2nd Workshop on Four Point Bending, Pais (ed.), © 2009. University of Minho.

ISBN 978-972-8692-42-1

The five-point bending test applied on wearing courses laid on

orthotropic steel decks

A. Houel
CETE de Lyon, LRPC de Lyon, Bron, France
L. Arnaud
Université de Lyon, Lyon, France; Ecole Nationale des Travaux Publics de l’Etat, Vaulx-en-Velin,
France

ABSTRACT: This paper deals with the evolving behaviour of wearing courses on steel ortho-
tropic decks, such as the French Millau viaduct bituminous mix or an ultra high performance
concrete (UHPC) pavement. This is of great importance when dealing with durability. A five-
point bending fatigue test was developed since 2003 at the ENTPE laboratory. It enables to test
various bituminous concrete mixes. Recent works on UHPC pavements on steel orthotropic
decks are considered to improve service life of such structures. This paper presents the conti-
nuous follow-up of material throughout the fatigue test and it enables to detect cracks through
the pavement. Two means are used: displacement sensors and ultrasonic measurements. The
second method clearly leads to follow-up the mechanical evolutionary behaviour of the wearing
course.

1 INTRODUCTION AND OBJECTIVES

1.1 Context
The metallic plates of steel decks are very flexible and consequently wearing courses applied to
such supports are submitted to very high levels of strain. Thus, a special design is of great im-
portance when dealing with steel structure durability. Usually wearing course on such bridges is
made of bituminous concrete: this is the case of the Millau Viaduct, the highest bridge in the
world recently built in France (Héritier et al., 2005). Two characteristic kinds of damage in the
bituminous concrete pavement are observed under traffic loads:
− on the one hand, fatigue cracks are generated in the thickness of the bituminous concrete
layer at right angles to the orthotropic plate stiffeners, due to the tensile stress created. They
propagate from the layer surface through the wearing course and can reach the sealing sheet
which protects steel plate from corrosion.
− on the other hand, shear cracks may appear at the interface between steel plate and wearing
course at the tack coat.
To improve the mechanical behaviour, a new kind of wearing course is also considered: an ultra
high performance concrete (UHPC) pavement with metallic fibers. This new solution seems to
be interested since the wearing course will contribute to the mechanical resistance of the deck.
For that, the mechanical transfer at the interface between steel and concrete is fundamental.
This paper presents the first measurements achieved at laboratory thanks on the fatigue test rep-
resentative of the UHPC behaviour laid on orthotropic steel decks.

1.2 Objectives
A five-point bending fatigue test – FPBT – (bending under negative moment) was developed by
the French Laboratoire Central des Ponts et Chaussées (LCPC) in the 1970s with bituminous

311
concrete pavements. It was shown that the results obtained from laboratory tests and in situ ob-
servations on real steel deck were very consistent (Hameau et al., 1981). Nevertheless, the main
disadvantage is that there was not any procedure that enables to obtain a continuous follow-up
of mechanical characteristics of asphalt concrete and a clear detection of cracks.
As a consequence, a new device has been developed since 2003 at the ENTPE laboratory. It
aims at improving the measurement of strains in the bituminous layer by the use of displacement
sensors of better accuracy than strain gages and also at detecting the appearance of cracks by an
objective way by means of a non-destructive test based on ultrasonic wave propagation.
In the first part of this paper, the experimental device and test conditions are described. The
instrumentation is then presented and the strain measurements allow mechanical evolution of
wearing course analysis. An original method based on ultrasonic wave propagation will be con-
sidered. This non-destructive test permits material mechanical evolution follow up throughout
the test. Moreover, under linear viscoelasticity assumption, an inverse analysis is possible. As a
consequence, intrinsic mechanical characteristics of the material may be calculated at each in-
stant during the test. Finally, new progress in these experimental studies will be summarized.

