SHEAR ANCHORING IN CONCRETE CLOSE TO THE EDGE

Norbert Randl* and Marcel John**

*Hilti AG, Kaufering, Germany
**Hilti AG, Schaan, Principality of Liechtenstein

Abstract
Since design rules for post-installed anchors are usually based on investigations in
unreinforced concrete, the design resistance of fixings set close to the concrete edge and
loaded in shear towards the edge is low. It is therefore necessary to consider the effective
strength of the reinforced concrete edge.
This paper first demonstrates that the current methods to take into account the strength
of the concrete edge are insufficient in many cases and proposes different edge
strengthening methods based on extensive laboratory testing. Finally a Eurocode 2
compatible design formula for cast-in hairpin reinforcement is derived.

1. Introduction

Limited space often requires the fixing of anchor

plates like baseplates of steel columns, railings or
lamp posts in the vicinity of the edge of the concrete
foundation.Though in practice the concrete usually
is reinforced, design rules for anchors are typically
based on investigations in unreinforced concrete.
The failure load of such fixings set close to the
concrete edge and loaded in shear towards the edge
is determined by a brittle concrete cone breaking out
in front of the anchor. The shear resistance therefore
reduces significantly with decreasing edge distance.
If the edge distance is very small, splitting due to
wedge forces must also be considered. Adhering to
prescribed minimum edge distances will typically
prevent this mode of failure.
Fig. 1 Anchoring of guard rail

251
Extensive testing has shown that standard slab lateral reinforcement is usually not
sufficient to significantly increase the shear resistance of anchors loaded towards the
edge. Higher shear resistance can be activated, for example, by using elongated holes in
the anchorage base plate for the anchors situated closest to the edge or by externally
reinforcing the edge. If no external reinforcement is to be applied, cast-in additional U-
shaped reinforcement in the concrete has been shown to be a very effective method to
strengthen the edge [1], [2].

For this strengthening method, a design formula based on the Eurocode safety concept as
described in [3] has been derived from test results. This allows the design engineer to
plan post-installed anchorages closer to the concrete edge, and with higher resistance
than would be possible with standard anchor design concepts.

2. Unreinforced Concrete

The analytical determination of the ultimate load capacity is difficult as it depends on the
behaviour of concrete under multiaxial stresses and has to consider the scatter in local
concrete strength, size effects etc. Most equations for the prediction of failure loads have
therefore been derived empirically, taking into account the observations from tests and
are available only for the case without any retaining hanger reinforcement. Based on
regression analyses of 147 tests Eligehausen and Fuchs propose the following equation
for the calculation of the average ultimate failure load in unreinforced uncracked
concrete [4, 5]:
0 ,2
0 ,5  h  (units: [N, mm]) (1)
Vum ,c = k ⋅ d nom ⋅ f cc0 ,5 ⋅  ef  ⋅ c11,5
 d nom 

with:
dnom outside diameter of anchor
hef anchorage depth [5]
c1 edge distance of anchor axis
fcc concrete cube strength
k constant factor (k = 1.0)

3. Effect of Standard Reinforcement

A concrete plate with reinforcement bars parallel to the edge and with ties along the edge
shall be considered as having standard reinforcement. Reinforced concrete is typically
assumed to be cracked concrete. To account for the cracks, the load bearing capacity of
uncracked concrete has to be multiplied by a global factor of 0.7 [5]. In the presence of
minimum edge reinforcement, this reduction can be partially compensated for by a
factor of 1.2 or 1.4, depending on the density of the edge reinforcement [6].
This is only a very rough approximation, and tests have shown that, for smaller anchor

252
diameters, it may even be unsafe. Small diameter anchors have a low bending resistance
and introduce the load right at the top of the
concrete surface. Thus, the failure cone is
pushed over the edge reinforcement (Fig. 2)
[7].
With greater anchor diameters the edge
≈ reinforcement may support the shear load
more effectively because the anchor is able to
introduce the load farther away from the
concrete surface. Figures 3 and 4 show test
results with edge reinforcement of diameters
6 – 12 mm and welded wire meshes.
Especially the test results with M16 anchors
≈ are somewhat higher than those expected in
unreinforced concrete, nevertheless the effect
is rather negligible.
Fig. 2: Small anchors with edge reinforcement

