Sunteți pe pagina 1din 23

What You Should Know About

Heat Straightening Repair of Damaged Steel


INTRODUCTION straightening. Rather, the thermal expansion/contraction is

an unsymmetrical process in which each cycle leads to a
T he repair of damaged steel through the application of
heat has long been considered an art rather than a sci-
ence. As such both the methodology and results have var-
gradual straightening trend. The process is characterized by
the following conditions, which must be maintained:
1. The maximum heating temperature of the steel does
ied considerably from one practitioner to another. Because
not exceed either (a) the lower critical temperature
steel tends to be a forgiving material, successful repairs out-
(the lowest temperature at which permanent molecu-
number failures. Yet material properties may be compro-
lar changes may occur), or (b) the temper limit for
mised resulting in a weakened in-place structure. In
quenched and tempered steels.
addition, it is not unusual for a fracture to occur during the
2. The stresses produced by applied external forces do
repair process. As a result, the use of heat straightening (or
not exceed the yield stress of the steel in its heated
flame straightening) has had limited application. Many
engineers are reluctant to allow heat straightening because
3. Only the regions in the vicinity of the plastically
of the lack of an established method of application as well
deformed zones are heated.
as the lack of quantification of steel properties after the
When these conditions are met, the material properties
repair is complete. The purpose of this paper is to provide
undergo relatively small changes and the performance of
some engineering guidelines for use of the heat-straighten-
the steel remains essentially unchanged after heat straight-
ing process.
ening. Properly conducted, heat straightening is a safe and
The ideas presented in this paper are referenced to sup-
economical procedure for repairing damaged steel.
porting research and provide the reader with a place to
begin when contemplating heat straightening repair. The
paper is intended to be a state-of-the-practice report, which
should be used to complement sound engineering judgment, The basic concept of heat straightening is relatively simple
not to replace it. In addition, many of the principles dis- and relies on two distinct properties of steel:
cussed here also apply to heat curving to produce camber or • If steel is stretched or compressed past yield, it does
sweep. not assume its original shape when released. Rather,
it remains partially elongated or shortened, depending
WHAT IS HEAT SRAIGHTENING? on the direction of the originally applied force.
• If steel is heated to relatively modest temperatures
Heat-straightening is a repair procedure in which a limited
(370-700°C or 700-1300°F), its yield value becomes
amount of heat is applied in specific patterns to the plasti-
significantly lower while at the elevated temperature.
cally deformed regions of damaged steel in repetitive heat-
To illustrate how steel can be permanently deformed
ing and cooling cycles to produce a gradual straightening of
using these two properties, consider the short steel bar in
the material. The process relies on internal and external
Figure 1a. First, the bar is placed in a fixture, much
restraints that produce thickening (or upsetting) during the
stronger than the bar itself, and clamped snug-tight (Figure
heating phase and in-plane contraction during the cooling
1b). Then the bar is heated in the shaded portion. As the
phase. Heat straightening is distinguished from other meth-
bar is heated it tries to expand. However, the fixture pre-
ods in that force is not used as the primary instrument of
vents expansion in the longitudinal direction. Thus, the fix-
ture exerts restraining forces on the bar as shown in Figure
1c. Since the bar is prevented from longitudinal expansion,
R. Richard Avent is Robert and Adele Anding distinguished
professor, department of civil and environmental engineer-
it is forced to expand a greater amount laterally and trans-
ing, Louisiana State University, Baton Rouge, LA. versely through its thickness than in an identical unre-
strained bar. Consequently, a bulge will occur in the heated
David J. Mukai is assistant professor, department of civil and
zone. Because the bulge has been heated, its yield value has
environmental engineering, Louisiana State University,
BatonRouge, LA.
been lowered, resulting in plastic deformations, which do


not occur in the unheated portions. When the heating straightening in that applied external force is used to
source is removed, the material will cool and contract three- straighten the damage. These applied forces produce
dimensionally. The clamp cannot prevent the bar from con- stresses well above yield, resulting in large movements dur-
tracting longitudinally. As cooling progresses the bar ing a single heat cycle. Often the member is completely
shortens and the bulge shrinks. However, a portion of the straightened by the continued application of a large force
bulge is permanent even after the bar has completely cooled during a single cycle. The results of this type of straighten-
and the bar is shorter than its original length, Figure 2.1d. ing are unpredictable and little research has been conducted
In essence a permanent redistribution of material has on this procedure. Specific concerns about hot mechanical
occurred in the heated zone leaving the bar slightly shorter straightening include:
with a small bulge. This permanent bulge, or thickening, in 1. Fracture may occur during straightening
the heated zone is called “upsetting”. The redistribution of 2. Material properties may be adversely affected
material is referred to as “plastic deformation” or “plastic 3. Buckles, wrinkles or crimps may result
flow”. The clamping force is often referred to as a restrain- The Engineer should recognize that hot mechanical
ing force. Through cycles of clamping, heating, and cool- straightening is an unproven method, which may lead to
ing, the bar could be shortened to practically any length damaged or degraded steel. As such, its use should be con-
desired. sidered only in special cases in which other methods are not
This simple example illustrates the fundamental princi- viable.
ples of heat straightening. However, most damage in steel Hot working is distinguished from heat straightening in
members is much more complex than stretching or shorten- that both large external forces and high heat are used. This
ing of a bar. Consequently, different damage conditions method is similar to hot mechanical straightening in that
require their own unique heating and restraining patterns. external forces are used. In addition, the steel is heated well
above the lower critical temperature and often glows cherry
WHY PLACE LIMITS ON HEATING red indicating a temperature above the upper critical tem-
TEMPERATURE AND JACKING STRESSES? perature. The results of this process are highly unpre-
dictable and may result in:
A clear distinction should be made between heat straighten-
1. Fracture during straightening
ing and two other methods often confused with heat
2. Severe changes in molecular structure which may not
straightening: hot mechanical straightening and hot work-
be reversible
ing. Hot mechanical straightening differs from heat
3. Severe changes in mechanical properties including a
high degree of brittleness
4. Buckles, wrinkles, crimps, and other distortions
Hot working should not be used to repair damaged steel.
Some practitioners will tend to over-jack and over-heat
yet claim to be heat straightening. The reader is cautioned
to be aware of these distinctions when specifying heat
straightening as opposed to either hot mechanical straight-
ening or hot working.


One of the most important and yet difficult-to-control
parameters of heat straightening is the temperature of the
heated metal. Factors affecting the temperature include size
and type of the torch orifice, intensity of the flame, speed of
torch movement, and thickness and configuration of the
member. Assuming that adequate control of the applied
temperature is maintained, the question arises as to what
temperature produces the best results in heat straightening
without altering the material properties.
The large majority of steels used for construction in the
United States are either carbon or low alloy steel. At ambi-
ent temperature, these steels have two major constituents:
Fig. 1. Conceptual example of shortening a steel bar.


ferrite grains and pearlite grains. Pearlite is itself composed (0.8 percent carbon). The volume fraction of cementite
of alternating laths of ferrite iron and cementite (Fe3C). An increases with increasing carbon content of steel. Low car-
iron-carbon equilibrium diagram is shown in Figure 2 illus- bon steels tend to be softer and more ductile because these
trating the relationship between the components of steel and are characteristics of ferrite. Cementite is hard and brittle
temperature. Solid steel is comprised of grains. The grains thus high carbon steels are harder and less ductile.
themselves are composed of iron atoms arranged on a crys- Temperatures greater than 727°C (1340°F) begin to pro-
tal lattice. The carbon (and for that matter any other alloyed duce a phase change in steel. This temperature is often
or residual atoms) do not molecularly bond with the iron, called the lower critical (or lower phase transition) temper-
but rather; nest within the spaces between the iron atoms ature. The ferritic and pearlite crystal structures begin to
(interstitial solid solution), or displaces an iron atom from a assume a face centered cubic form (austenite). Thus, as fer-
lattice node and occupies the position itself (substitutional rite and pearlite are heated to the austenitic range and their
solid solution). Ferrite is essentially pure iron. However, crystal lattices change to face-centered cubic, all contained
by examining the iron-carbon Phase diagram (Figure 2), it carbon goes into solution. When steel cools below the
can be seen that ferrite has a carbon solubility capacity up lower critical temperature, it attempts to return to its former
to .008 percent. Cementite is an orthorhombic crystal struc- ferritic or pearlitic structure. Since this change requires a
ture comprised of iron and carbon. Chemically, cementite specified time frame, rapid cooling may not permit the
contains 6.67 percent carbon dissolved in the crystal matrix complete molecular change to occur. Under these circum-
(but is not molecularly bonded to the iron). The ratio of fer- stances, a hard, strong and brittle phase called martensite
rite to cementite in pearlite formed under slow cooled and occurs. The steel in this form may have reduced ductility
under equilibrium conditions from a eutectoid steel is and be more sensitive to brittle fracture under repeated
approximately 6:1. However, under other conditions (slow loads.
cool as not to form martensite or bainite) and compositions, The upper critical (or upper phase transition) temperature
pearlite has been observed to contain ferrite in a 3:1 to 2:1 is the level at which the molecular change in structure is
ratio with cementite. A low carbon steel has less than 0.8 complete. At this temperature (around 815 to 925°C or
percent carbon, too little carbon to develop a 100 percent 1500 to 1700°F for most steels, depending on carbon con-
pearlite structure, resulting in pearlite plus ferrite. This fer- tent) the steel assumes the form of a uniform solid solution
rite can exist as both a grain and as a constituent of pearlite. called austenite. It is at temperatures between the lower and
Carbon plus iron alloys with less than 0.8 percent carbon upper critical that a wide range of mill hot rolling and work-
are hypoeutectic steels, commonly known as ferritic steels. ing can occur. As long as the temperature is lowered slowly
These steels have a larger volume fraction of ferrite grains, in a controlled manner from these levels, the steel assumes
body-centered crystal structure (BCC), than pearlite. High its original molecular configuration and properties. This
carbon steels (carbon content between 0.8 and 2.0 percent) temperature control is more difficult to maintain at a fabri-
are hypereutectic steels, commonly known as pearlite cation shop or in the field when conducting heat straighten-
steels. These steels are predominately pearlite (orthorhom- ing repairs. Consequently, if the temperature during heat
bic cementite and BCC ferrite). The volume fraction of straightening is not kept below the lower critical tempera-
cementite and ferrite are equal at the eutectic composition ture, undesirable properties may be produced during cool-
It has been shown by Avent, Fadous, and Boudreaux
(1991), Avent and Mukai (1998), and Roeder (1986), that
the amount of movement during heat straightening of dam-
aged flexural members is directly proportional to heating
temperature up to at least 870°C (1800°F). Permanent
movement can occur with heating temperatures as low as
370°C (700°F). To avoid the possibility of detrimental
changes in material properties, it is recommended that the
steel temperature be kept somewhat below the lower phase
transition temperature of approximately 700°C (1300°F).
The maximum temperature recommended by most
researchers such as Holt (1971), Avent (1995), Roeder
(1986) and Shannafelt and Horn (1984), is 650°C (1200°F)
for all but quenched and tempered high-strength steels. The
limiting temperature of 650°C (1200°F) allows for about
Fig. 2. Iron-carbon equilibrium diagram. 55°C (100°F) of temperature variation, which was found to


