You are on page 1of 10


Title no. 105-S43

Design and Analysis of Heavily Loaded Reinforced

Concrete Link Beams for Burj Dubai
by Ho Jung Lee, Daniel A. Kuchma, William Baker, and Lawrence C. Novak

This paper presents a study on the shear capacity and behavior of The center hex reinforced concrete core walls provide the
reinforced concrete link beams designed for the Burj Dubai Tower, torsional resistance of the structure similar to a closed pipe
which is the tallest building in the world and will be ready for or axle, as shown in Fig. 3. The center hex walls are
occupancy in 2009. Several thousand reinforced concrete link
beams were used in this structure to interconnect structural walls;
in some cases, the factored shear forces in these link beams were
up to three times the traditional nominal ACI shear force strength
limit. This study presents an examination of the factors that control
the design and behavior of heavily loaded reinforced concrete link
beams. Nonlinear finite element analysis methods were used to
validate and inform the design of the reinforced concrete link
beams for Burj Dubai and to examine if the ACI nominal sectional
shear force limit is appropriate for this type of member. The results
illustrate the undue conservatism of the ACI design provisions and
the role of nonlinear analyses in design.

Keywords: beam; design; nonlinear finite element analysis; reinforced

concrete; shear; strut-and-tie model.

The Burj Dubai Tower, when completed, will be the
world’s tallest structure. Whereas the final height of this
multi-use skyscraper is a “well-guarded secret,” it will
comfortably exceed the current record holder of 509 meter Fig. 1—Tower rendering.
(1671 ft) tall Taipei 101. The 280,000 m2 (3,000,000 ft2)
reinforced concrete tower will be used for retail, an Armani
hotel, residences, and offices. The goal of the Burj Dubai
Tower is not simply to be the world’s highest building—it is
to embody the world’s highest aspirations.
Designers purposefully shaped the structural concrete for
the Burj Dubai to be Y-shaped in plan to reduce the wind
forces as well as to keep the structure simple and foster
constructibility. The structural system can be described as a
buttressed core, as shown in Fig. 1 and 2. Each wing, with its
own high-performance concrete core and perimeter
columns, buttresses the others via a six-sided central core, or
hexagonal hub. The result is a tower that is extremely stiff
torsionally. The design team purposely aligned all the
common central core and column elements to form a
building with no structural transfers.
Each tier of the building steps back in a spiral pattern that
causes the tower’s width to change at each setback. The
advantage of this stepping and shaping is to “confuse the
wind.” The wind vortexes never become organized because
at each new tier the wind encounters a different building
shape that reduces the overall wind loads on the structure.
Due to the tapering of the tower, the primary demand on the Fig. 2—Construction photo of tower.
link beams is from gravity load redistribution, flow from the
taller core to the perimeter of the structure. The 280,000 m2 ACI Structural Journal, V. 105, No. 4, July-August 2008.
(3,000,000 ft2) tower and 185,000 m2 (2,000,000 ft2) MS No. S-2007-030 received January 19, 2007, and reviewed under Institute
publication policies. Copyright © 2008, American Concrete Institute. All rights reserved,
podium structures are currently under construction, as shown including the making of copies unless permission is obtained from the copyright proprietors.
Pertinent discussion including author’s closure, if any, will be published in the May-
in Fig. 2. The project is scheduled for completion in 2009. June 2009 ACI Structural Journal if the discussion is received by January 1, 2009.

ACI Structural Journal/July-August 2008 451

concrete strengths ranging from C80 to C60 cube strength
Ho Jung Lee is an Engineer with SC Solutions, Inc., Sunnyvale, CA. He received his
BS and MS from the Seoul National University, Seoul, South Korea, and his PhD from and contained portland cement and fly ash. Local aggregates
the University of Illinois at Urbana-Champaign, Urbana, IL. were used for the concrete mixture design. The wall and
column sizes were optimized using virtual work/LaGrange
Daniel A. Kuchma, FACI, is an Assistant Professor in the Department of Civil and
Environmental Engineering at the University of Illinois at Urbana-Champaign. He multiplier methods.1 This results in a very efficient structure.
received his PhD from the University of Toronto, Toronto, ON, Canada. He is a The structure was analyzed for gravity (including P-Δ
member of ACI Subcommittee 318-E, Shear and Torsion.
analysis and creep and shrinkage), wind, and seismic loads
William Baker is a Partner of Skidmore, Owings & Merrill LLP, Chicago, IL. by a three-dimensional analysis model that consisted of
the reinforced concrete walls, link beams, slabs, mats, piles,
Lawrence C. Novak is an Associate Partner of Skidmore, Owings & Merrill LLP. and the spire structural steel system. The model consisted of
over 73,500 shells and 75,000 nodes.
buttressed by the wing walls and hammer head walls, which
behave as the webs and flanges of a beam to resist the wind RESEARCH SIGNIFICANCE
shears and moments. Outriggers at the mechanical floors This paper examines methods for the design and analysis
allow the columns to participate in the lateral load resistance for reinforced concrete link beams that are cast integral with
of the structure; hence, all of the vertical concrete is used to wall piers. In particular, the appropriateness of the
support both gravity and lateral loads. The walls had strut-and-tie method and shear design stress limits for this
class of member are reviewed. The results of this investigation
indicate that much higher shear stress levels should be
permitted in ACI 318, as this would greatly extend the utility
of this class of member.


