Professional Documents
Culture Documents
FrameworkofDynamicAnalysisofTransportPackages-preprint - Demonstração Dos Parâmetros Do CDP
FrameworkofDynamicAnalysisofTransportPackages-preprint - Demonstração Dos Parâmetros Do CDP
________________________________________________________
aEngineering Mechanics Department, Ontario Power Generation Inc., 889 Brock Rd,
Pickering, Ontario, Canada L1W 3J2
bNuclear Waste Design Engineering, Ontario Power Generation Inc., 1340 Pickering
*Corresponding author
________________________________________________________
ABSTRACT
1
Keywords: Transport package, True stress, True strain, Triaxiality, Strain rate,
Constitutive model
________________________________________________________
1. Introduction
2
flow rules are established to relate the plastic defamation increments to stresses and
perhaps other quantities via the plastic potential function. Material parameters in the
plastic potential function should be calibrated based on the overserved material
behavior in elementary stress states.
In conventional designs under normal design loads, design inputs to a structural
system are generally external loads or forces. The output of a structural analysis is
generally the stresses that are checked against permissible stress criteria. Since the
plastic deformation is small up to the permissible stress, the engineering stress and
strain are generally used because the difference between the engineering and true
stresses or strains is negligibly small.
In large structural impact scenarios, kinetic energy is normally the input to a structural
system. Large deformation in some parts of the system is required to transfer the
kinetic energy into high strain energy. The acceptance criteria for a containment
boundary component such as a DSC liner is the permissible fracture strain to prevent
leakage of radioactive materials. In this case, true stress and strain should be used in
the dynamic analysis to quantify large deformation. For steel materials, the true stress is
increasingly higher than the engineering stress with increasing strain and the true
fracture strain is larger than the engineering fracture strain.
Static characteristics of materials are much different from the dynamic characteristics,
especially when materials are subjected to short duration impact with high strain rates.
Based on the investigation of a transport package under the 9-meter drop scenario
presented in this paper, the maximum strain rate of the impact limiter steel shell can
reach approximately 300/s, and that of the impact limiter polyurethane foam 500/s.
Consequently, the yield strengths of the impact limiter steel and foam can increase by
more than 50%. This increase is considered as very significant.
There is no steadfast rule to judge whether or not using engineering stress-strain
relationship and ignoring strain rate effect could lead to conservative result. On the one
hand, ignoring strain rate effect reserves conservatism of the material strengths. On the
other hand, ignoring the strain rate effect softens and weakens the actual mechanical
behavior of the materials that are directly in contact with targets, which overestimate of
the deformation and strain energy of the energy absorbent materials such as impact
limiters and target objects, and consequently underestimate the impact load transferred
to the more critical DSC. Reduction of strain energy in some parts of a structural
assembly increases the strain energy in the remaining parts due to the conservation of
energy. Using engineering stress-strain relationship and ignoring strain rate effect can
significantly alter the distribution of the strain energy and, in turn, result in unrealistic
strain and stress of the materials. To reveal the realistic effect of accident drop and
impact scenarios on the safety of the DSC as a containment boundary, the true stress
and strain with realistic strain rate effect should be incorporated in the dynamic analysis.
3
Lo Frano et al (2014) performed a numerical analysis of a type IP-2 package under a
9-meter flat end drop scenario. The mechanical behavior of the packaging (steel shells
encasing concrete) was assumed to be elastic-perfectly plastic, and that of the steel
materials of the container a piece wise linear stress-strain curve without description of
the data used. No details of strain rate effect and fracture criteria were provided. Qiao et
al (2012) presented a finite element model for evaluation of the performance of
transport and storage casks. The analysis model was validated only by comparing the
computed local decelerations with the recorded deceleration from a test. However, no
details of stress-strain relationships, strain rate effect, material constitutive models, and
fracture criteria were provided. Sprung et al (2000) developed an in-house finite element
program PRONTO-3D for nonlinear transient analysis of generic spent fuel casks.
Idealized Ramberg-Osgood stress-strain relationship and bi-linear elastic-plastic
material model are used to represent mechanical behavior of steels and energy
absorbent materials. No details of strain rate effect and triaxial constitutive models were
provided. The EPRI reports 1009929 and 1011817 (EPRI, 2005a and 2005b)
incorporated true stress and strain. Fracture criteria were provided, but not associated
with stress triaxiality. No detail of strain rate effect was provided.
In general, materials used in a radioactive material container and transport package
include materials of the impact limiter steel shell, crushable foam, impact limiter
connection components, DSC steel liners/flanges, weld and/or bolt metals, concrete,
reinforcing steel, fuel modules, and fuel bundles. For these materials, establishments of
material constitutive models, uniaxial true stress-strain relationships, effect of strain
rate, and fracture and damage criteria incorporated in the dynamic analysis framework
are discussed, respectively, in the following sections.
In this paper, stress and modulus of elasticity are in the metric unit, MPa, and density
in kg/m3. Compressive stresses assume negative values, unless noted otherwise.
Stress and strain refers to true stress and strain, respectively, unless indicated
otherwise.
4
where 𝜎 and 𝜀 are the true stress and strain, respectively; 𝑠 and 𝑒 are the engineering
stress and strain, respectively; and 𝜐(𝑒) is the Poisson’s ratio which is a function of
strain. Eqs. (1) are valid for brittle materials under tension and compression, ductile
materials under tension up to the peak engineering stress, and ductile materials under
compression.
For a uniaxial tensile test of a specimen made of ductile materials, initiation of cross-
section necking occurs at the peak tensile load or peak engineering stress. Until the
neck forms, the deformation is essentially uniform throughout the gauge length of the
specimen. After necking, all subsequent deformation takes place in the necking region,
and the engineering strain varies with the change of the gauge length of the specimen.
In the plastic-flow regime following yield, ductile materials flow with negligible change in
volume. The true strain after necking is then measured as
𝐴
𝜀 = 𝑙𝑛 ( 𝐴0 ) (2)
where 𝐴0 and 𝐴 are the initial and current minimum cross-sectional area in the necking
region.
After necking occurs, the specimen necking increases with increasing extension. The
state of stress in the necking region always becomes increasingly non-uniaxial. The true
stress is lower than the average value over the minimum cross-sectional area due to the
triaxial stress state in the necking region. The true stress can be calculated based on
the geometry of the necking as discussed by Yan et al (2018). Alternatively, an
empirical form is proposed based on the work by Chen (2010) to correct the average
stress after the initiation of necking for the true stress:
𝐹 𝐴 1+𝑒
𝑢
𝜎 = 𝛾𝑛 (𝐴) (1 + √𝐴 ) = 𝑠 ∙ 𝛾𝑛 exp(𝜀) (1 + √exp(𝜀) ) (3)
𝑝
𝐸𝑇
𝜀̇𝑝𝑙 = (1 − ) 𝜀̇ (4)
𝐸
in which 𝐸 is the static modulus of elasticity and 𝐸𝑇 the static tangent modulus. This
equation is used to convert experimental data that uses the total strain rate into analysis
inputs in terms of the plastic strain rate as required in analysis software.
5
The most promising crushable foam material model is the model with volumetric
hardening which assumes that the evolution of the yield surface is controlled by the
volumetric compacting. The yield surface function adopted in ABAQUS (2017) is
defined in the meridional plane as
𝑝𝑙 𝑝𝑙 2
𝐹(𝜎𝑖𝑗 ) = √𝑞(𝜎𝑖𝑗 )2 + 𝛼 2 (𝑝(𝜎𝑖𝑗 ) − 𝑝0 (𝜀𝑣𝑜𝑙 )) − 𝐵(𝜀𝑣𝑜𝑙 )=0 (5)
where
1
𝑝(𝜎𝑖𝑗 ) = − 3 𝑡𝑟𝑎𝑐𝑒(𝜎𝑖𝑗 ) , the hydrostatic pressure being the function of the stress tensor 𝜎𝑖𝑗 ;
3
𝑞(𝜎𝑖𝑗 ) = √2 𝑆𝑖𝑗 𝑆𝑖𝑗 , the Mises stress being the function of the deviatoric stress tensor 𝑆𝑖𝑗 =
𝜎𝑖𝑗 + 𝛿𝑖𝑗 𝑝(𝜎𝑖𝑗 );
𝑝𝑙
𝑝𝑙 𝑝𝑐 (𝜀𝑣𝑜𝑙 )−𝑝𝑡
𝑝0 (𝜀𝑣𝑜𝑙 ) = , the centre of the yield surface on the p-axis;
2
𝑝𝑙
𝑝𝑐 (𝜖𝑣𝑜𝑙 )
is the evolved hydrostatic yield stress under compaction as a function of the
𝑝𝑙
volumetric compacting plastic strain 𝜀𝑣𝑜𝑙 ;
𝑝𝑡 is the yield stress in hydrostatic tension;
𝑝𝑙
𝑝𝑙 𝑝𝑐 (𝜀𝑣𝑜𝑙 )+𝑝𝑡
𝐵(𝜀𝑣𝑜𝑙 )=𝛼 , a yield surface evolution parameter with the hardening variable
2
being the volumetric plastic strain;
3𝑘𝑐
𝛼= )(3−𝑘 )
, a material constant that defines the shape of the yield surface;
√(3𝑘𝑡 +𝑘𝑐 𝑐
𝑘𝑐 = 𝜎𝑐0 /𝑝𝑐0 and 𝑘𝑡 = 𝜎𝑡0 /𝑝𝑐0;
𝜎𝑐0 is the initial uniaxial compressive strength;
𝜎𝑡0 the initial uniaxial tensile strength; and
𝜎𝑝0 the initial hydrostatic compressive strength.
