Procedia Engineering

Volume 143, 2016, Pages 419426

Advances in Transportation Geotechnics 3 . The 3rd
International Conference on Transportation Geotechnics
(ICTG 2016)

Investigation of the Mechanical Behaviour of the

Interface between Soil and Reinforcement, via
Experimental and Numerical Modelling
Elena Kapogianni1*, Michael Sakellariou1, Jan Laue2 and Sarah
Springman3#
1

National Technical University of Athens, Athens, Greece.

2
Lule University of Technology, Lule, Sweden.
3
ETH Zrich, Zrich, Switzerland.
elkapogianni@gmail.com, mgsakel@mail.ntua.gr, jan.laue@itu.se,
sarah.springman@igt.baug.ethz.ch

Abstract
The purpose of this study is to investigate the interface properties between soil and reinforcement,
via experimental and numerical modelling of reinforced slopes. In particular, several scale models
were built and tested under enhanced gravity in the geotechnical drum centrifuge at ETH Zurich and
corresponding prototype numerical models were analyzed via a finite element stress analysis code.
Optical fibre sensors were attached on the reinforcement layers of the experimental scaled models in
order to measure linear strain during the increase of the g-level, and the results were compared to
linear strain that was derived by the numerical analysis of the correspondent prototype reinforced
slopes. The interface between soil and reinforcement was expressed in terms of normal and shear
stiffness on the soil-reinforcement boundary and different values were tested in order to achieve
validation of the experimental and numerical results.
Keywords: Physical Modelling, Numerical Modelling, Centrifuge Model Tests, Reinforced Slopes, Optical Fibre
Sensors, Interface Properties

Postdoctoral Researcher
Professor Emeritus

Chaired Professor
# Rector

Selection and peer-review under responsibility of the Scientic Programme Committee of ICTG 2016
c The Authors. Published by Elsevier B.V.


doi:10.1016/j.proeng.2016.06.053

419

Investigation of the Mechanical Behaviour of the Interface between Soil ...

Kapogianni et al.

1 Introduction
Reinforced soil slopes are traditionally investigated by applying experimental, numerical and
analytical methods with the main goal being to define their behaviour and failure mechanisms due to
static and seismic loading. A more comprehensive analysis can be made for reinforced slopes with
various geometrical and mechanical characteristics by combining the above methods so that useful
conclusions can be derived about the methods applied.
Experimental investigation requires detailed physical modelling of the planned construction and
large or small scale models can be built and tested. The main forces that are applied and define the
behaviour of soil structures, and reinforced slopes in particular, are those of self-weight. The simple
gravity field is often inadequate when small scale models are examined. Numerical modelling in
geotechnics requires detailed simulation of the mechanical and geometrical characteristics of the
models. Finite element stress analysis is traditionally applied, and special attention should be given to
the simulation of the interface between soil and reinforcement for the case of reinforced slopes.
Analytical investigation of reinforced slopes is also applied for the design of reinforced slopes that are
subjected to static and seismic loading and in order to examine their vulnerability. The analytical
methods are based on the limit equilibrium method (Terzaghi, 1943) and on the limit analysis method
(Drucker & Prager, 1952), and are applied for the design and evaluation of reinforced slopes
(Kapogianni & Sakellariou, 2008).

