Nhi13061 4-6 PDF

printed on June 24, 2003
Lecture - 4-1
LECTURE 4 - LOADS II
4.1 OBJECTIVE OF THE LESSON
The purpose of this lesson is to continue the developments
of the loads for bridge design required by the LRFD Specification by
providing information on ice loads, earth pressures.
4.2 ICE LOADS
4.2.1 General
The Specification requires consideration of the following
types of ice action.
Dynamic pressure due to moving sheets or floes of ice being
carried by stream flow, wind or currents and striking a pier.
Static pressure due to thermal movements of ice sheets.
Static forces may be caused by the thermal expansion of ice
in which a pier is embedded, or by irregular growth of the ice
field.
Pressure resulting from hanging dams or jams of ice.
Hanging dams are the phenomenon of frazil ice passing
under the surface layer of ice and accumulating under the
surface ice at the bridge site. The frazil ice comes typically
from rapids or waterfalls upstream. The hanging dam can
cause a back-up of water, which exerts pressure on the pier,
and can also cause scour around or under the piers as the
water flows at an increased velocity.
Static uplift or vertical load resulting from adhering ice in
waters of fluctuating level.
The behavior of ice and the forces that it generates is a very
complex issue and is not yet fully understood. When undergoing
long-term changes in temperature and sustained loadings, ice can
behave in a relatively plastic manner. Ice which is moving with a
current can either create a substantial impact load on a pier or
breakup when it hits the pier reducing the load. For purposes of
classification, the commentary of the Specification uses the types
of ice failure developed by Montgomery in 1984. The reference to
Montgomery is given in the Specification. Thus, the following types
of ice failure are considered:
Lecture - 4-2
crushing - the ice fails by local crushing across the width of
the pier. The crushed ice is continually cleared from a zone
around the pier as the floe moves past,
bending - for piers with inclined noses, a vertical reaction
component acts on the impinging ice floe. This reaction
causes the floe to rise up the pier nose and fail as flexural
cracks form,
splitting - where a comparatively small floe strikes a pier,
stress cracks propagating from the pier into the floe split the
floe into smaller parts,
impact - if the floe is small, it is brought to a halt when
impinging on the nose of the pier before it has failed by
crushing over the full-width of the pier, by bending or by
splitting, and
buckling - for very wide piers, where a large floe cannot
clear the pier as it fails, compressive forces cause the floe
to fail by buckling in front of the pier nose.
4.2.2 Design for Ice
The design for ice typically starts with the determination of
the effective crushing strength of the ice (p). The following values
are specified when cite specific information is not available.
0.38 MPa where break-up occurs at melting temperatures
and the ice is substantially disintegrated in its structure,
0.77 MPa where break-up occurs at melting temperatures
and the ice is somewhat disintegrated in its structure,
1.15 MPa where break-up or major ice movement occurs at
melting temperatures, but the ice moves in large pieces and
is internally sound, and
1.53 MPa where break-up or major ice movement occurs
with the ice temperature, averaged over its depth,
measurably below the melting point.
As indicated in the commentary to the Specification, the values
identified above are considerably less than ice values which may be
measured under laboratory conditions. The lowest value is
appropriate for piers where experience with similar sites indicates
that ice forces are very low, but ice should be considered in the
design at that location. The maximum value of 1.53 MPa is based
on an observed history of bridges that have survived significant
Lecture - 4-3
icing conditions, and these are detailed in the reference list in
Section 3 of the Specification.
Once the effective ice strength, P, has been determined for
the given site, breakup conditions, the next step is to determine the
horizontal force, F, resulting from the pressure of moving ice. This
horizontal force is related to the effective strength by a series of
expressions in S3.9.2.2, which are also a function of the pier
geometry. A distinction is made as to whether the horizontal force
is caused by an ice flow failing by compression over the full-width
of the pier, or whether the ice flow fails by flexure as it rides up an
inclined ice breaking pier nose.
The design expressions require an estimate of the ice
thickness. The preferred method to determine the ice thickness is
historical records of actual ice thickness at a potential bridge site
over some number of years. When this data is not available, an
empirical expression is provided in the commentary to S3.9.2.2,
which provides a means of estimating ice thickness, depending on
the accumulation of freezing days at the site, per year, and the
particular conditions of wind and snow apt to occur at the site.
Snow cover is found to be an important factor in determining ice
thickness. The snow cover in excess of 150 mm in thickness has
been shown to reduce ice thickness by almost one-half. The
specification permits the reduction in ice forces on small streams
not conducive to the formation of large ice flows. The reduction is
limited to not more than 50% of the design force.
Once the ice force, F, has been determined, it is necessary
to consider combinations of longitudinal and transverse forces
acting on a pier. It would be unrealistic to expect ice to move
exactly parallel to a pier, so that a minimum lateral component of
15% of the longitudinal force is specified. Piers are usually aligned
in the direction of stream flow, usually assumed to be the direction
of ice movement.
Two design cases are investigated as follows:
a longitudinal force equal to F shall be combined with a
transverse force of 0.15F, or
a longitudinal force of 0.5F shall be combined with a
transverse force of F
t
.
The transverse force, F
t
, shall be taken as:
(4.2.2-1)
F
t
F
2 tan(/2
f
)
Lecture - 4-4
where:
= nose angle in a horizontal plane, for a round nose
taken as 100 (DEG)
f
= friction angle between ice and pier nose (DEG)
Both the longitudinal and transverse forces shall be assumed to act
at the pier nose.
Where the longitudinal axis of a pier is not parallel to the
principal direction of ice action, or where the direction of ice action
may shift, the total force on the pier shall be determined on the
basis of the projected pier width and resolved into components.
Under such conditions, forces transverse to the longitudinal axis of
the pier is taken to be at least 20% of the total force.
4.2.3 Static Ice Loads on Piers
Ice pressures on piers frozen into ice sheets shall be
investigated where the ice sheets are subject to significant thermal
movements relative to the pier where the growth of shore ice is on
one side only, or other situations which may produce substantial
unbalanced forces on the pier. Unfortunately, little guidance is
available for predicting static ice loads on piers. Under normal
circumstances, the effects of static ice forces on piers may be strain
limited, but expert advice should be sought if there is reason for
concern.
4.2.4 Hanging Dams and Ice Jams
The frazil accumulation in a hanging dam may be taken to
exert a pressure of 0.0096 to 0.096 MPa as it moves by the pier.
An ice jam may be taken to exert a pressure of 0.96x10
-3
to 9.6x10
-3
MPa.
The wide spread of pressures quoted reflects both the
variability of the ice and the lack of firm information on the subject.
4.2.5 Vertical Forces due to Ice Adhesion
The vertical force on a bridge pier due to rapid water level
fluctuation is given by:
for a circular pier, in N:
(4.2.5-1)
F
v
0.3t
2
0.0169Rt
1.25
Lecture - 4-5
for an oblong pier, in N/mm of pier perimeter:
(4.2.5-2)
F
v
2.3x10
3
t
1.25
where:
t = ice thickness (mm)
R = radius of circular pier (mm)
Equations 4.2.5-1 and 4.2.5-2 neglect creep and are,
therefore, conservative for fluctuations occurring over more than a
few minutes, but they are also based on the assumption that failure
occurs on the formation of the first crack, which is non-conservative.
4.2.6 Ice Accretion and Snow Loads on Superstructures
No specific ice accretion or snow loads are specified in the
LRFD Specification. However, Owners in areas where unique
accumulations of snow and/or ice are possible should specify
appropriate loads for that condition.
The following discussion of snow loads is taken from Ritter
(1991).
Snow loads should be considered where a bridge is located
in an area of potentially heavy snowfall. This can occur at high
elevations in mountainous areas with large seasonal
accumulations. Snow loads are normally negligible in areas of the
United States that are below 600 000 mm elevation and east of
longitude 105 W, or below 300 000 mm elevation and west of
longitude 105 W. In other areas of the country, snow loads as
large as 0.034 MPa may be encountered in mountainous locations.
The effects of snow are assumed to be offset by an
accompanying decrease in vehicle live load. This assumption is
valid for most structures, but is not realistic in areas where snowfall
is significant. When prolonged winter closure of a road makes snow
removal impossible, the magnitude of snow loads may exceed
those from vehicular live loads. Loads also may be notable where
plowed snow is stockpiled or otherwise allowed to accumulate. The
applicability and magnitude of snow loads are left to Designer's
judgment.
Snow loads vary from year to year and depend on the depth
and density of snow pack. The depth used for design should be
based on a mean recurrence interval or the maximum recorded
depth. Density is based on the degree of compaction. The lightest
accumulation is produced by fresh snow falling at cold
temperatures. Density increases when the snow pack is subjected
Lecture - 4-6
to freeze-thaw cycles or rain. Probable densities for several snow
pack conditions are as follows, ASCE (1980):
Table 4.2.6-1 - Snow Density
CONDITION OF
SNOW PACK
PROBABLE
DENSITY (kg/m
3
)
Freshly Fallen 96
Accumulated 300
Compacted 500
4.3 EARTH LOADS
4.3.1 General
The Specification defines several broad classifications of
walls which are also referred to herein. For reference, these
definitions are repeated below:
Abutment - A structure that supports the end of a bridge span, and
provides lateral support for fill material on which the roadway rests
immediately adjacent to the bridge.
Anchored Wall - An earth retaining system typically composed of
the same elements as non-gravity cantilevered walls, and which
derive additional lateral resistance from one or more tiers of
anchors.
Mechanically Stabilized Earth Wall - A soil retaining system,
employing either strip or grid-type, metallic or polymeric tensile
reinforcements in the soil mass, and a discrete modular precast
concrete facing which is either vertical or nearly vertical.
Non-Gravity Cantilever Wall - A soil retaining system which
derives lateral resistance through embedment of vertical wall
elements and support retained soil with facing elements. Vertical
wall elements may consist of discrete elements, e.g., piles,
caissons, drilled shafts or auger-cast piles spanned by a structural
facing, e.g., lagging, panels or shotcrete. Alternatively, the vertical
wall elements and facing may be continuous, e.g., diaphragm wall
panels, tangent piles or tangent drilled shafts.
Prefabricated Modular Wall - A soil retaining system employing
interlocking soil-filled timber, reinforced concrete or steel modules
Lecture - 4-7
or bins to resist earth pressures by acting as gravity retaining walls.
Prefabricated modular walls consist of individual structural units
assembled at the site into a series of hollow bottomless cells known
as cribs. The cribs are filled with soil, and their stability depends not
only on the weight of the units and their filling, but also on the
strength of the soil used for the filling. The units themselves may
consist of reinforced concrete, fabricated metal, or timber.
Rigid Gravity, Semi-Gravity and Cantilever Retaining Walls - A
structure that provides lateral support for a mass of soil and that
owes its stability primarily to its own weight and to the weight of any
soil located directly above its base. This classification of walls
includes:
A gravity wall depends entirely on the weight of the stone
or concrete masonry and of any soil resting on the masonry
for its stability. Only a nominal amount of steel is placed
near the exposed faces to prevent surface cracking due to
temperature changes.
A semi-gravity wall is somewhat more slender than a
gravity wall and requires reinforcement consisting of vertical
bars along the inner face and dowels continuing into the
footing. It does not rely on the weight of the overlying soil
for stability. It is provided with temperature steel near the
exposed face.
A cantilever wall consists of a concrete stem and a
concrete base slab, both of which are relatively thin and fully
reinforced to resist the moments and shears to which they
are subjected.
A counterfort wall consists of a thin concrete face slab,
usually vertical, supported at intervals on the inner side by
vertical slabs or counterforts that meet the face slab at right
angles. Both the face slab and the counterforts are
connected to a base slab, and the space above the base
slab and between the counterforts is backfilled with soil. All
the slabs are fully reinforced.
Several of these types of walls are illustrated in Figure 4.3.1-
1.
Lecture - 4-8
Figure 4.3.1-1 - Illustration of Several Wall Types (from Das, B. M.,
Principles of Foundation Engineering, Brooks/Cole Engineering
Division, 1984)
Retained earth exerts lateral pressure on retaining walls and
abutments. In general, the magnitude and distribution of the lateral
earth pressure on such structures is a function of the composition
and consistency of the retained earth and the magnitude of external
loads applied to the retained soil mass. Typically, development of
a design earth pressure considers the following:
The type, unit weight, shear strength and creep
characteristics of the retained earth;
The anticipated or permissible magnitude and
direction of lateral deflection at the top of the wall or
abutment;
The degree to which backfill soil retained by the wall
is to be compacted;
Lecture - 4-9
The location of the groundwater table within the
retained soil;
The magnitude and location of surcharge loads on
the retained earth mass; and
The effects of horizontal acceleration of the retained
earth mass during an earthquake.
The degree to which a wall (or abutment) is permitted to
deflect laterally, and the characteristics of the retained earth are the
two most significant factors in the development of lateral earth
pressure distributions. Walls which are permitted to tilt or move
laterally away from the retained soil permit the development of an
active state of stress in the retained soil mass and should be
designed for the active earth pressure. Walls which are restrained
against movement (e.g., integral abutments) or walls for which
lateral deflection and associated ground movements may adversely
impact adjacent facilities (typically within a distance behind the wall
less than about one-half the wall height) should be designed to
resist the at-rest earth pressure, which may be 50% greater than
the magnitude of the active pressure. Walls which may deflect
laterally into the retained soil should be designed to resist the
passive earth pressure, which can be 10 to 20 times greater than
the active pressure. (The passive state of stress is limited, for all
practical purposes, to lateral deflection of the embedded portions of
flexible cantilever retaining walls into the supporting soil.)
The lateral wall movement required to permit development
of the minimum active earth pressure or maximum passive earth
pressure is affected by the type of soil retained, as shown in Table
4.3.1-1
where:
= lateral movement at top of wall
(achieved through rotation or
translation) required for development
of active or passive earth pressure
(mm)
H = wall height (mm)
Lecture - 4-10
Table 4.3.1-1 - Approximate Values of Relative Movements
Required to Reach Minimum Active or Maximum Passive Earth
Pressure Conditions, Clough (1991)
Type of Backfill
Values of /H
Active Passive
Dense sand 0.001 0.01
Medium dense sand 0.002 0.02
Loose sand 0.004 0.04
Compacted silt* 0.002 0.02
Compacted lean clay* 0.010 0.05
Compacted fat clay* 0.010 0.05
*Not typically used to backfill highway structures
Nearly all conventional retaining walls of typical proportions,
except very short walls, deflect sufficiently to permit development
of active earth pressures. Gravity and semi-gravity walls designed
with a sufficient mass to support only active earth pressures will
deflect (tilt or translate) in response to more severe loading
conditions (e.g., at-rest earth pressures) until stresses in the
retained soil mass are relieved sufficiently to permit development of
an active state of stress in the retained soil. The most significant
potential for development of at-rest earth pressures on such walls
is on the stems of cantilevered retaining walls, where a rigid stem-
to-base connection may prevent lateral deflection of the stem with
respect to the base in response to the lateral pressure of backfill soil
retained above the base. For such a condition, excessive lateral
earth pressures on the stem could conceivably lead to a structural
failure of the stem or stem-to-base connection.
A comparison of estimated actual lateral deflections to
deflections required to mobilize active earth pressure on the stem
of a cantilevered retaining wall backfilled with dense sand is
provided in Tables 4.3.1-2 and 4.3.1-3. Table 4.3.1-2 assumes that
the full section modulus of the stem is effective in resisting bending,
and that the base of the stem is fixed to the foundation slab. Table
4.3.1-3 assumes that a reduced section modulus is effective in
resisting bending to account for cracking and creep of the concrete.
Neither Table 4.3.1-2 nor Table 4.3.1-3 includes lateral deflections
that would occur due to differential settlement of the wall base slab.
Lecture - 4-11
Table 4.3.1-2 Cantilever Wall Stem Deflections Using Full Section
Modulus
Estimated Stem
Deflection
Wall
Height
H
(mm)
Average
Wall Stem
Thickness
t
(mm)
Under
Active
Earth
Pressure
/H
(dim)
Under At-
Rest
Earth
Pressure
/H
(dim)
Deflection
Required to
Mobilize
Active Earth
Pressure
/H
(dim)
1500 120 0.00013 0.00021 0.001
4500 300 0.00074 0.00119 0.001
7600 490 0.00159 0.00255 0.001
9100 610 0.00151 0.00241 0.001
Note: Assumes Backfill = 37 , k
a
= 0.25, k
o
= 0.40 and f'
c
= 27.6
MPa
Table 4.3.1-3 - Cantilever Wall Stem Deflections Using Reduced
Section Modulus
Estimated Stem Deflection
Wall
Height
H
(mm)
Average
Wall Stem
Thickness
t
(mm)
Under Active
Earth
Pressure
/H
(dim)
Under At-
Rest Earth
Pressure
/H
(dim)
Deflection
Required to
Mobilize
Active Earth
Pressure
/H
(dim)
1500 120 0.00066 0.00106 0.001
4500 300 0.00371 0.00594 0.001
7600 490 0.00795 0.01273 0.001
9100 610 0.00486 0.00779 0.001
Note: Assumes Backfill = 37 , k
a
= 0.25, k
o
= 0.40 and f'
c
= 27.6
MPa
Tables 4.3.1-2 and 4.3.1-3 indicate that, with the exception
of very short walls, cantilever wall stems will generally deflect, crack
Lecture - 4-12
and creep sufficiently to permit mobilization of active earth
pressures in backfill soils composed of relatively dense sand, which
is the typical backfill material for such walls. For walls bearing on
yielding (soil) foundation materials, tilting of the base slab due to
settlement of the foundation soils will result in even greater
deflections, such that active earth pressures will be mobilized even
on the stems of very short walls. It is likely that, unless walls are
otherwise restrained against rotation or translation or have a
massive cross-section, active earth pressure conditions will be
achieved on the stems of nearly all cantilever retaining walls
backfilled with compacted granular soil, with the exception of very
short walls (less than 1.5 m tall) bearing directly on rock.
For walls retaining cohesive soils, the effects of soil creep
may prevent the permanent establishment of active and passive
earth pressure. Under stress conditions producing the minimum
active or maximum passive earth pressure, cohesive soils
continually creep, such that shear stresses within the soil mass are
partially received. As a result, the movements indicated in Table
4.3.1-1 are produced only temporarily. Without further movement,
the lateral earth pressure exerted by a cohesive soil initially in the
active stress state will increase eventually to a value approaching
the at-rest earth pressure. Likewise, the lateral earth pressure
exerted by a cohesive soil initially in the passive stress state will
decrease eventually to a value approaching approximately 40% of
the passive earth pressure.
4.3.2 Compaction
When mechanical compaction equipment is operated within
a distance behind a retaining wall equal to about one-half of the wall
height, additional lateral earth pressures are induced on the wall
due to the compaction effort. Excessive backfill compaction can
increase lateral earth pressures to values significantly greater than
the active or even at-rest lateral earth pressure. Such compaction-
induced pressures continue to act, even after the compaction
equipment has been removed due to the inelastic behavior of the
soil.
A typical lateral earth pressure distribution for an unyielding
wall, including compaction-induced residual pressures, is shown in
Figure 4.3.2-1.
Lecture - 4-13
Figure 4.3.2-1 - Residual Earth Pressure after Compaction of
Backfill Behind an Unyielding Wall (after Clough and Duncan, 1991)
The induced residual pressures would be somewhat less on
a flexible or unrestrained wall subjected to the same compaction
loading conditions since lateral deflection or movement of the wall
would permit partial relief of the stress in the retained soil.
The heavier the compaction equipment and the closer it
operates to the wall, the greater are the compaction-induced
pressures. Therefore, the use of soils which are difficult to compact
(e.g., fine-grained, moisture sensitive soils) and heavy compaction
equipment immediately behind earth retaining structures is likely to
cause unacceptably large lateral soil pressures and should be
avoided.Use of free-draining granular soils and light compaction
equipment within a distance of H/2 behind retaining walls is usually
specified to preclude development of excessive compaction-
induced lateral earth pressures. If the use of heavy static or
vibratory compaction equipment behind a retaining wall cannot be
avoided, the residual compaction-induced lateral earth pressures
should be estimated by available procedures (e.g., Clough and
Duncan, 1991).
Lecture - 4-14
4.3.3 Earth Pressure
As described in Article 4.3.1, the magnitude and distribution
of lateral earth pressure on a retaining structure is primarily a
function of the retained soil characteristics and the degree to which
the wall tilts or translates in response to the loading. For most
abutments and conventional retaining walls, the earth pressure
distribution is assumed to increase linearly with depth in accordance
with the following:
p = k
h
g z 10
-9
(4.3.3-1)
where:
= unit density of soil (kg/m
3
)
p = lateral earth pressure (MPa)
k
h
= lateral earth pressure coefficient taken as k
a
or k
o
,
depending on the magnitude of lateral deflection
(dim) (see Article 4.3.1)
z = depth below backfill surface (mm)
g = gravitational constant (m/sec
2
)
Although the lateral earth pressure due to the retained soil
is assumed to increase linearly with depth, the resultant lateral load
due to the earth pressure is assumed to act at a height of 0.4H
above the base of the wall for conventional gravity retaining walls
(where H is the total wall height measured from the top of the
backfill to the base of the footing) rather than 0.33H, as would be
expected for a linearly proportional (triangular) distribution. As a
conventional gravity wall deflects laterally (translates) in response
to lateral earth loading, the backfill behind the wall must slide down
along the back of the wall for the retained soil mass to achieve the
active state of stress. Experimental results indicate that the backfill
arches against the upper portion of the wall as the wall translates,
causing an upward shift in the location at which the resultant of the
lateral earth load is transferred to the wall (Terzaghi, 1934; Clausen,
1972, and Sherif, 1982)
For non-gravity cantilever retaining walls or other flexible
walls which tilt or deform laterally in response to lateral loading,
significant arching of the backfill against the wall does not occur,
and the resultant lateral load due to earth pressure is assumed to
act at a height of 0.33H above the base of the wall.
Lecture - 4-15
4.3.3.1 AT-REST PRESSURE COEFFICIENT, k
o
When a retaining wall is restrained against lateral movement
or lateral movement of the wall is unacceptable, the lateral earth
pressure coefficient (k
h
) in Equation 4.3.3.1-1 is taken as k
o
.
For a normally consolidated soil, the at-rest lateral earth
pressure coefficient, k
o
, can be computed by the following:
k
o
= 1 - sin
f
(4.3.3.1-1)
where:
f
= effective stress angle of internal friction of the
drained soil
For overconsolidated soils, the at-rest lateral earth pressure
coefficient is generally considered to vary as a function of the stress
history, the value of k
o
increasing with increasing degree of
overconsolidation in accordance with the following:
k
o
= (1 - sin
f
)(OCR)
sin
f
(4.3.3.1-2)
where:
OCR = overconsolidation ratio (dim)
Values of k
o
for various soil types and degrees of
overconsolidation are presented in Table 4.3.3.1-1.
Lecture - 4-16
Table 4.3.3.1 - Typical Coefficients of At-Rest Lateral Earth
Pressure
Soil Type
Coefficient of Lateral Earth Pressure, k
o
OCR = 1 OCR = 2 OCR = 5 OCR = 10
Loose Sand 0.45 0.65 1.10 1.60
Medium
Sand
0.40 0.60 1.05 1.55
Dense Sand 0.35 0.55 1.00 1.50
Silt (ML) 0.50 0.70 1.10 1.60
Lean Clay
(CL)
0.60 0.80 1.20 1.65
Highly
Plastic Clay
(CH)
0.65 0.80 1.10 1.40
4.3.3.2 ACTIVE PRESSURE COEFFICIENT, k
a
When a retaining wall deflects laterally in response to
loading by the retained earth, a wedge of the retained soil moves
laterally and downward along the back of the wall, as shown in
Figure 4.3.3.2-1.
Lecture - 4-17
Figure 4.3.3.2-1 - Active Failure Wedge for Conventional Gravity
and Cantilever Retaining Walls, Coulomb Analysis
As the soil wedge moves, the shear strength of the soil is
gradually mobilized along the failure plane shown in Figure 4.3.3.2-
1. When the full shear strength of the soil is mobilized, the
additional force required to maintain the stability of the wedge (and
the corresponding force acting on the wall) reaches a minimum
value equal to the active earth pressure, P
a
. Prior to wall movement
and mobilization of shear strength in the soil, the wall must support
the at-rest earth pressure (Article 4.3.3.1).
One of two theories is generally used to estimate the active
earth pressure on retaining walls. Rankine earth pressure theory
neglects the vertical friction force applied to the surface of the wall
by the retained soil wedge as it moves downward along the back of
the wall. For the Rankine earth pressure theory, the active earth
pressure resultant is assumed to have a line of action parallel to the
backfill surface. Coulomb earth pressure theory accounts for the
friction force exerted on the wall by the retained earth, which results
in an inclination of the earth pressure resultant of with respect to
the back face of the wall (for a gravity wall) or to the vertical
pressure surface extending up from the heel of the wall (for a
cantilever wall), as shown in Figure 4.3.3.2-1. Typical values of ,
Lecture - 4-18
the friction angle between the wall and backfill, are presented in
Table 4.3.3.2-1.
Table 4.3.3.2-1 - Friction Angles Between Dissimilar
Materials
Interface Materials Friction
Angle,
(deg)
Mass concrete on the following foundation
materials:
Clean sound rock
Clean gravel, gravel-sand mixtures,
coarse sand
Clean fine to medium sand, silty
medium to coarse sand,
silty or clayey gravel
Clean fine sand, silt or clayey fine to
medium sand
Fine sandy silty, non-plastic silt
Very stiff and hard residual or
preconsolidated clay
Medium stiff and stiff clay and silty
clay
Masonry on foundation materials has same
friction factors
35
29 to 31
24 to 29
19 to 24
17 to 19
22 to 26
17 to 19
Steel sheet piles against the following soils:
well-graded rock fill with spalls
Clean sand, silty sand-gravel mixtures,
single-size hard rock fill
Silty, sand, gravel, or sand mixed with
silty or clay
Fine sandy silt, non-plastic silt
22
17
14
11
Interface Materials Friction
Angle,
(deg)
Lecture - 4-19
Formed or precast concrete or concrete
sheet piling against the following soils:
well-graded rock fill with spalls
Clean sand, silty sand-gravel mixtures,
single-size hard rock fill
Silty sand, gravel or sand mixture with
silt or clay
Find sandy silt, non-plastic silt
22 to 26
17 to 22
17
14
Various structural materials:
Masonry on masonry, igneous and
metamorphic rocks:
dressed soft rock on dressed soft
rock
dressed hard rock on dressed soft
rock
dressed hard rock on dressed hard
rock
Masonry on wood in direction of cross
grain
Steel on steel at sheet pile interlocks
35
33
29
26
17
Both AASHTO Standard Specification (AASHTO 1992) and
the LRFD Specification (AASHTO 1993) employ the Coulomb earth
pressure theory. For the typical case when a retaining wall is
permitted to deflect sufficiently to develop the active state of stress
in the retained soil, the lateral earth pressure coefficient (k
h
) in
Equation 4.8-1 is, therefore, taken as the Coulomb active earth
pressure coefficient, k
a
. For the case of a vertical retaining wall and
a horizontal backfill surface, the value of k
a
can be obtained from
Figure 4.3.3.2-2.
For theoretical solutions like those shown in Table 1 and
Figure 1, the angle of internal friction is denoted simply as . The
value of , shown in these solutions, is to be interpreted as the
effective stress friction angle,
f
, determined from a drained shear
test, when analyses are performed using effective stresses, and the
total stress friction angle , determined from an undrained shear
test, when analyses are performed using total stresses. For long-
term conditions, the earth pressures should be calculated using
effective stresses, and adding water pressures as appropriate.
Lecture - 4-20
Figure 4.3.3.2-2 - Active and Passive Pressure Coefficients for
Vertical Wall and Horizontal Backfill - Based on Log Spiral Failure
Surfaces
For the more general case of an inclined wall face and
sloping backfill surface, the value of the k
a
can be obtained from
Table 4.3.3.2-2.
Lecture - 4-21
Table 4.3.3.2-2 - Value of k
a
for Log Spiral Failure Surface
(DEG)
i
(DEG)

