Sunteți pe pagina 1din 9

Proc Indian natn Sci Acad 79 No. 3 September 2013 Spl. Issue, pp.

Printed in India.

Research Paper
Impact Simulation of Rocks under SHPB Test
TANUSREE CHAKRABORTY1
1Assistant Professor, Department of Civil Engineering, Indian Institute of Technology (IIT) Delhi

(Received 16 January 2013; Accepted 26 March 2913)

In the present study, an attempt has been made to understand the impact induced stress-strain response of rocks up to 100/
sec strain rate through numerical simulation of uniaxial and triaxial split Hopkinson pressure bar (SHPB) tests. The finite
element (FE) software Abaqus has been used with explicit time integration scheme to simulate the SHPB tests on rocks.
The modelled FE geometry of the SHPB test consists of a steel incident bar, the rock sample and a steel transmission bar.
High strain rate material constitutive models have been used to characterise the rock sample. The impact of striker bar on
the incident bar is simulated by applying an impact induced incident stress pulse on the incident bar. The analysis results
have been compared with the laboratory test data collected from the literature. The numerical simulation of SHPB test on
rocks resulted in good correlation with the experimental results. It is observed from the simulation results that the peak
axial stress of rocks increases with increasing stiffness of rocks and increasing confining pressure.

Key Words: Finite Element; Impact; Numerical Simulation; Rock; Split Hopkinson Pressure Bar Test

Introduction In rock mechanics, the SHPB test has been used


to determine the dynamic compressive and tensile
Blast and impact resistant design of underground civil
strength of rocks [13-27]. Lu et al. [16] showed that
infrastructure in rock, e.g. railway stations, power
the dynamic increase factor (DIF) for rocks, which
house caverns, ammunition storage structures and
is defined as the ratio of the dynamic strength to the
energy storage caverns require characterisation of the
quasi-static strength in uniaxial compression,
surrounding rock under impact loading. The split
increases gradually with strain rate in the low strain
Hopkinson pressure bar (SHPB) has been widely used
rate regime (104-101/sec), however turns to rapid
to experimentally study the response of materials
increase in the intermediate (101-102/sec) and high
under impact loading. After Hopkinson, Kolsky and
(102-104/sec) strain rate regimes. Lindholm et al. [22]
Lindholms early work [1-5], the split Hopkinson
showed that energy required in fracturing basalt rock
pressure bar has gradually become a widely accepted
increased with increasing rate of strain. Li et al. [26]
facility to test the mechanical properties of materials
observed that the use of bigger samples is required
under dynamic loading at high strain rate, e.g. 102-
to reach stress uniformity in heterogeneous rocks.
104/sec [6-12]. The SHPB setup usually consists of a
Moreover, reduction in loss of incident wave was
striker bar, an incident bar and a transmission bar
observed in the presence of pre-stress. Cho et al. [27]
with all bars having the same diameter and cylindrical
performed dynamic tensile SHPB test on rock and
shape. The sample is sandwiched between the
showed that the dynamic tensile strength increases
incident and the transmission bars; the striker bar hits
with strain rates.
end of the incident bar in the axial direction with an
initial velocity of V0. The SHPB tests performed on rocks reported

