University of Colorado

Department of Civil Engineering

CVEN 6831

Behavior and Design of Steel Anchor Systems

in Cementitious Concrete

Scott Hamel

Instructor:
Professor Victor E. Soauma

December 13, 2004

1.0 Scope
There are many facets to the concept of connecting structural elements to
concrete with steel anchors. The designer must consider anchor type, applied forces,
condition of the concrete, and proximity to other anchors.
There are four major types of anchors, as seen in Figure 1, including Headed (a),
Keying (b), Bonding (c), and Friction (d). Keying and headed anchors undercut the
concrete to provide mechanical resistance, bonding anchors use a chemical adhesive to
provide resistance, and friction anchors generally expand near the tip to provide
frictional resistance. Headed anchors are cast in situ, while the other three types are
applied to hardened concrete. The force being applied to the fasteners are characterized
as tension, shear, or both. And the concrete may be cracked or uncracked at the
fastener location.
Meanwhile, the researcher (and code-writer) must consider the failure mode for
each of the above cases, and its associated mechanics and limitations.
Each of the above anchor types, in tension and shear, and in cracked and
uncracked concrete, are addressed in the ACI Code, and have been examined using
fracture mechanics based 3-D finite element analysis [1,2,3]. In an attempt to limit the
complexity of the subject, this paper will focus primarily on headed anchor studs, both

Figure 1: Types of Anchorages

single and in groups, subjected to tension only in uncracked concrete. This, and the
corresponding shear cases, are the predominant conditions in new construction, where
large numbers of connections may be required, requiring economically optimized (i.e.
not over-conservative) connections.

2.0 History

The current accepted code, referenced by the 2003 International Building Code
(IBC), for the design of headed anchor studs is Appendix D of ACI 318-02. This
document describes the design procedure, which is covered in section 3.0 of this paper
and references materials that support the method.
There was a flurry of activity in the mid 90’s, which attempted to (and eventually
succeded in) replacing the then current ACI 349 code (1985), which was based on a
method developed in the mid 70’s. Eligehausen, who led the charge, conducted most of
the experimental research in Germany, and based the new method, called the Concrete
Capacity Design (CCD) [4], on the European version (kappa method). One of the few
very vocal dissenting voices was Cannon [5], a practicing engineer in Tennessee.
The CCD approach was primarily based on extensive testing, unavailable when
the previous code was written [6], with consideration given to the principles of fracture
mechanics.
The interesting aspect of the CCD method is that its development has been
somewhat backwards of the normal code development. Instead of being an analytic
theory verified by increasingly complex and encompassing experimental results (like
plastic design for example), it has experienced the opposite. As finite-element
simulations of materials subjected to elasto-plastic fracture mechanics have improved
(along with a plunge in the cost of computing power), the CCD method has been verified
by multiple researchers, including Fox [7] at the University of Colorado. The exception,
discussed in Section 4.0, is for cases difficult to test, such as deep anchors in large
masses of concrete or anchors with relatively large heads.

3.0 ACI Design Method (CCD)

3.1 Single Anchor

The ACI Code [8] specifies that the nominal breakout strength of an anchor is:
AN
Ncb = ⋅ Ψ2 Ψ3 N b (3.1)
ANO
where:
AN = Projected area of the failure surface
ANO= Projected area of a the failure surface of a single anchor remote from edges
Y2= Modification for edge effects
Y3= Modification for Cracked Concrete
Nb= basic concrete breakout strength for a single anchor in cracked concrete:

1.5
N b = 24 f ' c hef (3.2)

where:

hef = effective anchor embedment depth (in)

f’c = Concrete compressive strength (psi)

At first glance, this equation does not seem to be based on anything except
empirical data. However, a review of the Code background paper [4] reveals differently.
That paper defines the breakout strength as:
1.5
N b = knc f ' c hef (3.3)

where

knc = k1 ⋅ k2 ⋅ k3 (3.4)

