquist JO. LS-DYNA® Keyword user’s manual volume II material models.
Hopkinson pressure bar. J Mater Civ Eng 2016;28(11):04016114. https://doi.org/ Livermore, California, USA; 2013.
10.1061/(ASCE)MT.1943-5533.0001580. [51] Su Yu, Li J, Wu C, Wu P, Li Z-X. Influences of nano-particles on dynamic strength of
[21] Ren L, Yu X, He Yu, Wang K, Yao H. Numerical investigation of lateral inertia effect ultra-high performance concrete. Compos B Eng 2016;91:595–609.
in dynamic impact testing of UHPC using a Split-Hopkinson pressure bar. Constr [52] Su Yu, Li J, Wu C, Wu P, Tao M, Li X. Mesoscale study of steel fibre-reinforced
Build Mater 2020;246:118483. https://doi.org/10.1016/j. ultra-high performance concrete under static and dynamic loads. Mater Des 2017;
conbuildmat.2020.118483. 116:340–51.
[22] Mao L, Barnett S, Begg D, Schleyer G, Wight G. Numerical simulation of ultra high [53] Malvar LJ, Crawford JE. Dynamic increase factors for concrete. Port hueneme CA:
performance fibre reinforced concrete panel subjected to blast loading. Int J Impact Naval Facilities Engineering Service Center; 1998.
Eng 2014;64:91–100. [54] Fujikake K, Senga T, Ueda N, Ohno T, Katagiri M. Effects of strain rate on tensile
[23] Zhang F, Wu C, Zhao X-L, Heidarpour A, Li Z. Experimental and numerical study of behavior of reactive powder concrete. J Adv Concr Technol 2006;4(1):79–84.
blast resistance of square CFDST columns with steel-fibre reinforced concrete. Eng [55] Larsen IL, Thorstensen RT. The influence of steel fibres on compressive and tensile
Struct 2017;149:50–63. strength of ultra high performance concrete: a review. Constr Build Mater 2020;
256:119459. https://doi.org/10.1016/j.conbuildmat.2020.119459.