*Author for Correspondence: E-mail: tanusree@civil.iitd.ac.in


2 Tanusree Chakraborty

in the literature are uniaxial in nature. However, simulation on soft rocks. By using the numerical
beneath the earth surface, rocks are subjected to high simulation a large number of studies can be performed
confining pressure. The compressive strength of rock which otherwise would be impossible to study
increases and crack density decreases with increasing through experiments involving considerable amount
confining stress. Hence, it is necessary to understand of resource, time and tediousness. In the present
the dynamic behaviour of rocks under multi-axial investigation, uniaxial and triaxial SHPB tests are
loading. Blanton [28] performed dynamic triaxial simulated using finite element (FE) software Abaqus
compression test on three rocks to study the effect of [31] to study (a) the stress-strain behaviour of rocks
strain rates on rocks up to 10/sec strain rate. However, and (b) the effect of radial confinement on the peak
the conventional dynamic triaxial compression tests axial stress of the rocks.
cannot reach strain rate higher than 10/sec [28].
Nemat-Nasser et al. [29] discussed the setup of Numerical Simulation
triaxial SHPB test. Li and Meng [30] investigated In the present investigation, finite element code
the influence of radial confinement in the SHPB test Abaqus is used for modelling rock samples under
on concrete like brittle materials and showed that dynamic compression in SHPB test. Fig. 1 shows a
compressive strength increases significantly with schematic diagram of the SHPB test setup. In SHPB
increasing confinement pressure as compared to the test, when the striker hits the incident bar in the axial
uniaxial compressive strength. However, triaxial direction, a rectangular stress pulse generates, which
SHPB test has not been performed on rocks in the propagates along the incident bar. When the length
previous investigations. of striker is short as compared to the total length of
Characterisation of rock behaviour under impact the incident bar and the transmission bar, the peak
loading for wide range of strain rates requires both rectangular incident stress is given by
experimental and numerical simulation of SHPB tests 1
on rock under uniaxial and triaxial conditions. i = bar cbarV0 (1)
2
Numerical simulation of SHPB tests on rocks
considering the elasto-plastic behaviour of rocks is and the duration of the stress pulse is given by
rather unavailable in the previous studies. Hence, the
2Ls
objectives of the present study are to understand the t = (2)
cbar
SHPB simulation technique for rocks and to identify
a suitable constitutive model for the impact load where bar is the mass density of the bar; cbar = Ebar

Fig. 1: Schematic diagram of SHPB test setup and a typical finite element mesh used in SHPB simulation
Impact Simulation of Rocks under SHPB Test 3

/bar is the elastic longitudinal stress wave velocity The C3D8R elements take into account the large mesh
in the bar; Ebar is the Youngs modulus of the bar; V0 distortion without affecting the results due to
is the velocity of the striker bar; and Ls is the length volumetric locking which is a common problem in
of the striker bar [32]. Transmitted and reflected stress large deformation analysis. Fig. 1 shows the 3D
pulses are generated and propagated in the meshes of the bars and the rock sample used in the
transmission bar and the incident bar, respectively. present investigation. The bar and sample meshes are
When there is force equilibrium between the samples selected through mesh convergence study. There are
incident and transmitted sides, the sample is 77958 elements in the incident bar, 16425 elements
uniformly deformed, and the stress-strain curves of in the rock sample and 33489 elements in the
the sample can be determined with one dimensional transmission bar. The rock sample is meshed using a
wave equation assumption given by global seed of 0.003 and bars are meshed using a
global seed of 0.05 considered accurate enough for
t = i + r (3)
obtaining the strain profile. A general contact
and algorithm is used to define contact between bars and
sample in the SHPB test. A central difference scheme
Abar Ebar t
= (4)
is used to integrate the equation of motion explicitly
Asp through time. The method is conditionally stable for
time increments (t) that are smaller than Courant
where t, i and r are the values of transmitted,
time limit, t < l/c, where l is the smallest element
incident and reflected signals in the bars, respectively.
dimension and c is the speed of sound wave in
The parameters Abar and Asp are the cross-sectional
medium in which it travels. Also, the artificial bulk
areas of bars and specimen respectively.
viscosity is activated to properly represent
In the current study, the incident bar, propagation of the induced compressive stress wave
transmission bar and the sample have been modelled by employing quadratic and linear functions of
to simulate the SHPB test, as shown in Fig. 1. The volumetric strain rates with default values of 1.2 and
length and diameter of the incident and the 0.06, respectively. Gauges in the form of reference
transmission bars are 2.13 m and 0.915 m, points are placed at several locations along the sample
respectively [33]. The sample has both length and length and bars in order to collect the data of incident,
diameter of 12.7 mm. The bars were made from high reflected and transmitted waves. The contact surface
strength maraging VM350 steel (Vasco Pacific, between the incident bar, sample and transmission
Montebello, CA) and have density, = 8100 kg/m3; bar are defined as frictionless and hard. The boundary
Youngs modulus, E = 200 GPa; and bar wave conditions are applied to bars and sample such that
velocity, c = 4970 m/sec. The striking velocity of movement in one direction is allowed to simulate the
striker bar is considered to be 13.9 m/sec. The bar one-dimensional wave propagation which is the basis
dimensions, bar material properties and the velocity of SHPB test. The uniaxial simulations are performed
of the striker bars are taken from [33]. The bars and for three different rock types, e.g., limestone, weak
the sample are placed in the same line. The end of sandstone and granite. The simulation results for
the transmission bar is kept fixed. The impact of the limestone are compared with the experimental data
striker bar on the incident bar is simulated by applying given by Frew et al. [33].
an incident stress pulse on the incident bar calculated
In the present investigation, the crushable foam
using equation (1) for time t calculated using
plasticity model in Abaqus has been used to model
equation (2).
soft rocks, e.g. limestone and weak sandstone [34].
Three dimensional (3D) solid eight node brick The performance of the model has also been tested
elements with reduced integration and hourglass for hard rock, e.g. granite. The crushable foam
control (C3D8R) are used for modelling the assembly plasticity constitutive model can simulate
of incident bar, transmission bar and the rock sample. compressive stress induced compaction behaviour.
4 Tanusree Chakraborty