2 THE FIVE-POINT BENDING TEST (FPBT)

2.1 Principle
The FPBT consists in testing a sample reproducing the area located on either side of an ortho-
tropic plate longitudinal stiffener. This is the area where the largest strains are generated be-
cause of traffic, and therefore where the most important fatigue damage in the wearing course is
observed. This test is a fatigue test over several millions cycles.
Samples are constituted generally with three main materials:
− a 12 or 14 mm-thick steel plate reinforced at the center with a welded stiffener,
− a 3 mm-thick sealing sheet,
− the wearing course, whose thickness is variable.
Each sample is held in its center on a rigid frame and is loaded using the device as presented
in Figure 1. The steel plate is embedding in the center, and its two extremities rest on two sim-
ple supports that are adjustable in order to correct the flatness defects of the steel plate. Above
the wearing course, the beam sets to apply the sine compression load. The load is composed by
compression sine cycles at a frequency of 4 Hz (Hameau et al. 1981). The amplitude of the max-
imal load is determined during a calibration phase so as to take into account the mechanical role
of the steel plate in the sample. This load, applied to the steel plate without layer, corresponds to
a stress of 120 MPa at right angles to the weld. Of course, the effort depends on the thickness of
the steel plate. For example, for calibrated 12 mm-thick plates, the maximal effort equals the
load when a strain of 625 µm/m are observed at right angles to the weld, that is to say 32 kN in
compression. Thus, the sine load, applied to the sample with the wearing course, ranges in be-
tween the maximal effort and 10% of this load. A counter measures the number of applied load
cycles. Based on the French bituminous concrete mix standardization, for each coating, one
sample is tested at -10°C, and another at +10°C which is very often the critical case, thanks to a
heat-regulated chamber. As far as UHPC pavements are concerned, each specimen is tested at
the ambient temperature (+23±2°C).
According to the French requirements, a bituminous concrete mix is considered "good" when
no damage can be observed at -10°C after 1 million cycles, and at +10°C after 2 million cycles
(AFNOR 2006). A UHPC pavement is considered as good if no damage can be observed after 2
million cycles at the ambient temperature. No damage means that no crack was generated and
observed with the soapy water. But there is no requirement for the mechanical evolution and the
breaking threshold of the material. That is why a special instrumentation based on ultrasonic
wave propagation is investigated.

312
Figure 1. Experimental device of the FPBT (Laajili 2003)

2.2 Instrumentation
Two strain gauges are pasted on the two side faces the closest of the top face as possible in or-
der to monitor the evolving strains of the wearing course where cracks are likely to appear
(Fig. 2). However, their installation is a delicate and difficult task, and it is possible that their
breaking does not exactly account for crack appearance because of the glue. That is why two
different displacement sensors, type LVDT (Linear Variable Differential Transformer) are posi-
tioned on the upper face of the sample at the center. Both accuracy and a large detection area are
ensured: a first sensor (LVDT 1: ±2.5 mm and Δl=60 mm) is certain to have a measurement
zone where cracks are likely to appear, and a second (LVDT 2: ±1.0 mm and Δl=30 mm) senses
the displacements more precisely.
Moreover several temperature sensors are positioned in the chamber and one at the center on
the upper face of the wearing course.
Finally wave transducers are placed in the central area on the side faces in order to conti-
nuously monitor the pavement modulus where the material is submitted to tensile stress and thus
where cracks are likely to appear.

Figure 2. Locations of sensors on specimens for the FPBT: strain gauges, displacement sensors (LVDT 1
and LVDT 2), and wave propagation transducers (Houel 2008)

2.3 Non-destructive tests by ultrasonic wave propagation

A non-destructive test has been developed to follow-up mechanical characteristics of the ma-
terial. This test consists in transmitting a wave with a first transducer and receiving this wave

313
that propagate throughout the considered medium (Fig. 2). Two kinds of ultrasonic waves are
studied: P-waves or compression waves, and S-waves or shear waves. Two pairs of P-wave
transducers are used to scan different excitation frequencies, and one pair of S-wave transducers
is also positioned and senses the center part of the bituminous concrete layer. From the recorded
signals, the below parameters are followed throughout fatigue tests (Fig. 3):
− the arrival time, which enables to know the wave velocity and by an inverse analysis the
complex modulus of the bituminous concrete,
− the maximum amplitude of ultrasonic waves, and
− the Fast Fourier Transform of received signal.
Few precautions must be taken into account: an appropriate excitation frequency needs to be
chosen carefully to avoid diffraction, scattering and so high attenuation because of the hetero-
geneity of bituminous concrete and the limited size of specimens. As a consequence, the wave-
length should be larger than the grain size (10 mm) (Van Hauwaert 1998, Houel 2004, Arnaud
& Houel 2007). Moreover, wave velocities depend on temperature. So, temperature effects must
be analyzed before choosing adequate excitation frequencies. Characteristics of wave propaga-
tion are presented in Table 1.

Figure 3. Emitted and received signals of P-wave propagation, determination of time delay and damping
coefficient (Houel & Arnaud 2007).

Table 1. Wave velocities, frequency excitations of transducers and wavelengths used and measured at the
beginning of fatigue tests
Wave velocity and wavelength Temperature
-10°C +10°C
Cp 4000 m/s 3600 m/s
P-wave (50kHz)
λ 7 cm 6 cm
Cs 2100 m/s 1800 m/s
S-wave (10kHz)
λ 15 cm 20 cm

2.4 Tested pavements

Three bituminous concrete formulations are presented in this paper. They are characterized by a
unique granulometry but are distinguished by different binders:
− a classic bituminous concrete with 7% polymer-modified binder (F2),
− a bituminous concrete with 7% pure binder (F3), and
− a bituminous concrete with 7% special polymer-modified binder called Styrelf and commer-
cialized by Total (F4).