2 1
N b = k1 f ' c ⋅ k2 hef ⋅ k3 (3.5)
hef

The first coefficient and term ( k1 f ' c ) represents the nominal tensile strength
2
of the concrete. The second coefficient and term ( k2 hef ) is the failure area, shaped

1
like a pyramid, and discussed later (figure 3). And the last term ( k3 ) incorporates
hef
size effect.
The authors further explain that since the strain gradient for concrete in
fastenings is very large, the size effect is maximized, and thus essentially behaves in a
Linear Elastic Fracture Mechanics (LEFM) fashion. This in turn means that the stress
(and the final force) is proportional to the square root of the size, here represented by
the effective depth. Thus, very large embedded anchors will have a much smaller
tensile stress.
Fuchs et al. go on to compare the CCD method with, ACI 349 and a large array
of test data. The older code uses a similar method of assuming a breakout shape (in
this case, a cone) and multiplying the surface area by a tensile stress capacity. This
curve was fitted to available data back in the 1970’s such that it formed a lower bound
of the test data. However, there was no fracture mechanics component to account for
size effect. The results of this can be seen in Figure 2. As the effective depth becomes
larger, ACI 349 becomes unconservative.

Figure 2 : Concrete Breakout Load for Cast-In Headed Studs.4

The authors of the Code background paper found that knc = 40 provided good
agreement with the data. It seems that the ACI committee felt this to be too risky, and
they lowered the number to knc =30 for headed anchor studs in uncracked concrete,
along with a strength reduction factor, ∅, generally around 0.7, depending on the

anchor type and load condition.

Once Nb has been found, the concrete breakout strength can be computed using
equation 3.1. The first term of this equation (AN/ANO) infers that the strength of a
fastener varies linearly with the ratio of its projected area to the area of a remote
fastener (unaffected by edges or other anchors). The area of a remote fastener is
defined as:
2
ANO = 9 hef (3.6)

which comes from Figure 3. The angle

of 35 degrees, which is the most
important factor in determining ANO is
“understood”, and was not proven by
any analytical means. This is discussed
further in Section 4.0. This linear
relationship has generally been found to
be true, with the exception of edge
effects and the influence of cracks, for
which the modification factors (Y) are

less than 1.0.

3.2 Multiple Anchors

The equation for the nominal

concrete breakout strength of multiple
anchors in a group is almost the same
as that for a single anchor (Eq 3.1), with
the addition of a modification factor for
eccentrically loaded groups (Y1):

Figure 3 : Idealized Concrete

Cone under tensile loading

AN
Ncb = ⋅ Ψ1Ψ2 Ψ3 N b (3.7)
ANO
The surprising aspect about this equation is that there is no modification factor
for closely spaced anchors. The reduced strength of a tightly spaced group of anchors
is entirely captured in the linear relationship between AN and ANO. Figure 4 shows an
example of how AN is calculated.
 Figure 4: Nominal Area (AN) of a Group of Anchors

While the fracture mechanics based approach yielded somewhat better results for case
of a single anchor with large effective depths, as discussed above, the CCD method was
found to predict tensile capacities far better than ACI 349 for anchor groups. This can
be seen in Figure 5.

Figure 5 : Actual and Predicted Loads for Cast-In Headed Studs as a function of
distance between outermost anchors
4.0 FEM Simulations
There have been several independent researchers that have undertaken the 3-D
finite element analysis of a single anchor [3,9,10], including the author of the original
CCD proposal [1,11]. Much of this work was done with the nuclear industry in mind,
which uses these types of connections in critical spots more than other industries.
Early attempts at modeling concrete breakout with fracture mechanics contained
rigorous analysis with the attempt of finding a simplified design method. On example is
presented below.
More recent research has focused on investigating the myriad of geometric and
material conditions, the physical testing of which would be too costly and time
consuming. This is done with the purpose expanding and revising the current code
(CCD method). One such example is presented in the following sections.

