INTERNATIONAL JOURNAL OF PAVEMENT ENGINEERING

2018, VOL. 19, NO. 10, 917–929

http://dx.doi.org/10.1080/10298436.2016.1224410

Numerical simulation tool for paving block structures assessed by means of full-scale
accelerated pavement tests
J. Füssla , H. Hengla , L. Eberhardsteinerb , W. Kluger-Eiglb and R. Blabb
a Institute for Mechanics of Materials and Structures, TU Wien, Vienna, Austria; b Institute of Transportation, TU Wien, Vienna, Austria

ABSTRACT ARTICLE HISTORY

Block pavements are an attractive alternative to asphalt and concrete pavements, especially in communal Received 5 September 2014
areas. Architects and urban planners would like to take advantage of various shapes, colours and Accepted 8 August 2016
textures of paving blocks in order to achieve a higher quality of urban space if the performance of KEYWORDS
block pavements could be better predicted to avoid large horizontal displacements of chipped stone Paving blocks; numerical
corners and rutting. Unfortunately, the computational performance prediction of paving block structures pavement simulation; joint
is more complex than that for flexible pavements with homogeneous surface layers of asphalt concrete. behaviour; Mohr–Coloumb
The influence of the large number of vertical joints between paving blocks on the overall mechanical friction; accelerated
performance has not been considered sufficiently within computational tools yet. The proposed numerical pavement testing
simulation tool is able to take into account the complex mechanical behaviour of sand-filled joints as
well as the non-linear mechanical behaviour of the underlying base courses. Joints are modelled using a
Mohr–Coulomb-type friction model with the normal stresses non-linearly related with joint opening. Three
different experimental set-ups were developed for the identification of the model parameters. The base
behaviour was modelled using the Drucker–Prager cap model. The paper shows that the proposed tool
predicts reasonable deformations and stresses in block pavements. The results of the simulations were
compared with measured stresses from the full-scale accelerated pavement testing and a good agreement
was observed.

1. Introduction and motivation (1989), Everseries Pavement analysis programs (1999), Huang
(2004). With these approaches stress states due to edge load-
Block pavements are an ecological, economical and visually
ing, especially in the upper base courses, cannot be captured
appealing type of road surface. Especially in communal areas,
realistically. For this reason, numerical simulation tools based
the large number of design possibilities they deliver, such as
on the finite element method were developed. One of the first
shape, colour and texture, are greatly appreciated. Nevertheless,
numerical tools was proposed by Huurman et al. (1992), in
the use of block pavements is lower than one would expect.
which rigid elastically bedded blocks are connected with springs
One of the reasons is a lack of adequate performance prediction
and in a subsequent calculation the loaded concrete block is
tools, especially for predicting damage in areas with high ver-
analysed. Further on, some similar models were developed,
tical and horizontal loads due to brake and acceleration forces.
(Nishizawa et al. 1984, Nishizawa 2003, Hassani 2006,
Responsible for the performance of a paving block structure are
Huurman 2006) which all had in common a linear elastic int-
(i) the material behaviour and the thickness of the base courses,
eraction behaviour between paving blocks and a rather simple
(ii) the material quality and the shape/dimension of the paving
material behaviour of the underlying base courses. An interest-
blocks and (iii) the type of vertical joints, defining the interaction
ing model has been developed by Ascher (2003), where the non-
behaviour between paving blocks. In many engineering design
linear elastic Dresdner model (Gleitz 1996) has been assigned to
concepts this quite complex structural system is not captured
the base course and the performance of the model was evaluated
realistically enough.
by experiments in Lerch (2005). Within these works also the
Since some design methods have evolved from, and hence
horizontal resistance of paving block structures has been studied
are heavily based on, concepts for asphalt pavement structures,
intensively.
for the determination of relevant stress states in the base courses,
This paper presents a finite element method-based numerical
blocks are modelled as continuous layer with an effective simulation tool for paving block pavements, with emphasis on
stiffness, e.g. in Janda (2004). In this procedure, stress states characterisation of the interaction behaviour between paving
within the base courses are obtained on the basis of closed- blocks. An orthotropic elasto-plastic material law in tangen-
form solutions for elastic half-space systems with infinite lateral tial direction and a non-linear elastic relationship in normal
dimension. The basic solution therefore was delivered by directions were finally used to describe the paving block int-
Boussinesq (Taylor 1963, Timoshenko and Goodier 1987), eraction within the numerical simulation tool. Additionally, a
which was extended by Burmister to multi-layered elastic sys- realistic description of all other parts of a paving block struc-
tems (Burmister 1943, 1958). The solutions of Burmister can be ture allowed for reliable stress predictions within the whole
found in a large number of software programs, e.g. in BISAR superstructure.
CONTACT J. Füssl josef.fuessl@tuwien.ac.at
© 2016 Informa UK Limited, trading as Taylor & Francis Group
918 J. FÜSSL ET AL.

