Академический Документы
Профессиональный Документы
Культура Документы
t -B
c/ Repairs of Damaged
U.S.Department
of Transportation
Federal Hlghway
Steel Bridges
Administration
A Technical Guide and
Manual of Practice
-..
Heat-Straightening
Repairs of Damaged
U.S.Department Steel Bridges
of Transportation
Federal Highway A Technical Guide and
Administration
Manual of Practice
Report No. FHWA-IF-99-004
Prepared by
Prepared for
October 1998
1. Report No. 2. Government Accession No. 3. Recipient’s Catalog No.
FHWA-IF-99-004
4. Title and Subtitle 5. Report Date
Heat-straightening Repairs of Damaged Steel Bridges October 1998
A Manual of Practice and Technical Guide 6. Performing Organization Code
Foreword
The manual is divided into three parts. Part I provides a background and -
overview of the heat-straightening process. It is intended for an audience ranging
from practitioner to contractor to supervisor to bridge engineer. Part II is a
technical guide intended primarily for engineers. This section provides detailed -
technical information on most aspects of heat straightening. Part Ill contains
guides, specifications and reference material. -
David Densmore
Bridge Office
Office of Infrastructure
Notice
Acknowledgements
The following project review committee provided invaluablereview and commentson
the manualduring its preparation.
KrishnaIS. Verma David L. McQuaid
Welding and StructuralEngineer D & L, Inc.
Bridge Office Four GatewayCenter, 12* Floor
Office of Infmstructure Pittsburgh,PA 15222
FederalHighway Administration
400-7thSt., S.W. Room 3203 Jon J. Edwards,Chief
Washington,DC 20590 ShopPlansandFabricationUnit
Bureauof Bridgesand Structures
Jeffrey W. Post Illinois Departmentof Transportation
J.W. Post andAssociates,Inc. 2300SouthDirksen Parkway
19834SundanceDr. Springfield,IL 62764
Humble, TX 77346
Fred Beckman
Dean C. Krause 167HawthorneSt.
SeniorMetallurgicalEngineer Chicago,IL 60411
BethlehemSteelCorp.
1170Eighth Ave. Jerry Hill
Bethlehem,PA 18016 G. Hill andAssociates
3843 SandhillRd.
Lansing,MI 48911
This project was a subcontractwith SouthernUniversity as a part of their contract with FHWA.
The principal investigatorswere:
Dr. I. K. Dabipi, Chairman Dr. JudyPerkins
Departmentof Electrical Engineering Departmentof Civil Engineering
SouthernUniversity University of New Orleans
BatonRouge,LA 70813 New Orleans,LA 70148
The following organizationsare acknowledgedfor the supportof the researchupon which most
of this manualis based.
Department of Civil and Environmental LouisianaTransportationResearchCenter --
Engineering LouisianaDepartmentof Transportationand
LouisianaStateUniversity Development
BatonRouge,LA 70803 Baton Rouge,LA 70803 -
LENGTH
-.
MASS (weight)
MASS (weight) -
-
VOLUME
.--
-.
PurposeofThisManual.. ...................................... .3
History of Heat Straightening.................................... 3
TypicalTypesofDamage ....................................... 5
1. CategoryS....................................... ...5
2. CategoryW......................................... 5
3. CategoryT....................................... ...6
4. CategoryL ....................................... ...8
Objectivesof This Manual ...................................... 8
HowtoUseThisManual.. ...................................... .
Chapter 2. Heat Straightening Basics. ................................. 11
-
Chapter 3. Assessing, Planning and Conducting Successful Repairs . . . . . . . .35
Introduction.................................................. 97
ExperimentalProgram..........................................lO 0
Evaluationof Resultsof ExperimentalProgram...................... 100
VeeAngle.............................................lO 1
DepthofVee...........................................lO 1
PlateThicknessand Geometry. ............................ 102
Temperature............................................lO 7
RestrainingForces..................................... .108
AnalyticalDevelopment ....................................... 112
Example5.1..................................................117
Significanceof PlateResponseto Heat Straightening................. 117
Key PointstoRemember...................................... .118
Introduction..................................................16 7
Responseof Columnsto Heat Straightening........................ .167
Heat StraighteningResponseof ColumnsWith CategoryW Damage..... 168
Responseof Columnsto Heat StraighteningFor CategoryS Damage..... 172
summary ................................................... 173
Key Pointsto Remember....................................... 173
Sectionl.General.............................................217
Section2. DamageAssessment.................................. .217
Section3. Material Assessment............................... ..
f .220
Section4. Designof Repair Sequence............................. .222
Section5. Field Supervisionof Repair. ............................ 225
Chapterl3.Glossary................................................23 3
Chapter 14. Nomenclature. .......................................... 239
Chapterl5.Bibliography............................................24 3
-.
