Reinforced Concrete Slab Subjected To Soft Missile Impact
Reinforced Concrete Slab Subjected To Soft Missile Impact
Reinforced Concrete Slab Subjected To Soft Missile Impact
ABSTRACT
This article provides a study of comparisons between experimental tests and numerical analysis
results for short-term dynamic loads on reinforced concrete slabs. Meppen Tests II/4 and II/5 in the series
of tests performed during the 80’s in Germany were selected as representatives of experimental tests and
Abaqus version 6.12-3 was used for the numerical FE analysis. These two test results were provided in
the international project “Improving Robustness Assessment Methodologies for Structures Impacted by
Missiles (IRIS_2010)” and Test II/4 was a subject of the simulation in the project.
In this article, parametric study for analytical parameters such as these in the inelastic concrete
material model “Concrete Damaged Plasticity” was carried out to figure out the effect of each parameter
and to obtain a set of reasonable parameters by mainly focusing on the Test II/4. And then Test II/5 was
simulated applying the same set of parameters. The conclusions were summarized based on the
comparison of experimental and simulation results with the idea of feasibility for simulating the two
separate tests, which had different level of damage using the same material constitutive model.
INTRODUCTION
The robustness against impact load has been considering more and more important for safety
related structures in nuclear facilities especially considering aircraft crash since September 11th, 2001.
Although several empirical formulas have been developed to for the design of reinforced concrete wall,
practically, more detail evaluations are required and several commercial FE codes are commonly used.
However, material constitutive models under high inelastic damage especially for quasi brittle materials
such as concrete are still under development and benchmark analysis simulating experimental impact tests
is important to figure out its ability and sensitivity.
Meppen Test II/4 and II/5 were selected to be simulated. The missiles used in both tests were
fabricated with thin steel parts such as pipes and plates based of the same design. For the target slab, steel
reinforced concrete slabs were built. Shear reinforcement density was the only difference between the
target slabs of Test II/4 and II/5 and more severe damage was observed in Test II/5 than in Test II/4. Test
II/4 was simulated in the first phase of the international project IRIS_2010 organized by the Committee
on the Safety of Nuclear Installations (CSNI) in conjunction with the Committee on Nuclear Regulatory
Activities (CNRA), both part of Nuclear Energy Agency (NEA) of the Organization for Economic Co-
operation and Development (OECD).
For the numerical FE code, Abaqus 6.12-3 and its explicit solver Abaqus/Explicit was used.
Concrete Damaged Plasticity (CDP) was selected among three standard concrete constitutive models
available in Abaqus, because this is the only model which can simulate both compressive and tensile
plastic behaviors with the explicit solver. During the simulation of Test II/5, numerical instability
problems were experienced because of the relatively sever damage. Therefore, the automatic conversion
of finite elements to mesh-free particles which is called “Smoothed Particle Hydrodynamics (SPH)”
method was applied to deal with this problem. In this study, only standard material models incorporated
in Abaqus were used and simple models which allowed obtaining of results within a few days were built.
Material test data were basically taken from IRIS_2010 project report and the required data which was not
available from the report was supplemented by civil standards and relating articles. These principles
23rd Conference on Structural Mechanics in Reactor Technology
Manchester, United Kingdom - August 10-14, 2015
Division V
correspond to the design of new structures when only limited knowledge of material properties especially
for concrete and limited time for solution are available.
TEST DISCRIPTION
For Test II/4 and II/5, the missiles with 607 mm outer diameter and total length of 5,990 mm
were fabricated with thin RST 37-2 (S 235 G3) steel pipes and plates based on the same design (Figure
1a). The missile weight and the impact velocity were 1,016 kg and 247.7 m/s for Test II/4 and 974 kg and
234.8 m/s for Test II/5. The targets were 6,500 mm horizontal, 6,000 mm vertical and 700 mm thick steel
reinforced concrete slabs (Figure 1b). The concrete type was B35 and the steel type of reinforcement was
BSt 420/500 RK. 20 mm diameter reinforcement steel was arranged on a side where the missile hit the
target (front) with a density of 273 mm2/m and 28 mm diameter reinforcement steel was arranged on the
other side (rear) with a density of 536 mm2/m. 20 mm diameter shear reinforcement steel was arranged
with 502 mm2/m2 for Test II/4 and 126 mm2/m2 for Test II/5. This shear reinforcement steel density
difference was the only difference between Test II/4 and Test II/5 target slabs. The target slabs were
supported by the blind cylinders and tie rods. The blind cylinders and the tie rods were set for the
compression and the tensile reaction forces, respectively (Figure 1bc).