2 Centrifuge Modelling of Reinforced Slopes

In order to recreate stress and strain due to gravity in a small scale model 1/n, it is necessary to
apply n times the gravity field. Enhanced gravity can be applied with the help of geotechnical
centrifuge technology. Whereas the idea of using a centrifuge originated from Philips (1869), the
application as a geotechnical centrifuge came from the 1930s (Pokrovsky & Fedorov, 1936; Bucky,
1931), and many drum and beam centrifuges have been built around the world since then, establishing
centrifuge technology as a powerful tool for the experimental investigation of geotechnical
constructions (Laue, 2014; Springman, 2014; Craig, 2014 & 2001; Mayne et al., 2009; Schofield,
1980).
For the simulation of the prototype model behaviour in the geotechnical centrifuge, several scaling
laws should be taken into account. Such laws have been presented in the past by Schofield (1980) and
can be used in order to create models with geometrical and mechanical characteristics similar to those
of prototype structures. Also, materials used for the model construction should satisfy the
corresponding scaling laws and boundary effects should be minimized (Laue, 2002). Especially for
reinforced slopes involving reinforcement layers, difficulties are encountered concerning the choice of
materials, mainly because the reinforcements used in prototype constructions are inappropriate for
small scale models, since they dont replicate the full-scale stress-strain behaviour and the geometrical
conditions at small-scale. Such scaling laws have been presented by Garnier et al. (2007),
Viswanadham and Mahajan (2007), Viswanadham et al. (2006), Zornberg et al. (1997), and are
applied for the purposes of the current study. The physical modelling of reinforced slopes in the
centrifuge has also been studied by Balachandran and Springman (1997) and Springman et al. (1997).
Several scaled wrap-around reinforced sand slope models were built and tested for the current
study, under enhanced gravity up to 50g, in the ETH Zurich drum geotechnical centrifuge (Springman
et al., 2001), with a facing slope inclination of 2V:1H (63.43) and a total height of 20 cm (Figure 1a).
The soil material used for the models was Perth sand from the west coast of Australia. Perth sand is
white, fine grained sand with rounded grains. The angle of deposition is between 31 0 to 330 and the
diameter of the grains lays between 0.125-0.5 mm. This soil is classified as uniform sand (Nater,
2006). The height of the reinforced soil mass was 18 cm and the wall was built on a 2 cm sand

420

Investigation of the Mechanical Behaviour of the Interface between Soil ...

Kapogianni et al.

foundation layer. The reinforcement layers were uniformly spaced every 2 cm and different
reinforcement materials were used, with length varying between 11-19 cm. The scale applied was
n=50.
Optical fibre strain sensors were glued onto the reinforcement layers in order to measure linear
strain during the increase of the g-level. The locations of the sensors can be noted in the cross-section
view of the experimental test set-up in Figure 1b. Sensors were placed on Layer No. 8, near the top of
the slope and on Layer No. 4 near the middle of the slope. The details of the the material chosen, and
the model making system set up, have been presented in Kapogianni et al. (2010, a & b). The physical
models were instrumented to monitor the structures during the increase in g-level, and this was applied
for the first time in the geotechnical centrifuge for the purposes of the current study. Innovative
monitoring techniques have also been applied for various geotechnical structures (Soga et al., 2010).

Figure 1a. Reinforced slope model (scaling factor n=50)

Figure 1b. Cross-section view (dimensions in mm)

The optical fibre sensors that were attached on the reinforcement layers successfully measured
linear strain during the increase in the g-level. Different values were recorded by the sensors placed on
Layer No. 8, near the top of the slope and on Layer No. 4, near the middle of the slope. Some
characteristic results are presented in Figure 2 for Layers No. 4 and 8. Linear strain increased as the glevel increased and remained stable as the g-level remained stable. Also, as expected, strains measured
on Layer No. 4 were higher than strains on Layer No. 8. It can also be seen that measurements taken
by the optical fibre sensors were similar at the same locations for different tests, and for the same
loading conditions. The time delay on the measurements was due to the fact that the optical fibre
sensors started recording measurements at different moments during each of the tests. However, the
time frame was the same when comparing the time that the strains started developing for the
comparable tests.
Comparison of strains measured 50g

Strain (strain)

1400

Layer No. 4

1200

1000
800

Layer No. 8

600
400
200
0
0

100

200

300

400

500

600

700

Time (sec)

Figure 2. Strain variation measured via o.f. sensors

421

Investigation of the Mechanical Behaviour of the Interface between Soil ...

Kapogianni et al.

3 Numerical Modelling of Reinforced Slopes

The mechanical and geometrical characteristics of the correspondent prototype numerical models,
created and analyzed by a finite element stress analysis code, are presented in the following. As can be
seen in Table 1, two different reinforcement materials have been chosen to represent the reinforcement
sheets. The scaling factor applied is n=50 for both materials, equal to the maximum g-level that the
experimental slope models were subjected to. The reinforcement tensile strength in kN/m can also be
seen for the stronger and the weaker material (Type 1 & 2 respectively). The soil Youngs Modulus
was defined equal to 30.000 kPa, the Poissons ratio 0.3 and the soil friction angle 31 0.
Reinforcement

Scale

Slope

Reinforcement

Reinforcement Tensile

Factor (n)

Height (m)

Length (m)

Strength (kN/m)

Type 1

50

5.5-9.5

82

Type 2

50

5.5-9.5

32.5

Material

Table 1. Mechanical and geometrical characteristics of prototype numerical models