...
Federal High way Administration XII1
-
--
--
--
--
-.
--
Little information hasbeen available search,it was indeed true that the artesian
in terms of quantification of the heat- practicing the trade was much more impor-
tant than the engineer. Consequently,heat- -
straighteningprocess. The earliest written
information found was tracedto JosephHolt straighteningrepair was often not considered
who defined some of the basic conceptsof on engineeredstructures.
-
heat straighteningin an unpublishedmanu- In recentyears,considerableresearch
script in 1938. Over the years since, more has been conducted to quantify the heat-
publicationsbegan to appearwhich tended straighteningprocess. The technical data
(a) Category W damage on a built-up double channel (b) Category W damage to main girders caused by
truss member. The damage was caused by a log fal- wind during construction of a Louisiana bridge.
ling from a truck on a bridge in North Louisiana.
3. Category T
This type refers to damageas a result below yield. If one flange is constrained -
of torsion or twisting about the longitudinal (such as the case of a composite bridge
axis of a member. For rolled or built-up girder), then the unconstrainedflange ele-
shapes,the flange elementstend to exhibit ment is subjectedto plastic deformation and -
flexural plastic deformation in oppositedi- yielding may also occur in the web. Exam-
rections. Theweb is often stressedat levels ples are shownin fig. 1.3. -
(b) Category T damage on a composite bridge girder impacted by an over-height vehicle in Wisconsin.
-
the stepsto be consideredTom assessment accessmaterialinteractively. It containsan
of damageto planning the repair processto overview section as well as sections on
supervisingthe repair. The technical speci- management,designand techniquesfor heat
fications are given in a form which may be straightening.An index providesa listing of
incorporatedinto contracts or may be suit- commonly usedheat straighteningpatterns.
ablefor inclusionin codesand standards. The program also provides a step-by-step
In conjunctionwith the development exampleof how to conduct a bridge repair.
of this manual, a multimedia instructional The video is available in VHS format. It
computerprogram and video havebeenpre- provides an overview of how to conduct
pared. The fundamentalsof heat straight- heat-straighteningrepairs. This manualpro-
ening are illustrated in both. The multime- vides background material and more in-
dia computer program is available on a depth information as a supplementto the
compact disc (CD) and allows the user to computerprogram andvideo.
.-
_-
Figure 2.6. Strip heat in progress with a completed strip heat in the foreground. .-
(fig. 2Sc), the restrainingmoments tend to metal is heated. A spot heat causesupset-
prevent transverseexpansionbelow the plate ting of the metal through the thicknessdue
centerline. In a manner similar to the vee to the restraint provided by the cool sur-
heat mechanism,the material thus tends to rounding material. On cooling, a spot heat
expand through the thickness,or “upset”. leavestensile stressesin all the radial direc-
Upon cooling, the restrainingmomentstend tions acrossthe heatedarea. During a spot
to magnify transverse contraction (fig. 2.5d). heat, the torch shouldnot be held at a par-
The speedof the travel of the torch is criti- ticular point for too long, asthe spotmay get
cal as it determinesthe temperatureattained. too hot and buckling may occur due to ex- -
Elwation
Wide flange bent about major axis Wide flange bent about weak axis
I
Plan Plon
IIon m sactlon
Elavaiion
Elevation
Channel bent about weak axis
Channel bent about major axis
-
(b) Category W damage (weak axis)
(a) Category S damage (strong axis)
_.
section
Elevation
Plan
..... .....” .. .
II
SOCiiO4l
Elwatlon
1 section
Elevation
-
Figure 2.12. Heating patterns for wide flanges and Figure 2.13. Wide flanges and channels with
channels bent about their minor axes twisting damage (Category T).
(Category W).
-
-
’ Llns heats ’ Vee ’ heat star pattern
(I heats (typ)
-
Figure 2.14. Typical heating patterns for local Figure 2.15. Heating patterns for angles.
damage.
---I
Figure 2.21. Characteristics of plastic flow and restraint during heat straightening.