The demands on the link beams vary greatly and are
dependent on the location of the link beam relative to a
setback with the largest shear forces being generated in the
beams closest to a setback. The typical link beams used in
the Burj Dubai are quite stocky with a shear-span ratio (l/2h)
of 0.85, a width of 650 mm (2.13 ft), and a height of 825 mm
(2.7 ft). For the design of reinforced concrete link beams, the
conventional deep beam design method in the ACI 318-992
and the strut-and-tie method in ACI 318-023 were used, with
Fig. 3—Three-dimensional view of single floor. Appendix A enabling the design of link beams somewhat
beyond the conventionally designed maximum deep beam
stress limit of 10 f c′ in psi (0.83 f c′ in MPa), which is
based on ACI 318-99,2 Section 11.8.4, as will be discussed
in the following. In the case of members subjected to very
large shear forces, embedded built-up structural steel
sections were provided within the core of the concrete link
beams to carry the entire shear and flexure demand.
This study was principally conducted during the structural
design of the Burj Dubai and used to check and inform the
design by the different methods and to examine the condition
and stiffness of the reinforced concrete link beams under
service and factored loads. In addition, a study was made of
the appropriateness of nominal shear stress limits for link
beams. A series of nonlinear analyses were conducted
that can account for the influence of many factors on
response including the amount of flexural reinforcement,
the distribution of vertical and horizontal web reinforcement,
the span-depth ratio, and the confinement provided by walls
at the ends of the link beams. The nonlinear finite element
analysis tools used in this investigation were ABAQUS,4
ADINA,5 and VecTor2.6


The geometry, factored loads, and design methods of four
Burj Dubai link beams, LB1 to LB4, are shown in Table 1.
These link beams, which have the same external dimensions,
capture the range of typical shear design force levels for
which different design solutions were used. Table 2 and Fig. 4
present details on Link Beams LB1 and LB2, as well as Link
Fig. 4—Design details for analyzed link beams. (Note: 1 mm = Beams LB2A and LBRCMAX that are more heavily reinforced
0.0394 in.) and hypothetical link beams whose behavior is evaluated in the