be a common range among experienced practitioners. The this assumption has been published. The senior writer has
American Welding Society Bridge Welding Code (1996) investigated this phenomenon (Avent and Fadous, 1989).
specifies maximum heating temperatures of 590°C Successively higher jacking forces were used in heat
(1100°F) for quenched and tempered steels and 650°C straightening minor axis damage in a W24×76 beam. This
(1200°F) for all others. For A514 and A709 (grades 100 beam suddenly fractured during the eighth heating cycle
and 100W), a minimum tempering temperature of 620°C with the highest jacking force applied. Based on the move-
(1150°F) is required. Thus, the 590°C (1100°F) limit pro- ment per heat as well as jacking force measurements, the
vides a 30°C (50°F) safety factor. However, for A709 grade jacking stresses exceeded the reduced yield (but not
70W the specified minimum tempering temperature is reduced tensile strength) in the heated zones. It was con-
590°C (1100°F). A maximum heating temperature of cluded that over-jacking, which produced stresses that were
565°C (1050°F) is recommended for this grade to provide a additive to residual stresses, produced this failure. Typi-
30°C (50°F) safety factor and to avoid property changes. cally, heat straightening practitioners do not use calibrated
To control the temperature, the speed of the torch move- jacks but rather control the force by “feel”. The number of
ment and the size of the orifice must be adjusted for differ- variables makes this approach likely to produce over-stress.
ent thicknesses of material. However, as long as the For example, consider a composite bridge girder. The fol-
temperature is rapidly achieved at the appropriate level, the lowing variables are associated with inducing stress by
contraction effect will be similar. Various methods can be jacking the bottom flange: diaphragm spacing, number of
used to monitor temperature during heating. Principal diaphragms, beam length, section modulus of beam, depth-
among these includes: visual observation of color of the to-web thickness ratio, and section modulus of lower
steel; use of special temperature sensing crayons or pyrom- flange. For a given stress level, the interaction of so many
eters; and infrared electronic temperature sensing devices. variables means that the jacking force could vary over a
wide range. By using the “feel” method, it is erroneously
HOW SHOULD THE STEEL BE COOLED assumed that the jacking force varies little from one struc-
AFTER HEATING? ture to the next. Thus, it would not be unusual to over-stress
such a beam with the likely result of a fracture during heat
To obtain the maximum effect of heat straightening, the
steel should be allowed to cool to approximately ambient
It is therefore recommended that stresses in the heated
temperature. If not, the desired temperature gradient and
zone due to jacking not exceed 50 percent of the nominal
distribution during the succeeding heating cycle will be
yield stress at ambient temperature. Since the yield stress
changed to produce a less effective pattern. It is recom-
at 1200°F is reduced by approximately 50 percent, this cri-
mended that the steel cool to at least 90°C (200°F) prior to
terion prevents the heated steel stress from becoming sig-
applying the next cycle.
nificantly larger than yield. In addition, because the entire
The method of cooling most often used is natural cooling
cross-section does not reach the maximum heating temper-
without artificial aids. However, this approach slows the
ature simultaneously, the 50 percent value includes an addi-
repair and increases costs. Theoretically, any artificial cool-
tional margin of safety. Thus, it is also recommended that
ing procedure could be used as long as the maximum tem-
jacks always be calibrated and the structure be analyzed to
perature in the steel does not exceed the lower phase
prevent over-stress.
transition temperature or the lowest tempering temperature
since changes in crystal structure would not occur. From a
practical viewpoint, it is difficult to ensure that heating tem-
perature can be perfectly controlled. Therefore, it is rec-
ommended that the steel be allowed to cool naturally to The primary equipment utilized for heat straightening is a
315°C (600°F) prior to using compressed air or water to heating torch. The heat source is typically an oxygen-fuel
accelerate cooling. mixture. Typical fuels include acetylene, propane, and nat-
ural gas. The appropriate fuel is mixed with oxygen under
WHAT LIMITS SHOULD BE PLACED ON THE pressure at the nozzle to produce a proper heating flame.
MAGNITUDE OF JACKING FORCES? Either a single or a multiple orifice tip may be used. The
size and type is dictated by the fuel selected and thickness
On several occasions, the writers have observed sudden
of material to be heated. A No. 8 single orifice tip is gener-
fractures during heat straightening. These types of fractures
ally satisfactory for thicknesses up to 20 to 25 mm (3/4 or 1
are not uncommon. The usual explanation is that small
in.) with acetylene. For thinner material a smaller tip is rec-
microcracks must have occurred as a result of initial dam-
ommended. If heavy sections are being heated, a single ori-
age. Thus, during heat straightening, one of these cracks
fice tip may not be adequate. For such cases, either heat
propagates. However, no experimental evidence validating


from both sides or use a rosebud or multiple orifice tip. The resists the normal thermal expansion of the steel in the lon-
size may vary depending on the material thickness. The gitudinal direction. As a result, the heated material will
determining factor is the ability to raise the through-thick- tend to expand, or upset, to a greater extent through the
ness steel temperature to the specified level. Note that thickness of the plate than longitudinally, resulting in plas-
whether a single or multiple orifice is used, the torch should tic flow. At the completion of the heat, the entire heated
be a heating torch and not a cutting torch. area is at a high and relatively uniform temperature. At this
point the plate has moved downward (Figure 3b) due to lon-
WHAT ARE THE BASIC TYPES OF HEATS? gitudinal expansion of material below the neutral axis, pro-
ducing positive bending. As the steel cools, the material
There are several basic types of heats that are frequently
contracts longitudinally to a greater degree than the expan-
used as illustrated by Avent et al. (1991), Avent and Mukai
sion during heating. Thus, a net contraction occurs.
(1998), Ciesicki and Bulter (1968), and Holt (1955, 1965,
Because the net upsetting is proportional to the width across
1971, 1977). For some applications only one type is
the vee, the amount of upsetting increases from top to bot-
required. However, in many cases a combination of these
tom of the vee. This variation produces a closure of the vee.
basic types is necessary. The three basic types most often
Bending is produced in an initially straight member, or
used in heat straightening are described as follows.
straightening occurs (if the plate is bent in the opposite
direction to that of the straightening movement, Figure 3c).
Vee Heat
For many applications, it is most efficient to utilize a vee
The vee heat is used to straighten strong axis bends in steel that extends over the full depth of the plate element, but
plate elements. As seen in Figure 3, a typical vee heat starts partial depth vees may be applicable in certain situations.
at the apex of the vee-shaped area using an oxy-fuel torch. When using partial depth vees, the open end should extend
When the desired temperature is reached (usually around to the edge of the element. The vee depth is varied by plac-
650°C or 1200°F for mild carbon steel), the torch is ing the apex at a partial depth location. For example, in
advanced progressively in a serpentine motion toward the deep plate elements the depth of the vee may result in too
base of the vee. This motion is efficient for progressively much cooling near the apex prior to completing the heat at
heating the vee from top to bottom. The plate will initially the open end. A 3/4 depth vee may be applicable in reducing
move upward (Figure 3a) as a result of longitudinal expan- this problem. Other examples of partial depth vees will be
sion of material above the neutral axis producing negative shown in later sections.
bending. The cool material adjacent to the heated area
Line Heats
Line heats are employed to repair a bend in a plate about its
weak axis. Such bends, severe enough to produce yielding
of the material, often result in yield lines. A line heat con-
sists of a single pass of the torch. The restraint in this case
is often provided by an external force although movement
will occur without external constraints. This behavior is
illustrated in Figure 4. A line heat is applied to the under-
side of a plate element subjected to bending moments pro-
duced by external forces (Figure 4a). As the torch is
applied and moved across the plate, the temperature distri-
bution decreases through the thickness (Figure 4b). The
cool material ahead of the torch constrains thermal expan-
sion, even if bending moment constraints are not present.
Because of the thermal gradient, more upsetting occurs on
the torch (or hotter) side of the plate. During cooling this
side consequently contracts more, creating a concave bend
on the torch side of the plate similar to that shown in Figure
4d. Thus, to straighten a plate bent about its weak axis, the
heat should be applied to the convex side of the damaged
plate. The movement can be magnified by the use of
applied forces, which produce bending moments about the
yield line (Figure 4c). Referring to a section through the
Fig. 3. Stages of movement during vee heat.