The yield surface of the model takes an elliptical Table 2: Properties of bars
shape in the mean stress (p) vs. deviatoric stress (q)
Rock Length Mass Modulus Poissons Material Refe-
plane. Inside the yield surface, the behaviour of the type L (m) den- of ratio, rences
rocks remains linear elastic. The elliptical yield sity, elasticity, v
surface equation of the model is given by (kg/m3) E (GPa)

F = q 2 + 2 ( p p0 ) B = 0
2 Incident 2.13 8100 200 0.3 High strength
(5) maraging
VM350 [33]
where p0 is given by (pc-pt)/2; pc and pt are the yield steel
strength values of the rocks under hydrostatic
Trans- 0.915 8100 200 0.3
compression and tension, respectively. The parameter mission
B is the magnitude of the intercept of the yield surface
with the vertical axis for deviatoric stress q; and the
Table 3: Dynamic increase factor (DIF) for different
parameter defines the shape of the yield surface in
materials [37]
the meridional plane. The volumetric hardening of
the model is defined by providing the experimental S.No. Material description Strain rate Dynamic
data for uniaxial compressive strength with axial ( ) increase
strain. The strain rate dependence of rocks is included (/sec) factor (DIF)

in the model by defining the increase of dynamic yield 1. Limestone 0 1


strength with respect to the static yield strength, i.e., 0.01 1.36
the dynamic increase factor (DIF) with the increase 0.1 1.47
1 1.59
in strain rate. The constitutive model uses a non- 10 1.71
associated flow rule with the plastic potential surface 100 1.83
(GP), given by 1000 1.95

2. Weak sandstone 0 1
(6) 0.01 1.4
0.1 1.85
1 2.5
The physical properties of the rocks and yield 10 3.2
strength of rocks in compression are given in Table 100 4.27
1000 5.63
1. The stress-strain curves for the limestone, weak
sandstone and granite are obtained from [33], [35] 3. Granite 0 1
and [36], and used as input in Abaqus. The DIF 1 1.36
5 1.4
parameters under high loading rates used in the 45 1.45
present study are given in Table 3. The DIF 145 1.59
parameters are obtained in the present study from Li 676 1.9
et al. [37]. 731 2.15
1023 2.27
2344 2.47
Table 1: Properties of rocks

Rock Mass Modulus Poissons Yield References


type density, of elasticity, ratio, v strength, Results and Discussions
(kg/m3) E (GPa) y (MPa)
Uniaxial SHPB Simulation
Limestone 2300 24 0.35 68 [33]
Fig. 2(a) presents the axial stress (a) - strain (a)
Weak 2291 13.6 0.35 23 [35]
sandstone
response of limestone under uniaxial SHPB test
simulation. The simulation results are compared with
Granite 2620 50 0.17 200 [36] the experimental data given in [33]. It is observed
Impact Simulation of Rocks under SHPB Test 5