452 ACI Structural Journal/July-August 2008

nonlinear analyses. Link Beam LB1 was designed by the The strut-and-tie model used for the design of Link Beam LB2
deep design method specified in ACI 318-99.2 Link Beam is shown in Fig. 5. The design yields the required amount of
LB2 was designed by the strut-and-tie model in Appendix A horizontal and vertical tie reinforcement and the required
of ACI 318-02.3 Vertical shear or tie reinforcement in Link strength of diagonal concrete struts and nodal zones. Minimum
Beams LB1 and LB2 was determined using the selected horizontal web reinforcement was provided in the link beams
design approaches to support their factored design loads. even though it was not specifically required when the strut-and
Horizontal web reinforcement close to the minimum amount tie design procedure is used. In this link beam design, the amount
suggested for deep beams in ACI 318-023 was used in Link of horizontal and vertical ties can simply be increased to support
Beams LB1 and LB2. The design of Link Beam LB2 using the larger shear loads until the strength is limited by the strength of
strut-and-tie method in the ACI 318-023 provisions, unchanged the diagonal struts and nodal zones.
in ACI 318-05,7 is described in the next section. As previously The use of Appendix A in ACI 318-023 and ACI 318-057
mentioned, a pure reinforced design concrete solution was not provides for a direct design of the diagonal strength of struts so
possible for all members by ACI 318-02,3 or would be by ACI as to avoid a diagonal compressive failure. This permits a
318-05,7 such that composite members with steel-embedded member to be designed for a higher shear stress than the limit for
sections were used to support the shear design in members deep beams in ACI 318-99.2 This deep beam stress limit was also
subjected to very large shear and flexural forces, including Link set to guard against diagonal compression failures but is not
Beams LB3 and LB4. An objective of this study was to investigate considered to be necessary for deep beams as will be discussed in
whether or not it was possible to develop a pure reinforced the following. In the design of Link Beam LB2 by
concrete solution to support the very large shear forces in Link Appendix A of ACI 318-02, the conservative assumption of
Beams LB3 and LB4. To this end, the behavior of more heavily
narrow bottle-shaped diagonal struts were made even though
reinforced members, Link Beams LB2A and LBRCMAX, as
the results of the analyses indicate that there is a uniform
described in Table 2 and Fig. 4, will also be examined.
field of diagonal compression throughout these members.
The concrete cylinder compressive strength used in the
Thus, it was considered that the nominal capacity calculated
design of these link beams was fc′ = 64 MPa (9280 psi); a
for this strut-and-tie model by ACI 318-02 would lead to a
concrete cube strength of 80 MPa (11600 psi) was specified
conservative design.
and actual cube and cylinder breaks indicate considerably
stronger concrete. The design yield strength of flexural rein-
forcement used in the link beams was 460 MPa (67 ksi) and MAXIMUM SHEAR STRENGTH SPECIFIED
of the vertical stirrups and horizontal web reinforcement was IN CODE PROVISIONS
420 MPa (61 ksi). Reinforcement with a yield strength of There is a large variation in the nominal shear design stress
460 MPa (67 ksi) was actually provided, but in accordance limit that is specified in codes of practice even though the
with Section 11.5.2 of ACI 318-02,3 the effective strength reason for this limit is the same as guarding against a diagonal
was taken as 420 MPa (61 ksi). Nonlinear finite element
analyses of these reinforced concrete link beams (LB1, LB2, Table 1—Geometry, loading, and design methods
LB2A, and LBRCMAX) were performed as will be presented. for link beams
Geometry Factored loads
STRUT-AND-TIE MODEL USED IN LINK BEAM Beam Width, Depth, Span, Shear, Moment
The strut-and-tie method has recently developed as a rational ID mm mm mm kN kN-m Design method used
method in the design of discontinuity (D)-regions in structural Conventional (ACI 318-99,
LB1 650 825 1400 1705 1194
concrete such as deep beam, squat walls, pile caps, and other Section 11.8)
elements in which plane sections do not remain plane. The Strut-and-tie
strut-and-tie method provides a conceptually simple design LB2 650 825 1400 2805 1164 (ACI 318-02,
Appendixes A and C)
methodology based on the lower-bound theorem of limit
LB3 650 825 1400 3750 2625 Steel plate
analysis.8 Provisions for using the strut-and-tie method were
LB4 650 825 1400 5250 3675 Built-up steel I-beam
included as Appendix A in ACI 318-02.3
Notes: Factored loads are equal for both ends of link beams. Ratio of ultimate load to
sustained day-to-day (gravity only) service loads is approximately 2.5. Walls adjacent
to link beams are 650 mm thick and are typically reinforced with a minimum of
T20mm at 350 mm on each vertical and horizontal face. 1 in. = 25.4 mm; 1 kip =
4.448 kN; and 1 ft-kip = 0.356 kN-m.

Table 2—Reinforcing details of link beams

analyzed in study
Reinforcement Stirrups
Top Bottom Side bars Spacing,
Beam ID bars bars each face Size mm Type
LB1 5-T32 5-T32 5-T12 T16 150 Two hoops
LB2 12-T32 12-T32 4-T12 T16 125 Two hoops
LB2A 18-T32 18-T32 4-T12 T16 80 Three hoops
LBRCMAX 27-T32 27-T32 4-T12 T16 75 Five hoops
Notes: T32, T20, T16, and T12 are deformed reinforcing bars with respective diameters of
32, 20, 16, and 12 mm. In LB2, top and bottom bars were used in two layers each. LB2A
and LRCMAX were not used in Burj Dubai project. They are included for purpose of
examining appropriateness of current ACI 318-05 limit on maximum shear stress. LB2A
Fig. 5—Strut-and-tie model used in design of Link Beam LB2. has significantly more longitudinal tension reinforcement and transverse reinforcement
than LB2. LBRCMAX is analyzed to figure out maximum shear capacity of reinforced
(Note: 1 in. = 25.4 mm; 1 kip = 4.448 kN; 1 psi = 6.89 × concrete link beam. In LB2A and LBRCMAX, top and bottom bars were placed in two
10–3 MPa; and 1 ft-kip = 1.356 kN-m.) layers and three layers each, respectively. 1 in. = 25.4 mm.