plate transverse to the line heat (Figure 4c), the restraining fashion across a strip for a desired length, Figure 5a. This
moments tend to prevent transverse expansion below the pattern sequentially brings the entire strip to the desired
plate centerline. In a manner similar to the vee heat mech- temperature. The orientation can be an important consider-
anism, the material thus tends to expand through the thick- ation. The strip heat may be initiated at the midpoint and
ness, or “upset”. Upon cooling, the restraining moments moved toward both edges simultaneously using two
tend to magnify transverse contraction (Figure 4d). The torches. A second alternative with similar effect is shown
speed of the travel of the torch is critical as it determines the in Figure 5b using a single torch and starting from one side.
temperature attained. With proper restraints and a uniform Depending on the structural configuration, the strip may
speed of the torch, a rotation will occur about the heated also be started at a free edge as shown in Figure 5c. How-
line. ever, without restraints, this orientation may produce some
weak axis bending. By alternating the initiation point to
Strip Heats opposite edges in successive heating cycles, the weak axis
bending can be minimized.
Strip heats, also called rectangular heats, are used to com-
plement a vee heat. Strip heats are similar to vee heats and
are accomplished in a like manner. Beginning at the initia-
tion point, the torch is moved back and forth in a serpentine
A convenient way to classify damage is to define four fun-
damental damage patterns (Avent, 1995). For each damage
pattern, there is a sequence of specific heating types that
will produce straightening. In most cases, actual damage to
steel structures is a combination of two or more of these
fundamental damage patterns. However, more complex
damage can be repaired by sequentially using the heating
patterns for these four fundamental damage patterns. For
purposes of defining heating patterns, it is convenient to
refer to the elements of a cross section as either primary or
stiffening elements. The primary elements are plate ele-
ments damaged by bending about their major axes. The
stiffening elements are those bent about their minor axes.
Typically, vee heats are applied to primary elements while
strip, line or no heat at all may be applied to stiffening ele-
ments. The combination of these heats, applied either
simultaneously or consecutively, represent a single cycle of
the heating pattern. Because the net change in curvature
after one heating cycle is small, a number of cycles are
required to straighten a damaged member. For each cycle

Fig. 4. Schematic of line heat mechanism. Fig. 5. Schematic of strip heat on the top flange of a wide flange beam.


the heating pattern should be shifted to a different location at a time) as shown in Figure 7. After heating these primary
within the yield zone region, although the same location can elements, a strip heat is applied to the web. The only excep-
be returned to after every third cycle. More heats should be tion is that no strip heat is applied to stiffening elements
placed in the region of largest curvature and fewer near located adjacent to the apex of a vee heated element since
extremities to reflect the difference in damage curvature. this element offers little restraint to the closing of the vee
The fundamental damage categories and their heating pat- during cooling. Note that the width of the strip heat is equal
terns are described in the following sections. to the width of the vee heat at the point of intersection. For
all cases the pattern is repeated by shifting over the vicinity
Category S of the yield zone until the member is straight.
This type refers to bending about the “strong” or major axis.
Category T
The heating pattern consists of a vee and strip heat as shown
in Figure 6. This pattern is superimposed on a typical yield This type refers to damage as a result of torsion or twisting
zone for a wide flange beam bent about the strong axis. The about the longitudinal axis of a member. For rolled or built-
strip heat and open end of the vee are placed on the flange up shapes, the flange elements tend to exhibit flexural plas-
damaged in tension. The vee heat is first applied to the tic deformation in opposite directions. The web is often
web. Upon completion, a strip heat is applied to the flange stressed at levels below yield. If one flange is constrained
at the open end of the vee. The width of the strip heat (such as the case of a composite beam), then only the
always equals the vee width at the point of intersection. unconstrained flange element is subjected to plastic defor-
This procedure allows the vee to close during cooling with- mation and yielding may also occur in the web. The heat-
out restraint from the stiffening element. No heat is applied ing pattern for this damage case is shown in Figure 8. The
to the flange at the apex of the vee. This vee/strip combi- vees on the top and bottom flange are reversed to reflect the
nation is repeated by shifting over the vicinity of the yield different directions of curvature of the opposite flanges.
zone until the member is straight. The pattern is shown for The vee heats are applied first and then the strip heat is
a wide flange and channel in Figure 6. applied. Note that for the channel, the strip heat need only
be applied to half depth. This half depth strip allows the
Category W lower flange vee to close with minimal restraint from the
web. If one flange is a composite connection, then that
This category refers to damage as a result of bending about
flange is not heated.
the “weak” or minor axis. For rolled or built-up shapes the
web is usually at, or near, the neutral axis. Consequently, it
may not deform into the inelastic range. The heating pat-
tern for this case is similar to the previous case but note the
primary and stiffening elements are reversed. The vee heat
is first applied to both flanges (either simultaneously or one

Fig. 6. Heating patterns for wide flanges and channels bent about their Fig. 7. Heating patterns for wide flanges and channels bent about their
major axes (Category S). minor axes (Category W).


Category L A common sequence is to heat the vee first, followed
immediately by the strip. An alternative used by some prac-
This category includes damage that is localized in nature.
titioners is to heat the web vee first and allow it to cool for
Local flange or web buckles, web crippling and small bends
a few minutes before heating the flange strip. It is felt by
or crimps in plate elements of a cross section typify this
some that this approach will reduce the residual stresses sig-
behavior. A local buckle or bulge reflects an elongation of
nificantly. The available research data and different
material. Restoration requires the bulging area to be short-
sequences used in practice indicate that more than one
ened. A series of vee or line heats can be used for this pur-
sequence can be successful for many cases. Since different
pose as shown in Figure 9. These vees are heated
sequences are successfully used in practice, it appears that
sequentially across the buckle or around the bulge. For web
heat straightening is not overly sensitive to the sequence
bulges either lines or vees may be used. If vees are used,
used. At this time there is not adequate documentation to
they are spaced so that the open end of the vees touch.
support one sequence over another for a particular heating
There is a tendency for practitioners to over-heat web
pattern. The experience of the practitioner is the most reli-
bulges. For most cases, too much heat is counter-produc-
able guide to proper sequencing. The sequencing patterns
tive. The preferred pattern is the line heats in the
shown here are based on those often successfully used in
spoke/wagon wheel pattern. For the flange buckle pattern
(Figure 9b), either lines or a combination of lines and vees
may be used. For most cases, the line pattern with few or
no vees tends to be most effective. Since the flange dam-
age tends to be unsymmetrical, more heating cycles are
required on the side with the most damage. When possible,
the heating should be done on the convex side.
Since typical damage is often a combination of these fun-
IS THE SEQUENCING OF HEATING IMPORTANT? damental damage categories, a combination of heating pat-
terns is often required. The key is to select the combination
When a combination of vee, strip or line heats is used, the
of patterns to fit the damage. When in doubt, a good policy
order of heating is referred to as the sequence. The
is to address one of the fundamental damage patterns at a
sequencing of heats may be important in some straightening
time. For example, remove the Category W damage prior
operations. However, little research has been conducted to
verify its effects. Some practitioners feel that proper
sequencing will accelerate the straightening and help keep
residual stresses to a minimum. Consider the case of an I-
beam with Category S damage requiring a vee heat in the
web and a strip heat in the flange as shown in Figure 6.

Fig. 8. Wide flanges and channels with twisting damage (Category T). Fig. 9. Typical heating patterns for local damage.


to addressing the Category L damage. It should be noted A final example is the case of multiple plastic hinges
that with proper combinations, several types of damage formed about the weak axis such as might occur for a beam
could be removed expeditiously. For example, suppose that continuous over interior supports. The heating pattern is
a wide flange section is impacted such that the bending shown in Figure 12. Note the reversed direction of the vees
occurs about an axis at an arbitrary angle to the principal to reflect the multiple curvature damage. The restraining
axes, i.e., bending occurs about both the strong and weak moments must also reflect the reverse curvature nature of
axis. The heating pattern, Figure 10, requires a vee heat on the damage as shown in the figure.
the web to restore the strong axis damage and vee heats on
the flanges to restore the weak axis damage. The heats WHEN EVALUATING DAMAGE, HOW SHOULD IT
should be executed sequentially as numbered in Figure 10. BE QUANTIFIED?
Note that no strip heat is required on the web since a vee is
It is important to quantify the level of damage prior to
used there. Restraining forces should be used to produce
deciding if heat straightening is the best repair approach.
bending moments about both the strong and weak axis as
Of particular interest is the degree to which the material has
indicated in Figure 10, tending to straighten the damage.
been strained past yield. A useful measure is the strain
Once the damage is corrected about one of the principal
ratio, µ, which is the ratio of the maximum actual strain to
axes, the heating pattern should revert to one of the funda-
the strain at initial yield. The maximum actual strain, εmax,
mental patterns until straightening is complete about the
can be expressed in terms of radius of curvature, R, as
other principal axis.
As a second example, consider a wide flange beam with 1
ε max = ymax (1)
weak axis bending damage combined with a local bulge in R
one flange. The heating pattern is shown in Figure 11. Vee
where ymax is the distance from the extreme fiber to the neu-
heats are used on the top and bottom flanges along with a
tral axis. The yield strain, εy, is
web strip heat similar to the standard weak axis pattern.
However, partial depth vees are used on the flange with the Fy
εy = (2)
bulge along with a series of line heats along bulge yield E
lines. Since a yield line is likely to occur at the lower web
where Fy is the yield stress and E is the modulus of elastic-
fillet, a line heat is also needed on the web. Restraining
ity. Combining these equations, the strain ratio is
forces are used to create bending moments about the weak
axis as shown in Figure 11. In addition, a jacking force εymax
µ= (3)
should be applied on the local bulge as shown on the cross RFy
section in Figure 11. The sequence of heats is also indi-
cated in the figure.