19
J. Liu et al. Engineering Structures 255 (2022) 113958

[56] Park SH, Kim DJ, Kim SW. Investigating the impact resistance of ultra-high- [71] Gurusideswar S, Shukla A, Jonnalagadda KN, Nanthagopalan P. Tensile strength
performance fiber-reinforced concrete using an improved strain energy impact test and failure of ultra-high performance concrete (UHPC) composition over a wide
machine. Constr Build Mater 2016;125:145–59. range of strain rates. Constr Build Mater 2020;258:119642. https://doi.org/
[57] Thomas RJ, Sorensen AD. Review of strain rate effects for UHPC in tension. Constr 10.1016/j.conbuildmat.2020.119642.
Build Mater 2017;153:846–56. [72] Park JK, Kim S-W, Kim DJ. Matrix-strength-dependent strain-rate sensitivity of
[58] Yoo D-Y, Banthia N. Mechanical and structural behaviors of ultra-high- strain-hardening fiber-reinforced cementitious composites under tensile impact.
performance fiber-reinforced concrete subjected to impact and blast. Constr Build Compos Struct 2017;162:313–24.
Mater 2017;149:416–31. [73] Pyo S, El-Tawil S, Naaman AE. Direct tensile behavior of ultra high performance
[59] Hou X, Cao S, Zheng W, Rong Q, Li G. Experimental study on dynamic compressive fiber reinforced concrete (UHP-FRC) at high strain rates. Cem Concr Res 2016;88:
properties of fiber-reinforced reactive powder concrete at high strain rates. Eng 144–56.
Struct 2018;169:119–30. [74] Pyo S, Wille K, El-Tawil S, Naaman AE. Strain rate dependent properties of ultra
[60] Fujikake K, Senga T, Ueda N, Ohno T, Katagiri M. Nonlinear analysis for reactive high performance fiber reinforced concrete (UHP-FRC) under tension. Cem Concr
powder concrete beams under rapid flexural loadings. J Adv Concr Technol 2006;4 Compos 2015;56:15–24.
(1):85–97. [75] Tai Y-S, El-Tawil S, Chung T-H. Performance of deformed steel fibers embedded in
[61] Al-Salloum Y, Almusallam T, Ibrahim SM, Abbas H, Alsayed S. Rate dependent ultra-high performance concrete subjected to various pullout rates. Cem Concr Res
behavior and modeling of concrete based on SHPB experiments. Cem Concr 2016;89:1–13.
Compos 2015;55:34–44. [76] Tran NT, Kim DJ. Synergistic response of blending fibers in ultra-high-performance
[62] Hassan M, Wille K. Experimental impact analysis on ultra-high performance concrete under high rate tensile loads. Cem Concr Compos 2017;78:132–45.
concrete (UHPC) for achieving stress equilibrium (SE) and constant strain rate [77] Tran NT, Tran TK, Kim DJ. High rate response of ultra-high-performance fiber-
(CSR) in Split Hopkinson pressure bar (SHPB) using pulse shaping technique. reinforced concretes under direct tension. Cem Concr Res 2015;69:72–87.
Constr Build Mater 2017;144:747–57. [78] Cadoni E, Forni D. Experimental analysis of the UHPFRCs behavior under tension
[63] Huang YJ, Yang ZJ, Chen XW, Liu GH. Monte Carlo simulations of meso-scale at high stress rate. Eur Phys J Special Top 2016;225(2):253–64.
dynamic compressive behavior of concrete based on X-ray computed tomography [79] Noldgen M, Riedel W, Thomas K, Fehling E. Properties of ultra high performance
images. Int J Impact Eng 2016;97:102–15. concrete (UHPC) in tension at high strain rates. In: Proceedings of the VIII
[64] Othman H, Marzouk H, Sherif M. Effects of variations in compressive strength and international conference on fracture mechanics of concrete and concrete structures
fibre content on dynamic properties of ultra-high performance fibre-reinforced (FraMCoS-8); 2013.
concrete. Constr Build Mater 2019;195:547–56. [80] Rong Z, Sun W. Experimental and numerical investigation on the dynamic tensile
[65] Ren GM, Wu H, Fang Q, Liu JZ. Effects of steel fiber content and type on dynamic behavior of ultra-high performance cement based composites. Constr Build Mater
compressive mechanical properties of UHPCC. Constr Build Mater 2018;164: 2012;31:168–73.
29–43. [81] Janota O, Foglar M. Experimental and numerical evaluation of UHPFRC slabs
[66] Su Yu, Li J, Wu C, Wu P, Li Z-X. Effects of steel fibres on dynamic strength of UHPC. subjected to contact and close-in explosion. Solid State Phenomena: Trans Tech
Constr Build Mater 2016;114:708–18. Publ 2020: 180–5.
[67] Tai YS. Uniaxial compression tests at various loading rates for reactive powder [82] McVay M. Spall damage of concrete structures. No. WES/TR/SL-88-22. 1988.
concrete. Theor Appl Fract Mech 2009;52(1):14–21. [83] Morishita M, Tanaka H, Ando T, Hagiya H. Effects of concrete strength and
[68] Wang S, Zhang M-H, Quek ST. Mechanical behavior of fiber-reinforced high- reinforcing clear distance on the damage of reinforced concrete slabs subjected to
strength concrete subjected to high strain-rate compressive loading. Constr Build contact detonations. Concr Res Technol 2004;15(2):89–98.
Mater 2012;31:1–11. [84] Yu X, Zhou B, Hu F, Zhang Yi, Xu X, Fan C, et al. Experimental investigation of
[69] Wu Z, Shi C, He W, Wang D. Static and dynamic compressive properties of ultra- basalt fiber-reinforced polymer (BFRP) bar reinforced concrete slabs under contact
high performance concrete (UHPC) with hybrid steel fiber reinforcements. Cem explosions. Int J Impact Eng 2020;144:103632. https://doi.org/10.1016/j.
Concr Compos 2017;79:148–57. ijimpeng.2020.103632.
[70] Caverzan A, Cadoni E, Di Prisco M. Dynamic behavior of HPFRCC at high strain
rate: the fiber role. In: High performance fiber reinforced cement composites 6.
Springer; 2012. p. 339–46.

20

You might also like