You are on page 1of 10

Magazine of Concrete Research Magazine of Concrete Research, 2014, 66(1), 50–59

Volume 66 Issue 1 http://dx.doi.org/10.1680/macr.13.00194


Paper 1300194
Gauged reactive confinement cell for Received 04/07/2013; revised 16/09/2013; accepted 23/09/2013
determination of concrete triaxial properties Published online ahead of print 08/11/2013

Sheridan and Pullen ICE Publishing: All rights reserved

Gauged reactive confinement


cell for determination of
concrete triaxial properties
A. James Sheridan Andrew D. Pullen
Principal Scientist, Air & Weapons Systems Department, Dstl Portsdown Research Fellow, Department of Civil and Environmental Engineering,
West, Fareham, UK Imperial College London, London, UK

A simple experimental method for determining a set of triaxial properties for concrete and other geologically based
materials such as rock and soil is described. The method involves the uniaxial loading of a cylindrical material
specimen that is radially confined by a steel tube. The steel tube is lined to reduce friction and is externally strain
gauged to determine axial and circumferential strains. Elastic thick-cylinder theory is then applied to determine the
radial stresses and strains induced in the specimen during uniaxial loading. The specimen load conditions are
approximately those of uniaxial strain, with lateral strains being two orders of magnitude lower than the axial
strains. The test is capable of directly outputting pressure against volumetric strain and stress difference against
mean effective stress in a form that, within a single test, is suitable for evaluation of the parameters for hydrocode
material models that can be employed in military equipment performance assessments, where triaxial compressive
stresses can be significantly greater than those typically resulting from structural service loads. This paper describes
specimen preparation, the test set-up, the test process and associated data capture, the theory of the stress analysis
and the application of the test data to simple hydrocode material models. A database of concrete, rock and soil
properties has been generated using the gauged reactive confinement (GREAC) cell and the method has also been
compared with the more traditional active confinement triaxial test method. The issues of strain rate are also
addressed.

Notation Introduction
a inside radius of GREAC cell steel confining The concept for the gauged reactive confinement (GREAC) cell
cylinder (Figure 1) was initially proposed as a method to provide the static
a0 initial undeformed inside radius of GREAC cell component of the pressure–volumetric strain behaviour of con-
steel confining cylinder crete under plain compressive shock wave conditions (Figure 2)
b outside radius of GREAC cell steel confining in order to investigate concrete response to high explosive
cylinder detonation (Sheridan, 1990; Sheridan et al., 1989). However,
E Young’s modulus of steel during the early development of the laboratory technique at
F axial load on GREAC cell specimen Imperial College London (ICL) (Kotsovos and Perry, 1984, 1986),
P pressure within GREAC cell specimen it soon became apparent that it had greater utility as a simple
1 axial strain in GREAC cell specimen low-cost triaxial loading technique. Specifically, the method
3 lateral strain in GREAC cell specimen allowed access to much higher stress states than typical active
Æ axial strain in GREAC cell steel confining cylinder confinement test cells, such as the Hoek cell (Franklin and Hoeck,
h hoop strain in GREAC cell steel confining cylinder 1970), making it more relevant to military research. However, just
 Poisson’s ratio of steel as important from a cost and time viewpoint, the more rigid
1 axial stress in GREAC cell specimen confinement enabled a continuous failure response to be evaluated
3 lateral stress in GREAC cell specimen without the test specimen going into unstable collapse.
Æ axial stress in GREAC cell steel confining cylinder
h hoop stress in GREAC cell steel confining cylinder The capability of the GREAC cell method to rapidly generate
r radial stress in GREAC cell steel confining cylinder continuous data for pressure–volumetric strain and stress differ-
y yield stress in GREAC cell specimen ence–pressure (Figure 3) readily lends it to exploitation in
 shear stress within GREAC cell specimen hydrocode modelling, where data are used to specify the equation
 shear strain within GREAC cell specimen of state and the strength (or failure) model (Sheridan, 1991;

50
Downloaded by [ Universidade Federal de Pernambuco] on [22/11/17]. Copyright © ICE Publishing, all rights reserved.
Magazine of Concrete Research Gauged reactive confinement cell for
Volume 66 Issue 1 determination of concrete triaxial
properties
Sheridan and Pullen