Fig. 10. Heating pattern and sequence for bending combination about Fig. 11. Heating pattern and sequence for combination weak axis bend-
both the strong and weak axis. ing and local flange bulge.


If the material properties are known, the strain ratio can (Brockenbrough and Johnston, 1968). However, the appli-
be computed once the radius of curvature has been deter- cation of heat tends to restore the original material charac-
mined. One convenient approach is to determine the radius teristics. Shanafelt and Horn (1984) recommended that the
of curvature by measuring the degree of damage. The maximum allowable strain be limited to 15 times the yield
degree of damage, ϕd is defined as the change in slope of an strain and/or 5 percent nominal strain for repair of tension
element over the yield zone as shown in Figure 13. ϕd can members. The limit approximately defines the delineation
be computed as: between the plastic region and the strain-hardening region.
However, these recommendations were not backed by spe-
y2 − y1 y − y4
ϕ d = tan −1 ( ) + tan −1 ( 3 ) (4) cific research data. No limits were suggested for compres-
L1 L2 sion members. A 5 percent nominal strain corresponds to a
where ϕd is the degree of damage or angle of permanent µ = 15 for Fy = 100 ksi. The writers have conducted a num-
deformation at the plastic hinge and yi is a measured offset ber of tests on plates and beams in which the maximum
as shown in Figure 13b. If C is defined as the chord length strain ratio, µ, was 100. Each of these members was suc-
between tangents at the edges of the yield zone, then cessfully heat straightened. Stress-strain curves taken from
tensile coupons after heat straightening indicated that there
1 2 ϕ
= ( )sin d (5) was no significant degradation of the material. These tests
R C 2 indicate that the limiting strain for heat straightening repair
should be equal to or greater than 100 times the yield stain.
A more accurate approach is to take offsets within the Based on anecdotal evidence even larger strains can proba-
damaged zone where curvature is the sharpest. From Fig- bly be repaired. However, research has not been conducted
ure 13a the radius of curvature, R, can be directly approxi- past 100 times the yield strain.
mated as
1 yr −1 − 2 yr + yr +1 WHAT IS THE IMPACT OF
Using either approach the strain ratio can be calculated
Most testing for the basic mechanical properties of heat-
from Equation 3 once R has been determined.
straightened plates and beams has been conducted on
undamaged material after 3 or 4 vee heats (Welding Engi-
neer, 1959; Nicholls and Weerth, 1972; Pattee, Evans, and
Monroe, 1969, 1970; Roeder, 1986; Rothman, 1973; and
Surprisingly little information is available on the effect of Rothman and Monroe, 1973). Researchers concluded from
damage level (strain history). It is known that cold bending these tests that: (1) little change occurred in modulus of
into the yield range reduces the ductility of steel in general

Fig. 12. Heating pattern for reverse curvature bending. Fig. 13. Offset measurements to calculate degree of damage and radius
of curvature.


elasticity, (2) slight increases were found in yield and ulti- DOES MEMBER SHORTENING OCCUR DURING
mate tensile stress, and (3) 10 to 25 percent reduction in HEAT STRAIGHTENING?
ductility was observed. Of more significance are the prop-
The subject of member shortening due to heat straightening
erties of damaged plates (or rolled shapes) after experienc-
has been mentioned in the literature but little research has
ing the large number of heats required to fully straighten the
been conducted. One researcher stated that using smaller
member. Avent, Robinson, Madan, and Shenoy (1993a),
vee depth ratios should result in less member shortening,
Avent and Mukai (1998) conducted tests on seven plates
given any particular damage situation, (Moberg, 1979).
and four beams which were damaged to strain ratios rang-
However, it could be argued that less shortening would
ing from 30 to 100 and then heat straightened. In general,
occur when using full-depth vee heats, since the top fibers
it was found that the yield stress in the heated regions
have been heated and are subjected to a tensile stress. In
increased approximately 10 percent over that of the
fact, the amount of shortening in a member can be quite sig-
unheated member. The largest increases were found at the
nificant, regardless of the vee depth used. If a plate is dam-
apex of the vee heats where yield increased by 15 to 20 per-
aged about its strong axis with a midpoint loading, the top
cent. The maximum tensile stress values showed a smaller
edge of the plate experiences compressive yielding (short-
increase of around 4 to 6 percent. This test data indicates
ening) and the bottom edge of the plate experiences tensile
that heat straightening has a small but significant heat-treat-
yielding (stretching). As the plate is subjected to the heat
ing effect on steel.
straightening process, the top edge experiences some
“restretching” in the longitudinal direction. However, these
positive strains are small in comparison to the simultaneous
shortening of the bottom edge of the plate. Tests indicate
that shortening at around 0.10 inches occurs during the
Tests on undamaged plate specimens have suggested that straightening of W6×9’s with a degree of damage, ϕd = 6°.
heat straightening had little effect on modulus of elasticity. Tests on plates have indicated that up to one-half inch of
However, the writers’ tests, Avent et al. (1993a), on actual shortening can occur for ϕd = 25°. Since member shorten-
damaged and repaired plates and beams indicated reduc- ing may be significant it should be anticipated when con-
tions on the order of 8 to 23 percent. The most significant ducting repairs.
changes occurred in ductility, which was typically reduced
by one-third. While this loss of ductility is significant, all HOW DOES HEAT STRAIGHTENING AFFECT
members tested still met ASTM specifications. FATIGUE LIFE?
Only one series of fatigue tests on flame straightened mem-
bers was found in the literature (AREA, 1946). In this case
three eye bars of A7 steel were heat shortened and then
Various types of notch toughness and hardness tests have fatigue cycled. When compared to similar specimens,
been conducted on heat-straightened steel by Rothman which had not been heated, the fatigue strength at both
(1973), Rothman and Monroe (1973), Pattee et al. (1970) 500,000 and 1,000,000 cycles was similar. Although data is
and summarized by Avent (1989). In general, little signifi- sparse, there is no indication that carbon steels will have a
cant change was found using Charpy V-notch or Rockwell shortened fatigue life after heat straightening.
Hardness tests. Only the quenched and tempered steels
showed evidence of small reductions in toughness. WHAT TYPE OF RESIDUAL STRESES OCCUR
Although residual stresses are often mentioned in literature
on heat straightening, there has been little documented
Since hardness tests may be inconclusive, there are cur- research in this area. Past research was conducted in the
rently no nondestructive testing procedures to determine the context of heat curving (not heat straightening), and thus is
condition of the steel after heat straightening. Visual somewhat limited in its applicability to heat straightening.
inspection during and after repair remains the most viable Some of the most notable research was conducted at the
option. It is particularly important to monitor the heating University of Washington (Roeder, 1986), where a finite
temperature, as over-heating is the most likely cause of element model was developed to predict the local behavior
damage to the steel. Other visual signs of possible prob- of a plate element subjected to a vee heat. Residual stresses
lems include pitting and blackened spots on the surface. were estimated using the model and experimental strains
Properly heated steel will have a silvery color after cooling. were also measured. An example of Roeder’s results are


