Articles

Thomas Lechner* DOI: 10.1002/stco.201610012

Sebastian Gehrlein
Oliver Fischer

Structural behaviour of composite dowels

in thin UHPC elements
Composite dowels with different shapes have been developed This article describes the experimental investigations
and used in composite members during the last two decades. At with push-out tests from which the basic characteristics of
the Chair of Concrete and Masonry Structures at Technical Uni- composite dowels in thin UHPC elements were acquired
versity of Munich (TUM), composite dowels with a clothoid shape as well as the preliminary results and the test setup for the
are used for filigree composite beams and columns. In both types experimental research into UHPC composite beams and
of application the composite dowels are used as external rein- columns. The research results for thin UHPC wall panels
forcement and Ultra-High Performance Concrete (UHPC) is cho- are given in [4].
sen instead of normal-strength concrete. This article describes
mainly the results of the preliminary push-out tests that were car- 2  Ultra-High Performance Concrete
ried out in order to determine the structural behaviour of compos-
ite dowels in thin UHPC elements and verify the influence of the Ultra-High Performance Concrete is an extremely dense
UHPC web thickness, the steel thickness of the shear connector, concrete with a compressive strength > 150 MPa. The com-
the concrete compressive strength and the influence of rein-
position of UHPC differs considerably from that of nor-
forcement on the loadbearing capacity and failure modes. The
mal- or high-strength concrete since the matrix porosity is
paper also includes an overview of and the outlook for the experi-
minimized with the help of a very low w/c ratio and a very
ments with composite beams and composite columns.
dense packing of the fine particles due to the use of differ-
ent mineral additions such as silica fume and quartz. The
1 Introduction B5Q mix as shown in Table 1 with a maximum grain size
of 8 mm and a regular fibre content of 2.5 % by vol. was
Due to the extremely dense microstructure of Ultra-High used for this research. It was developed during priority
Performance Concrete (UHPC), the durability and the programme 1182 of the German Research Foundation [5].
compressive strength are significantly better than those of The steel fibres chosen had a length h = 9.0 mm and a di-
normal- or high-strength concrete. These characteristics ameter d = 0.15  mm, which gives an h/d ratio of 60. A
are the main reason why UHPC is the perfect material for certain amount of steel fibres assures a ductile behaviour
filigree elements with small concrete cover and hence less and enhances the tensile as well as the bending strength of
weight compared with standard concrete elements. UHPC. The water/cement ratio was 0.24.
In composite structures the steel carries the tensile
forces and the concrete is mainly arranged in the compres- 3  Composite dowels
sion zone. This kind of task sharing leads to structural ele-
ments that are typically made of normal- or high-strength Composite dowels have been used in normal-strength con-
concrete. If UHPC is used for such structures, a significant crete since the development of the perfobond rib in 1985 [6].
reduction in the overall dimensions and the dead weight is
possible. For this reason, the behaviour of composite dow-
els in filigree UHPC members (beams, columns and wall Table 1. Composition of UHPC
panels) was investigated at the Chair of Concrete and Ma- Constituent Amount in kg/m³
sonry Structures at TUM. The clothoid shape (CL-shape)
CEM I 52.5 R HS-NA 650
according to the German technical approval [1] was cho-
Silica Fume 177
sen from the existing different shapes of composite dowels.
Clothoid strips (CL-strips) offer a high shear strength and Quartz I 325
a very ductile load-deformation behaviour [2], which are Quartz II 131
necessary to take full advantage of the high concrete Sand 0.125/0.5 354
strength [3].
Basalt 2/8 597
Superplasticizer 30.4

* Corresponding author: Steel fibres 2.5 % by vol.

thomas.lechner@tum.de Water 158

132 © Ernst & Sohn Verlag für Architektur und technische Wissenschaften GmbH & Co. KG, Berlin · Steel Construction 9 (2016), No. 2
 T. Lechner/S. Gehrlein/O. Fischer · Structural behaviour of composite dowels in thin UHPC elements

Fig. 1.  Use of composite dowels in

plates and slender beams and the
principal failure mechanisms:
concrete splitting and steel failure

During the last two decades, research mainly focused on

the optimization of the geometry of this shear connector in
order to achieve a higher shear capacity [7] and a better
behaviour under cyclic loading. Experimental investiga-
tions with composite dowels demonstrated excellent
load-deformation behaviour when tested with UHPC
beams or plates [8]. The loadbearing capacity and also the
failure mechanism of composite dowels that are positioned
parallel to the concrete surface (e.g. in concrete beams)
with only a few centimetres of concrete cover differ from
the usual use in plates (see Fig. 1). Whereas the concrete
failure in plates is mainly governed by concrete pry-out, the
failure in thin elements is mostly caused by splitting failure
due to the small concrete cover and splitting tensile forces
resulting from the concentrated load introduction of the
steel dowel. Furthermore, concrete shearing might occur in
the case of slender beams.