Fig. 2: Comparison of simulation results with experimental data for limestone

Fig. 3: Comparison of axial stress, incident and transmitted strain in different rocks
6 Tanusree Chakraborty

from the figure that the simulation results compare of experiment. The difference between the simulation
reasonably well with the experimental data in the results and experimental data may be attributed to
initial part of the stress-strain curve at low strains the application of stress pulse in simulation as
and in predicting the peak stress. Insignificant opposed to the impact of striker bar in the experiment
mismatch in the results is seen in the strain softening without pulse shaping.
region. The peak axial stress magnitude for limestone
Fig. 3(a) shows the stress-strain plots of
is 119.881 MPa. Fig. 2(b) compares the incident (I)
limestone, weak sandstone and granite under uniaxial
and transmitted (T) strain time histories in the SHPB
SHPB test. Maximum peak axial stress is observed
test. The peak strain magnitudes from the simulation
in granite, e.g., 145.678 MPa. Limestone and
results compare well with the experimental data. The
sandstone show 119.881 and 95.19 MPa peak axial
time duration of the peak strain is shorter in the case

Fig. 4: Effect of confining pressure on SHPB stress-strain response for different rocks
Impact Simulation of Rocks under SHPB Test 7

sample in a static analysis and (ii) simulation of SHPB


test in a dynamic explicit analysis. The stress state at
the end of static analysis is imported in the dynamic
analysis. Figs. 4(a) to (c) show the triaxial SHPB
simulation results for limestone, weak sandstone and
granite respectively at 0, 5 and 20 MPa confining
pressures. The peak axial stress values for all the three
rocks at three different confining pressures are noted
in Fig. 4. It is observed that the peak axial stress of
all the three rocks increased with increasing confining
pressure. Also, the initial stiffness of the rocks
increases with increasing confining pressure. Figure
5 summarises the peak axial stress of the rock samples
with increasing confinement. The rate of increase of
the peak axial stress is higher for soft rocks, e.g. weak
sandstone and lime stone and lower in hard rock, e.g.
Fig. 5: Effect of confining pressure on SHPB peak stress response granite.

Conclusions
stresses, respectively. Higher peak stress is observed
for rocks with higher stiffness. The initial slope of In the present study, an attempt has been made to
the stress-strain curves also clearly shows the effect understand the impact induced stress-strain response
of stiffness of the rocks, i.e. the initial slope decreases of rocks up to 100/sec strain rate through numerical
with decreasing stiffness of rock. For granite, the simulation of uniaxial and triaxial split Hopkinson
loading and unloading paths remain same because pressure bar (SHPB) tests using the finite element
granite does not exhibit elasto-plastic response. (FE) software Abaqus/Explicit. From these
However, plastic strain is observed in the cases of simulations, the following conclusions are drawn:
limestone and weak sandstone which is expected. (1) The simulation results compare reasonably
Figs. 3(b) and (c) compare the incident and well with the experimental data for the initial part of
transmitted strains for three different rocks. The the stress-strain curve at low strain level and for the
incident strains are observed to be same in all the peak stress prediction. Insignificant mismatch in the
three rocks. However, the transmitted strains are results is obtained in the strain softening region. It
different for the three different rocks. The highest may be concluded that the strain rate dependent
amount of transmitted strain is observed in granite crushable foam plasticity constitutive model
whereas the lowest amount of transmitted strain is simulates the high strain rate behaviour of rocks with
observed in sandstone. Higher amount of strain in reasonable accuracy.
the transmission bar in the case of granite is due to
elastic behaviour of granite in SHPB test and thus (2) The initial slope of the stress-strain curve
the absence of plastic dissipation. Plastic dissipation decreases with decreasing stiffness of rock. For
is present in the cases of limestone and sandstone granite, the loading and unloading paths remain same
which results in lower amount of strain transmission because granite does not exhibit elasto-plastic
in these rocks. response. However, plastic strain generates in the
cases of limestone and weak sandstone which is
Triaxial SHPB Simulation expected.
The triaxial SHPB test is simulated in two steps - (i) (3) The amount of transmitted strain decreases
application of confining pressure around the rock with decreasing stiffness of rocks. Higher amount of
8 Tanusree Chakraborty