The plastic flow is non-associative with the plastic flow potential being defined as
9
𝐺(𝜎𝑖𝑗 ) = √𝑞(𝜎𝑖𝑗 )2 + 2 𝑝(𝜎𝑖𝑗 )2 (6)
6
in the uniaxial compression, where 𝜐𝑓𝑐 is the plastic Poisson’s ratio of crushable foams.
It is noted that 𝜐𝑓𝑐 is conventionally assumed to be zero for crushable foams with the
volumetric hardening model. With a zero plastic Poisson’s ratio, Eq. (8) reduces to the
𝑝𝑙 𝑝𝑙
equation 𝜀𝑣𝑜𝑙 = 𝜀𝑎𝑥𝑖𝑎𝑙 , which may have been used in the built-in constitutive models in
analysis software. To use this built-in model for foams of non-zero plastic Poisson’s
ratio, the hardening rule should be specified as a compressive stress - volumetric plastic
𝑝𝑙
strain relationship, 𝜎𝑐 (𝜀𝑣𝑜𝑙 ) , which can be obtained from modifying the uniaxial
𝑝𝑙
relationship 𝜎𝑐 (𝜀𝑎𝑥𝑖𝑎𝑙 ) by scaling the axial plastic strain using Eq. (8).
where
𝑝𝑙
is the initial value of 𝑝0 (𝜀𝑣𝑜𝑙 ), i.e., the initial center of the yield surface. It is noted that
the uniaxial compressive strength, 𝜎𝑐0 , is a positive value.
exp(𝛿𝑒)−1
𝑠 = 𝑎 exp(𝛽𝑒)+𝑏 + exp(𝑐)(𝑒𝑥𝑝(𝛾𝑒) − 1) (11)
where 𝑠 and 𝑒 are uniaxial engineering compressive stress and strain, respectively; a, b,
c, δ, β and γ are model parameters related to foam density. If the uniaxial compressive
stress-strain relationship is established using engineering strain as the common
representation of the laboratory test results, it should be converted into true stress-strain
relationship using Eqs. (1) for large deformation analysis.
7
3.4 Compressive Poisson’s ratio of polyurethane foams
To convert an engineering stress-strain relationship to a true stress-strain relationship,
the Poisson’s ratio used in Eqs. (1) should be given. For low-density foams, Kraus et al
(2013) proposed a general form of the Poisson’s ratio as
𝜈𝑓0
𝑓𝑜𝑟 𝑒<0
1+exp(𝛽𝑣 (𝜀𝑜𝑐 −𝑒))
𝜈𝑘 (𝑒) = { 𝜈 (12)
𝜈𝑓0 + 1+exp(𝛼𝑓𝑒(𝜀 −𝑒)) 𝑓𝑜𝑟 𝑒≥0
𝑣 𝑜𝑡
where 𝑒 is the engineering strain. Under large uniaxial compressive strain, the Poisson’s
ratio approaches to zero. Kraus et al (2013) proposed that 𝜈𝑓0 = 0.3, 𝜈𝑓𝑒 = 0.2, 𝛼𝑣 = 20,
𝛽𝑣 = 14, 𝜀𝑜𝑡 = 0.25, and 𝜀𝑜𝑐 = −0.28 to fit the experimental results of the low-density
conventional foams by Choi and Lakes (1992).
For high-density foams, the Poisson’s ratio may not approach to zero under high
compressive strain. For example, the experimental result of the Poisson’s ratio of the
LAST-A-FOAM® FR-3718 provided by General Plastics Manufacturing Company is
shown in Fig. 2. The density of the foam is 𝜌 = 288 𝑘𝑔/𝑚3 . When the foam is densified,
the Poisson’s ratio reaches the plastic Poisson’s ratio. For high-density foams, the
Poisson’s ratio is proposed to take the form
𝑒 𝑚𝑐 𝜈 −𝜈
𝜈𝑓𝑐 + (𝜈𝑓0 − 𝜈𝑓𝑐 )exp[(𝑒 ) ln(𝜈 𝑐𝑖 −𝜈𝑓𝑐 )] 𝑓𝑜𝑟 𝑒<0
𝑐𝑖 𝑓0 𝑓𝑐
𝜈𝑓 (𝑒) = { 𝑚𝑡 𝜈 −𝜈
(13)
𝑒
𝜈𝑓𝑡 + (𝜈𝑓0 − 𝜈𝑓𝑡 )exp[(𝑒 ) ln(𝜈 𝑡𝑖 −𝜈𝑓𝑡 )] 𝑓𝑜𝑟 𝑒≥0
𝑡𝑖 𝑓0 𝑓𝑡
where (𝑒𝑐𝑖 , 𝜈𝑐𝑖 ) and (𝑒𝑡𝑖 , 𝜈𝑡𝑖 ) are the inflection points on the compression and tension
branches of the Poisson’s ratio - engineering strain relation curve, respectively; 𝑚𝑐 and
𝑚𝑡 the material parameters controlling the curvature of the compression and tension
braches, respectively; 𝜈𝑓𝑐 and 𝜈𝑓𝑡 the plastic compressive and tensile Poisson’s ratios at
high strain, respectively; and 𝜈𝑓0 the elastic Poisson’s ratio. For the LAST-A-FOAM® FR-
3718, the material parameters are taken as 𝑒𝑐𝑖 = 0.12, 𝜈𝑐𝑖 = 0.30 , 𝜈𝑓0 = 0.34, 𝜈𝑓𝑐 =
0.24, and 𝑚𝑐 = 3 to fit the experimental results as shown in Fig. 1.
0.36 Test
0.34 This paper
0.32
Poisson's ratio
0.3
0.28
0.26
0.24
0.22
0.2
0 0.1 0.2 0.3 0.4
Engineering strain
Fig. 1. Poisson’s ratio of polyurethane foam LAST-A-FOAM® FR-3718
8
3.5 True stress-strain relationship of polyurethane foams
For the LAST-A-FOAM® FR-3718, the true stress-strain relationship can be obtained
by converting the engineering stress-strain using Eqs. (1) with the proposed Poisson’s
ratio in Eq. (13). The engineering and true stress vs plastic strain relationships are
depicted in Fig. 2. It can be found from the figure that the foam in the densification
regime with true stress-plastic strain relationship behaves softer than that indicated by
the engineering stress-plastic strain relationship.
In Fig. 3, the relationship of the true stress vs the volumetric plastic strain or the
modified axial plastic strain multiplied by (1 − 2𝜐𝑓𝑐 ) in accordance with Eq. (8) is also
shown. This relationship is required to be used as the hardening rule of the compressive
strength of the foams with volumetric hardening.
35 Engineering stress-axial plastic strain (Test)
30 True stress-axial plastic strain
True stress-volumetric plastic strain
25
20
Stress (MPa)
15
10
0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 1.3
Plastic Strain
Fig. 2. Stress-strain relationships of polyurethane foam LAST-A-FOAM® FR-3718
𝑝𝑙
It is noted that 𝜀̇ = 𝜀̇𝑎𝑥𝑖𝑎𝑙
𝑝𝑙
for the uniaxial compression and tension.
𝑝𝑙 𝑝𝑙
The strain-rate-dependent size of the yield surface, 𝐵̃ (𝜀 , 𝜀̇ ) , can be calibrated by
𝑝𝑙 𝑝𝑙
𝑝𝑙 𝐵̃(𝜀 ,𝜀̇ ) 𝑝𝑙 3𝑘 +𝑟(𝜀 ̇ 𝑝𝑙 )[𝑘 +𝑘 (3−𝑘 )]
𝑅 (𝜀̇ ) = 𝑝𝑙 = 1 + (𝑟(𝜀̇ ) − 1) 𝑡 𝑐 𝑡
𝑝𝑙
𝑐
(15)
𝐵(𝜀 ) (1+𝑘𝑡 )(3𝑘𝑡 +𝑟(𝜀̇ )𝑘𝑐 )
9
In this paper, the DIF of the yield stress in the plateau regime of the uniaxial
compression curve is proposed in the following form similar to that presented by Song
et al (2009):
𝑝𝑙
𝑝𝑙 𝑎𝑥𝑖𝑎𝑙 𝜀̇
𝑟(𝜀̇𝑎𝑥𝑖𝑎𝑙 ) = 1 + 𝑟𝑐 𝑙𝑜𝑔( 10−3
) ≥ 1.0 (16)
For polyurethane foams of the high-density approximately 𝜌 = 300 𝑘𝑔/𝑚3 , the model
parameter in Eq. (16) can be taken as 𝑟𝑐 = 0.12 to fit the experimental results of the
plateau stress by Kasparek et al (2012), Lu (2014), Song et al (2009), and Maji et al
(1995) (perpendicular to rise), as illustrated in Fig. 3. It was demonstrated by Kasparek
et al (2011) that the DIF of the yield stress in the stress plateau regime can be applied
to the densification regime of compaction.