314
 These mixes were tested on orthotropic specimen. The bituminous mixture laid on the Millau
viaduct (F1) on orthotropic specimen was also tested in laboratory. This special mixture was
specially developed by Eiffage TP (Héritier et al. 2005).
Moreover, an innovative wearing course on an orthotropic specimen has been tested: a
35mm-thick UHPC pavement.

2.5 Results
Maximal and minimal strains are monitored throughout the fatigue test. As expected, both bi-
tuminous concrete mixtures show that strain amplitudes are greater at +10°C than -10°C (Fig. 4)
(Houel 2008). The main differences between the four bituminous mixtures are as follows:
− As far as the Millau viaduct bituminous concrete mixture, we firstly observed that ampli-
tudes increase during the test until approximately 1,000,000 cycles, then a sudden variation
of slopes is recorded from the most accurate sensor.
− As regards the bituminous concrete mixtures F2, F3 and F4 throughout both fatigue tests at -
10°C and +10°C, amplitudes corresponding to one load cycle increase quickly at the begin-
ning of the test, then a sudden and strong increase appears, and finally strain amplitudes do
not increase any more but are relatively constant (Somda 2007, Houel 2008, N’Guyen 2008).
So, as regards the fatigue behaviour and the appearance of cracks, the special bituminous
concrete mixture F1 has a stronger resistance to cracking than the others.
In each case, variations are determined by linear regression, and the intersection of two
straight lines defines the cycle number when cracks appear on the upper side of the bituminous
concrete layer. This experimental set up is very sensitive to the appearance of cracks initiated
from the top face of the bituminous concrete layer.
Figure 5 compares results for the UHPC wearing course and the bituminous concrete layer.
We clearly observed that strain amplitudes are dramatically lower for UHPC layer whereas its
thickness is half the one of bituminous layers. This important result proves that the UHPC layer
supports a large part of the mechanical load and as a consequence leads to strain decrease in the
steel plate. Thus, this results in new designing and this UHPC wearing course could improve
durability of such steel structures.
Table 2 sums up crack detection results from displacement measurements. Hence, displace-
ment evolution shows a sudden and significant variation in the strain amplitudes during the test
that proves the appearance of fatigue cracks in the sensed zone for the bituminous concrete. Be-
sides, tests based on ultrasonic wave propagation will confirm this method. Results are consis-
tent for the same bituminous mixture.

Figure 4. Experimental shapes of strain amplitudes throughout the FPBT measured by LVDT 2 on the
up-per face of specimens at +10°C (Houel & Arnaud 2008b)

315
Figure 5. Experimental shapes of strain amplitude variation throughout the FPBT measured by LVDT 2
on both wearing courses (bituminous concrete mix F1 and UHPC)

Table 2. Number of cycles when crack is detected (Fig. 4)

Wearing course layer Temperature
-10°C +10°C
BC mix F1 900,000 950,000
(Millau viaduct BC) 1,200,000 1,050,000
BC mix F2 - 200,000
BC mix F3 200,000 100,000
BC mix F4 1,000,000 850,000
UHPC No crack at ambient temperature

An example of such results in Figures 6 and 7 shows the mechanical evolution of the bitu-
minous concrete and UHPC layers respectively from P-wave and S-wave transducers through-
out the FPBT. P and S-wave velocities and their amplitude decrease clearly as the number of
load cycles increases. Then signals stabilize before significant amplitudes decrease.
For the Millau viaduct bituminous concrete mixture (F1), the calculated modulus is about
28,500 MPa at the beginning, and decreases to 27,400 MPa after 800,000 cycles. Then, it stabi-
lizes until 3 million cycles before significant decreases. This last stage reveals the appearance of
cracks that are created and propagated as far as the middle of the thickness of the bituminous
concrete pavement. In the second case, the decrease appears throughout the two hundred thou-
sand first cycles like results from LVDT measurements, decreasing from 30,000 MPa to
26,500 MPa.
For the composite structure made of UHPC and steel plate, we expect and observe an initial
higher value for the Young’s modulus (about 60,000 MPa). Then with the fatigue mechanical
evolution, the modulus continuously decreases to 50,000 MPa: one can deduce that the layers
assembly (steel plate and concrete) evolves significantly as function of load cycles. The same
remark is correct for the shear modulus even if measurements are less precise. Plastic effects
appear around the steel connectors in concrete, they are detected by the wave propagation. But
finally, these fatigue effects do not lead to a macro crack in the UHP concrete layer.
Therefore, wave propagation is an efficient way to compare and to monitor the evolutionary
behaviour of different wearing courses on the FPBT, and then it is possible to plot the damage
curve.