(a) Missile
!= Load cells
"= Blind cylinders
x = Tie rods
Figure 1. Meppen Test II/4 and II/5 – Missile, Target and Target Support Condition
MODEL DISCRIPTION
Each model comprised a one-quarter symmetric representation of the missile and reinforced
concrete slab. The gravity was ignored and the general contact algorithm in Abaqus/Explicit was applied.
(a) Missile
(b) Target (Front) (c) Target (Rear View) (e) Tie Rod and Plate
ఌሶ ఈ
ܨܫܦൌ ቀ ቁ (1)
ଵషర
ߙ ൌ ͲǤͲͶ െ ͲǤͲͶͲ (2)
ସଵସ
ߙೠ ൌ ͲǤͲͳͻ െ ͲǤͲͲͻ (3)
ସଵସ
and the strain rate ε -1 is in s-1 (1/second), and fy is the bar yield strength in MPa. This formulation is valid
for bars with yield stresses between 290 and 710 MPa and for strain rates between 10-4 and 225 s-1. The
steel density was set to fit the missile’s total weight in the test.
ϲϬϬ
ϱϬϬ
ϰϬϬ
^ƚƌĞƐƐDWĂ
ϯϬϬ
^ƚƌĂŝŶƌĂƚĞс
Ϭ͘ϬϬϬϭ;ƐƚĂƚŝĐͿ
ϮϬϬ
Ϭ͘Ϭϭ
ϭ
ϭϬϬ
ϭϬ
ϮϮϱ
Ϭ
Ϭ ϱ ϭϬ ϭϱ ϮϬ Ϯϱ
^ƚƌĂŝŶй
(a) Test Data (b) Input Data
The concrete material model was represented by CDP. Figure 4 shows the compressive concrete
behaviour in stress-strain relationship inputted to the model. Since no stress-strain curve was available
from the experimental test, the static compression relationship was developed using the following
relationship in Eurocode 2:
ఙ ijିఝమ
ൌ (4)
ଵାሺିଶሻఝ
where ߮ ൌ ߝ Τߝ, ߝ is the strain at peak stress, k = 1 .05 E cm × ε c1 / f cm . Ecm = 29.053 GPA was from
the test and ߝଵ ൌ ͲǤͲͲʹʹ and fcm = 38 MPa were picked up from Eurocode 2. The concrete compressive
strength of cubic specimen was provided as 37.2 MPa in the test, thus values relating to C30/C37 were
adapted. Since the equation (4) is valid only in the strain range Ͳ ൏ ߝ ൏ ߝ௨ଵ, the range over the nominal
ultimate strain ߝ௨ଵ ൌ ͲǤͲͲ͵ͷ was plotted with engineering judgment. The strain-rate effect for the
concrete compressive behaviour was taken into consideration using the following equations (L. Javier
Malvar and John E. Crawford, 1998):
ఌሶ ଵǤଶఈೞ
݂ Ȁ݂௦ ൌ ቀ ቁ for ߝሶ ͵Ͳି ݏଵ (5)
ఌೞሶ
Τ
ఌሶ ଵ ଷ
ൌ ߛ௦ ቀ ቁ for ߝሶ ͵Ͳି ݏଵ (6)
ఌೞሶ
where fc is dynamic compressive strength at ε , fcs is static compressive strength at ߝ௦ሶ ൌ ͵Ͳ ൈ ͳͲି,
݂Τ݂௦ is compressive strength DIF, ε is strain rate in the range of 30 ×10 to 300 s-1, logγ = 6.156α –
−6
ϭϰϬ ϭϰϬ
ϯϬϬƐͲϭ ϯϬϬƐͲϭ
ϭϮϬ ϭϬϬƐͲϭ ϭϮϬ ϭϬϬƐͲϭ
ϯϬƐͲϭ ϯϬƐͲϭ
ϭϬϬ ϭƐͲϭ ϭϬϬ ϭƐͲϭ
ϭĞͲϮƐͲϭ ϭĞͲϮƐͲϭ
^ƚƌĞƐƐDWĂ
^ƚƌĞƐƐDWĂ
ϴϬ ϭĞͲϰƐͲϭ ϴϬ ϭĞͲϰƐͲϭ
^ƚĂƚŝĐ ^ƚĂƚŝĐ
ϲϬ ϲϬ
ϰϬ ϰϬ
ϮϬ ϮϬ
Ϭ Ϭ
Ϭ Ϭ͘Ϭϱ Ϭ͘ϭ Ϭ͘ϭϱ Ϭ͘Ϯ Ϭ͘Ϯϱ Ϭ͘ϯ Ϭ͘ϯϱ Ϭ Ϯ ϰ ϲ ϴ ϭϬ ϭϮ ϭϰ ϭϲ ϭϴ ϮϬ
^ƚƌĂŝŶй ^ƚƌĂŝŶй
(a) Developed by Equation (4) (b) Whole Input Data
Fracture energy cracking model was applied for the concrete tensile behavior. Since only the tensile
strength value 4.8 MPa was availabel from the material test, a linear loss of strength after cracking was
assumed. Abaqus Manual refers to typical values for the Fracture Energy, Gf = 40 N/m and 120 N/m for
typical construction concretes which have approximate compressive strengths 20 MPa and 40 MPa,
respectively. However, it is described in Eurocode 2 that the concrete which has a compressive strenght
about 40 MPa has a tensile strength only around 3.0 MPa. Due to the significant difference between these
tensile strengths from the test and Eurocode 2, the Fracture Energy Gf was studied as a parameter with
the values 120, 160 and 200 N/m. There are several parameters to define concrete plastic behavior in CDP,
however, defaut values were applied for the most of the parameters because of the insuficiency of the
material test data available in IRIS_2010 report. Only Dilation Angle for the flow potentioal was studied
with the values 30, 40 and 50 degrees, because no default value is specified in Abaqus Manual.