1.9
This paper
1.5
1.4
1.3
1.2
1.1
1
0.001 0.01 0.1 1 10 100 1000 10000
Strain Rate
Fig. 3. Dynamic increase factor of stress of polyurethane foam of 𝜌 = 300 𝑘𝑔/𝑚3
10
crack strain; 5) dilation angle in meridional plane; 6) ratio of equibiaxial compressive
strength to uniaxial compressive strength; and 7) ratio of the second stress invariant on
the tensile meridian to that on the compressive meridian.
where
3(1−𝐾 )
𝛾 = 2𝐾 −1𝑐 , 0.5 ≤ 𝛾 ≤ 1.0 ;
𝑐
𝐾𝑐 is the ratio of the second stress invariant on the tensile meridian to that on the
compressive meridian;
(𝑓 ′ /𝑓′ )−1
𝛼 = 2(𝑓𝑏′ /𝑓𝑐 ′ )−1 , 0 ≤ 𝛼 ≤ 0.5;
𝑏 𝑐
𝑓𝑏′ /𝑓𝑐′ is the ratio of initial equibiaxial compressive strength to initial uniaxial compressive
strength;
𝑝𝑙
̅𝑐 (𝜀̃𝑐 )
𝜎
𝛽(𝜀̃𝑝𝑙 ) = 𝑝𝑙 (1 − 𝛼) − (1 + 𝛼) ;
̅𝑡 (𝜀̃𝑡 )
𝜎
𝑝𝑙
𝜎𝑐 (𝜀̃𝑐 )
𝜎̅𝑐 (𝜀̃𝑐𝑝𝑙 ) = 𝑝𝑙 , the effective compressive stress corresponding to the compressive
1−𝑑𝑐 (𝜀̃𝑐 )
stress 𝜎𝑐 (𝜀̃𝑐𝑝𝑙 ) being determined based on the compressive equivalent plastic strain, 𝜀̃𝑐𝑝𝑙 ;
𝑝𝑙
𝑝𝑙 𝜎𝑡 (𝜀̃𝑡 ) 𝑝𝑙
𝜎̅𝑡 (𝜀̃𝑡 ) = 𝑝𝑙 , the effective tensile stresses corresponding to the tensile stress 𝜎𝑡 (𝜀̃𝑡 )
1−𝑑 (𝜀̃ ) 𝑡 𝑡
Lim et al (2016) calibrated from experimental results the following equations for the
ratio, 𝑓𝑏′ /𝑓𝑐′ , and the parameter, 𝐾𝑐 :
𝑓𝑏′ −0.09
= 1.57𝑓𝑐′ (18)
𝑓𝑐′
−0.025
𝐾𝑐 = 0.71𝑓𝑐′ (19)
where 𝑠𝑐 and 𝑒𝑐 are uniaxial engineering compressive stress and strain, respectively; 𝜀𝑐𝑝 is
the strain at the peak engineering stress, 𝑓𝑐′ ;
𝑓𝑐′
𝑛1 = (1.02 − 1.17 𝐸 )−0.74 𝑓𝑜𝑟 𝑒𝑐 ≤ 𝜀𝑐𝑝
𝑛={ 𝑐 𝜀𝑐𝑝 (21)
𝑛1 + 𝑎 + 28𝑏 𝑓𝑜𝑟 𝑒𝑐 > 𝜀𝑐𝑝
−911
in which 𝑎 = 3.5(12.4 − 0.0166𝑓𝑐′ )−0.46 and 𝑏 = 0.83exp( ) ; and
𝑓𝑐′
in which 𝛾𝑐 is the density of concrete in kg/m3. The strain at the peak compressive
stress, 𝜀𝑐𝑝 , can be considered as either engineering or true strain due to its small vale.
Based on the study by Samani and Attard (2012) for normal-weight concrete, the strain
at the peak compressive stress is proposed as
0.75
𝑓𝑐′ 𝛾
𝜀𝑐𝑝 = 𝜇𝑎 𝑐
(2300 )𝜑 (23)
𝐸𝑐
where 𝜇𝑎 = 4.26 for crushed aggregate or 𝜇𝑎 = 3.78 for rounded aggregate (Samani
and Attard, 2012); and the exponent, 𝜑 , is to incorporate heavyweight concrete. Based
on the limited experimental results performed in author’s organization for the
heavyweight concrete of density 3700 kg/m3, the exponent may be taken as 𝜑 = 0.3.
12
𝜎𝑡 𝑤 3 𝑤 𝑤
= [1 + (𝑐1 𝑤 ) ] exp (−𝑐2 𝑤 ) − 𝑤 (1 + 𝑐1 3 )exp(−𝑐2 ) (24)
𝑓𝑡 𝑐 𝑐 𝑐
where 𝑐1 = 3 ; 𝑐2 = 6.93 ; 𝜎𝑡 is the uniaxial tensile stress; 𝑓𝑡 the initial uniaxial tensile
strength of concrete; 𝑤 the crack opening displacement; and 𝑤𝑐 the critical crack
opening displacement with free of stress determined by
𝐺𝐹
𝑤𝑐 = 5.136 (25)
𝑓𝑡
in which 𝐺𝐹 is the fracture energy that equals the area under the entire stress-crack
opening curve. By fitting the experimental results, Bazant and Becq-Giraudon (2002)
proposed the fracture energy of concrete as
𝜎 0.46 𝑑 0.22
𝑐𝑢
𝐺𝐹 = 𝛼𝑎 (0.051) 𝑎
(1 + 11.27) (𝑤/𝑐)−0.30 (26)
where 𝑑𝑎 is maximum aggregate size, mm; 𝑤/𝑐 the water cement ratio by weight; and
𝛼𝑎 = 1.0 for rounded aggregate or 𝛼𝑎 = 1.44 for crushed aggregate.
In lieu of an exponential cohesion law by Eq (24), a linear descending cohesion law
with a constant slope (de Oliveira e Sousa and Gettu, 2006) may be assumed:
𝜎𝑡 𝑤
𝑓𝑡
=1−𝑤 (27)
𝑒𝑐
13
𝑝𝑙
By definition the compressive damage index, 𝑑𝑐 (𝜀̃𝑐 ) , indicating the degradation of
the unloading stiffness as illustrated in Fig. 4, can be expressed as 𝑑𝑐 (𝜀̃𝑐𝑝𝑙 ) = 1 −
𝐸𝑐𝑢𝑛 /𝐸𝑐 . With substitution of Eq. (26) into this expression, the compressive damage
index can be derived as follow in terms of inelastic compressive strain 𝜀̃𝑐𝑖𝑛 = 𝜀𝑐 − 𝜎𝑐 /𝐸𝑐 :
|𝜀̃𝑐𝑖𝑛 |
𝑑𝑐 (𝜀̃𝑐𝑖𝑛 ) = 𝜎 ≤ 0.8 (30)
|𝜀̃𝑐𝑖𝑛 + 𝑐 |+0.57𝜀𝑐𝑝
𝐸𝑐
where a limit of 0.8 is imposed to maintain a minimum stiffness of 0.2𝐸𝑐 . The inelastic
compressive strain can be converted to the equivalent plastic strain using 𝜀̃𝑐𝑝𝑙 = 𝜀̃𝑐𝑖𝑛 −
𝑑𝑐 (𝜀̃𝑐𝑖𝑛 ) 𝜎𝑐
. Eq. (30) is required as input to the concrete damage plasticity model.
1−𝑑𝑐 (𝜀̃𝑐𝑖𝑛 ) 𝐸𝑐
𝜆𝐸𝑐 𝜀̃𝑡𝑐𝑟
𝑑𝑡 (𝜀̃𝑡𝑐𝑟 ) = 𝜆𝐸 ̃ 𝑡𝑐𝑟 +𝜎𝑡
≤ 0.8 (32)
𝑐𝜀
where the damage index is limited to 0.8 to maintain a minimum stiffness of 0.2𝐸𝑐 . The
𝑑 (𝜀̃ 𝑖𝑛 ) 𝜎
crack strain can be converted to the equivalent plastic strain via 𝜀̃𝑡𝑝𝑙 = 𝜀̃𝑡𝑖𝑛 − 1−𝑑𝑡 (𝜀𝑐̃ 𝑖𝑛) 𝐸𝑡 .
𝑡 𝑐 𝑐
Eq. (32) is required as input to the concrete damage plasticity model.