316
Figure 6. Norm of the complex modulus calculated from P-wave velocities and their damping by inverse
analysis throughout the FPBT performed on two bituminous mixtures at +10°C (Houel & Arnaud 2008a,
b)

Figure 7. Young’s modulus throughout the FPBT on UHPC layer at ambient temperature (N’ Guyen
2008)

3 CONCLUSIONS

A special design is necessary for wearing courses on steel orthotropic deck, because it is sub-
jected to very severe strains due to vehicle traffic and flexibility of steel structure.
In this paper, the laboratory test, the five-point bending test, is presented and used to charac-
terize the mechanical behaviour of various wearing courses.
The FPBT in laboratory is representative of solicitations applied on wearing courses on steel
orthotropic decks. An accurate instrumentation is defined in this paper and makes it possible to
obtain a continuous follow-up of mechanical evolutionary characteristics of the wearing course
throughout the fatigue test.
Tests achieved on various bituminous concrete mixes lead to assess the fatigue behaviour for
each mix whereas for UHPC layer, the plastic effects in concrete are detected but no macro

317
crack is noticed. This test constitutes an indispensable step for the design of wearing course on
steel orthotropic bridges.

4 REFERENCES

AFNOR 2006. Méthodes d’essai sur banc de fatigue en flexion sous moment négatif. Détermination de la
résistance en fatigue d’une étanchéité/roulement sur tôle métallique, Essais relatifs aux chaussées –
Produits d’étanchéité pour ouvrages d’art, French standard method (NF P 98-286). [In French]
Arnaud L. & Houel A. 2006. Fatigue damage and cracking of asphalt pavement on orthotropic steel
bridge deck, Transportation Research Board Annual meeting Compendium of Papers CD-ROM,
Washington D.C.
Arnaud L. & Houel A. 2007. Fatigue damage of asphalt pavement on an orthotropic bridge deck: me-
chanical monitoring with ultrasonic wave propagation. International Journal of Road Materials and
Pavement Design 8-3.
Hameau G., Puch C. & Ajour A.-M. 1981. Comportement à la fatigue en flexion sous moment négatif,
Bulletin de liaison des Ponts et Chaussées 111. [In French]
Héritier B., Olard F., Saubot M. & Krafft S. 2005. Design of a specific bituminous surfacing for ortho-
tropic steel bridge decks: application to the Millau viaduct, 7th Symposium on Bearing Capacity of
Roads, Railways and Airfields, Trondheim, Norway.
Houel A. 2004. Comportement à la fatigue et fissuration mécanique des enrobés bitumineux sur dalle or-
thotrope, Master in Civil Engineering. Ecole doctorale MEGA-ENTPE, France. [In French]
Houel A. & Arnaud L. 2007. Damage characterization of asphalt concrete specimens by ultrasonic P and
S-wave propagation in laboratory. Advanced Characterization of Pavement and Soil Engineering Ma-
terials. London: Loizos, Scarpas & Al-Qadi eds. Taylor & Francis Group.
Houel A. & Arnaud L. 2008a. A five-point bending test for asphalt cracking on steel plates, Proceedings
of the International RILEM Symposium, Chicago.
Houel A. & Arnaud L. 2008b. The five-point bending test: a way to the dimensioning of the asphalt layer
on steel orthotropic decks, International Orthotropic Bridge Conference, Sacramento.
Houel A. 2008. Endommagement et fissuration mécanique des enrobés bitumineux sur dalle orthotrope.
Ph. D. Ecole doctorale MEGA-ENTPE, Lyon, France. [In French]
Laajili H. 2003. Caractérisation des enrobés bitumineux comme couche de roulement sur tabliers
d’ouvrage d’art métallique, Master in Civil Engineering, Ecole doctorale MEGA-ENTPE, France. [In
French]
N’Guyen T. L. 2008. Revêtements de dalle de pont métallique orthotrope, comparaison d’essais à la fati-
gue entre béton bitumineux et BFUP, Master in Civil Engineering, ENTPE, France. [In French]
Somda A. R. 2007. Revêtement de tablier de pont métallique: caractérisation et dimensionnement lors
d’essai de fatigue, Master in Civil Engineering, ENTPE, France. [In French]
Hauwaert A. V., Thimus J.-F. & Delannay F. 1998. Use of ultrasonics to follow crack growth. Ultrason-
ics 36: 209-217.

318