4  Experimental investigations with push-out tests

The experimental investigations with push-out tests were Fig. 2.  Geometry and dimensions of push-out specimen
designed to check the ductility behaviour and failure loads
for small CL-strips in thin UHPC elements with and with- specimens were equipped with stirrups (∅ 6 mm between
out conventional bar reinforcement. Another main objec- each steel dowel) and longitudinal bars (∅ 6 mm), while the
tive of this research was to determine the steel and con- others were plain fibre-reinforced concrete without conven-
crete failure mechanisms that can occur when filigree tional bar reinforcement. The detailed parameters investi-
UHPC members are used. gated and the results obtained within this series are summa-
rized in Table 2. Two push-out specimens were produced
4.1  Test setup and tested for each setup. The results in Table 2 are mean
values for the load and minimum values for the character-
Due to the special shear behaviour of CL-strips in thin istic ductility duk, which was evaluated according to EC 4
UHPC elements, the first push-out tests were designed to [9].
analyse the factors influencing concrete failure and the In all tests a brittle failure occurred after the splitting
second test series focused on steel failure. The geometry tensile strength was reached. Only for POCL 1 (with rein-
and dimensions of the push-out specimens are shown in forcement) was a satisfactory ductility reached. It can be
Fig. 2. The composite dowels have a height of 40 mm and seen from Fig. 3 that the thinner the UHPC panel and the
a longitudinal distance ex = 100  mm between the steel higher the shear capacity of the steel, the more brittle was
dowels. The compressive strength of the UHPC was be- the load-deformation behaviour. Consequently, specimens
tween 130 and 198 MPa for a 150 mm high cube. Defor- with a steel thickness of 10 mm are more brittle than those
mation-controlled loading allowed the measurement of the with only 5 mm CL-strips (Table 2). But, in addition, they
post-failure behaviour even when a brittle failure occurred. reach a significantly higher maximum load.
The slip d between the shear connectors and the UHPC In order to achieve a ductile steel failure, the shear
members was measured with displacement transducers. strength of the CL-strip was reduced by using thin steel
with a low yield strength. Table 3 shows the test pro-
4.2 Results gramme and the results for specimens where steel failure
occurred. The use of conventional bar reinforcement and
The influence of the concrete compressive strength fc,cube,150, the reduction in concrete thickness had only a minor influ-
the yield strength of the CL-strip fy,CL, the thickness of the ence on the loadbearing capacity and all these specimens
steel tCL and the UHPC panel tUHPC as well as the fibre ra- showed a pronounced ductility (see Table 3 and Fig. 4)
tio were studied in the first test series. Furthermore, some with a characteristic ductility duk > 6 mm.

Steel Construction 9 (2016), No. 2 133

T. Lechner/S. Gehrlein/O. Fischer · Structural behaviour of composite dowels in thin UHPC elements

Table 2.  Test programme and results for push-out specimens with concrete failure
Test Parameter Results
fc,cube,150 fy,CL tCL tUHPC Fibre ratio Stirrups/ max. load duk
No.
[MPa] [MPa] [mm] [mm] [% p.V.] bars [kN] [mm]
POCL-1 136 438 5 60 2.5 w 645 6.1
POCL-2 144 438 5 60 2.5 w/o 578 1.7
POCL-3 160 438 5 50 2.5 w/o 590 1.1
POCL-4 159 438 5 40 2.5 w/o 577 1.3
POCL-5 137 384 10 60 2.5 w 1013 2.5
POCL-6 143 384 10 60 2.5 w/o 901 0.8
POCL-7 148 384 10 50 2.5 w/o 878 1.0
POCL-8 138 384 10 40 2.5 w/o 798 0.6
POCL-9 146 384 10 60 0.9 w/o 891 0.6