Representative results of the analysis performed via the finite element stress code (Phase2), are
presented and discussed in the following section. At first, the Strength Reduction Method is applied,
and the Strength Reduction Factor (SRF) that expresses the available extra strength on the
constructions is defined. An example of a reinforced slope can be seen in Figure 3a, where the shear
strain and failure mechanism for n=50 and reinforcement material Type 1 is presented. The failure
mechanism is located near the end of the reinforced soil mass, as expected. The SRFs that have been
calculated for the two models are presented in Table 2, and it can be noted that the SRF value is higher
for the stronger reinforcement material, Type 1. The SRFs defined are higher than 1 for both cases,
indicating that the models are far from failure, which also validates the behaviour of the correspondent
experimental models. The numerical models were analysed again with expanded boundaries in order
to investigate the influence of the boundary on the failure mechanism, and on the strains developed.
As can be noted in Figure 3b and Table 3, smaller SRF values and higher shear strains were calculated
for the models with expanded boundaries. However, this variation is not very high and models with
limited boundaries were selected for further investigation, representing the experimental models more
accurately. The interface friction along the walls was investigated and restrain at x axis and free
restrain at y axis, for model with limited boundaries, while restrain at both x and y axis for model with
expanded boundaries, were chosen. Further details have been presented in Kapogianni (2013).

Figure 3. Shear strain in a 63.43 sand slope, due to reducing strength in the reinf. by the factor SRF

422

Investigation of the Mechanical Behaviour of the Interface between Soil ...

Kapogianni et al.

Material Type

Prototype Slope Height (m)

SRF

Type 1

50

1.80

Type 2

50

1.70

Table 2. Strength Reduction Factor (SRF) for models with limited boundaries

Material Type

Prototype Slope Height (m)

SRF

Type 1

50

1.70

Type 2

50

1.50

Table 3. Strength Reduction Factor (SRF) for models with expanded boundaries

The numerical models were analysed again with pre-defined SRF equal to 1, in order to
demonstrate the behaviour of the slopes due to self-weight loading. The results of this analysis were
compared to the corresponding experimental results, and are presented in the following section. An
example of the slope model with SRF equal to 1 is shown in Figure 4. As can be noted, shear strains
developed are smaller for the lower SRF=1, as expected.

Figure 4. Shear strain in a 63.43 slope, due to reducing strength in the reinf. by the factor SRF=1

4 Comparison of Numerical and Experimental Results

The results obtained by analysis of the prototype slope models via the finite element code, provided
useful information about the model strength, failure mechanisms and shear strains developed. The soil
was modeled as linear elastic-perfectly plastic and the angle of friction was 31 0. Additionally, linear
strains of the reinforcement layers were defined, on the corresponding locations of the optical fibre
sensors that were attached onto the experimental model reinforcement. In particular, linear strains
were calculated by the division of the horizontal elongation dl and the corresponding length l of
the reinforcement layers. Length l was defined as the distance between the nodes created during the
mesh generation of the finite element models and dl the corresponding elongation.
The interface between soil and reinforcement was investigated in terms of normal and shear
stiffness. Normal stiffness kn equals the slope of the normal stress , versus the normal displacement
of interface v, and shear stiffness equals the slope of the shear stress-shear displacement curve until

423

Investigation of the Mechanical Behaviour of the Interface between Soil ...

Kapogianni et al.

slip. In particular, various values of normal and shear stiffness were selected during the creation and
analysis of the numerical models, in order to define those for which the numerical results were similar
to the experimental ones. The limitations placed were that normal stiffness should be smaller than the
Youngs modulus, and shear stiffness smaller than normal stiffness. The results are presented for
reinforcement in Tables 4 and 5 for layers No. 4 and 8, near the middle and edge of the slopes for
reinforcement material Type 1.
Normal Stiffness 20000 (kPa)
Shear Stiffness (kPa)
Layer No.4

Strain via o.f. (scaled model)

0.0013

Strain via n.m. (prototype model)

Middle

Edge

5000

0.001313

-1.8E-06

10000

0.00114

0.000214

15000

0.001071

0.000318

20000

0.001033

0.000376

Layer No. 8

0.0007

5000

0.001351

0.000502

10000

0.000870

0.000288

15000

0.000707

0.000259

20000

0.000608

0.000222

Table 4. Comparison of linear strain of scaled models, via optical fibre sensors (o.f.) and corresponding prototype
numerical models (n.m.), for various shear stiffness values and normal stiffness=20000kPa (limited boundaries).
Normal Stiffness 25000 kPa
Shear Stiffness (kPa)

Strain via o.f. (scaled model)

Layer No.4

0.0013

Strain via n.m. (prototype model)

Middle

5000

0.001300

10000

0.001150

15000

0.001069

20000

0.001034

Layer No. 8

Edge

0.0007

5000
10000

0.000914

15000

0.000726

20000

Table 5. Comparison of linear strain of scaled models, via optical fibre sensors (o.f.) and corresponding prototype
numerical models (n.m.), for various shear stiffness values and normal stiffness= 25000 kPa (limited boundaries).