shown in Figure 14 where tension stresses are positive in center of the heated zone was similar to Figure 15. The
this and other figures of residual stress distribution. large number of repetitive heats tended to slightly reduce
Experimental research was conducted by Brockenbrough the residual stresses in comparison to those of the undam-
(1970b) to back up earlier theoretical residual stress studies, aged plates having only four heats. Overall, maximum
(Brockenbrough, 1970a) on heat-curved plate girders sub- compression residual stresses were approximately 45 per-
jected to line heats. These stresses, determined by the “sec- cent of yield at the flange tips for either case and maximum
tioning method,” were reasonably consistent with the tensile stresses varied from 20 to 30 percent of yield at the
theoretical values. Similar theoretical methods were used center of the plate.
on vee-heated plate elements by Nicholls and Weerth The maximum residual stresses tend to be higher in
(1972) and on wide flange beams by Horton (1973). How- rolled shapes. Typical residual stress distributions are
ever, the results were not supported by any experimental shown in Figures 16 through 19 for typical rolled shapes
data. Avent et al. (1993a) and Avent and Mukai (1998) con- with a negative sign indicating compression. For angles,
ducted a study on residual stresses that included both plates high compression residual stresses (approaching yield in
and rolled shapes; variations in vee angle, vee depth, and some cases) were found at the toes and heel as shown in
level of restraining forces; and degree of initial damage. Figure 16. Similarly high values were found at the heels
Residual stress patterns were determined by using the sec- and toes of channels (Figure 17). Note that strips were not
tioning method (Avent and Wells, 1982). The change in measured at the web-flange junctures resulting in disconti-
strain before and after heating was measured on small strips nuities in the graphs. For the Category W heating pattern,
cut from the heated zone. The strains were converted to wide flange beams tended to have tensile stresses across
stresses using an E = 29,000 ksi. most of both flanges and compression stresses in the web,
Shown in Figure 15 are residual stress patterns for ini- as shown in Figure 18. When the beam was re-damaged
tially undamaged plates, which were vee heated four times. and heat straightened multiple times, the residual stresses
A four-inch (100mm)-gauge length centered in the heated were mitigated. For Category S damaged wide flange
zone was used. For the eight plates tested, relatively little beams, the pattern reversed with compression stresses in
difference in residual stresses were found for: vee angles both flanges and tension in the web, as shown in Figure 19.
between 20 to 82°, moments due to jacking forces between Residual stresses were measured for a single W6×9 beam
zero and 50 percent of member capacity, and vee depths with Category S damage which was repaired using the stan-
between 3/4 and full depth of the plate. The average values dard pattern. The maximum compressive stresses in the
are shown in Figure 15 and were similar to those obtained flanges approached yield while those in the web were some-
by Roeder (1985) as shown in Figure 14. what less. A comparison of the residual stresses for the
A second series of tests were conducted on plates which undamaged and damaged beams showed a reasonably good
had been damaged over the range of 6 to 24° corresponding correlation for Category S.
to a strain ratio, µ, ranging between 30 and 100. The num- The large residual stresses created during heat straighten-
ber of heating cycles required to straighten the plates ranged ing have several implications. First, if the member is a
from 25 to 100. The distribution of residual stresses at the compression element, the high residual stresses are similar

Fig. 14. Experimental strain and theoretical residual stress distribution

for 2/3 depth, 45° vee heated plate subjected to 1000°F temperature Fig. 15. Average residual stress values for vee heated plates which were
(Roeder, 1985). originally undamaged.


to welded built-up members. Since U.S. codes use a single ment could be reduced or even reversed, if the jacking force
column curve concept, these members are all treated the moment does not compensate for the residual stresses.
same and no capacity reduction would be assumed. How-
ever, if multiple column curves are used (typical of many DOES THE DEPTH OF THE VEE INFLUENCE THE
European countries), then heat straightened columns would AMOUNT OF MOVEMENT DURING HEAT
fall in a lower strength curve after heating due to residual STRAIGHTENING?
stresses. Consequently, there would be some loss of design
Past researchers (Nicholls and Weerth, 1972; Roeder, 1985)
have concluded that the plastic rotation (change in ϕd after
Second, high tensile residual stresses may reduce the
one heating cycle) is proportional to the depth ratio, Rd,
effectiveness of jacking forces by effectively canceling out
which is the ratio of vee depth, dv to plate width, W. A
the restraining compression associated with jacking. Move-
review of Roeder’s test data in the range of 650°C (±80°) or

Fig. 16. Stresses in angle L4×4

(45° vee, apex at heel, Mj/Mp = 0.50, depth ratio = 1.00).

Fig. 18. Residual stress distribution damaged, Category W wide flange

beams (assumed E = 200,000 MPa or 29,000 ksi).

Fig. 17. Stresses in channel IX-6 Fig. 19. Residual stresses in Category S damaged wide flange beam
(45° vee, Mj/Mp = 0.50, depth ratio = 1.00). (45° vee, Mj/Mp = 0.50, depth ratio = 1.00).


1200°F (±150°) is inconclusive as to vee depth effect. Rec- HOW DOES THE GEOMETRY OF THE PLATE
ognizing that the data was sparse, neither the depth ratio of ELEMENT EFFECT HEAT STRAIGHTENING?
0.75 nor 0.67 produced plastic rotations that were consis-
Researchers have generally considered plate thickness to
tently proportional to vee depth. To further evaluate this
have a negligible affect on plastic rotation. The only reser-
behavior, a series of tests was conducted for depth ratios of
vation expressed has been that the plate should be thin
0.5, 0.75, and 1.0 and vee angles ranging from 20° to 60°.
enough to allow a relatively uniform penetration of the heat
At least three heats were conducted on initially straight
through the thickness. The practical limiting value is on the
plates for each case and the results averaged. The results
order of 19 to 25 mm (3/4 to 1 in.). Thicker plates can be
are shown in Figure 20 for a combination of three depth
heated on both sides simultaneously to ensure a uniform
ratios, three vee angles and two jacking ratios. The jacking
distribution through the thickness or a rosebud tip can be
ratios reflect that a jacking force was used to create a
used. Similarly, researchers such as Roeder (1985) have
moment at the vee heat equal to either 25 percent or 50 per-
heated plates of various widths while keeping other param-
cent of the ultimate bending capacity of the plate. As can
eters constant. These tests showed no clear relationship
be seen from Figure 20, the depth ratios of 75 percent and
between the plastic rotation and plate width.
100 percent track each other well. In fact the 75 percent
depth ratio resulted in slightly larger plastic rotations in all
but one of the six cases. The 50 percent depth ratio resulted
in erratic behavior when compared to the other two. In
three of the six cases the 50 percent depth ratio produced
much smaller plastic rotations. In the other three cases, the One of the most important and yet difficult to control
plastic rotations were similar. parameters of heat straightening is the through-thickness
To further verify this behavior, a series of plates was temperature of the heated metal. Factors affecting the tem-
damaged and straightened. The degree of damage was large perature include: size of torch orifice, intensity of the flame,
enough that at least 20 heats were required for most of these speed of torch movement, and thickness of the plate. In his
plates. Therefore, more statistically significant average experiments Roeder (1985) made careful temperature
plastic rotations were obtained from these tests. Again the measurements of the heats produced by knowledgeable
pattern of plastic rotations does not have a direct correlation practitioners. He found that these individuals, when judg-
to the vee depth ratios. ing temperature by color, commonly misjudged by 56°C
Therefore, even though it would seem intuitive that (100°F) and, in some cases, as much as 111°C (200°F).
increasing the vee depth would increase the plastic rotation, Thus, there are considerable variations in temperature con-
there is no experimental justification for such a general trol, even with knowledgeable users.
statement. It can be concluded that the variation of vee Assuming adequate control of the applied temperature is
depth ratio between 0.75 and 1.0 has little influence on plas- maintained, the question arises as to what temperature pro-
tic rotation. However, a vee depth ratio of 50 percent may duces the best results in heat straightening without altering
reduce the plastic rotations. the material properties. Previous investigators have dif-
fered in answering this question. For example, Shanafelt
and Horn (1984) state that heats above 650°C (1200°F) on
carbon and low alloy steels will not increase plastic rota-
tion. Rothman and Monroe (1973) concluded that reheat-
ing areas where previous spot heats were performed would
Plastic Rotation (milliradians)

not produce any useful movements. However, Roeder

(1985) has shown that the resulting plastic rotation is gen-

erally proportional to the heating temperature up to at least
870°C (1600°F). To more clearly define the behavior sug-
gested by a limited number of data points in Roeder’s study,
2 a series of heats were applied to plates in which the heating
1 temperature was varied from 370 to 815°C (700° to
0.5 0.75 1 0.5 0.75 1 0.5 0.75 1 0.5 0.75 1 0.5 0.75 1 0.5 0.75 1
1500°F) in increments of 56°C (100°F). The results as
Depth Ratio
45 0 0
20 0 0 0 shown in Figure 21 establish a clear and regular progression
20 60 45 60
Jacking Ratio 0.5 0.5 0.5 0.25 0.25 0.25 of increased plastic rotation with increasing temperature.
Because increasing the temperature leads to more move-
Fig. 20. Influence of vee depth on plastic rotations of originally straight
plates for various vee angles and jacking ratios
ment, practitioners sometimes tend to overheat. As dis-
(heating temperature = 650°C or 1200°F). cussed previously, it is important to limit the maximum


temperature in order to insure that the material properties taken of both material and jacking constraints. For line
remain unchanged. heats in plates (particularly thin plates), jacking dominates
because thermal gradients are small.
While practitioners have long recognized the importance of A series of Category W damaged beams were re-damaged
applying jacking forces during the heat-straightening and straightened up to eight times, Avent, et. al., (1993a).
process, little research has been conducted to quantify its The beams had a degree of damage of approximately 7°.
effect. A series of tests designed to evaluate this parameter While the beams were successfully straightened, material
involved applying a jacking force to a plate such that a properties were noticeably affected by the fourth cycle of
moment is created about the strong axis in a direction tend- damage and repair. Both yield and tensile strength
ing to close the vee. This moment (at ambient temperature) increased, especially in the vicinity of the apex of the vee
is non-dimensionalized for comparison purposes by form- heats. The difference between tensile strength and yield
ing a ratio of the moment at the vee due to the jacking force, decreased with each damage-repair cycle indicating a trend
Mj, to the plastic moment, Mp, of the cross section, that is toward more brittleness. Ductility also decreased signifi-
Mj/Mp. This term is referred to as the jacking ratio. The cantly with successive damage-repair cycles. In another
tests included jacking ratios ranging from zero to 50 percent series of tests on composite girders (Avent, Madan, and
with four different vee angles and the vees extending over Shenoy, 1993b), a crack formed along a line heat during the
either 3/4 or full depth of the plate. The results shown in Fig- fourth damage-repair cycle.
ure 22 indicate how jacking can increase movement per Based on these results it is recommended that the same
cycle. regions not be heat straightened more than twice if re-dam-
Roeder (1985) also studied the effect of the jacking ratio aged. Should additional damage occur, the changes in
variation and found a similar pattern of behavior. However, mechanical properties as well as the possibility of cracking
the number of data points was limited. In general, the should be taken into account when deciding whether heat
amount of movement is proportional to the jacking ratio. straightening is more viable than member replacement.
Experimental evidence indicates that this relationship is
true for various cross sectional shapes. Several examples CAN THE AMOUNT OF STRAIGHTENING PER
are shown in Figures 23 to 25. This data shows that the HEAT BE PREDICTED?
restraint necessary during heating can result from: jacking
Two general approaches have been used to develop an ana-
forces, the temperature gradient of the cool surrounding
lytical procedure for predicting member response during a
material, or a combination of both. For any combination
heat-straightening repair. One approach involves finite ele-
producing the same restraint, the effect on the steel is simi-
ment/finite strip thermal and stress analyses including
lar. For example, in vee heated elements, advantage is
Plastic Rotation (milliradians)