(DEG)
(DEG)
20 25 30 35 40 45
-15
-10
0
10
0.37
0.42
0.45
0.30
0.35
0.39
0.24
0.29
0.34
0.19
0.24
0.29
0.14
0.19
0.24
0.11
0.16
0.21
0 0
-10
0
10
0.42
0.49
0.55
0.34
0.41
0.47
0.27
0.33
0.40
0.21
0.27
0.34
0.16
0.22
0.28
0.12
0.17
0.24
15
-10
0
10
0.55
0.65
0.75
0.41
0.51
0.60
0.32
0.41
0.49
0.23
0.32
0.41
0.17
0.25
0.34
0.13
0.20
0.28
-15
-10
0
10
0.31
0.37
0.41
0.26
0.31
0.36
0.21
0.26
0.31
0.17
0.23
0.27
0.14
0.19
0.25
0.11
0.17
0.23
0
-10
0
10
0.37
0.44
0.50
0.30
0.37
0.43
0.24
0.30
0.38
0.19
0.26
0.33
0.15
0.22
0.30
0.12
0.19
0.26
15
-10
0
10
0.50
0.61
0.72
0.37
0.48
0.58
0.29
0.37
0.46
0.22
0.32
0.42
0.17
0.25
0.35
0.14
0.21
0.31
Studies have shown that the failure surface defining the soil
wedge loading the wall is approximated more closely by a log spiral
curve than a straight line. The values of k
a
, provided in Figure
4.3.3.2-1 and Table 4.3.3.2-2, were, therefore, obtained from
analyses using log spiral surfaces (Caquot and Kerisel, 1948).
Both the AASHTO Standard and LRFD Specifications also
provide guidance for estimating earth pressures on special types of
earth retaining structures (e.g., non-gravity cantilevered walls,
anchored walls and mechanically stabilized earth walls) and walls
subjected to unusual loading conditions (e.g., passive earth
pressures). Where unusual backfill geometries or surcharge
conditions exist, the active pressure may be estimated using a
graphical trial wedge procedure.
Lecture - 4-22
4.3.3.3 EQUIVALENT FLUID PRESSURE
For simplicity, lateral earth pressure is often estimated as an
equivalent fluid pressure, wherein the resultant of the earth
pressure is equivalent to the resultant of a fictitious fluid exerting
hydrostatic pressure on the wall. Where equivalent fluid pressure
is used, the unit earth pressure (in MPa) at any depth is taken as:
p =
eq
g Z 10
-9
(4.3.3.3-1)
where:
eq
= equivalent fluid unit density of soil, not less than 480 (kg/m
3
)
Typical values of
eq
for design of walls up to 6.5 m in height
are shown in Table 4.3.3.3-1.
Table 4.3.3.3-1 - Typical Values for Equivalent Fluid Unit Densities
of Soil
Type of Soil
Level Backfill
Backfill with i = 25
At-Rest
eq
(kg/m
3
)
Active
(/H =
1/240)
eq
(kg/m
3
)
At-Rest
eq
(kg/m
3
)
Active
(/H =
1/240)
eq
(kg/m
3
)
Loose sand
or gravel
880 640 1040 800
Medium
dense sand
or gravel
800 560 960 720
Dense sand
or gravel
720 480 880 640
Compacted
silt (ML)
960 640 1120 800
Compacted
lean clay
(CL)
1120 720 1280 880
Compacted
fat clay
(CH)
1280 880 1440 1040
The resultant lateral earth load due to the equivalent fluid
pressure is assumed to act at a height of 0.4H above the base of
Lecture - 4-23
the wall footing for conventional gravity and semi-gravity retaining
walls.
4.3.4 Presence of Water
If soil mass retained by a retaining wall or abutment contains
groundwater and the groundwater is not eliminated through
drainage, a water table will develop behind and exert lateral
pressure on the structure above the water table, the horizontal
pressure is given as:
P = k
h
g z 10
-9
(4.3.4-1)
The presence of the water table behind the wall has two additional
effects, as indicated below and in Figure 4.3.4-1.
The unit weight of the retained soil is reduced to its
submerged or buoyant value:
s
=
s
-
w
(4.3.4-2)
As a result, the lateral earth pressure below the water table
is reduced to:
P = (k
h

s
(z-z
w
) + k
h
'
s
z) g 10
-9
(4.3.4-3)
The retained water exerts a horizontal hydrostatic
water pressure equal to:
P
w
=
w
g z
w
10
-9
(4.3.4-4)
where:
s
= submerged unit density of soil (kg/m
3
)
s
= total unit density of soil (kg/m
3
)
w
= unit density of water (kg/m
3
)
k
h
= horizontal earth pressure coefficient (dim)
z = depth below backfill surface (mm)
z
w
= depth below water table (mm)
P = horizontal earth pressure (MPa)
P
w
= hydrostatic water pressure (MPa)
Lecture - 4-24
Figure 4.3.4-1 - Effect of Groundwater Table on Earth Pressure
With some algebra, the equations for pressure below the
groundwater surface can be rearranged for ease of computation as:
P = [k
h
z +
w
z
w
- k
h

w
z
w
] g x 10
-9
(4.3.4-5)
The location of the resultant corresponding terms containing
"k
h
" is taken at 0.4 z or 0.4 z
w
above the design section defined by
z and z
w
. The location of the resultant of the term
w
z
w
is taken as
z
w
/3 above the design section.
Where possible, the development of hydrostatic water
pressures should be prevented through the use of free-draining
granular backfill and/or by providing a positive means of backfill
drainage, such as weep holes, perforated and solid pipe drains,
gravel drains or geofabric drains.
When groundwater levels differ on opposite sides of a
retaining wall, seepage occurs beneath the wall. The effect of
seepage forces is to increase the load on the back of the wall (and
decrease any passive resistance in front of the wall). Pore
pressures in the backfill soil can be approximated through
development of a flow net or using various analytical methods, and
must be added to the effective horizontal earth pressures to
determine the total lateral pressures on the wall.
4.3.5 Surcharge
Surcharge loads on the retained earth surface produce
additional lateral earth pressure on retaining walls. Where the
surcharge is uniform over the retained earth surface, the additional
lateral earth pressure due to the surcharge is assumed to remain
constant with depth and has a magnitude, p, of:
p = k
s
q
s
(4.3.5-1)
where:
Lecture - 4-25
p = constant horizontal earth pressure due to uniform
surcharge (MPa)
k
s
= coefficient of earth pressure due to surcharge
q
s
= uniform surcharge applied to the upper surface of the
active earth wedge (MPa)
For active earth pressure conditions, k
s
is taken as k
a
, and
for at-rest conditions, k
s
is taken as k
o
.
Where vehicular traffic is anticipated within a distance
behind a wall equal to about the wall height, a live load surcharge
is assumed to act on the retained earth surface. The uniform
increase in horizontal earth pressure due to live load surcharge is
typically estimated as:
p = k
s

s
g h
eq
10
-9
(4.3.5-2)
where:
p = constant horizontal earth pressure due to uniform
surcharge (MPa)
s
3
)
k = coefficient of earth pressure
h
eq
= equivalent height of soil for the design live load (mm)
Equivalent heights of soil, h
eq
, for highway loading on
retaining walls of various heights can be taken from Table 4.3.5-1.
Table 4.3.5-1 - Equivalent Height
of Soil for Vehicular Loading
Wall Height (mm) h
eq
(mm)
1500 1700
3000 1200
6000 760
9000 610
The tabulated values of h
eq
were determined based on
evaluation of horizontal pressure distributions produced on retaining
walls by the vehicular live loads specified in the LRFD Specification
(AASHTO 1993). The pressure distributions were obtained from
Lecture - 4-26
elastic half-space solutions with Poisson's ratio of 0.5, doubled to
account for the non-deflecting wall.
Alternatively, the increase in horizontal earth pressure, p,
on a retaining wall resulting from a live load surcharge, p, can be
taken as:
(4.3.5-3)
p
2p
( sincos( 2))
where:
p = live load intensity (MPa)
= angle illustrated in Figure 4.3.5.1 (RAD)
= angle illustrated in Figure 4.3.5.1 (RAD)
Figure 4.3.5-1 - Horizontal Pressure on Wall Caused by Uniformly
Loaded Strip
4.3.6 Effect of Earthquake
Lateral earth pressures on retaining structures are amplified
during an earthquake due to horizontal acceleration of the retained
earth mass. The Mononobe-Okabe method of analysis is a
pseudo-static method which is commonly used to develop an
Lecture - 4-27
equivalent static fluid pressure to model seismic earth pressure on
retaining walls. The Mononobe-Okabe method is contingent upon
the following assumptions (Barker, et al, 1991):
The wall is unrestrained and capable of deflecting
sufficiently to mobilize the active earth pressure
condition in the backfill;
The backfill is cohesionless and unsaturated;
The failure surface defining the active wedge of soil
loading the wall is planar; and
Accelerations are uniform throughout the soil mass.
As indicated above, the Mononobe-Okabe method assumes
that backfill soils are unsaturated and, as such, not susceptible to
liquefaction. The potential for liquefaction should be evaluated
where saturated soils may be subjected to earthquake or other
cyclical or instantaneous dynamic loadings.
The Mononobe-Okabe method (AASHTO 1992 and
AASHTO 1993) considers equilibrium of the soil wedge retained by
the wall as shown in Figure 4.3.6-1.
Figure 4.3.6-1 - Definition Sketch for Seismic Loading (after Barker,
1991)
The result of the combined active static and seismic earth
pressures is taken as:
P
AE
= 0.5 x K
AE
g (1-k
v
) H
2
10
-9
(4.3.6-1)
where the seismic active earth pressure coefficient, K
AE
, is:
Lecture - 4-28
s
2
( )
2
cos( )
1
sin( )si
cos( )
(4.3.6-2)
where:
3
)
H = height of soil face (mm)
= angle of internal of soil (deg)
= arc tan (k
h
/(1-k
v
))
= angle of friction between soil and wall (deg) (refer to
Article 4.3.6.2)
k
h
= horizontal acceleration coefficient (dim)
k
v
= vertical acceleration coefficient (dim)
= slope of backfill surface (deg)
= slope of wall back face (deg)
g = gravitational constant (m/sec
2
)
For estimation of lateral earth pressures under earthquake
conditions, the vertical acceleration coefficient, k
v
, is commonly
assumed equal to zero, and the horizontal acceleration coefficient,
k
h
, is taken as:
k
h
= 0.5 (4.3.6-3)
for walls designed to move horizontally 250 (mm), and
k
h
= 1.5 (4.3.6-4)
for walls designed for zero horizontal displacement
where:
= A/100
A = horizontal earthquake acceleration (percent of g)
g = acceleration of gravity (m/sec
2
)
Lecture - 4-29
Values of the earthquake acceleration coefficient, A, are
shown in Figures S3.10.2-1, S3.10.2-2 and S3.10.2-3.
Although the Mononobe-Okabe method of analysis does not
specify the point of application of the horizontal seismic earth
pressure resultant, the resultant of the dynamic component of the
earth pressure (P
AE
) is significantly higher on the wall than the
static active earth pressure resultant (P
a
), as indicated in Figure
4.3.6-1. Both the AASHTO Standard Specifications (AASHTO
1992) and the LRFD Specifications (AASHTO 1993) indicate that
the combined static and seismic lateral earth pressure can be
assumed to be uniformly distributed with a resultant (P
AE
) acting at
the mid-height of the wall.
The AASHTO Standard Specification and the LRFD
Specification also provide guidelines for determination of passive
seismic earth pressure on retaining structures which are being
forced horizontally into the backfill, and methods for design of
retaining structures for a limited tolerable displacement under
earthquake loading rather than for zero permanent displacement as
assumed in the Mononobe-Okabe method.
As with many current methods of seismic analysis, the
Mononobe-Okabe method neglects inertial forces due to the mass
of the retaining structure, concentrating only on the inertia of the
retained soils mass. For gravity structures which rely solely on their
mass for stability, this assumption is unconservative. The effects
of wall inertia on the behavior of gravity retaining walls is discussed
further by Richards and Elms (1979), who conclude that the
structure inertia forces should not be neglected. Richards and Elms
suggest a design approach based on limiting wall displacements to
tolerable levels, rather than designing for no movement, and
computing the weight of wall required to prevent movements
greater than specified. The work of Richards and Elms is
incorporated into the tolerable displacement procedure presented
in the AASHTO Specifications.
4.3.7 Reduction due to Earth Pressure
In some cases, lateral earth pressures may reduce the
effects of other loads and forces on culverts, bridges and their
components. One such case is that of the top slab of a box culvert,
for which the maximum bending moment in the top slab is reduced
due to the effect of lateral earth pressure on the side walls. Such
reductions in loading should be limited to the extent that the applied
earth pressures can be expected to be permanent. In lieu of more
precise methods of estimating the lateral earth pressure force
effects, the AASHTO Standard Specification (AASHTO 1992) and
the LRFD Specification (AASHTO 1993) permit a load effect
reduction of 50% where earth pressures are present.
Lecture - 4-30
4.3.8 Downdrag
When a point bearing pile or drilled shaft penetrates a soft
layer subject to settlement (e.g., where an overlying embankment
may cause settlement of the layer), the soil settling around the shaft
exerts a frictional force, or downdrag, around the perimeter of the
pile or shaft. This frictional force acts as an additional axial load on
the pile or shaft and, if sufficient in magnitude, could cause a
structural failure of the foundation element or a bearing capacity
failure at the tip.
The methods used to estimate downdrag loads are the same
as those used to estimate the side resistance of shafts and piles
due to skin friction, as the same load transfer mechanism is
responsible for both.
4.3.9 Design of a Cantilever Retaining Wall
The purpose of this example is to illustrate the application of
the various load factors and load combinations. In order to fully
develop the example, references to provisions for wall and footings,
and references to geotechnical textbooks are required and so
noted.
Some rounding of numbers has been done so it may not be
possible to exactly duplicate all values to the full precision shown.
The cantilever retaining wall below has been proposed to
support an embankment.
Lecture - 4-31
Figure 4.3.9-1 - Schematic of Example
During the subsurface exploration, it was determined that the
foundation soils are predominantly clay to a depth of 6 m below the
proposed bottom of footing, and, therefore, a 150 mm blanket of
compacted granular material was placed below the footing. Dense
sand and gravel underlies the clayey foundation soils. Assume
elastic settlement of the dense sand and gravel to be negligible.
The proposed wall backfill will consist of a free draining granular fill.
Assume the seasonal high water table to be at the bottom of the
footing.
Apply the vehicular live load surcharge (LS) on the backfill
as indicated in the figure above.
Determine the lateral pressure distribution on the wall and
estimate the bearing capacity for the proposed design. Check the
design for adequate protection against sliding and estimate the
consolidation settlement of the underlying clay.
Solution:
Step 1: Calculate the Unfactored Loads with = 1.0
(A) Dead Load of Structural Components and Nonstructural
Attachments (DC)
Unit Density of Concrete = 2,400 kg/m
3
Lecture - 4-32
Figure 4.3.9-2 - Retaining Wall Area Designation
Weight of Concrete
W1 = (0.3)(4.5)(2,400)g = 31,784 N per m of length
W2 = (0.5)(4.5)(0.2)(2,400)g = 10,595 N per m of length
W3 = (0.5)(3.0)(2,400)g = 35,316 N per m of length
DC = Sum of W1 - W3 = 77,695 N per m of length
(B) Vertical Earth Pressure (EV)
Weight of Soil
P
EV
= W = (2.0)(4.5)(1,920)g = 169,517 N per m of length
(C) Live Load Surcharge (LS)
The live load surcharge shall be applied where vehicular
load is expected to act on the backfill within a distance equal to the
wall height behind the wall (S3.11.6.2).
An equivalent height of soil for the design vehicular loading
(heq) is estimated using Table 4.3.9-1, repeated below, and a wall
height of 5 m.
Lecture - 4-33
Table 4.3.9-1 - Equivalent Height
of Soil for Vehicular Loading
Wall Height
(mm)
h
eq
(mm)
1500 01700
3000 1200
6000 760
h
eq
= 907 mm
Use soil density of the backfill = 1920 kg/m
3
Vertical Component of LS = (1920)(907)(9.81) x 10
-9
m
3
/mm
3
=
0.0171 MPa
For a width of 2m, P
LSV
= 34,167 N per m of length
Assume an active earth pressure coefficient k
a
using Figure
4.3.9-3 with a wall friction angle, = , repeated below.
Lecture - 4-34
Figure 4.3.9-3 - Active and Passive Pressure Coefficients for
Vertical Wall and Horizontal Backfill - Based on Log Spiral Failure
Surfaces
k = k
a
= 0.29
using Equation S3.11.6.2-1:
Lecture - 4-35
p
= (k) ('
1
) (g) (h
eq
) (10
-9
)
p
= 0.0050 MPa
Using a rectangular distribution, the live load horizontal earth
pressure acting over the entire wall will be:
P
LSH
= (0.0050)(5)(10
6
) = 24,771 N per m of length
(D) Horizontal Earth Pressure (EH)
The basic earth pressure should be assumed to be linearly
proportional to the depth of earth and is given by Equation
S3.11.5.1-1:
p = k('
1
)gz(10
-9
)
Use k = k
a
(as above)
At the base of the footing:
p = (0.29)(1,920)(9.81)(5000)(10
-9
)
p = 0.0273 MPa
The horizontal earth pressure acting over the entire wall
will be:
P
EH
= (0.5)(0.0273)(5)(10
6
) = 68,278 N per m of length
Note triangular pressure distribution
(E) Summary of Unfactored Loads
Table 4.3.9-2 - Vertical Loads and Resisting Moments - Unfactored
Item
V
N/m Arm about 0 m
Moment about 0
N-m/m
W1 31,784 0.85 27,017
W2 10,595 0.63 6,710
W3 35,316 1.5 52,974
P
EV
169,517 2.0 339,034
P
LSV
34,167 2.0 68,334
TOTAL 281,379
Lecture - 4-36
Table 4.3.9-3 - Horizontal Loads and Overturning Moments
Moment - Unfactored
Item
H
N/m Arm about 0 m
Moment about
0 N-m/m
P
LSH
24,771 2.5 61,928
P
EH
68,278 2.0 136,555
For location of resultant for lateral earth pressure see
S3.11.5.1.
Step 2: Determine the Appropriate Load Factors
Using Tables S3.4.1-1 and S3.4.1-2 given in this manual as Tables
2.4.1.2-1 and 2.4.1.2-2, respectively:
Table 4.3.9-4 - Load Factors
Group DC EV LSv P
LSH
EH
(active) Probable Use
Strength
I-a
0.9 1.0 1.75 1.75 1.5 Bearing Capacity
(eccent.)&Sliding
Strength
I-b
1.25 1.35 1.75 1.75 1.5 Bearing Capacity
(max. value)
Strength
IV
1.5 1.35 - - 1.5 Bearing Capacity
(max. value)
Service I 1.0 1.0 1.0 1.0 1.0 Settlement
By inspection, the following conclusions can be drawn for the case
of bearing capacity (maximum) Strength I-b will probably govern,
however, the factored loads will also be checked for Strength IV.
Lecture - 4-37
Step 3: Calculate the Factored Loads
Table 4.3.9-5 - Factored Vertical Loads
Group/
Item
Units
W1
N/m
W2
N/m
W3
N/m
P
EV
N/m
P
LSV
N/m
Total
N/m
V (Unf.) 31,784 10,595 35,316 169,517 34,167 281,379
Strength
I-a
28,606 9,535 31,784 169,517 59,792 299,235
Strength
I-b
39,731 13,244 44,145 228,848 59,792 385,759
Strength
IV
47,677 15,892 52,974 228,848 0 345,390
Service I 31,784 10,595 35,316 169,517 34,167 281,379
Table 4.3.9-6 - Factored Moments M
v
Group/
Item
Units
W1
N-m/m
W2
N-m/m
W3
N-m/m
P
EV
N-m/m
P
LSV
N-m/m
Total
N-m/m
M
v
(Unf.)
27,017 6,710 52,974 339,034 68,334 494,068
Strength
I-a
24,315 6,039 47,677 339,034 119,585 536,649
Strength
I-b
33,771 8,388 66,218 457,695 119,585 685,656
Strength
IV
40,525 10,065 79,461 457,695 0 587,747
Service I 27,017 6,710 52,974 339,034 68,334 494,068
Lecture - 4-38
Table 4.3.9-7 - Factored Horizontal Loads
Group/Item
Units
P
LSH
N/m
P
EH
N/m
Total
N/m
H (Unf.) 24,771 68,278 93,049
Strength I-a 43,349 102,416 145,766
Strength I-b 43,349 102,416 145,766
Strength IV 0 102,416 102,416
Service I 24,771 68,278 93,049
Table 4.3.9-8 - Factored Moments M
h
Group/Item
Units
P
LSH
N-m/m
P
EH
N-m/m
Total
N-m/m
M
v
(Unf.) 61,928 136,555 198,483
Strength I-a 108,374 204,833 313,206
Strength I-b 108,374 204,833 313,206
Strength IV 0 204,833 204,833
Service I 61,928 136,555 198,483
This example will be continued in Lecture 15 in which a
foundation design, eccentricity check, sliding check, and settlement
calculation will be added.
Lecture - 4-39
REFERENCES
Clough, G.W., and J.M. Duncan, 1991, "Earth Pressures," Chapter
6, Foundation Engineering Handbook, 2nd Edition, edited by H.Y.
Fang, Van Nostrand Reinhold, New York, NY.
Clausen, C.J.F. and S. Johansen, 1972, "Earth Pressures
Measured Against a Section of a Basement Wall," Proceedings, 5th
European Conference on Soil Mechanics and Foundation
Engineering, Madrid, pp. 515-516.
Sherif, M.A., I. Ishibashi and C.D. Lee, 1982, "Earth Pressures
Against Rigid Retaining Walls," Journal of Geotechnical
Engineering, ASCE, Vol. 108, GT5, pp. 679-695.
Terzaghi, K., 1934, "Retaining Wall Design for Fifteen-Mile Falls
Dam, Engineering News Record, May, pp. 632-636.
Barker, R.M., J.M. Duncan, K.B. Rojiani, P.S.K. Ooi, C.K. Tan and
S.G. Kim, 1991, "Manuals for the Design of Bridge Foundations,"
NCHRP Report 343, Transportation Research Board, Washington,
D.C.
American Association of State Highway and Transportation Officials
(AASHTO), 1993, "Draft LRFD Bridge Design Specifications,"
prepared by Modjeski and Masters, Inc. Consulting Engineers, Inc.
American Association of State Highway and Transportation
Officials, 1992, "Standard Specifications for Highway Bridges,"
Fifteenth Edition, AASHTO, Washington, D.C., 1992.
Richards, R., and D.G., Elms, 1979, "Seismic Behavior of Gravity
Retaining Walls," Journal of the Geotechnical Engineering Division,
ASCE, Volume 105, No. GT4.
Caquot, A. and J. Kerisel, 1948, "Tables for the Calculation of
Passive Pressure, Active Pressure and Bearing Capacity of
Foundations," Gauthier-Villars, Imprimeur-Libraire, Libraire du
Bureau des Longitudes, de L'Ecole Polytechnique, Paris, 120 pp.
Lecture - 5-1
LECTURE 5 - LOADS III
The objective of this lesson is to continue to provide a student
with the background for the provisions in Section 3, Loads and Load
Factors, of the AASHTO LRFD Specification.
The lesson includes:
the forces due to superimposed deformations resulting from
uniform and nonuniform temperature changes, differential
settlement, creep and shrinkage,
discussion of braking and centrifugal forces,
discussion of wind loads, and
discussion of water loads.
5.2 FORCE EFFECTS DUE TO SUPERIMPOSED DEFORMATIONS
Internal force effects in a component due to creep and
shrinkage shall be considered. The effect of temperature gradient
should be included where appropriate. Force effects resulting from
resisting component deformation, displacement of points of load
application and support movements is included in the analysis.
5.2.1 Uniform Temperature
In the absence of more precise information, the ranges of
temperature shall be as specified in Table 5.2.1-1. The difference
between the extended lower or upper boundary and the base
construction temperature assumed in the design is used to calculate
thermal deformation effects.
Table 5.2.1-1 - Temperature Ranges
CLIMATE
STEEL OR
ALUMINUM CONCRETE WOOD
Moderate -18 to 50 C -12 to 27 C -12 to 24 C
Cold -35 to 50 C -18 to 27 C -18 to 24 C
A moderate climate may be determined by the number of
freezing days per year. If the number of freezing days is less than 14,
the climate is considered to be moderate. Freezing days are days
when the average temperature is less than 0 C.
Lecture - 5-2
The actual air temperature averaged over the 24 hour period
immediately preceding the setting event can be used in installing
expansion bearings and deck joints.
5.2.2 Temperature Gradient
The load factor for temperature gradient should be determined
based on:
the type of structure, and
the limit state being investigated.
There is general agreement that in situ measurements of
temperature gradients have yielded a realistic distribution of
temperatures through the depth of some types of bridges, most
notably concrete box girders. There is very little agreement on the
significance of the effect of that distribution. It is generally
acknowledged that cracking, yielding, creep and other non-linear
responses diminish the effects. Therefore, load factors of less than
1.0 should be considered, and there is some basis for lower load
factors at the strength and extreme event limit states than at the
service limit state.
Similarly, open girder construction and multiple steel box
girders have traditionally, but perhaps not necessarily correctly, been
designed without consideration of temperature gradient, i.e.,
TG
= 0.0.
Temperature gradient is included in various load combinations
in Table S3.4.1-1. This does not mean that it need be investigated for
all types of structures. If experience has shown that neglecting
temperature gradient in the design of a given type of structure has not
lead to structural distress, the Owner may choose to exclude
temperature gradient. Multi-beam bridges are an example of a type
of structure for which judgment and past experience should be
considered.
The vertical temperature gradient in concrete and steel
superstructures with concrete decks may be taken as shown in Figure
5.2.2-1.
The dimension "A" in Figure 5.2.2-1 shall be taken as:
300 mm for concrete superstructures, that are 400 mm or
more in depth
for concrete sections shallower than 400 mm, "A" shall be 100
mm less than the actual depth
300 mm for steel superstructures, where t = depth of the
concrete deck
Lecture - 5-3
Temperature value T
3
shall be taken as 0 C, unless a site-
specific study is made to determine an appropriate value, but shall not
exceed 3 C.
Figure 5.2.2-1 - Positive Vertical Temperature Gradient in Concrete
and Steel Superstructures
This temperature gradient given herein is a modification of that
proposed in Imbsen (1985) which was based on studies of concrete
superstructures. The addition for steel superstructures is patterned
after the temperature gradient for that type of bridge in the Australian
bridge specifications, Austroads (1992).
Temperatures for use with Figure 5.2.2-1 may be taken from
Table 5.2.2-1.
Lecture - 5-4
Table 5.2.2-1 - Basis of
Temperature Gradients
Zone T
1
( C) T
2
( C)
1 30 7.8
2 25 6.7
3 23 6
4 21 5
Figure 5.2.2-2 - Solar Radiation Zones for the United States
Positive temperature values for the zones shall be taken as
specified for various deck surface conditions in Table 5.2.2-1.
Negative temperature values shall be obtained by multiplying the
values specified in Table 5.2.2-1 by -0.3 for decks with the concrete
top surface exposed and -0.2 for decks with an asphalt overlay.
The temperatures given in Table 5.2.2-1 form the basis for
calculating the change in temperature with depth in the cross-section,
not absolute temperature.
Where temperature gradient is considered, internal stresses
and structure deformations due to both positive and negative
temperature gradients may be determined by dividing into three
effects as follows:
AVERAGE AXIAL EXPANSION - This is due to the uniform
component of the temperature distribution which should be
Lecture - 5-5
considered simultaneously with the uniform temperature
specified in Article S3.12.2. It may be calculated as:
(5.2.2-1)
T
UG
1
A
c
T
G
dw dz
The corresponding total uniform axial strain from both T
u
and
T
G
is:
(5.2.2-2)
u
[T
UG
T
u
]
FLEXURAL DEFORMATION - Since plane sections remain
plane, a curvature is imposed on the superstructure so as to
accommodate the linearly variable component of the
temperature gradient. The rotation per unit length
corresponding to this curvature may be determined as:
(5.2.2-3)