Applied load σ1
Concrete specimen

Rayleigh line
Steel confining a
cylinder Hugoniot curve

Wave normal stress


b

Static curve
Heat
σα
σr(a)

σh(r)

σh(r) Work

σr(b)
σα Volumetric strain

Figure 2. Pressure–volumetric strain response of concrete under


planar compressive shock

(a) Sheridan et al., 1992). It can also be correlated to standard


unconfined compression testing in that the specimen follows the
uniaxial stress path to this initial point of failure (point 1 in the
plots shown in Figure 3). The specimen then continues to fail
under multiaxial confinement as further load is applied. Clearly,
there is a strain rate issue that needs to be addressed, which will
be done later in the paper, but such data have been successfully
employed in predicting the performance of military weapons
against hardened concrete structures (hard targets) (e.g. Clegg et
al., 1997).

Up to the end of 2009, the GREAC cell test method was


performed at ICL using a 3000 kN Amsler uniaxial loading
machine, providing the capacity for a maximum axial stress of
680 MPa on a 75 mm diameter cylindrical specimen. In practice,
the stresses were constrained to a maximum of about 500 MPa
axially by the elastic properties of the axial loading platens, and
to approximately 180 MPa radially by those of the confining
cylinder with an outer diameter of 100 mm. During the early
development of the method, this device was used to generate a
large database of materials, including a range of coarse aggregate
materials, varying mix proportions and free water/cement ratios,
the rock materials themselves that were used to produce the
coarse aggregates and a range of granular materials, including
gravel, sand and clay (Pullen and Newman, 1993).

Since early 2010, the GREAC cell test has been carried out using
an Instron SATEC 3500 kN servo-hydraulic test machine. The
primary advantage of this more up-to-date apparatus is the ability
to accurately control loading/unloading rate and transition points.
(b)

Figure 1. GREAC cell: (a) schematic illustration; (b) in Amsler GREAC cell design
uniaxial load device The original GREAC cell design comprised a cylindrical high-
strength EN24T steel hollow cylinder of 150 mm length, 75 mm

51
Downloaded by [ Universidade Federal de Pernambuco] on [22/11/17]. Copyright © ICE Publishing, all rights reserved.
Magazine of Concrete Research Gauged reactive confinement cell for
Volume 66 Issue 1 determination of concrete triaxial
properties
Sheridan and Pullen

300
geologic material specimen only, together with the external axial
and hoop strains of the confining cylinder (Figure 4). In practice,
the specimen response is almost pure uniaxial strain, with the
lateral strain being two orders of magnitude lower than the axial
5 strain. However, the small lateral strains measured on the outside
200
of the confining cylinder are sufficient, and necessary, to deter-
Pressure: MPa

mine the radial stresses and strains at the sample–confinement


6
interface by employing elastic thick-cylinder theory. The maxi-
4
mum specimen stresses and strains that can be achieved during a
100 test are therefore constrained by the need to ensure that the
confining cylinder behaves elastically.
7
2
Preparation and details of apparatus used/
1 3 required for GREAC cell testing
0 The cylindrical confinement is simply a hollow steel cylinder,
0 0·02 0·04 0·06 about 2 mm longer than the intended specimens. The inner
Volumetric strain
(a) surface is polished to minimise frictional axially aligned inter-
ference with the specimen. The inner diameter is ideally just
Uniaxial stress path Failure envelope greater than the intended specimen diameter to allow for the
250
thickness of the polythene lining sheet and to facilitate installa-
4 5 tion of the specimen in the cell.

150
A diametrically opposed pair of 0/90 strain-gauge rosettes are
installed on the outer surface at mid-height, with one gauge
Stress difference: MPa

