Geotech Geol Eng
DOI 10.1007/s10706-010-9368-7
ORIGINAL PAPER
A Simplified Approach to Bridge Substructure Design
Thomas Nicholas • Karl E. Barth •
David M. Boyajian
Received: 15 July 2005 / Accepted: 31 August 2009
 Springer Science+Business Media B.V. 2010
Abstract Strut-and-tie modeling is an analysis and
design tool for reinforced concrete elements in which
it may be assumed that internal stresses are trans-
ferred through a truss mechanism. The tensile ties and
compressive struts serve as truss members connected
by nodal zones. The internal truss idealized by the
strut-and-tie model implicitly account for the distri-
bution of both flexure and shear. In 1998, the
AASHTO LRFD Bridge Specifications (1998) incor-
porated the strut-and-tie modeling procedure for the
analysis and design of deep reinforced concrete
members where sectional design approaches are not
valid. In most instances, hammerhead piers can be
defined as deep reinforced concrete members and
therefore, should be designed using the strut-and-tie
modeling approach. However, most bridge designers
have not embraced the strut-and-tie model due to the
unfamiliarity with the design procedure, the inability
to check the truss model’s validity (without
T. Nicholas (&)  D. M. Boyajian
The University of North Carolina at Charlotte, 9201
University City Blvd., Charlotte, NC 28223-0001, USA
e-mail: tnichola@uncc.edu
D. M. Boyajian
e-mail: dboyajia@uncc.edu
K. E. Barth
Jack H. Samples Distinguished Professor, Civil and
Environmental Engineering, West Virginia University,
P.O. Box 6103, Morgantown, WV 26506-6103, USA
e-mail: kebarth@mail.wvu.edu
laboratory tests or a finite element model), and the
time it takes to complete the strut-and-tie model
analysis and design. Therefore, with the formulation
of a well-defined strut-and-tie modeling procedure,
practicing engineers will become more comfortable
with the design method and employ the method more
often and more consistently. This paper presents a
uniform design procedure for employing strut-and-tie
modeling for hammerhead piers that can be readily
utilized by practicing engineers. Additionally, a
comparison between the strut-and-tie method and
standard sectional design practices will be performed
for a typical hammerhead pier.
Keywords Bridge substructure design 
Strut and tie model  Bridge substructures 
Bridge construction  Concrete analysis
1 Introduction
Throughout the past several years, researchers have
sought to reliably and accurately predict the behavior
of deep structural members. For a number of years
designers have been using ‘‘good’’ engineering judg-
ment and ‘‘detailing’’ to handle theses situations.
Now, due to the implementation of strut-and-tie
modeling into the AASHTO LRFD Bridge Specifi-
cations and the ACI Design Code (MacGregor 2000),
designers have the tools to more accurately design
these regions.
123