I
c
T
G
z dw dz
1
R
If the structure is externally unrestrained, i.e., simply supported
or cantilevered, no external force effects are developed due to
this superimposed deformation.
The axial strain and curvature may be used in both flexibility
and stiffness formulations. In the former,
u
may be used in
place of P/AE, and may be used in place of M/EI in
traditional displacement calculations. In the latter, the fixed-
end force effects for a prismatic frame element may be
determined as:
N = EA
c
u
(5.2.2-4)
M = EI
c
(5.2.2-5)
An expanded discussion with examples may be found in Ghali
(1989).
Strains induced by other effects such as shrinkage and creep
may be treated in a similar manner.
INTERNAL STRESS - Internal stresses in addition to those
corresponding to the restrained axial expansion and/or rotation
may be calculated as:
E
= E [ T
G
- T
UG
- z] (5.2.2-6)
where:
T
G
= temperature gradient ( C)
T
UG
= temperature averaged across the cross-section
( C)
Lecture - 5-6
T
u
= uniform specified temperature ( C)
A
c
= cross-section area - transformed for steel
beams (mm
2
)
I
c
= inertia of cross-section - transformed for steel
beams (mm
4
)
= coefficient of thermal expansion (mm/mm/ C)
E = modulus of elasticity (MPa)
R = radius of curvature (mm)
w = width of element in cross-section (mm)
z = vertical distance from center of gravity of cross-
section (mm)
Note that a positive value of
E
in Equation 5.2.2-6 denotes
compression.
For example, the flexural deformation part of the gradient
flexes a prismatic superstructure into a segment of a circle in the
vertical plane. For a two-span structure with span length L, in mm, the
unrestrained beam would lift off from the central support by = L
2
/2R
mm. Forcing the beam down to eliminate would develop a moment
whose value at the pier would be:
(5.2.2-7)
M
c
3
2
EI
c
Therefore, the moment is the function of the beam rigidity and

imposed flexure. As rigidity approaches 0.0 at the strength limit state,
M
c
tends to disappear. This behavior also indicates the need for
ductility which ensures structural integrity as rigidity decreases.
A "common sense" explanation of the stress given by Equation
5.2.2-6 follows.
Consider the effect of a temperature gradient on a set of
disconnected layers within the structure. Each layer will expand an
amount required by the temperature gradient.
Lecture - 5-7
Equation 5.2.2-6 can be developed in two steps.
Step 1: Consider only the layer restraint stresses. For
illustration, consider the top layer:
1
T
G
L
But, the layers are not really disconnected, so apply a force to each
layer to make the displacement equal to zero, with compression taken
as positive.
P
1

1
A
1
E
L
T
G1
A
1
E
T
G
E
But, P
1
+ P
2
+ P
3
... cannot result in a net force on the unrestrained
end.
Similarly, P
i
Y
iCG
cannot result in a net external moment.
Now apply as notional external loads to P P
i
& M P
i
Y
iCG
result in a net external axial load and moment is zero. The total effect
is then:
So far, even for a simply supported beam:
i
P
i
A
i
P
i
A
yP
i
y
i
I
T
G
E
1
A
T
G
EdA
1
I
T
G
EydA
Since these are notional force effects, they appear to exist even at the
free edge of the beam.
Step 2: Add the effect of restrained indetermanent reactions,
if any, caused by the global end rotation in the
redundant structure.
! Apply end rotations as shown below:
Lecture - 5-8
! The moment diagram with redundant reactions removed is a
rectangle, i.e., constant moment from end-to-end. This
moment results in a displaced shape as shown below.
! Calculate reactions required to have no displacement at
redundant supports
! The moment diagram due to only the redundnat reactions is
shown below:
! The force effects due to redundant reactions is added to the
force effects obtained in Step 1.
Alternatively, if the axial force and moment given by Equations
5.2.2-4 and 5.2.2-5 are used as input into a structural analysis
package, the stresses from those two factors and the redundant
reaction are calculated directly, and only the first term of Equation
5.2.2-6 need be added to those results to get a complete solution.
5.2.3 Differential Shrinkage
Where appropriate, differential shrinkage strains between
concretes of different age and composition, and between concrete and
steel or wood, are determined in accordance with the provisions of
Section S5.
The designer may specify timing and sequence of construction
in order to minimize stresses due to differential shrinkage between
components.
5.2.4 Creep
Creep strains for concrete and wood are given by the
provisions of Section S5 and Section S8, respectively. Traditionally,
only creep of concrete is considered. Creep of wood is addressed
only because it applies to prestressed wood decks. In determining
force effects and deformations due to creep, dependence on time and
changes in compressive stresses shall be taken into account.
Lecture - 5-9
5.2.5 Settlement
Force effects due to extreme values of differential settlements
among substructure and within individual substructures units is
considered. Estimates of settlement may be made in accordance with
the provision of Article S10.7.2.3. Force effects due to settlement may
be reduced by considering creep.
5.3 OTHER LIVE LOAD EFFECTS
5.3.1 General
Lecture 3 contained information on the development of the
notional live load model. This lesson focuses on other aspects of the
live load, other than the weight and axle configuration of the notional
load, per se.
5.3.2 Centrifugal Force
The only addition to centrifugal force, compared to previous
editions of Standard Specification, is the inclusion of the factor 4/3 in
Equation S3.6.3-1. As explained in the commentary, the group of
exclusion vehicles, described in detail in Lecture 3, produced force
effects which are generally at least 4/3 of that caused by the design
truck alone on short- to medium-span bridges. This is the origin of the
4/3 factor in Equation S3.6.3-1. It is an approximation used to account
for vehicles whose gross vehicle weight is greater than the 325 kN
design truck.
Centrifugal force is applied to the design truck or to the tandem
axle, but not to the lane load on the basis that the lane load is used to
account for disbursed traffic, which normally does not contribute
significantly to a single pier, except on longer spans. Since centrifugal
force is applied to all loaded lanes, the multiple presence factors
apply.
5.3.3 Braking Force
The braking force requirement of Article S3.6.4 is 25% of the
axle weights or the design tandem in each lane with a multiple
presence factor applied. This is substantially greater fraction than
previous editions of the standard specifications. However, it is only
applied to the design load or design tandem, not the uniform load.
The 25% specified is based on improved braking capability of modern
trucks. The commentary to this article indicates the set of parameters
from which the 25% was developed, an exercise in engineering
judgment. The braking force is not applied to the lane load on the
basis that vehicles are apt to break out-of-phase.
Lecture - 5-10
5.3.4 Vehicular Collision Forces
Portions of structures deemed to be protected are not required
to be designed for vehicular collision loads. Structures which fall into
this category are those which are protected by an embankment, a
structurally independent crashworthy ground-mounted barrier, as
defined in Article S3.6.5.1, or a properly designed barrier positioned
to protect the structural elements. Barriers are to be designed for the
forces indicated in Section 13, Railings. Vehicular and railroad
collision loads are defined as a static equivalent force of 1800 kN
acting at a height 1200 mm above the ground. This force was
developed from information from full-scale tests involving the
redirection of tractor trailer trucks by barriers and from analytical work
identified in the commentary.
5.4 WATER LOADS
There are no substantially new requirements for static
pressure, buoyancy or stream pressure. The issue of debris moving
in a stream and impacting piers, or even elements of superstructure,
is the subject of an NCHRP Research Project, just getting underway
at this writing (Spring 1994). Wherever possible, the amount of
freeboard on superstructure and the spacing of piers should be
established to permit reasonable floating debris, determined on the
basis of review of the flood plain, from impacting the superstructure.
Floating debris does tend to collect around piers, forming what
is called a debris raft. Debris rafts tend to increase the apparent area
of the pier subject to stream pressure. Pending future U. S. research
on the design parameters for debris rafts, the commentary contains a
provision from the New Zealand Highway Bridge Design Specification
which may be used until better data becomes available.
In addition to the pressure, identified above, the Specification
contains a requirement to design for wave loads where structures are
exposed to that type of environment. Reference is given in the
commentary to the Shore Protection Manual as a source for wave
design information.
Also, new to the Specification is a requirement to design the
foundations for scour. Scour is a change in foundation condition
resulting from the design flood for scour and is applicable as both a
requirement for strength and service limit states, but also as an
extreme event limit state under the check flood for scour, usually a
500-year flood. In this case, the requirement is that the structure
should remain stable and intact at its full nominal resistance.
Lecture - 5-11
5.5 WIND LOADS
5.5.1 General Wind Provisions
Article S3.8 establishes wind loads which are consistent with
the format and presentation currently used in meteorology. Wind
pressures are established with consideration to a base wind velocity
(V
b
) of 160 km per hour corresponding to the 100 mph wind common
in past specifications. If no better information is available, the wind
velocity at 10 000 mm above the ground, a distance presumed to be
above the immediate effects of the ground in open terrain, may be
taken as the base wind V
b
. Alternatively, the base wind speed may be
taken from the Basic Wind Speed Charts available in the literature or
site specific wind surveys may be used to establish V
10
. Using
characteristics available in Table S3.8.1.1-1, repeated below, used to
describe the type of terrain over which approach winds move, two
characteristic values, V
0
and Z
0
, can be determined. These are
meteorological terms known as the friction velocity and friction length,
respectively.
CONDITION OPEN COUNTRY SUBURBAN CITY
V
0
(km/hr) 13.2 15.2 19.4
Z
0
(mm) 70 1000 2500
Using the information determined above and the height of the
structure above ground or water, if it is over 10 000 mm, it is possible
to calculate a design wind velocity, V
dz
, using Equation S3.8.1.1-1,
which is repeated below.
(5.5.1-1)
V
DZ
2.5 V
0
V
10
V
B
ln
Z
Z
o
Equation S3.8.1.1-1 provides a correction for structure
elevation with similar intent to that used by the 1/7 power rule used by
designers in the past, but agrees with the current meteorological
theories.
Given the design wind speed, now corrected for approach
conditions and height above the referenced datum of 10 000 mm, it is
possible to calculate the design wind pressure, P
D
, based on base
pressure, P
B
, given in Table S3.8.1.2-1, repeated below, for various
structural components. The base wind pressures, specified in that
table, are established for the case where V
B
is equal to 160 km per
hour. The base wind velocity and the wind velocity at the elevation in
question are used to customize the base wind pressures given in
Table S3.8.1.2-1 for the particular site conditions. Additionally, certain
minimum design wind pressures, comparable to those in past editions
of the Standard Specification, are also required.
Lecture - 5-12
STRUCTURAL
COMPONENT
P
B
WINDWARD LOAD,
MPa
P
B
LEEWARD LOAD,
MPa
Trusses,Column and
Arches
0.0024 0.0012
Beams 0.0024 NA
Large Flat Surfaces 0.0019 NA
Figure 5.5.1-1 shows the variation in design pressure with
height for the three upwind conditions.
0
0.001
0.002
0.003
0.004
8
0
0
0
1
2
0
0
0
1
6
0
0
0
2
0
0
0
0
2
4
0
0
0
2
8
0
0
0
Z(mm)
P
D

-

M
P
a
country PD
Suburban
City
Figure 5.5.1-1 Design Wind Pressure, PD, Vs. Height for P
B
= 0.0024
MPa
Consider the following example data point in Figure 5.5.1-1:
Assume V
B
= V
10
= 160 km/hr
Assume suburban setting for which V
o
= 15.2 km/hr and Z
o
=
300 mm
Assume windward pressure on a beam for which P
B
= 0.0024
MPa
Assume Z = 19 000 mm
V
DZ
2.5 V
o
V
10
V
B
n
Z
Z
o
V
DZ
2.5(15.2)
160
160
n
19 000
300
Lecture - 5-13
V
DZ
158
km
hr
P
D
0.0024
(158)
2
(160)
2
0.002 34 MPa
Wind pressure is also applied to vehicular live load, as was the
case in past editions of the Specification. No substantial changes in
this provision have been made.
5.5.2 Wind Blowing at an Angle
Where the wind is not taken as normal to the structure, the
base wind pressures, P
B
, for various angles of wind direction may be
taken as specified in the table below and shall be applied to a single
place of exposed area. The skew angle is measured from a
perpendicular to the longitudinal axis of the bridge. The wind direction
for design shall be that which produces the extreme force effect on the
component under investigation. The transverse and longitudinal
pressures are to be applied simultaneously. For trusses, columns,
and arches, the base wind pressures specified in the table are the
sum of the pressures applied to both the windward and leeward areas.
Base Wind Pressures, P
B
, for Various Angles of Attack and
V
B
= 160 km/hr
Columns and Arches Girders
Skew Angle
of Wind
Lateral
Load
Longitudinal
Load
Lateral
Load
Longitudinal
Load
Degrees MPa MPa MPa MPa
0 0.0036 0 0.0024 0
15 0.0034 0.0006 0.0021 0.0003
30 0.0031 0.0013 0.0020 0.0006
45 0.0023 0.0020 0.0016 0.0008
60 0.0011 0.0024 0.0008 0.0009
5.5.3 Wind Forces Applied Directly to the Substructure
The transverse and longitudinal forces to be applied directly to
the substructure shall be calculated from an assumed base wind
pressure of 0.0019 MPa. For wind directions taken skewed to the
substructure, this force shall be resolved into components
perpendicular to the end and front elevations of the substructure. The
component perpendicular to the end elevation shall act on the
exposed substructure area as seen in end elevation, and the
component perpendicular to the front elevation shall act on the
Lecture - 5-14
exposed areas and shall be applied simultaneously with the wind
loads from the superstructure.
5.5.4 Wind Pressure on Vehicles
When vehicles are present, the design wind pressure shall be
applied to both structure and vehicles. Wind pressure on vehicles
shall be represented by an interruptible, moving force of 1.46 N/mm
acting normal to, and 1800 mm above, the roadway and shall be
transmitted to the structure.
When wind on vehicles is not taken as normal to the structure,
the components of normal and parallel force applied to the live load
may be taken as specified in the table below with the skew angle
taken as referenced normal to the surface.
Wind Components on Live Load
Skew Angle
Normal
Componen
t
Parallel
Component
Degrees N/mm N/mm
0 1.46 0
15 1.28 0.18
30 1.20 0.35
45 0.96 0.47
5.5.5 Vertical Wind Pressure
Standard Specification has required the use of a windward 1/4
point vertical load, and this requirement is continued in the current
Specification. The purpose of this requirement is to account for the
change in pressure caused by the interruption of a horizontal wind
stream created by the bridge superstructure.
5.5.6 Aeroelastic Stability
The provisions of this article require that components whose
span to width or depth ratio exceeds 30 be considered wind-sensitive
and that aeroelastic effects should be taken into account for this type
of member or structure. The choice of the value of 30 is somewhat
arbitrary and was set so that most conventional composite girder
construction would not be affected by this provision. This is based
solely on experience. There have been cases where girders have
vibrated during construction, but this has been relatively rare.
All flexible structures or structural components should be
investigated for resistance to vortex shedding excitation, wake
Lecture - 5-15
buffeting and divergence and flutter, as appropriate. In the past, it has
typically been assumed that suspension bridges and cable-stayed
bridges were the type of structures for which aeroelastic effects could
be significant. However, other relatively flexible modern structures
may also be susceptible. Similarly, components of structures, notably
long, unbraced structural members and cables, may also be subject
to wind-induced vibrations. In at least one case, a low amplitude
vibration of a tied-arch appeared to be driving distortion-induced
fatigue cracking of welds inside the tie girder.
In response to these phenomena, the Specification requires
that bridges should be designed to be free of divergences and flutter
for wind speeds up to 1.2 times the design wind velocity at the bridge
deck height, and that structural components and bridges should be
free of fatigue damage due to vortex shedding or galloping.
It is of some interest to note that the same problems seem to
appear over and over again. Consider the four structures discussed
below, each of which had hanger vibrations due to vortex shedding.
The Tacony-Palmyra Bridge, designed in Circa 1928, has
hangers for the deck system which are structural members. These
hangers vibrated in the wind, probably to vortex shedding. In this
case, the solution was to reduce the length of the member by putting
a horizontal strut across the bridge intercepting each of the hangers.
The Robert Moses Causeway Bridge, Circa 1958, is a similar
configuration with long structural-shape hangers supporting the
roadway. These too vibrated in the wind, probably also from vortex
shedding. In this case, the members were studies in the wind tunnel,
and it was decided to use an aerodynamic solution by attaching sheet
metal deflectors to the corners of the I-shaped hangers, thus,
streamlining the shape. The deflectors contained a projection at 45
bent into the body of the hanger. This retrofit has been quite effective.
The Commodore Barry Bridge, Circa 1970, also has long
vertical members supporting the deck which vibrated in the wind and
caused significant fatigue damage to these members. In this case,
tuned mass dampers were added to the members and their
effectiveness was verified in the wind tunnel prior to installation. This
retrofit also appears to have been successful.
The I-470 Bridge at Wheeling, Circa 1982, used structural
strand hangers in groups of four to support the roadway. All these
hangers vibrated. On the longer hangers, it was possible to clearly
see 7th and 8th load vibrations in moderate winds. This vibration
caused damage to some of the wires resulting in wire breaks where
the structural strands were supported at the deck level. Wooden
wedges driven between the strands and their structural member
supports was found to add sufficient damping to stop these vibrations.
In this case, retrofit consisted of a steel and neoprene collar added to
the ends of the cables to simulate the effect of the wooden wedges,
and spacers between the cables and a set of four to add brace points
and to create damping by the action of one cable vibrating against
another.
Lecture - 5-16
The frequency of the vortex shedding and, hence, the pulsating
pressure, is given by:
(5.5.3-1)
f
VS
D
where:
V = the wind speed in mm/sec
D = a characteristic dimension, in mm
S = the Strouhal Number
A table of Strouhal Numbers for sections is given in "Wind Forces on
Structures", Transactions of the ASCE, Volume 126, Part II, Page
1180, and is repeated below.
Table 5.5.3-1 - Strouhal Number for Various Sections
Wind Profile Proportion Value of S Profile Proportion Value of S
0.120 0.200
0.137
0.144
b/d
2.5
2.0
1.5
1.0
0.7
0.5
0.060
0.080
0.103
0.133
0.136
0.138
0.145
0.147
Lecture - 5-17
Self-exciting oscillations of the member in the direction
perpendicular to the wind stream may result when the frequency of
vortex shedding coincides with a natural frequency of the obstruction.
Thus, determining the torsional frequency and bending frequency in
the plane perpendicular to the wind and substituting those frequencies
into the Strouhal equation leads to an estimate of wind speeds at
which resonance may occur. This motion has led to fatigue cracking
of some truss and arch members, particularly cable hangers and I-
shaped members. The vortex shedding design approach, described
herein, is oriented towards providing sufficient stiffness to reasonably
preclude vibrations. It does not directly lead to a solution for the
amplitude of vibration and, hence, it does not directly lead to a solution
for vibratory stresses. Solutions for amplitude are available in the
literature.
The following approximate procedure for estimating bending
and torsional frequencies is an excerpt from "Natural Frequencies of
Axially Loaded Bridge Members" by C. C. Ulstrup, Journal of the
Structural Division, ASCE, 1978.
The general approximate formula for members whose shear
center and centroid coincide is as follows:
(5.5.3-2)
f
n
a
2
k
n
l
l
2
1
p
Kl
2
1
2
in which:
f
n
= natural frequency of member for each mode
corresponding to n = 1, 2, 3, etc.
k
n
l = eigenvalue for each mode (see table below)
K = effective length factor (see table below)
l = length of the member
a = coefficient dependent on the physical properties of the
member, given as a
b
or a
t
e
p
= coefficient dependent on the physical properties of the
member and on the axial force, i.e., positive for
tension, negative for compression, given as e
pb
or e
pt
Lecture - 5-18
Table 5.5.3-2 - Eigenvalue k
n
l and Effective Length Factor K
Support Condition k
n
l K
n = 1 n = 2 n = 3 n = 1 n = 2 n = 3
2 3 1.000 0.500 0.333
3.927 7.069 10.210 0.700 0.412 0.292
4.730 7.853 10.996 0.500 0.350 0.259
1.875 4.694 7.855 2.000 0.667 0.400
For bending:
(5.5.3-3) a
b
EIg
A
1
2
(5.5.3-4)
e
pb
P
EI
For torsion:
(5.5.3-5) a
t
EC
w
g
I
p
1
2
(5.5.3-6)
e
pt
GJ PI
p
A
1
EC
w
in which:
E = Young's modulus
G = shear modulus
= weight density of member
g = gravitational acceleration
P = axial force (tension is positive)
I = moment of inertia about relevant axis
A = area of member cross-section
C
w
= warping constant
J = torsion constant
I
p
= polar moment of inertia
In the design of a member, the frequency of vortex shedding
for the section is set equal with the bending and torsional frequency
and the resulting equation solved for the wind speed V. This is the
wind speed at which resonance occurs and the design should be such
that V exceeds the velocity at which the wind is expected to occur by
a reasonable margin.
Lecture - 5-19
REFERENCES
Committee on Ship-Bridge Collisions, "Ship Collisions with Bridges -
The Nature of the Accident, their Prevent and Mitigation", Marine
Board, Commission on Engineering and Technical Systems, National
Research Council, National Academic Press, Washington, D.C., 1983
Modjeski and Masters, Inc., Consulting Engineers, "Criteria for: The
Design of Bridge Piers with Respect to Vessel Collision in Louisiana
Waterways", Prepared for the Louisiana Department of Transportation
and Development and the Federal Highway Administration, New
Orleans, Louisiana, November 1984
American Association of State Highway and Transportation Officials
(AASHTO), "Guide Specification and Commentary for Vessel Collision
Design of Highway Bridges", 1991, Washington, D. C.
Larsen, O. D., "Ship Collision with Bridges - The Interaction between
Vessel Traffic and Bridge Structures", International Association for
Bridge and Structural Engineering (IABSCE), Structural Engineering
Documents, 1993
Lecture - 6-1
LECTURE 6 - ANALYSIS I
The objectives of this lesson are to acquaint the student with:
the various analysis techniques required and/or recommended
for determining the force effects and components of bridges,
and
the use of approximate and refined methods for the
determination for force effects in conventional girder-type
structures.
The background on the development of new, improved,
distribution factors which were developed under NCHRP Project 12-26
has been included for reference in an Appendix.
The use of grid and finite element types of analysis for multi-
beam bridges is also recommended in the LRFD Specification. These
methods require considerable care in structural modeling, and several
examples of the large effects of seemingly small errors in structural
models will be presented.
6.2 ACCEPTABLE METHODS OF STRUCTURAL ANALYSIS
Article S4.4 contains a list of methods of analysis that are
considered suitable for analysis of bridges. These include the
classical force and displacement methods, such as virtual work,
moment distribution, slope deflection, the so-called general method,
the more modern finite element, finite strip and plate analogy-type
methods, analysis based on series expansions and the yield-line
method for the non-linear analysis of plates and railings. Some of
these methods of analysis are suitable for hand calculations, but for
any problem of large size, some sort of computer solution will almost
always be required for practical design purposes. This is because
almost all of these methods, with the possible exception of the series
methods and the yield-line methods, will eventually require the
solution of large sets of simultaneous equations. The series method,
while elegant from a mathematical point of view, will typically require
a computer program to expand the series sufficiently to yield good
results in a practical time frame. Yield-line methods, which could be
considered the extension of plastic design to two-dimensional
surfaces, are typically a hand calculation procedure.
The use of computer programs in bridge design brings up a
philosophical problem as to the responsibility for error. Almost all
vendors of commercial computer programs disavow any responsibility
for error. A release from liability is usually implicit in their use and may
even be an explicit part of obtaining a license. This means that an
organization using a computer program must be relatively certain of
Lecture - 6-2
the results that it obtains. It is not necessary for every engineer in a
large design section to have personally confirmed every computer
program, but it is necessary that some verification testing be done or
that the results of previous verification testing be obtained in order to
produce the required level of confidence. Computer programs can be
verified against universally accepted closed- form solutions, other
computer programs which have been previously verified, or the results
of testing.
Many computer programs for design use also contain code
checking capabilities. Others have portions of the applicable design
specification embedded in the coding of the program. In order to
identify the specification edition which may have been tied to a given
release of a program and also to provide a means for determining
which structures may have been designed with a version of a program
later found to contain errors, the specification requires that a name,
version and a release date of software be identified in the contract
drawing.
6.3 PRINCIPLES OF MATHEMATICAL MODELING
6.3.1 Structural Material Behavior
The LRFD Specification recognizes both elastic and inelastic
behavior of materials for analysis purposes. Inelastic material
behavior is implicit in many of the equations and procedures specified
for the calculation of cross-sectional resistance, such as calculating
the nominal resistance of a concrete beam or column, the nominal
plastic moment resistance of a compact and adequately braced steel
cross-section, or the bearing capacity of a spread footing. Often, the
force effects to which this resistance will be compared will be
calculated on a basis of a linear structural analysis with elastic
material properties having been assumed. This dichotomy has existed
in the bridge specification since the introduction of load factor design
in the early 1970's. It continues through the LRFD Specification.
On the other hand, there are certain assumed failure modes at
extreme events and the use of mechanism and unified autostress
design procedures for steel girders, where permitted, which require
analysis based on non-linear behavior. Many times, this analysis will
take a form analogous to plastic design of steel frames. For example,
seismic design may be based on the formation of plastic hinges at the
top and bottom of the columns of a bent. Ship collision forces may be
absorbed in a comparable inelastic manner. Furthermore, it is
anticipated that future seismic design provisions will be based on
extensive research currently underway to develop a step-by-step non-
linear force displacement relationship for the lateral displacement of
piers.
Where inelastic analysis is used, the Designer must be certain
that a ductile failure mode is obtained through proper detailing. Rules
Lecture - 6-3
for achieving this are presented in the sections for steel and concrete
design.
6.3.2 Geometry
6.3.2.1 GENERAL
Most analyses done for the purpose of designing bridges are
based on the assumption that the displacements caused by the loads
are relatively small and, therefore, it is suitably accurate to base the
calculations on the undeformed shape. This is typically referred to as
the small deflection theory, and it is routinely used in the design of
beam-type structures and bridges which resist loads through a couple
whose tensile and compressive forces remain essentially in fixed
positions relative to each other while the bridge deflects. This will be
the case for a truss and for tied-arches.
For other types of structures and components and for certain
types of analysis, the effect of the deflections must be considered in
the development of the force equilibrium equations, i.e., the equations
of equilibrium are written for the displaced shape. Almost all
engineers are aware that the study of structural stability requires
consideration of the displaced shape, in fact, if the displaced shape is
not part of the original formulation of the problem, one would never be
able to determine that a column, shell or plate can buckle. Consider
for a moment the simple pin-ended column. Unless the deflected
shape of the column is taken into account, the moment caused by the
axial load acting on the displaced shape would not be accounted for.
It is this moment which causes the column to move laterally, i.e., to
buckle.
Almost a century ago, it was found that the only reasonably
accurate way to calculate force effects in suspension bridges of any
size was to include the deflection of the cable in the formulation of the
problem and, therefore, the displacement of a stiffening truss or
stiffening girder. As conventional, i.e., not tied, arches became longer
and more slender, an effect directly analogous to that observed in
suspension bridges can become significant enough that it must be
accounted for in the design of the arch rib. In fact, because the arch
rib is in compression and can buckle, the effect of large deflections
can be especially important.
Finally, the compression members of frames in bents can also
be susceptible to this phenomenon.
Where non-linear effects arriving either out of material non-
linearity or large deflections become significant, then super position of
forces does not apply. This means that each load case under
investigation must be studied separately under the full effect of all of
the factored loads that make up the load combination under study.
This is a very significant effect on most practical design calculations.
Commonly, a Bridge Engineer calculates the force effect from a
variety of individual loads and then combines, or superimposes, the
Lecture - 6-4
force effects calculated for each individual load to make up whatever
group combination of loadings are needed. For non-linear analysis,
each combination must be investigated, i.e., analyzed, individually.
6.3.2.2 APPROXIMATE METHODS
To simplify analysis and to partially bypass the need to analyze
each load combination separately, as identified above, certain
approximate methods have been developed to allow the designer to
add a correction to a set of force effects calculated in a linear manner.
These are sometimes called single-step adjustment methods, the
most commonly used of which is moment magnification for beam
columns, which has been part of the AASHTO Specifications since the
early 1970's.
For beam columns, the moment magnification process is given
by the equations below.
M
c
=
b
M
2b
+
s
M
2s
(6.3.2.2-1)
f
c
=
b
f
2b
+
s
f
2s
(6.3.2.2-2)
for which:
(6.3.2.2-3)
b
C
m
1
P
u
P
e
1.0
(6.3.2.2-4)
s
1
1
P
u
P
e
where:
P
u
= factored axial load (N)
P
e
= Euler buckling load (N)
= resistance factor for axial compression as specified in
Specification Sections 5, 6 and 7, as applicable
M
2b
= moment on compression member due to factored
gravity loads that result in no appreciable sidesway
calculated by conventional first order elastic frame
analysis, always positive (N mm)
f
2b
= stress corresponding to M
2b
(MPa)
M
2s
lateral or gravity loads that result in sidesway, ,
Lecture - 6-5
greater than
u
/1500, calculated by conventional first
order elastic frame analysis, always positive (N mm)
f
2s
2s
(MPa)
It may appear that the moment magnification factor contains the Euler
buckling load, P
e
. However, P
e
is only a convenient substitution for a
group of terms related to the displacement of the beam column.
A derivation of the moment magnification equation can be
found in many textbooks on steel or concrete design.
For cases where the shape of the beam column is expected to
be radically different from that given by the simply-supported case, or
the loads significantly different from those indicated above, then it is
possible to make an adjustment to account for a different initial elastic
shape through the factor c
m
.
The moment magnification procedure has also been extended
to arches, and this has been available in the AASHTO Specifications
for many years and is reproduced as Article S4.5.3.2.2c with no further
refinement.
6.3.2.3 REFINED METHODS
The effect of large deflections can also be rigorously
accounted for through iterative solutions of equilibrium equations,
taking into account updated positions of the structure, or by using
geometric stiffness terms. In some cases, e.g., the suspension bridge,
solutions are available to the differential equations of equilibrium which
can be solved in a trial and error fashion, or through series expansion.
6.3.3 Modeling Boundary Conditions
Points of expansion or other forms of articulation in the
structure are commonly idealized as frictionless units. Where past
practice indicates that this has been a reasonable conservative
approach, continued use is warranted. There are other instances
where the potential for nonfunctional expansion devices and/or the
possibility that joints may close should also be investigated. This
might be the case, for example, in a seismic analysis where analysis
may be made, assuming that expansion joints are operable and open,
and then another analysis might be made, assuming that they are
closed and nonfunctional in order to simulate, or bound, the effects of
joints reaching the limits of travel during the earthquake. The
possibility of reaching the limit of expansion travel should also be
investigated when evaluating non-linear effects on substructure
elements. It may be possible that the amount of moment
magnification may be reduced because expansion dams will close,
jamming the structure against the abutments before the full movement
implicit in the moment magnification factor can be reached. This will
reduce the moment magnification factor and, hence, the design
moment.
Lecture - 6-6
Similarly, the effect of boundary conditions at foundation units
should also be evaluated. Foundation units are seldom fully fixed or
fully pinned, and an evaluation of the potential movement of a
foundation unit may be necessary in order to properly assess
response, as well as secondary moments caused by change in
geometry. Here again, bounding of the range of probable movement
may be the only practical way to attack such a problem.
6.4 STATIC ANALYSIS
6.4.1 The Influence of Plan Geometry
Article S4.6.1 deals with two simplifications which can be made
based on the plan geometry of the superstructure.
The first simplification involves the possibility of replacing the
superstructure for analysis purposes with a single-line element called
a spine beam. This may be done when the transverse distortion of the
superstructure is small in comparison with the longitudinal
deformation. Generally, if the superstructure is a torsionally stiff
closed section or sections whose length exceeds 2.5 times their width,
it may be idealized as a line element whose dimensions may be
determined as given in the Specification. This can be used to
significantly simplify analysis models.
The second simplification deals with when it is possible to
consider curved superstructures as straight for the purpose of
analysis. If the superstructure is a torsionally stiff closed section and
the central angle of a segment between piers is less than 12 , then
the segment may be considered straight. If the superstructure is
made of torsionally weak open sections, then the effects of curvature
may be neglected when the subtended angle is less than that given in
Table 6.4.1-1.
Table 6.4.1-1 - Limiting Central Angle
for Neglecting Curvature in Determining
Primary Bending Moments
No. of Beams
Angle for One
Span
Angle for
Multiple Spans
2 2 3
3 or 4 3 4
5 or more 4 5
Lecture - 6-7
6.4.2 Approximate Methods for Load Distribution
6.4.2.1 DECK SLABS AND SLAB-TYPE BRIDGES
The Specification permits the approximate analysis of deck
slabs by analyzing a strip of deck as a continuous beam. Provisions
are made for determining the width of that strip at
the unsupported edge of the slab and at points interior from the edges.
If the spacing of supporting components in the secondary
direction exceeds 1.5 times the spacing in the primary direction, then
all of the wheel loads applied to the deck are considered to be applied
to the primary strip. The secondary strip is designed on a basis of
percentage of reinforcement in the primary strip.
If the spacing of the supporting components on the secondary
direction is less than 1.5 times that in the primary direction, then a
crossed sticks analogy is used. The width of the equivalent strips in
each direction is provided by Table S4.6.2.1.3-1 and the wheel load
is distributed between two idealized intersecting strips according to the
relative stiffness of each strip.
Once the wheel loads have been assigned to the strips, for
either case identified above, then the force effects are calculated
based on a continuous beam. For the purpose of analyzing the
continuous beam, the span length of each span is taken as a center-
to-center of supporting components. For the purpose of calculating
moment and shear at a design section, some offset from the
theoretical center of support is permitted as given in the Specification.
Decks which form an integral part of a cellular cross-section
are supported on webs which are monolithic with the deck. Therefore,
when the deck rotates, the web of the box girder rotates giving rise to
bending stresses throughout the cross-section. For the purpose of
analyzing this effect, a cross-sectional frame action procedure is
identified in the Specification.
In the case of fully filled and partially filled grids, the results of
recent research are incorporated in LRFD Article S4.6.2.1.8 to given
bending moments per unit length of grid.
6.4.2.2 BEAM SLAB BRIDGES
6.4.2.2.1 General
The Specification provides a series of empirical rules for
assigning portions of a design lane to a supporting component. These
are commonly called distribution factors. It is important to remember
that the approximate distribution factors, specified in the LRFD
Specification, are on a lane, i.e., axle basis, not a wheel basis. The
distribution factors are given for the various kinds of bridges shown in
Figure 6.4.2.2.1-1.
Lecture - 6-8
SUPPORTING
COMPONENTS TYPE OF DECK
TYPICAL
CROSS-
SECTION
Steel Beam - Revised
Factors
Cast-in-place concrete slab,
precast concrete slab, steel grid,
glued/spiked panels, stressed
wood
Closed Steel or Precast
Concrete Boxes - Revised
Factors
Cast-in-place concrete slab
Open Steel or Precast
Factors
precast concrete deck slab
Cast-in-Place Concrete Multi-
cell Box - Revised Factors
Monolithic concrete
Cast-in-Place Concrete Tee
Beam - Revised Factors
Monolithic concrete
Precast Solid, Voided or
Cellular Concrete Boxes with
Shear Keys - Revised
Factors
Cast-in-place concrete overlay
Cellular Concrete Box with
Shear Keys and with or
without Transverse
Post-Tensioning - Revised
Factors (in some cases)
Integral concrete
Precast Concrete Channel
Sections with Shear Keys
Precast Concrete Double Tee
Section with Shear Keys and
with or without Transverse
Post-Tensioning
Integral concrete
Precast Concrete Tee
Post-Tensioning
Integral concrete