Table 3.  Test programme and results for push-out specimens with steel failure
Test Parameter Results
fc,cube,150 fy,CL tCL tUHPC Fibre ratio Stirrups/ max. load duk
No.
[MPa] [MPa] [mm] [mm] [% p.V.] bars [kN] [mm]
POCL-24 164 176 3 60 2.5 w/o 331 21.5
POCL-26 164 176 3 40 2.5 w/o 294 14.3
POCL-32 177 280 5 60 2.5 w 554 17.6
POCL-33 183 280 5 60 2.5 w/o 532 10.5
POCL-35 183 280 5 40 2.5 w/o 498 7.2
POCL-40 198 387 3 40 2.5 w/o 365 8.6
POCL-41 198 387 3 60 2.5 w/o 391 15.4

Fig. 3.  Load-slip diagram for push-out tests with concrete Fig. 4.  Load-slip diagram for push-out tests with steel fail-
failure and 10 mm thick steel dowels ure and 3 or 5 mm thick steel dowels

4.3 Evaluation where factor bel = 0.25, which was derived from FE analy-
ses and experimental testing of CL strips in normal-
The results of the push-out tests were evaluated by comparing strength  concrete. The distance between two steel teeth
the experimental results Pexp with the theoretical, character- ex = 100 mm and tCL and fy,CL are given in Table 3.
istic values for steel and concrete failure. For steel failure, the The Pexp/Ppl,k ratio is plotted against the correspond-
German technical approval [1] gives the following formula: ing shear strength fy,CL × tCL in Fig. 5 (left). For low values
of shear strength especially, the experimental results are
Ppl,k = βel ⋅ e x ⋅ t CL ⋅ fy,CL (1) much higher than the theoretical values and even push-out

134 Steel Construction 9 (2016), No. 2

 T. Lechner/S. Gehrlein/O. Fischer · Structural behaviour of composite dowels in thin UHPC elements

specimens with pure concrete failure reached higher loads mation behaviour for beams with complete and partial
than expected by applying Eq. (1). It seems obvious that shear connections. Furthermore, the maximum shear and
Eq. (1) is conservative but not very economical for thin bending capacity of UHPC composite beams with external
composite dowels in UHPC elements. Obviously, factor bel reinforcement were evaluated with these four-point bend-
has to be chosen bigger than 0.25 in the case of a strong ing tests. The following section gives a brief overview over
concrete where no pulverization occurs under pressure in these experiments, whereas the detailed test setup, test pro-
the contact area. gramme and results are described in [11].
No design formulas are given for splitting failure in [1]
because splitting is prevented by transverse reinforcement 5.1  Test setup, test programme and results
and a minimum concrete thickness. In [10] the splitting
failure is described by the following equation: The test programme consisted of 12 T-beams with compos-
ite dowels as external reinforcement. All beams had a span
1 0.4A d + 0.6e x ⋅ co of 3.0 m and were tested under four-point bending. The
Pc,splitting = ⋅ ⋅ fct,UHPC (2) beam height was 48 cm for two beams and 32 cm for the
0.3  t CL 
1 −  remaining 10 beams. A typical cross-section, the dimen-
 t 
 UHPC  sions and the test setup are shown in Figs. 6 and 7. It
should be mentioned that the UHPC web thickness was
where Ad is the area between two steel teeth (~2310 mm²) chosen to be 60 mm for all but one beam, where a web
and co the concrete cover above the CL-strip. Since the thickness of only 40  mm (in combination with a 3  mm
cracks in the concrete appeared about 20  mm above steel dowel) was used.
the composite dowels, this value was chosen for co. The The most important parameters varied during the
concrete tensile strength fct was evaluated in direct tensile tests were CL-strip thickness and steel grade. The maxi-
tests to be lower than 8 MPa. The results of the push-out mum shear capacity as well as the ductility of the com-
specimen Pexp are compared with the theoretical values posite dowels is dependent on these parameters. There-
Pc,splitting in Fig. 5 (right). POCL-5/6/7/8/9 (tCL = 10 mm) fore, they have a strong influence on the value of partial
in particular show resistances to splitting that are signifi- shear, which in turn influences the overall behaviour of
cantly higher than predicted values, whereas the results the composite beams. Whereas beams with a full shear
from POCL-1/2/3/4 (tCL = 5 mm) are close to the theoret- interaction fail by a cross-sectional failure such as bend-
ical values. ing or shear failure of the UHPC web, composite beams
Hence, it can be stated that improvements to both with a partial shear interaction fail primarily by shear
formulas are necessary for this special type of application failure of the composite dowels.
so that design is not only conservative but also economi- Another parameter that had an influence on the fail-
cal. These formulas will be derived after some additional ure mechanism (shear or bending) was the conventional
push-out tests are carried out. Meanwhile, the experimen- reinforcement (longitudinal bars and stirrups, ∅6/10). A
tal research on beams and columns was planned with the brittle shear failure occurred for a beam without reinforce-
results from the push-out tests. ment and complete shear connection. The same beam with
reinforcement failed in a very ductile way by yielding of
5  Composite beams the external reinforcement.
All beams were equipped with displacement transduc-
After the shear capacity of composite dowels was deter- ers to measure the relative displacement between the
mined with the help of push-out tests, beam experiments UHPC and the external reinforcement. Strain gauges were
were designed and performed to evaluate the behaviour of used at the steel dowels to evaluate the load distribution
composite dowels as external reinforcement for beams. Ex- over the length of the beam and at cross-sections B and M
perimental investigations on UHPC composite beams to measure the strain distribution at the designated
were necessary to obtain information about the load-defor- cross-sections.