For the reinforcement steel, Classical Metal Plasticity was used. The strain-stress relationship of
the reinforcement steel was provided with mechanical calacteristic values, Elastic limit σe = 430 MPa,
Plastic limit σp = 500 MPa at plastic strain 0.2 % and Upper resistance σmax = 620 MPa at plastic strain
3 % (Figure 5a). From these data, the stress-strain curve was plotted considering the true stress – true
strain relationship with strain-rate effect (Figure 5b). The equ
ations (1) to (3) were also used for the reinforcement steel.
ϴϬϬ
ϳϬϬ
ϲϬϬ
ϱϬϬ
^ƚƌĞƐƐDWĂ
^ƚĂŝŶƌĂƚĞс
ϰϬϬ
ϴ͘ϱͬƐ
ϯϬϬ Ϯ͘ϬͬƐ
Ϭ͘ϬϮͬƐ
ϮϬϬ
Ϭ͘ϬϬϭͬƐ;ƐƚĂƚŝĐͿ
ϭϬϬ
Ϭ
Ϭ Ϯ ϰ ϲ ϴ ϭϬ
^ƚƌĂŝŶй
(a) Test Data (b) Input Data
SIMULATION RESULTS
Missile Model
The missile model was calibrated before simulating its impact against the target slab. Table 2
shows the analysis cases. The missile model was impacted to a rigid wall and the displacement and
velocity of the rear of the missile (Figure 6) and its deformation shape (Figure 7) were compared with the
test results. It was assumed that, although the test results were of its impact against the RC slab, the
impact to the rigid wall could represent the missile’s similar response, because the RC slab was not
damaged severely and could be considered relatively stiff in Test II/4. Since Case M4 showed relatively
close results to the test results among these cases, this missile model was decided to be used to simulate
Meppen Test II/4 and II/5.
Ϯϳϱ ϲ
ϮϱϬ ϱ
ϮϮϱ ϰ
ϮϬϬ ϯ
Ϯ
ϭϳϱ
ŝƐƉůĂĐĞŵĞŶƚŵ
ϭ
sĞůŽĐŝƚLJŵͬƐ
ϭϱϬ
Ϭ
ϭϮϱ
Ͳϭ
ϭϬϬ
ĂƐĞDϭ ͲϮ
ϳϱ
ĂƐĞDϮ Ͳϯ
ϱϬ Ͳϰ
ĂƐĞDϯ
Ϯϱ Ͳϱ
ĂƐĞDϰ
Ϭ Ͳϲ
ͲϮϱ Ͳϳ
Ϭ ϭϬ ϮϬ ϯϬ ϰϬ
dŝŵĞŵƐ
(a) Test Result (Meppen Test II/4) (b) Case M1-M4
(a) Test result (b) Case M1 (c) Case M2 (d) Case M3 (e) Case M4
Ϭ͘ϳŵ Ϭ͘ϳŵ
tϮ tϭ
tϲ tϱ
tϴ tϳ
(a) Test Result (W3 – W6) (b) Test Result (W1, W2, W7 and W8)
ϱ ϱ
ĂƐĞϭ ĂƐĞϮ ĂƐĞϯ ĂƐĞϰ ĂƐĞϱ ĐĂƐĞϲ
ŝƐƉůĂĐĞŵĞŶƚĐŵ
Equivalent to W3-W6
ŝƐƉůĂĐĞŵĞŶƚĐŵ
ϰ ϰ
Equivalent to W3-W6
ϯ ϯ
Equivalent to W1, W2, W7 and W8 Equivalent to W1, W2, W7 and W8
Ϯ Ϯ
ϭ ϭ
Ϭ Ϭ
Ϭ ϮϬ ϰϬ ϲϬ ϴϬ ϭϬϬ Ϭ ϮϬ ϰϬ ϲϬ ϴϬ ϭϬϬ
dŝŵĞŵƐ dŝŵĞŵƐ
(c) Case 1 – Case 3 (d) Case 4 – Case 6
ϵϬϬ
ϴϬϬ
<ϴ
ϳϬϬ
<ϳ
ϲϬϬ
ZĞĂĐƚŝŽŶ&ŽƌĐĞŬE
<ϲ
ϱϬϬ
<ϱ
ϰϬϬ
ϯϬϬ
ϮϬϬ
ϭϬϬ
Ϭ
ͲϭϬϬ Ϭ ϮϬ ϰϬ ϲϬ ϴϬ ϭϬϬ
dŝŵĞŵƐ
(a) Test Result (b) Case 6
23rd Conference on Structural Mechanics in Reactor Technology
Manchester, United Kingdom - August 10-14, 2015
Division V
Figure 9. Meppen Test II/4 Reaction Force at Load Cells K5 – K8
&ƌŽŶƚ
^ĞĐƚŝŽŶϭ ^ĞĐƚŝŽŶϮ ^ĞĐƚŝŽŶϭ
ZĞĂƌ
(a) Crack Pattern (Rear & Sides) (c) Case 6 Damege (t = 100 ms)
(a) Crack Pattern (Rear) (b) Crack Pattern (Section) (c) Damage (t = 30 ms)
23rd Conference on Structural Mechanics in Reactor Technology
Manchester, United Kingdom - August 10-14, 2015
Division V
Figure 11. Meppen Test II/5 Target Response
Đŵ Đŵ
ϯϬ ϲ
ϮϬ ϰ
ϭϬ Ϯ
ϭ
ŵƐ ŵƐ ŵƐ
Ϭ ŵƐ
Ϯϱ Ϭ
Ϯϱ
Ϭ Ϯϱ Ϭ Ϯϱ
(a) Test and Simulation Results (W3 – W6) (b) Test and Simulation Results (W2, W7 and W8)