Concrete I or Bulb-Tee
Sections - Revised Factors
Cast-in-place concrete, precast
concrete
Wood Beams - Revised
Factors
Cast-in-place concrete or plank,
glued/spiked panels or stressed
wood
Figure 6.4.2.2.1-1 - Common Deck Superstructures Covered in LRFD Specification
Articles 4.6.2.2.2 and 4.6.2.2.3
Lecture - 6-9
Some of the distribution factors are new to the Specification as a result
of an extensive project on load distribution known as NCHRP Project
12-26. Where the distribution factors for a given type of cross-section
have been developed under that project and are new to the
Specification, the words "revised factors" appear in the column
identified as "supporting components". Where those words do not
appear, the distribution factors have been retained from earlier
editions of the AASHTO Standard Specifications.
Some simplifications have been made in utilizing the
distribution factors from NCHRP 12-26. In particular, correction
factors for various aspects of structural action, typically involving
continuity, which were less than 5%, were omitted from the LRFD
Specifications. Similarly, an increase in moments over piers, thought
to be on the order of 10%, was not included because stresses at or
near internal bearings have been shown to be reduced below that
calculated by simple analysis techniques due to an action known as
"fanning". The distribution factors, given in the LRFD Specification,
are also different from those given in NCHRP 12-26, because the
multiple presence factors, given in Lecture 3, are built into the
distribution factors, whereas, the multiple presence factors in earlier
editions of the AASHTO Standard Specifications are built into the
NCHRP 12-26 factors. Additionally, the factors appropriate for the
LRFD Specification are based on a lane of live load, rather than a "line
of wheels". Finally, when the SI version of the LRFD Specification is
used, conversion to that system of units has also been accounted for.
Various limits on span, spacing and other characteristics are
provided in the Specifications for each of the distribution factors.
These parameters identify the range for which the factors were
developed. They were not evaluated for factors beyond the ranges
indicated. Therefore, for structures which do not comply with these
limitations, a rigorous analysis by grid or finite elements should be
used. Furthermore, the distribution factors usually apply for structures
which are:
essentially constant in deck width,
have four or more beams, unless noted,
have beams which are parallel and approximately of the same
stiffness,
have overhangs that do not exceed 0.9 m, unless specifically
noted,
have in-plan curvatures less than those specified above, and
have a cross-section consistent with one of the cross-sections
identified in Table 6.4.2.2.1-1.
Since the distribution factors, developed under NCHRP 12-26,
are new to the Specification, it is appropriate to review the background
Lecture - 6-10
and development of the new distribution factors. The discussion
below was taken from the NCHRP Research Results Digest No. 187,
a summary of the Final Report of Project 12-26 as summarized by
Ian M. Friedland, NCHRP Project Coordinator.
Live load distribution on highway bridges is a key response
quantity in determining member size and, consequently, strength and
serviceability. It is of critical importance both in the design of new
bridges and in the evaluation of the load-carrying capacity of existing
bridges.
Using live load distribution factors, engineers can predict
bridge response by treating the longitudinal and transverse effects of
live loads as uncoupled phenomena. Empirical live load distribution
factors for stringers and longitudinal beams have appeared in the
AASHTO Standard Specifications for Highway Bridges with only minor
changes since 1931. Findings of recent studies suggest a need to
update these specifications in order to provide improved predictions
of live load distribution.
Live load distribution is a function of the magnitude and
location of truck live loads and the response of the bridge to these
loads. The NCHRP 12-26 study focused on the second factor
mentioned above: the response of the bridge to a predefined set of
loads, namely, the HS family of trucks.
In Project 12-26, three levels of analysis were considered for
each bridge type. The most accurate level, Level 3, involves detailed
modeling of the bridge deck. Level 2 includes either graphical
methods, nomographs and influence surfaces, or simplified computer
programs. Level 1 methods provide simple formulas to predict lateral
load distribution, using a wheel load distribution factor applied to a
truck wheel line to obtain the longitudinal response of a single girder.
The major part of the research in Project 12-26 was devoted
to the Level 1 analysis methods because of its ease of application,
established use, and the surprisingly good correlation with the higher
levels of analysis in their application to a majority of bridges. The
formulas presented in the current AASHTO specifications were
evaluated, and alternative formulas were developed that offer
improved accuracy, wider range of applicability, and in some cases,
easier application than the current AASHTO formulas. These
formulas were developed for interior and exterior girder moment and
shear load distribution for single or multiple lane loadings. In addition,
correction factors for continuous superstructures and skewed bridges
were developed.
The formulas presented in previous AASHTO Specifications,
although simpler, do not present the degree of accuracy demanded by
today's Bridge Engineers. In some cases, these formulas can result
in highly unconservative results (more than 40%), in other cases they
may be highly conservative (more than 50%). In general, the formulas
developed in Project 12-26 are within 5% of the results of an accurate
Lecture - 6-11
analysis. Table 6.4.2.2.1-1 shows comparisons with moment
distribution factors obtained from AASHTO, Level 1, Level 2 and Level
3 methods for simple span bridges.
Table 6.4.2.2.1-1 - Comparison of interior girder moment distribution
factors by varying levels of accuracy using the "average bridge" for
each bridge type
Bridge Type AASHTO
NCHRP 12-26
(Level 1)
Grillage
(Level 2)
Finite Element
(Level 3)
Beam-and-slab
a
1.413
(S/1700)
1.458 1.368 1.378
Box girder
a
1.144 1.143 0.970 1.005
Slab
b
1820 1710 1900 1890
Multi-box beam
a
0.646 0.597 0.540 0.552
Spread box beam
a
1.564 1.282 1.248 1.241
a
Number of wheel lines per girder
b
Wheel line distribution width, in mm
In addition, the study resulted in recommendations for use of
computer programs to achieve more accurate results. The
recommendations focus on the use of plane grid analysis, as well as
detailed finite element analysis, where different truck types and their
combinations may be considered.
6.4.2.2.2 Influence of Truck Configuration
The formulas developed in Project 12-26 for the Level 1
analysis were based on the standard AASHTO "HS" trucks. A limited
parametric study conducted as part of the research showed that
variations in the truck axle configuration or truck weight do not
significantly affect the wheel load distribution factors. The group of
axle trains used for this study are shown in Figure 6.4.2.2.2-1. It is
anticipated that smaller gage widths would result in larger distribution
factors, and larger gage widths would result in smaller distribution
factors. Table 6.4.2.2.1-1 gives the variation of wheel load distribution
factors with different axle configurations applied to a number of beam-
and-slab bridges. The differences were below 1% in many cases and,
in all cases, the formulas resulted in good predictions. Therefore, with
some caution, these formulas may be applied to other truck types.
Obviously, Levels 2 and 3 analyses may also be applied for trucks
significantly different from the AASHTO family of trucks.
Lecture - 6-12
Figure 6.4.2.2.2-1 - Axle Configurations for Truck Types Considered
in Study
Table 6.4.2.2.1-1 - Effect of Load Configuration on Distribution Factor
DISTRIBUTION
FACTOR (g)
PERCENT DIFFERENCE
WITH HS-20
HS-20 HTL-57 4A-66 B-141
NCHRP
12-26
HTL-57 4A-66 B-141
NCHR
P 12-
26
Average
*
1.293 1.261 1.285 1.268 1.304 -2.4 -0.6 -1.9 +0.9
Max. S
5000 mm
2.220 2.162 2.205 2.178 2.308 -2.6 -0.7 -1.9 +4.0
Min. S
1000 mm
0.713 0.717 0.713 0.715 0.755 +0.6 0.0 +0.3 +5.9
Max. L
60 000 mm
0.982 0.958 0.983 0.952 1.033 -2.4 +0.1 -3.1 +5.2
Min. L
6000 mm
1.630 1.625 1.624 1.623 1.807 -0.3 -0.3 -0.4 +10.9
*
S = 2200 mm
L = 20 000 mm
t
s
= 185 mm
K
g
= 2.33 x 10
11
mm
4
Lecture - 6-13
6.4.2.2.3 Simplified Methods
6.4.2.2.3a Simplified Formulas for Beam-and-Slab Bridges
This type of bridge has been the subject of many previous
studies, and many simplified methods and formulas were developed
by previous researchers for multi-lane loading moment distribution
factors. The AASHTO formula, the formulas presented by other
researchers, and the formulas developed in the study are discussed
in the following according to their application.
Table 6.4.2.2.3a-1 is taken from the specifications and
summarized criteria for moment in interior beams or elements for
various types of cross-sections. Similar tables exist for moment in
exterior griders, for shear in interior girders and shear in exterior
girders.
Table 6.4.2.2.3a-2 describes how the term L (length) may be
determined for use in the live load distribution factor equations given
in Table 6.4.2.2.3a-1.
In the rare occasion when the continuous span arrangement
is such that an interior span does not have any positive uniform load
moment (i.e., no uniform load points of contraflexure), the region of
negative moment near the interior supports would be increased to the
centerline of the span, and the L used in determining the live load
distribution factors would be the average of the two adjacent spans.
Lecture - 6-14
Table 6.4.2.2.3a-1 - Distribution of Live Loads Per Lane for Moment in Interior Beams
Type of Beams Applicable
Cross-Section
from Table
4.6.2.2.1-1
Distribution Factors Range of Applicability
Wood Deck on Wood
or Steel Beams
a, l See Table S4.6.2.2.2a-1
Concrete Deck on
Wood Beams
l One Design Lane Loaded: S/3700
Two or More Design Lanes Loaded: S/3000
S 1800
Concrete Deck, Filled
Grid, or Partially Filled
Grid on Steel or
Concrete Beams;
Concrete T-Beams, T-
and Double T-Sections
a, e, k and
also i, j
if sufficiently
connected to
act as a unit
One Design Lane Loaded:
0.06
S
4300
0.4
S
L
0.3
K
g
Lt
3
s
0.1
Two or More Design Lanes Loaded:
0.075
S
2900
0.6
S
L
0.2
K
g
Lt
3
s
0.1
1100 S 4900
110 t
s
300
6000 L 73 000
N
b
4
use lesser of the values obtained from the
equation above with N
b
= 3 or the lever rule
N
b
= 3
Multicell Concrete Box
Beam
d One Design Lane Loaded:
1.75
S
1100
300
L
0.35
1
N
c
0.45
13
N
c
0.3
S
430
1
L
0.25
2100 S 4000
18 000 L 73 000
N
c
3
If N
c
> 8 use N
c
= 8
Concrete Deck on
Concrete Spread Box
Beams
b, c One Design Lane Loaded:
S
910
0.35
Sd
L
2
0.25
S
1900
0.6
Sd
L
2
0.125
1800 S 3500
6000 L 43 000
450 d 1700
N
b
3
Use Lever Rule S > 3500
Cross-Section
from Table
4.6.2.2.1-1
Lecture - 6-15
Concrete Beams used
in Multibeam Decks
f One Design Lane Loaded:
k
b
2.8L
0.5
I
J
0.25
where: k 2.5(N
b
)
0.2
1.5
k
b
7600
0.6
b
L
0.2
I
J
0.06
900 b 1500
6000 L 37 000
5 N
b
20
g
if sufficiently
connected to
act as a unit
h Regardless of Number of Loaded Lanes: S/D
where:
C = K (W/L) K
D = 300 [11.5 - N
L
+ 1.4N
L
(1 - 0.2C)
2
] when C
5
D = 300 [11.5 - N
L
] when C > 5
Skew 45
N
L
6
g, i, j
if connected
only enough to
prevent relative
vertical
displacement at
the interface
K =
(1 ) I
J
for preliminary design, the following values of
K may be used:
Beam Type K
Nonvoided rectangular beams 0.7
Rectangular beams with
circular voids: 0.8
Box section beams 1.0
Channel beams 2.2
T-beam 2.0
Double T-beam 2.0
Steel Grids on Steel
Beams
a One Design Lane Loaded:
S/2300 If t
g
< 100 mm
S/3050 If t
g
100 mm
S/2400 If t
g
< 100 mm
S/3050 If t
g
100 mm
S 1800 mm
S 3200 mm
Concrete Deck on
Multiple Steel Box
Girders
b, c Regardless of Number of Loaded Lanes:
0.05 0.85
N
L
N
b
0.425
N
L
0.5
N
L
N
b
1.5
Lecture - 6-16
Table 6.4.2.2.3a-2 - L for Use in Live Load Distribution Factor Equations
FORCE EFFECT L (mm)
Positive Moment The length of the span for which
moment is being calculated.
Negative Moment - End spans of continuous spans,
from end to point of contraflexure under a uniform
load on all spans
The length of the span for which
Negative Moment - Near interior supports of
continuous spans, from point of contraflexure to point
of contraflexure under a uniform load on all spans
The average length of the two
adjacent spans.
Positive Moment - Interior spans of continuous
spans, from point of contraflexure to point of
contraflexure under a uniform load on all spans
Shear The length of the span for which
shear is being calculated.
Exterior Reaction The length of the exterior span.
Interior Reaction of Continuous Span The average length of the two
adjacent spans.
Moment Distribution to Interior Girders, Multi-Lane Loading
The AASHTO formula for moment distribution for multi-lane
loading is given as S/1800 for reinforced concrete T-beam bridges
with girder spacing up to 3000 mm, and as S/1700 for steel girder
bridges and prestressed concrete girder bridges with girder spacing
up to 4300 mm, where S is the girder spacing. When the girder
spacing is larger than the specified limit, simple beam distribution is to
be used to calculate the load distribution factors.
Marx, et al, at the University of Illinois, developed a formula for
wheel load distribution for moment which included multiple lane
reduction factors and is applicable to all beam-and-slab bridges. The
formula is based on girder spacing, span length, slab thickness and
bridge girder stiffness.
A formula which does not consider a reduction for multi-lane
loading was developed at Lehigh University. The Lehigh formula
includes terms for the number of traffic lanes, number of girders, girder
spacing, span length and total curb-to-curb deck width.
Sanders and Elleby (NCHRP Report 83) developed a simple
formula based on orthotropic plate theory for moment distribution on
beam-and-slab bridges. Their formula includes terms for girder
Lecture - 6-17
spacing, number of traffic lanes and a stiffness parameter based on
bridge type, bridge and beam geometry and material properties.
A full-width design approach, known as Henry's Method, is
used by the State of Tennessee. Henry's Method includes factors for
number of girders, total curb-to-curb bridge deck width and a reduction
factor based on number of lanes.
A formula developed as part of NCHRP Project 12-26 includes
the effect of girder spacing, span length, girder inertia and slab thick-
ness. The multiple lane reduction factor is built into the formula. This
formula, applicable to cross-sections with four or more beams, is given
by:
(6.4.2.2.3a-1)
g 0.075
S
2900
0.6
S
L
0.2
K
g
Lt
3
s
0.1
where:
S = girder spacing (1100 mm S 4900 mm)
L = span length (6000 mm L 73 000 mm)
K
g
= n(I+Ae
g
2
) (4 x 10
8
K
g
3 x 10
12
mm
4
)
n = modular ratio of girder material to slab material
I = girder moment of inertia
e
g
= eccentricity of the girder (i.e., distance from centroid of
girder to mid-point of slab)
t
s
= slab thickness (110 mm t
s
300 mm)
This formula is dependent on the inertia of the girder and, thus, a
value for K
g
must be assumed for initial design. For this purpose,
K
g
/Lt
S
3
may be taken as unity.
All of the above formulas were evaluated using direct finite
element analysis with the GENDEK-5 program and a database of 30
bridges; subsequently, they were evaluated using the MSI method and
database of more than 300 bridges. It was found that Equation 6 and
the Illinois formulas are accurate and produce results that are as
accurate as the Level 2 methods.
Moment Distribution to Exterior Girders, Multi-Lane Loading
Previous AASHTO Specifications recommend a simple beam
distribution of wheel loads in the transverse direction for calculating
wheel load distribution factors in edge girders. Any load that falls
Lecture - 6-18
outside the edge girder is assumed to be acting on the edge girder,
and any load that is between the edge girder and the first interior
girder is distributed to these girders by assuming that the slab acts as
a simple beam in that region. Any wheel load that falls inside of the
first interior girder is assumed to have no effect on the edge girder.
the exterior girder based on certain assumptions in the placement of
loads and may not be applicable to all bridges. This formula includes
terms similar to those used in their formula for moment distribution to
interior girders.
A formula, depending on wheel position, alone was developed
as part of this study which results in a correction factor for the edge
girder. The factor must be applied to the distribution factor for the
interior girder to obtain a distribution factor for the edge girder. This
formula is given by:
(6.4.2.2.3a-2) e 0.77
d
e
2800
where:
d
e
= distance from edge of the roadway, usually the face of
curb, to the center of the exterior web of the exterior
cell, in mm
If the edge of the lane is outside of the exterior girder, the
distance is positive; if the edge of the lane is to the interior side of the
girder, the distance is negative.
It was found that the formula developed in Project 12-26
resulted in accurate correction factors and was simpler than the
previous AASHTO procedure.
Moment Distribution to Interior Girders, Single-Lane Loading
The literature search performed in this study did not reveal any
simplified formula for single-lane loading of beam-and-slab bridges.
The formula developed as part of the study is as follows:
(6.4.2.2.3a-3)
g 0.06
S
4300
0.4
S
L
0.3
K
g
Lt
3
s
0.1
where the parameters are the same as those given for Equation
6.4.2.2.3a-1.
This formula is applicable to interior girders only.
Lecture - 6-19
Moment Distribution to Exterior Girders
Simple beam distribution in the transverse direction should be
used for single-lane loading of edge girders.
One other investigation, applicable to load distribution for both
shear and moment, is required for exterior beams of beam-slab
bridges with diaphragms or cross-frames. This addition was not
developed as part of NCHRP 12-26, but was added by the NCHRP
12-33 Editorial Committee. This distribution is based on treating the
cross-section as a transversely rigid unit which deflects and rotates as
a straight line. The live load is positioned for maximum effect on an
exterior beam (one lane, two lane, three lane, etc., each with its
appropriate multiple presence factor). The total vertical force and
moment about the centroid of the cross-section is applied to the area
of the cross-section, i.e., the number of beams, and the section
modulus, i.e., the sum of the square of the distances of each beam
from the centroid of the beams divided by the distance to the exterior
beam. The specification puts this in equation form as:
(6.4.2.2.3a-4)
R
N
L
N
b
X
ext
N
L
e
v
N
b
x
2
where:
R = reaction on exterior beam in terms of lanes
N
L
= number of loaded lanes under consideration
e
v
= eccentricity of a design truck or a design lane
load from the center of gravity of the pattern of
girders (mm)
x = horizontal distance from the center of gravity of
the pattern of girders to each girder (mm)
X
ext
= horizontal distance from the center of gravity of
the pattern of girders to the exterior girder (mm)
N
b
= number of beams or girders
Shear Distribution
No formula was found from previous research for the
calculation of wheel load distribution factors for shear. Therefore, the
formulas developed as part of the 12-26 study are reported for
different cases as follows.
The formula for multi-lane loading of interior girders is:
Lecture - 6-20
(6.4.2.2.3a-5) g 0.2
S
3600
S
10 700
2
The correction formula for multi-lane loading edge girder shear
is:
(6.4.2.2.3a-6) e 0.6
d
e
3000
The formula for shear distribution factor due to single-lane
loading is:
(6.4.2.2.3a-7)
g 0.36
S
7600
Equation 6.4.2.2.3a-7 is applicable to interior girders only.
Simple beam distribution in the transverse direction should be used for
single-lane loading of edge girders.
Correction for Skew Effects
Previous AASHTO Specifications did not include approximate
formulas to account for the effect of skewed supports. However, some
researchers have developed correction factors for such effects on
moments in interior girders.
Marx, et al, at the University of Illinois, developed four
correction formulas for skew, one each for skew angles of 0, 30, 45
and 60 degrees. Corrections for other values of skew are obtained by
straight-line interpolation between the two enveloping skew values.
These correction formulas are based on girder spacing, span length,
slab thickness and bridge girder stiffness.
A formula for a correction factor for prestressed concrete
I-girders was developed as part of the research performed at Lehigh
University. This formula is based on the number of traffic lanes,
number of girders, girder spacing, span length and total curb-to-curb
deck width, and includes a variable term for skew angle.
A correction factor for moment in skewed supports was also
developed as part of Project 12-26. This formula is:
(6.4.2.2.3a-8) r 1 c
1
(tan)
1.5
where, for > 30 ,:
Lecture - 6-21
(6.4.2.2.3a-9)
c
1
0.25
K
g
Lt
3
s
0.25
S
L
0.5
If 30 , c
1
is taken as zero. In calculating c
1
, should not be taken
as greater than 60 . The other parameters are as defined previously.
From the literature review, no correction formulas were
obtained for shear effects due to skewed supports. In Project 12-26,
it was found that shear in interior girders need not be corrected for
skew effects; that is, the shear distribution to interior girders is similar
to that of a straight bridge. A correction formula for shear at the
obtuse corner of the exterior girder of two girder systems and all
girders of a multi-girder bridge was developed as part of this study and
is given as:
(6.4.2.2.3a-10) r 1 0.2
Lt
3
s
k
g
0.3
tan
where the parameters are defined in Equation 6.4.2.2.3a-1.
Equation 6.4.2.2.3a-7 is to be applied to the shear distribution
factor in the exterior girder of non-skewed bridges. Therefore, the
product of factors g, e and r must be obtained to find the obtuse corner
shear distribution factor in a beam-and-slab bridge.
The distribution factors calculated for moments are plotted as
a function of girder spacing for Spans 9, 18, 27, 36 and 60 m in Figure
6.4.2.2.3a-1. For comparisons, AASHTO (1989) distribution factors
are also shown. Girder distribution factors, specified by AASHTO
(1989), are conservative for larger girder spacing. For shorter spans
and girder spacings, AASHTO (1989) produces smaller distribution
factors than calculated values.
Lecture - 6-22
Figure 6.4.2.2.3a - Calculated Distribution Factors
6.4.2.2.3b Simplified Formulas for Box Girder Bridges
Research on box girder bridges has been performed by
various researchers in the past. Bridge deck behavior has been well
studied and many recommendations have been made for detailed
analysis of these bridges. However, there is a limited amount of
information on simplified wheel load distribution formulas in the
literature.
In this context, a girder is a notional I-shape consisting of one
web of a multi-cell box and the associated half-flanges on each side
of the web.
Moment Distribution to Interior Girders
Scordelis, at the University of California, Berkeley, presented
a formula for prediction of wheel load distribution for moment distribu-
tion in prestressed and reinforced concrete box girder bridges. The
formula is based on modification of distribution factors obtained for a
rigid cross-section. The formula predicts load distribution factors in
reinforced concrete box girders with high accuracy and for prestressed
concrete box girders with acceptable accuracy.
Sanders and Elleby also presented a simple formula for
moment distribution factors which is similar to their formula for beam-
and-slab bridges.
The following formulas, developed as part of NCHRP 12-26,
may be used to predict the moment load distribution factors in the
interior girders of concrete box girder bridges due to single-lane and
multi-lane loadings. These formulas are applicable to both reinforced
Lecture - 6-23
and prestressed concrete bridges, and the multiple presence factor is
accounted for.
For single-lane loading:
(6.4.2.2.3b-1)
g 1.75
S
1100
300
L
0.35
1
N
c
0.45
For multi-lane loading:
(6.4.2.2.3b-2)
g
13
N
c
0.3
S
430
1
L
0.25
where:
S = girder spacing, in mm
L = span length, in mm
N
c
= number of cells
The factor for load distribution for exterior girders shall be
W
e
/4300 mm, where W
e
is the width of the exterior girder, taken as the
top slab width measured from the mid-point between girders to the
edge of the slab.
Shear Distribution
No formula for shear load distribution was obtained from
previous research for box girder bridges, but the following were
developed as part of NCHRP 12-26.
The shear distribution factor for interior girder multi-lane
loading of reinforced and prestressed concrete box girder bridges is:
(6.4.2.2.3b-3) g
S
2200
0.9
d
L
0.1
where:
d = girder depth, in mm
Lecture - 6-24
The distribution factor for shear in the interior girders due to
single-lane loading may be obtained from:
(6.4.2.2.3b-4) g
S
2900
0.6
d
L
0.1
where the parameters are as defined in Equation 6.4.2.2.3b-3.
A correction formula for shear in the exterior girder for multi-
lane loading is:
(6.4.2.2.3b-5) e 0.64
d
e
3800
where:
d
e
cell, in mm
The following formula was developed for correction of moment
due to skewed supports for values of from 0 to 60 :
(6.4.2.2.3b-6) r 1.05 0.25 (tan) 1.0
If > 60 , use 60 in Equation 6.4.2.2.3b-6.
Another formula was developed in Project 12-26 for correction
of shear at the obtuse corner of an edge girder. It must be applied to
the shear distribution factor for the edge girder of a non-skewed bridge
and must, therefore, be used in conjunction with the edge girder
correction factor of Equation 6.4.2.2.3b-5. This formula, applicable for
values of up to 60 , is:
(6.4.2.2.3b-7)
r 1 c
1
(tan)
where:
c
1
= 0.25 + L/(70d)
d = bridge depth, in mm
Lecture - 6-25
6.4.2.2.3c Simplified Formulas for Slab Bridges
The literature search did not reveal any simplified formulas for
wheel load distribution in slab bridges other than those recommended
by AASHTO. Therefore, the following are formulas that were
Moment Distribution, Multi-Lane Loading
Equation 6.4.2.2.3c-1 was developed to predict wheel load
distribution (distribution design width) for moment in slab bridges due
to multi-lane loading. Multiple presence factors are already accounted
for in the formula:
(6.4.2.2.3c-1)
E 2100 0.12 L
1
W
1
0.5
W
N
L
where:
E = the transverse distance over which a wheel line is
distributed
L
1
= L 18 000 mm
W
1
= W 18 000 mm
W = bridge width, in mm, edge-to-edge
Moment Distribution, Single-Lane Loading
The equation below predicts wheel load distribution for
moment due to single-lane loading.
(6.4.2.2.3c-2) E 250 0.42 (L
1
W
1
)
0.5
where the parameters are as defined in Equation 6.4.2.2.3c-1.
Equation 6.4.2.2.3b-6 may be used to account for the reduction
of moment in skewed bridges.
According to previous AASHTO Specifications, slab bridges
are adequate for shear if they are designed for moment. A quick
check of this assumption was made and it was concluded that it is a
valid assumption. Therefore, no formula or method is presented for
calculation of shear in slab bridges.
Lecture - 6-26
6.4.2.2.3d Simplified Formulas for Multi-Beam Decks which are
Sufficiently Interconnected to Act as a Unit
Only one formula, other than those presented in the previous
AASHTO Specifications, was obtained for load distribution in multi-
beam decks. This formula, developed by Arya at the University of
Illinois, is applicable to both box and open section multi-beam bridges
and predicts interior beam moment responses due to single-lane and
multi-lane loading. However, a number of simplified formulas
developed in the study are valid only for multi-box beam decks and do
not apply to open sections. Therefore, the response of multi-beam
decks made of open members, such as channels, may or may not be
accurately predicted by the formulas developed in that study.
The formula developed by Arya for interior girder load
distribution in multi-beam decks includes terms for the maximum
number of wheels that can be placed on a transverse section of the
bridge, number of beams, beam width and span length. A variation of
the formula was also proposed for multi-beam decks made of
channels, which includes consideration of the overall depth of the
channel section and its average thickness, defined as its area divided
by its length along the centerline of the thickness.