Fig. 5.  Comparison between ex-

perimental and theoretical loads
for steel failure (left) and concrete
splitting failure (right)

Steel Construction 9 (2016), No. 2 135

T. Lechner/S. Gehrlein/O. Fischer · Structural behaviour of composite dowels in thin UHPC elements

umn with conventional reinforcement will be tested. The

height of the columns is between 2.5 and 3.0 m. All speci-
mens are tested with a two-sided simple support and an
eccentricity between 1.0 and 3.0 cm. The test setup and the
cross-section for most of the specimens are displayed in
Fig. 8. Whereas the CL-strip thickness will be varied be-
tween 3.0 and 5.0  mm within the test programme, the
outer dimension of the specimens (h = 145 mm) will stay
the same throughout the whole test programme.
Besides the parameters CL-strip thickness, eccentric-
ity and column height already mentioned, the influence of
the steel grade and the steel fibre content within the UHPC
will be tested experimentally. Furthermore, the number of
steel dowels, and therefore the stiffness of the connection
between steel and UHPC, will be varied within the test pro-
Fig. 6.  Cross-section of a UHPC T-beam with external rein- gramme. According to this test programme, the experimen-
forcement and a height of 320 mm tal investigations should result in, on the hand, conclusions
regarding the maximum loadbearing capacity of UHPC
The results proved that composite dowels can be used composite columns. On the other hand, it should be possi-
in UHPC beams as external reinforcement with full and ble to determine the minimum quantity of steel dowels and
partial shear connection even if composite dowels with a minimum CL-strip thickness needed in order to ensure full
ductility lower than 6 mm are used. However, especially activation of the steel flange as external reinforcement.
for composite beams without conventional reinforcement, To measure the slip between the steel flange and the
the shear capacity of the UHPC can limit the loadbearing UHPC, all composite columns tested were equipped with
capacity since a brittle failure of the UHPC web might displacement transducers. In addition, the activation of the
occur. steel flange as well as the stress in the steel dowels were
measured with strain gauges.
6  Composite columns First results documented a high loadbearing capacity of
the UHPC composite columns. The maximum load for an
Another possible field of application for composite dowels eccentricity of 1.0 cm and with a column height of 3.0 m was
in UHPC is their use in composite columns. Similarly to 1.6 MN. At this load, a very ductile buckling failure mode
the composite beams, the composite dowels and the steel occurred. Furthermore, the results of the measurements sug-
flange can be considered as external reinforcement. In or- gest that the use of additional stirrups should have no signif-
der to investigate the loadbearing behaviour of composite icant influence on the maximum loadbearing capacity.
columns as well as the degree up to which the composite
dowels can be used as external reinforcement, experimen- 7  Summary and outlook
tal tests were indispensable. A brief summary of the exper-
imental investigations regarding composite columns, Composite dowels can be used in filigree precast UHPC
which are currently planned and being carried out at the elements (beams and columns) as external reinforcement.
Chair of Concrete and Masonry Structures at TUM will be Push-out tests with CL-strips and thin UHPC members were
given in the next section. conducted and it was found that composite dowels can
carry very high loads even in slender elements only 40 mm
6.1  Test setup, test programme and first results thick. However, it has to be considered that a rather brittle
concrete failure can occur if the composite dowel thickness
The planned and already partially carried out test pro- is large compared with the UHPC thickness. For specimens
gramme includes 10 octagonal composite columns that with steel failure, lower maximum shear forces were ob-
were made of UHPC (according to mix B5Q) and four served than for specimens with concrete failure, but in re-
CL-strips each. In addition, one composite column made turn, the ductility of this small-scale shear connector im-
of high-performance concrete (HPC) and one UHPC col- proved clearly. Further investigations will focus on the de-