0 10 20 30 40 50 60 70 80 90

Vee Angle (degrees)

Fig. 22. Influence of jacking ratio on average plastic rotation for ¾

Fig. 21. Influence of heating temperature on plastic rotation for 3/4 depth depth vee heats and 650°C (1200°F) heating temperatures
vee heats and a jacking ratio of 0.16. (lines represent a least squares curve fit).


6 inelastic behavior (Roeder, 1985). The stress and strain
equilibrium is evaluated over small time steps and takes
Plastic Rotation (milliradians)

Jacking Ratio=0.0 Jacking Ratio=0.50

5 Jacking Ratio=0.25
into account the influence of the non-uniform temperature
Jacking Ratio=0.50 distribution. This approach is a lengthy computational task,
Theoretical Eq. 6.1
Least squares curve fit
which is only possible using computer techniques and a typ-
ical analysis for a single vee heat can require extensive set
3 Jacking Ratio=0.25 up and computer time.
The other approach considers the global action of the
2 vee. The Holt equation is based on such an approach and
Jacking Ratio=0.0
assumes that perfect confinement is provided at all times
during the heating phase. As a result, longitudinal dis-
placements through the vee are linear. With this equation
the number of vee heats required to remove a bend in a steel
0 10 20 30 40 50
member can be simply calculated. However, the Holt equa-
Vee Angle (degrees) tion neglects the effect of restraining forces and temperature
Fig. 23. Experimental and theoretical plastic rotations for a
variation. A set of assumptions that are consistent with the
C 6x8 channel with Category S damage. observed behavior of vee heated flexural members include:
(1) longitudinal plastic strain occurs only in the middle two-
thirds of the vee heat zone (and in a reflected vee about the
Jacking Ratio=0.0
apex for partial depth vees); (2) at any specified distance
Plastic Rotation (milliradians)

Jacking Ratio=0.25 from the neutral axis of the plate, the strains are constant in
20 Jacking Ratio=0.50
Theoretical Eq. 6.1
Jacking Ratio=0.50 the longitudinal direction over the middle two-thirds of the
Least squares curve fit vee; (3) the planes defined by the sides of the vee remain
15 plane after heating and rotate about the apex of the vee; (4)
the plastic rotation per heat varies linearly with the jacking
10 Jacking Ratio=0.25 ratio, Mj/Mp; (5) the degree of internal confinement (exclud-
ing any jacking force) equals 60 percent of perfect confine-
5 ment; (6) the maximum heating temperature averages
1200°F (650°C); (7) the plastic rotation per vee heat varies
Jacking Ratio=0.0
linearly with the ratio of the product of the stiffening ele-
0 10 20 30 40 50
ment width, bs, and its distance from the apex of the vee, ds,
to the square of the vee depth, d; and (8) the movement per
Vee Angle (degrees) vee heat is a linear function of the ratio of plastic to elastic
Fig. 24. Experimental and theoretical plastic rotations for Category W section modulus, Z/S, about the axis of bending. Using
damage of a C 6x8.2 with the open end of vees at flange-web-juncture. these assumptions, an expression for the plastic rotation, ϕ,
per vee heat for Category S or W damaged wide flange
10 beams or channels or angles bent about their x or y axis is
Jacking Ratio=0.50
Jacking Ratio=0.0 given by
Plastic Rotation (milliradians)

9 Jacking Ratio=0.25

8 Jacking Ratio=0.50 ϕ p = Fl Fs Fa ϕ b (7)

Theoretical Eq. 6.1
7 Experimental least squares curve fit
where Fl , the factor associated with the external jacking
6 force is given by
5 Mj
4 F l= 0.6 + 2 (8)
Jacking Ratio=0.25 Mp
Fs, a factor reflecting the shape of the cross section, is
Jacking Ratio=0.0
given by
1 bd
0 Fs = 1 + ( s 2 s ) (9)
2 d
0 10 20 30 40 50

Vee Angle (degrees) Fa, a stress factor, is given by

Fig. 25. Plastic rotation versus vee angle for W6x9 using the Category LM FG 2 IJ FG Z IJ OP M
Fa = 1 − 2 1 −
j (10)
W heating pattern (Temperature = 650°C or 1,200°F).
N H 3K H S K Q M p


and ϕb is the basic plastic rotation factor for a rectangular on the beams or girders of composite deck-girder bridges.
plate having no external jacking force and is given by Heat straightening is an attractive repair alternative because
of its low cost and minimal disruption of traffic (Avent and
ϕ b = 0.0147 sin (11) Brakke, 1996). Typical damage includes: (1) a flexural
3 yield zone at the impact point in the lower flange due to
bending about the strong axis of the flange; (2) a yield line
HOW DO AXIALLY LOADED COMPRESSION or yield zone in the web due to bending of the web element
MEMBERS RESPOND TO HEAT about its weak axis; (3) reverse curvature flexural yield
STRAIGHTENING? zones in the lower flange at diaphragms on either side of the
impact point; and (4) localized bulges in the lower flange
For columns and axially loaded members it is important to
and web. Conceptually, vee heats are used to repair plate
consider the P-∆ effect. If an axially compressed member
elements with plastic bending about the major axis, while
is damaged by lateral loads as shown in Figure 26, a
line heats are applied to repair plate elements with flexural
moment is generated which is equal to P∆. This moment is
damage about the minor axis. Hence, a vee heat on the bot-
in the opposite direction to the moment generated by a jack-
tom flange in conjunction with a line heat on the web,
ing force during the straightening process. The moment due
applied to their respective plastically yielded portions, are
to the P∆ will oppose the restoration movement during heat
the proper heat patterns to repair composite beams
impacted by high loads (Figure 27). Care must be taken to
A series of full-scale tests were conducted on both Cate-
continually adjust the span of the line heats, so that only
gory W and S damaged axially loaded columns (Avent and
those portions of the web are heated that show plastic cur-
Mukai, 1998). Seven straightening procedures were con-
vature after the last heating cycle. Similarly, the vee heats
ducted with at least 10 heats used for each set of parame-
are confined to the portion of the bottom flange with plastic
ters. The axial load imposed during the straightening
deformations. In addition, a half-depth web strip heat is
process was typically 35 percent of the design allowable
applied. The purpose of this heat is to reduce the differen-
value. The jacking force was modified to cancel the P∆
tial shortening between web and flange. By heating the
moment effect in addition to providing the specified jacking
web with a half-depth strip, the web can deform and relieve
ratio. The response for both cases was significantly less
some of these stresses. The application of strip heats on the
than that predicted for a wide flange beam without axial
web do not influence the average plastic rotations apprecia-
load. From these tests it is concluded that heat straighten-
bly but, does tend to reduce the buckling of the web near the
ing is effective for axially loaded members using the same
center of damage.
heating patterns as for cases without axial loads. However,
the movement per heat will be somewhat smaller than if no
axial loads are present.
FOR LATERAL IMPACT OF THE LOWER When a jacking force is applied to the lower flange of a
FLANGE OF A COMPOSITE GIRDER? composite beam, the internal redundancy makes it difficult
to calculate the stress in the lower flange. Such difficulties
Perhaps the most typical type of damage found on steel
may occur on other types of indeterminate structures as
bridge members results from impact of vehicles or freight

Fig. 26. P∆ effect on an axially loaded column. Fig. 27. Heating patterns for composite girder.