The accuracy of the finite element calculations, at least the

Table 1. Test sections.
response due to vertical loads, could be assessed by compar-
ing the results to measurement data from full-scale accelerated Test section Type of paving block Instrumentation
pavement tests (APT) on two paving block test sections. These 5 Rectangular interlocking 4 SPC
tests were conducted by an international cooperation consisting block 200/200/100 (mm)
6 Double T block 100 (mm) None
of two research institutes, the Vienna University of Technology 7 (pervious concrete) Rectangular interlocking 2 SPC
(VUT) and the Swiss Federal Laboratories of Material Science block 200/200/100 (mm)
and Technology (EMPA), and one industrial partner, whose Note: SPC – soil pressure cells.
main activity is the development, production and sale of con-
crete block and slab pavers.
In the following section, the APT, including the layout, Table 2. Test program (load 65 kN, 1.06 N/mm2 , super single tire).
instrumentation, test program and results, are described. The Number of rollovers Max. rutting depth
identification experiments for the mechanical behaviour of the Test section Projected Applied Loading classa (mm)
vertical joints, the unbound base courses and the concrete blocks 5 285,000 250,000 IV > 20.0
are proposed in Section 3. The identified material properties 6 285,000 285,000 III 19.5
serve as input to a numerical simulation tool which is presented 7 285,000 285,000 III 3.5
in Section 4. In Section 5, the simulation results are compared a According to Austrian design standard (RVS 03.08.63 2008).

to measured values from the APT. Finally, a summary of the

results is given and conclusions are drawn in Section 6.
long and 5.0 m wide, as illustrated in Figure 1(a). The length of
each test section was adjusted to the length between the corner
2. Full-scale APT jacks (7.2 m) of the MLS10 (see Figure 2), to avoid the machine
standing on the pavement during traffic simulation. All sections
Within full-scale APT the whole traffic load of the expected
were surrounded with a concrete foundation, serving as support
service life of a pavement structure is simulated in real scale.
for the MLS10 and as confinement for the pavement sections
This leads to two major advantages, namely, (i) the ‘total system’
themselves. The superstructure of the test sections, illustrated
response is provided within (ii) a relatively short period of time,
in Figure 1(b), was chosen according to the requirements in
compared with mechanical testing at a smaller scale and the
regulation RVS 03.08.63 (2008), describing block/slab pave-
many years required for long-term monitoring, respectively. For
ments structures without compound effect on unbound base
this reason, full-scale accelerated pavement testing has a long
courses. Bedding sand (2/4 mm), jointing sand (0/4 mm) and
tradition and has been conducted for almost 50 years (Powell
base materials (0/63 and 0/32 mm) were chosen according to
2012; Steyn 2012a). A comprehensive overview of the so far per-
the requirements in regulations RVS 08.18.01 (2009) and RVS
formed APTs and the obtained significant findings can be found
08.15.01 (2010).
in the two NCHRP synthesis reports Hugo and Martin (2004)
Test Sections 5 and 7 were instrumented in the main loading
and Steyn (2012b), respectively. According to these reports,
line of the MLS10 as shown schematically in Figure 1(a). Soil
most of the APTs have been performed on asphalt pavements
pressure cells (SPC) of type EBKE 20/30 K10 A were installed
or dowelled concrete slab pavements. The few APTs performed
on the surface of the subgrade and the upper base course (see
on paving block structures have been carried out in Australia,
Figure 1(a) and (b)).
New Zealand, South Africa and Japan. A good overview of these
developments together with insights on design and construction
of paving block structures based on his extensive work in this
2.2. Accelerated pavement tests
field for decades has been given by Shackel (1990). Details of his
performed heavy vehicle simulations on block pavements can For the traffic simulation, shown in Figure 2, the Mobile Load
be found in Shackel (1980, 1982, 1980) and in the referenced Simulator MLS10 from Switzerland (Partl 2008) was used. The
publications therein. To the knowledge of the authors, the last MLS10 was equipped with Goodyear 455/40 R22.5 super single
APTs on block pavements were performed by Sharp around tires, exhibiting a tire pressure of 1.06 N/mm2 (10.6 bar) and
2000. Some information about these experiments can be found able to apply a wheel load of 65 kN. At the beginning of the test
in Hugo and Martin (2004), Sharp et al. (1999). Within the procedure, this wheel load was adjusted with a calibrated static
present work, the Mobile Load Simulator (MLS10) has been scale and controlled with a mobile weight in motion device
applied for the first time to paving block superstructures. The (WIM), ensuring an exact wheel load of 65 kN. Before the traffic
MLS10 itself, the test sections and the testing program will be simulation a pre-loading phase with reduced wheel load and a
presented in the following. speed of 7.2 km/h was performed on each test section for about
15 min (1000 passes). After this initial phase the full load of
65 kN was applied and the speed increased to 22 km/h. During
2.1. Test sections
the tests no lateral wandering or application of water was carried
The APT sections and instrumentation were designed and con- out. The MLS10 was able to simulate 38,400 load rollovers per
structed in Summer 2010. Four different pavements sections day. Table 2 gives the number of the projected and applied
with paving slabs and three sections with paving blocks were rollovers of each test section as well as the respective loading
built. Within this paper only the paving block sections, Sections class according to the Austrian design standard (RVS 03.08.63
5–7, will be considered (see Table 1). Each test section were 6.4 m 2008).
 INTERNATIONAL JOURNAL OF PAVEMENT ENGINEERING 919

