Download as pdf or txt
Download as pdf or txt
You are on page 1of 10

Displacement measurement with laser scanning in triaxial

testing apparatuses
Sophie MESSERKLINGER, Ernst BLEIKER, Adrian ZWEIDLER &
Sarah M. SPRINGMAN
Institute for Geotechnical Engineering, Swiss Federal Institute of Technology Zurich

ABSTRACT
A laser scanning device for measurement of displacement is introduced for application in
triaxial testing apparatuses. The equipment provides contactless measurement of the radial
displacement of the sample over the whole sample height. Additionally, a method is proposed
for determining the initial sample volume from the scanned sample profiles as well as the
change in volume during test performance. These methods are then evaluated, based on test
results from a drained stress path test on Swiss lacustrine clay.

1. INTRODUCTION
In Switzerland, normally consolidated varved post glacial lacustrine deposits form the
dominating soft soil. In previous years, research has been undertaken to investigate the peak
and residual shear resistance, and the likely failure mechanisms in these soils, e.g. Heil
(2002). A comprehensive study about the characterisation of typical Swiss lacustrine clays
was recently published in Trausch-Giudici (2004). However it is becoming increasingly
important to model the deformation behaviour and hence to investigate the stress-strain
response before failure, and especially the small strain behaviour more precisely. The method
which was chosen for these investigations is triaxial testing, as many common constitutive
models are based on triaxial test results, e.g. critical state model (Schofield & Wroth 1968).

2. TEST APPARATUSES
The triaxial test apparatuses were constructed in-house at the Institute (IGT) and are described
more detailed in Arenson et al. (2004). These apparatuses consist of three units, as shown in
Fig 1(a), the cell pressure unit (2), the back pressure unit (1) and the triaxial cell (3). The cell
pressure unit and back pressure unit regulate the corresponding pressure with a piston system,
whereby the change in cell and pore water volume is measured. The sample (7) is placed in
the triaxial cell (3) and the axial load is applied to the sample through the top cap, via the load
cell (6) and the loading frame (5). The top cap is mounted on the loading frame so that
extension tests can be performed easily. All three units are powered by stepping motors (4),
which have a resolution of 400 steps per millimetre. In addition, a gear box is installed, which
reduces the vertical movement per step by a factor of 16. The resulting resolution is 1.5610-4
mm/step for axial movement and 1.227 mm3/step for the volume measurement. The axial
displacements are measured inside the cell by the external Linear Variable Displacement
Transformer (LVDT) (8), which is mounted above the top cap and has a resolution of
0.003 mm. The nonlinearity is 0.1 % of the total measurement range, which is 50 mm. The
triaxial apparatuses are placed in a temperature controlled room and the tests are performed at
a temperature of 10C ( 0.5C). Samples with a diameter of 50 mm and a height of 100 mm
are tested.
250 mm

5
6 8

7
1 9 2 (b)
1 back pressure unit
2 triaxial cell
3 cell pressure unit
4 4
4 stepping motors
5 loading frame
4 4 4
6 load cell
7 sample
8 external LVDT
(a) 9 laser measurement device

Fig. 1: (a) Standard triaxial testing apparatus (b) Construction of the radial displacement
measurement device.

Analysing the displacement measurement devices with respect to their accuracy, the
following errors are recognisable. In the axial direction, the LVDT (8) is mounted above the
top cap (but below the load cell (6)) and measures down to the bottom of the pedestal.
Therefore compliance in the top cap and pedestal, seating errors of the top cap on the porous
stone and bedding errors due to surface irregularities are incorporated into the measured data.
In the measurement of pore water volume, the main error is assumed to be due to the
extension of the tubes and the regulation piston system itself.
In order to investigate the small strain stiffness response of soft Swiss lacustrine clays,
options for more accurate displacement measurement methods were considered. Scholey et al.
(1995) present a comprehensive review of local strain measurement devices for triaxial tests,
pointing out that LVDTs mounted locally on the sample will give the best results for axial
displacement measurement. Four methods were discussed for measurement of radial
displacement, although they did not specify the accuracy for any of them.