well. The question is: how much of the jacking force is HOW CAN LOCALIZED DAMAGE
translated into lower flange moment and how much is trans- BE QUANTIFIED?
ferred through the web and into the deck? When applying
The focus of past heat straightening research has been on
a lateral load to the bottom flange near the center of the
various aspects of repairing global damage. However, it is
composite girder, the moment produced is transferred to the
a rare situation when localized damage doesn’t occur con-
end reactions by two mechanisms: the bottom flange acts as
currently with global damage. Yet, little published infor-
a flexural beam supported at the ends, and the web acts as a
mation has been available on how to repair local damage by
flexural plate (and, at large deformations, a membrane
heat straightening. As a result, localized damage is often
plate) supported by the deck and end diaphragms. As
repaired improperly by cold mechanical straightening and
demonstrated by Avent and Fadous (1989), a shallower
hot mechanical straightening, as well as heat straightening.
girder has greater stiffness and thus lower flange moments
Local damage patterns display two main characteristics:
than that of the deeper girder subjected to the same lateral
large plastic strains (usually tensile) in the damaged zone,
load. By adopting a jacking ratio definition which assumes
and bending of plate elements about their weak axis. If the
that all of the applied force is transferred into a lower flange
local damage is to be repaired, shortening must be induced
moment, Mj, which is resisted only by the plastic moment
in the damaged area equal to the elongation caused when
of the bottom flange, Mp, the implication is that most of the
the element was damaged. In addition, the distortion along
lateral stiffness is provided by the bottom flange. This
the yield lines must be removed as part of the repair
apparent jacking ratio can be misleading (Avent et al.,
process. Studies on global damage repair have shown that
1993b). It is more relevant to use an effective jacking ratio,
vee heated regions shorten significantly during cooling and
Mf, which reflects the actual moment in the flange to its
that line heats can be used to induce bending about the yield
plastic capacity, Mp. This relationship has been developed
lines. Thus a combination of line and vee heats can be used
as a function of the ratio of beam depth, d, to web thickness
to repair localized damage. It is helpful to separate local-
tw in the form
ized damage into two distinct categories: Category L/S
Mf = γ Mj (12) denoting bulges in plate elements with stiffening elements
on two sides (such as a web), and Category L/U denoting
bulges in plate elements having one unstiffened edge (such
d tw as a flange). Examples are shown in Figure 28.
γ= (15 + 2.75 d t w ) (13)
Based on sudden fractures that have occurred at higher
jacking ratios, it is recommended that the jacking ratio
Mf/Mp be limited to one-third.
While every damaged unstiffened plate element will have
CAN THE MOVEMENT PER HEAT OF its own unique shape, there is a set of common characteris-
COMPOSITE BEAMS BE PREDICTED? tics, which can be used to determine the heating pattern.
An equation for plastic rotation has been developed by
Avent and Mukai (1998) in the form of that for rolled
shapes, Equation 7, where ϕb is defined by Equation 11,
Fs = 1.0, γ by Equation 13, and
Fa =
FG d / t IJ
w (14)
H 46 K
Fl = 0.6 + 2
Mf I
GH Mp JK (15)

Fl = 0.6 + 2 γ (16)

Fig. 28. Typical localized damage classified as Category L.


Category L/U damage is commonly caused by impact to may form at the F side fillet. The angle formed by this yield
a flange on one side of the web. An idealized zone of dam- line is θf and the rotation of the F flange is θw - θf. The iden-
age is shown in Figure 29. The impacted side of the dam- tification of these yield lines is important in the repair pro-
aged flange will be referred to as the near side (N). The cedure.
non-impacted side of the flange on the other side of the web The heating/jacking pattern to straighten this damage
will also typically incur damage. This damage on the far will depend on how the geometry changes as heat straight-
side of the flange (F) has geometry similar to that of type N ening progresses. It is usually most effective to have jack-
but usually of lower magnitude. The damaged flange typi- ing forces on both the near and far sides of the flange.
cally undergoes rotation about a clearly defined yield-line However, straightening can be conducted with jacking only
near the fillet of the web. The impacted side of the flange on the near (impacted) side. The specific steps are:
(side N) usually deforms in a folded plate pattern. This
flange is shown deforming toward the web in Figure 29b. 1. Restraining forces
The deformation usually results in yield lines, which define
Place jacking forces on both the near and far sides of the
the edges of the folded plate (Figure 29c). In some cases,
damaged flange in the direction tending to restore the
particularly in regions of high curvature, the deformation
flange to its original condition. As shown in Figure 30a,
pattern may be one of a flexural yield zone rather than a
a convenient arrangement on the near side is to place a
series of yield lines. These zones result from plate element
jack, Pn, between the top and bottom flange. The far side
flexure and tend to spread over the surface as the degree of
jack, Pf, requires a clamping type force which is often
damage increases. Such zones will be referred to here as
more difficult to arrange in field applications. If the
yield surfaces. The other half of the same flange usually
clamping force cannot be anchored from the opposite
deforms in a similar pattern in the opposite direction, even
flange, a spreader beam arrangement can be used, as
if not directly impacted. The folded plate pattern, Figure
shown in Figure 30d, to anchor the reaction to the
29d, tends to have smaller deformations, thus δn > δf.
straight portions of the far side flange. An alternative is
Because the web is thinner than the flange, a yield line often
to only jack from the near side.
forms in the web near the fillet. The section shown in Fig-
ure 29b illustrates this behavior. The tee section at the However, the average movement per cycle tends to be
flange/web juncture remains a rigid right angle. The yield lower than similar cases jacked on both sides. Care
line forming in the web fillet allows this tee to rotate should be taken to limit the magnitude of jacking forces
through an angle θw. The yield line at the flange fillet on the used.
impacted side of the flange (side N) results from the addi-
tional rotation, θn, thus the total rotation of the N flange is
2. Vee heats
θw + θn. The other half of the flange (side F) tends to act as
a reaction against rotation thus a second flange yield line Although vee heats may not be necessary, a limited num-

Fig. 29. Heat straightening local flange damage (Category L/U).


ber may be used to assist in the flange shortening effort. section at a spacing of approximately bf/4 where bf is the
The vees should be approximately half depth and applied flange width. Similarly, line heats may also be used
to both the near and far sides of the flange to eliminate instead of vee heats on section BCDE. The order of heat-
global curving of the member. The vee should be narrow ing the yield lines tends to have a minor impact although
with an angle of 20° or less and the open end of the vees it is good practice to heat the ones at the largest damage
should be at the flange tips. It is best to place the vee locations first. It is also recommended to heat the near
heats in regions where no line heats are required. No side lines prior to the far side.
more than two vees should be used (preferably only one)
in one heating cycle. The location should be shifted with
4. Web line heat
each heating cycle so the same location is not re-heated
for at least three cycles. A typical arrangement is shown The web yield line should be heated last. It is typically
in Figure 31b. located at the fillet as shown in Figure 31c.
These four steps complete the cycle. The cycle should
be repeated until the flange is straight.
3. Line heats
Quite often this procedure can be used to nearly
A pattern of line heats should be placed so that the line straighten the section. However, the progress of the
heats correspond to yield lines. These should be heated movement should be observed to insure that over-
on the convex surface if possible. A typical pattern is straightening does not take place. If the near flange
shown in Figure 31a. For yield surfaces of continuous movement progresses too quickly, then θn may become
plastic strain, such as often occurs in regions such as zero prior to θw. This situation is shown in Figure 30b.
ABC in Figure 31a, line heats should be spaced over the Should this behavior occur, a modification should be
made in Step 3 for line heats. Rather than heating all
lines on the near side (31a), line 4 should not be heated.
If straightening progresses to the point that θf = θw,
then the far flange may over-straighten with the continu-
ation of heating. The pattern should be changed. The sit-
uation is depicted in Figure 30c. The modification is to
reverse the direction of the far side jacking force while
continuing the pattern. The force Pf will prevent over-

Fig. 30. Arrangement of restraining forces during various

stages of repair. Fig. 31. Arrangement of vee and line heats.


straightening while allowing the near flange and web to 2. Final Heating Pattern
continue corrective movement. Judgement is required in
As the crown section flattens, the heating pattern should
selecting the number of line heats to fit the specific dam-
be expanded into the reverse curvature regions. One
age. As straightening progresses, the number of lines
alternative is to expand the number of ring heats and
may be adjusted to fit the changing shape of the damage.
extend the radial heats as described in Step 1 and shown
by dashed lines in Figure 32b. The second alternative is
to use four-point star vee heats along with ring heats
instead of radials. The star vee pattern is shown in Fig-
ure 33. The vees are heated first, working from tip to
Stiffened elements of a beam are defined as those supported center. However, the central portion past where adjacent
on both sides by perpendicular elements. The web of a
wide flange beam is a typical example in that the two
flanges provide the support or stiffening effect. When stiff-
ened elements are damaged, the pattern formed is usually a
dish shaped bulge. This damage frequently approximates a
shallow spherical shell.
Judgement is required when selecting the heating pat-
terns for damaged stiffened elements. Heating the regions
of sharpest curvature with combinations of lines and/or nar-
row vees is the most effective approach. Always follow the
practice of heating only in the vicinity of regions with plas-
tic curvature. As straightening progresses, this region may
become smaller. The following methodology is recom-
mended for bulges in stiffened elements. The jacking force
is typically applied at the crown of the bulge using a
spreader plate.

1. Initial Heating Pattern

The typical bulge will have reverse curvature bending as
shown in Figure 32. The crown region should be heated
first with the torch on the convex side. As movement
progresses, the heating patterns can be expanded into the
reverse curvature region again with the torch on the con-
Fig. 32. Curvature and line heating patterns for category L/S damage.
vex side. The initial heating patterns should consist of
radial and ring line heats as illustrated in Figure 32. The
exact number of ring heats will depend on the size of this
region. It is recommended that the diameter of the small-
est ring be no less 51 mm (2 in.) and that spacing between
rings be at least 51 mm (2 in.). For large bulges the ring
spacing should be larger than 51 mm (2 in.). For cases
where the curvature is relatively uniform, equally spaced
rings may be used. However, a ring heat should be
placed at locations where sharp changes in curvature are
indicated. Heat the outer ring on the convex side first
and work inward. After the rings are heated, the radial
lines should be heated. Again, work from the outside in
but do not run the radial lines inside the last ring. Con-
tinue this pattern cyclically until the crown region begins
to flatten. Allow the steel to completely cool between
heating cycles.