20/20/10 cm concrete blocks

3 cm sandbed

2 soil pressure cells on top

of upper base course
U2/O2 20 cm unbound base course (Section 5/6)
20 cm pervious concrete (Section 7)

6.4 m
U1/O1

30 cm unbound base course

5.0 m
2 soil pressure cells on top
soil pressure cells of subgrade
subgrade

type EBKE 20/30 K10A (gloetzl.de)

(a) (b)

Figure 1. (a) Layout of one test section (Section 5) including position of SPC, and (b) pavement structure.

electr.
control generator

4 loading wheels

corner transport
jack wheels

7.2 m 65 kN

Figure 2. Mobile Load Simulator MLS10 with super single tires.

During testing the main problem was the rapid loss of joint- Section 6 exhibited a rut depth of 20 mm at the intended
ing sand, since the sand joints within the newly constructed number of load passes, but no damaged blocks could be iden-
slab test sections had no time to consolidate and to develop tified. Small horizontal deformations and contact between two
conventional inter-locking status. Therefore, it was necessary to pavers were observed in the touchdown of the super single tires.
add jointing sand during the maintenance stops of the MLS10. The joint width of almost all joints in the middle of the load line
However, inspection and refilling of joints are regular mainte- decreased to below 1 mm.
nance actions for paving block structures during their lifetime. The pavement structure with pervious concrete (Section 7) is
For further tests on these types of pavements it is, thus, recom- a relatively new construction method in Austria for which only
mended that new test sections are exposed to environmental limited knowledge and experience is available. For this reason, it
conditions (rain, dust) for a few weeks before the start of the was interesting to observe that the performance of this pavement
APT until the jointing material consolidates. Nevertheless, it was excellent. Only small deformations could be measured and
can be stated that the execution of the APT and the projected no visible damaged was observed.
load applications were completed successfully (see Table 2). In summary, it can be assumed that the performance of
On Section 5 the APT was stopped before reaching the pro- the test sections showed a realistic behaviour during the APTs
jected load passes because of a rut depth larger than 20 mm, compared to real traffic situations. A significant amount of time
which is the threshold value in Austria and also the maximum could be saved compared to test sections on public roads with
tolerance for the operating of the MLS10. The largest defor- high traffic volumes. The only shortcoming of the MLS10 is that
mation was induced directly above the SPC installed at the no realistic breaking or acceleration forces, which often have
formation level of the unbound base course. It was concluded a major impact on the performance of a pavement structure,
that the compaction of the base course in the surrounding of can be applied on the road surface. In future, this disadvan-
the instrumentation was unsatisfactory. Although considerable tage could be overcome by using a numerical simulation tool,
deformation was measured, no blocks were broken and only where a variety of loads can be applied without much additional
minimal spalling on two loaded stones occurred. The width of effort.
joints in the middle of the load line after construction was less The obtained results from the installed measuring devices
than the allowable tolerances, which may have contributed to and their evaluation will be presented in the following
the exhibited performance. section.
920 J. FÜSSL ET AL.