Force Vu [kN]
70

60

50

40
Equ. (1)
30
unreinforced
20 series1
series2
10
Rk (steel failure)
0
0 25 50 75 100 125 150 175 200 Fig. 3 Tests with anchors
c1 [mm] HSL M12
Force Vu [kN]
120

100

80

60
Equ. (1)
unreinforced
40
series1
series2
20
series3
Rk (steel failure)
0
0 25 50 75 100 125 150 175 200 225 250 Fig. 4 Tests with anchors
c1 [mm] HSL M16

253
4. Subsequent Constructive Measures

4.1 External Edge Support

column
anchorage baseplate

HSL HSL
HAS with HIT-HY 150
or HVU

failure cone

Fig. 5 Scheme of back anchorage Fig. 6 Shear support

A steel bar reinforces the concrete edge by anchoring the concrete edge behind the
expected failure cone. The support steel bar is designed according to the rules of steel
construction, assuming a uniformely distributed load. The anchorage is composed of
bonded anchors reaching behind the expected failure cone (Fig. 5). The shear anchors
(HSL) developed a very high resistance and typically failed by yielding of the steel rod
in shear (Fig. 6).

4.2 Use of Elongated Holes in Anchorage Baseplate

If the anchorage baseplate has two or more rows of anchors, the holes for the anchors
closest to the edge should be elongated, directed towards the edge. Thus, these anchors
do not take any shear loads. The anchors of the row farther away can activate a bigger
failure cone and, therefore, a higher load capacity.

5. Cast-In Hairpin Reinforcement

5.1 Research program

A comprehensive test program was carried out at the laboratories of Hilti AG [7] and the
Institute for Concrete Construction at the University of Innsbruck [8] in order to quantify
the effect of cast-in hairpin reinforcement and to develop corresponding design rules.
The U-shaped hairpins were set with an inclination of 5°- 10°.
The following parameters have been varied:
hairpins: - diameter ds [mm]: 12, 16
- diameter of bend resp. distance e of hairpin legs: e = 88 - 134 mm
- concrete cover to front side (csv = 10 - 30 mm) and top surface
(cso = 7 - 33 mm)
anchors: - anchor type: expansion anchors HSL, bonded anchors HVU
- anchor diameter d [mm]: 12, 16, 20

254
 - edge distance c1: 55 - 150 mm
- excentricity of anchor axis (in relation to axis of the hairpin)

The concrete cube strength fcc was between 25 and 30 N/mm² for all tests. The loading
was displacement controlled.
A total of 62 tests with cast-in hairpin reinforcement and 14 tests without any
reinforcement in the anchorage area were carried out.
LINEAR POTENTIO M ETER
The load was introduced by a
steel plate parallel to the
PTFE-LAYER
concrete surface (Fig. 7).
V PTFE layers were put
~10 30 c so between steel plate and
c sv concrete in order to reduce
HAIRPIN friction.
ANCHO R BO LT

Fig. 7 Test setup

5.2 Test results:

Failure usually occurs by formation of a failure cone followed by pullout or breakage of
the anchor. Typically, the maximum load
is reached, when the failure cone starts to
break out. The cracks start from the
anchor and run towards the edge. Their
inclination to the concrete edge is 30° -
45° . Generally, the angle becomes smaller
towards the edge (Fig. 8).

Fig. 8 Concrete cone failure of specimen

12.2 (Dowel HVU 16, hairpin ø16)

Some tests showed a second increase of the load after displacements of 10mm and more:
One reason is the tensile capacity of the deformed, inclinded anchor (kinking effect) and
the second is that anchors set very close to the edge touched the hairpin reinforcement.
Four tests with bonded anchors yielded in shear failure of the anchor without formation
of a concrete cone (steel failure). These tests have not been considered in the
development of the design recommendation for the resistance at the concrete edge.
The tests confirm that cast-in hairpin reinforcement can significantly increase the
loadbearing capacity. Moreover, the post peak load behaviour becomes much more
ductile. The most significant increase of up to 200% is observed if the edge distance is
very small (Fig. 9). Due to the reinforcement, the first cracks appear at about 50% higher
loads than in unreinforced concrete.