element aligned parallel to the cylinder axis and the other


circumferential (Figure 4(b)).
50 1
The axial loading platens are plain cylinders with their ends
2 6
ground flat, comparable with those used for uniaxial compressive
0 100 200 300 testing, except that their diameter must allow a close running fit
3 Pressure: MPa
⫺50 inside the confining cylinder. The axial load applied to the
specimen is monitored by a compression load cell, typically
7
either placed immediately below the lower axial platen or integral
to the test machine.
⫺150
(b) Axial shortening of the specimen is monitored by a diametrically
opposed pair of displacement transducers. Shortening of the axial
Figure 3. GREAC cell continuous data for (a) pressure–volumetric loading platens and other components is determined in advance of
strain and (b) stress difference–pressure (Sheridan et al., 1992) testing by loading a plain steel cylinder and is used in the
subsequent data analysis (a compliance correction). Potentiometer-
type transducers have generally been used, but linear variable
internal diameter and 12.5 mm thickness. Three samples of the differential transformers (LVDTs) would also be appropriate. The
steel were tested in tension to determine the Young’s modulus E displacement data generated by the test rig can be used, but this
(¼ 202.95 GPa), Poisson’s ratio  (¼ 0.288) and yield stress y tends to increase the magnitude of the required compliance
(¼ 590 MPa), which were used in the data analysis. The steel correction to the extent that the resulting specimen shortening data
confining cylinder is lined with grease and polyethylene sheet to are degraded and therefore less accurate.
reduce friction at the interface between the specimen and the
confinement. The material specimens were originally cast into the The strain gauges, load cell and displacement transducers are
lined cylinder or cored and potted into it using an epoxy resin connected to an appropriate data acquisition and recording
filler. Current practice involves casting or coring cylindrical system. A Datascan modular system (Measurement Systems Ltd,
specimens sized to achieve a push fit in existing strain-gauged UK) was used for the tests carried out at ICL, using one Datascan
confining cylinders. 7220 module for all transducer voltage signals and one Datascan
7221 module for direct monitoring of strain gauges.
The method is dependent on measuring the load and displacement
of plain cylindrical platens, which are applied to the end of the Up until 2009, a DOS-Basic graphical software, developed at

52
Downloaded by [ Universidade Federal de Pernambuco] on [22/11/17]. Copyright © ICE Publishing, all rights reserved.
Magazine of Concrete Research Gauged reactive confinement cell for
Volume 66 Issue 1 determination of concrete triaxial
properties
Sheridan and Pullen

ICL, was used to monitor, process, display and store data during
the GREAC cell test. The data acquisition rate was approximately
one sample per second. The software includes a pacer to ensure
that a target axial strain rate (typically 103 /s) was approximately
achieved by manual control of the valves on the test rig. Since
2010, Visual Basic graphical software was developed and used
for data acquisition and processing, with all rate control being
carried out by the Instron control hardware and software.

Conditioning (pre-stressing) of the


confinement cylinder
Analysis of the data from a test relies on the steel material of the
confining cylinder behaving elastically. Conditioning, therefore,
involves increasing the internal stress (the radial stress at the
specimen interface) to the extent that the steel on the outer
surface of the cylinder reaches its elastic limit. The material
within the cylinder will then have experienced permanent plastic
deformation. Any new confining cylinder should be subjected to
this level of loading for a few cycles before being used for
material testing. This can be achieved during the calibration
process or by sacrificing one specimen.

Calibration of the confining cylinder


The relationship between the strains monitored on the outer
surface of the confining cylinder and the radial stress at its inner
surface can be predicted using thick-cylinder theory. This requires
(a) accurate knowledge of the dimensions of the cylinder and its
elastic properties. This can be supplemented by empirical calibra-
tion. Ideally, this would be achieved by applying a fluid pressure
to the inner surface of the cylinder. More practically, a very wet
and therefore weak (in shear) clay specimen was sealed into the
cell and subjected to testing. If the material shear strength is
assumed to be zero (negligible) then the radial stress will be
equal to the axial applied stress. This assumption can be checked
by observation of the difference in strain gauge response when
loading and unloading.

Test procedures
Specimen preparation and installation in the GREAC cell
Concrete specimens have become available for GREAC cell
testing from a variety of sources. During the initial development
and use, concrete was cast directly into the confining cylinders,
which were first pre-lined with two layers of greased polythene
sheet.

More recently, concrete has been cast into a set of cylinder


moulds known to produce specimens of a diameter that will suit
an existing conditioned strain-gauged confinement cylinder. Al-
ternatively, cores have been taken from blocks cast for a specific
purpose, or from existing structures and trials targets, preferably
(b) using a core barrel known to produce cores of a predictable and
Figure 4. General set-up (a) and external strain gauges (b) of uniform diameter. After an appropriate period of curing, the
GREAC cell cylindrical specimens are sawn to length and their planar ends
ground flat and parallel, normal to the cylinder axis.