3 75 Rosebud 5
>4 100 Rosebud 5
29
Federal Highway Administration
Practical Considerations of heating. A less intense flame allows for
This description of the heat straightening a slower pace as the torch is moved along __
process provides the basic methodology. the path. The intensity may be adjusted so
However, the proper application of heat is a as to compensate for variables encountered
skill requiring practice and experience and at in the field. -
this juncture, the art of heat straightening Material configuration.-The pace
meets the technology. The practitioner of moving the torch along the path will be a
needs to understand the variables involved function of the configuration of the member,
in the process and how to control them. location of damage and fabrication details.
Some of the more important variables are At the initiation of heating, the torch typi-
discussed here. cally remains on a single spot as the tern-
Torch Tip Size and Intensity.-The perature rises. Once the heating temperature
amount of heat applied to a steel surface is a is reached, a steady movement along the
function of the type of fuel, the number and path of heating can usually be maintained.
size of the orifices as well as the adjustment However, backside attachments such as
of the fuel pressure and intensity at the noz- stiffeners may serve as a heat sink requiring
zle tip. Selecting the appropriate tip size is the slowing of the torch movement over
primarily a function of the thickness of the certain zones.
material. The goal is to rapidly bring the One typical example is the heating of -
steel in the vicinity of the torch tip to the the flange of a rolled beam where the web-
specified temperature, not just at the surface, flange juncture must be heated more slowly
but through out the thickness. Once this since the web draws heat away from the
condition is obtained at the initial heating flange. Sometimes the pace must be quick-
location, the torch should be moved along ened to maintain a uniform heat. A common
the path at a rate that brings successive sec- example is at the conclusion of a vee heat at -
tions of steel to the specified temperature. A a free edge. By the last pass along that edge,
tip that is too small for the thickness will the wave of heat moving down the vee al- _
result in an inadequate heat input at the sur- most overtakes the torch. As a result, the
face that does not penetrate adequately last pass is usually conducted very quickly.
through the thickness. If the tip is too large, Practice heats will allow you to develop a -
there will be a tendency to input heat into feel for how to vary the torch speed over
the region so quickly that it is difficult to various configurations.
control the temperature and distortion. table Judging the Temperature.-In the- -
2.1 is a general guide for selecting a tip size. ory, control of temperature may seem easy:
However, this table is only a guide. Inten- watch the color of the steel and use tem-
sity of the torch, ambient temperature, steel perature crayons. In practice, temperature -
configuration, access, and fabrication details control is quite difficult. First, the satiny
may influence the choice of tips. Adjust- silver color of steel indicating 650°C
ments can also be made in the torch intensity (1200°F) is often obscured. The torch flame -
to improve the heating response. A hotter often reacts with surface impurities includ-
flame is helpful if the configuration of the ing paint, oil or previous temperature crayon __
steel tends to draw heat away from the spot
-.
R R
(b) w
Figure 3.2. Offset measurements to calculate de- Figure 3.3. Relationship of degree of damage to
gree of damage and radius of curvature. radius of curvature and cord length.
R L2
R= cd
110
100
90
80
70
60
50
40
30
20
10
0
1 10 100 1000
R/Y max
Figure 3.5. Strain ratio vs. normalized radius of curvature (p vs. IUymd.
where R is the actual radius of curvature in one measureof the extent to which the steel -
the damagedregion. has been damaged. From eqs. 3.7 and 3.8,
the strainratio is -
Since damage measurements are
taken at discretelocations,the radiusof cur-
vature can be approximatedfrom eq. 3.2 or -
3.3. Oncethe smallestradiusof curvatureis (Eq. 3.9)
determined in the damaged region, the
maximum strain can be computed from eq. -
3.7 andcomparedto the yield strain Since E is a constant for all steel grades
(200,000 MPa or E=29,000 ksi), p can be
obtained graphically in terms of the ratio -
E, =- 5 (Eq. 3.8) WYmaxand F, for various steel grades as
E shownin fig. 3.5. A similar approachcanbe
used for localized bulges, buckling or --
crimps.
The ratio of maximum strain to yield strain,
referred to as the strain ratio, p, is used as -
Figure 3.6. Offset measurements for example 3.1(25.4 mm = 1 in and 0305 m = 1 ft).
44
Federal Highwcry Administration -
-.
Horizontal Ditplacsment Component, a (in.) Hotizonlol Dtrplacrment Component, a (in.)