14
taken as 1.0 as given in the CEB-FIB Model Code 2010 (FIB, 2012), i.e., the splitting
tensile strength is approximately equal to the direct tensile strength.
For concrete without accelerating admixtures and made with crushable aggregates,
the tensile strength in MPa is provided by Arioglu et al (2006) by fitting experimental
results:
0.630
𝑓𝑡 = 𝑓𝑠𝑝 = 0.387𝑓𝑐′ (33)
2
𝐺(𝜎𝑖𝑗 ) = √(𝜖𝑓𝑡 tan 𝜓)2 + 𝑞(𝜎𝑖𝑗 ) − 𝑝(𝜎𝑖𝑗 ) tan 𝜓 (34)
where 𝜓 is the dilation angle measured in the meridional plane at high hydrostatic
pressure. 𝜖 is the eccentricity that defines the rate at which the hyperbolic plastic
potential function approaches the asymptote.
Malm (2006) and Ferrotto et al (2018) studied the effect of the dilation angle in the
range between 10𝑜 and 56.3𝑜 on the flexural response of reinforced concrete
specimens. The results indicate that the dilation angle strongly influences the flexural
strength and ductility of the specimens. The larger the dilation angle, the more ductile of
the specimen is, and the higher strength the specimen has. For normal strength
concrete, different authors in the past assumed different values of the dilation angle
between 13° and 49° independently. Lee and Fenves (1998) specified it as 31° for the
verification of the model; Nana et al (2017) suggested 37° based on good agreement
with the shear test results; Jankowiak and Lodygowski (2005) calibrated it to be 38°;
and Malm (2009) showed that a dilation angle between 30° and 40° for normal strength
concrete can have the best agreement with the experimental response.
In this paper, the dilation angle, 𝜓 , is proposed to be
6(𝜈𝑐0 −𝜈𝑐𝑒 )
𝜓 = 𝑡𝑎𝑛−1 [3𝐸𝑐𝜀𝑐𝑝 ] (35)
+2(𝜈𝑐0 −𝜈𝑐𝑒 )−3
𝑓′𝑐
where 𝜀𝑐𝑝 is the strain at the peak compressive engineering stress as given by Eq. (23);
𝜈𝑐𝑒 the elastic (initial) Poisson’s ratio; and 𝜈𝑐0 the Poisson’s ration at the peak
compressive engineering stress. The dilation angle may be limited to the concrete
material angle, 56.3𝑜 (Malm, 2006), but this cap value corresponds to a very high
compressive strength (greater than 140 MPa) which may not be justified due to scarcity
of the experimental results at this high strength.
Derivation of the plastic flow potential with respect to the hydrostatic pressure results
in the inclination of the plastic flow potential curve:
𝑑𝑞 𝑓 2
= tan(𝜓)√1 + ( 𝑞𝑡 𝜖tan(𝜓)) (36)
𝑑𝑝
15
Jiang and Wu (2012) derived a relationship between this inclination and plastic
volumetric and shear strain increments for the case of axial compression with uniform
lateral confinement. Since the inclination of the asymptote is constant, this relationship
can be extended as follow for the total inelastic strains under high hydrostatic pressure:
𝑖𝑛 𝑖𝑛 𝑖𝑛
𝑑𝑞 3𝑑𝜀𝑣𝑜𝑙 3 ∫ 𝑑𝜀𝑣𝑜𝑙 3𝜀𝑣𝑜𝑙
=− 𝑖𝑛 −𝜀 𝑖𝑛 ) =− 𝑖𝑛 −𝜀 𝑖𝑛 ) = 𝑖𝑛 𝑖𝑛 ) (37)
𝑑𝑝 2𝑑(𝜀𝑐𝑎 𝑐𝑙 2 ∫ 𝑑(𝜀𝑐𝑎 𝑐𝑙 2(𝜀𝑐𝑙 −𝜀𝑐𝑎
𝑖𝑛 𝑖𝑛 𝑖𝑛
where 𝜀𝑣𝑜𝑙 , 𝜀𝑐𝑎 , and 𝜀𝑐𝑙 are the inelastic volumetric, axial, and lateral strains,
respectively. To calibrate the dilation angle, the peak engineering stress state under
uniaxial compression is considered since the hydrostatic pressure under uniaxial
compression reaches the maximum as intended for the determination of the dilation
angle. Let the Poisson’s ratio at the peak engineering stress (−𝑓𝑐′ ) under the uniaxial
𝑖𝑛
compression be 𝜈𝑐0 , the inelastic volumetric strain is then 𝜀𝑣𝑜𝑙 = 2(𝜈𝑐0 − 𝜈𝑐𝑒 )𝑓𝑐′ /𝐸𝑐 . In
𝑖𝑛 𝑖𝑛 𝑖𝑛
addition, 𝜀𝑐𝑎 = −𝜀𝑐𝑝 + 𝑓𝑐′ /𝐸𝑐 , 𝜀𝑐𝑙 = (𝜀𝑣𝑜𝑙 𝑖𝑛
− 𝜀𝑐𝑎 )/2 . Substituting these values into Eq.
(37) and neglecting the second term in the square root of Eq. (36) due to small value of
𝑓𝑡 /𝑓𝑐′ result in Eq. (35).
Fig. 4 shows the variation of the dilation angle with the compressive strength in the
range between 20 MPa and 100 MPa for concrete made with crushed aggregate. The
dilation angle falls in the range between 29𝑜 and 42𝑜 for normal strength of concrete
(20-60 MPa) and increases with the increasing the compressive strength.
16
𝜈𝑐𝑒 𝑓𝑜𝑟 |𝑒𝑐 | ≤ 𝜀𝑒𝑙
𝑛
𝜈𝑐 (|𝑒𝑐 |) = { |𝑒 |−𝜀 𝜈𝑐𝑝 −𝜈𝑐0 (38)
𝜈𝑐𝑝 − (𝜈𝑐𝑝 − 𝜈𝑐𝑒 ) exp [(𝜀 𝑐 −𝜀𝑒𝑙) ln(𝜈 )] 𝑓𝑜𝑟 |𝑒𝑐 | > 𝜀𝑒𝑙
𝑐𝑝 𝑒𝑙 𝑐𝑝 −𝜈𝑐𝑒
2
where 𝑒𝑐 is the uniaxial engineering compressive strain; 𝜈𝑐𝑒 = 8 × 10−6 𝑓𝑐′ + 0.0002𝑓𝑐′ +
0.138, the elastic (initial) Poisson’s ratio (Candappa et al, 2001); 𝜈𝑐0 = 0.5, the Poisson’s
ratio of concrete at the peak engineering stress; 𝜈𝑐𝑝 is the ultimate Poisson’s ratio at
large plastic strains to be determined; 𝜀𝑒𝑙 = 0.4𝑓𝑐′ /𝐸𝑐 , the strain limit to the elastic regime
under uniaxial compression. 𝑛 is a material constant being taken as 𝑛 = 3 to
approximately fit the experimental results of Candappa et al (2001), Binici (2005), and
Samani and Attard (2014) for small to moderate compressive deformation.
The strain-related Poisson’s ratio is proposed to be expressed as 𝜈𝑐 = 𝜈𝑐𝑒 (1 − 𝜀 𝑝 /𝜀) +
𝜈𝑐𝑝 (𝜀 𝑝 /𝜀) , where 𝜈𝑐𝑒 and 𝜈𝑐𝑝 are the elastic and plastic Poisson’s ratios, respectively.
As the strain increases, the stress decreases and the contribution of the elastic
Poisson’s ratio to the Poisson’s ratio is reduced. The plastic Poisson’s ratio is the
ultimate value when the elastic strain is negligibly small.
The plastic Poisson’s ratio is related to the yielding mechanism. Concrete is a
composite material consisting of a heterogeneous matrix of mortar and aggregates.
Under uniaxial compression of a concrete specimen, plasticity occurs when meso-
cracks are formed in the direction of the load. The extent of plasticity is increased as the
meso-cracks start to propagate and finally coalesce to form macro-cracks. Then the
specimen disintegrates into many parts separated by the macro-cracks, which are
considered as individual separated structures. Additional meso-cracks are formed in the
separated parts as the concrete exhibits a strain softening behavior. Lateral expansion
measured at the exterior surface of the specimen is very sensitive to the macro-crack
width, or the clear distance of the separated parts. Direct measurement of the lateral
expansion at the exterior surface may not be objective, because 1) the separated parts
may be subjected to non-uniform compression owing to non-uniform damage to and
different profiles of the individual parts, which lead to their different structural behavior
and cause different macro-crack widths; 2) the direct measurement at the exterior
surface is also affected by loading and restraining conditions at the ends of the
specimen; and 3) the individual structural behavior and interaction of the separated
parts should be captured by structural analysis, not by material constitutive model.
Therefore, the material behavior should be limited within the matrix of the individual
parts with meso-cracks and the lateral expansion of the specimen is to exclude the
macro-crack width. The plastic Poisson’s ratio represents the lateral plastic dislocation
of grains in the matrix and may be assumed to be constant during the material
softening.