2.3. Results of APT thickness direction (direction of joint width) δun was obtained,
reading
As mentioned before, Section 5 was equipped with 4 SPC, ins-
σn = E0 δunn , (2)
talled at the top surface of the subgrade (U1 and U2) and
the top surface of the upper base course (O1 and O2) and where E0 denotes a secant-bedding modulus and n is a dimen-
Section 7 with 2 SPC at the top of the upper base course (O1 sionless parameter.
and O2). The position and arrangement of these measurement A detailed description of the test-setup, the test program
devices are shown in Figure 1. The SPC delivered measure- and the evaluation of all three experiments can be found in
ment values with a frequency of 100 Hz during the whole test Füssl et al. (2015a). The identified parameters as input for the
period. The obtained data within one illustrative second (100 numerical simulation tool are finally: ϕh = 1.21, ϕv = 0.58,
measured values per SPC) is shown in Figure 3, for the SPC τmax = 2.41 N/mm2 for the tangential joint behaviour, and
O1 and U1 of Test Section 5. It can be seen that the peak E0 =11.482 N/mm3 and n = 1.795 for the normal joint
values of the compressive stresses due to a load pass can be behaviour.
captured with sufficient accuracy, even if a higher frequency
could be recommended for the SPC O1. For further evaluation, 3.2. Concrete paving block properties
the maximum and minimum pressure values of each minute of
the whole testing time were used. Figure 4 shows this data for The stiffness of the concrete slabs were obtained by ultrasonic
the SPC U1 and U2 of Section 5. The obtained minute minima measurements. The ultrasonic measurement set-up and a con-
and maxima of the other SPC can be found in Figures A1 crete specimen with the two measuring directions are shown in
and A2 of the Appendix. In Section 5, the mean value and stan- Figure 7. By measuring the transit time T of the ultrasonic wave
dard deviation of the pressure amplitudes (difference between through the specimen, the wave velocity v was determined and
minute maxima and minima) are compared to values obtained further the modulus of elasticity E could be obtained by
from finite element simulations.
ρv 2 (1 + ν)(1 − 2ν)
E= , (3)
(1 − ν)
3. Identification experiments
Responsible for the overall performance of a pavement structure where ρ denotes the mass density of concrete, which was
with paving blocks are the following parts: (i) the joint behaviour measured for each sample (mean value 2.37 t/m3 ), and ν is
between blocks, (ii) the mechanical properties of the blocks itself the Poisson’s ratio assumed to be 0.15. From 24 measurements
and (iii) the behaviour of the base courses including the sub- on 4 samples a mean value of the modulus of elasticity of
grade. For a reliable simulation tool the mechanical behaviour 45,334 N/mm2 with a standard deviation of 7.3% was
of all these parts needs to be characterised. For this reason, obtained.
different identification experiments have been carried out which
will be presented in the following. 3.3. Behaviour of base courses
Unbound base course materials often exhibit non-linear elastic
3.1. Joint behaviour and plastic material behaviour that cannot be described with
For the determination of the joint behaviour, three different the linear elastic model. The non-linear (due to the granular
new experimental set-ups were developed, which are illustrated material structure) elastic behaviour is often modelled by the
schematically in Figure 5. From the first and second test set- simple two-parameter k − φ model (Hicks and Monismith
up (Figure 5(a) and (b)), the frictional behaviour in vertical 1971), which suggests that the resilient modulus is propor-
and horizontal direction, respectively, was obtained. By varying tional to the mean stress raised to a power, or further devel-
the confinement force in normal direction to the joint and by oped models like the three-parameter model according to Uzan
assuming a frictional behaviour according to Mohr–Coulomb, (1022), the Dresden model (Gleitz 1996) and the isotropic or
a vertical and horizontal friction angle describing the tangential anisotropic Boyce model (Boyce 1980, Hornych et al. 1998). A
joint behaviour could be determined from these experiments. good overview of the basic models for the resilient response
Moreover, the amount of tangential deformation γe before plas- of unbound layers is given in Lekarp et al. (2000). The plastic
tic effects occur as well as a maximum shear stress τmax , which is material behaviour is either described by classical failure criteria,
independent of the confinement stress, were identified. With all according to Mohr–Coulomb or Drucker–Prager, or/and by the
this information the tangential joint behaviour can be described ‘shakedown’ concept described in Werkmeister et al. (2001).
by an anisotropic friction criterion, illustrated in Figure 6 and A comprehensive model for granular layers, from non-linear
reading elastic material behaviour up to plastic shakedown considera-
|τh |2 |τv |2 tions, and the implementation into a finite-element code can be
2
+ = 1, (1) found in Chazallon et al. (2009). The drawback of such com-
τhcrit τvcrit 2
plex material models is the large number of material properties
with τvcrit = ϕv σn ≤ τmax and τhcrit = ϕh σn ≤ τmax , where required for these models, which are mainly obtained from
ϕv and ϕh denote the vertical and horizontal friction angle, comprehensive static and cyclic triaxial testing programs, e.g.
respectively, obtained from experiments. presented in COURAGE (1999).
From the third experiment (see Figure 5(c)), a relationship For this reason and since the main focus of this work is
between the joint normal stress σn and the joint deformation in on the interaction behaviour between paving blocks, a simple
 INTERNATIONAL JOURNAL OF PAVEMENT ENGINEERING 921

Figure 3. Data obtained within one illustrative second from SPC O1 and U1 of Test Section 5.

Figure 4. Data obtained from SPC U1 and U2, installed on top of subgrade, of Section 5.

(a) (b) (c)

Figure 5. Schematic illustration of identification experiments of mechanical joint behaviour in (a) horizontal, (b) vertical and (c) normal direction.

Figure 6. Anisotropic friction model implemented in Abaqus to describe the tangential behaviour between vertical joints.
922 J. FÜSSL ET AL.

oscilloscop measuring
directions
pulser−receiver
16 cm
6 cm

transducer

(a) (b) (c)

Figure 7. (a) Ultrasonic measurement setup with dummy sample, (b) ultrasonic transducer with honey as contact agent, and (c) tested concrete samples.

(a) (b)

Figure 8. (a) Loading plate test results on the top of the subrade, the lower base course and the upper base course compared to the corresponding numerical results, and
(b) numerically obtained vertical deformation fields for a loading plate test at the top of the upper base course.
 INTERNATIONAL JOURNAL OF PAVEMENT ENGINEERING 923

Figure 9. Geometry, load and boundary conditions of finite element model.