53
Downloaded by [ Universidade Federal de Pernambuco] on [22/11/17]. Copyright © ICE Publishing, all rights reserved.
Magazine of Concrete Research Gauged reactive confinement cell for
Volume 66 Issue 1 determination of concrete triaxial
properties
Sheridan and Pullen

A typical specimen installation comprises the following steps. A set of two Instron test methods, which will not exceed this
maximum axial force, are then created. One of these tests repeats
(a) Ensure that the test specimen will fit in the confinement the loading of the first test, but precisely stops loading at the
cylinder and is slightly (2 mm) shorter; the fit should be a maximum force and then unloads to zero force; this load cycle is
close but easy running fit. repeated three times. The second test method will load/unload to
(b) Wrap an oversized piece of polythene sheet around the an increasing series of axial forces up to the maximum force.
specimen so that there is an overlap in the circumferential
direction and at least a specimen diameter of excess material The rate of displacement for all tests is calculated to result in an
at each end. axial strain rate of approximately 103 /s. Signals from all of the
(c) Cut the polythene in an axial direction along the overlap so load cells, strain gauges and transducers are recorded at intervals
that extraneous sheet can be removed to leave one piece of 0.2 s. These are saved by the controlling PC as the test
covering the circumferential surface. progresses. These data are also processed (in real time) to
(d ) Apply a single piece of thin adhesive tape along the join, produce estimates of the stresses and strains in the specimen and
beyond the ends of the specimen. confining cylinder, which are represented graphically on the PC
(e) At one end, wrap the excess sheet over the planar end and screen (Figure 5). Key features of this representation, which may
tape in place, then cut off any material that extends beyond be used by the operator to assess the success of the test, include
the circumference. This will be the end pushed into the
confinement. j an indication of the initial ‘unconfined’ failure, which is
( f ) Wipe a grease or similar viscous lubricant over the inner comparable with the unconfined cylinder strength of the
surface of the confining cylinder, then wipe most of this concrete (point 1 in Figure 3)
away. A lubricant known not to react with the polythene j a record of stress and strain components (Figure 5)
lining sheet should be chosen. j a record of load histories (e.g. unload/reload)
(g) Push the wrapped end of the specimen into the confining j identification of limitations being approached (confinement
cylinder. Ideally, this will be achieved using only manual yield, maximum stresses).
force and will be such that the specimen does not slip under
its own weight. If necessary, minimal mechanical force can Analysis of gauge measurements
be used. If the polythene sheet is observed to be damaged The stresses and strains within the GREAC cell material speci-
during emplacement, then it is likely that the specimen will men are determined from the load and displacement of the end
need to be reduced in diameter before testing; this can be platens, the external strains in the steel confining cylinder and the
achieved by uniformly abrading the circumferential surface. application of elastic thick-cylinder theory. The steel confining
(h) If the specimen appears loose, additional sheets of polythene cylinder has an internal radius a and an external radius b. Starting
can be added but these then start to become a significant part from the generalised Hooke’s law and assuming a zero stress at
of the specimen. the outer surface (r (b) ¼ 0), the required stress components on
(i) When the specimen is correctly positioned in the confining the outside and inside of the steel confining cylinder are
cylinder, excess polythene sheet is cut away, ensuring that the (Sheridan, 1991)
planar surfaces of the specimen are clear of polythene.
( j) Place the specimen and cylinder on the lower platen. If the E
confining cylinder drops under its own weight, use notched  h (b) ¼ [h (b) þ Æ ]
1: 1  2
balsa sticks/rods to hold it in place.
(k) Place the upper platen on top of the specimen and a spherical
seating above this. Adjust the pair of displacement
transducers if necessary, to ensure they will remain a2  b2
operational throughout the test. 2:
 r (a) ¼  h (b)
2a2
Performance of test
The following description of the test conduct refers to current
practice – that is, using modern PC-controlled equipment. A a2 (1  ) þ b2 (1 þ )
typical test series will consist of three tests on three specimens of  Æ (a) ¼
a2 (1 þ )(1  2) þ b2 (1 þ )
the same concrete. The first test will be conducted with the rate " #
of displacement controlled by the Instron controller, but with end/ 2a2
transition points set (in real time) by the operator when one or 3 E Æ þ 2 h (a)
3: a (1  ) þ b2 (1 þ )
other of the apparatus stress limits is reached. This determines
the maximum axial force that can be applied to the specimen
without causing inelastic strains in the end platens or the
confining cylinder. The associated deformed inside radius a is calculated from