The plastic Poisson’s ratio of concrete, 𝜈𝑐𝑝 , is then determined using the plastic
deformation at the peak engineering stress of concrete prior to the onset of macro
cracks. It is the ratio of the inelastic lateral strain increment to the inelastic axial strain
𝑑𝜀 𝑖𝑛
𝑐𝑙
increment, i.e., 𝜈𝑐𝑝 = − 𝑑𝜀𝑖𝑛 . Using Eqs. (36) and (37), the plastic Poisson’s ratio of
𝑐𝑎
3+2𝑡𝑎𝑛(𝜓)
concrete can be obtained as 𝜈𝑐𝑝 = 6−2𝑡𝑎𝑛(𝜓) . Based on the proposed dilation angle in
Eq. (35), the plastic Poisson’s ratio is given by
17
𝐸𝑐 𝜀𝑐𝑝 −2𝜈𝑐𝑒 𝑓𝑐′
𝜈𝑐𝑝 = (39)
2(𝐸𝑐 𝜀𝑐𝑝 −𝑓𝑐′ )
For normal strength concrete made with crushed aggregates, the plastic Poisson’s ratio
is approximately 1.0.
For the 40 MPa concrete, the proposed Poisson’s ratio is displaced in Fig. 5. This
figure also shows the derived Poisson’s ratios based on 1) the lateral strain model by
Samani and Attard (2014) which fits experimental results of the specimens of standard
geometry with presence of macro-cracks; 2) the Poisson’s ratio model by Candappa et
al (2001) which fits the experimental results under moderate inelastic lateral expansion;
3) the Poisson’s ratio model by Binici (2005) who calibrated the model such that the
curve crosses through the Poisson ratio (0.5) at the peak engineering stress and fits the
experimental results with presence of macro-cracks. The plots show good agreement of
the proposed Poisson’s ratio model with the experimental results in small to moderate
compressive strains up to 0.0025. At high compressive strains, the models with macro-
cracks included are not appropriate as discussed previously. Ferretti (2004) also
indicates that lateral strain measurements acquired on specimen surface are not real
lateral expansion of concrete material due to macro-cracks propagating through the
specimen. Therefore, the results in high strain range proposed by Binici (2005),
Candappa et al (2001) and Samani and Attard (2014) may have overestimated the
realistic lateral strain of the concrete due to presence of macro-cracks.
3
Samani and Attard (2014)
Binici (2005)
3
Candappa et al (2001)
This paper
2
Poisson's ratio
0
0 0.001 0.002 0.003 0.004 0.005 0.006 0.007 0.008 0.009 0.01
Engineering strain
Fig. 5 Comparison of Poisson’s ratio models
18
displacement relationship can be used for the corresponding dynamic relationships with
the updated materials properties due to effect of strain rates.
𝜀̇
(𝜀̇ 𝑐 )1.026𝛼𝑠 ≥ 1.0 𝑓𝑜𝑟 𝜀̇𝑐 ≤ 30𝑠 −1
𝑓̃𝑐′ (𝜀̇ 𝑐 )
𝑟𝑐 (𝜀̇𝑐 ) = = { 𝑐0 (40)
𝑓𝑐′ 𝜀̇
𝛾𝑠 (𝜀̇ 𝑐 )1/3 𝑓𝑜𝑟 𝜀̇𝑐 > 30𝑠 −1
𝑐0
where 𝜀̇𝑐0 = 30 × 10−6 𝑠 −1 , the static strain rate; 𝛼𝑠 = 1/(5 + 9𝑓𝑐′ /𝑓𝑐𝑜 ) ; 𝑙𝑜𝑔(𝛾𝑠 ) =
6.15𝛼𝑠 − 2 ; and 𝑓𝑐𝑜 = 10 𝑀𝑃𝑎. Since 𝐸𝑇 (𝜀𝑐 ) = 0 at the peak stress, the DIF in terms of
𝑝𝑙 𝑝𝑙
the plastic strain rate, 𝜀̇𝑐 , is 𝑟 𝑝𝑙 (𝜀̇𝑐 ) = 𝑟(𝜀̇𝑐 ) as per Eq. (4).
𝜀̇
(𝜀̇ 𝑡 )𝛿𝑠 ≥ 1.0 𝑓𝑜𝑟 𝜀̇𝑡 ≤ 1𝑠 −1
𝑓̃𝑡 (𝜀̇ 𝑡 )
𝑟𝑡 (𝜀̇𝑡 ) = = { 𝑡0 𝜀̇ (41)
𝑓𝑡
𝛽𝑠 (𝜀̇ 𝑡 )1/3 𝑓𝑜𝑟 𝜀̇𝑡 > 1𝑠 −1
𝑡0
where 𝜀̇𝑡0 = 10−6 𝑠 −1, the static strain rate; 𝛿𝑠 = 1/(1 + 8𝑓𝑐′ /𝑓𝑐𝑜 ) ; and 𝑙𝑜𝑔(𝛽𝑠 ) = 6𝛿𝑠 − 2.
Since 𝐸𝑇 (𝜀𝑐 ) = 0 at the peak stress, the DIF in terms of the plastic strain rate, 𝜀̇𝑡𝑝𝑙 , is
𝑟 𝑝𝑙 (𝜀̇𝑡𝑝𝑙 ) = 𝑟(𝜀̇𝑡 ) as per Eq. (4).
𝐸̃𝑐 (𝜀̇ 𝑢 ) 𝜀̇
𝑟𝐸 (𝜀̇𝑢 ) = = (𝜀̇ 𝑢 )0.026 ≥ 1.0 (42)
𝐸𝑐 𝑐0
The DIF of the modulus of elasticity cannot be presented in terms of the plastic strain
rate since the modulus is a quantity measured in the elastic regime. It is noted that there
may be no built-in method in a finite element analysis software to incorporate the strain-
rate-dependent modulus of elasticity. Then a user-defined subroutine may be needed to
be implemented in the software for the modulus to account for the strain rate effect.
19
𝜀̃𝑐𝑝 (𝜀̇ 𝑐 ) 𝜀̇
𝑟𝜀𝑝 (𝜀̇𝑐 ) = = (𝜀̇ 𝑐 )0.02 ≥ 1.0 (43)
𝜀𝑐𝑝 𝑐0
The DIF in terms of the plastic strain rate, 𝜀̇𝑐𝑝𝑙 , is assumed to be 𝑟𝐸 (𝜀̇𝑐𝑝𝑙 ) = 𝑟𝐸 (𝜀̇𝑐 ) since
𝐸𝑇 (𝜀𝑐 ) = 0 as per Eq. (4).
𝑤̇
𝐺̃𝑓 (𝜀̇ 𝑔 )
(𝑤̇ )𝛿𝑔 ≥ 1.0 𝑓𝑜𝑟 𝑤̇ ≤ 𝑤̇𝑖
𝑟𝐺 (𝑤̇ ) = = { 0 𝑤̇ (44)
𝐺𝑓
𝛽𝑔 (𝑤̇ )𝜆𝑔 𝑓𝑜𝑟 𝑤̇ > 𝑤̇𝑖
0
where 𝑤̇ is the crack opening velocity; 𝑤̇0 the reference crack opening velocity; 𝑤̇𝑖 the
𝑤̇
crack opening velocity at the trend transition; and 𝛽𝑔 = (𝑤̇ 𝑖 )𝛿𝑔−𝜆𝑔 . To fit the experimental
0
results as displayed in Fig. 6, the model parameters are taken as 𝑤̇0 = 0.01 𝑚𝑚/𝑠, 𝛿𝑔 =
0.08 , 𝜆𝑔 = 0.62 and 𝑤̇𝑖 = 200 𝑚𝑚/𝑠 . The comparison between the experimental results
and the proposed Eq. (44) is illustrated in Fig. 6.
20
Brara et al (2007)
25
DIF of fracture enegy
Cadoni et al (2013)
This paper
1
1E+0 1E+1 1E+2 1E+3 1E+4 1E+5
Crack opening velocity (mm/s)
Fig. 6 DIF of fracture energy in terms of crack opening velocity
100
Static test
(Engineering)
80 Static (True)
60
Stress (MPa)
40
20
0
0 0.001 0.002 0.003 0.004 0.005 0.006
Total Strain
21
5.1 Average stress-strain relationship of rebars
Hsu and Mo (2010) proposed a bilinear average stress-strain curve for rebars
embedded in concrete by fitting experimental data. The bilinear stress-strain
relationship may be sufficient if the rebars experience small to moderate inelastic
deformation. It may overestimate the strength of the rebars under large inelastic strain.