Figure 10. Exemplary finite element mesh, where the loaded blocks are finer discretised.

Figure 11. Vertical deformation fields of Section 5, for the two load cases (A and B) shown in Figure 9.

yet realistic approach was chosen to describe the constitutive model. In Figure 8(a), the experimentally and numerically obt-
relationship for the base courses in the proposed numerical ained load deformation relationships are compared with each
simulation tool. Five conventional loading plate tests were car- other. It can be seen that by assigning a modified Drucker–
ried out during the construction process on the top of the Prager cap model to all base courses, a calibration of the mat-
subgrade, the lower base course and the upper base course. erial properties allows for a very good reproduction of the
The experimentally obtained load deformation relationships loading plate tests numerically. The modified Drucker–Prager
were used to determine elasto-plastic material parameters of cap model and the required parameters are described in the
all three layers by reproducing the tests with a finite element Appendix.
924 J. FÜSSL ET AL.

Figure 12. Relevant vertical stress states σ33 at the top of the base courses and the subgrade for Test Section 5, and the position of the relevant SPC.

4. Finite element model

In this section, a finite element model for the test sections on
which the APT have been carried out is presented. For the
performance evaluation of the model, the obtained simulation
results are then compared to the measurements taken during
the APT.
The finite element model of Test Section 5 with the two
investigated load cases and all boundary conditions is shown
in Figure 9. Load case A represents a wheel load directly above
a continuous transverse joint while the wheel load according
to load case B is located exactly in-between two such joints.
The dimensions of the analysed paving blocks are 200/200 mm
with a thickness of 100 mm. The elastic properties of the blocks
were obtained from ultrasonic measurements as described in
Section 3.2. The interaction between slabs (joint behaviour) is
considered by an anisotropic friction criterion in tangential dir-
ection and with a non-linear elastic relationship in normal dir-
ection. The identification of this interaction behaviour and the Figure 13. Mean values and standard deviation of maximum pressure values
corresponding material properties were proposed in Section 3.1. obtained from SPC compared to corresponding results from finite element
simulations.
To the upper as well as lower base course and the subgrade a
modified Drucker–Prager cap model, described in Section 3.3,
was assigned. Since for the sandbed no identification experi- lifting of the paving blocks was allowed. All other layers were
ments were available, the same plastic behaviour as for the upper tied together. The interaction with the surrounding base courses
base course was assumed and a slightly stiffer elastic behaviour and subgrade material (lateral boundary of the modelled area
with an elasticity modulus of E = 350 N/mm2 and a Poisson’s shown in Figures 9 and 10, respectively) is roughly modelled
ratio of ν = 0.3. Between the paving blocks and the sandbed a by lateral elastic foundations with a lateral modulus of sub-
friction coefficient of 0.6 was assigned and in normal direction grade reaction according to suggestions of Austrian regulations.
 INTERNATIONAL JOURNAL OF PAVEMENT ENGINEERING 925

(a) (b)

Figure 14. Future applications of the simulation tool: (a) study of the horizontal resistance of block pavement superstructures, and (b) determination of the influence of
arch supporting effects on the load bearing capacity of paving block superstructures.

A lateral pressure is applied to the concrete slabs, describing course, respectively. At Section 7 two SPC were installed at the
the restraining of the test sections and to activate a low coh- top of the subgrade. The measurement data was consolidated to
esion within the vertical joints, as seen in joint experiments mean values and the associated standard deviation of amplitude
(Füssl et al. 2015a). values (min/max values).
The discretisation is done with 20-node hexahedron ele- Figure 13 shows the comparison between the compressive
ments for the sandbed and 8-node hexahedron elements for all stress values obtained from the SPC and the corresponding
other parts. An exemplary mesh with the corresponding model numerical results. The FEM values represent the mean vertical
parameters is shown in Figure 10. pressure over an area of 200/300 mm, equal to the measurement
Subsequently, static calculations in Abaqus for different pos- surface of the SPC. The individual lower values denote the
itions of the super single tire were performed. The wheel load mean pressure of a fictive SPC built-in with an eccentricity
of 65 kN was applied linearly within a maximum of 10 steps, of 50/50 mm at the upper base course and 200/200 mm at the
and no more than 16 equilibrium iterations had to be carried subgrade, respectively.
out per step. The vertical deformation fields, obtained for the For all measuring points, the experimentally and numerically
two load cases (A and B), are shown in Figure 11. An analysis obtained pressure values are in reasonable agreement. For a
of Figure 11 suggests that the models predictions of the vertical profound validation of the simulation tool, however, additional
displacements fields are reasonable. Mainly responsible for this experimental data would be required.
is an appropriate implementation of the joint behaviour and the
interaction between the slabs and the sandbed.
In the following, the relevant load cases and the resulting 6. Summary and concluding remarks
stress states obtained from the finite element simulations, which Within this paper, APT using the Mobile Load Simulator
deliver extreme values at the positions of the SPC in the APT (MLS10) on block pavements with sand filled joints were pre-
test sections, are presented. Figure 12 shows the relevant vertical sented. Two test sections were equipped with SPC, allowing the
stress states at the top of the base courses and the subgrade. An recording of relevant pressure stress states in the base courses.
influence of the load case on the distribution of vertical pressure These stress states were then compared to values obtained from
can only be observed at the top of the upper base course. Under a numerical simulation tool based on the finite element method,
load case A, where the wheel load is directly above a continuous where the necessary material properties were acquired from dif-
transverse joint (see Figure 11), the maximum vertical stress is ferent identification experiments. The main results and findings
slightly higher compared with the one under load case B. The can be concluded as follows:
same effect was also obtained from the APTs and can be seen
in Figure A1. However, in contrast to concrete slab structures • The operation of the MLS10 on the three different test
(investigated in Füssl et al. (2015, 2015b)), the load position has sections worked properly, and thus, this testing method
no major impact on the vertical stress distributions in the base can be recommended also for superstructures with paving
layers. blocks. The installed SPC on the top of the upper base
In the next section, the measured stresses during the APT are course and the subgrade were able to capture compressive
compared to the numerically obtained results. stress peaks due to a tire rollover.
• On the basis of three different joint experiments, an app-
ropriate characterisation of the tangential as well as nor-
5. Comparison of experimental and numerical results
mal mechanical behaviour of the sand filled joints was
In the following, the results obtained from the two test sections possible. The obtained mechanical behaviour could be
(equipped with measuring devices) of the APT are compared to implemented accurately into the finite element model,
results from finite element simulations. Four SPC were installed using Mohr–Coulomb friction in tangential direction and
in Section 5, two at the top of the subgrade and the upper base a non-linear elastic material law in normal direction.
926 J. FÜSSL ET AL.