4.1 Two-Domain Boundary Element Pull-out Test

Chahrour and Ohtsu [9] give an extremely

in depth description of their analysis of different
geometries using the Two-Domain Boundary
Element. A Two-domain boundary element is a
formulation of two elastic domains with a
boundary around each, and shared in one area
(Figure 6). The interface boundary maintains the
compatibility condition and thus the traction and
displacement can be related:

u1j (Q ) = u 2j (Q ) (4.1)
Figure 6: Two-Domain
t1j (Q ) = −t 2j (Q ) (4.2) Boundary Element

These can be discretized along the boundary, and then numerically integrated over 6
gauss points. The relationship (now in matrix form) can then be converted into a
standard finite element formulation to find the nodal forces and displacements.
Once the analysis is complete, the stress intensity factors, KI and KII at the crack
tip are computed from the relative displacements of A and B (Smith’s one point
formulae), as seen in Figure 7a. And then the direction of the crack propagation is
determined from the maximum circumferential stress (Figure 7b). The continuum is
then re-meshed and the process continued.
 Figure 7: (a) Crack Tip Elements (b) Crack Growth Direction

This method was used to analyze specimens such as the one shown in Figure 8.
Using the results of the simulations, Chahrour and Ohtsu proposed the following
formula for the capacity of anchor bolts:

γ
⎛a⎞
Fu = q ⋅ b(EGF ) d ⎜ ⎟
1/ 2 1.5
(4.3)
⎝d ⎠

Where:

Fu = pullout capacity
b = specimen thickness
d= diameter of bolt
a= distance to support
q and g = constants depending on restraints (given in 9)

This equation was never adopted due to its complexity and lack of generality,
but one can easily see that it is fracture mechanics based, and it illustrates that
solutions to such complex problems can be proposed using only analytical tools.
Figure 8: Simulation Setup. (a) crack trajectory (b) load-anchor head displacement
curve (c) normalized S.I.F.

4.2 3D FE Analysis of Anchor Bolts

with Large Embedment Depths
Ozbolt et al. [11] conducted numerical
testing to examine the effects of larger heads for
cast in place studs, and for very long embed
depths. The finite element code used is called
MASA. The code uses the microplane model and a
smeared crack approach. Steps were taken to
relate the total energy consumption to the concrete
fracture energy, thus make the meshing size
independent.
Only one quarter of the anchor bolt is
modeled due to symmetry (Figure 9). The results
of multiple analyses using this method have been
compared with the CC-method (same as the CCD Figure 9: Typical Crack
Patterns: a) small head b) large
head
 Figure 10: Relative Concrete Cone Resistance as a Function of the
Embedment Depth

method, metric units in this case). These results generally agrees with the conclusions
drawn above, that the CCD method accurately predicts the tensile capacity of the
anchors. In addition, the resulting FEM failure planes (figure 9) support the
“assumption” that the failure cone is at 35 degrees. One discover of this analysis,
however, is that the CCD method is overly conservative for anchors with large heads.
This can be seen in Figure 10 and Table 2.

As noted in Section 3, the CCD method is based on size effect and behaves in an
LEFM manner. The authors give a fairly simple explanation for why the small anchor
heads agree with CCD predictions, and large ones don’t:

The reason why for fasteners with small anchor head size the size
effect agrees well with the size effect prediction according to LEFM is
 due to the fact that for all embedment depths the crack patterns at
peak load are similar – the crack length is relatively small and
approximately proportional to the embedment depth. The main
assumption of LEFM…is fulfilled and therefore size effect is maximal.
On the contrary, for fasteners with larger heads the crack pattern for
different embedment depths is not proportional, This is the case for
both the crack length at peak load, as well as for the shape of the
failure cone. Consequently, the size effect on the concrete cone failure
load is smaller.

And thus, larger headed anchor studs move up the size-effect curve, and are better
described by Non-Linear Fracture Mechanics (NLFM). The authors also propose the
following addition to the CCD method (metric):

Pu = 15.5 f 'c hef

1.5
(4.4)

Pu = 15.5 ⋅ γ 0.35 ⋅
1.5
f 'c hef (4.5)

where:
A PU
γ = ; A0 = (4.6)
A0 15 f 'c
and:
A= bearing area under the head of the anchor
A0= bearing area under the head of the anchor such that for Pu from Eq 4.4, the ratio
PU
A = 15
f 'c

Ozbolt et al. claim that this modification accounts for the discrepancy. Note that Eq
4.4, valid for small heads is equal to Eq 3.3. This equation has yet to be adopted by an
American code, and studs with larger heads are not addressed.