Therefore, a slight modification to this relationship by limiting the stress to the ultimate
stress, 𝑓𝑠𝑢 , is proposed and given by
′
𝐸𝑠 𝜀𝑠 𝑓𝑜𝑟 𝜀𝑠 ≤ 𝜀𝑠𝑦
𝜎𝑠 = { ′ (45)
(0.91 − 2𝐵𝑎 )𝑓𝑠𝑦 + (0.02 + 0.25𝐵𝑎 )𝐸𝑠 𝜀 ≤ 𝑓𝑠𝑢 𝑓𝑜𝑟 𝜀𝑠 > 𝜀𝑠𝑦
where 𝑓𝑠𝑦 and 𝑓𝑠𝑢 are the engineering yield stress and tensile strength of bare rebars,
′
𝑓𝑠𝑦
′ ′
respectively; 𝜀𝑠𝑦 = , the average (smeared) yield strain; 𝑓𝑠𝑦 = (0.93 − 2𝐵𝑎 )𝑓𝑠𝑦 , the
𝐸𝑠
1 𝑓
average yield stress; and 𝐵𝑎 = 𝜌 (𝑓 𝑡 )1.5 , in which 𝑓𝑡 is the tensile strength of concrete
𝑠𝑦
and 𝜌 (≥ 0.15%) the reinforcement ratio.
It is noted that the truss stress-strain relationship of rebars is identical to the
engineering stress-strain relationship if the rebars are modeled as truss or beam
elements without considering cross-sectional Poisson’s effect.
𝑓̃𝑠𝑦 (𝜀̇ 𝑠 ) 𝜀̇
𝑟𝑠𝑦 (𝜀̇𝑠 ) = 𝑠
= (10−4 )𝛼𝑠𝑦 ≥ 1.0 (46)
𝑓𝑠𝑦
𝑓𝑠𝑦
where 𝛼𝑠𝑦 = 0.074 − 0.04 414 . The DIF of the yield stress can be expressed in terms of
𝑝𝑙
the plastic strain rate, 𝜀̇𝑎𝑥𝑖𝑎𝑙 , which is assumed to be 𝑟𝑠𝑦 (𝜀̇𝑠𝑝𝑙 ) = 𝑟𝑠𝑦 (𝜀̇𝑠 ) since 𝐸𝑇 (𝜀𝑠𝑦 ) =
0 for rebars at the yield stress plateau regime.
𝑓̃𝑠𝑢 (𝜀̇ 𝑠 ) 𝜀̇
𝑟𝑠𝑢 (𝜀̇𝑠 ) = 𝑠
= (10−4 )𝛼𝑠𝑢 ≥ 1.0 (47)
𝑓𝑠𝑢
22
𝑓𝑠𝑦
where 𝛼𝑠𝑢 = 0.019 − 0.009 414 . The DIF of the tensile strength can be expressed in
terms of the plastic strain rate, 𝜀̇𝑠𝑝𝑙 , which is assumed to be 𝑟𝑠𝑢 (𝜀̇𝑠𝑝𝑙 ) = 𝑟𝑠𝑢 (𝜀̇𝑠 ) since
𝐸𝑇 (𝜀𝑠𝑢 ) = 0 for rebars at the peak engineering stress (also true stress as discussed in
Section 5.1).
𝑝𝑙
where 𝜎𝑦 (𝜀 ) is the evolved yield stress with the hardening variable being taken as the
𝑝𝑙
equivalent plastic strain, 𝜀 . As the associated plastic flow shows no plastic volume
change, the equivalent plastic strain simply coincides with the uniaxial plastic strain, i.e.,
𝑝𝑙 𝑝𝑙
𝜀 = |𝜀𝑎𝑥𝑖𝑎𝑙 | , which is given by a uniaxial test.
For cases involving gross plastic straining such as plastic deformation under impact
loading, an isotropic hardening rule can be assumed. For ductile metals, associated
flow rule is commonly applied, and the plastic flow potential function is identical to the
yield surface function.
𝑝 𝑝 𝛼𝑠1 1
𝜀̅𝑐𝑟 = 𝜀𝑐𝑟 𝑒𝑥𝑝 [− (1+𝑚 ) (𝜂 − 3)] (49)
2
where 𝜂 = −𝑝(𝜎𝑖𝑗 )/𝑞(𝜎𝑖𝑗 ) is the stress triaxiality factor defined as the ratio of the
hydrostatic pressure stress to the equivalent stress (𝜂 = 1/3 corresponds to uniaxial
tension); 𝛼𝑠1 = 2.2 for ferritic steel, or 𝛼𝑠1 = 0.6 for stainless steel; 𝑚2 = 0.6(1 − 𝑓𝑠𝑦 /𝑓𝑠𝑢 )
for ferritic steel, or 𝑚2 = 0.75(1 − 𝑓𝑠𝑦 /𝑓𝑠𝑢 ) for stainless steel; and 𝑓𝑠𝑦 and 𝑓𝑠𝑢 the
𝑝
engineering yield stress and engineering ultimate strength, respectively. 𝜀𝑐𝑟 is the
critical plastic strain at the onset of fracture to be determined based on uniaxial test
result of a standard specimen, or determined using Eq. (50) as per ASME BPVC
(ASME, 2017b):
23
𝑝 𝐸𝑓 𝑓𝑠𝑦
𝜀𝑐𝑟 = 𝐴𝜀 𝑙𝑛 (1 + 100) ≥ 𝐵𝜀 (1 − 𝑓 ) (50)
𝑠𝑢
𝑓𝑠𝑦
𝐸𝑠 𝑒 𝑓𝑜𝑟 𝑒 < 𝐸𝑠
𝑓𝑦𝑠
𝑠 = 𝑓𝑠𝑦 𝑓𝑜𝑟 𝐸𝑠
≤ 𝑒 ≤ 𝑒𝑠ℎ (51)
112(𝑒−𝑒 )+2 𝑒−𝑒𝑠ℎ 𝑓𝑠𝑢
𝑓𝑠𝑦 [ 60(𝑒−𝑒 𝑠ℎ)+2 + 𝑒 ( − 1.7)] 𝑓𝑜𝑟 𝑒𝑠ℎ < 𝑒 ≤ 𝑒𝑢
{ 𝑠ℎ 𝑢 −𝑒𝑠ℎ 𝑓𝑠𝑦
where 𝑒 and 𝑠 are uniaxial engineering strain and stress, respectively; 𝑓𝑠𝑦 and 𝑓𝑠𝑢 the
uniaxial engineering yield and peak stresses, respectively; 𝐸𝑠 the modulus of elasticity;
and 𝑒𝑠ℎ and 𝑒𝑢 the engineering strains at onset of strain hardening and that at the peak
engineering stress, respectively.
where 𝐸1 , 𝐸2 , and 𝛿 are material constants, which are used for a combination of a
Ramberg-Osgood exponential strain hardening at small to moderate plastic strains and
𝑝
a linear strain hardening correction at large plastic strains; 𝑟𝑖 = 1 + 𝜀𝑠ℎ − 𝐸1 𝜀𝑠ℎ −
24
𝑝 𝛿 𝑝
𝐸2 (𝜀𝑠ℎ ) ; 𝜀𝑠ℎ = 𝜀𝑠ℎ − 𝑓𝑠𝑦 /𝐸𝑠 ; 𝜀𝑠ℎ the strain at onset of strain hardening in the
engineering stress-strain curve (e.g., the strain at the end of the yield stress plateau in
the engineering stress-strain curve); 𝐸𝑠 the modulus of elasticity; 𝑓𝑠𝑦 the yield stress
defined as the elastic limit stress below which stress-strain behavior is linear (it is noted
𝑝
that yield stress is not the proof or offset yield stress); and 𝜀𝑐𝑟 is determined by test or
given by Eq. (50).
For mild steels, the material constants 𝐸1 , 𝐸2 and 𝛿 in Eq. (52) are proposed to be
determined for the true stress-plastic strain curve to cross through the following three
true plastic strain-stress points: 〈𝜀𝐿𝑝 , 𝜎𝐿 〉; 〈𝜀𝑢𝑝 , 𝜎𝑢 〉; and 〈𝜀ℎ𝑝 , 𝜎ℎ 〉:
112𝑒 +2 𝑒𝑠ℎ
𝜎𝐿 = (1 + 2𝑒𝑠ℎ ) [ 60𝑒 𝑠ℎ+2 𝑓𝑠𝑦 + 𝑒 (𝑓𝑠𝑢 − 1.7𝑓𝑠𝑦 )]
𝑠ℎ 𝑢 −𝑒𝑠ℎ
𝜎𝐿
𝜀𝐿𝑝 = ln(1 + 2𝑒𝑠ℎ ) −
𝐸𝑠
𝜎𝑢 = 𝑓𝑠𝑢 (1 + 𝑒𝑢 ) (53)
𝜎
𝜀𝑢𝑝 = 𝑙𝑛(1 + 𝑒𝑢 ) − 𝐸𝑢
𝑠
1+𝑒
𝜎ℎ = 0.5𝑓𝑠𝑢 exp(2𝜀𝑢𝑝 )(1 + √exp(2𝜀𝑢𝑝 ))
𝑢
𝜀ℎ𝑝 = 2𝜀𝑢𝑝
in which 𝜀𝐿𝑝 and 𝜀𝑢𝑝 are the true plastic strains at the true stress, 𝜎𝐿 , corresponding the
engineering stress determined using Eq. (51) with respect to the engineering strain of
2𝑒𝑠ℎ and the true stress, 𝜎𝑢 , corresponding to the peak engineering stress, 𝑓𝑠𝑢 ,
𝑝
respectively. The true stress, 𝜎ℎ , is derived using Eq. (3) along with the fact that 𝜀ℎ ≅ 𝜀ℎ
for the large strain and that the engineering stress is very close to 𝑓𝑠𝑢 at the true plastic
strain of 𝜀ℎ𝑝 or approximately the corresponding engineering strain of 1 − 1/(1 + 𝑒𝑢 )2 .