255
 Test no. 5.1 V [kN] Test no. 5.2
V [kN] HSL 12, c1 = 65 mm, unreinforced HSL 12, c1 = 65 mm, hairpin ø12, cso=11mm
50
50
45 40,8 kN
45 ∇
40
40
35
35
30
30
25 25
st
20 ∇ 1 crack
20 12,9 kN
15 15
∇
10 10
5 5
0 0
0 2 4 6 8 10 12 14 16 18 20 0 2 4 6 8 10 12 14 16 18 20
s [mm] s [mm]

Fig. 9 Load-displacement curves for specimens without and with hairpin reinforcement

The parameter most strongly influencing the peak load is the distance of the hairpin
reinforcement from the concrete surface (cso), especially with small reinforcement
diameters and with anchors set very close to the edge. While the peak load can be
increased by a factor of 3 with a concrete cover of 10mm for 12mm diameter hairpins,
the influence of the same reinforcement with a cover of 30mm is scarcely observable.
The tests also show that the peak load increases with the diameter of the cast-in
reinforcement. However, this increase is not directly proportional to the increase of the
steel area. The effect of the hairpin reinforcement is independent of the exact position of
the anchor between its legs.

5.3 Theoretical considerations and prediction of failure load

Anchors subjected to shear loads experience bending, shear and with increasing lateral
displacement also axial stresses. Small loads are directly introduced from the shaft into
the surface concrete. The load-displacement curve is steep and shows a linear-elastic
behaviour. The transmission of the shear force from the anchor bolt to the concrete takes
place within a depth measured from the surface of 1 to 2 times the dowel diameter
[9,10]. A hanger reinforcement makes sense only in this area, because the resulting
compression strut then finds a support. Due to the locally high pressure in front of the
bolt, the surface concrete plasticizes under increasing load and flexural stresses are
generated in the anchor shaft. For anchors situated near the edge this finally leads to the
formation of a concrete failure cone.
Some of the tests have also been analysed by means of a 3-dimensional finite element
modelling using non-linear material laws and a smeared crack approach. The computer
simulation confirmed that the failure is due to cracking and crushing of the concrete in
front of the anchor. Provided that the anchor has sufficient embedment length, this leads
to the development of a plastic hinge. The hinge is closer to the surface if the hanger
reinforcement has less concrete cover and thus significantly reduces the lever arm of the
anchor. Moreover the calculations demonstrated that the hairpins remain elastic until the
peak load is reached.
The increase of the ultimate load is not proportional to the steel area because the

256
centerpoint of the support for the compression strut moves down with increasing hairpin
diameter at constant concrete cover.
The results of the tests in unreinforced concrete correspond well to formula (1). The best
agreement is reached with a factor k = 1,0, the coefficient of variation results in a rather
low value of 14 %.
The cast-in reinforcement delays the formation of cracks starting from the anchor as well
as their propagation. The additional energy required for crack growth corresponds to the
possible increase of the load bearing capacity. Therefore, it is best represented by an
additional term to equation (1). The effectiveness of the hanger reinforcement is about
inversely proportional to the concrete cover and decreases with increasing distance
between the anchor and the hairpin reinforcement bend. Moreover, the anchor stiffness,
characterized by dowel diameter and embedment depth, has an influence on the
loadbearing behaviour.
The combination of all relevant parameters with respect to their effects leads to the
following approach for the prediction of the failure load:
k
  2
( )
Vum = Vum ,c + κ ⋅ A s,h ⋅ f y ,h ⋅ 1 − f 1 (l proj , c1 , h ef ) ⋅ 
k ⋅
d nom
c + d
 < Vum ,s (2)
 1 so s 
with:
κ ...... effectiveness factor taking into account that the
concrete will crush before the ultimate capacity c1
of the hairpin is reached csv
k1, 2 ... constants
f1 ....... empirical function
ds ..... nominal diameter of hairpin reinforcement
hef .... anchorage length [5] lproj e
dnom ... outside diameter of anchor
lproj .... projective length: can be approximated in terms
of the edge distance and the concrete cover
ahead of the hairpin (Fig. 10):
lproj ≈ 1,7 ⋅ (c1 - csv) ≤ e
As,h.... total cross section of both hairpin legs Fig. 10 Projective length
fy,h..... yield strength of hairpin steel
Steel failure due to a combination of shear and bending of the anchor shaft represents an
upper bound on the shear capacity: Vum,s = α ⋅ As ⋅ fy with α ≈ 0,6 - 0,7 [5]. (AS = cross
section and fy = yield strength of anchor)
The formation of the concrete cone is considered as the failure criterion. This generally
corresponds to the first peak load or the first horizontal branch of the load displacement
curve. The systematic variation of the different parameters and the subsequent statistical
evaluation finally leads to the following form of equation (2):
 l proj0,5 ⋅ c10,5   d nom 
1,5
Vum = Vum ,c + 0,4 ⋅ A s,h ⋅ f y ,h ⋅ 1 − 0,5 ⋅ ⋅   (3)
 h ef  1,2 ⋅ c so + d s
   