• For the first time, the complex bedding and interaction Everseries Pavement analysis programs, 1999. Washington State
behaviour of paving blocks could be considered within a Department of Transportation. Olympia, WA.
numerical simulation tool realistically. A comparison bet- Füssl, J., et al., 2015. Mechanical performance of pavement structures
with paving slabs – Part II: Numerical simulation tool validated by
ween the experimentally- and numerically-obtained pres- means of full-scale accelerated tests. Engineering Structures, 98, 221–229.
sure stress states in the base courses showed a reasonable doi:10.1016/j.engstruct.2014.10.055.
agreement. Nevertheless, the punctual conducted pressure Füssl, J., Kluger-Eigl, W., and Blab, R., 2015a. Experimental identification
measurements cannot be considered as sufficient for a and mechanical interpretation of the interaction behaviour between
profound validation of the proposed numerical simulation concrete paving blocks. International Journal of Pavement Engineering,
17(6), 478–488. doi:10.1080/10298436.2014.993205.
tool. Füssl, J., Kluger-Eigl, W., and Blab, R., 2015b. Mechanical performance of
In a next step, this simulation tool will be used to study pavement structures with paving slabs – Part I: Full-scale accelerated tests
as validation for a numerical simulation tool. Engineering Structures, 98,
two important load transfer mechanisms in block pavement
212–220. doi:10.1016/j.engstruct.2014.10.054.
structures which were not considered sufficiently so far: (i) the Gleitz, T., 1996. Beitrag zur rechnerischen Erfassung des nichtlinearen
load transfer of horizontal loads into the structure (exemplarily Spannungs-Verformungsverhaltens ungebundener Tragschichtmateri-
shown in Figure 14(a)) and (ii) the influence of arch support- alien in flexiblen Straßenkonstruktionen [Contribution to the
ing effects on the load bearing capacity (exemplarily shown in computational detection of non-linear stress-strain behavior of unbound
base course materials in flexible road constructions]. Dissertation am
Figure 14(b)). Basic numerical investigations of the latter mech-
Lehrstuhl Straßenbau der Technischen Universität Dresden.
anism have already been performed and can be found in Hengl Hassani, A., 2006. Modeling and structural design of a concrete block
and Füssl (2016). Both mechanisms are strongly influenced by pavement system. In: 8th International Conference on Concrete Block
the joint behaviour, the block shape and dimensions, and the Paving. San Francisco, CA.
laying pattern. All this can be considered very well with the Hengl, H.L. and Füssl, J., 2016. The influence of superelevated profiles
of paving block structures on their load-bearing behavior. Engineering
developed simulation tool, and thus, meaningful results can be
Structures, 117, 195–203. doi:10.1016/j.engstruct.2016.03.003.
expected. Hicks, R.G. and Monismith, C.L., 1971. Factors influencing the resilient
behavior of granular materials. Transportation Research Record, 345,
15–31.
Acknowledgements Hornych, P., Kazai, A., and Piau, J.M., 1998. Study of the resilient behaviour
The writers thank TVFA Vienna for the good cooperation, helpful com- of unbound granular materials. In: Proceedings of Fifth Conference on
ments and the conduction of the experiments. Financial support by the FFG Bearing Capacity of Roads and Airfields. Trondheim, vol. 3, 1277–1287.
Austrian Research Promotion Association is gratefully acknowledged. Huang, Y.H., 2004. Pavement analysis and design. 2nd ed. Upper Saddle
River, NJ: Pearson-Prentice Hall.
Hugo, F. and Martin, A.L.E., 2004. Significant findings from full-scale
Disclosure statement accelerated pavement testing. NCHRP SYNTHESIS 325, Transportation
Research Board. Washington, DC.
No potential conflict of interest was reported by the authors. Huurman, M., 2006. Validation of the new dutch design method for
concrete block road pavements. In: 8th International Conference on
Concrete Block Paving. San Francisco, CA.
Funding Huurman, M., Houben, L.J.M., and Kok, A.W.M., 1992. Development of
This work was supported by FFG Austrian Research Promotion a three-dimensional finite element model for concrete block pavements.
Association. In: Proceedings Fourth International Conference on Concrete Block
Paving. Auckland.
Janda, G., 2004. Erarbeitung eines Bemessungsvorschlages für Pflaster- und
References Plattenbefestigungen [Development of a design approach for paving
block and slab pavements]. Diplomarbeit am Institut für Straßenbau
Ascher, D., 2003. Generierung von FALT-FEM Eingabedaten für und Straßenerhaltung der Technischen Universität Wien, Fakultät für
Betonpflasterbefestigungen [Generation of FALT-FEM input data for Bauingenieurwesen, Wien.
concrete block pavements]. Diplomarbeit. Technische Universität Lekarp, F., Isacsson, U., and Dawson, A., 2000. State of the Art I: resilient
Dresden. response of unbound aggregates. Journal of Transportation Engineering,
BISAR, 1989. Shell pavement design methodology. Netherlands: Shell 126(1), 66–75.
Petroleum Company. Lerch, T., 2005. Untersuchungen zum Verformungsverhalten von
Boyce, J.R., 1980. A non linear model for the elastic behaviour of granular Betonpflasterkonstruktionen unter praxisrelevanter Belastung [Investiga-
materials under repeated loading. In: International Symposium on Soils tion of the deformation behaviour of concrete block pavements under
under Cyclic and Transient Loading. Swansea, 285–294. simulated traffic loading]. Dissertation. Technische Universität Dresden.
Burmister, D.M., 1943. The theory of stresses and displacements in layered Nishizawa, T., 2003. A tool for structural analysis of block pavements
systems and applications to the design of airport runways. In: Proceedings based on 3DFEM. In: Proceedings of the 7th International conference
of the Highway Research Board. Washington, DC. on Concrete Block Paving. Sun City, South Africa, 1–4.
Burmister, D.M., 1958. Evaluation of pavement systems of the WASHO Nishizawa, T., Matsono, S., and Komura, M., 1984. Analysis of interlocking
road test by layered systems methods. In: Highway Research Board, block pavements by finite element method. In: Proceedings of the
Bulletin 177. Washington, DC. 2nd International Conference on Concrete Block Paving. Delft, The
Chazallon, C., et al., 2009. Modelling of rutting of two flexible pavements Netherlands, 80–85.
with the shakedown theory and the finite element method. Computers Partl, M.N., 2008. Full-scale accelerated pavement testing APT at
and Geotechnics, 36, 798–809. EMPA – the new mobile load simulator. Duebendorf, Switzerland: EMPA
Chiroux, R.C., et al., 2005. Three-dimensional finite element analysis of soil Duebendorf.
interaction with a rigid wheel. Applied Mathematics and Computation, Powell, R.B., 2012. A decade of full-scale accelerated pavement testing. In:
162, 707–722. Jones, D., Harvey, J., Al-Qadi, I.L., and Mateos, A., eds. Advances in
COURAGE – Construction with Unbound Road Aggregates in Europe, pavement design through full-scale accelerated pavement testing. London:
1999. Final Report, European Commission – DGVII, 4th Framework Taylor & Francis Group. 13–22.
Programme. Road Transport Research.
 INTERNATIONAL JOURNAL OF PAVEMENT ENGINEERING 927