Previous researchers focused on understanding the
internal distribution of forces in a reinforced concrete
structure have defined two specific regions; B-regions
and D-regions. The B-regions of a structure (where B
stands for a region where Bernoulli Beam theory may
be employed) have internal states of stress that are
easily derived from the sectional forces e.g. bending,
shear, etc. For structural members that do not exhibit
plain strain distribution, e.g. the strain distribution is
non-linear, the sectional force approach in not
applicable. These regions are called D-regions (where
D stands for discontinuity, disturbance, or detail).
The D-regions of a structure are normally corners,
corbels, deep sections, and areas near concentrated
loads. When D-regions crack the treatments used
such as ‘‘detailing’’, ‘‘past experience’’, and ‘‘good
practice’’ often prove inadequate and inconsistent
(Schlaich et al. 1987).
Truss model geometry and the detailing of the
truss elements (nodal zones, tension ties, and com-
pression struts) is an iterative and subjective process.
The tension ties in the truss model may represent one
or several layers of flexural reinforcement in the deep
section. The locations of the tension ties normally are
defined at the centroid of the reinforcing mat. The
struts are concrete compressive stress fields with the
prevailing compression acting along the direction of
the strut (Kuchma and Tjhin 2001). Nodal zones are
the truss joints in the strut-and-tie model. Nodal
zones are formed where the tension ties, compression
struts, and exterior loads intersect.
Figure 1 provides a simple strut-and-tie model
applied to a simply supported deep beam. In this
figure, the lighter shaded regions represent concrete
Fig. 1 Typical truss model for a deep beam. Adapted from
MacGregor 1997
123
Geotech Geol Eng
compressive struts, the steel reinforcing bar repre-
sents a tensile tie, and the dark shaded regions
represent nodal zones.
For further in-depth information on the application
of strut-and-tie modeling for general structures, the
reader is referred to Schlaich et al. (1987), Collins
and Mitchell (1991), Adebar and Zhou (1996), and
MacGregor (1997, 2000). It should be noted while
research has been widely performed on the various
parts of a structure, Schlaich et al. (1987) combined
these individual pieces to allow for the entire
structure to be modeled using the strut-and-tie
approach.
2 Design Procedure
The AASHTO LRFD Specifications states that strut-
and-tie models may be used to determine internal
force effects near supports and the points of appli-
cation of concentrated loads at strength and extreme
event limit states. Additionally, the strut-and-tie
model should be considered for the design of deep
footings and pile caps or other situations in which the
distance between the centers of the applied load and
the supporting reactions is less than twice the
member thickness. Strut-and-tie modeling is covered
by Articles 5.6.3.2 through 5.6.3.6.
2.1 Load Development
The external loads acting on the pier at the nodal zone
locations are the superstructure dead load and live
load reactions. Members contributing to the dead load
reactions are the beam, intermediate diaphragms,
deck, pier diaphragm, parapet, and future wearing
surface. The dead load reactions should be calculated
for the interior and exterior beams separately due to
the difference in effective slab widths. The live load
reactions arise in accordance to the AASHTO LRFD
HL-93 Truck Loading (AASHTO 1998). The live
load reactions should be determined by considering
the structure to be continuous. When placing the
truck component of the HL-93 live load, the designer
should place the second wheel directly over the pier
to insure the maximum reaction is achieved as
illustrated by Fig. 2.
In turn, the HL-93 reaction should be placed so
that to induce the maximum moment on the
Geotech Geol Eng
Fig. 2 HL-93 Reaction
calculation configuration
cantilever of the hammerhead pier. For the design
study, the maximum moment is produced by placing
the truck component of the HL-93 reaction two feet
from the face of the curb or parapet as illustrated by
Fig. 3. The lane load component of the HL-93 is
placed at the face of the curb to produce the
maximum moment as illustrated by Fig. 4. Additional
superstructure load configurations may control and
should be considered when applicable.
Furthermore, the load effects from water, wind,
self-weight, and wind on the live load should be
considered when applicable. The pressure loads, as
well as the self-weight of the pier cap, are distributed
evenly as point loads to each node by dividing the
total pressure load by the total number of nodes.
Fig. 3 Illustration of Wheel Placement
Fig. 4 Illustration of Lane Loading Placement
2.2 Truss Model Background
Adequate representation of the truss model requires a
level of skill and engineering judgment and typically
requires an iterative procedure to produce an ade-
quate reinforcement pattern for a given member. The
process of defining the truss begins by defining the
flow of forces in the member and specifying the nodal
zone locations at points where the external loads act
and the loads are transferred between structural
members, e.g. the pier cap to pier column or at the
supports. The tension ties and compression struts can
then be defined once the nodal zones have been
specified. The tension ties lie along the assumed
centroid of the tensile reinforcement beginning and
terminating at nodal zones. The compression struts
are defined to coincide with the compressive field
and, as with the tension ties, begin and terminate at
the nodal zones (Collins and Mitchell 1991).
The truss should exhibit equilibrium at each node
and should portray an acceptable truss model.
Figure 5 illustrates the difference between an accept-
able model and a poor model. The truss model
represented by Fig. 5b is classified as poor due to the
larger number of tension ties, denoted by the solid
lines.
Furthermore, it is helpful to bear in mind that loads
are transferred through the path with the least forces
and deformations. Since reinforced ties have a greater
potential to deform than their concrete compressive
strut counterparts, the model with the least and
shortest ties should provide the most favorable
model.
Schlaich et al. (1987), proposed a simple criterion
for optimizing a model as derived from the principle
of minimum strain energy for linear elastic behavior
of the struts and ties following the event of cracking.
The contribution of the concrete struts can generally
be omitted because the strains of the struts are usually
123