25
800
400
Engineering (Chen, 2010)
True (Chen, 2010)
200 This paper
True with specified properties
0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8
Plastic Strain
Fig. 8 Stress-plastic strain relationship of CSA G40.20/21 300W
where 𝑓𝑠𝑝 is the proportionality limit taken as the proof stress at 0.01% strain offset.
Based on the engineering stress-strain relationship included in Annex C of Eurocode 3
(2006) for the constitutive modeling of stainless steel material behavior, 𝑓𝑠𝑝 falls in the
range of 0.6 to 0.7 times the yield or proof stress at 0.2% strain offset, 𝑓𝑠𝑦 .
Approximately, it can be taken as 𝑓𝑠𝑝 = 0.65𝑓𝑠𝑦 . 𝐸1 , 𝐸2 and 𝛿 are material constants
which are proposed to be determined for the true stress-plastic strain curve to cross
through the following three true plastic strain-stress points: 〈𝜀𝑎𝑝 , 𝜎𝑎 〉; 〈𝜀𝑢𝑝 , 𝜎𝑢 〉; and
〈𝜀ℎ𝑝 , 𝜎ℎ 〉:
𝜎𝑎 = 0.5(𝑓𝑠𝑦 + 𝑓𝑠𝑢 )(1 + 𝑒𝑎 )
𝜎𝑎
𝜀𝑎𝑝 = 𝑙𝑛(1 + 𝑒𝑎 ) −
𝐸𝑠
𝜎𝑢 = 𝑓𝑠𝑢 (1 + 𝑒𝑢 ) (55)
𝜎
𝜀𝑢𝑝 = 𝑙𝑛(1 + 𝑒𝑢 ) − 𝐸𝑢
𝑠
1+𝑒
𝜎ℎ = 0.57𝑓𝑠𝑢 exp(2𝜀𝑢𝑝 )(1 + √exp(2𝜀𝑢𝑝 ))
𝑢
𝜀ℎ𝑝 = 2𝜀𝑢𝑝
26
𝑓 𝑓𝑠𝑦 𝑓 1−𝑓𝑠𝑦 /𝑓𝑠𝑢
𝑒𝑎 = 0.007 𝑓𝑠𝑢 − 0.005 + 𝑠𝑢
+ 2𝐸 + 1+3.5𝑓𝑠𝑦 /𝑓𝑠𝑢
𝑠𝑦 𝐸𝑠 𝑠 2
where 𝜀𝑦𝑝 and 𝜀𝑢𝑝 are the true plastic strains at the true yield stress, 𝜎𝑦 , and the true
stress, 𝜎𝑢 , corresponding to the peak engineering stress, respectively. 𝑒𝑎 is the
engineering strain corresponding to the engineering stress of 0.5(𝑓𝑠𝑦 + 𝑓𝑠𝑢 ) which is
determined based on the engineering stress-stain curve adopted in Annex C of
Eurocode 3 (2006). The proposed strain, 𝜀𝑢𝑝 , is derived based on the engineering strain,
𝑒𝑢 = 1 − 𝑓𝑠𝑦 /𝑓𝑠𝑢 , corresponding to the peak engineering stress, , adopted in Annex C
of Eurocode 3 (2006). The true stress, 𝜎ℎ , is derived using Eq. (3) along with the fact
that 𝜀ℎ ≅ 𝜀ℎ𝑝 for the large strain and that the engineering stress approximately equals to
𝑓𝑠𝑢 at the true plastic strain of 𝜀ℎ𝑝 or approximately the corresponding engineering strain
2
of 1 − [𝑓𝑠𝑢 /(2𝑓𝑠𝑢 − 𝑓𝑠𝑦 )] ≈ 𝑒𝑢 for 𝑓𝑠𝑢 = (1 𝑡𝑜 3)𝑓𝑠𝑦 .
For AISI 304L, Blandford et al (2007) performed experimental investigation of the true
stress-strain relationship. The tensile properties of the 304L (average of three tests) are
𝑓𝑠𝑦 = 277 𝑀𝑃𝑎 and 𝑓𝑠𝑢 = 660 𝑀𝑃𝑎. Using these tensile properties, and letting the true
stress-strain curve given in Eq. (54) cross through the proposed three stress-strain
points with Eqs. (55), the material parameters in Eq. (54) are determined to be 𝐸1 =
2.532, 𝐸2 = 5.727 and 𝛿 = 0.569. The comparison between the resulted true stress-
plastic strain curve and the experimental result by Blandford et al (2007) is shown in
Fig. 9.
2400
2000
1600
Stress (MPa)
1200
This paper
800 Blandford et al…
400
0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 1.3 1.4 1.5 1.6 1.7
Plastic Strain
Fig. 9 True stress-strain relationship of stainless steel AISI 304L
As per ASME/BPVC Section II, Part D (ASME, 2017a), the specified tensile properties
for AISI 304L are 𝑓𝑠𝑦 = 170 𝑀𝑃𝑎 and 𝑓𝑠𝑢 = 485 𝑀𝑃𝑎. Accordingly, the corresponding
material parameters for the true stress-plastic strain relationship in Eq. (54) are 𝐸1 =
−45.549, 𝐸2 = 55.778 and 𝛿 = 0.941.
27
the strain rate does not have a significant effect on the engineering strain at the peak
engineering stress and fracture strain.
𝑝 𝑝 𝑝 𝛿1 𝑝 𝛿2
̃𝑠 (𝜀̇ 𝑠 ,𝜀𝑠 )
𝜎 𝜀̇ 𝜀
𝑟𝑠 (𝜀̇𝑠𝑝 , 𝜀𝑠𝑝 ) = 𝑝 = 1 + (𝜀̇ 𝑝𝑠 ) [1 − (𝜀𝑝𝑠 ) ] (56)
𝜎𝑠 (𝜀𝑠 ) 𝑠0 𝑠0
𝑝 𝑝
where 𝜀̇𝑠0 , 𝛿1 , 𝜀𝑠0 , and 𝛿2 are material constants. For the mild steel CSA G40.20/21
𝑝 𝑝
300W, the material constants are determined to be 𝜀̇𝑠0 = 44.409 , 𝛿1 = 0.267 , 𝜀𝑠0 =
0.55 , and 𝛿2 = 0.4 to fit the experimental results by Chen (2010).
𝜎 𝜎 𝑛
𝜀 = 𝐸 + 0.002 (𝑓 ) (57)
𝑠 0.2
where 𝜀 and 𝜎 are uniaxial true strain and stress, respectively; 𝑛 = 2.4 +
2.9(𝑓0.2 /𝑓𝑢 )/(1 − 0.95𝑓0.2 /𝑓𝑢 ) , in which 𝑓0.2 is the 0.2% proof stress and 𝑓𝑢 the peak
engineering stress. In the absence of the 0.2% proof stress, this ratio can be taken as
𝑓0.2 /𝑓𝑢 = 1/[1.07 + (350/𝑓𝑢 )4.8 ].
𝑝 0.04
̃ (𝜀 )
𝜎 𝜀̇
𝑟𝑠 (𝜀̇𝑠𝑝 , 𝜀𝑠𝑝 ) = 𝜎𝑠 (𝜀𝑠 ) = (1 + 0.2
𝑠
) (58)
𝑠 𝑠
28
where 𝜎𝑠 (𝜀𝑠 ) is the static true stress and 𝜎̃𝑠 (𝜀𝑠 ) the dynamic true stress. This DIF is
assumed to be applicable to general weld metals due to scarcity of experimental results.
29
Mechanical contacts among all package components are established. The contact
setup is assigned to the interfaces between the concrete-steel liners, cask-lid flanges,
steel liners-impact limiter shells, impact limiter shell-shell, crushable foam-impact limiter
shell, rigid plane-impact limiter shell, and self-contact of materials. Rebars are
embedded in the host concrete by using the embedded constraint with the rebar steel
being assigned with a smeared stress-strain relationship.
Strain-rate-dependent material constitutive models in terms of true stress and train are
specifically established based on the equations presented in this paper. Fracture criteria
are incorporated in the progressive failure rule in conjunction with the material
constitutive models. Attention is brought to the tensile stress-crack opening relationship
of the concrete constitutive model. This relationship is converted into the tensile stress-
crack strain relationship based on the mesh size.
30
example, the initial stiffness multiplication factor to scale the default contact
stiffness may be given a value of 0.0001.