Fig. 33. Star vee heat pattern.


Table 1. Recommended Tolerances for Heat Straightening Repair
Member Type Recommended Minimum Tolerance
English (in.) SI (mm)
Beams, Truss Members,
or Columns:
overall ½ in over 20 ft 13 mm over 6 m
sections including ¾ in over 20 ft 19 mm over 6 m
impact point
Local Web Deviations d/100 but not less than ¼ in. d/100 but not less than 6 mm
Local Flange Deviations b/100 but not less than ¼ in. b/100 but not less than 6 mm
Units of member depth, d, and flange width, b, are inches and millimeters, respectively, for
English and SI units
Tolerances for curved or cambered members should account for the original shape of the

vees intersect should not be heated. Vees should be nar- In order to provide some general guidelines, a set of rec-
row with an angle of 20° or less. After star vee heating, ommended tolerances is given in Table 1. These tolerance
the rings should be heated, starting with the outermost levels are less restrictive than specified in the Bridge Weld-
ring. A maximum of three ring heats should be used at ing Code.
one time with the star vees. Rotate the star vee pattern
30° after each cycle so that the same vees are not repeti- SUMMARY AND CONCLUSIONS
tively heated. Rings may be repetitively heated or
The goal of this paper has been to summarize the state-of-
shifted, depending on the degree of plastic curvature.
the-art in heat straightening repair of damaged steel. The
The steel should completely cool before the next heating
format used was to list the commonly asked questions and
cycle begins.
to provide the answers based on the latest research and
development. The principles discussed here can also be
WHAT TYPE OF TOLERANCES SHOULD BE applied to heat curving members to produce camber, sweep
USED IN HEAT STRAIGHTENING? or curved beams. Heat straightening is a safe and econom-
ical repair procedure when properly executed. It is a skill
In most cases any desired tolerance level can be obtained by
requiring practice and experience. The proper placement
heat straightening. One approach to specifying the toler-
and sequencing of heats combined with control of the heat-
ance for heat straightening repair is to use a standard spec-
ing temperature and jacking forces distinguishes the expert
ification for manufacturing or fabrication. One commonly
practitioner. In the past, heat straightening has been more
referenced standard is the American Welding Society
art than science. With the information presented here, the
Bridge Welding Code (1996 or latest edition). Member
practicing engineer will have guidelines quantifying the
dimensional tolerances are given in Section 3.5 and flat
most suitable procedures.
plate tolerances are given in Section 9.19. The use of this
or similar construction standards for heat straightening tol-
erances may be too limiting for the following reasons: (1)
The tighter the tolerances, the longer the repair will take. American Railway Engineering Association (AREA) (1946), “The
Costs and delays in using the structure will thus increase. Shortening of Eyebars to Equalize the Stress,” Bulletin No.
460, American Railway Engineering Association, Chicago, IL,
(2) In many cases, the area of damage is small and not nec- April.
essarily located at the most highly stressed region. Less American Welding Society (1996), Bridge Welding Code,
restrictive tolerances can safely be used in such instances ANSI/AASHTO/AWS D1.5-96, Miami, FL.
since the engineer can assess the specific structural config- Avent, R. R. (1989), “Heat-Straightening of Steel: Fact and
uration. (3) The steel in the vicinity of impact may have Fable,” Journal of Structural Engineering, ASCE, 115(11), pp.
material distortions, such as changes in thickness, which 2773-2793.
may preclude highly restrictive tolerances. As a conse- Avent, R. R. (1995), “Engineered Heat Straightening,” Proceed-
quence, engineering judgement for the specific situation is ings, National Steel Construction Conference, AISC, San Anto-
recommended as the most effective approach in setting tol- nio, TX, May.


Avent, R. R. and Brakke, B. C. (1996), “Anatomy of Steel Bridge Holt, R. E. (1965), “Flame Straightening Basics,” Welding Engi-
Heat Straightening Project,” Transportation Research Record, neer, 50(9), pp. 49-53.
No. 1561, TRB, National Research Council, Washington, D.C., Holt, R. E. (1971), “Primary Concepts in Flame Bending,” Weld-
pp. 26-36. ing Engineer, 56(6), pp. 416-424.
Avent, R. R. and Fadous, G. M. (1988), “Heat Straightening Pro- Holt, R. E. (1977), “How to Control and Correct Warping,” Weld-
totype Damaged Bridge Girders,” Journal of Structural Engi- ing Design and Fabrication, 49(6), pp. 98-102.
neering, ASCE, Vol. 15, No. 7, July, pp. 1631-1649.
Horton, D. L. (1973), “Heat Curved Mild Steel Wide Flange Sec-
Avent, R. R., Fadous, G. M., and Boudreaux, R. J. (1991), “Heat tions: An Experimental and Theoretical Analysis,” Thesis pre-
Straightening of Damaged Structural Steel in Bridges,” Trans- sented to the University of Washington, at Seattle, WA, in
portation Research Board, No. 1319 TRB, National Research partial fulfillment of the requirements for the Master of Science
Council, Washington, D.C., pp. 86-93. degree.
Avent, R. R., and Mukai, D. J. (1998), Heat Straightening Repairs Moberg, K. L. (1979). “Damage Assessment and Contraction
of Damaged Steel for Bridges, FHWA Report, U.S. Dept. of Straightening of Steel Structures,” thesis presented to the Uni-
Transportation, Federal Highway Administration, Washington, versity of Washington, at Seattle, Wash., in partial fulfillment of
D.C. the requirements for Master of Science Degree.
Avent, R. R., Robinson, P. F., Madan, A., and Shenoy, S. (1993), Nicholls, J. I., and Weerth, D. E. (1972), “Investigation of Trian-
“Development of Engineering Design Procedures for Heat gular Heats Applied to Mild Steel Plates,” Engineering Journal,
Straightening Repair of Damaged Structural Steel in Bridges,” Oct., pp. 137-141.
LTRC 251, Louisiana Transportation Research Center, Baton
Rouge, LA. Pattee, H. E., Evans, R. M., and Monroe, R. E. (1969), “Flame
Straightening and its Effect on Base Metal Properties,” Sum-
Avent, R. R., Madan, A., and Shenoy, S. (1993), “Design and mary Report to Ship Structure Committee concerning first
Implementation of Heat-Straightening Repair for Composite phase of Project SR-185, Straightening Distorted Weldments,
Deck-Girder Bridges,” Transportation Research Record 1392, Battelle Memorial Inst., Columbus, Ohio, August.
TRB, National Research Council, Washington, D.C., pp. 90-98.
Pattee, H. E., Evans, R. M., and Monroe, R. E. (1970), “Effect of
Avent, R. R. (1995), “Engineered Heat Straightening Comes of Flame and Mechanical Straightening on Material Properties of
Age,” Modern Steel Construction, Vol. 35, Weldments,” Summary Report on Ship Structure Committee on
No. 2, February, pp. 32-39. Project SR-185, Straightening Distorted Weldments, Battelle
Avent, R. R. and Wells, S. (1982), “Experimental Study of Thin- Memorial Inst., Columbus, OH.
Web Welded H Columns,” Journal of the Structural Division, Roeder, C. W. (1985), “Use of Thermal Stress for Seismic Dam-
ASCE, Vol. 108, No. ST7, July. age Repair,” Final Report on NSF Grant CEE-82-05260, Univ.
Brockenbrough, R. L. (1970a), “Theoretical Stresses and Strains of Washington, Seattle, WA, October.
from Heat Curving,” Journal of the Structural Division, ASCE, Roeder, C. W. (1986), “Experimental Study of Heat Induced
96(7), pp. 1421-1444. Deformation,” Journal of Structural Engineering, ASCE,
Brockenbrough, R. L. (1970b). “Criteria for Heat Curving Steel 112(10), pp. 2247-2262.
Beams and Girders,” Journal of the Structural Division, ASCE, Rothman, R. L. (1973), “Flame Straightening Quenched and
96(10), pp. 2209-2226. Tempered Steels in Ship Construction,” Report No. 247, Ship
Brockenbrough, R. L., and Johnston, B. G. (1968), USS Steel Structs. Committee, U. S. Coast Guard, Washington D.C.
Design Manual, U. S. Steel Corp., Pittsburgh, PA. Rothman, R. L., and Monroe, R. E. (1973), “Effect of Temperature
Ciesicki, H., and Butler, S. (1968), “Flame and Camber Your Own and Strain Upon Ship Steels,” Report No. 235, Ship Structs.
Beams,” Industry and Welding, 31(4), pp. 76-79. Committee, U. S. Coast Guard, Washington D.C.
Welding Engineer (1959), “Flame Buckled This Steel and Flame Shanafelt, G. O., and Horn, W. G. (1984), “Guidelines for Evalu-
Straightened It, Part 1,” Welding Engineer, 44(2), p. 43. ation and Repair of Damaged Steel Bridge Members,” NCHRP
Holt, J. E. (1955), “Flame Straightening: A Friend in Need,” Report No. 271, Transportation Research Board, National
Welding Engineering, 40(10), 44-46, (12), pp. 30-31. Research Council, Washington, D.C., June.