54
Downloaded by [ Universidade Federal de Pernambuco] on [22/11/17]. Copyright © ICE Publishing, all rights reserved.
Magazine of Concrete Research Gauged reactive confinement cell for
Volume 66 Issue 1 determination of concrete triaxial
properties
Sheridan and Pullen

Figure 5. Real-time monitoring of stresses and strains

 
a  h (b) b2 The radial stress r (a) within the steel (Equation 2) is compres-
¼ (1 þ ) þ (1  ) sive and balanced by a lateral stress of equal magnitude within
a0 2E a2
the material specimen
[a2 (1  ) þ b2 (1 þ )]Æ  22 a(a  a0 )
 þ1
4: a2 (1 þ )(1  2) þ b2 (1 þ )
6:  3 ¼  r (a)

Equation 4 is cubic in a. However, because – in practice – the


first two terms on the right-hand side are much smaller than the
The parameters required for determining hydrocode material
last term (i.e. unity), a first estimate of a can be produced by
models are pressure, yield stress, shear stress and shear strain,
substituting a0 for a in the right-hand terms and using this to
given respectively as
recalculate Equation 4. The procedure shows very rapid conver-
gence.

1
The axial strain Æ in the steel is also compressive, though it acts P ¼ ( 1 þ 2 3 )
7: 3
against the force F transmitted by the platens directly onto the
specimen, leading to an axial stress 1 within the specimen given by

F b2  a2
1 ¼   Æ (a)  y ¼ j 1   3 j
5: a2 a2 8:

55
Downloaded by [ Universidade Federal de Pernambuco] on [22/11/17]. Copyright © ICE Publishing, all rights reserved.
Magazine of Concrete Research Gauged reactive confinement cell for
Volume 66 Issue 1 determination of concrete triaxial
properties
Sheridan and Pullen

1 250
9:  ¼ j 1   3 j
2
200

Pressure: MPa
  150
 1 1  1
 ¼  2 tan
10: 2 1 þ 3
100

50
The above data analysis is undertaken in an Excel spreadsheet.
0
Application of data to hydrocode material models 0 0·02 0·04 0·06
The analysed data can be applied directly to hydrocode modelling Volumetric strain
(a)
methods – similar to finite-element techniques – and used to
undertake numerical calculation, or simulations, of weapons 250
effects against concrete. Example data for two samples of
concrete specimen DRA9254 are shown in Figure 6 by the 200
Stress difference: MPa

continuous and dotted curves. Example hydrocode data are shown


as the thick dashed lines connecting the reduced set of datapoints
150
– and are also generated within the Excel spreadsheet.

The authors’ principal direct experience of hydrocodes lies mainly 100


in the use of Autodyn; the GREAC cell data can be applied very
readily to the simplest of concrete material models available in 50
this code (i.e. the porous equation of state and the Drucker–
Prager strength model). The tabulated porous equation of state is
0
derived from the plot of pressure against volumetric strain (Figure
0 100 200 300
6(a)) and the Drucker–Prager strength model is determined from Pressure: MPa
the plot of stress difference (yield stress) against pressure (Figure (b)
6(b)). The slope of shear stress against shear strain, during 300
unloading, is used to quantify the shear modulus (Figure 6(c)). It
250
should be noted that these data are supplemented with a tensile
Stress difference: MPa