ACI Structural Journal/July-August 2008 453

compression failure. For the limits shown in Table 3, the beams is investigated in the nonlinear finite element analyses
range in maximum permitted shear stress is more than a that are presented in the next section.
factor of two for the 64 MPa (9280 psi) concrete used in the
design of the Burj Dubai. The ACI conventionally designed NONLINEAR FINITE ELEMENT ANALYSIS
beam shear limit of 10 f c′ (psi) (0.83 f c′ [MPa]) was The structural behavior of reinforced concrete link beams
derived from simply supported beam tests in which there is is influenced by many factors including the amount of flexural
a funneling and thus magnification of diagonal compressive reinforcement, vertical and horizontal shear reinforcement, the
stresses as they flow from the top of the beam to its support span-depth ratio, and the confinement effect of the adjacent
and in which anchorage failures are common. The AASHTO pier walls. Nonlinear finite element analyses can be used to
LRFD9 and CSA (Canadian) code 10 shear stress limit of evaluate the effect of these parameters. In this section, the
0.25fc′ for reinforced concrete members was derived from nonlinear finite analyses of the Burj Dubai link beams (LB1
the modified compression field theory.11) In this approach, and LB2) and the more heavily reinforced link beams (LB2B
the influence of diagonal tensile strains on the capacity of a and LBRCMAX), as described in Table 2, are presented.
uniform diagonal compression field is directly considered. Three different programs, ABAQUS, ADINA, and
The flow of diagonal forces in link beams, which are VecTor2, were used for the finite element analysis of these
supported over their ends, are expected to be uniform as reinforced concrete link beams. In ABAQUS, the concrete
illustrated later and thus the limit in the LRFD and CSA damaged plasticity model was selected. This model is based
provisions is more appropriate. For the reinforced concrete on the work of Lubliner et al.12 and Lee and Fenves.13 The
link beams used in the Burj Dubai, there is the further structural concrete damaged plasticity model is intended to provide a
advantage of the confinement effect from the pier walls. The general capability for the analysis of concrete structures
walls on both sides help to prevent vertical expansion of the link subjected to static and dynamic loading under low confining
beams at their ends, which limits diagonal cracking. The pressures. It is based on the combination of nonassociated
use of a higher shear design stress limit in these link multi-hardening plasticity and scalar damaged elasticity
models. The program ADINA provided the option to adjust
various concrete strength and ductility parameters as well as
failure envelopes to better represent the concrete behavior
for a specific application. The program VecTor2 implements
the rotating-angle smeared-crack model of the modified
compression field theory and quadrilateral elements are
used in the analyses.
Before conducting the finite element analyses of these link
beams, the finite element models were calibrated and validated
using experimental results that accounted for the effects of
complex geometries, loadings, and edge effects. Because
this study focuses on the strength of the link beams, it was
investigated to see if the selected parameters for the concrete
material model in ABAQUS, ADINA, and VecTor2 predict
well the strength of link beams tested to failure in laboratories.
One of the comparisons used the link beam tested under
monotonic loading by Galano et al.14 that had a shear span
Fig. 6—Geometry and detailing of link beam tested by Galano ratio of 0.75. This ratio was the closest to that of the Burj
et al.14 (Note: 1 in. = 25.4 mm; and 1 psi = 6.89 × 10–3 MPa.) Dubai link beams (shear ratio = 0.85) for the identified test
data. The geometry and reinforcement details of the tested

Table 3—Maximum nominal shear design stress

permitted by code provisions
Maximum shear strength,
Design code Vn, max

ACI 318-02 and ACI 318-05 V c + 8 f c′ b w d

AASHTO STD1 V c + 8 f c ′b w d

AASHTO LRFD 3 0.25f c′ b v d v + V p

AASHTO segmental bridges 2 12 f c ′b w d

CSA (1994)10 0.25f c′ b v d v + V p

CSA (2004)11 0.25f c′ b v d v + V p

JSCE14 15 f c ′b w d
Notes: Vc equals nominal shear resistance provided by concrete, fc′ equals concrete
compressive strength, bw equals width, d equals distance from compression face to
Fig. 7—Predicted load-deformation response of link beam centroid of longitudinal tension reinforcement, bv equals web width including adjustment
tested by Galano et al.14 (Note: 1 in. = 25.4 mm; and 1 kip = for presence of ducts, dv equals effective shear depth, and Vp equals component in
direction of applied shear of effective prestressing force.
4.448 kN.)