Fig. 5 (a) An Acceptable
Model (b) A Poor Model.
Adapted from Schlaich
et al. (1987)
much smaller than those of the steel ties. The
minimum number of ties required for the model can
be found by the following minimization:
X As already mentioned, the first step in defining the
Fi li 2mi ¼ Minimum ð1Þ
where Fi = force in strut or tie i, li = length of
member i, 2mi = mean strain of member i.
Liang et al. (2002) developed a performance-based
strut-and-tie modeling procedure for reinforced con-
crete citing the inefficiency of the trial-and-error
iterative process based on designer intuition and
experience. Their optimization procedure consisted
of eliminating the lowest stressed portions from the
structural concrete member to find the actual load
path. Liang et al. (2002) proposed that minimizing
the strain energy is equivalent to maximizing the
overall stiffness of a structure and that the strut-and-
tie system should be based on system performance
(overall stiffness) instead of component performance
(compression struts and tension ties).
Fig. 6 Typical section—
Three Beam Superstructure
123
Geotech Geol Eng
2.3 Simplified Truss Definition Procedure
for Hammerhead Pier Caps
truss involves specifying the nodal zones. These
occur at locations where external loads, e.g. beam
reactions, act on the pier cap and where the stress is
transferred from the cap to the column. The design
example used in this paper involves the three beam
superstructure illustrated in Fig. 6.
Since the three beams produce three reactions at
their centerlines, the nodal zones may then be defined
as shown in Fig. 7. Note that the lower nodal zones
occur where the cap to column steel reinforcement is
located.
The tension ties should be modeled at the
predicted location of the tension reinforcement while
the compression struts represent the primary com-
pressive stress and should be defined accordingly.
Both the tension ties and the compression struts
Geotech Geol Eng

defining tension ties and compressive struts. It should

also be noted that the solid lines of the truss model
represent the tension ties while the dashed lines
represent the compressive struts.

2.4 Dimensioning of Truss Elements: Tensile

Ties, Compressive Struts, and Nodal Zones

The dimensioning of the compression strut, tension

tie, and nodal zones are governed by Articles 5.6.3.2
Fig. 7 Location of Nodal Zones for hammerhead pier cap
through 5.6.3.6 of the AASHTO LRFD Specifications
(1998). For ease of calculation, the governing equa-
should begin and terminate at the nodal zones. The
tions can easily be performed using a spreadsheet.
final truss model should be represented by an
Axial members in the truss model most explicitly
acceptable truss model and have the least number
satisfy force limitations as provided by the following
of tension ties possible.
generalized expression:
Due to the modeling of the truss based partially on
the location of the reinforcing, care should be taken PU  /Pn ð2Þ
to insure that the final reinforcing pattern represents
where: Pn = nominal resistance of strut or tie, /
the truss model. For example, in the design study the
= resistance factor for tension or compression as
flexural reinforcing was assumed to be located at
specified in Article 5.5.4.2
76.200 mm from the face of the concrete. In turn, the
The nominal resistance of a tension tie in a
tension tie was modeled at a depth of 76.200 mm.
hammerhead pier cap should be calculated using the
Therefore, the final detail should show reinforcing at
following equation
the three-inch level. If the location of the flexural
steel exceeds the 76.200 mm depth, then the model Pn ¼ fy Ast ð3Þ
should be resized based on the new centroid of the
where: Ast = total area of longitudinal mild steel
reinforcing mat. The diameter of reinforcing bars
reinforcement in the tie, fy = yield strength of mild
used also dictates the location of the centroid. Smaller
steel longitudinal reinforcement, fpe = stress in pre-
reinforcing bars will normally produce a deeper
stressing steel due to prestress after losses.
centroid and the opposite occurs for larger diameter
Using Pn C Pu // and solving for Ast, the area of
bars. However, care should be taken when specifying
steel required to resist the tensile load can be found,
the larger diameter bars due to violating flexural steel
from which the area of reinforcement can be specified
distribution to control cracking.
based on AASHTO LRFD Specifications.
A typical hammerhead pier cap for a three beam
The first step in determining the capacity of the
superstructure is illustrated in Fig. 8. Additionally,
compression struts is calculating the limiting com-
Fig. 8 presents the optimal truss model achieved by
pressive stress, fcu:
applying the previously discussed methodology for
fc0
fcu ¼  0:85fc0 ð4Þ
0:8 þ 170e1
where:
e1 ¼ es þ ðes þ 0:002Þ cot2 as ð5Þ
and: as = the smallest angle between the compressive
strut and adjoining tension ties (deg), es = the tensile
strain in the concrete in the direction of the tension tie,
fc0 = specified concrete compressive strength.
Utilizing the value of fcu found in Eq. 4, the nominal
Fig. 8 Hammerhead pier cap with acceptable truss model resistance can be calculated depending on the