2.1. Local axial strain measurement device


Consequently, it was expedient to install local LVDTs for local axial displacement
measurement in the triaxial apparatuses. Mountings were constructed similar to those
described in Cuccovillo & Coop (1997) to fix these LVDTs to the soil sample, although the
pins were omitted in this case. Gens (1982) deduced that the rubber membrane does not slide
on the sample surface, and so the displacement measurement is not improved by putting pins
through the rubber membrane into the soil sample. However the risk of creating a leak is
much higher with pins. Therefore the mountings were stuck on to the rubber membrane with
the superglue LOCTITE 460, which is effective in water and under pressure, and is very
stiff. An initial axial distance of 70 mm was guaranteed because stiff spacers were fixed
between the mountings while they were being glued onto the rubber membrane. The local
LVDTs have a nonlinearity of 0.1 % of the measurement range, which is 5 mm.

2.2. Local radial strain measurement device


Methods of carrying out radial displacement measurement without disturbing the sample were
considered. Although proximity transducers would have been suitable, an additional aim was
to measure the radial displacements over the whole sample height. In order to investigate
unsaturated soils under non-isothermal conditions, Romero et al. (1997) had mounted lasers
outside the Perspex triaxial cell and used them to scan the sample to obtain radial
measurement of displacement. This idea was implemented within the triaxial testing apparatus
by installing two lead screws (diameter 20 mm; pitch 4 mm/rotation) into the triaxial cell,
which are powered by stepping motors. A connecting plate was mounted onto these lead
screws and the lasers were fixed to it. A three-dimensional picture of the two lead screws with
the plate and the mounted lasers is shown in Fig 1(b). This plate is moved upwards and
downwards along the lead screws together with the lasers, therefore the radial distance to the
surface of the sample can be scanned. The movement speed is chosen to be 1 mm per second
and the data of the radial distance are logged every second. Additionally, it was decided to do
the data logging during upwards movement of the lasers and to move them back down
without data logging to avoid any hysteretic effects due to slippage on the driving screws
following change of direction. The vertical zero position of the lasers was set below the
bottom of the sample, to exclude hysteretic effects due to the initiation of movement of the
stepping motors. It was decided to place three lasers around the sample, spaced at 120, to
observe any bending or other irregularities in the sample displacements. The construction of
the mounting unit and the arrangement of the lasers are shown in the top view of Fig. 2(a).
Three lasers were installed to enable the initial sample volume to be determined. Up to now,
the initial sample dimensions were measured with a sliding calliper, which is especially
inaccurate for soft samples because the sample deforms as soon as the calliper touches it. The
lasers (Micro-Epsilon ILD1400-5) have a resolution of 1 m. The measurement range is 5
mm, the mean measurement distance to the sample is 22.5 mm and the nonlinearity is 0.2 %
of the measurement range.
measurement evaluation

lead screw

top
plate

connecting plate
r +
=> => V0 +V
profile
housing
laser &

B B
laser &
sample housing
o 50 mm

bottom pedestal

10 mm
Section A - A
La
se
r2

F2 [xL2,yL2]
m2

[x3,y3]
120
A 120 A [x0,y0] m1 F1 [xL1,yL1]
R
Laser 1
120 [x1,y1]
x
[x2,y2]
m3
r3