454 ACI Structural Journal/July-August 2008

link beam are shown in Fig. 6. Four-node plane stress isopara- concrete). These input parameters define the stress-strain
metric elements were used in the analyses. Reinforcing steel curve of concrete.
was modeled as two-node truss elements. The top section of In ABAQUS, the compressive and tensile response was
the left wall is restrained and the shear force is applied to the defined similarly as in ADINA, but the prediction by
bottom section of the right wall by displacement control; the ABAQUS was based on the concrete damaged plasticity
same prescribed vertical displacement is used along the model. In this model, the value of the flow potential eccentricity
section. The analysis results are presented in Fig. 7 and the parameter has a significant effect on the concrete response. This
strengths predicted by ABAQUS, ADINA, and VecTor2 are value was set to be 20 based on the calibration with experimental
shown to predict the experimentally measured capacity data, whereas the dilation angle was similarly selected to be 36.
reasonably well. The results of other validation studies were Default values were used for the other input parameters in
similar. Therefore, the selected parameters for the concrete the concrete damaged plasticity model.
model in ABAQUS, ADINA, and VecTor2 were taken to be
sufficiently accurate for capacity evaluation of the four link DISCUSSION OF PREDICTED LINK
beams in this study. BEAM BEHAVIOR
The load and boundary conditions of the finite element The finite element analysis results of Link Beam 1 (LB1)
model for the Burj Dubai link beams are shown in Fig. 8. The are shown in Fig. 9 through 11. ABAQUS, ADINA, and
wall boundary is extended vertically up to the depth of the VecTor2 all predicted similar responses for Link Beam LB1.
link beam and horizontally up to the length of the link beam. According to Fig. 9, the predicted capacity is much larger
This extension with the constraints put in the finite element than the factored design load. Therefore, the use of the deep
model provides a realistic boundary for the investigation of beam design method as specified in the ACI 318-992 Code is
link beam capacity that was expected to lead to conservative considered to lead to a very conservative design for this link
(lower bound) estimations of capacity. The load and beam. Crack patterns at the loading steps similar to the
boundary conditions are the same as that of the link beam factored design load and at the ultimate load are shown in
tested by Galano et al.14 except the stiff perimeter steel was Fig. 10. At the factored design load, flexural cracks occur at
used to account for the constraining effect of the wall as the the boundary regions of the link beam under tension,
boundary of the model. The additional benefit of axial
compression in the walls, as investigated by other analyses
not presented in this paper, was found to only have a very
minimal beneficial effect on the capacity of the link beams;
it is the link beam end restraint rather than axial compression
in the wall piers that influences the behavior of the link
beams. Thus, the axial load acting on the walls was not
considered in the reported analyses. The strain variation of
reinforcing steel identified in Fig. 8 was investigated as well
as the crack patterns and capacity of the link beams.
Additional information on the finite element models is
now presented and this is followed in the next section by the
predicted capacity and behavior of the link beams. In
creating the finite element models, it was necessary to select
and specify material models and select convergence limits.
In all analyses, the response of the reinforcement was linear
elastic and perfectly plastic. The default value of the
convergence limit in VecTor2 was used, a convergence
value of up to 0.01 was used in ADINA, whereas the stability Fig. 8—Load and boundary conditions used in modeling of
command was used in ABAQUS, but its use was found to Burj Dubai link beams.
have little effect on the predicted behavior. The philosophy
of VecTor2 is that the user is not required to select nonbasic
material parameter values and thus only the compressive
cylinder strength was specified and all other default values
were accepted. ADINA and ABAQUS require the user to
select many other parameter values. The selections made in
these programs are given in the following.
In ADINA, the required concrete material input parameters
were tangent modulus at zero strain (44,000 MPa [6380 ksi]),
Poisson’s ratio (0.15), uniaxial cut-off tensile stress (4.83 MPa
[700 psi]), post-cracking uniaxial cut-off tensile stress (4.83 MPa
[700 psi]), uniaxial maximum compressive stress (–64 MPa
[–9280 psi] for unconfined concrete; –102.4 MPa [–14848 psi] for
confined concrete) and corresponding compressive
strains (–0.002 for unconfined concrete; –0.0032 for confined
concrete), ultimate uniaxial compressive stress (–54.4 MPa
[–7888 psi] for unconfined concrete; –87 MPa [–12615 psi]
for confined concrete), and ultimate uniaxial compressive Fig. 9—Predicted load-deformation response of Link
strain (–0.003 for unconfined concrete; –0.0048 for confined Beam LB1. (Note: 1 in. = 25.4 mm; and 1 kip = 4.448 kN.)