strain in the transmission bar in the case of granite is Also, the initial stiffness of the rocks increases with
due to elastic behaviour of granite in SHPB test and increasing confining pressure.
thus the absence of plastic dissipation. Plastic
dissipation is present in the cases of limestone and Acknowledgements
sandstone which results in lower amount of strain The technical discussions with Dr. Manjit Singh,
transmission in these rocks. Director, Terminal Ballistic Research Laboratory
(4) The peak axial stress increases with (TBRL), Chandigarh had been greatly useful in the
increasing confining pressure on the rock sample. numerical simulations and results reported in this
manuscript.

References (Materials Park, OH: ASM International) 8 (2000)


13. Cai M, Kaiser P K, Suorineni F and Su K A study on the
1. Hopkinson B A method of measuring the pressure in the
dynamic behaviour of the Meuse/Haute-Marne argillite
deformation of high explosives or by the impact of bullets
Phys Chem Earth 32 (2007) 907-916
Phil Trans Roy Soc A213 (1914) 437-452
14. Li X B, Lok T S, Zhao J and Zhao P J Oscillation
2. Kolsky H An investigation of the mechanical properties
elimination in the Hopkinson bar apparatus and resultant
of materials at very high rates of loading Proc Phys Soc
complete dynamic stress strain curves for rocks Int J Rock
62 (1949) 676-700
Mech Min Sci 37(7) (2000) 1055-1060
3. Kolsky H Stress waves in solids Clarendon Press Oxford
15. Olsson W A The compressive strength of tuff as a function
(1953) pp 215
of strain rate from 10-6 to 103/sec. Int J Rock Mech Min Sc
4. Lindholm U S Some experiments with the split Hopkinson
Abs 28(1) (1991) 115-118
pressure bar J Mech Phys Solids 12 (1964) 317-335
16. Lu Y B, Li Q M and Ma G W Numerical investigation of
5. Duffy J, Campbell J D and Hawley R H On the use of a
the dynamic compressive strength of rocks based on split
torsional split Hopkinson bar to study rate effects in 1100-
Hopkinson pressure bar tests Int J Rock Mech Min Sc 47
0 aluminium ASME J Appl Mech 38 (1971) 83-91
(2010) 829-838
6. Nicholas T Material behaviour at high strain rates Imp
17. Goldsmith W, Sackman J L and Ewert C Static and
Dyn (New York: Wiley) chapter 8 1982
dynamic fracture strength of Barre granite Int J Rock Mech
7. Ellwood S, Griffiths L J and Parry D J Materials testing Min Sc 13 (1976) 303-309
at high constant strain rates J Phys E Sci Instrum 15 (1982)
18. Perkins R D, Green S J and Friedman M Uniaxial stress
280-282
behaviour of porphyritic tonalite at strain rate to 103/s Int
8. Franz C E, Follansbee P S and Wright W J New J Rock Mech Min Sc Abs 7 (1970) 527-535
experimental techniques with the split Hopkinson pressure
19. Logan J M, Handin J Triaxial compression testing at
bar 8th Int. Conf. on High Energy Rate Fabrication Pres
intermediate strain rates In: (Ed. Clark GB) Proc 12th US
Ves Pip Div ASME (San Antonio, TX, 17-21 June) ed I
Rock Mech Symp, New York, American Inst. Min. Met.
Berman and J W Schroeder, (1984)
Petr. Engg, (1971) 167-194
9. Follansbee P S The Hopkinson bar Mechanical Testing,
20. Green S J, Perkins R D Uniaxial compression tests at
Metals Handbook vol 8, 9th edn (Metals Park, OH:
strain rates from 10-4/s-04/s on three geologic materials
American Society for Metals), (1985) 198-217
In: (Ed. Gray KE) Proc 10th US Rock Mech Symp New
10. Nemat-Nasser S, Isaacs J B and Starrett J E Hopkinson York: American Institute of Mining and Metallurgical
techniques for dynamic recovery experiments Proc R Soc Petroleum Engineering, (1972) 35-54
Lond A 435 (1991) 371-391
21. Stowe R L, Ainsworth D L Effect of rate of loading on
11. Gray G T Classic split-Hopkinson pressure bar technique, strength and Youngs modulus of elasticity of rock In: (Ed:
ASME Handbook Mech Test Eval Materials Park, OH: Gray KE) Proc 10th US Rock Mech Symp New York,
ASM International) 8 (2000) American Inst. of Min. Met. Petr. Engg, (1972) 3-34
12. Gray G T, Blumenthal W R Split-Hopkinson pressure bar 22. Lindholm U S, Yeakley L M and Nagy A The dynamic
testing of soft materials ASME Handbook Mech Test Eval strength and fracture properties of dresser basalt Intl J Rock
Impact Simulation of Rocks under SHPB Test 9