It can be seen from Table 1 that the effect of strain rate is significant. Constitutive
models for the materials used in the transport packages must include the effect of strain
rate.
31
Based on the fracture criteria that are stress-triaxiality-dependent as given in Eq. (49),
no fracture occurs in the DSC containment boundary components including steel liners,
flanges, and welds, although local and minor yielding occurs in the DSC steel liner.
Fracture does occur in the stainless steel shells of the impact limiter directly in contact
with the rigid surface. Cracks occur in the lid concrete, but the shielding capacity is
judged not to be significantly affected because the crack widths are small and cracks
will be closed after the dynamic impact since the rebars in the lid concrete are still in the
elastic regime, except one reinforcing tie is slightly yielded.
8. Conclusions
This study proposed a framework for dynamic analysis of radioactive material
transport packages under accident drop scenarios to meet the nuclear licensing
requirements. The true stress and strain, effect of strain rate, triaxial stress states,
realistic constitutive models, and damage and fracture criteria of materials are
incorporated into the dynamic analysis framework to address large deformation of
materials occurring in a very short impact duration. Critical material parameters
associated with material constitutive models, true stress-strain relationships, and effect
of strain rate are studied and determinations of these parameters are proposed. Based
on the study, the following conclusions can be drawn:
Materials in the transport packages are subjected to large deformation under the
accident 9-meter drop scenario. Therefore, true stress and strain should be used
to realistically reveal the actual strain energy distribution among the package
components.
Effect of strain rates is significant even with impact limiters being attached to the
radioactive material containers. The strain rates of the materials subjected to the
accident condition can cause the increase of the yield stresses by 50% to 160 %.
Stress conditions in the transport packages are very complex. Therefore,
material constitutive models should be established for highly triaxial stress states.
Parameters associated with the material constitutive models, true stress-strain
relationships, and effect of strain rate are appropriately calibrated and
determined based on experimental results or theoretically derived based on
observed material behavior in elementary stress states.
Stress triaxiality-dependent damage and fracture criteria of materials are
incorporated into the dynamic analysis framework to limit the plastic deformation
to material fracture and to safeguard the containment boundary of the DSC that
contains radioactive materials.
Ignoring effect of strain rate and using engineering stress-strain overestimate the
strain energy absorption in the impact limiters and underestimate the strain
energy distribution to the critical radioactive material containers.
The proposed framework is sufficient to capture the realistic mechanical behavior of
materials and the structural behavior of the transport packages and radioactive material
containers. It can be conveniently implemented in analysis software for the dynamic
analysis of the transport packages under various accident impact scenarios.
32
Acknowledgements
The authors acknowledge Ontario Power Generation Inc. for providing resources to
write this paper. The opinions and conclusions expressed in this paper are of the
authors and do not necessarily reflect the official position of Ontario Power Generation
Inc.
References
N. Arioglu, Z. C. Girgin, and E. Ariglu, 2006. Evaluation of ratio between splitting tensile
strength and compressive strength for concrete up to 120 MPa and its application in
strength criterion. ACI Materials Journal, Vol. 103, No. 1, pp.18-24.
F. Aslani and R. Jowkarmeimandi, 2012. Stress-strain model for concrete under cyclic
loading. Magazine of Concrete Research, Vol. 64, No. 8, pp. 673-685.
ASME, 2017a. Boiler and pressure vessel code 2017, Section II: Materials, Part D:
Properties. American Society of Mechanical Engineers.
ASME, 2017b. Boiler and pressure vessel code 2017, Section VIII: Rules for
Construction of Pressure Vessels, Division 2: Alternative Rules. American Society of
Mechanical Engineers.
A. Brara and J. R. Klepaczko, 2007. Fracture energy of concrete at high loading rates in
tension. International Journal of Impact Engineering, Vol. 34, pp. 424-435.
33
E. Cadoni, G. Solomos and C. Albertini, 2013. Concrete behavior in direct tension tests
at high strain rates. Magazine of Concrete Research, Vol. 65, No. 11, pp. 660-672.
J. Chen, 2010. An experimental study of strain rate effects on mild steel. Master Thesis,
Carleton University, Ottawa, Ontario, Canada.
J. B. Choi and R. S. Lakes, 1992. Non-linear properties of polymer cellular material with
a negative Poisson’s ratio. Journal of Materials Science. Vol. 27, pp. 4678-4684.
CSA, 2012. Design and assessment of buildings subjected to blast loads. Standard
CSA S850-12. Canadian Standard Association (CSA).
EPRI, 2005a. Spent fuel transportation applications: fuel rod failure evaluation under
simulated cask side drop conditions. Report No. 1009929, Electric Power Research
Institute, Palo Alto, California, USA.
EPRI, 2005b. Spent fuel transportation applications: global forces acting on spent fuel
rods and deformation patterns resulting from transportation accidents. Report No.
1011817, Electric Power Research Institute, Palo Alto, California, USA.
Eurocode 3, 2006. Design of steel structures – Part 1-4: General rules – Supplementary
rules for stainless steels. EN 1993-1-4, European Standad.
Federation Internationale du Beton (FIB), 2012. CEB-FIP Model Code 2010 – final draft.
FIB Bulletin 65, Vol. 1, Lausanne, Switzerland.
34
T. T. C. Hsu and Y. L. Mo, 2010. Unified Theory of Concrete Structures. New York:
John Wiley & Sons.
IAEA Safety Standards, 2012. Regulations for the Safe Transport of Radioactive
Material, Specific Safety Requirements No. SSR-6, International Atomic Energy
Agency.
J. Lee and G. L. Fenves, 1998. Plastic-damage model for cyclic loading concrete
structures. Journal of Engineering Mechanics, Vol. 124, No. 8, pp. 892-900.
J. A. Lichtenfeld, M. Mataya, and C. J. Van Tyne, 2006. Effect of strain rate on stress-
strain behavior of alloy 309 and 304L austenitic stainless steel. Metallurgical and
Material Transactions A, Vol. 37A, No. 1, pp. 147-161.
35
Q. Liu, and G. Subhash, 2004. A phenomenological constitutive model for foams under
large deformations. Polymer Engineering and Science, Vol. 44, No. 3, pp. 463-473.
R. Malm, 2009. Predicting shear type crack initiation and growth in concrete with non-
linear finite element method. TRITA-BKN, Bulletin 97, Royal Institute of Technology,
Division of Structural Design and Bridges, Stockholm, Sweden.
L. J. Malvar, 1998. Review of static and dynamic properties of steel reinforcing bars.
ACI Materials Journal, Vol. 95, No. 5, pp. 609-616.
L. J. Malvar and C. A. Ross, 1998. Review of strain rate effect for concrete in tension.
ACI Materials Journal, Vol. 95, No. 6, pp. 735-739.
K. Pack and D. Mohr, 2017. Combined necking & fracture model to predict ductile
failure with shell finite elements. Engineering Fracture Mechanics, Vol. 182, pp. 32-51.
36
L. Qiao, U. Zencker, H. Völzke, F. Wille, and A. Musolff, 2012. Validation of numerical
simulation models for transport and storage casks using drop test results. Proceedings
of the Eleventh International Conference on Computational Structures Technology,
B.H.V. Topping, (Editor), Civil-Comp Press, Stirlingshire, Scotland.
A. K. Samani, and M. M. Attard, 2012. A stress-strain model for uniaxial and confined
concrete under compression. Engineering Structures, Vol. 41, pp. 335-349.
A. K. Samani, and M. M. Attard, 2014. Lateral strain model for concrete under
compression. ACI Structural Journal, Vol. 111, No. 2, pp. 441-452.
B. Song, W. Y. Lu, and C. J. Syn, 2009. The effects of strain rate, density and
temperature on the mechanical properties of polyurethane diisocyanate (PMDI)-based
rigid polyurethane foams during compression. Journal of Materials Science, Vol. 44, No.
2, pp. 351-357.
P. Soroushian and K.–B. Choi, 1987. Steel mechanical properties at different strain
rates. ASCE Journal of Structural Engineering, Vol. 113, No. 4, pp. 863-872.
M. G. Stout and P. S. Follansbee, 1986. Strain rate sensitivity, strain hardening, and
yield behavior of 304L stainless steel. ASME Journal of Engineering Materials and
Technology, Vol. 108, pp. 344-353.
J. G. M. Van Mier, 1986. Multiaxial strain softening of concrete. Part I: fracture. Material
s and structures, Col. 19, No. 111, pp. 179-190.
37
J. Weerheijm and I. Vegt, 2010. The dynamic fracture energy of concrete: Review of
test methods and data comparison. Proceedings of the 7th International Conference on
Fracture Mechanics of Concrete and Concrete Structures, Jeju Island, Korea.
S. Yan, X. Zhao, and A. Wu, 2018. Ductile Fracture Simulation of Constructional Steels
Based on Yield-to-Fracture Stress–Strain Relationship and Micromechanism-Based
Fracture Criterion. ASCE Journal of Structural Engineering, vol. 144, No. 3, 04018004.
38