failure strength, typically derived from the Brazilian indirect 200


tension test, or from any appropriate tensile strength test. 150

100
Materials database
A wide range of concretes and geologically based materials have 50
been tested in the GREAC cell, including rocks, soils, bricks and 0
adobe materials. A much wider range of other materials (e.g. 0 0·02 0·04 0·06
⫺50
graphite, foamed plastic, simulant energetic materials) has also
been tested, typically when the end user has been able to obtain ⫺100 Strain difference
cylindrical specimens and has needed to populate a predomi- (c)
nantly compressive hydrocode material model.
Figure 6. Hydrocode material model data extracted from plots of
In the early development of the technique in the early 1990s, a (a) pressure–volumetric strain, (b) stress difference–pressure and
database was created for use by the Defence Research Agency (c) shear stress–strain for concrete DRA9254
(DRA) (Pullen and Newman, 1993). A range of concrete mix
designs was specified by the designator DRA92XY, where X
represents the aggregate materials (0–8 in Table 1) and Y repre- cell test method by comparing two samples of concrete
sents the mix number (1–6 in Table 2). The database also includes DRA9254. Figure 7 shows the variation in material properties
tests on the mortars, aggregates and solid rock materials that were that arises for differing mix numbers DRA925Y.
used in the concrete formulations, as well as a range of soils.
Comparison with active confinement cells
The mix design used as the baseline for plots in this paper is The GREAC cell utilises reactive confinement to perform multi-
DRA9254. Figure 6 demonstrates the repeatability of the GREAC axial compression (1 . 2 ¼ 3 ) of geologically based material

56
Downloaded by [ Universidade Federal de Pernambuco] on [22/11/17]. Copyright © ICE Publishing, all rights reserved.
Magazine of Concrete Research Gauged reactive confinement cell for
Volume 66 Issue 1 determination of concrete triaxial
properties
Sheridan and Pullen

250
X Aggregate
DRA9251
1 Flint DRA9253
200
2 Portland limestone DRA9254
3 Plas Gwilyn limestone

Pressure: MPa
4 Trevone granite 150
5 Budenwood granodiorite
6 Cragmill dolerite
7 Lytag 100
8 Liapor

Table 1. Aggregates used in concretes designated DRA92XY 50

0
Free water/cement ratio Mix number Y 0 0·01 0·02 0·03 0·04 0·05 0·06
Volumetric strain
Coarse aggregate by weight: %
(a)
300
DRA9251
40 50 60
DRA9253
250
0.3 — 1 — DRA9254
0.4 — 2 — 200
Stress difference: MPa

0.5 6 3 5
0.6 — 4 —
150
Table 2. Mix number Y used for concretes designated DRA92XY
100

50
specimens. Active confinement cells, such as the Hoek cell, used
for testing cylindrical specimens, perform a similar multiaxial
0
compression loading. However, unlike active cells, the GREAC 0 50 100 150 200 250
cell is limited to approximately uniaxial strain conditions and the Pressure: MPa
⫺50 (b)
specimen does not go into unstable collapse once the triaxial
failure surface has been reached. The latter point makes the
300 DRA9251
GREAC cell a low-cost solution to producing a continuous failure
surface up to the elastic limits of the steel confining cylinder. The DRA9253
uniaxial strain constraint is worthy of more detailed considera- 250
DRA9254
tion.
200
Stress difference: MPa

It was found that a concrete specimen compressed within a thin


steel cylinder could sustain a greater stress difference for a given 150
mean effective stress, or pressure, than for the thick steel cylinder
of the GREAC cell (Kinoshita, 1992). It was conjectured that 100
allowing the concrete specimen to laterally dilate resulted in a
more supportive stress redistribution within material components. 50

This was investigated further by subjecting concrete DRA9254 to


0
a series of Hoek cell tests. In such tests, the specimens are often 0 0·01 0·02 0·03 0·04 0·05 0·06 0·07
initially loaded hydrostatically (1 ¼ 2 ¼ 3 ) to a predefined
⫺50 Strain difference
pressure and the end platens are then further loaded while holding (c)
the lateral stresses constant (1 . 2 ¼ 3 ). Figure 8 shows the
results of two Hoek cell tests (each of which were terminated at Figure 7. Specimen variability for DRA925Y concrete mixes:
specimen failure) compared with the GREAC cell results for the (a) pressure–volumetric strain; (b) yield stress–pressure; (c) shear
same concrete. Clearly, the GREAC cell failure surface is not the stress–strain
ultimate failure surface.