F3 [xL3,yL3]
se
La

Section B - B
(a) (b)
Fig. 2: (a) Front view and top view of the construction of the laser scanning device (b) Evaluation
methods of the laser data.
2.3. Proposed sample volume determination method
An attempt has been made to determine the volume of the sample from surface scans on three
sides of the sample. The most straightforward solution is to place a circle through these three
points, Grn (2004). Therefore a general coordinate system was introduced on a horizontal
plane as shown in Fig. 2(b) and the centre of the sample [x0,y0] was fixed at a defined distance
from the origin. Data logging was done every 1 mm of laser movement in an axial direction.
Therefore, a horizontal plane was introduced at each measurement cross section. For the
definition of the position of the lasers in the coordinate plane (F1 to F3, Fig 2(b)), a zero
reading was necessary. An aluminium cylinder with a known diameter of 50 mm within an
accuracy of 0.01 mm was prepared for calibration. A laser scan of the cylinder was made
and the coordinates of the laser position at each cross section (xL1 to xL3 and yL1 to yL3) was
determined. Consequently from the radial distance measurement of the laser (m1 to m3,
Fig. 2(b)) to any sample, placed in the triaxial cell, the coordinates of the three surface points
(x1 to x3 and y1 to y3) can be determined. Having three coordinates of the sample, the
coordinates of the actual sample centre [xa,ya] and the radius R of the circle can be determined
using Eq. (1). The cross sectional area and the volume of the sample slice can be calculated
from the radius of the circle.
R2 = (x1(2,3) xa)2 + (y1(2,3) ya)2 (1)
The sample volume can be determined by summing up the volume of all slices over the
sample height (considering the top slice height corresponding to the actual sample height).
This method was implemented in the test control program Labview. Calibration tests on
various steel cylinders of known dimensions allowed the measurement accuracy of the lasers
and the effect of these positions during the scan to be evaluated. The accuracy of the volume
determined was in a range of 0.1 %, without using the circular slices approach mentioned
above. This approach was validated by testing a natural sample of Swiss lacustrine clay (see
chapter 3).

3. DRAINED STRESS PATH TEST


A drained stress path test was carried out by loading natural Swiss lacustrine clay sample
along a stress ratio of = 0.75 ( = q/p; q = 1 3; p = (1 + 2*3)/3) up to p = 300 kPa.
Subsequently, the sample was unloaded along the same stress ratio down to p = 150 kPa. The
sample was reloaded along the same stress ratio of = 0.75 to detect stiffness reduction for
stress path reversal.
3.1. Evaluation of the local axial strain measurement device
The measurement data are presented in Fig. 3 to Fig. 5. A comparison of the axial
displacements measured by the local LVDTs to the measurement data of the external LVDT
is presented in Fig. 3 (a) & (b). The axial strains a are plotted against the natural logarithm of
mean effective stresses p. The axial strains a are defined as the change in sample height
divided by the actual sample height. In Fig. 3(a), the displacements due to reconsolidation and
unloading are shown. It can be seen that the values of strains measured by the two local
LVDTs and the external one are similar. It is assumed that the effects of seating and bedding
errors are not that dominant for this triaxial test apparatus setup and the sample. A yield point
lying above an equivalent intrinsic compression line (Burland (1990)) followed by
destructuration of the natural sample is clearly visible.
In Fig. 3 (b), the displacements due to reloading are shown. A high fluctuation in the strains
measured by the external LVDT is visible, while the measurement data of the two local
LVDTs give quite a smooth curve. The external LVDT is mounted at the load frame above
the top cap. These high fluctuations are thought to be due to stress adjustments through the
load frame with the stepping motors in order to follow the prescribed stress path.
Summarising the results of axial displacement measurements using the triaxial apparatus,
comparable results for plastic straining can be produced using external and internal LVDTs.
For elastic straining, where the strain values are much lower, the external LVDT is unsuitable.

ln p' ln p'
1 2 3 4 5 6 5.0 5.1 5.2 5.3 5.4
0 0.00

2 0.02

0.04
4
0.06
[%]

[%]

6
0.08
8
0.10
local LVDT 1 local LVDT 1
10 local LVDT 2 0.12 local LVDT 2
external LVDT external LVDT
12 0.14