As can be noted in Table 4, linear strains calculated for reinforcement Layer No. 4 near the middle
of the slope are similar to the corresponding experimental results for normal stiffness equal to 20000
kPa, and for shear stiffness equal to 5000 kPa. However, values near the edge of the slope are not
similar due to the simulation method used. The reinforcement layers were placed using a wraparound
technique in order to retain the soil during the construction of the experimental models, and this
technique could not be applied in the simulation of the numerical models. It can be noted that linear

424

Investigation of the Mechanical Behaviour of the Interface between Soil ...

Kapogianni et al.

strains that were derived from the numerical analysis are similar to the corresponding experimental
results for reinforcement Layer No. 8, for normal stiffness equal to 20000 kPa and shear stiffness
equal to 15000 kPa. This identification occurs for values near the middle of the slope also for this
case, however significant differences are noted near the edge of the slope. The results for normal
stiffness equal to 25000 kPa are presented in Table 4 and strains that were derived for this case from
the numerical analysis are similar to the experimental ones for shear stiffness equal to 5000 kPa for
Layer No. 4 and for 15000 kPa for Layer No. 8. Results near the edge of the slope models are not
presented for this case, due to the differences in the simulation method used that resulted in high
deviation of the results.

5 Conclusions
The combination of experimental and numerical investigation of geotechnical models, and
reinforced slopes in particular, can provide useful information about the performance of the
constructions. Failure mechanisms can be defined by applying centrifuge technology on small scale
models, and accurate measurements of strains developed can be recorded, by using the appropriate
measuring techniques. Also, numerical models via a finite element analysis code can give useful
information about the behaviour of the corresponding prototype models and the results can be
compared with the experimental data.
In the current study, optical fibre sensor technology was applied on reinforced slope models tested
under enhanced gravity, and the linear strains were measured. Also, corresponding prototype
numerical models were used in order to define the interface properties between soil and reinforcement.
Calibration of the numerical model results via the optical fibre measurements was achieved by
combining both physical and numerical modelling techniques and the interface was expressed in terms
of normal and shear stiffness. It was noted that smaller values of the shear stiffness were activated for
lower layers of reinforcement and linear strains developed were higher, while higher values of shear
stiffness were activated for upper layers, and strains developed were smaller.
After the definition of the interface properties between soil and reinforcement, the numerical
models can be built and analysed again, using the new interface properties values. This technique can
be applied to various geotechnical constructions and optical fibre linear strain measurements can be
compared with corresponding linear strain results that were derived from numerical analysis.
ACKNOWLEDGEMENTS
The first author is grateful for support provided by the State Scholarships Foundation of Greece
(IKY). The authors would like to thank the technical support group of IGT at ETH Zurich and
particular, Markus Iten, Heinz Richner and Ernst Bleiker for their support.

References
Balachandran, S. & Springman, S.M., 1997. Prediction of front wall deformation of reinforced soil
walls. International Conference on Ground Improvement Techniques, Macau, China. pp. 41-51.
Bucky, P.B., 1931. The use of models for the study of mining problems. Technical Publication
425, American Institute of Mining and Metallurgical Engineering. New York, USA.
Craig, W.H., 2014. Modelling slope failures by 'gravity turn-on'. Proceedings of the 8th
International Conference on Physical Modelling in Geotechnics, Perth, Australia. (1). pp. 1203-1209.
Craig, W.H., 2001. The seven ages of Physical Modelling. Workshop on Constitutive and
Centrifuge Modelling: Two Extremes, Monte Verit, Switzerland. pp. 165-174.

425

Investigation of the Mechanical Behaviour of the Interface between Soil ...

Kapogianni et al.