The following formula was developed in Project 12-26 to
predict load distribution factors for interior beam moment due to multi-
lane loading. The multiple presence reduction factor is already
accounted for in the formula.
(6.4.2.2.3d-1) g k
b
7600
0.6
b
L
0.2
I
J
0.06
where:
k = 2.5(N
b
)
0.2
1.5
b = beam width, in mm
N
b
= number of beams
I = moment of inertia of a beam (mm)
4
J = torsional constant of a beam (mm)
4
This formula is dependant on the inertia and torsional constant of a
beam; an estimated value for these properties must, therefore, be
used in preliminary design. The term I/J may be taken as unity for this
case.
Lecture - 6-27
Arya also presented a load distribution formula for multi-beam
decks designed for one traffic lane. The formulation and parameters
were similar to those presented for multi-lane loading. A variation of
that equation was also presented for calculation of the interior beam
moment distribution factor for a single-lane, channel section, multi-
beam deck. It should be noted that Arya's equations are not
applicable to cases of only one-lane loading with more than one traffic
lane.
A formula for wheel load distribution for moment in the interior
girders due to single-lane loading was also developed in NCHRP 12-
26. This formula is as follows:
(6.4.2.2.3d-2) g k
b
2.8L
0.5
I
J
0.25
All parameters are defined in Equation 6.4.2.2.3d-1. Equation
6.4.2.2.3c-2 is also dependent on inertia and torsional constants, and
a value of 1.0 may be used as an approximation for the term I/J during
preliminary design.
The moment in the edge girder due to multi-lane loading in
multi-beam decks comprised of box units is obtained by using a
correction factor applied to the interior girder distribution factors for
multi-lane loading. This correction factor may be found from the
following formula:
(6.4.2.2.3d-3) g 1.04
d
e
7600
where:
d
e
= distance from edge of the lane to the center of the exterior web
of the exterior girder, in mm
For exterior beams of sufficiently interconnected multi-beam
bridge decks comprised of T-shaped units subjected to multi-lane
loading, Equation 6.4.2.2.3b-2 applies.
For single-lane loading and for multi-beam decks comprised of
either box units or units other than box units, the lever rule is used.
Lecture - 6-28
Shear Distribution
Distribution factors for shear in interior girders of multi-beam
decks in "Bridge Decks Comprised of Box Units" due to multi-lane
loading may be calculated from the following formula:
(6.4.2.2.3d-4) g
b
4000
0.4
b
L
0.1
I
J
0.05
where the parameters are as defined in Equation 6.4.2.2.3d-1.
Distribution factors for shear in the interior girders of multi-
beam decks in "Bridge Decks Comprised of Box Units" due to single-
lane loading are obtained from the following formula:
(6.4.2.2.3d-5) g 0.70
b
L
0.15
I
J
0.05
where the parameters are again as defined in Equation 6.4.2.2.3d-1.
Note that Equations 6.4.2.2.3d-4 and 6.4.2.2.3d-5 are
dependent on inertia and torsional constants, and a value of 1.0 may
be used as an approximation for the term I/J during preliminary
design.
The shear in the edge girder of multi-beam deck in "Bridge
Decks Comprised of Box Units" due to multi-lane loading can be found
using a correction factor applied to interior girder distribution factors.
This correction factor is obtained from the formula:
(6.4.2.2.3d-6) e 1.02
d
e
15 000
where:
d
e
= distance from edge of lane to the center of exterior web
For shear in exterior beams of sufficiently interconnected multi-
beam bridge decks comprised of T-shaped units, Equations
6.4.2.2.3b-4 through 6.4.2.2.3b-5 and the lever rule should be used,
where appropriate.
The moment in any beam in a skewed bridge may be obtained
by using a skew reduction factor given by Equation 6.4.2.2.3b-6.
Lecture - 6-29
The shear in the interior beams of a skewed multi-beam bridge
comprised of box beams is usually of the same order as that of the
shear in the obtuse corner and must be obtained by applying a
correction factor to the response of the edge girder in a straight bridge.
This correction factor may be calculated from the formula:
(6.4.2.2.3d-7) r 1 c
1
(tan)
0.5
where:
(6.4.2.2.3d-8)
c
1
L
90d
6.4.2.2.3e Simplified Formulas for Multi-Beam Decks which are not
The LRFD Specification contains the same provisions for load
distribution in this type of bridge superstructure as appeared in recent
editions of the Standard Specification, and as repeated below for
completeness.
The key difference between bridges treated herein, as
compared to Section 6.4.2.2.3d, is the degree of transverse
interconnection of units. If box, T, channel or other precast units are
interconnected through a structural slab, or sufficiently transversely
post-tensioned to produce a similar level of continuity, then the
discussion of Section 6.4.2.2.3d applies. If the interconnection
between the units is expected to transmit shear, but relatively little
moment over the bridge service life, then the provisions herein apply.
The Specification provides for the computation of a bending
moment distribution factor, regardless of the number of lanes, given
by:
g
S
300D
for which:
D = 300 [11.5 - N
L
+ 1.4 N
L
(1 - 0.2C)
2
] when C 5
D = 300 [11.5 - N
L
] when C > 5
C = K (W/L)
K =
(1 )I
J
Lecture - 6-30
where:
= Poisson ratio
I = moment of inertia (mm)
4
J = St. Venant's constant (mm)
4
L = span length (mm)
N
L
= number of lanes
S = spacing of units (mm)
W = edge-to-edge width of bridge (mm)
6.4.2.2.3f Simplified Formulas for Spread Box Beam Bridges
Only one formula, other than those recommended by
AASHTO, was obtained from previous research for determining load
distribution factors in spread box beam bridges. This formula was
developed at Lehigh University for predicting the response of interior
beams due to multi-lane loading and was later adopted by AASHTO.
A correction factor for skewed bridges was also presented. In
addition, a number of simple formulas were developed as part of
NCHRP Project 12-26.
Moment Distribution to Interior Beams, Multi-Lane Loading
A formula developed in Project 12-26 for moment in interior
spread box beams due to multi-lane loading is as follows:
(6.4.2.2.3f-1)
g
S
1900
0.6
Sd
L
2
0.125
where
S = girder spacing (mm)
d = beam depth (mm)
Moment Distribution to Interior Beams, Single-Lane Loading
A similar formula was developed for distribution to interior
beams due to single-lane loading:
Lecture - 6-31
(6.4.2.2.3f-2)
g
S
910
0.35
Sd
L
2
0.25
where the parameters are as defined in Equation 6.4.2.2.3f-1.
The moment in edge girders due to multi-lane loading may be
calculated by applying a correction factor to the interior girder
distribution factor:
(6.4.2.2.3f-3) e 0.97
d
e
8700
where:
d
e
of the exterior girder (mm)
The distribution factor for moment in the edge girder due to
single-lane loading may be obtained by simple-beam distribution, i.e.,
the lever rule, in the same manner as was described for beam-and-
slab bridges.
Shear Distribution
multi-lane loading may be calculated from the following:
(6.4.2.2.3f-4) g
S
2250
0.8
d
L
0.1
where the parameters are as defined previously.
(6.4.2.2.3f-5) g
S
3050
0.6
d
L
0.1
where the parameters are again as defined previously.
The shear in the edge girder due to multi-lane loading can be
found by applying a correction factor to the interior girder equation.
This correction factor is:
Lecture - 6-32
(6.4.2.2.3f-6) e 0.8
d
e
3050
where:
d
e
The wheel load distribution factor for shear in the edge girder
due to single-lane loading may be obtained by simple-beam
distribution in the same manner as was described for beam-and-slab
bridges, i.e., the lever rule.
Research at Lehigh University also resulted in a formula for
correction of wheel load distribution for moment in interior girders due
to multi-lane loading in skewed bridges. NCHRP 12-26 concludes that
Equation 6.4.2.2.3b-6 was also applicable to this case.
The shear in the interior beams of a skewed bridge is the same
as that of a straight bridge. However, the shear in the obtuse corner
must be obtained by applying a correction factor to the distribution
factor for the edge girder in a straight bridge, given by Equation
6.4.2.2.3b-7, which C
1
is taken as:
(6.4.2.2.3f-7)
c
1
(Ld )
0.5
6S
6.4.2.2.3g Response of Continuous Bridges
The response of continuous bridges was studied by modeling
a number of two-span continuous bridges where each span is similar
to the average bridge. The wheel load distribution factor for each case
was compared to that of a simple bridge and correction factors for
continuity were obtained. In the case of beam-and-slab bridges, a
complete parameter study was performed, and it was found that the
correction factor is generally independent of bridge geometry. These
factors are given in the table below.
When the NCHRP 12-26 factors were incorporated into the
LRFD Specification, it was decided that 5% corrections were
unwarranted given that the distribution factors are an approximation
of actual behavior and are, therefore, subject to some variability. The
continuity correction for negative moment, a 10% increase, was also
neglected on the basis that experimentally observed "fanning" of the
reaction tends to reduce the negative moment as compared to a
typical beam calculation.
Lecture - 6-33
Table 6.4.2.2.3f-1 - Continuity Correction
Factors
Beam-and-Slab Bridges
Positive moment
Negative moment
Shear at simply-supported end
Shear at continuous bent
c = 1.05
c = 1.10
c = 1.00
c = 1.05
Box Girder Bridges
Positive moment
Negative moment
c = 1.00
c = 1.10
c = 1.00
c = 1.00
Slab Bridges
Positive moment
Negative moment
c = 1.00
c = 1.10
Multi-Beam Bridges
Positive moment
Negative moment
c = 1.00
c = 1.10
c = 1.00
c = 1.05
Spread box beam bridges
Positive moment
Negative moment
c = 1.00
c = 1.10
c = 1.00
c = 1.05
6.4.2.3 TRUSS AND ARCH BRIDGES
6.4.2.3.1 General
The approximate method for load distribution to lines of trusses
and arches is a so-called "lever rule", which is simply a matter of
summing moments about one line of trusses or arches to find the
reaction on the other line. This approach is illustrated by calculations
in Lecture 7.
Lecture - 6-34
6.5 REFINED METHODS
6.5.1 Deck Slabs
Where refined analysis of deck slabs is desirable, finite
element analysis is recommended. Elements should be chosen to
simulate both bending and in-plane or membrane effects. If the
analysis utilizes only plate/membrane or shell elements and has only
one or two elements through the thickness of the deck, then the
refined analysis will report an essentially bending-type response in the
deck. There has been much experimental and analytic work that
suggests that bending is not the primary source of strength in bridge
decks, but that the development of membrane action, analogous to a
shallow arch or dome load path within the deck is the primary source
of strength. This type of action will only be determined through a very
rigorous modeling of the deck.
6.5.2 Beam Slab Bridges
Relatively rigorous models of beam slab bridges can be
developed using general purpose commercial finite element programs,
finite strip programs or special purpose greater finite element-based
computer programs which have been specifically developed to simplify
the analysis of bridge-type structures. These more custom-oriented
programs often contain mesh generating capabilities, automatic load
placement capabilities and code checking.
Detailed bridge deck analysis using a finite element computer
program may be used to produce accurate results. However, extreme
care must be taken in preparation of the model, or inaccurate results
will be obtained. Important points to consider are selection of a
program capable of accurately modeling responses being investi-
gated, calculation of element properties, mesh density and support
conditions. Every model should be thoroughly checked to ensure that
nodes and elements are generated correctly.
Another important point is the loading. Truck loads should be
placed at positions that produce the maximum response in the
components being investigated. In many cases, the truck location is
not known before preliminary analysis is performed and, therefore,
many loadings should be investigated. This problem is more
pronounced in skewed bridges.
Many computer programs have algorithms that allow loads to
be placed at any point on the elements. If this feature is not present,
equivalent nodal loads must be calculated. Distribution of wheel loads
to various nodes must also be performed with care, and the mesh
should be fine enough to minimize errors that can arise because of
load approximations.
Many computer programs, especially the general purpose finite
element analysis programs, report stresses and strains, not shear and
moment values. Calculation of shear and moment values from the
Lecture - 6-35
stresses must be carefully performed, usually requiring an integration
over the beam cross-section. Some programs report stresses at node
points rather than Gaussian integration points. Integration of stresses
at node points is normally less accurate and may lead to inaccurate
results.
Many graphical and computer-based methods are available for
calculating wheel load distribution. One popular method consists of
design charts based on the orthotropic plate analogy, similar to that
presented in the Ontario Highway Bridge Design Code. As computers
become readily available to designers, simple computer-based
methods, such as SALOD, become more attractive than nomographs
and design charts. Also, grillage analysis presents a good alternative
to other simplified bridge deck analysis methods and will generally
produce more accurate results.
The grillage analogy may be used to model any one of the five
bridge types studied in this research. Each bridge type requires
special modeling techniques. A major advantage of plane grid
analysis is that shear and moment values for girders are directly
obtained and integration of stresses is not needed. Loads normally
need to be applied at nodal points, and it is recommended that simple-
beam distribution be used to distribute wheel loads to individual
nodes. If the loads are placed in their correct locations, the results will
be close to those of detailed finite element analysis.
As indicated previously, the designer has to be responsible for
constructing a suitable model and determining that the results are
accurate. It is possible to make seemingly small errors in computer
models which can have dramatic effects on the results which are
obtained.
6.5.3 Example of Modeling Errors
The modeling of diaphragms and boundary conditions at
supports and bearings is vital to obtaining the proper results when
using these sophisticated programs. The burden of correctly handling
these factors rests with the designer. Consider the following example
which shows how a very small modeling error produced very
erroneous results.
The framing plan shown on Figure 6.5.3-1 represents an actual
bridge that was designed using a grid-type approach. The designer
had a good model for this structure, except that the rotational degree
of freedom corresponding to the global "x" axis was fixed at all of the
bearings. This did not allow the diaphragms at the piers and
abutments to respond correctly to the imposed loadings and
deformations, and also had the effect of producing artificially stiff ends
on the girders by virtue of vector resolution between global and local
systems. The effect of this condition on the reactions obtained at the
abutments and piers was dramatic. Modest uplift was reported at the
acute corner along the near abutment shown in Figure 6.5.3-1, and a
very substantial uplift was reported at the acute angle at the far
Lecture - 6-36
abutment. This is shown in the table in Figure 6.5.3-1, as is a moment
diagram for non-composite dead loads which reflects the incorrect
reactions. Also shown in the table of reactions on Figure 6.5.3-1 are
the correct reactions determined when the structure was modeled
using the generic STRESS Computer Program with proper boundary
conditions at the supports. In this case, a positive reaction is found at
all bearings, and a significantly different moment diagram for non-
composite dead load also resulted. The correct reactions and moment
diagram are also shown on Figure 6.5.3-1.
The modeling of the degrees of freedom at the lines of support
on this structure was also investigated utilizing a relatively complete
three-dimensional finite element analysis and the SAPIV Computer
Program. The model used is illustrated in Figure 6.5.3-2, which shows
how the deck slab, girders and cross-frames were modeled in their
proper relative positions in the cross-section which extended along the
bridge from end-to-end. Also shown on this figure is a comparison of
the reactions obtained from STRESS and from SAPIV by applying all
of the non-composite loads in a single loading. The comparison
between these reactions is excellent.
In order to verify that the order of pouring the deck slab units
would not contribute to an uplift situation, the pouring sequence was
replicated in a three-dimensional SAPIV analysis. The results of the
analysis of the three stages of the pouring sequence are also shown
in Figure 6.5.3-2, as well as the total accumulated load at the end of
the pour. Comparison of the sequential loading with the application of
a single loading of non-composite dead load also showed relatively
good agreement in this case.
The important point demonstrated in the example of Figures
6.5.3-1 and 6.5.3-2 is that seemingly small errors in modeling of the
structure can result in very substantial changes in the reactions,
shears and moments. The designer must be aware of this potential
when utilizing the more refined analysis techniques.
Incidentally, there are cases in which an uplift reaction due to
skew and/or curvature is possible. The simple span bridge shown in
Figure 6.5.3-3 and reported on in the November 1, 1984, issue of
Engineering News-Record, was analyzed at the request of the owner.
In this case, the uplift reactions computed by the designer were
verified.
Sometimes modeling problems occur because User's Manuals
are not clear, or a "bug" exists, of which the program's author/vendor
is not aware. Such a case is illustrated for the simply-supported,
partially-curved and skewed bridge in Figure 6.5.3-4. Initially, this
bridge was modeled with extra joints at locations other than
diaphragms in an effort to improve live load determination. As a
result, the number of points along each girder were not equal, but
there was no indication in the program's descriptive literature that this
type of arrangement of program input could potentially cause a
problem. The resulting moment envelopes for the middle and two
Lecture - 6-37
exterior girders are shown in Figure 6.5.3-4, and are obviously
unusual in shape and also in their order of maximum moment, i.e., #4,
#5 and #3.
It was found that the live load processor was not responding
properly to the unequal number of nodes per girder, that nodes should
be essentially "radial", and that it was not certain that nodes to which
diaphragms were not connected were legitimate. The revised model,
shown in Figure 6.5.3-4, produced clearly better results, as shown in
the indicated moment envelopes.
Lecture - 6-38
Figure 6.5.3-1 - Framing plan, comparative live load reactions and
moment envelopes showing effect of proper and improper rotational
boundary condition, as reflected in grid analysis.
Lecture - 6-39
Figure 6.5.3-2 - Finite element idealization and reactions obtained for
structure shown in Figure 6.
Lecture - 6-40
Figure 6.5.3-3 - Framing Plan of Curved Span with Skewed Piers
Figure 6.5.3-4 - Comparative moment envelopes for the middle and
two outside girders of curved skewed system showing the results of
apparent "bug" in algorithm for applying live load.
6.5.4 Other Types of Bridges
The Specification contains additional requirements for the
rigorous analysis of cellular-type structures, truss bridges, arch
bridges, cable-stayed bridges and suspension bridges. Generally
speaking, refined analysis will involve a computer model which
accurately affects the geometry, relative component stiffnesses,
Lecture - 6-41
boundary conditions and load supply to the structure. Suspension
bridges will almost always be analyzed using a large deflection theory.
The deflection theory may also be applied to arches and cable-stayed
bridges. In the case of the cable-stayed bridge of moderate span, it
may be sufficiently accurate to evaluate the second order effects on
the deck system of the tower by supplementary calculations in
providing a correction factor, developed on a bridge-specific basis.
The change in stiffness of the cables caused by change in sag as the
cable load changes can be accounted for using the so-called "Ernst"
equations, given in the Specification, for modified modulus of
elasticity.
Lecture - 6-42
REFERENCES
Jones, 1976, "A Simple Algorithm for Computing Load Distribution in
Multi-Beam Bridge Decks", Proceedings, 8th ARRB Conference, 1976
APPENDIX A
The Load Distribution Problem and its Solution in NCHRP 12-26
Lecture - 6-A1
Older editions of the AASHTO Specifications allow for simplified analysis of bridge
superstructures using the concept of a load distribution factor for bending moment in interior girders
of most types of bridges, i.e., beam-and-slab, box girder, slab, multi-box beam and spread-box
beam. This distribution factor is given by:
(A-1)
g
S
300D
where:
g = a factor used to multiply the total longitudinal response of the bridge due to a single
longitudinal line of wheel loads in order to determine the maximum response of a
single girder
S = the center-to-center girder spacing (mm)
D = a constant that varies with bridge type and geometry
A major shortcoming of the previous specifications is that the piecemeal changes that have
taken place over the last 55 years have led to inconsistencies in the load distribution criteria
including: inconsistent consideration of a reduction in load intensity for multiple lane loading;
inconsistent changes in distribution factors to reflect the changes in design lane width; and,
inconsistent approaches for verification of live load distribution factors for various bridge types.
The past AASHTO simplified procedures were developed for non-skewed, simply-supported
bridges. Although it was stated that these procedures apply to the design of normal (i.e., supports
oriented perpendicular to the longitudinal girders) highway bridges, there are no other guidelines
for determining when the procedures are applicable. Because modern highway and bridge design
practice requires a large number of bridges to be constructed with skewed supports, on curved
alignments, or continuous over interior supports, it is increasingly important that the limitations of
load distribution criteria be fully understood by designers.
Advanced computer technology has become available in recent years which allows detailed
finite element analysis of bridge decks. However, many computer programs exist which employ
different formulations and techniques. It is important that the computer methodology and
formulation that produces the most accurate results be used to predict the behavior of bridge decks.
In order to identify the most accurate computer programs, data from full-scale and prototype bridge
load tests were compiled. The bridge tests were then modeled by different computer programs and
the experimental and computer results were compared. The programs that produced the most
accurate results were then considered as the basis for evaluation of the other method levels, i.e.,
Levels 2 and 1 methods.
An important part of the development or evaluation of simplified methods is range of
applicability. In order to ensure that common values of various bridge parameters were considered,
a database of actual bridges was compiled. Bridges from various states were randomly selected
in order to achieve national representation. This resulted in a database of 365 beam-and-slab
bridges, 112 prestressed concrete and 121 reinforced concrete box girder bridges, 130 slab
bridges, 67 multi-box beam bridges and 55 spread-box beam bridges. This bridge database was
studied to identify the common values of various parameters, such as beam spacing, span length,
slab thickness, and so on. The range of variation of each parameter was also identified. A
hypothetical bridge that has all the average properties obtained from the database, referred to as
Lecture - 6-A2
the "average bridge" was created for each of the beam-and-slab, box girder, slab, multi-box beam
and spread-box beam bridge types. For the study of moment responses in box girder bridges,
separate reinforced concrete and prestressed concrete box girder average bridges were also
prepared.
In evaluating simplified formulas, it is important to understand the effect of various bridge
parameters on load distribution. Bridge parameters were varied one at a time in the average bridge
for the bridge type under consideration. Load distribution factors for both shear and moment were
obtained for all such bridges. Variation of load distribution factors with each parameter shows the
importance of each parameter. Simplified formulas can then be developed to capture the variation
of load distribution factors with each of the important parameters. A brief description of the method
used to develop such formulas is as follows.
In order to derive a formula in a systematic manner, certain assumptions must be made.
First, it is assumed that the effect of each parameter can be modeled by an exponential function
of the form ax
b
, where x is the value of the given parameter, and a and b are coefficients to be
determined based on the variation of x. Second, it is assumed that the effects of different
parameters are independent of each other, which allows each parameter to be considered sep-
arately. The final distribution factor will be modeled by an exponential formula of the form: g =
(a)(S
b1
)(L
b2
)(t
b3
)(...) where g is the wheel load distribution factor; S, L, and t are parameters included
in the formula; a is the scale factor; and b1, b2, and b3 are determined from the variation of S, L,
and t, respectively. Assuming that for two cases, all bridge parameters are the same, except for
S, then:
g
1
= (a)(S
1
b1
)(L
b2
)(t
b3
)(...) (A-2)
g
2
= (a)(S
2
b1
)(L
b2
)(t
b3
)(...) (A-3)
therefore:
(A-4)
g
1
g
2
S
1
S
2
b1
or:
(A-5)
b1
ln
g
1
g
2
ln
S
1
S
2
If n different values of S are examined and successive pairs are used to determine the value
of b1, n 1 different values for b1 can be obtained. If these b1 values are close to each other, an
exponential curve may be used to accurately model the variation of the distribution factor with S.
In that case, the average of n 1 values of b1 is used to achieve the best match. Once all the power
factors, i.e., b1, b2, and so on, are determined, the value of "a" can be obtained from the average
bridge, i.e.,
Lecture - 6-A3
(A-6)
a
g
o
S
b1
o
L
b2
o
t
b3
o
(...)
This procedure was followed during the entire course of the NCHRP 12-26 study to develop
new formulas as needed. In certain cases where an exponential function was not suitable to model
the effect of a parameter, slight variation from this procedure was used to achieve the required
accuracy. However, this procedure worked quite well in most cases and the developed formulas
demonstrate high accuracy.
Because certain assumptions were made in the derivation of simplified formulas and some
bridge parameters were ignored altogether, it is important to verify the accuracy of these formulas
when applied to real bridges. The database of actual bridges was used for this purpose. Bridges
to which the formula can be applied were identified and analyzed by an accurate method. The
distribution factors obtained from the accurate analysis were compared to the results of the
simplified methods. The ratio of the approximate to accurate distribution factors was calculated and
examined to assess the accuracy of the approximate method. Average, standard deviation, and
minimum and maximum ratio values were obtained for each formula or simplified method. The
method or formula that has the smallest standard deviation is considered to be the most accurate.
However, it is important that the average be slightly greater than unity to assure slightly
conservative results. The minimum and maximum values show the extreme predictions that each
method or formula produced when a specific database was used. Although these values may
change slightly if a different set of bridges is used for evaluation, the minimum and maximum values
allow identification of where shortcomings in the formula may exist that are not readily identified by