123

reinforcing pattern used in the hammerhead pier. For
the unreinforced compressive strut, AASHTO gives
the following equation for the nominal resistance:
Pn ¼ fcu Acs ð6Þ
where: Pn = nominal resistance of a compressive
strut, fcu = limiting compressive stress as specified in
Article 5.6.3.3.3, Acs = effective cross-sectional area
of strut as specified in Article 5.6.3.3.2
Using Pn C Pu // and solving for Acs, the result of
Eq. 6 is the area of concrete required to resist the
compressive load.
AASHTO provides the following equation should
the compressive strut contain reinforcement parallel
to the strut and detailed in such a way to develop its
yield stress in compression. For cases where rein-
forcement is present, the nominal resistance of the
strut shall be taken as:
Pn ¼ fcu Acs þ fy Ass ð7Þ
where: Ass = area of reinforcement in the strut,
Acs = effective cross-sectional area of strut as spec-
ified in Article 5.6.3.3.2, fcu = limiting compressive
stress as specified in Article 5.6.3.3.3, fy = yield
strength of steel.
Again, using Pn C Pu // and solving for Acs, the
result of Eq. 7 is the area of concrete required to
resist the compressive load. The value of the required
Acs can then be compared to the available Acs. Based
on the calculated Acs required, a required effective
depth of the compressive strut is calculated as:
Acs
DR ¼ ð8Þ
Width Compressive Strut
where: DR = required effective depth, Acs =
required area of concrete in compression, Width of
Compressive Strut = taken as the width of the pier
cap (Oliva 1997).
The width and the depth of the nodal zone can be
taken as the width of the required bearing area
(previously calculated in the superstructure design)
(Oliva 1997). Using the smallest angle between the
compressive strut and adjoining tension ties as, the
available effective depth can be calculated as follows:
DA ¼ ðW sin as Þ þ ðD cos as Þ ð9Þ
where: DA = available effective depth, W = width
of the nodal zone, D = depth of the nodal zone.
123
Geotech Geol Eng
After dimensioning the tension ties and compres-
sion struts, the stress levels in the nodal zones must
be checked. AASHTO LRFD Specifications require
the concrete compressive stress in the nodal regions
of the strut to not exceed the following:
• For nodal regions bounded by compressive struts
and bearing areas: 0:85/ fc0
• For nodal regions anchoring a one-direction
tension tie: 0:75/ fc0
• For nodal regions anchoring tension ties in more
than one direction: 0:65/ fc0
Truss models implicitly carry both flexural and
shearing type forces through compressive and tensile
axial force elements. Therefore, it is apparent that
shear reinforcement could be omitted when employ-
ing the strut-and-tie model to reinforced concrete
members, based on how the flexural and shearing
stresses are treated by the truss model.
Kani et al. (1979) suggested that the shear
behavior of a beam depends on the ‘‘shear span’’.
The shear span is defined as the distance from the
support of the structure to the load acting on the
structure. A simply supported beam can resist higher
levels of shear at locations closer to the support. Tests
showed that for span-to-depth ratios from 1 to 2.5 the
shear is carried by strut-and-tie action; however, for a
ratio exceeding 2.5, a sectional model should be
considered to transfer the shearing stress. The find-
ings of Kani et al. (1979) further supports the ability
of the truss model to transfer the shear in D-regions
near supports and point loads.
However, it would be imprudent not to provide
shear stirrups in the design of reinforced concrete.
The shear design should be accomplished using a
sectional approach provided by AASHTO LRFD
Article 5.8.3.3. Additionally, distributed steel should
be provided in accordance with appropriate state
transportation department procedures and AASHTO
LRFD Specifications.
3 Design Comparison
3.1 Overview
For the purposes of comparing design calculations,
the Shepherd Bridge located in Wheeling, West
Geotech Geol Eng