5.0 Further Study

One very important factor, though not directly related to fracture mechanics, to
the strength of multiple anchor connections (such as column baseplates), is the relative
stiffness of the plate that connects the anchorages. When a moment is applied to such
a system, the connecting element can not be too stiff or too flexible. This will result in a
lessening of the moment arm and an increase in the force on the anchors, possibly
causing failure [12]. Thus, the designer must consider that the tensile capacity of the
anchorages alone may not necessarily govern the connection. The relationship between
the anchorages, their connecting element, and the corresponding strengths and
stiffnesses is an important one and has not been studied in depth.
Another area that has been as yet neglected is that of fatigue behavior of
anchorages. With continuing sophistication of the models, and the current trend of
investigating the capacity of anchorages subject to high temperatures, there is no doubt
that this problem will be addressed by fracture mechanics based finite element
simulation in the not too distant future.
References

1 Li, Y., Elingehausen, R., Lehr, B, and Ozbolt, J. “Fracture Analysis of Quadruple
Fastenings with Bonded Anchors”. Fracture Mechanics of Concrete Structures. De
Borst, R., Mazars, J. Pijauduer-cabot, G., Van Mier, J. (Editors). 2001, Swets &
Zeitlinger, Lisse, Switzerland

2 Eligehausen, R., Balogh, T. “Behavior of Fasteners Loaded in Tension in Cracked

Reinforced Concrete”. ACI Structural Journal. 1995. Vol 92, No 3, pp 365-379.

3 Nienstedt, J., Matner, R. “Three Dimensional Modeling of Anchoring Systems in

Concrete”. Fracture Mechanics of Concrete Structures. De Borst, R., Mazars, J.
Pijauduer-cabot, G., Van Mier, J. (Editors). 2001, Swets & Zeitlinger, Lisse,
Switzerland.

4 Fuchs, W., Eligehausen, R., Breen, J. “Concrete Capacity Design (CCD) Approach for
Fastening to Concrete” ACI Structural Journal. 1995. Vol 92, No. 1, pp 73-94.

5 Cannon, R. “Straight Talk About Anchorages to Concrete” 1995, Vol 92, No. 5 pp,
580-586 and N0. 6, pp. 724-734.

6 Orr, Richard “Disc. 92-S9” ACI Structural Journal. 1995. Vol 92, No. 6, pp 792-793.

7 Fox, K. “Fracture Mechanics Analyses of Anchor Bolts”. Unpublished Thesis,

University of Colorado, 1990.

8 Building Code Requirements for Structural Concrete (ACI-318-02), Appendix D.

9 Chahrour, A., Outsu, M. “Analysis of Anchor Bolt pull-out tests by a two-domain

Boundary Element Method.” Materials and Structures. 1995, Vol 28. pp 201-209

10 Hoffman, J., Eligehausen, R., Ozbolt, J. “Behavior and design of Fastenings with
Headed Anchors at the Edge Under Tension and Shear Load”. Fracture Mechanics of
Concrete Structures. De Borst, R., Mazars, J. Pijauduer-cabot, G., Van Mier, J.
(Editors). 2001, Swets & Zeitlinger, Lisse, Switzerland.

11 Ozbolt, J., Eligehausen, R., Periskie, G., Mayer, U. “3D FE analysis of anchor botls
with large embedment depths.” Fracture Mechanics of Concrete Structures. Li, V,
Leung, C., Willam, K., Billington, S. (Editors). FraMCos, 2004. Vail Colorado.

12 Fichtner, S. and Eligehausen, R. “Stiffness Requirements of Baseplates” Fracture

Mechanics of Concrete Structures. Li, V, Leung, C., Willam, K., Billington, S. (Editors).
FraMCos, 2004. Vail Colorado.