(a) (b)
Fig. 3: Axial strains measured by the two local LVDTs and the external LVDT, plotted against
the natural logarithm of the mean effective stresses p (stress unit is kPa) (a) for the
reconsolidation and unloading stress path (b) for the reloading stress path.
3.2. Evaluation of the volume determination approach
As a next step, the initial volume of the sample is determined. Following laser measurement
and evaluation of the data with the method described above (Fig. 2 (b)), led to an initial
sample volume of 223.717 cm3. In comparison, the initial sample volume determined using
the sliding calliper, (by taking the mean values of the height and the top, middle and bottom
dimension) gave a sample volume of 223.57 cm3, which varies only by 0.07 %.
The volume reduction curve due to reconsolidation and unloading is shown in Fig. 4. Three
methods are evaluated. Firstly, the cell water volume measurement, which determines the
change of the sample volume by the change in the cell water volume, is derived. The volume
change is measured by the cell pressure piston system due to the adjustment of the cell
pressure. Owing to the volume of water in the cell (~ 60 litres) poor deaeration of this water
and environmental effects, especially temperature fluctuations have a major influence. The
second method measures relative the pore water volumes and is based on the same system as
the cell water volume measurement. The third method, using the lasers, represents the result
of the volume calculation out of the scanned sample surface. This method was already
described above and is summarized in Fig. 2. Comparing the results of these three methods, it
can be seen that all methods indicate a similar value of yield stress. As expected, the cell
water volume measurement overestimates the volume loss significantly. The stiffest sample
response was measured with the laser volume measurement method.
225

220

215

210
volume [cm ]
3

205

200
Laser volume
195 measurement
Pore water volume
190 measurement
Cell water volume
measurement
185
1 2 3 4 5 6
ln p'

Fig. 4: Comparison of the volume measurement of the three volume measurement devices, plotted
against the natural logarithm of the mean effective stresses p for the reconsolidation and
unloading stress path.

Summarising the results of sample volume measurement, it can be said that the new laser
volume measurement method gives encouraging results from preliminary tests, although the
apparent steps in the data (Fig. 4) are being investigated and dealt with (occasional skipping
of one data capture step). With this method, the saturation of the sample is no longer
necessary, in terms of obtaining accurate volume measurement and errors due to expansion of
the triaxial devices or leaking are eliminated.

3.3. Evaluation of radial strain measurement device


In Fig. 5, the data of the laser measurements are evaluated in terms of radial strains. In
Fig. 5 (a) the radial strains r are plotted against the sample height. The radial strains r are
defined as the change of the sample dimension in a radial direction divided by the actual
sample radius. A positive value of radial strain corresponds to a radial expansion.
90
0
80
deviator stress q [kPa]

1
70
2
60 3
50 4
40 5
30 6
7
20
8
10
9
0
0 50 100 150 200
(a) mean effective stress p' [kPa]

Laser 1 Laser 2 Laser 3


100 100 100

90 90 90

80 80 80

70 70 70
sample height [mm]

60 60 60
local LVDTs

50 50 50

40 40 40

30 0 30 0 30 0
1 1 1
2 2 2
20 3 3 20 3
20
4 4 4
5 5 5
6 6 6
10 10 10
7 7 7
8 8 8
9 9 9
0 0 0
-2 -1 0 1 2 -1 0 1 2 3 -1 0 1 2 3
(b) radial strain r [%] radial strain r [%] radial strain r [%]