ACI Structural Journal/July-August 2008 455

whereas no shear cracking was observed in the central analysis of Link Beam LB2, the capacity of Link Beam LB2
portion of the link beam. The development of straining in the is slightly decreased due to concrete crushing at the ends of
longitudinal and shear reinforcement is presented in Fig. 11. the link beam. This is not expected in practice because the
The longitudinal tensile strain increased gradually with pier walls provide significant restraint to the vertical expansion
increasing shear force until the capacity of the member was of the link beam and this increases the concrete compressive
reached at the point of longitudinal yielding. As shown in strength at the ends of these link beams. VecTor2 directly
Fig. 11(b), the transverse reinforcement strain rapidly considers concrete confinement and softening without
increased with the formation of diagonal cracks, but the additional input from users. When the confined concrete
transverse strain at the capacity of the member was less than
the yield strain of 0.0021. At the ultimate load, severe flex-
ural cracks occurred and extensive shear cracking was
predicted. In summary, the finite element analysis results
predict that the failure mode of Link Beam LB1 was yielding
of longitudinal tension reinforcement and that shear failures
did not occur. It is also observed that Link Beam LB1, as
designed by the deep beam design method in ACI 318-99,2
was more conservatively designed for shear than for flexure.
The predicted load-deformation response of Link Beam 2
(LB2) by the nonlinear finite element analyses is shown in
Fig. 12 in which the predicted capacity is considerably larger
than the factored design load. The strut-and-tie method specified
in ACI 318-057 was therefore observed to lead to a conservative
design for this link beam. ABAQUS, ADINA, and VecTor2
provide similar predictions of behavior of this link beam.
Two different concrete models were used in the ADINA
analyses. One is an unconfined concrete stress-strain model
and the other is a confined concrete model. ADINA and
ABAQUS do not directly consider the confinement effect by
the pier walls at the end of the link beams. Thus, if the
unconfined model is used in ABAQUS and ADINA for the

Fig. 11—Predicted development of reinforcement strains in

Link Beam LB1. (Note: 1 kip = 4.448 kN.)

Fig. 12—Predicted load-deformation response of Link

Fig. 10—Predicted crack patterns by ADINA for Link Beam LB2. (Note: 1 psi = 6.89 × 10–3 MPa, 1 in. = 25.4 mm;
Beam LB1. (Note: 1 kip = 4.448 kN.) and 1 kip = 4.448 kN.)

456 ACI Structural Journal/July-August 2008

model is used in ABAQUS and ADINA for the analysis of strains were considerably smaller than yield strain at the
the link beams, it gives similar results to those of the VecTor2 point of yielding of the longitudinal reinforcement.
default model. The predicted development of reinforcement Therefore, it is predicted that the capacity of these link
strains for Link Beam LB2 is shown in Fig. 13. For Link beams is limited by their flexural capacities.
Beam LB2, the strain in the transverse reinforcement Sectional analyses of Link Beams LB2A and LBRCMAX
reaches yield before yielding of the longitudinal reinforcement. were also performed using the sectional analysis program
The capacity and mode of failure of Link Beam 2 is predicted Response 2000.15 Response 2000 15 is also based on the
to occur at the point of yielding of both the longitudinal and modified compression field theory and employs the engineering
transverse reinforcement. beam theory assumption that plane sections remain plane.
The finite element analysis results of hypothetical Link Unlike with the nonlinear finite element analyses,
Beams LB2A and LBRCMAX, which contained higher Response 200015 cannot account for the beneficial effects of
levels of longitudinal and shear reinforcement, are shown in confinement and strut action. Under combined shear and
Fig. 14 through 17. ABAQUS, ADINA, and VecTor2 bending loads, the sectional capacities predicted by
provide similar predictions of the behavior of these link beams. Response 200015 for Link Beams LB2A and LBRCMAX
Without consideration of the confinement effect by the walls, were 6026 and 7600 kN (1355 and 1709 kips), respectively.
the analysis results predict a maximum shear stress capacity The state of cracking in Link Beam LBRCMAX was also
limit that is 70% larger than the conventionally designed predicted by program Response 200015 as shown in Fig. 18
beam shear stress limit of 10 f c′ (psi) (0.83 f c′ [MPa]) that is at its ultimate capacity. The member analysis by Response
specified in ACI 318-05.7 If the confinement effect by the 200015 for Link Beam LBRCMAX supports the conclusion
pier walls is considered in the analysis through the use of from the nonlinear finite element analyses that a shear
the confined concrete stress-strain model, the shear stress force close to the LRFD and CSA limit can be supported
capacity of the reinforced concrete link beam can be 2.5 times by a reinforced concrete link beam of the dimensions used
the ACI shear stress limit. This is illustrated in the in the Burj Dubai.
predicted behavior of Link Beam LBRCMAX in which The strength results of this investigation are summarized
very significant amounts of longitudinal and transverse in Fig. 19, which presents a comparison of the factored shear
reinforcement were provided. For both Link Beams LB2A
and LBRCMAX, the predicted transverse reinforcement