Virginia that spans the Big Wheeling Creek River is
considered. The bridge is a replacement structure of a
former Whipple Truss bridge constructed in 1882
which remains in place due to historical preservation.
The bridge, along with the new roadway alignment,
provides improved access to the neighboring com-
munity. The structure consists of three spans totaling
61.57 m, with a superstructure (see Fig. 9) involving
three spread-box prestressed concrete beams support-
ing a 203.2 mm deck. The beams are supported at the
beginning and at the end of bridge bearings by
abutments that are integral with the deck. The
integral abutments are supported by pilings that are
embedded into bedrock. Additionally, the bridge
utilizes two hammerhead piers as intermediate sup-
ports. The piers are located in the Big Wheeling
Creek River and have an overall height of 11.61 m
that are positioned on spread footings keyed into
bedrock. The ratio of the cantilever to width of the
column is 0.5.
3.2 Original Analysis and Design
The loading for the original design was completed
using an HS-25 truck and an AASHTO Standard
Specifications lane loading. However, the larger of
the two live load reactions, the lane loading, was used
for the live load on the pier. Placing lane loads in
different locations on the superstructure generated
multiple live load cases. A total of seven live load
cases were considered that included the loading of
one and two lanes. Other loads that were considered
included dead loads on the pier cap, buoyancy, wind
loads, temperature, and water pressure.
The original analysis of the hammerhead pier was
performed using the Georgia Pier Program (1984).
The original analysis yielded seven #32 bars for the
tension reinforcing in the pier cap. Furthermore, the
original design also specified #16 shear stirrups
spaced at 171.45 mm.
3.3 Strut-and-Tie Analysis and Design
The strut-and-tie analysis and the pier design were
completed using the procedure previously defined.
After performing several iterations, a truss model
illustrated in Fig. 10 was considered as optimal and
was used for the pier cap analysis. Figure 10 also
shows the resulting forces obtained from the truss
analysis.
The truss analysis was performed using the
software program RISA 3D (2007) and checked
by manual calculations. A spreadsheet was used for
sizing the reinforcement of the tension ties and
calculating the required compression area (see
Table 1.)
As can be seen in Table 1, the required area of the
tensile steel is 7009.66 m2. Using the area of a #32
bar, the final design requires nine #32 bars providing
7,371 mm2 of tensile reinforcement. Furthermore, the
Fig. 10 Shepherd Bridge Truss Model Dimensions and
Solutions

Fig. 9 Typical section—

Shepherd Bridge

123
 Geotech Geol Eng

Table 1 Tensile reinforcement calculations and concrete compression requirements—Shepherd Bridge

Tie* Pu / Pn As Req’d Bars As Prov’d es

Required tension capacities

1 1826.82 0.70 2609.74 7009.66 9-#32 7,371 0.001282
2 1826.82 0.70 2609.74 7009.66 9-#32 7,371 0.001282
Strut* as es e1 Pu Pn Req’d Acu Req’d

Required compression capacities

1 42.879 0.00128 0.0051 2359.52 3370.74 286,651.04
2 68.962 0.00128 0.0018 574.51 820.72 43 664.43
(Pn Req’d = Pu /.7)
* Refer to Fig. 10

• Comparing DR to DA, respectively, yields1292.27

[ 235.11 mm ) no reinforcement required.
The final step in the pier cap design involves
details for shear stirrups and the provision of evenly
distributed temperature steel. The shear reinforce-
ment required by the sectional method specified #16
shear stirrups spaced at 225.25 mm.