RVS 03.08.63, 2008. Straßenplanung – Bautechnische Details – Oberbaube- Sharp, K.G., Johnson-Clarke, J.R., and Fossey, D.W., 1999. A review of the
messung [Road design - structural details - design of superstruc- Australian ALF program. In: Presented at International Conference on
ture], Österreichische Forschungsgesellschaft Straße-Schiene-Verkehr, Accelerated Pavement Testing. Reno, October.
Vienna, Austria. Steyn, W.J.vdM., 2012a. A history of modern accelerated performance
RVS 08.15.01, 2010. Technische Vertragsbedingungen – Unterbauplanum testing of pavement structures. In: Jones, D., Harvey, J., Al-Qadi, I.L.,
und ungebundene Tragschichten [Technical specifications - subgrade and and Mateos, A., eds. Advances in pavement design through full-scale
unbound base courses], Österreichische Forschungsgesellschaft Straße- accelerated pavement testing. London: Taylor & Francis Group. 3–12.
Schiene-Verkehr, Vienna, Austria. Steyn, W.J.vdM., 2012b. Significant findings from full-scale accelerated
RVS 08.18.01, 2009. Technische Vertragsbedingungen – Pflasterarbeiten, pavement testing. NCHRP SYNTHESIS 433, Transportation Research
Randbegrenzungen [Technical specifications - paving block construc- Board. Washington, DC.
tions, surrouding bordering], Österreichische Forschungsgesellschaft Taylor, D.W., 1963. Fundamentals of soil mechanics. New York: Wiley.
Straße-Schiene-Verkehr, Vienna, Austria. Timoshenko, S.P. and Goodier, J.N., 1987. Theory of elasticity. 3rd ed.
Shackel, B., 1980. An Experimental investigation of the roles of the bedding New York: McGraw-Hill.
and jointing sands in the performance of interlocking of concrete block Uzan, J., 1985. Characterization of granular materials. Transportation
pavements. Concrete/Beton No. 19, Johannesburg, South Africa. Research Record, 1022, 52–59.
Shackel, B., 1982. An experimental investigation of factors influencing Werkmeister, S., Dawson, A., and Wellner, F., 2001. Permanent
the design of interlocking concrete block pavement for roads. In: deformation behavior of granular materials and the shakedown concept,
Proceedings of the Australian Road Research Board. Vermont South, Transportation Research Record 1757. Journal of the Transportation
Victoria, Australia, 11, Part 2, 6–15. Research Board, 75–81.
Shackel, B., 1990. Design and construction of interlocking concrete block
pavements. London: Elsevier Applied Science.
Shackel, B., 1980. The performance of interlocking pavements under
accelerated trafficking. In: Proceedings First International Conference on
Concrete Block Paving. London: Concrete Publishing Co.
928 J. FÜSSL ET AL.