Mech Min Sc Geomech Abs 11 (1974) 181-191 of concrete-like materials in a split Hopkinson pressure
23. Zhao J, Li H B, Wu M B and Li T J Dynamic uniaxial bar test Intl J Solids and Str 40 (2003) 343-360
compression tests on a granite Intl J Rock Mech Min Sc 31. Abaqus v6.11 users manual (2011), Abaqus Inc., DS
(1999) 36 273-277 Simulia, Providence, RI, USA.
24. Zhai Y, Ma G W, Zhao J H and Hu C M Dynamic failure 32. Wang Z G, Meyer L W On the plastic wave propagation
analysis on granite under uniaxial impact compressive load along the specimen length in SHPB test Exp Mech 50
Chinese J Geotech Engg (2007) 385-390 (2010) 1061-1074
25. Xia K, Nasseri M H B, Mohanty B, Lu F, Chen R and Luo 33. Frew D J, Forrestal M J and Chen W A Split Hopkinson
S N Effects of microstructures on dynamic compression pressure bar technique to determine compressive stress-
of barre granite Intl J Rock Mech Min 45 (2008) 879-887 strain data for rock materials Exp Mech 41 (2001) 40-46
26. Li X B, Hong L, Yin T B, Zhou Z L and Ye Z Y 34. Deshpande V S, Fleck N A Isotropic constitutive model
Relationship between diameter of split Hopkinson pressure for metallic foams J Mech Phys Solids 48 (2000) 1253-
bar and minimum loading rate under rock failure J Centr 1276
S Univ Tech 15 (2008) 218-223 35. Hoge L R, Shugar T A Three-dimensional finite element
27. Cho S H, Ogata Y J and Kaneko K Strain-rate dependency modelling of confinement stress for projectiles embedded
of the dynamic tensile strength of rock Intl J Rock Mech in rock - a preliminary study NCEL Tech note AD-A258
Min Sc 40 (2003) 763-777 755 (1992)
28. Blanton T L Effects of strain rates from 10-2 to 10 s-1 in 36. Brady B G H and Brown E T Rock mechanics for
triaxial compression tests on three rocks Intl J Rock Mech underground mining (Second ed.). Chapman & Hall
Min Sc Geomech Abs 18 (1981) 47-62 (1985)
29. Nemat-Nasser S, Isaacs J and Rome J Triaxial Hopkinson 37. Li Q, Crawford J and Hao H Impact and blast effects:
techniques ASME Mech Test Eval Handbook 8 (2000) 516- Theory, analysis and design Course literature for three
518 day short course, The University of Manchester, UK, 27-
30. Li Q M, Meng H About the dynamic strength enhancement 29th September 2010.

S-ar putea să vă placă și