4 Summary and Conclusions

The specific objectives of this study was to compare

Fig. 11 Effective depth calculation for Shepherd Bridge pier the reinforcing requirements of the strength design
cap approach (AASHTO LFD) for flexure and shear with
the strut-and-tie modeling method and to develop a
required area of compression concrete in the first simplified procedure for modeling a hammerhead pier
compressive strut is 286,651.04 mm2. Referring to cap by practicing engineers. The results from this strut-
Fig. 11, the calculation of the required effective and-tie model were found to be a viable analysis and
depth and the available effective depth for the first design tool in the design of hammerhead piers. The
compressive is as follows design procedure involves the placement of the loads
to induce the maximum moment in the cantilever
• Assuming a 1219.20 mm width of the compres-
section of the hammerhead pier. The design procedure
sive strut (width of pier cap) the required effective
also demonstrated the process for defining the tension
depth is
ties, compression struts, and nodal zones. In summary,
Acs the following steps were used for designing hammer-
DR ¼
Width Compressive Strut head pier caps by the strut-and-tie method:
286651:04 mm2 • Determine the reactions of the superstructure
DR ¼ ¼ 235:11 mm
1219:20 mm based on the governing load combinations.
• The available effective depth is therefore • Define all nodal zones at the beam reactions and
the cap to column reinforcing locations.
DA ¼ ðW  sin as Þ þ ðD  cos as Þ • Define the tension ties and compression struts
DA ¼ ð914:4  sin 42:879Þ þ ð914:4  cos 42:879Þ from each nodal zone and at depths equal to the
¼ 1292:27 mm approximate location of the reinforcing pattern.

123
Geotech Geol Eng
• Check truss continuity at each nodal zone.
• Solve internal truss forces for tension ties and
compression struts.
• Determine reinforcing requirements for tension
ties and check compressive strut regions.
• Check stress in nodal zones.
• Revise truss as required.
• Provide shear stirrups and distributed steel for the
hammerhead pier cap.
Although only one example was presented in this
paper, multiple three-beam bridges were analyzed to
check the methodology provided previously. As the
procedure was performed it was quiet evident that
while not dimensioned exactly, the truss models used
for all three-beam superstructures were similar. The
repeating of truss geometries can be attributed to the
process of defining the truss elements in the same
manner for each hammerhead pier cap. The signif-
icance of repeating truss geometries is that it provides
consistency to practicing engineers as it pertains to
defining a proper truss model for hammerhead piers.
While the strut-and-tie model did require more
flexural steel (29%) than the sectional design proce-
dure, its major contribution, notwithstanding this,
may be attributed to the increase in live load of
AASHTO LRFD over the live load generated by
AASHTO LFD.
References
AASHTO LFD (1996) Standard specifications, 16th edn.
American Association of State Highway and Transporta-
tion Officials, Washington DC
AASHTO LRFD (1998) Bridge design specifications, 2nd edn.
American Association of State Highway and Transporta-
tion Officials, Washington DC
Adebar P, Zhou Z (1996) Design of deep pile caps using strut-
and-tie models. ACI Struct J 93:437–448
Collins MP, Mitchell D (1991) Prestresed concrete structures.
Prentice-Hall, Inc., Englewood Cliffs 766 pp
Kani MW, Huggins MW, Wittkopp RR (1979) Kani on shear
in reinforced Concrete. Department of Civil Engineering,
Toronto
Kuchma AD, Tjhin NT (2001) CAST (Computer Aided Strut-
and-Tie) design tool. J Struct Eng
Liang QQ, Xie YM, Steven GP (2002) Performance-based
optimisation for strut-tie modeling of structural concrete.
J Struct Eng
MacGregor JG (1997) Reinforced concrete: mechanics and
design, 3rd edn. Prentice-Hall, Inc., Englewood Cliffs 939
pp
MacGregor JG (2000) ACI Committee 318E (Shear and Tor-
sion), Appendix X and CE49: Strut-and-Tie models, draft
of code and commentary to be considered for 2002 ACI
Building Code, approx. 30 pp
Oliva M (1997) Pier design presentation. University of Wis-
consin, Madison
RISA 3D (2007) RISA technologies, structural engineering
software
Schlaich J, Schäfer K, Jennewein M (1987) Toward a consis-
tent design of structural concrete. J Prestress Concr Inst
32(3):74–150
Yun YM, Ramirez JA (1996) Strength of struts and nodes in
strut-tie model. J Struct Eng 122:20–29
123