57
Downloaded by [ Universidade Federal de Pernambuco] on [22/11/17]. Copyright © ICE Publishing, all rights reserved.
Magazine of Concrete Research Gauged reactive confinement cell for
Volume 66 Issue 1 determination of concrete triaxial
properties
Sheridan and Pullen

200
The Defence Evaluation and Research Agency (DERA) under-
took a series of dynamic GREAC cell tests using a DARTEC
175 350 kN high rate load device. This demonstrated that the stress
GREAC-A difference against pressure results were not sensitive to strain
150 GREAC-B rates up to 10/s. Further to this, static GREAC cell data were
Active confinement 1 used to characterise the concrete specimen used in a split
Active confinement 2 Hopkinson pressure bar. The static data were successfully used to
Stress difference: MPa

125
predict the peak stress transmitted through the specimen (Pullen
et al., 1993). Nevertheless, there are limited data to suggest that
100 there may be a true material strain rate effect at the much higher
rates associated with plate impact tests (,104 /s) – possibly
associated with internal energy states.
75

The effect of strain rate on the pressure–volumetric strain


50 relationship has been investigated using thicker and stronger steel
confining cylinders (Neal, 2012; Vogler et al., 2007). This
25 research did not utilise the data from any external strain gauging
of the confining cylinder and assumed pure uniaxial strain. The
cells were also of a much smaller diameter (,30 mm) to achieve
0
0 25 50 75 100 125 150 much higher stresses. However, they were used to compare static
Pressure: MPa data with those generated dynamically by plate impact using a
gas gun. Idealised brittle granular materials employing soda-lime
Figure 8. Comparison of Hoek cell and GREAC cell failure glass microspheres showed no strain rate effect, whereas a quartz
surfaces for concrete DRA9254 sand did reveal a significant strain rate effect (Neal, 2012).

Exploitation
When considering the application of triaxial data to the hydro- GREAC cell data for pressure–volumetric strain can be applied
code modelling of weapons effects, it should be noted that the to the equation of state and stress difference–pressure curves
nature of the concrete response is determined by the geometry of used to describe the constitutive properties, or strength model,
the propagated stress wave of concrete for direct application to hydrocode modelling.
Experience suggests that the data are suitable to describe the
j a plane wave for a large stand-off explosion compressive behaviour of concrete at strain rates up to at least
j a spherical wave for a close-in explosion 102 /s. In practice, the data have been successfully used to
j a cylindrical wave for a target penetrator. predict warhead penetration of concrete and other geologically
based target materials at velocities up to about 300 m/s (Clegg
Clearly, there is a need to better understand the relationship et al., 1997). In particular, hydrocodes have matched the
between the failure surface and the degree of lateral dilation transients recorded by on-board accelerometers within about
associated with the geometry of the applied load. 10% and can predict depth of penetration and even residual
velocity through thinner targets. However, they are less effective
It should be noted that is feasible to ‘test’ the failure surface at a at predicting the failure of the target (such as the size and shape
number of points, using a Hoek cell, if the axial load is rapidly of the front face crater and rear face spall), which requires a
removed before the onset of collapse, though tracking along the more accurate tensile failure model than exists at this time.
surface would not be credible. The authors have investigated the With a suitable tensile failure model to supplement the GREAC
use of steel stiffening rods to stabilise the specimen and this has cell characterisation, it would also be possible to exploit
been used with considerable success. hydrocodes to predict beam, column and slab responses to
stand-off explosive loading. With a suitable strain rate ‘enhance-
Rate effects ment’ to the GREAC cell data, it may be possible to exploit this
Unconfined compression tests and, to a lesser extent, active information in hydrocodes to predict high-velocity penetration
triaxial compression tests are subject to an apparent strain rate and contact explosive loading.
effect associated with specimen inertia. Essentially, the maximum
load that can be supported is determined by the speed at which GREAC cell characterisation of the compressive properties of
the load is applied compared with the rate at which the specimen concrete (and masonry) is expected to be suitable for exploitation
goes into unstable collapse. However, because the GREAC cell in a wide range of non-military applications including earthquake
specimen remains stably confined it is unable to collapse and is defences, hazard engineering and counter-terrorism. Again, it
therefore unaffected by inertial effects. requires supplementation with a suitable tensile failure model.