the average or standard deviation values.
It was previously mentioned that different subsets of the database of bridges were used to
evaluate different formulas. When a subset included a large number of bridges (100 or more), a
Level 2 method was used as the basis of comparison. When it included a smaller number of
bridges (less than 100), a Level 3 method was used. As a result, LANELL (an influence surface
method) was used for verification of formulas for moment distribution in box girder bridges, and a
Multi-dimensional Space Interpolation (MSI) method was used for verification of formulas for
straight beam-and-slab and slab bridges.
Findings
Level 3 Methods: Detailed Bridge Deck Analysis
Recent advances in computer technology and numerical analysis have led to the
development of a number of computer programs for structural analysis. Programs that can be
applicable to bridge deck analysis can be divided into two categories. One includes general
purpose structural analysis programs such as SAP, STRUDL and FINITE. The other category is
specialized programs for analysis of specific bridge types, such as GENDEK, CURVBRG and
MUPDI.
In the search for the best available computer program for analysis of each bridge type, all
suitable computer programs (general and specific) that were available at the time of the 12-26
research were evaluated. In order to achieve meaningful comparisons and assess the level of
accuracy of the programs, a number of field and laboratory tests were modeled by each program.
The results were then compared in three ways:
by visual comparison of the results plotted on the same figure,
Lecture - 6-A4
by comparison of the averages and standard deviations of the ratios of analytical to
experimental results, and
by comparison of statistical differences of analytical and experimental results. Five bridge
types were considered: beam-and-slab, box girder, slab, multi-box beam, and spread-box
beam.
For analysis of beam-and-slab bridges, the following computer programs and models were
evaluated: GENDEK-PLATE, GENDEK-3, GENDEK-5, CURVBRG, SAP and MUPDI. It was found
that, in general, GENDEK-5 analysis using plate elements for the deck slab and eccentric beam
elements for the girders is very accurate. This program is also general enough to cover all typical
cases, i.e., straight, skew, moment and shear. However, for analysis of curved open girder steel
bridges, CURVBRG was the most accurate program. MUPDI was also found to be a very accurate
and fast program; however, skewed bridges cannot be analyzed with this program and shear values
near the point of application of load, or near supports, lack accuracy. GENDEK-5 was, therefore,
selected to evaluate Level 2 and Level 1 methods.
For analysis of box girder bridges, computer programs MUPDI, CELL-4 and FINITE were
evaluated. MUPDI was the fastest and most practical program for analysis of straight bridges for
moment, but FINITE was found to be the most practical program for skewed bridges and for
obtaining accurate shear results. Therefore, MUPDI was selected for the evaluation of LANELL (a
Level 2 method for moment in straight bridges which was, in turn, used for evaluation of Level 1
methods) and FINITE was selected for other cases.
For the analysis of slab bridges, computer programs MUPDI, FINITE, SAP and GENDEK
were evaluated. Shear results cannot be obtained accurately in slab bridges and, therefore, were
not considered. The GENDEK-5 program, without beam elements, proved to be very accurate.
However, MUPDI was found to be the most accurate and practical method for non-skewed
prismatic bridges and was selected to evaluate Level 2 and Level 1 methods.
For the analysis of multi-beam bridges, the following computer programs were evaluated:
SAP, FINITE and a specialized program developed by Professor Powell at the University of
California, Berkeley, for analysis of multi-beam bridges (referred to as the POWELL program
herein). Various modeling techniques were studied using different grillage models and different
plate elements. The program that is capable of producing the most accurate results in all cases,
i.e., straight and skewed for shear and moment, was the FINITE program. This program was later
used in evaluation of more simplified methods. POWELL is also very accurate in reporting
moments in straight bridges, but it uses a finite strip formulation, similar to MUPDI, and, therefore,
is incapable of modeling skewed supports, and shear results near supports and load locations can-
not be accurately obtained. This program was used to evaluate simplified methods for straight
bridges.
For analysis of spread-box beam bridges, computer programs SAP, MUPDI, FINITE and
NIKE-3D were evaluated. FINITE produced the most accurate results, especially when shear was
considered. MUPDI was selected to evaluate simplified methods for calculation of moments in
straight bridges, and FINITE was selected for all other cases.
Level 2 Methods: Graphical and Simple Computer-Based Analysis
Nomographs and influence surface methods have traditionally been used when computer
methods have been unavailable. The Ontario Highway Bridge Design Code uses one such method
based on orthotropic plate theory. Other graphical methods have also been developed and
reported. A good example of the influence surface method is the computer program SALOD
Lecture - 6-A5
developed by the University of Florida for the Florida Department of Transportation. This program
uses influence surfaces, obtained by detailed finite element analysis, which are stored in a
database accessed by SALOD. One advantage of influence surface methods is that the response
of the bridge deck to different truck types can be readily computed.
A grillage analysis using plane grid models can also be used with minimal computer
resources to calculate the response of bridge decks in most bridge types. However, the properties
for grid members must be calculated with care to assure accuracy. Level 2 methods used to
analyze the five bridge types (beam-and-slab, box girder, slab, multi-beam and spread-box beam
bridges) are discussed below.
The following methods were evaluated for analysis of beam-and-slab bridges: plane grid
analysis, the nomograph-based method included in the Ontario Highway Bridge Design Code
(OHBDC), SALOD and Multi-dimensional Space Interpolation (MSI). All of these methods are
applicable for single- and multi-lane loading for moment. The OHBDC curves were developed for
a truck other than HS-20, and using the HS-20 truck in the evaluation process may have introduced
some discrepancies. The method presented in OHBDC was also found to be time consuming, and
inaccurate interpolation between curves was probably a common source of error. SALOD can be
used with any truck and, therefore, the "HS" truck was used in its evaluation. The MSI method was
developed based on HS-20 truck loading for single- and multiple-lane loading. MSI was found to
be the fastest and most accurate method and was, therefore, selected for the evaluation of Level
1 methods. This method produces results that are generally within 5% of the finite element
(GENDEK) results.
In the analysis of box girder bridges, OHBDC curves and the LANELL program were
evaluated. The comments made about OHBDC for beam-and-slab bridges are valid for box girder
bridges as well. As LANELL produced results that were very close to those produced by MUPDI,
it was selected for evaluation of Level 1 methods for moment.
OHBDC, SALOD and MSI were evaluated for the analysis of slab bridges. MSI was found
to be the most accurate method and, thus, was used in the evaluation of Level 1 methods. SALOD
also produced results that were in very good agreement with the finite element (MUPDI) analysis.
Results of OHBDC were based on a different truck and, therefore, do not present an accurate
evaluation.
In the analysis of multi-beam bridges, a method presented in Jones, 1976, was evaluated.
The method is capable of calculating distribution factors due to a single concentrated load and was
modified for this study to allow wheel line loadings. The results were found to be in very good
agreement with POWELL. However, because this method was only applicable for moment
distribution in straight single-span bridges, it was not used for verification of Level 1 methods.
In the analysis of spread-box beam bridges, only plane grid analysis was considered as a
Level 2 method.
In general, Level 2 graphical and influence surface methods generated accurate and
dependable results. While these methods are sometimes difficult to apply, a major advantage of
some of them is that different trucks, lane widths, and multiple presence live load reduction factors
may be considered. Therefore, if a Level 2 procedure does not provide needed flexibility, its use
is not warranted because the accuracy of it is on the same order as a simplified formula. MSI is an
example of such a method for calculation of load distribution factors in beam-and-slab bridges.
A plane grid analysis would require computer resources similar to those needed for some
of the methods mentioned above. In addition, a general purpose plane grid analysis program is
Lecture - 6-A6
available to most bridge designers. Therefore, this method of analysis is considered a Level 2
method. However, the user has the burden of producing a grid model that will produce sufficiently
accurate results. As part of NCHRP Project 12-26, various modeling techniques were evaluated,
and it was found that a proper plane grid model may be used to accurately produce load distribution
factors for each of the bridge types studied.
Level 1 Methods: Simplified Formulas
The current AASHTO Specifications recommend use of simplified formulas for determining
load distribution factors. Many of these formulas have not been updated in years and do not
provide optimum accuracy. A number of other formulas have been developed by researchers in
recent years. Most of these formulas are for moment distribution for beam-and-slab bridges
subjected to multi-lane truck loading. While some have considered correction factors for edge
girders and skewed supports, very little has been reported on shear distribution factors or
distribution factors for bridges other than beam-and-slab.
The sensitivity of load distribution factors to various bridge parameters was also determined
as part of the study. In general, beam spacing is the most significant parameter. However, span
length, longitudinal stiffness and transverse stiffness also affect the load distribution factors.
Figures 6.4.2.2.3-2 through 6.4.2.2.3-6 show the variation of load distribution factors with various
bridge parameters for each bridge type. Ignoring the effect of bridge parameters, other than beam
spacing, can result in highly inaccurate (either conservative or unconservative) solutions.
A major objective of the research in Project 12-26 was to evaluate older AASHTO
Specifications and other researchers' published work to assess their accuracy and develop
alternate formulas whenever a more accurate method could be obtained. The formulas that were
evaluated and developed are briefly described below, according to bridge type; i.e., beam-and-slab,
box girder, slab, multi-beam and spread-box beam.
Figure A-1 - Effect of Parameter Variation on Beam-and-Slab Bridges
Lecture - 6-A7
Figure A-2 - Effect of Parameter Variation on Box Girder Bridges
Figure A-3 - Effect of Parameter Variation on Slab Bridges
Figure A-4 - Effect of Parameter Variation on Multi-Box Beam Bridge
Lecture - 6-A8
Figure A-5 - Effect of Parameter Variation on Spread-Box Beam Bridges
Lecture - 6-1
LECTURE 6 - ANALYSIS I
The objectives of this lesson are to acquaint the student with:
the various analysis techniques required and/or recommended
for determining the force effects and components of bridges,
and
the use of approximate and refined methods for the
determination for force effects in conventional girder-type
structures.
The background on the development of new, improved,
distribution factors which were developed under NCHRP Project 12-26
has been included for reference in an Appendix.
The use of grid and finite element types of analysis for multi-
beam bridges is also recommended in the LRFD Specification. These
methods require considerable care in structural modeling, and several
examples of the large effects of seemingly small errors in structural
models will be presented.
6.2 ACCEPTABLE METHODS OF STRUCTURAL ANALYSIS
Article S4.4 contains a list of methods of analysis that are
considered suitable for analysis of bridges. These include the
classical force and displacement methods, such as virtual work,
moment distribution, slope deflection, the so-called general method,
the more modern finite element, finite strip and plate analogy-type
methods, analysis based on series expansions and the yield-line
method for the non-linear analysis of plates and railings. Some of
these methods of analysis are suitable for hand calculations, but for
any problem of large size, some sort of computer solution will almost
always be required for practical design purposes. This is because
almost all of these methods, with the possible exception of the series
methods and the yield-line methods, will eventually require the
solution of large sets of simultaneous equations. The series method,
while elegant from a mathematical point of view, will typically require
a computer program to expand the series sufficiently to yield good
results in a practical time frame. Yield-line methods, which could be
considered the extension of plastic design to two-dimensional
surfaces, are typically a hand calculation procedure.
The use of computer programs in bridge design brings up a
philosophical problem as to the responsibility for error. Almost all
vendors of commercial computer programs disavow any responsibility
for error. A release from liability is usually implicit in their use and may
even be an explicit part of obtaining a license. This means that an
organization using a computer program must be relatively certain of
Lecture - 6-2
the results that it obtains. It is not necessary for every engineer in a
large design section to have personally confirmed every computer
program, but it is necessary that some verification testing be done or
that the results of previous verification testing be obtained in order to
produce the required level of confidence. Computer programs can be
verified against universally accepted closed- form solutions, other
computer programs which have been previously verified, or the results
of testing.
Many computer programs for design use also contain code
checking capabilities. Others have portions of the applicable design
specification embedded in the coding of the program. In order to
identify the specification edition which may have been tied to a given
release of a program and also to provide a means for determining
which structures may have been designed with a version of a program
later found to contain errors, the specification requires that a name,
version and a release date of software be identified in the contract
drawing.
6.3 PRINCIPLES OF MATHEMATICAL MODELING
6.3.1 Structural Material Behavior
The LRFD Specification recognizes both elastic and inelastic
behavior of materials for analysis purposes. Inelastic material
behavior is implicit in many of the equations and procedures specified
for the calculation of cross-sectional resistance, such as calculating
the nominal resistance of a concrete beam or column, the nominal
plastic moment resistance of a compact and adequately braced steel
cross-section, or the bearing capacity of a spread footing. Often, the
force effects to which this resistance will be compared will be
calculated on a basis of a linear structural analysis with elastic
material properties having been assumed. This dichotomy has existed
in the bridge specification since the introduction of load factor design
in the early 1970's. It continues through the LRFD Specification.
On the other hand, there are certain assumed failure modes at
extreme events and the use of mechanism and unified autostress
design procedures for steel girders, where permitted, which require
analysis based on non-linear behavior. Many times, this analysis will
take a form analogous to plastic design of steel frames. For example,
seismic design may be based on the formation of plastic hinges at the
top and bottom of the columns of a bent. Ship collision forces may be
absorbed in a comparable inelastic manner. Furthermore, it is
anticipated that future seismic design provisions will be based on
extensive research currently underway to develop a step-by-step non-
linear force displacement relationship for the lateral displacement of
piers.
Where inelastic analysis is used, the Designer must be certain
that a ductile failure mode is obtained through proper detailing. Rules
Lecture - 6-3
for achieving this are presented in the sections for steel and concrete
design.
6.3.2 Geometry
6.3.2.1 GENERAL
Most analyses done for the purpose of designing bridges are
based on the assumption that the displacements caused by the loads
are relatively small and, therefore, it is suitably accurate to base the
calculations on the undeformed shape. This is typically referred to as
the small deflection theory, and it is routinely used in the design of
beam-type structures and bridges which resist loads through a couple
whose tensile and compressive forces remain essentially in fixed
positions relative to each other while the bridge deflects. This will be
the case for a truss and for tied-arches.
For other types of structures and components and for certain
types of analysis, the effect of the deflections must be considered in
the development of the force equilibrium equations, i.e., the equations
of equilibrium are written for the displaced shape. Almost all
engineers are aware that the study of structural stability requires
consideration of the displaced shape, in fact, if the displaced shape is
not part of the original formulation of the problem, one would never be
able to determine that a column, shell or plate can buckle. Consider
for a moment the simple pin-ended column. Unless the deflected
shape of the column is taken into account, the moment caused by the
axial load acting on the displaced shape would not be accounted for.
It is this moment which causes the column to move laterally, i.e., to
buckle.
Almost a century ago, it was found that the only reasonably
accurate way to calculate force effects in suspension bridges of any
size was to include the deflection of the cable in the formulation of the
problem and, therefore, the displacement of a stiffening truss or
stiffening girder. As conventional, i.e., not tied, arches became longer
and more slender, an effect directly analogous to that observed in
suspension bridges can become significant enough that it must be
accounted for in the design of the arch rib. In fact, because the arch
rib is in compression and can buckle, the effect of large deflections
can be especially important.
Finally, the compression members of frames in bents can also
be susceptible to this phenomenon.
Where non-linear effects arriving either out of material non-
linearity or large deflections become significant, then super position of
forces does not apply. This means that each load case under
investigation must be studied separately under the full effect of all of
the factored loads that make up the load combination under study.
This is a very significant effect on most practical design calculations.
Commonly, a Bridge Engineer calculates the force effect from a
variety of individual loads and then combines, or superimposes, the
Lecture - 6-4
force effects calculated for each individual load to make up whatever
group combination of loadings are needed. For non-linear analysis,
each combination must be investigated, i.e., analyzed, individually.
6.3.2.2 APPROXIMATE METHODS
To simplify analysis and to partially bypass the need to analyze
each load combination separately, as identified above, certain
approximate methods have been developed to allow the designer to
add a correction to a set of force effects calculated in a linear manner.
These are sometimes called single-step adjustment methods, the
most commonly used of which is moment magnification for beam
columns, which has been part of the AASHTO Specifications since the
early 1970's.
For beam columns, the moment magnification process is given
by the equations below.
M
c
=
b
M
2b
+
s
M
2s
(6.3.2.2-1)
f
c
=
b
f
2b
+
s
f
2s
(6.3.2.2-2)
for which:
(6.3.2.2-3)
b
C
m
1
P
u
P
e
1.0
(6.3.2.2-4)
s
1
1
P
u
P
e
where:
P
u
= factored axial load (N)
P
e
= Euler buckling load (N)
= resistance factor for axial compression as specified in
Specification Sections 5, 6 and 7, as applicable
M
2b
gravity loads that result in no appreciable sidesway
calculated by conventional first order elastic frame
analysis, always positive (N mm)
f
2b
2b
(MPa)
M
2s
lateral or gravity loads that result in sidesway, ,
Lecture - 6-5
greater than
u
/1500, calculated by conventional first
order elastic frame analysis, always positive (N mm)
f
2s
2s
(MPa)
It may appear that the moment magnification factor contains the Euler
buckling load, P
e
. However, P
e
is only a convenient substitution for a
group of terms related to the displacement of the beam column.
A derivation of the moment magnification equation can be
found in many textbooks on steel or concrete design.
For cases where the shape of the beam column is expected to
be radically different from that given by the simply-supported case, or
the loads significantly different from those indicated above, then it is
possible to make an adjustment to account for a different initial elastic
shape through the factor c
m
.
The moment magnification procedure has also been extended
to arches, and this has been available in the AASHTO Specifications
for many years and is reproduced as Article S4.5.3.2.2c with no further
refinement.
6.3.2.3 REFINED METHODS
The effect of large deflections can also be rigorously
accounted for through iterative solutions of equilibrium equations,
taking into account updated positions of the structure, or by using
geometric stiffness terms. In some cases, e.g., the suspension bridge,
solutions are available to the differential equations of equilibrium which
can be solved in a trial and error fashion, or through series expansion.
6.3.3 Modeling Boundary Conditions
Points of expansion or other forms of articulation in the
structure are commonly idealized as frictionless units. Where past
practice indicates that this has been a reasonable conservative
approach, continued use is warranted. There are other instances
where the potential for nonfunctional expansion devices and/or the
possibility that joints may close should also be investigated. This
might be the case, for example, in a seismic analysis where analysis
may be made, assuming that expansion joints are operable and open,
and then another analysis might be made, assuming that they are
closed and nonfunctional in order to simulate, or bound, the effects of
joints reaching the limits of travel during the earthquake. The
possibility of reaching the limit of expansion travel should also be
investigated when evaluating non-linear effects on substructure
elements. It may be possible that the amount of moment
magnification may be reduced because expansion dams will close,
jamming the structure against the abutments before the full movement
implicit in the moment magnification factor can be reached. This will
reduce the moment magnification factor and, hence, the design
moment.
Lecture - 6-6
Similarly, the effect of boundary conditions at foundation units
should also be evaluated. Foundation units are seldom fully fixed or
fully pinned, and an evaluation of the potential movement of a
foundation unit may be necessary in order to properly assess
response, as well as secondary moments caused by change in
geometry. Here again, bounding of the range of probable movement
may be the only practical way to attack such a problem.
6.4 STATIC ANALYSIS
6.4.1 The Influence of Plan Geometry
Article S4.6.1 deals with two simplifications which can be made
based on the plan geometry of the superstructure.
The first simplification involves the possibility of replacing the
superstructure for analysis purposes with a single-line element called
a spine beam. This may be done when the transverse distortion of the
superstructure is small in comparison with the longitudinal
deformation. Generally, if the superstructure is a torsionally stiff
closed section or sections whose length exceeds 2.5 times their width,
it may be idealized as a line element whose dimensions may be
determined as given in the Specification. This can be used to
significantly simplify analysis models.
The second simplification deals with when it is possible to
consider curved superstructures as straight for the purpose of
analysis. If the superstructure is a torsionally stiff closed section and
the central angle of a segment between piers is less than 12 , then
the segment may be considered straight. If the superstructure is
made of torsionally weak open sections, then the effects of curvature
may be neglected when the subtended angle is less than that given in
Table 6.4.1-1.
Table 6.4.1-1 - Limiting Central Angle
for Neglecting Curvature in Determining
Primary Bending Moments
No. of Beams
Angle for One
Span
Angle for
Multiple Spans
2 2 3
3 or 4 3 4
5 or more 4 5
Lecture - 6-7
6.4.2 Approximate Methods for Load Distribution
6.4.2.1 DECK SLABS AND SLAB-TYPE BRIDGES
The Specification permits the approximate analysis of deck
slabs by analyzing a strip of deck as a continuous beam. Provisions
are made for determining the width of that strip at
the unsupported edge of the slab and at points interior from the edges.
If the spacing of supporting components in the secondary
direction exceeds 1.5 times the spacing in the primary direction, then
all of the wheel loads applied to the deck are considered to be applied
to the primary strip. The secondary strip is designed on a basis of
percentage of reinforcement in the primary strip.
If the spacing of the supporting components on the secondary
direction is less than 1.5 times that in the primary direction, then a
crossed sticks analogy is used. The width of the equivalent strips in
each direction is provided by Table S4.6.2.1.3-1 and the wheel load
is distributed between two idealized intersecting strips according to the
relative stiffness of each strip.
Once the wheel loads have been assigned to the strips, for
either case identified above, then the force effects are calculated
based on a continuous beam. For the purpose of analyzing the
continuous beam, the span length of each span is taken as a center-
to-center of supporting components. For the purpose of calculating
moment and shear at a design section, some offset from the
theoretical center of support is permitted as given in the Specification.
Decks which form an integral part of a cellular cross-section
are supported on webs which are monolithic with the deck. Therefore,
when the deck rotates, the web of the box girder rotates giving rise to
bending stresses throughout the cross-section. For the purpose of
analyzing this effect, a cross-sectional frame action procedure is
identified in the Specification.
In the case of fully filled and partially filled grids, the results of
recent research are incorporated in LRFD Article S4.6.2.1.8 to given
bending moments per unit length of grid.
6.4.2.2 BEAM SLAB BRIDGES
6.4.2.2.1 General
The Specification provides a series of empirical rules for
assigning portions of a design lane to a supporting component. These
are commonly called distribution factors. It is important to remember
that the approximate distribution factors, specified in the LRFD
Specification, are on a lane, i.e., axle basis, not a wheel basis. The
distribution factors are given for the various kinds of bridges shown in
Figure 6.4.2.2.1-1.
Lecture - 6-8
SUPPORTING
COMPONENTS TYPE OF DECK
TYPICAL
CROSS-
SECTION
Steel Beam - Revised
Factors
precast concrete slab, steel grid,
glued/spiked panels, stressed
wood
Closed Steel or Precast
Factors
Cast-in-place concrete slab
Open Steel or Precast
Factors
precast concrete deck slab
Cast-in-Place Concrete Multi-
cell Box - Revised Factors
Monolithic concrete
Cast-in-Place Concrete Tee
Beam - Revised Factors
Monolithic concrete
Cellular Concrete Boxes with
Shear Keys - Revised
Factors
Cellular Concrete Box with
Shear Keys and with or
without Transverse
Post-Tensioning - Revised
Factors (in some cases)
Integral concrete
Precast Concrete Channel
Sections with Shear Keys
Precast Concrete Double Tee
Post-Tensioning
Integral concrete
Precast Concrete Tee
Post-Tensioning
Integral concrete