Fig. 5: (a) Shear stresses q versus mean effective stresses p plot for the reconsolidation stress
path; indication of the stress states corresponding to the strain profiles in (b); (b) Radial
strains r plotted against the sample height for the stress states indicated in (a); Please note
that the sample height reduced with ongoing consolidation.
The profiles presented in Fig. 5 are numbered continuously. Profile number zero represents
the data of the sample surface scan after sample setup. For the following profiles, the
corresponding stress state is given in the q p plot in Fig. 5 (a). The results of the zero
reading give a smooth surface profile for all three scans Fig. 5 (b). Profile number 1 gives a
distinct expansion of the sample. For the other profiles, expansion continues more slowly for
laser 2 and 3, although apparent contraction was measured with laser 1. This is maybe due to
slight translation of the sample from laser 1 towards laser 2 & 3.
With ongoing consolidation, the surface profile seems to become a zigzag line. Interpreting
this result, two possible explanations can be given. Firstly, it can be due to influences of the
construction as the plate is moved by a lead screw which could produce cyclic influences. But
as this scattering does not occur for the readings at beginning and becomes more significant
with ongoing consolidation, it was not thought to be the reason. Secondly, it might be due to
the influence of the sample. A natural sample of Swiss lacustrine clay was used to evaluate
this method. Lacustrine clay is known to have a distinct stratification due to deposition. The
thickness of the layers depends on the deposition region. For this clay sample, the silt layers
have a thickness of 2 to 4 mm and the clay layers are slightly thinner. Evaluating the data of
the measured radial strains in Fig. 5, the zigzag profile could be due to the higher
compressibility of the clay layers. For a detailed evaluation of this effect, the data logging for
laser scanning has to be done ten times quicker to reduce spacing between readings and then
compared with the sample post-test. It is clear that the radial displacement measurement at
one single location would not have been successful.

4. DISCUSSION
One of the new IGT triaxial apparatus was equipped with small strain measurement devices.
For measurement of axial displacement, LVDTs mounted locally on the sample have been
installed and for determination of radial and volume displacement measurement a laser
scanning device was developed and constructed. A drained stress path test on a natural Swiss
lacustrine clay sample was performed to evaluate accuracy of the new displacement
measurement system. It was seen that for axial displacement measurement, the amount of
strains determined by external and local measurement devices is quite similar (perhaps
coincidentally), while the scattering of the data, which is especially important in the small
strain region, is significantly lower for the local axial measurement devices. The volumetric
measurement method based on scanned sample profiles was proved to be effective for the
determination of the initial sample volume as well as for the measurement of volume decrease
during performance of the drained stress path test. Radial strain measurement with the laser
scanning device turned out to be most suitable for layered soils.

ACKNOWLEDGEMENTS
The authors are very grateful to Heinz Buschor who built the housings and frames for the
laser scanning devices in the IGT workshop.

REFERENCES
[1] Heil M. (2002): Localisation and critical state. Constitutive and Centrifuge
Modelling: Two Extremes, Workshop, Ed. S.M. Springman, Ascona, 199-208.

[2] Trausch-Giudici J. (2004): Stress strain characterisation of Swiss Seebodenlehm. PhD


thesis, Swiss Federal Institute of Technology, Zurich.

[3] Schofield A.N. & Wroth C.P. (1968): Critical state soil mechanics. McGraw Hill,
London.

[4] Arenson L.U., Trausch-Giudici J., Messerklinger S., Zweidler A., Bleiker E. &
Springman S.M. (2004): A new computer controlled triaxial test apparatus for frozen
soils and soft clays. Submitted to ASTM Geotechnical Testing Journal.

[5] Scholey G.K., Frost J.D., Lo Presti D.C.F. & Jamiolkowski M. (1995): A review of
instrumentation for measuring small strains during triaxial testing of soil specimens.
Geotechnical Testing Journal, Vol. 18, No. 2, 137-156.

[6] Cuccovillo T. & Coop M. R. (1997): TECHNICAL NOTE: The measurement of local
axial strains in triaxial tests using LVDTs. Gotechnique, Vol. 47, No. 1, 167-171.

[7] Gens A. (1982): Stress-strain and strength characteristics of a low plasticity clay. PhD
thesis, London University.

[8] Romero E., Facio J.A., Lloret A., Gens A. & Alonso E.E. (1997): A new suction and
temperature controlled triaxial apparatus. 4th ICSMFE Hamburg, Vol. 1, 185-188.

[9] Grn A. (2004): Professor for Photogrammetry at the Institute of Geodesy and
Photogrammetry at the Swiss Federal Institute of Technology Zurich. Personal
Communication.

[10] Burland J.B. (1990): On the compressibility and shear strength of natural clays.
Gotechnique, Vol. 40, No. 3, 329-378.

You might also like