Appendix 1. Data obtained from SPC

Figure A1. Data obtained from SPC O1 and O2, installed on top of upper base course, of Section 5.

Figure A2. Data obtained from SPC U1 and U2, installed on top of subgrade, of Section 7.

Appendix 2. Drucker–Prager cap model

(a) (b)

Figure B1. Yield surface of the modified Drucker–Prager cap model (a) along the hydrostatic axis and (b) in the meridial plane.

The modified Drucker–Prager cap model is illustrated in Figure B1. It consists of a failure surface describing shear failure, which reads

Fs = t − p tan β − d = 0 (B1)

and a surface defining the plastic behaviour under predominantly volumetric compressive stress states, reading

 2
Rt
Fc = (p − pa )2 + − R(d + pa tan β) = 0, (B2)
(1 + α − α/ cos β)
 INTERNATIONAL JOURNAL OF PAVEMENT ENGINEERING 929

Table B1. Cap hardening stress/strain relationship (derived from experimental data in Chiroux et al. (2005)).
Hydrostatic yield Volumetric plastic Hydrostatic yield Volumetric plastic
stress (N/mm2 ) strain (–) stress (N/mm2 ) strain (–)
0 0 0.10 0.0845
0.01 0.0141 0.15 0.1021
0.03 0.0373 0.30 0.1241
0.05 0.0551 0.40 0.1335
0.07 0.0690 0.50 0.1405

Table B2. Material parameters.

Density Elasticity Modified Drucker–Prager/cap model
ρ E ν d β R 0 α K
(t/m3 ) (N/mm2 ) (–) (N/mm2 ) (◦ ) (–) (–) (–) (–)
Upper base course 1.8 240 0.4 1.0 45 0.0005 0 0.03 0.8
Lower base course 1.8 180 0.4 0.8 45 0.0005 0 0.03 0.8
Subgrade 1.5 60 0.4 0.3 35 0.0005 0 0.03 0.8

√
where p = −1/3σii is the equivalent pressure stress, q = 3/2(σ d : σ d ) the Mises equivalent stress and r = (9/2σ d · σ d : σ d )1/3 the third invariant of
deviatoric stress with σ d = σ + pI. A deviatoric stress measure is defined as
    
q 1 1 r
t= 1+ − 1− . (B3)
2 K K q

The evolution parameter, pa , reads

pb − Rd
pa = , and (B4)
(1 + R tan β)
the transition surface Ft between Fs and Fc is given by

 2
α
Ft = (p − pa )2 + t − (1 − )(d + pa tan β) − α(d + pa tan β) = 0. (B5)
cos β

The evolution of hardening is derived from the experimental results in Chiroux et al. (2005). The implemented relation between the hydrostatic yield
stress and the volumetric plastic strain is given in Table B1. Finally, with the parameters shown in Table B2 the loading plate tests could be reproduced
very well.
Copyright of International Journal of Pavement Engineering is the property of Taylor &
Francis Ltd and its content may not be copied or emailed to multiple sites or posted to a
listserv without the copyright holder's express written permission. However, users may print,
download, or email articles for individual use.