0 4.1 8.2 12 15.4 l&4 20.6 22.6 23.6 24 0 1.7 3.4 5 6.4 7.6 6.6 9.4 9.8 10
0.4
0.3
0.2
0.1
0 “0 10 20 30 40 50 60 70 60 90
0 10 20 30 40 50 60 70 80 90
Angle of Rotation, 0 (degrees)
Angle of Rotation. 0 (dograw)
Figure 3.7. Reduction in effective section modulus Figure 3.8. Reduction in effective section modulus
for a W 24x76 beam subjected to varying degrees for a W 10x39 beam subjected to varying degrees
of idealized damage (Note: 25.4 mm = 1 in). of idealized damage (Note: 25.4 mm = 1 in).
The bottom flange is modeledin two ways: spondsto a rotation of approximately 10”.
eitherit remainsparallelto the top flange, or In referenceto the field tests conductedon
it remainsperpendicularto the web. Actual the WlO X 39 and W24 X 76 beamsas de-
combinationsof damageoften fall between scribed in Chapter 7, the damage induced
thesetwo conditions. Plotted in fig. 3.7 and was consideredto be moderate. In both
3.8 are the variations in the selection cases the flange remained almost perpen-
p4 modulus(for bending about the strong axis) dicular to the web. The calculatedbending
associatedwith different levels of damage strength reduction for the WlO X 39 beam
for two beams:a W24 X 76 and WlO X 39. was calculatedin the range of 25 percent.
The case of the bottom flange remaining The reduction for the W24 X 76 beam was
perpendicularto the web is the more critical on the order of 9 percent.
casefor the comparisonof sectionmodulus
P-A Effects.-For compressionmem-
values. As can be observed, the section
bers, the square of the minimum radius of
modulusdips fairly rapidly with an increase
in the crosssectionrotation. A 10” rotation gyration is the section property associated
results in a strength reduction within the with the strengthof the member. The effect
range of 8-15 percent, depending on the of the two idealizedcasesof damageprevi-
section, while at 20”, strength reduction is ously described is plotted in fig. 3.9 and
3.10. In this case, the configuration in
between 18 and 29 percent. Although an
which the bottom flange remainsparallel to
engineershould evaluatethe specific condi-
the top flange is the more critical. The
tions and config.urationof eachcase,a good
curves are very similar for both wide-flange
generalguidelineis to repair the member if
sections. The reduction in strength, as
the sectionmodulus is reducedby 10 per-
measuredby the squareof the radius of gy-
cent. This level of damagetypically corre-
ration, is not quite as large as the
0.8 _._
\ --
0.7 0.7
\ \, -b a
0.6 0.6 \
\
0.5 0.5
\
0.4 0.4
0.3 0.3
0.2 0.2
I I I I I -\I- I
0.1 0.1
.
0
“0 10 20 30 40 50 60 70 80 90 0 10 20 30 40 50 60 70 80 90
Angle of Rotdon, 0 (degrees) Angle of Rotation, @ (degrees)
Figure 3.9. Reduction in the square of the effective Figure 3.10. Reduction in the square of the effec-
minimum radius of gyration for a W 24x76 beam tive minimum radius of gyration for a W 10x39
subjected to varying degrees of idealized damage beam subjected to varying degrees of idealized
(Note: 25.4 mm = 1 in). damage (Note: 25.4 mm = 1 in). -
-.
_-
0 I I I I I I I I I I I
0 2 4 6 8 10 12 14 16 18 20
Amplification Facior
Figure 3.11. Effect of amplification factor for lateral deflections on compression members.
-
-_
corresponding case for section modulus. lowable values. In deciding upon this
The reductionis only about 5 percentfor the value, it was assumedthat relatively small
10” rotation and about 14 percent at 20” ro- initial values of lateral deflections would
tation. However, another aspectthat must exist due to lateral loads or fabrication im-
be consideredwhen evaluatingcompression perfections, e.g., within the elastic range.
membersis the strength reductiondue to the When a compression member has larger
P-deltaeffect. If a simply supportedcolumn permanentdeformationswell into the plastic
has an initial midpoint deflection,y,, due to or strain-hardeningrange due to damage,
impact damage, then the deflection (and then the effective strength of the member is
bending moment) is amplified accordingto reducedby a larger factor than expressedby
the amplification factor eventhe columnsafety factor.