58
Downloaded by [ Universidade Federal de Pernambuco] on [22/11/17]. Copyright © ICE Publishing, all rights reserved.
Magazine of Concrete Research Gauged reactive confinement cell for
Volume 66 Issue 1 determination of concrete triaxial
properties
Sheridan and Pullen

Current status and recommendations for Passive Confinement and their Use in Structural Analysis.
further development PhD/DIC thesis, Imperial College London, London, UK.
A 30 mm diameter GREAC cell confining cylinder made of Kotsovos MD and Perry SH (1984) Determination of the Stress–
maraging steel is under development, which will allow the Strain Relationships for Confined Concrete. Final Report to
specimen stresses to reach 2 GPa. This could be scaled up to a Royal Aircraft Establishment Farnborough. Imperial College
diameter greater than 75 mm to allow full-scale concretes to be London, London, UK.
tested. Kotsovos MD and Perry SH (1986) Behaviour of concrete
subjected to passive confinement. Matériaux et Constructions
Future research on the characterisation of the plain strain behav- 19(112): 259–264.
iour of concrete should be directed at Neal WD (2012) The Role of Particle Size in Shock Compaction
of Brittle Granular Materials. PhD thesis, Imperial College of
j the development of a more robust static GREAC cell capable Science, Technology and Medicine, London, UK.
of significantly higher stresses Pullen AD and Newman JB (1993) Dynamic Properties of
j benchmarking the GREAC cell against more materials Shocked Geologic Materials. Final Technical Report for the
subjected to active triaxial testing Defence Research Agency Farnborough. Imperial College,
j development of a dynamic GREAC cell using the split London, UK.
Hopkinson pressure bar, for intermediate strain rates Pullen AD, Sheridan AJ and Newman JB (1993) Dynamic
j undertaking plate impact and/or contact explosive loading for compressive behaviour of concrete at strain rates up to 103 /s
highest strain rates to evaluate the dynamic shear behaviour – comparison of physical experiments with hydrocode
as well as the Hugoniot conditions. simulations. Proceedings of 6th International Symposium on
the Interaction of Conventional Munitions with Protective
It is hoped that, ultimately, it will only be necessary to undertake Structures, Panama City Beach, FL, USA. Wright Laboratory,
a static GREAC cell test as a low-cost routine methodology to Tyndall Air Force Base, Florida, pp. 182–187.
adequately characterise concrete and other geologically based Sheridan AJ (1990) Response of Concrete to High Explosive
materials by the application of factors to accommodate material Detonation. PhD thesis, Imperial College London, London,
lateral dilation and rate effects. UK.
Sheridan AJ (1991) Application of Concrete Triaxial Data to
Acknowledgements Hydrocodes. Defence Research Agency, Farnborough, UK,
The GREAC cell development and exploitation into defence Working paper MTC-91-WP-26.
research has been funded by the Ministry of Defence. # Crown Sheridan AJ, Stephens WH and Perry SH (1989) The behaviour of
Copyright 2013. Published with the permission of the Defence concrete under explosive shock in relation to its compaction
Science and Technology Laboratory on behalf of the Controller under uniaxial strain loading. Proceedings of Structures
of HMSO. Under Shock and Impact (SUSI) Conference, Boston, MA,
USA. Computational Mechanics Publications, Southampton,
REFERENCES UK, pp. 87–96.
Clegg RA, Sheridan J, Hayhurst CJ and Francis NJ (1997) The Sheridan AJ, Pullen AD and Newman JB (1992) The search
application of SPH techniques in Autodyn-2D to kinetic for a general geologic material model for application to
energy penetrator impacts on multi-layered soil and concrete finite element methods and hydrocodes. Proceedings of
targets. Proceedings of the 8th International Symposium on Structures Under Shock and Impact, Portsmouth, UK.
Interaction of the Effects of Munitions with Structures, Computational Mechanics Publications, Southampton, UK,
Alexandria, VA, USA. pp. 199–210.
Franklin JA and Hoeck E (1970) Developments in triaxial testing Vogler TJ, Lee MY and Grady DE (2007) Static and dynamic
technique. Rock Mechanics 2(4): 223–228. compaction of ceramic powders. International Journal of
Kinoshita M (1992) Constitutive Relations of Concrete under Solids and Structures 44(2): 636–658.

WHAT DO YOU THINK?


To discuss this paper, please submit up to 500 words to
the editor at www.editorialmanager.com/macr by 1 July
2014. Your contribution will be forwarded to the
author(s) for a reply and, if considered appropriate by
the editorial panel, will be published as a discussion in a
future issue of the journal.

59
Downloaded by [ Universidade Federal de Pernambuco] on [22/11/17]. Copyright © ICE Publishing, all rights reserved.

You might also like