Fig. 7.  Test setup and dimensions of UHPC T-beam with a height of 320 mm

136 Steel Construction 9 (2016), No. 2

 T. Lechner/S. Gehrlein/O. Fischer · Structural behaviour of composite dowels in thin UHPC elements

[3]  Döinghaus, P.: Zum Zusammenwirken hochfester Baustoffe

in Verbundträgern. Dissertation, Faculty of Civil Engineering,
RWTH Aachen, 2001.
[4]  Lechner, T.; Fischer, O.: Stabilitätsverhalten von schlanken
Wandscheiben aus Normal- und Ultrahochleistungsbeton. Be-
ton- und Stahlbetonbau 110 (2015), No. 5, pp. 328–339. DOI:
10.1002/best.201500011.
[5]  Deutscher Ausschuss für Stahlbeton: Sachstandsbericht ul-
trahochfester Beton. Schriftenreihe des DAfStb, No. 561:
Beuth Verlag, Berlin, 2008.
[6] Andrä, H.-P.: Neuartige Verbundmittel für den Anschluss
von Ortbetonplatten an Stahlträger. Beton- und Stahlbeton-
bau 80 (1985), No. 12, pp. 325–328.
[7]  Seidl, G.: Behaviour and load bearing capacity of composite
dowels in steel-concrete composite girders. Dissertation, Wro-
claw University of Technology, 2009.
[8]  Gallwoszus, J.; Hegger, J.; Heinemeyer, S.: Cyclic Behaviour
of Puzzle Strips in UHPC. Ultra-High Performance Concrete
and Nanotechnology in Construction: Proc. of HiPerMat,
2012.
Fig. 8.  Test setup and cross-section of composite columns [9]  DIN EN 1994-1-1: 2010-12: Design of composite steel and
made of UHPC (thickness of composite dowel tCL = 10 mm) concrete structures – Part 1-1: General rules and rules for
buildings, Beuth Verlag, Berlin, 2010.
[10]  Heinemeyer S.: Zum Trag- und Verformungsverhalten von
velopment of design formulas for steel and splitting failure Verbundträgern aus ultrahochfestem Beton mit Verbundle-
in thin UHPC elements. Especially interesting is the high isten. Dissertation, Faculty of Civil Engineering, RWTH
loadbearing capacity of composite dowels in UHPC com- Aachen, 2011.
pared with high-strength concrete for all specimens. [11]  Lechner, T.; Gehrlein, S.; Fischer, O.: Modulares Bauen mit
Verbundträgern aus ultrahochfestem Beton. Tagungsband
The research concerning the behaviour of UHPC
zum 19. Münchener Massivbau Seminar, 2015.
composite beams with external reinforcement and com-
[12]  Haensel, J.; Kina, J.; Schaumann, P.: Zur Erweiterung des
posite dowels under four-point bending has proved that Anwendungsbereiches von Stahlträgerverbundkonstruk-
UHPC composite beams can be designed according to the tionen. Stahlbau 63 (1994), No. 4, pp. 279–283.
results from the push-out tests and, subsequently, also ac-
cording to the corresponding design formulas. In addition, Keywords: composite dowels; composite beams; composite col-
UHPC composite beams with full and partial shear con- umns; UHPC; push-out tests; experimental testing
nection possess a high moment-carrying capacity and a
pronounced ductility.
The first experimental investigations on UHPC compos-
ite columns showed a high loadbearing capacity of the test
specimens and that a ductile failure mode could be achieved.
Further experiments will be carried out to confirm these re- Authors
Dipl.-Ing. Thomas Lechner
sults and quantify the influence of different parameters.
Sebastian Felix Gehrlein, M.Sc.
sebastian.gehrlein@tum.de
References Prof. Dr.-Ing. Dipl.-Wirt. Ing. Oliver Fischer
oliver.fischer@tum.de
[1]  Deutsches Institut für Bautechnik: Allgemeine Bauaufsicht-
liche Zulassung Z-26.4-56: Verbunddübelleisten, May 2013. All authors:
[2] Berthellemy, J.; Lorenc, W.; Mensinger, M.; Rauscher, S.; Technische Universität München
Seidl, G.: Zum Tragverhalten von Verbunddübeln – Teil 1: Lehrstuhl für Massivbau – MPA Bau
Tragverhalten unter statischer Belastung. Stahlbau 80 (2011), Theresienstr. 90
No. 3, pp. 172–184. DOI: 10.1002/stab.201101408. 80333 München

Steel Construction 9 (2016), No. 2 137