Drucker D.C. & Prager, W. 1952. Soil Mechanics and plastic analysis or limit design. Quarterly of
Applied Mathematics, 10, No. 2. pp. 157-172.
Garnier, J. Gaudin, C., Springman, S.M., Culligan, P.J., Goodings, D., Konig, D., Kutter, B.,
Philips, R., Randolph, M.F. & Thorel, L., 2007. Catalogue of Scaling laws and similitude questions in
geotechnical centrifuge modelling. International Journal of Physical Modelling in Geotechnics. 7(3).
pp. 1-23.
Kapogianni, E., 2013. Analytical Experimental and Computational investigation of reinforced
slopes, due to static and seismic loading. PhD Thesis, NTUA (in Greek).
Kapogianni, E., Laue, J. & Sakellariou, M., 2010. Reinforced slope modelling in a geotechnical
centrifuge. Proceedings of the 7th International Conference on Physical Modelling in Geotechnics,
Zurich, Switzerland. pp. 1125-1130.
Kapogianni, E., Laue, J., Sakellariou, M., Springman, S.M., 2010. The use of optical fibers in the
centrifuge. Proceedings of the 7th International Conference on Physical Modelling in Geotechnics,
Zurich, Switzerland. pp. 343-348.
Kapogianni, E., & Sakellariou, M., 2008. Analytical solution of multi-step reinforced soil slopes
stability due to static and seismic loading. Proceedings of the 1st International Conference on
Transportation Geotechnics, Nottingham, UK. pp. 445-451.
Laue, J., Springman, S.M., Gautray, J., Morales, W.F., Iten, M. & Arnold, A., 2014. 15 years of
experience using a physical model exercise in a Masters course. Proceedings of the 8th International
Conference on Physical Modelling in Geotechnics, Perth, Australia. pp. 445-450.
Laue, J. 2002. Centrifuge Technology. In Springman, S.M. (ed.). Constitutive and Centrifuge
Modelling: Two Extremes. pp. 75-112. Balkema.
Mayne, P.W., Coop, M.R., Springman, S., Huang, A-B., & Zornberg, J., 2009. GeoMaterial
Behavior and Testing, State-of-the-Art Paper. Proceedings of the 17th International Conference in Soil
Mechanics & Geotechnical Engineering, Vol. 4, Alexandria, Egypt. pp. 2777-2872.
Nater, P., 2006. Belastungs und Verformungsverhalten von geschichteten Bodensystemen unter
starren Kreisfundationen. PhD Thesis, ETH Zurich.
Pokrovsky, G.I. & Fedorov, I.S., 1936. Studies of soil pressures and soil deformations by means of
a centrifuge. Proceedings of the 1st International Conference on Soil Mechanics and Foundation
Engineering, Vol. 1, E-5. pp. 70.
Schofield, 1980. Cambridge geotechnical centrifuge operations. Geotechnique 30 (3). pp. 227-268.
Soga, K., Chaiyasarn, K., Viola, F., Yas, J., Seshia, A. & Cipiola, R., 2010, Innovation in
Monitoring Technology for Underground Structures. Information Technology in Geo-Engineering,
Proceedings of the 1st International Conference (ICITG), Shanghai, pp. 3-18.
Springman, S.M., Laue, J., Herzog, R. & El-Hamalawi, A., 2014, The evolution of a physical
modelling course over two decades. Proceedings of the 8th International Conference on Physical
Modelling in Geotechnics, Perth, Australia. (1). pp. 433-439.
Springman, S.M., Balachandran, S. & Jommi, C., 1997. Modelling prefailure deformation
behaviour of reinforced soil walls. Gotechnique. 48(3). pp. 653-663.
Springman, S.M., Laue, J., Boyle, R., White, J. & Zweidler, A., 2001. The ETH Zurich
geotechnical drum centrifuge. International Journal of Physical Modelling in Geotechnical
Engineering. 1(1):59-70.
Viswanadham, B.V.S. & Mahajan, R.R., 2007. Centrifuge model tests on geotextile-reinforced
slopes. Geosynthetics International. 14(6). pp. 365-379.
Viswanadham, B.V.S., Konig, D. & Triantafyllidis, T., 2006. Centrifuge model tests on geotextile
reinforced slope subjected to differential settlements. In Ng et al. (eds.), Physical Modelling in
Geotechnics-6th ICPMG. Taylor and Francis Group. London. pp. 597-602.
Zornberg, J.G., Mitchell, J.K., & Sitar, N., 1997. Testing of Reinforced Soil Slopes in a
Geotechnical Centrifuge. ASTM Geotechnical Testing Journal, Vol. 20, No. 4, December, pp. 470480.

426