Fig. 14—Predicted load-deformation response of hypothetical

Link Beam LB2A. (Note: 1 in. = 25.4 mm; and 1 kip =
4.448 kN.)

Fig. 15—Predicted load-deformation response of hypothetical

Fig. 13—Predicted development of reinforcement strains in Link Beam LBRCMAX. (Note: 1 in. = 25.4 mm; and 1 kip =
Link Beam LB2. (Note: 1 kip = 4.448 kN.) 4.448 kN.)

ACI Structural Journal/July-August 2008 457

design loads for link beams (LB1 to LB4) in the Burj Dubai for Link Beam LB1 that was designed by the ACI 318-992
and the predicted shear capacities by the nonlinear finite deep beam provisions and for Link Beam LB2 that was
element analyses for Link Beams LB1, LB2, LB2A, and designed by the strut-and-tie model of Appendix A in ACI
LBRCMAX. The shear force corresponding to the ACI 318-02.3 In the Burj Dubai, composite link beams with large
318-057 conventionally designed beam shear stress limit of embedded sections were used to support the shear and
10 f c′ (psi) (0.38 f c′ [MPa]) is also shown. The results of flexure demand when this demand was in excess of what
the analyses predict that the designs are very conservative could be designed for by ACI 318 sectional or strut-and-tie
design procedures. The results of the analyses for hypothetical
Link Beams LB2A and LBRCMAX suggest that it would have
been possible to support the large shear force demands in Link
Beams LB3 and LB4 with reinforced concrete link beams;
however, reinforcing bar congestion and constructibility
concerns would need to be evaluated for Link Beams LB2A and

In this paper, the design and capacity of heavily loaded
reinforced concrete link beams were investigated by
nonlinear finite element analyses using the programs
ABAQUS,4 ADINA,5 and VecTor2.6 The following is a
summary of results from this study:
1. The effect of the pier walls at the ends of the link beams
was predicted by the analyses to reduce the vertical transverse
expansion at the ends of the members and thereby reduce the
demands on transverse reinforcement. The pier walls also
provided confinement that enabled the link beams to support
larger compressive stresses at their ends. Furthermore, they
enabled a more uniform field of diagonal compression and
vertical distribution of shear over the depth of the member
throughout the entire length of the link beam;

Fig. 16—Predicted development of reinforcement strains in

hypothetical Link Beam LB2A. (Note: 1 kip = 4.448 kN.)
Fig. 18—Predicted cracking in hypothetical Link Beam
LBRCMAX by Response 2000 at ultimate load (V = 1709 kips
[7600 kN]). (Note: 1 in. = 25.4 mm; and 1 kip = 4.448 kN.)

Fig. 17—Predicted development of reinforcement strains in

hypothetical Link Beam LBRCMAX (Location A in Fig. 8). Fig. 19—Comparison of shear demands and capacities in
(Note: 1 kip = 4.448 kN.) link beams.