257
Fig. 11 shows the comparison of predicted and observed failure loads for undercut
anchors type HSL in function of the edge distance c1. Considering all specimens with
introduced hairpin reinforcement except for the 4 tests with early shear failure of the
dowels the average ratio of actual to predicted failure load is 1.05. With a coefficient of
variation of 19 % and a factor of correlation of 89%, the scatter zone is within acceptable
limits (Fig. 12).

HSL ø16
(Hairpin reinforcement ø12 and ø16)
V [kN]
120
Eq. (1) (unreinforced)
110
100 Eq. (3), hairpin ø16,
90 cso ≈ 9 mm
80 Eq. (3), hairpin ø12,
70 cso ≈ 11 mm
60 Vks (steel failure)
50
HSL 16, hairpin ø16,
40 cso ≈ 9 mm
30 HSL 16, hairpin ø12,
20 cso ≈ 11 mm
10 unreinforced
0
50 70 90 110 130 150 170 190 210
Edge distance [mm]

Fig. 11 Predicted (calculated) and observed failure loads in function of c1

Vexp [kN]
120,0

100,0

80,0

60,0

40,0

20,0 Vcalc =Vexp

0,0 Fig. 12 Comparison of calculated and

0,0 20,0 40,0 60,0 80,0 100,0 120,0 experimental results
Vcalc [kN]

258
5.4 Design formula and recommendations for execution
For practical design purposes the transition from mean ultimate loads to characteristic
values (fractiles) is required. The characteristic resistance is derived as the 5%-fractile of
the mean value of strength with a confidence level of 90%. Admitting a log-normal
distribution, the statistical evaluation of the tests conducted in reinforced concrete yields
a global reduction factor ψ = 0,72. This corresponds to the factor of ψ = 0,7 proposed
by Eligehausen in [4] for unreinforced concrete.
The material properties are also considered by their lower fractile; this additional safety
will not be used in the design formula.
The design resistance is derived in accordance with the safety concept of MC 90 [3] and
Eurocode 2 by applying appropriate partial safety factors.
In unreinforced, cracked concrete, the design resistance is the mean ultimate load
multiplied by ψ = 0,7 for the statistical evaluation of the 5%-fractile [4], multiplied
again by 0,7 accounting for the effect of cracks [4, 5] and divided by the partial safety
factor γMc [6]:
0 ,2
1 0 ,5 0 ,5  h ef  (units: [N, mm]) (4)
VRd ,c = ⋅ 0,5 ⋅ d nom ⋅ f ck ⋅   ⋅ c11,5
γ Mc  d nom 
with:
fck ...... characteristic cylinder compressive strength of concrete
γMc .... safety factor for system with normal installation safety:
γMc = γ1⋅ γ2 = 1,8⋅1,2 = 2,2 [6].