Figure 12. Meppen Test II/5 Displacement (measurement positions were the same as in Test II/4)
CONCLUDIONS
Meppen Test II/4 and II/5 which had different level of damage were simulated with the
commercially available finite element code Abaqus. The material constitutive models and their
parameters were defined according to the Test II/4 results and using these parameters, the response and
damage observed in Test II/4 were reasonably predicted. On the other hand, the simulation of Test II/5
using the same material models used for Test II/4 experienced numerical instability. After SPH method
was additionally applied to the concrete material model, the analysis was successfully done. However, the
punching failure in Test II/5 was not captured and the damage could not be reproduced precisely enough.
Therefore, the simulation of the two tests which had different damage levels using the same set of
material constitutive models was not achieved reasonably within this study. The followings were found as
technical difficulties in the simulation of Test II/5 using SPH method: (1) the only absolute principal
strain and stress can be specified for the thresholds respectively for strain-based and stress-based criteria
to convert finite element to SPH particle but the separate thresholds for compressive and tensile principal
strain and stress might be needed to predict the response more accurately, and (2) the embedded beam
elements don’t have contact capability and after they are released from the host solid element due to SPH
conversion, they cannot contact other elements. Apart from problems with SPH, there are some more
difficulties - the material test data was not sufficiently available to define the material parameters and the
material constitutive models available in the Abaqus without incorporating user material models are not
sufficient enough to simulate impact against reinforced concrete slab with high level of damage.
From this study it seems that it is still needed to consider different approaches to simulate
different levels of damage in reinforced concrete slabs caused by impact. For a new design, it is important
to benchmark the capability of used code and material models by simulating a set of experimental impact
tests covering expected range of structure response and level of damage.
REFERENCES
Abaqus Manual (2012), Version 6.12-3, Dassault Systemes.
Improving Robustness Assessment Methodologies for Structures Impacted by Missiles (IRIS_2010),
Final Report, NEA/CSNI/R(2011)8.
EUROPEAN STANDARD, EN 1992-1-1, Eurocode 2: Design of concrete structures - Part 1-1: General
rules and rules for buildings.
L. Javier Malvar and John E. Crawford (1998), “DYNAMIC INCREASE FACTORS FOR STEEL
REINFORCING BARS,” Twenty-Eighth DDESB Seminar Orlando, FL, August 98.
L. Javier Malvar and John E. Crawford (1998), DYNAMIC INCREASE FACTORS FOR CONCRETE,
Twenty-Eighth DDESB Seminar Orlando, FL, August 98.