Concrete I or Bulb-Tee
Sections - Revised Factors
Cast-in-place concrete, precast
concrete
Wood Beams - Revised
Factors
Cast-in-place concrete or plank,
glued/spiked panels or stressed
wood
Figure 6.4.2.2.1-1 - Common Deck Superstructures Covered in LRFD Specification
Articles 4.6.2.2.2 and 4.6.2.2.3
Lecture - 6-9
Some of the distribution factors are new to the Specification as a result
of an extensive project on load distribution known as NCHRP Project
12-26. Where the distribution factors for a given type of cross-section
have been developed under that project and are new to the
Specification, the words "revised factors" appear in the column
identified as "supporting components". Where those words do not
appear, the distribution factors have been retained from earlier
editions of the AASHTO Standard Specifications.
Some simplifications have been made in utilizing the
distribution factors from NCHRP 12-26. In particular, correction
factors for various aspects of structural action, typically involving
continuity, which were less than 5%, were omitted from the LRFD
Specifications. Similarly, an increase in moments over piers, thought
to be on the order of 10%, was not included because stresses at or
near internal bearings have been shown to be reduced below that
calculated by simple analysis techniques due to an action known as
"fanning". The distribution factors, given in the LRFD Specification,
are also different from those given in NCHRP 12-26, because the
multiple presence factors, given in Lecture 3, are built into the
distribution factors, whereas, the multiple presence factors in earlier
editions of the AASHTO Standard Specifications are built into the
NCHRP 12-26 factors. Additionally, the factors appropriate for the
LRFD Specification are based on a lane of live load, rather than a "line
of wheels". Finally, when the SI version of the LRFD Specification is
used, conversion to that system of units has also been accounted for.
Various limits on span, spacing and other characteristics are
provided in the Specifications for each of the distribution factors.
These parameters identify the range for which the factors were
developed. They were not evaluated for factors beyond the ranges
indicated. Therefore, for structures which do not comply with these
limitations, a rigorous analysis by grid or finite elements should be
used. Furthermore, the distribution factors usually apply for structures
which are:
essentially constant in deck width,
have four or more beams, unless noted,
have beams which are parallel and approximately of the same
stiffness,
have overhangs that do not exceed 0.9 m, unless specifically
noted,
have in-plan curvatures less than those specified above, and
have a cross-section consistent with one of the cross-sections
identified in Table 6.4.2.2.1-1.
Since the distribution factors, developed under NCHRP 12-26,
are new to the Specification, it is appropriate to review the background
Lecture - 6-10
and development of the new distribution factors. The discussion
below was taken from the NCHRP Research Results Digest No. 187,
a summary of the Final Report of Project 12-26 as summarized by
Ian M. Friedland, NCHRP Project Coordinator.
Live load distribution on highway bridges is a key response
quantity in determining member size and, consequently, strength and
serviceability. It is of critical importance both in the design of new
bridges and in the evaluation of the load-carrying capacity of existing
bridges.
Using live load distribution factors, engineers can predict
bridge response by treating the longitudinal and transverse effects of
live loads as uncoupled phenomena. Empirical live load distribution
factors for stringers and longitudinal beams have appeared in the
AASHTO Standard Specifications for Highway Bridges with only minor
changes since 1931. Findings of recent studies suggest a need to
update these specifications in order to provide improved predictions
of live load distribution.
Live load distribution is a function of the magnitude and
location of truck live loads and the response of the bridge to these
loads. The NCHRP 12-26 study focused on the second factor
mentioned above: the response of the bridge to a predefined set of
loads, namely, the HS family of trucks.
In Project 12-26, three levels of analysis were considered for
each bridge type. The most accurate level, Level 3, involves detailed
modeling of the bridge deck. Level 2 includes either graphical
methods, nomographs and influence surfaces, or simplified computer
programs. Level 1 methods provide simple formulas to predict lateral
load distribution, using a wheel load distribution factor applied to a
truck wheel line to obtain the longitudinal response of a single girder.
The major part of the research in Project 12-26 was devoted
to the Level 1 analysis methods because of its ease of application,
established use, and the surprisingly good correlation with the higher
levels of analysis in their application to a majority of bridges. The
formulas presented in the current AASHTO specifications were
evaluated, and alternative formulas were developed that offer
improved accuracy, wider range of applicability, and in some cases,
easier application than the current AASHTO formulas. These
formulas were developed for interior and exterior girder moment and
shear load distribution for single or multiple lane loadings. In addition,
correction factors for continuous superstructures and skewed bridges
were developed.
The formulas presented in previous AASHTO Specifications,
although simpler, do not present the degree of accuracy demanded by
today's Bridge Engineers. In some cases, these formulas can result
in highly unconservative results (more than 40%), in other cases they
may be highly conservative (more than 50%). In general, the formulas
developed in Project 12-26 are within 5% of the results of an accurate
Lecture - 6-11
analysis. Table 6.4.2.2.1-1 shows comparisons with moment
distribution factors obtained from AASHTO, Level 1, Level 2 and Level
3 methods for simple span bridges.
Table 6.4.2.2.1-1 - Comparison of interior girder moment distribution
factors by varying levels of accuracy using the "average bridge" for
each bridge type
Bridge Type AASHTO
NCHRP 12-26
(Level 1)
Grillage
(Level 2)
Finite Element
(Level 3)
Beam-and-slab
a
1.413
(S/1700)
1.458 1.368 1.378
Box girder
a
1.144 1.143 0.970 1.005
Slab
b
1820 1710 1900 1890
Multi-box beam
a
0.646 0.597 0.540 0.552
Spread box beam
a
1.564 1.282 1.248 1.241
a
Number of wheel lines per girder
b
Wheel line distribution width, in mm
In addition, the study resulted in recommendations for use of
computer programs to achieve more accurate results. The
recommendations focus on the use of plane grid analysis, as well as
detailed finite element analysis, where different truck types and their
combinations may be considered.
6.4.2.2.2 Influence of Truck Configuration
The formulas developed in Project 12-26 for the Level 1
analysis were based on the standard AASHTO "HS" trucks. A limited
parametric study conducted as part of the research showed that
variations in the truck axle configuration or truck weight do not
significantly affect the wheel load distribution factors. The group of
axle trains used for this study are shown in Figure 6.4.2.2.2-1. It is
anticipated that smaller gage widths would result in larger distribution
factors, and larger gage widths would result in smaller distribution
factors. Table 6.4.2.2.1-1 gives the variation of wheel load distribution
factors with different axle configurations applied to a number of beam-
and-slab bridges. The differences were below 1% in many cases and,
in all cases, the formulas resulted in good predictions. Therefore, with
some caution, these formulas may be applied to other truck types.
Obviously, Levels 2 and 3 analyses may also be applied for trucks
significantly different from the AASHTO family of trucks.
Lecture - 6-12
Figure 6.4.2.2.2-1 - Axle Configurations for Truck Types Considered
in Study
Table 6.4.2.2.1-1 - Effect of Load Configuration on Distribution Factor
DISTRIBUTION
FACTOR (g)
PERCENT DIFFERENCE
WITH HS-20
HS-20 HTL-57 4A-66 B-141
NCHRP
12-26
HTL-57 4A-66 B-141
NCHR
P 12-
26
Average
*
1.293 1.261 1.285 1.268 1.304 -2.4 -0.6 -1.9 +0.9
Max. S
5000 mm
2.220 2.162 2.205 2.178 2.308 -2.6 -0.7 -1.9 +4.0
Min. S
1000 mm
0.713 0.717 0.713 0.715 0.755 +0.6 0.0 +0.3 +5.9
Max. L
60 000 mm
0.982 0.958 0.983 0.952 1.033 -2.4 +0.1 -3.1 +5.2
Min. L
6000 mm
1.630 1.625 1.624 1.623 1.807 -0.3 -0.3 -0.4 +10.9
*
S = 2200 mm
L = 20 000 mm
t
s
= 185 mm
K
g
= 2.33 x 10
11
mm
4
Lecture - 6-13
6.4.2.2.3 Simplified Methods
6.4.2.2.3a Simplified Formulas for Beam-and-Slab Bridges
This type of bridge has been the subject of many previous
studies, and many simplified methods and formulas were developed
by previous researchers for multi-lane loading moment distribution
factors. The AASHTO formula, the formulas presented by other
researchers, and the formulas developed in the study are discussed
in the following according to their application.
Table 6.4.2.2.3a-1 is taken from the specifications and
summarized criteria for moment in interior beams or elements for
various types of cross-sections. Similar tables exist for moment in
exterior griders, for shear in interior girders and shear in exterior
girders.
Table 6.4.2.2.3a-2 describes how the term L (length) may be
determined for use in the live load distribution factor equations given
in Table 6.4.2.2.3a-1.
In the rare occasion when the continuous span arrangement
is such that an interior span does not have any positive uniform load
moment (i.e., no uniform load points of contraflexure), the region of
negative moment near the interior supports would be increased to the
centerline of the span, and the L used in determining the live load
distribution factors would be the average of the two adjacent spans.
Lecture - 6-14
Table 6.4.2.2.3a-1 - Distribution of Live Loads Per Lane for Moment in Interior Beams
Cross-Section
from Table
4.6.2.2.1-1
Wood Deck on Wood
or Steel Beams
a, l See Table S4.6.2.2.2a-1
Concrete Deck on
Wood Beams
l One Design Lane Loaded: S/3700
Two or More Design Lanes Loaded: S/3000
S 1800
Concrete Deck, Filled
Grid, or Partially Filled
Grid on Steel or
Concrete Beams;
Concrete T-Beams, T-
and Double T-Sections
a, e, k and
also i, j
if sufficiently
connected to
act as a unit
One Design Lane Loaded:
0.06
S
4300
0.4
S
L
0.3
K
g
Lt
3
s
0.1
0.075
S
2900
0.6
S
L
0.2
K
g
Lt
3
s
0.1
1100 S 4900
110 t
s
300
6000 L 73 000
N
b
4
use lesser of the values obtained from the
equation above with N
b
= 3 or the lever rule
N
b
= 3
Multicell Concrete Box
Beam
d One Design Lane Loaded:
1.75
S
1100
300
L
0.35
1
N
c
0.45
13
N
c
0.3
S
430
1
L
0.25
2100 S 4000
18 000 L 73 000
N
c
3
If N
c
> 8 use N
c
= 8
Concrete Deck on
Concrete Spread Box
Beams
b, c One Design Lane Loaded:
S
910
0.35
Sd
L
2
0.25
S
1900
0.6
Sd
L
2
0.125
1800 S 3500
6000 L 43 000
450 d 1700
N
b
3
Use Lever Rule S > 3500
Cross-Section
from Table
4.6.2.2.1-1
Lecture - 6-15
Concrete Beams used
in Multibeam Decks
f One Design Lane Loaded:
k
b
2.8L
0.5
I
J
0.25
where: k 2.5(N
b
)
0.2
1.5
k
b
7600
0.6
b
L
0.2
I
J
0.06
900 b 1500
6000 L 37 000
5 N
b
20
g
if sufficiently
connected to
act as a unit
h Regardless of Number of Loaded Lanes: S/D
where:
C = K (W/L) K
D = 300 [11.5 - N
L
+ 1.4N
L
(1 - 0.2C)
2
] when C
5
D = 300 [11.5 - N
L
] when C > 5
Skew 45
N
L
6
g, i, j
if connected
only enough to
prevent relative
vertical
displacement at
the interface
K =
(1 ) I
J
for preliminary design, the following values of
K may be used:
Beam Type K
Nonvoided rectangular beams 0.7
Rectangular beams with
circular voids: 0.8
Box section beams 1.0
Channel beams 2.2
T-beam 2.0
Double T-beam 2.0
Steel Grids on Steel
Beams
a One Design Lane Loaded:
S/2300 If t
g
< 100 mm
S/3050 If t
g
100 mm
S/2400 If t
g
< 100 mm
S/3050 If t
g
100 mm
S 1800 mm
S 3200 mm
Concrete Deck on
Multiple Steel Box
Girders
b, c Regardless of Number of Loaded Lanes:
0.05 0.85
N
L
N
b
0.425
N
L
0.5
N
L
N
b
1.5
Lecture - 6-16
Table 6.4.2.2.3a-2 - L for Use in Live Load Distribution Factor Equations
FORCE EFFECT L (mm)
Positive Moment The length of the span for which
Negative Moment - End spans of continuous spans,
from end to point of contraflexure under a uniform
load on all spans
Negative Moment - Near interior supports of
continuous spans, from point of contraflexure to point
of contraflexure under a uniform load on all spans
The average length of the two
adjacent spans.
Positive Moment - Interior spans of continuous
spans, from point of contraflexure to point of
contraflexure under a uniform load on all spans
Shear The length of the span for which
shear is being calculated.
Exterior Reaction The length of the exterior span.
Interior Reaction of Continuous Span The average length of the two
adjacent spans.
The AASHTO formula for moment distribution for multi-lane
loading is given as S/1800 for reinforced concrete T-beam bridges
with girder spacing up to 3000 mm, and as S/1700 for steel girder
bridges and prestressed concrete girder bridges with girder spacing
up to 4300 mm, where S is the girder spacing. When the girder
spacing is larger than the specified limit, simple beam distribution is to
be used to calculate the load distribution factors.
wheel load distribution for moment which included multiple lane
reduction factors and is applicable to all beam-and-slab bridges. The
formula is based on girder spacing, span length, slab thickness and
bridge girder stiffness.
A formula which does not consider a reduction for multi-lane
loading was developed at Lehigh University. The Lehigh formula
includes terms for the number of traffic lanes, number of girders, girder
spacing, span length and total curb-to-curb deck width.
Sanders and Elleby (NCHRP Report 83) developed a simple
formula based on orthotropic plate theory for moment distribution on
beam-and-slab bridges. Their formula includes terms for girder
Lecture - 6-17
spacing, number of traffic lanes and a stiffness parameter based on
bridge type, bridge and beam geometry and material properties.
A full-width design approach, known as Henry's Method, is
used by the State of Tennessee. Henry's Method includes factors for
number of girders, total curb-to-curb bridge deck width and a reduction
factor based on number of lanes.
A formula developed as part of NCHRP Project 12-26 includes
the effect of girder spacing, span length, girder inertia and slab thick-
ness. The multiple lane reduction factor is built into the formula. This
formula, applicable to cross-sections with four or more beams, is given
by:
(6.4.2.2.3a-1)
g 0.075
S
2900
0.6
S
L
0.2
K
g
Lt
3
s
0.1
where:
S = girder spacing (1100 mm S 4900 mm)
L = span length (6000 mm L 73 000 mm)
K
g
= n(I+Ae
g
2
) (4 x 10
8
K
g
3 x 10
12
mm
4
)
n = modular ratio of girder material to slab material
I = girder moment of inertia
e
g
= eccentricity of the girder (i.e., distance from centroid of
girder to mid-point of slab)
t
s
= slab thickness (110 mm t
s
300 mm)
This formula is dependent on the inertia of the girder and, thus, a
value for K
g
must be assumed for initial design. For this purpose,
K
g
/Lt
S
3
may be taken as unity.
All of the above formulas were evaluated using direct finite
element analysis with the GENDEK-5 program and a database of 30
bridges; subsequently, they were evaluated using the MSI method and
database of more than 300 bridges. It was found that Equation 6 and
the Illinois formulas are accurate and produce results that are as
accurate as the Level 2 methods.
Moment Distribution to Exterior Girders, Multi-Lane Loading
Previous AASHTO Specifications recommend a simple beam
distribution of wheel loads in the transverse direction for calculating
wheel load distribution factors in edge girders. Any load that falls
Lecture - 6-18
outside the edge girder is assumed to be acting on the edge girder,
and any load that is between the edge girder and the first interior
girder is distributed to these girders by assuming that the slab acts as
a simple beam in that region. Any wheel load that falls inside of the
first interior girder is assumed to have no effect on the edge girder.
the exterior girder based on certain assumptions in the placement of
loads and may not be applicable to all bridges. This formula includes
terms similar to those used in their formula for moment distribution to
interior girders.
A formula, depending on wheel position, alone was developed
as part of this study which results in a correction factor for the edge
girder. The factor must be applied to the distribution factor for the
interior girder to obtain a distribution factor for the edge girder. This
formula is given by:
(6.4.2.2.3a-2) e 0.77
d
e
2800
where:
d
e
cell, in mm
If the edge of the lane is outside of the exterior girder, the
distance is positive; if the edge of the lane is to the interior side of the
girder, the distance is negative.
It was found that the formula developed in Project 12-26
resulted in accurate correction factors and was simpler than the
previous AASHTO procedure.
The literature search performed in this study did not reveal any
simplified formula for single-lane loading of beam-and-slab bridges.
The formula developed as part of the study is as follows:
(6.4.2.2.3a-3)
g 0.06
S
4300
0.4
S
L
0.3
K
g
Lt
3
s
0.1
where the parameters are the same as those given for Equation
6.4.2.2.3a-1.
This formula is applicable to interior girders only.
Lecture - 6-19
Simple beam distribution in the transverse direction should be
used for single-lane loading of edge girders.
One other investigation, applicable to load distribution for both
shear and moment, is required for exterior beams of beam-slab
bridges with diaphragms or cross-frames. This addition was not
developed as part of NCHRP 12-26, but was added by the NCHRP
12-33 Editorial Committee. This distribution is based on treating the
cross-section as a transversely rigid unit which deflects and rotates as
a straight line. The live load is positioned for maximum effect on an
exterior beam (one lane, two lane, three lane, etc., each with its
appropriate multiple presence factor). The total vertical force and
moment about the centroid of the cross-section is applied to the area
of the cross-section, i.e., the number of beams, and the section
modulus, i.e., the sum of the square of the distances of each beam
from the centroid of the beams divided by the distance to the exterior
beam. The specification puts this in equation form as:
(6.4.2.2.3a-4)
R
N
L
N
b
X
ext
N
L
e
v
N
b
x
2
where:
R = reaction on exterior beam in terms of lanes
N
L
= number of loaded lanes under consideration
e
v
= eccentricity of a design truck or a design lane
load from the center of gravity of the pattern of
girders (mm)
x = horizontal distance from the center of gravity of
the pattern of girders to each girder (mm)
X
ext
= horizontal distance from the center of gravity of
the pattern of girders to the exterior girder (mm)
N
b
= number of beams or girders
Shear Distribution
No formula was found from previous research for the
calculation of wheel load distribution factors for shear. Therefore, the
formulas developed as part of the 12-26 study are reported for
different cases as follows.
The formula for multi-lane loading of interior girders is:
Lecture - 6-20
(6.4.2.2.3a-5) g 0.2
S
3600
S
10 700
2
The correction formula for multi-lane loading edge girder shear
is:
(6.4.2.2.3a-6) e 0.6
d
e
3000
The formula for shear distribution factor due to single-lane
loading is:
(6.4.2.2.3a-7)
g 0.36
S
7600
Equation 6.4.2.2.3a-7 is applicable to interior girders only.
Simple beam distribution in the transverse direction should be used for
single-lane loading of edge girders.
Previous AASHTO Specifications did not include approximate
formulas to account for the effect of skewed supports. However, some
researchers have developed correction factors for such effects on
moments in interior girders.
Marx, et al, at the University of Illinois, developed four
correction formulas for skew, one each for skew angles of 0, 30, 45
and 60 degrees. Corrections for other values of skew are obtained by
straight-line interpolation between the two enveloping skew values.
These correction formulas are based on girder spacing, span length,
slab thickness and bridge girder stiffness.
A formula for a correction factor for prestressed concrete
I-girders was developed as part of the research performed at Lehigh
University. This formula is based on the number of traffic lanes,
number of girders, girder spacing, span length and total curb-to-curb
deck width, and includes a variable term for skew angle.
A correction factor for moment in skewed supports was also
developed as part of Project 12-26. This formula is:
(6.4.2.2.3a-8) r 1 c
1
(tan)
1.5
where, for > 30 ,:
Lecture - 6-21
(6.4.2.2.3a-9)
c
1
0.25
K
g
Lt
3
s
0.25
S
L
0.5
If 30 , c
1
is taken as zero. In calculating c
1
, should not be taken
as greater than 60 . The other parameters are as defined previously.
From the literature review, no correction formulas were
obtained for shear effects due to skewed supports. In Project 12-26,
it was found that shear in interior girders need not be corrected for
skew effects; that is, the shear distribution to interior girders is similar
to that of a straight bridge. A correction formula for shear at the
obtuse corner of the exterior girder of two girder systems and all
girders of a multi-girder bridge was developed as part of this study and
is given as:
(6.4.2.2.3a-10) r 1 0.2
Lt
3
s
k
g
0.3
tan
where the parameters are defined in Equation 6.4.2.2.3a-1.
Equation 6.4.2.2.3a-7 is to be applied to the shear distribution
factor in the exterior girder of non-skewed bridges. Therefore, the
product of factors g, e and r must be obtained to find the obtuse corner
shear distribution factor in a beam-and-slab bridge.
The distribution factors calculated for moments are plotted as
a function of girder spacing for Spans 9, 18, 27, 36 and 60 m in Figure
6.4.2.2.3a-1. For comparisons, AASHTO (1989) distribution factors
are also shown. Girder distribution factors, specified by AASHTO
(1989), are conservative for larger girder spacing. For shorter spans
and girder spacings, AASHTO (1989) produces smaller distribution
factors than calculated values.
Lecture - 6-22
Figure 6.4.2.2.3a - Calculated Distribution Factors
6.4.2.2.3b Simplified Formulas for Box Girder Bridges
Research on box girder bridges has been performed by
various researchers in the past. Bridge deck behavior has been well
studied and many recommendations have been made for detailed
analysis of these bridges. However, there is a limited amount of
information on simplified wheel load distribution formulas in the
literature.
In this context, a girder is a notional I-shape consisting of one
web of a multi-cell box and the associated half-flanges on each side
of the web.
Moment Distribution to Interior Girders
Scordelis, at the University of California, Berkeley, presented
a formula for prediction of wheel load distribution for moment distribu-
tion in prestressed and reinforced concrete box girder bridges. The
formula is based on modification of distribution factors obtained for a
rigid cross-section. The formula predicts load distribution factors in
reinforced concrete box girders with high accuracy and for prestressed
concrete box girders with acceptable accuracy.
Sanders and Elleby also presented a simple formula for
moment distribution factors which is similar to their formula for beam-
and-slab bridges.
The following formulas, developed as part of NCHRP 12-26,
may be used to predict the moment load distribution factors in the
interior girders of concrete box girder bridges due to single-lane and
multi-lane loadings. These formulas are applicable to both reinforced
Lecture - 6-23
and prestressed concrete bridges, and the multiple presence factor is
accounted for.
For single-lane loading:
(6.4.2.2.3b-1)
g 1.75
S
1100
300
L
0.35
1
N
c
0.45
For multi-lane loading:
(6.4.2.2.3b-2)
g
13
N
c
0.3
S
430
1
L
0.25
where:
N
c
= number of cells
The factor for load distribution for exterior girders shall be
W
e
/4300 mm, where W
e
is the width of the exterior girder, taken as the
top slab width measured from the mid-point between girders to the
edge of the slab.
Shear Distribution
No formula for shear load distribution was obtained from
previous research for box girder bridges, but the following were
The shear distribution factor for interior girder multi-lane
loading of reinforced and prestressed concrete box girder bridges is:
(6.4.2.2.3b-3) g
S
2200
0.9
d
L
0.1
where:
d = girder depth, in mm
Lecture - 6-24
(6.4.2.2.3b-4) g
S
2900
0.6
d
L
0.1
where the parameters are as defined in Equation 6.4.2.2.3b-3.
A correction formula for shear in the exterior girder for multi-
lane loading is:
(6.4.2.2.3b-5) e 0.64
d
e
3800
where:
d
e
cell, in mm
The following formula was developed for correction of moment
due to skewed supports for values of from 0 to 60 :
(6.4.2.2.3b-6) r 1.05 0.25 (tan) 1.0
If > 60 , use 60 in Equation 6.4.2.2.3b-6.
Another formula was developed in Project 12-26 for correction
of shear at the obtuse corner of an edge girder. It must be applied to
the shear distribution factor for the edge girder of a non-skewed bridge
and must, therefore, be used in conjunction with the edge girder
correction factor of Equation 6.4.2.2.3b-5. This formula, applicable for
values of up to 60 , is:
(6.4.2.2.3b-7)
r 1 c
1
(tan)
where:
c
1
= 0.25 + L/(70d)
d = bridge depth, in mm
Lecture - 6-25
6.4.2.2.3c Simplified Formulas for Slab Bridges
The literature search did not reveal any simplified formulas for
wheel load distribution in slab bridges other than those recommended
by AASHTO. Therefore, the following are formulas that were
Moment Distribution, Multi-Lane Loading
Equation 6.4.2.2.3c-1 was developed to predict wheel load
distribution (distribution design width) for moment in slab bridges due
to multi-lane loading. Multiple presence factors are already accounted
for in the formula:
(6.4.2.2.3c-1)
E 2100 0.12 L
1
W
1
0.5
W
N
L
where:
E = the transverse distance over which a wheel line is
distributed
L
1
= L 18 000 mm
W
1
= W 18 000 mm
W = bridge width, in mm, edge-to-edge
Moment Distribution, Single-Lane Loading
The equation below predicts wheel load distribution for
moment due to single-lane loading.
(6.4.2.2.3c-2) E 250 0.42 (L
1
W
1
)
0.5
where the parameters are as defined in Equation 6.4.2.2.3c-1.
Equation 6.4.2.2.3b-6 may be used to account for the reduction
of moment in skewed bridges.
According to previous AASHTO Specifications, slab bridges
are adequate for shear if they are designed for moment. A quick
check of this assumption was made and it was concluded that it is a
valid assumption. Therefore, no formula or method is presented for
calculation of shear in slab bridges.
Lecture - 6-26
6.4.2.2.3d Simplified Formulas for Multi-Beam Decks which are
Only one formula, other than those presented in the previous
AASHTO Specifications, was obtained for load distribution in multi-
beam decks. This formula, developed by Arya at the University of
Illinois, is applicable to both box and open section multi-beam bridges
and predicts interior beam moment responses due to single-lane and
multi-lane loading. However, a number of simplified formulas
developed in the study are valid only for multi-box beam decks and do
not apply to open sections. Therefore, the response of multi-beam
decks made of open members, such as channels, may or may not be
accurately predicted by the formulas developed in that study.
The formula developed by Arya for interior girder load
distribution in multi-beam decks includes terms for the maximum
number of wheels that can be placed on a transverse section of the
bridge, number of beams, beam width and span length. A variation of
the formula was also proposed for multi-beam decks made of
channels, which includes consideration of the overall depth of the
channel section and its average thickness, defined as its area divided
by its length along the centerline of the thickness.
The following formula was developed in Project 12-26 to
predict load distribution factors for interior beam moment due to multi-
lane loading. The multiple presence reduction factor is already
accounted for in the formula.
(6.4.2.2.3d-1) g k
b
7600
0.6
b
L
0.2
I
J
0.06
where:
k = 2.5(N
b
)
0.2
1.5
b = beam width, in mm
N
b
= number of beams
I = moment of inertia of a beam (mm)
4
J = torsional constant of a beam (mm)
4
This formula is dependant on the inertia and torsional constant of a
beam; an estimated value for these properties must, therefore, be
used in preliminary design. The term I/J may be taken as unity for this
case.
Lecture - 6-27
Arya also presented a load distribution formula for multi-beam
decks designed for one traffic lane. The formulation and parameters
were similar to those presented for multi-lane loading. A variation of
that equation was also presented for calculation of the interior beam
moment distribution factor for a single-lane, channel section, multi-
beam deck. It should be noted that Arya's equations are not
applicable to cases of only one-lane loading with more than one traffic
lane.
A formula for wheel load distribution for moment in the interior
girders due to single-lane loading was also developed in NCHRP 12-
26. This formula is as follows:
(6.4.2.2.3d-2) g k
b
2.8L
0.5
I
J
0.25
All parameters are defined in Equation 6.4.2.2.3d-1. Equation
6.4.2.2.3c-2 is also dependent on inertia and torsional constants, and
a value of 1.0 may be used as an approximation for the term I/J during
preliminary design.
The moment in the edge girder due to multi-lane loading in
multi-beam decks comprised of box units is obtained by using a
correction factor applied to the interior girder distribution factors for
multi-lane loading. This correction factor may be found from the
following formula:
(6.4.2.2.3d-3) g 1.04
d
e
7600
where:
d
e
= distance from edge of the lane to the center of the exterior web
For exterior beams of sufficiently interconnected multi-beam
bridge decks comprised of T-shaped units subjected to multi-lane
loading, Equation 6.4.2.2.3b-2 applies.
For single-lane loading and for multi-beam decks comprised of
either box units or units other than box units, the lever rule is used.
Lecture - 6-28
Shear Distribution
Distribution factors for shear in interior girders of multi-beam
decks in "Bridge Decks Comprised of Box Units" due to multi-lane
loading may be calculated from the following formula:
(6.4.2.2.3d-4) g
b
4000
0.4
b
L
0.1
I
J
0.05
where the parameters are as defined in Equation 6.4.2.2.3d-1.
Distribution factors for shear in the interior girders of multi-
beam decks in "Bridge Decks Comprised of Box Units" due to single-
lane loading are obtained from the following formula:
(6.4.2.2.3d-5) g 0.70
b
L
0.15
I
J
0.05
where the parameters are again as defined in Equation 6.4.2.2.3d-1.
Note that Equations 6.4.2.2.3d-4 and 6.4.2.2.3d-5 are
dependent on inertia and torsional constants, and a value of 1.0 may
be used as an approximation for the term I/J during preliminary
design.
The shear in the edge girder of multi-beam deck in "Bridge
Decks Comprised of Box Units" due to multi-lane loading can be found
using a correction factor applied to interior girder distribution factors.
This correction factor is obtained from the formula:
(6.4.2.2.3d-6) e 1.02
d
e
15 000
where:
d
e
For shear in exterior beams of sufficiently interconnected multi-
beam bridge decks comprised of T-shaped units, Equations
6.4.2.2.3b-4 through 6.4.2.2.3b-5 and the lever rule should be used,
where appropriate.
The moment in any beam in a skewed bridge may be obtained
by using a skew reduction factor given by Equation 6.4.2.2.3b-6.
Lecture - 6-29
The shear in the interior beams of a skewed multi-beam bridge
comprised of box beams is usually of the same order as that of the
shear in the obtuse corner and must be obtained by applying a
correction factor to the response of the edge girder in a straight bridge.
This correction factor may be calculated from the formula:
(6.4.2.2.3d-7) r 1 c
1
(tan)
0.5
where:
(6.4.2.2.3d-8)
c
1
L
90d
6.4.2.2.3e Simplified Formulas for Multi-Beam Decks which are not
The LRFD Specification contains the same provisions for load
distribution in this type of bridge superstructure as appeared in recent
editions of the Standard Specification, and as repeated below for
completeness.
The key difference between bridges treated herein, as
compared to Section 6.4.2.2.3d, is the degree of transverse
interconnection of units. If box, T, channel or other precast units are
interconnected through a structural slab, or sufficiently transversely
post-tensioned to produce a similar level of continuity, then the
discussion of Section 6.4.2.2.3d applies. If the interconnection
between the units is expected to transmit shear, but relatively little
moment over the bridge service life, then the provisions herein apply.
The Specification provides for the computation of a bending
moment distribution factor, regardless of the number of lanes, given
by:
g
S
300D
for which:
D = 300 [11.5 - N
L
+ 1.4 N
L
(1 - 0.2C)
2
] when C 5
D = 300 [11.5 - N
L
] when C > 5
C = K (W/L)
K =
(1 )I
J
Lecture - 6-30
where:
= Poisson ratio
I = moment of inertia (mm)
4
J = St. Venant's constant (mm)
4
N
L
= number of lanes
S = spacing of units (mm)
W = edge-to-edge width of bridge (mm)
6.4.2.2.3f Simplified Formulas for Spread Box Beam Bridges
Only one formula, other than those recommended by
AASHTO, was obtained from previous research for determining load
distribution factors in spread box beam bridges. This formula was
developed at Lehigh University for predicting the response of interior
beams due to multi-lane loading and was later adopted by AASHTO.
A correction factor for skewed bridges was also presented. In
addition, a number of simple formulas were developed as part of
NCHRP Project 12-26.
Moment Distribution to Interior Beams, Multi-Lane Loading
A formula developed in Project 12-26 for moment in interior
spread box beams due to multi-lane loading is as follows:
(6.4.2.2.3f-1)
g
S
1900
0.6
Sd
L
2
0.125
where
S = girder spacing (mm)
d = beam depth (mm)
Moment Distribution to Interior Beams, Single-Lane Loading