According to equation (3) the shear load capacity with cast-in hairpin reinforcement
becomes:
1  l proj 0,5 ⋅ c10,5   d nom 
1,5
VRd = VRd ,c + ⋅ 0,3 ⋅ A s,h ⋅ f yk ,h ⋅ 1 − 0,5 ⋅ ⋅   (5)
γ Ms  h ef   1,2 ⋅ c so + d s 
 
(valid for fyk,h ≤ 600 N/mm², fck ≤ 40 N/mm², 4 dnom ≤ hef ≤ 8 dnom, dnom ≤ 25 mm)
with:
γMs = γ1⋅ γ2⋅ γ3 = 1,15 ⋅ 1,5 ⋅ 1,2 = 2,1
γ1 .... partial safety factor for steel in tension
γ2 .... partial safety factor taking into account deviations in the height position of the
reinforcement and uncertainties due to anchor installation
γ3 .... partial safety factor accounting for scatter of failure loads and model uncertainties

Eq. (5) is valid for single anchors and sufficient thickness of the concrete member
(h > 1,5 c1) as well as an edge distance c2 perpendicular to the direction of the shear load
measured from the axis of the anchor c2 > 1,5 c1.

For execution special attention should be paid to the following aspects:

The hairpin reinforcement should be anchored outside the assumed failure cone and
consist of ribbed reinforcing bars with a diameter not larger than 16 mm. It should be

259
inclined with respect to the concrete cover. Thus the concrete cover of the hairpin legs is
bigger than that of the bend, which improves the anchorage. Corrosion generally should
not be relevant, since the baseplate of the fixed part is typically grouted over the bend of
the hairpin reinforcement. If not, the reinforcement must be protected by galvanization,
special coatings or the use of stainless steel. Sufficient bending radius will compensate
for the positioning tolerances of the hairpin reinforcement and thus make sure that the
anchors will be positioned within the hairpin.

6. Summary

After reviewing some alternative methods to most effectively utilize the concrete edge
strength, the reader has been presented with a formula to design anchors under shear
loads when hairpin reinforcement is present in the concrete. This formula is compatible
with modern design concepts and has been adapted to Eurocode 2. Moreover it has been
shown, that global design concepts do not always clearly represent the complex situation
near a concrete edge. Therefore it is strongly recommended to use engineering
judgement and well based design concepts in the planning of safety relevant fixings near
a concrete edge.

7. References

1. Paschen, H., and Schönhoff, Th., ‘Untersuchungen über in Beton eingelassene

Scherbolzen aus Betonstahl’, Deutscher Ausschuß für Stahlbeton 346 (Wilhelm
Ernst & Sohn, Berlin, 1983) 105-147.
2. Klingner, R., Mendonca, J., and Malik, J., ‘Effect of reinforcing details on the shear
resistance of anchor bolts under reversed cyclic loading’, Journal of the American
Concrete Institute 79 (1) (1982) 3-12.
3. Comité Euro-International du Béton, CEB-FIP Model Code 1990 (Thomas Telford,
London, 1993) 437 pp.
4. Eligehausen, R., Mallée, R., Rehm, G., ‘Befestigungstechnik’, Betonkalender 1997,
part 2, 609-753.
5. Comité Euro-International du Béton, ‘Fastenings to concrete and masonry
structures’, Bulletín 216 (Thomas Telford, London, 1994) 249 pp.
6. Comité Euro-International du Béton, ‘Design of fastenings in concrete’, Bulletín
233 (Thomas Telford, London, 1997) 83 pp.
7. Hartmann, M., Silva, J., ‘Versuche zur Erhöhung der Tragfähigkeit von randnahen
Ankern’ (HILTI AG Konzernforschung, Report No. A-IF6-8/97, 1999).
8. Fritsche, G., Wicke, M., ‘Versuche zur Prüfung von Metalldübeln’, Institut für
Betonbau, Universität Innsbruck (Report No. 22, 1998).
9. Randl, N., Wicke, M., Schubübertragung zwischen Alt- und Neubeton, Beton- und
Stahlbetonbau 95 (8) (2000) 461-473.
10. Randl, N., Untersuchungen zur Kraftübertragung zwischen Alt- und Neubeton bei
unterschiedlichen Fugenrauhigkeiten, Doctoral thesis (Universität Innsbruck, 1997)
369 pp.

260