458 ACI Structural Journal/July-August 2008

2. The use of the deep beam design method in the ACI REFERENCES
318-992 provisions leads to very conservative shear designs 1. Baker, W.; Novak, L.; Sinn, R.; and Viise, J., “Structural Optimization
for link beams as suggested by the results of multiple of 2000’ Tall 7 South Dearborn Building,” Proceedings of the 2000 ASCE
nonlinear finite element analyses. The ACI 318-992 shear Structures Congress, Track: 14th Analysis and Computational Conference,
stress limitation for deep beams, which is a function of the 2000.
span-depth ratio and imposed to guard against a diagonal 2. ACI Committee 318, “Building Code Requirements for Structural
compression failure, unnecessarily limits the load-carrying Concrete (ACI 318-99) and Commentary (318R-99),” American Concrete
capacity of stocky link beams supported over their heights Institute, Farmington Hills, MI, 1999, 369 pp.
such as those used in the Burj Dubai project; 3. ACI Committee 318, “Building Code Requirements for Structural
Concrete (ACI 318-02) and Commentary (318R-02),” American Concrete
3. The strut-and-tie model in Appendix A of ACI 318-023
Institute, Farmington Hills, MI, 2002, 443 pp.
and ACI 318-057 permits reinforced concrete link beams to be
4. ABAQUS, Version 6.4.3, Hibbitt, Karlsson & Sorensen, Inc.,
designed for substantially higher loads than would be possible
Pawtucket, RI, 2004.
by the use of ACI sectional design methods. In the use of the
5. ADINA, “A Finite Element Program for Automatic Dynamic Incre-
strut-and-tie model, a direct check is made to ensure that the mental Nonlinear Analysis,” Version 8.3, ADINA R & D, Inc., 2006.
diagonal compressive struts have adequate capacity. There-
6. Vecchio, F. J., “VecTor2, Nonlinear Finite Element Analysis Program
fore, a design completed by the strut-and-tie model negates the of Reinforced Concrete,” University of Toronto, Toronto, ON, Canada, 2002.
need to satisfy any sectional shear stress limitation; 7. ACI Committee 318, “Building Code Requirements for Structural
4. The results of nonlinear finite element analyses indicate Concrete (ACI 318-05) and Commentary (318R-05),” American Concrete
that reinforced concrete link beams have significantly larger Institute, Farmington Hills, MI, 2005, 430 pp.
shear capacities than the nominal strengths calculated by the 8. Schlaich, J.; Schafer, K.; and Jennewein, M., “Toward a Consistent
strut-and-tie model of ACI 318-05.7 The capacity of the Design of Structural Concrete,” PCI Journal, V. 32, No. 3, 1987, pp. 75-149.
members was predicted to be controlled by yielding of the 9. AASHTO, “AASHTO LRFD Bridge Design Specifications,” third
longitudinal reinforcement; the compressive and nodal stress edition, Washington, DC, 2004, 1450 pp.
limits in ACI 318-057 were found to be conservative; and 10. CSA Committee A23.3, “Design of Concrete Structures (CSA
5. There is a wide variation in the maximum shear stress A23.3-04),” Rexdale, ON, Canada, 2004.
limits in codes of practice. The difference is more than a 11. Vecchio, F. J., and Collins, M. P., “The Modified Compression Field
factor of two between the sectional design models in ACI Theory for Reinforced Concrete Elements Subjected to Shear,” ACI
318-057 (10 f c′ [psi] [0.83 f c′ (MPa)]) and the Canadian JOURNAL, Proceedings V. 83, No. 2, Feb. 1986, pp. 219-231.
and AASHTO LRFD Design Codes (0.25f c′ ) for link beams 12. Lubliner, J.; Oliver, J.; Oller, S.; and Oñate, E., “A Plastic-Damage
cast with 64 MPa (9280 psi) concrete, as used in the Burj Model for Concrete,” International Journal of Solids and Structures, V. 25,
Dubai. The higher shear stress limit in the Canadian and 1989, pp. 299-329.
AASHTO LRFD code was found to be appropriate for short 13. Lee, J., and Fenves, G. L., “Plastic-Damage Model for Cyclic
link beams that are supported over their height at their Loading of Concrete Structures,” Journal of Engineering Mechanics,
ends by continuous wall piers. ASCE, V. 124, No. 8, 1998, pp. 892-900.
14. Luciano, G., and Vignoli, A., “Seismic Behavior of Short Coupling
Beams with Different Reinforcement Layouts,” ACI Structural
The analyses presented in this paper were performed for Skidmore, Journal, V. 97, No. 6, Nov.-Dec. 2000, pp. 876-885.
Owings & Merrill LLP (SOM) LLP, Chicago, IL, under the direction of 15. Bentz, E. C., and Collins, M. P., “Response 2000,” http://
W. Baker, L. Novak, A. Ozkan, and S. Korista., 2000.

ACI Structural Journal/July-August 2008 459

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.