A similar formula was developed for distribution to interior
beams due to single-lane loading:
Lecture - 6-31
(6.4.2.2.3f-2)
g
S
910
0.35
Sd
L
2
0.25
where the parameters are as defined in Equation 6.4.2.2.3f-1.
The moment in edge girders due to multi-lane loading may be
calculated by applying a correction factor to the interior girder
distribution factor:
(6.4.2.2.3f-3) e 0.97
d
e
8700
where:
d
e
The distribution factor for moment in the edge girder due to
single-lane loading may be obtained by simple-beam distribution, i.e.,
the lever rule, in the same manner as was described for beam-and-
slab bridges.
Shear Distribution
multi-lane loading may be calculated from the following:
(6.4.2.2.3f-4) g
S
2250
0.8
d
L
0.1
where the parameters are as defined previously.
(6.4.2.2.3f-5) g
S
3050
0.6
d
L
0.1
where the parameters are again as defined previously.
The shear in the edge girder due to multi-lane loading can be
found by applying a correction factor to the interior girder equation.
This correction factor is:
Lecture - 6-32
(6.4.2.2.3f-6) e 0.8
d
e
3050
where:
d
e
The wheel load distribution factor for shear in the edge girder
due to single-lane loading may be obtained by simple-beam
distribution in the same manner as was described for beam-and-slab
bridges, i.e., the lever rule.
Research at Lehigh University also resulted in a formula for
correction of wheel load distribution for moment in interior girders due
to multi-lane loading in skewed bridges. NCHRP 12-26 concludes that
Equation 6.4.2.2.3b-6 was also applicable to this case.
The shear in the interior beams of a skewed bridge is the same
as that of a straight bridge. However, the shear in the obtuse corner
must be obtained by applying a correction factor to the distribution
factor for the edge girder in a straight bridge, given by Equation
6.4.2.2.3b-7, which C
1
is taken as:
(6.4.2.2.3f-7)
c
1
(Ld )
0.5
6S
6.4.2.2.3g Response of Continuous Bridges
The response of continuous bridges was studied by modeling
a number of two-span continuous bridges where each span is similar
to the average bridge. The wheel load distribution factor for each case
was compared to that of a simple bridge and correction factors for
continuity were obtained. In the case of beam-and-slab bridges, a
complete parameter study was performed, and it was found that the
correction factor is generally independent of bridge geometry. These
factors are given in the table below.
When the NCHRP 12-26 factors were incorporated into the
LRFD Specification, it was decided that 5% corrections were
unwarranted given that the distribution factors are an approximation
of actual behavior and are, therefore, subject to some variability. The
continuity correction for negative moment, a 10% increase, was also
neglected on the basis that experimentally observed "fanning" of the
reaction tends to reduce the negative moment as compared to a
typical beam calculation.
Lecture - 6-33
Table 6.4.2.2.3f-1 - Continuity Correction
Factors
Beam-and-Slab Bridges
Positive moment
Negative moment
c = 1.05
c = 1.10
c = 1.00
c = 1.05
Box Girder Bridges
Positive moment
Negative moment
c = 1.00
c = 1.10
c = 1.00
c = 1.00
Slab Bridges
Positive moment
Negative moment
c = 1.00
c = 1.10
Multi-Beam Bridges
Positive moment
Negative moment
c = 1.00
c = 1.10
c = 1.00
c = 1.05
Spread box beam bridges
Positive moment
Negative moment
c = 1.00
c = 1.10
c = 1.00
c = 1.05
6.4.2.3 TRUSS AND ARCH BRIDGES
6.4.2.3.1 General
The approximate method for load distribution to lines of trusses
and arches is a so-called "lever rule", which is simply a matter of
summing moments about one line of trusses or arches to find the
reaction on the other line. This approach is illustrated by calculations
in Lecture 7.
Lecture - 6-34
6.5 REFINED METHODS
6.5.1 Deck Slabs
Where refined analysis of deck slabs is desirable, finite
element analysis is recommended. Elements should be chosen to
simulate both bending and in-plane or membrane effects. If the
analysis utilizes only plate/membrane or shell elements and has only
one or two elements through the thickness of the deck, then the
refined analysis will report an essentially bending-type response in the
deck. There has been much experimental and analytic work that
suggests that bending is not the primary source of strength in bridge
decks, but that the development of membrane action, analogous to a
shallow arch or dome load path within the deck is the primary source
of strength. This type of action will only be determined through a very
rigorous modeling of the deck.
6.5.2 Beam Slab Bridges
Relatively rigorous models of beam slab bridges can be
developed using general purpose commercial finite element programs,
finite strip programs or special purpose greater finite element-based
computer programs which have been specifically developed to simplify
the analysis of bridge-type structures. These more custom-oriented
programs often contain mesh generating capabilities, automatic load
placement capabilities and code checking.
Detailed bridge deck analysis using a finite element computer
program may be used to produce accurate results. However, extreme
care must be taken in preparation of the model, or inaccurate results
will be obtained. Important points to consider are selection of a
program capable of accurately modeling responses being investi-
gated, calculation of element properties, mesh density and support
conditions. Every model should be thoroughly checked to ensure that
nodes and elements are generated correctly.
Another important point is the loading. Truck loads should be
placed at positions that produce the maximum response in the
components being investigated. In many cases, the truck location is
not known before preliminary analysis is performed and, therefore,
many loadings should be investigated. This problem is more
pronounced in skewed bridges.
Many computer programs have algorithms that allow loads to
be placed at any point on the elements. If this feature is not present,
equivalent nodal loads must be calculated. Distribution of wheel loads
to various nodes must also be performed with care, and the mesh
should be fine enough to minimize errors that can arise because of
load approximations.
Many computer programs, especially the general purpose finite
element analysis programs, report stresses and strains, not shear and
moment values. Calculation of shear and moment values from the
Lecture - 6-35
stresses must be carefully performed, usually requiring an integration
over the beam cross-section. Some programs report stresses at node
points rather than Gaussian integration points. Integration of stresses
at node points is normally less accurate and may lead to inaccurate
results.
Many graphical and computer-based methods are available for
calculating wheel load distribution. One popular method consists of
design charts based on the orthotropic plate analogy, similar to that
presented in the Ontario Highway Bridge Design Code. As computers
become readily available to designers, simple computer-based
methods, such as SALOD, become more attractive than nomographs
and design charts. Also, grillage analysis presents a good alternative
to other simplified bridge deck analysis methods and will generally
produce more accurate results.
The grillage analogy may be used to model any one of the five
bridge types studied in this research. Each bridge type requires
special modeling techniques. A major advantage of plane grid
analysis is that shear and moment values for girders are directly
obtained and integration of stresses is not needed. Loads normally
need to be applied at nodal points, and it is recommended that simple-
beam distribution be used to distribute wheel loads to individual
nodes. If the loads are placed in their correct locations, the results will
be close to those of detailed finite element analysis.
As indicated previously, the designer has to be responsible for
constructing a suitable model and determining that the results are
accurate. It is possible to make seemingly small errors in computer
models which can have dramatic effects on the results which are
obtained.
6.5.3 Example of Modeling Errors
The modeling of diaphragms and boundary conditions at
supports and bearings is vital to obtaining the proper results when
using these sophisticated programs. The burden of correctly handling
these factors rests with the designer. Consider the following example
which shows how a very small modeling error produced very
erroneous results.
The framing plan shown on Figure 6.5.3-1 represents an actual
bridge that was designed using a grid-type approach. The designer
had a good model for this structure, except that the rotational degree
of freedom corresponding to the global "x" axis was fixed at all of the
bearings. This did not allow the diaphragms at the piers and
abutments to respond correctly to the imposed loadings and
deformations, and also had the effect of producing artificially stiff ends
on the girders by virtue of vector resolution between global and local
systems. The effect of this condition on the reactions obtained at the
abutments and piers was dramatic. Modest uplift was reported at the
acute corner along the near abutment shown in Figure 6.5.3-1, and a
very substantial uplift was reported at the acute angle at the far
Lecture - 6-36
abutment. This is shown in the table in Figure 6.5.3-1, as is a moment
diagram for non-composite dead loads which reflects the incorrect
reactions. Also shown in the table of reactions on Figure 6.5.3-1 are
the correct reactions determined when the structure was modeled
using the generic STRESS Computer Program with proper boundary
conditions at the supports. In this case, a positive reaction is found at
all bearings, and a significantly different moment diagram for non-
composite dead load also resulted. The correct reactions and moment
diagram are also shown on Figure 6.5.3-1.
The modeling of the degrees of freedom at the lines of support
on this structure was also investigated utilizing a relatively complete
three-dimensional finite element analysis and the SAPIV Computer
Program. The model used is illustrated in Figure 6.5.3-2, which shows
how the deck slab, girders and cross-frames were modeled in their
proper relative positions in the cross-section which extended along the
bridge from end-to-end. Also shown on this figure is a comparison of
the reactions obtained from STRESS and from SAPIV by applying all
of the non-composite loads in a single loading. The comparison
between these reactions is excellent.
In order to verify that the order of pouring the deck slab units
would not contribute to an uplift situation, the pouring sequence was
replicated in a three-dimensional SAPIV analysis. The results of the
analysis of the three stages of the pouring sequence are also shown
in Figure 6.5.3-2, as well as the total accumulated load at the end of
the pour. Comparison of the sequential loading with the application of
a single loading of non-composite dead load also showed relatively
good agreement in this case.
The important point demonstrated in the example of Figures
6.5.3-1 and 6.5.3-2 is that seemingly small errors in modeling of the
structure can result in very substantial changes in the reactions,
shears and moments. The designer must be aware of this potential
when utilizing the more refined analysis techniques.
Incidentally, there are cases in which an uplift reaction due to
skew and/or curvature is possible. The simple span bridge shown in
Figure 6.5.3-3 and reported on in the November 1, 1984, issue of
Engineering News-Record, was analyzed at the request of the owner.
In this case, the uplift reactions computed by the designer were
verified.
Sometimes modeling problems occur because User's Manuals
are not clear, or a "bug" exists, of which the program's author/vendor
is not aware. Such a case is illustrated for the simply-supported,
partially-curved and skewed bridge in Figure 6.5.3-4. Initially, this
bridge was modeled with extra joints at locations other than
diaphragms in an effort to improve live load determination. As a
result, the number of points along each girder were not equal, but
there was no indication in the program's descriptive literature that this
type of arrangement of program input could potentially cause a
problem. The resulting moment envelopes for the middle and two
Lecture - 6-37
exterior girders are shown in Figure 6.5.3-4, and are obviously
unusual in shape and also in their order of maximum moment, i.e., #4,
#5 and #3.
It was found that the live load processor was not responding
properly to the unequal number of nodes per girder, that nodes should
be essentially "radial", and that it was not certain that nodes to which
diaphragms were not connected were legitimate. The revised model,
shown in Figure 6.5.3-4, produced clearly better results, as shown in
the indicated moment envelopes.
Lecture - 6-38
Figure 6.5.3-1 - Framing plan, comparative live load reactions and
moment envelopes showing effect of proper and improper rotational
boundary condition, as reflected in grid analysis.
Lecture - 6-39
Figure 6.5.3-2 - Finite element idealization and reactions obtained for
structure shown in Figure 6.
Lecture - 6-40
Figure 6.5.3-3 - Framing Plan of Curved Span with Skewed Piers
Figure 6.5.3-4 - Comparative moment envelopes for the middle and
two outside girders of curved skewed system showing the results of
apparent "bug" in algorithm for applying live load.
6.5.4 Other Types of Bridges
The Specification contains additional requirements for the
rigorous analysis of cellular-type structures, truss bridges, arch
bridges, cable-stayed bridges and suspension bridges. Generally
speaking, refined analysis will involve a computer model which
accurately affects the geometry, relative component stiffnesses,
Lecture - 6-41
boundary conditions and load supply to the structure. Suspension
bridges will almost always be analyzed using a large deflection theory.
The deflection theory may also be applied to arches and cable-stayed
bridges. In the case of the cable-stayed bridge of moderate span, it
may be sufficiently accurate to evaluate the second order effects on
the deck system of the tower by supplementary calculations in
providing a correction factor, developed on a bridge-specific basis.
The change in stiffness of the cables caused by change in sag as the
cable load changes can be accounted for using the so-called "Ernst"
equations, given in the Specification, for modified modulus of
elasticity.
Lecture - 6-42
REFERENCES
Jones, 1976, "A Simple Algorithm for Computing Load Distribution in
Multi-Beam Bridge Decks", Proceedings, 8th ARRB Conference, 1976
APPENDIX A
The Load Distribution Problem and its Solution in NCHRP 12-26
Lecture - 6-A1
Older editions of the AASHTO Specifications allow for simplified analysis of bridge
superstructures using the concept of a load distribution factor for bending moment in interior girders
of most types of bridges, i.e., beam-and-slab, box girder, slab, multi-box beam and spread-box
beam. This distribution factor is given by:
(A-1)
g
S
300D
where:
g = a factor used to multiply the total longitudinal response of the bridge due to a single
longitudinal line of wheel loads in order to determine the maximum response of a
single girder
S = the center-to-center girder spacing (mm)
D = a constant that varies with bridge type and geometry
A major shortcoming of the previous specifications is that the piecemeal changes that have
taken place over the last 55 years have led to inconsistencies in the load distribution criteria
including: inconsistent consideration of a reduction in load intensity for multiple lane loading;
inconsistent changes in distribution factors to reflect the changes in design lane width; and,
inconsistent approaches for verification of live load distribution factors for various bridge types.
The past AASHTO simplified procedures were developed for non-skewed, simply-supported
bridges. Although it was stated that these procedures apply to the design of normal (i.e., supports
oriented perpendicular to the longitudinal girders) highway bridges, there are no other guidelines
for determining when the procedures are applicable. Because modern highway and bridge design
practice requires a large number of bridges to be constructed with skewed supports, on curved
alignments, or continuous over interior supports, it is increasingly important that the limitations of
load distribution criteria be fully understood by designers.
Advanced computer technology has become available in recent years which allows detailed
finite element analysis of bridge decks. However, many computer programs exist which employ
different formulations and techniques. It is important that the computer methodology and
formulation that produces the most accurate results be used to predict the behavior of bridge decks.
In order to identify the most accurate computer programs, data from full-scale and prototype bridge
load tests were compiled. The bridge tests were then modeled by different computer programs and
the experimental and computer results were compared. The programs that produced the most
accurate results were then considered as the basis for evaluation of the other method levels, i.e.,
Levels 2 and 1 methods.
An important part of the development or evaluation of simplified methods is range of
applicability. In order to ensure that common values of various bridge parameters were considered,
a database of actual bridges was compiled. Bridges from various states were randomly selected
in order to achieve national representation. This resulted in a database of 365 beam-and-slab
bridges, 112 prestressed concrete and 121 reinforced concrete box girder bridges, 130 slab
bridges, 67 multi-box beam bridges and 55 spread-box beam bridges. This bridge database was
studied to identify the common values of various parameters, such as beam spacing, span length,
slab thickness, and so on. The range of variation of each parameter was also identified. A
hypothetical bridge that has all the average properties obtained from the database, referred to as
Lecture - 6-A2
the "average bridge" was created for each of the beam-and-slab, box girder, slab, multi-box beam
and spread-box beam bridge types. For the study of moment responses in box girder bridges,
separate reinforced concrete and prestressed concrete box girder average bridges were also
prepared.
In evaluating simplified formulas, it is important to understand the effect of various bridge
parameters on load distribution. Bridge parameters were varied one at a time in the average bridge
for the bridge type under consideration. Load distribution factors for both shear and moment were
obtained for all such bridges. Variation of load distribution factors with each parameter shows the
importance of each parameter. Simplified formulas can then be developed to capture the variation
of load distribution factors with each of the important parameters. A brief description of the method
used to develop such formulas is as follows.
In order to derive a formula in a systematic manner, certain assumptions must be made.
First, it is assumed that the effect of each parameter can be modeled by an exponential function
of the form ax
b
, where x is the value of the given parameter, and a and b are coefficients to be
determined based on the variation of x. Second, it is assumed that the effects of different
parameters are independent of each other, which allows each parameter to be considered sep-
arately. The final distribution factor will be modeled by an exponential formula of the form: g =
(a)(S
b1
)(L
b2
)(t
b3
)(...) where g is the wheel load distribution factor; S, L, and t are parameters included
in the formula; a is the scale factor; and b1, b2, and b3 are determined from the variation of S, L,
and t, respectively. Assuming that for two cases, all bridge parameters are the same, except for
S, then:
g
1
= (a)(S
1
b1
)(L
b2
)(t
b3
)(...) (A-2)
g
2
= (a)(S
2
b1
)(L
b2
)(t
b3
)(...) (A-3)
therefore:
(A-4)
g
1
g
2
S
1
S
2
b1
or:
(A-5)
b1
ln
g
1
g
2
ln
S
1
S
2
If n different values of S are examined and successive pairs are used to determine the value
of b1, n 1 different values for b1 can be obtained. If these b1 values are close to each other, an
exponential curve may be used to accurately model the variation of the distribution factor with S.
In that case, the average of n 1 values of b1 is used to achieve the best match. Once all the power
factors, i.e., b1, b2, and so on, are determined, the value of "a" can be obtained from the average
bridge, i.e.,
Lecture - 6-A3
(A-6)
a
g
o
S
b1
o
L
b2
o
t
b3
o
(...)
This procedure was followed during the entire course of the NCHRP 12-26 study to develop
new formulas as needed. In certain cases where an exponential function was not suitable to model
the effect of a parameter, slight variation from this procedure was used to achieve the required
accuracy. However, this procedure worked quite well in most cases and the developed formulas
demonstrate high accuracy.
Because certain assumptions were made in the derivation of simplified formulas and some
bridge parameters were ignored altogether, it is important to verify the accuracy of these formulas
when applied to real bridges. The database of actual bridges was used for this purpose. Bridges
to which the formula can be applied were identified and analyzed by an accurate method. The
distribution factors obtained from the accurate analysis were compared to the results of the
simplified methods. The ratio of the approximate to accurate distribution factors was calculated and
examined to assess the accuracy of the approximate method. Average, standard deviation, and
minimum and maximum ratio values were obtained for each formula or simplified method. The
method or formula that has the smallest standard deviation is considered to be the most accurate.
However, it is important that the average be slightly greater than unity to assure slightly
conservative results. The minimum and maximum values show the extreme predictions that each
method or formula produced when a specific database was used. Although these values may
change slightly if a different set of bridges is used for evaluation, the minimum and maximum values
allow identification of where shortcomings in the formula may exist that are not readily identified by
the average or standard deviation values.
It was previously mentioned that different subsets of the database of bridges were used to
evaluate different formulas. When a subset included a large number of bridges (100 or more), a
Level 2 method was used as the basis of comparison. When it included a smaller number of
bridges (less than 100), a Level 3 method was used. As a result, LANELL (an influence surface
method) was used for verification of formulas for moment distribution in box girder bridges, and a
Multi-dimensional Space Interpolation (MSI) method was used for verification of formulas for
straight beam-and-slab and slab bridges.
Findings
Level 3 Methods: Detailed Bridge Deck Analysis
Recent advances in computer technology and numerical analysis have led to the
development of a number of computer programs for structural analysis. Programs that can be
applicable to bridge deck analysis can be divided into two categories. One includes general
purpose structural analysis programs such as SAP, STRUDL and FINITE. The other category is
specialized programs for analysis of specific bridge types, such as GENDEK, CURVBRG and
MUPDI.
In the search for the best available computer program for analysis of each bridge type, all
suitable computer programs (general and specific) that were available at the time of the 12-26
research were evaluated. In order to achieve meaningful comparisons and assess the level of
accuracy of the programs, a number of field and laboratory tests were modeled by each program.
The results were then compared in three ways:
by visual comparison of the results plotted on the same figure,
Lecture - 6-A4
by comparison of the averages and standard deviations of the ratios of analytical to
experimental results, and
by comparison of statistical differences of analytical and experimental results. Five bridge
types were considered: beam-and-slab, box girder, slab, multi-box beam, and spread-box
beam.
For analysis of beam-and-slab bridges, the following computer programs and models were
evaluated: GENDEK-PLATE, GENDEK-3, GENDEK-5, CURVBRG, SAP and MUPDI. It was found
that, in general, GENDEK-5 analysis using plate elements for the deck slab and eccentric beam
elements for the girders is very accurate. This program is also general enough to cover all typical
cases, i.e., straight, skew, moment and shear. However, for analysis of curved open girder steel
bridges, CURVBRG was the most accurate program. MUPDI was also found to be a very accurate
and fast program; however, skewed bridges cannot be analyzed with this program and shear values
near the point of application of load, or near supports, lack accuracy. GENDEK-5 was, therefore,
selected to evaluate Level 2 and Level 1 methods.
For analysis of box girder bridges, computer programs MUPDI, CELL-4 and FINITE were
evaluated. MUPDI was the fastest and most practical program for analysis of straight bridges for
moment, but FINITE was found to be the most practical program for skewed bridges and for
obtaining accurate shear results. Therefore, MUPDI was selected for the evaluation of LANELL (a
Level 2 method for moment in straight bridges which was, in turn, used for evaluation of Level 1
methods) and FINITE was selected for other cases.
For the analysis of slab bridges, computer programs MUPDI, FINITE, SAP and GENDEK
were evaluated. Shear results cannot be obtained accurately in slab bridges and, therefore, were
not considered. The GENDEK-5 program, without beam elements, proved to be very accurate.
However, MUPDI was found to be the most accurate and practical method for non-skewed
prismatic bridges and was selected to evaluate Level 2 and Level 1 methods.
For the analysis of multi-beam bridges, the following computer programs were evaluated:
SAP, FINITE and a specialized program developed by Professor Powell at the University of
California, Berkeley, for analysis of multi-beam bridges (referred to as the POWELL program
herein). Various modeling techniques were studied using different grillage models and different
plate elements. The program that is capable of producing the most accurate results in all cases,
i.e., straight and skewed for shear and moment, was the FINITE program. This program was later
used in evaluation of more simplified methods. POWELL is also very accurate in reporting
moments in straight bridges, but it uses a finite strip formulation, similar to MUPDI, and, therefore,
is incapable of modeling skewed supports, and shear results near supports and load locations can-
not be accurately obtained. This program was used to evaluate simplified methods for straight
bridges.
For analysis of spread-box beam bridges, computer programs SAP, MUPDI, FINITE and
NIKE-3D were evaluated. FINITE produced the most accurate results, especially when shear was
considered. MUPDI was selected to evaluate simplified methods for calculation of moments in
straight bridges, and FINITE was selected for all other cases.
Level 2 Methods: Graphical and Simple Computer-Based Analysis
Nomographs and influence surface methods have traditionally been used when computer
methods have been unavailable. The Ontario Highway Bridge Design Code uses one such method
based on orthotropic plate theory. Other graphical methods have also been developed and
reported. A good example of the influence surface method is the computer program SALOD
Lecture - 6-A5
developed by the University of Florida for the Florida Department of Transportation. This program
uses influence surfaces, obtained by detailed finite element analysis, which are stored in a
database accessed by SALOD. One advantage of influence surface methods is that the response
of the bridge deck to different truck types can be readily computed.
A grillage analysis using plane grid models can also be used with minimal computer
resources to calculate the response of bridge decks in most bridge types. However, the properties
for grid members must be calculated with care to assure accuracy. Level 2 methods used to
analyze the five bridge types (beam-and-slab, box girder, slab, multi-beam and spread-box beam
bridges) are discussed below.
The following methods were evaluated for analysis of beam-and-slab bridges: plane grid
analysis, the nomograph-based method included in the Ontario Highway Bridge Design Code
(OHBDC), SALOD and Multi-dimensional Space Interpolation (MSI). All of these methods are
applicable for single- and multi-lane loading for moment. The OHBDC curves were developed for
a truck other than HS-20, and using the HS-20 truck in the evaluation process may have introduced
some discrepancies. The method presented in OHBDC was also found to be time consuming, and
inaccurate interpolation between curves was probably a common source of error. SALOD can be
used with any truck and, therefore, the "HS" truck was used in its evaluation. The MSI method was
developed based on HS-20 truck loading for single- and multiple-lane loading. MSI was found to
be the fastest and most accurate method and was, therefore, selected for the evaluation of Level
1 methods. This method produces results that are generally within 5% of the finite element
(GENDEK) results.
In the analysis of box girder bridges, OHBDC curves and the LANELL program were
evaluated. The comments made about OHBDC for beam-and-slab bridges are valid for box girder
bridges as well. As LANELL produced results that were very close to those produced by MUPDI,
it was selected for evaluation of Level 1 methods for moment.
OHBDC, SALOD and MSI were evaluated for the analysis of slab bridges. MSI was found
to be the most accurate method and, thus, was used in the evaluation of Level 1 methods. SALOD
also produced results that were in very good agreement with the finite element (MUPDI) analysis.
Results of OHBDC were based on a different truck and, therefore, do not present an accurate
evaluation.
In the analysis of multi-beam bridges, a method presented in Jones, 1976, was evaluated.
The method is capable of calculating distribution factors due to a single concentrated load and was
modified for this study to allow wheel line loadings. The results were found to be in very good
agreement with POWELL. However, because this method was only applicable for moment
distribution in straight single-span bridges, it was not used for verification of Level 1 methods.
In the analysis of spread-box beam bridges, only plane grid analysis was considered as a
Level 2 method.
In general, Level 2 graphical and influence surface methods generated accurate and
dependable results. While these methods are sometimes difficult to apply, a major advantage of
some of them is that different trucks, lane widths, and multiple presence live load reduction factors
may be considered. Therefore, if a Level 2 procedure does not provide needed flexibility, its use
is not warranted because the accuracy of it is on the same order as a simplified formula. MSI is an
example of such a method for calculation of load distribution factors in beam-and-slab bridges.
A plane grid analysis would require computer resources similar to those needed for some
of the methods mentioned above. In addition, a general purpose plane grid analysis program is
Lecture - 6-A6
available to most bridge designers. Therefore, this method of analysis is considered a Level 2
method. However, the user has the burden of producing a grid model that will produce sufficiently
accurate results. As part of NCHRP Project 12-26, various modeling techniques were evaluated,
and it was found that a proper plane grid model may be used to accurately produce load distribution
factors for each of the bridge types studied.
Level 1 Methods: Simplified Formulas
The current AASHTO Specifications recommend use of simplified formulas for determining
load distribution factors. Many of these formulas have not been updated in years and do not
provide optimum accuracy. A number of other formulas have been developed by researchers in
recent years. Most of these formulas are for moment distribution for beam-and-slab bridges
subjected to multi-lane truck loading. While some have considered correction factors for edge
girders and skewed supports, very little has been reported on shear distribution factors or
distribution factors for bridges other than beam-and-slab.
The sensitivity of load distribution factors to various bridge parameters was also determined
as part of the study. In general, beam spacing is the most significant parameter. However, span
length, longitudinal stiffness and transverse stiffness also affect the load distribution factors.
Figures 6.4.2.2.3-2 through 6.4.2.2.3-6 show the variation of load distribution factors with various
bridge parameters for each bridge type. Ignoring the effect of bridge parameters, other than beam
spacing, can result in highly inaccurate (either conservative or unconservative) solutions.
A major objective of the research in Project 12-26 was to evaluate older AASHTO
Specifications and other researchers' published work to assess their accuracy and develop
alternate formulas whenever a more accurate method could be obtained. The formulas that were
evaluated and developed are briefly described below, according to bridge type; i.e., beam-and-slab,
box girder, slab, multi-beam and spread-box beam.
Figure A-1 - Effect of Parameter Variation on Beam-and-Slab Bridges
Lecture - 6-A7
Figure A-2 - Effect of Parameter Variation on Box Girder Bridges
Figure A-3 - Effect of Parameter Variation on Slab Bridges
Figure A-4 - Effect of Parameter Variation on Multi-Box Beam Bridge
Lecture - 6-A8
Figure A-5 - Effect of Parameter Variation on Spread-Box Beam Bridges

Nhi13061 4-6 PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Nhi13061 4-6 PDF

Uploaded by

Copyright:

Available Formats

printed on June 24, 2003

Therefore, the moment is the function of the beam rigidity and

You might also like