In light of theseconsiderations,even
relatively small permanent deformations
A.F.= ” (Eq. 11.6) shouldbe repairedfor compressionmembers
l-pKuk* unlessadditionalbracing is addedor a sta-
bility analysisis performed to justify that the
strengthreductionis small.
where P is the axial load and Peuler is the
Euler buckling load. This factor is taken Residual forces.-The analysisof re-
into accountin design codesby an adjust- sidual forcesin damagedsystemsrequiresa
ment in the safetyfactor for columns. plastic analysis. To illustrate the procedure,
a bridge girder laterally supportedby dia-
Consider the DISC code (1989), for
phragms will be used. For lateral impact,
example. The long column formula (eq. E2-
such asmight occur with an over-heightve-
2) is the classicalEuler buckling formula,
hicle, the girder acts as a continuousbeam
divided by a safety factor of 23/12. Con-
with the diaphragmsas interior supports. If
versely, the safety factor for tension mem-
an impact load occurs, the lower flange has
bers is given as 5/3. The reason for the
positive bending at the impact point and
higher safety factor for compressionmem-
negativebendingat the adjacentdiaphragm
bers is to accountfor the P-deltamagnifica-
supports. A layout modeling this type of
tion effect. A plot of the amplification fac-
girder is shownin fig. 3.12. During impact,
tor is given in fig. 3.11. As the load ap-
it is assumedthat plastic hinges form at the
proachesthe critical buckling load, the de-
impact point and at adjacent supports.
flection (and consequentlythe moment) ap-
Thesehingesform a mechanismfrom which
proachesinfinity. Failure must thereforebe
the impact load can be computed. Using a
defined as the point where the deflection
plastic analysis,the load, P,, can be calcu-
(and consequentlythe moment) remains fi-
lated and a moment diagram constructed,
nite but becomes excessively large. The
fig. 3.12b. The impact load is now applied
safety factor for column buckling was there-
in the reversedirection and an elasticanaly-
fore increasedby 0.25 above that used in
sis performed,fig. 3.12~. The superposition
tensionmembers. As can be seenfrom fig. of these two diagrams (b and c) give the
3.11, this extra safety factor accounts for
0.08 of the total load ratio reduction to al-
restrolnlng A
moments from
applied
restraining A mom’“‘~~~)
‘yg
(\~j)
C ” (d) Bottom flange showing
(c) Bottom Flange line heats only
Figure 3.14. Heating pattern and sequence for Figure 3.15. Heating pattern and sequence for
bending combination about both the combination weak axis bending and local
strong and weak axis. flange bulge.
4. Select Heating Patterns and Pa- fig.3.15. Vee heatsare usedon the top and
rameters bottom flangesalong with a web strip heat _
Typical Heating Patterns.-The fun- similar to the standard weak axis pattern.
damental heating patterns have been de- However, partial depth vees are usedon the
scribedin Chapter2. Sincetypical damage flange with the bulge along with a seriesof _
is often a combinationof thesefundamental line heats along bulge yield lines. Since a
damage types, a combination of heating yield line is likely to occur at the lower web
patternsis often required. The key is to se- fillet, a line heat is also neededon the web. -
lect the combination of patterns to fit the Restrainingforcesareusedto createbending
damage. When in doubt, a good policy is to moments about the weak axis as shown in
addressthe attention to one of the basic fig. 3.15. In addition, a jacking force should -.
heatingpatternsat a time. For example,re- be applied on the local bulge as shown on
move the Category W damageprior to ad- the cross sectionin fig. 3.15. The sequence
-
dressingthe CategoryL damage. It should of heatsis alsoindicatedin the figure.
be notedthat with proper combinations,sev- A third exampleis damageresulting
eral types of damagecan be removed expe- from impact of a composite bridge girder -
ditiously. For example,supposethat a wide which produces weak axis damage to the
flange section is impacted such that the bottom flange and twisting due to the re-
bendingoccursabout an axis at an arbitrary straint of the top flange. Theheatingpattern --
angleto the principal axes,i.e., bending oc- is shown in fig. 3.16 consistingof a bottom
curs about both the strong and weak axis. flange vee heat, a web strip heat and a line
The heatingpattern, fig. 3.14, requiresa vee heatat the top fillet of the web. The heating -
heat on the web to restore the strong axis sequenceis shown in fig. 3.16 aswell as the
damageand vee heats on the flanges to re- restraining moment required on the bottom -.
store the weak axis damage. The heats flange.
should be executed sequentially as num- A final exampleis the caseof multi-
bered in fig. 3.14. Note that no stip heat is ple plastic hinges formed about the weak -
required on the web since a vee is used axis such as might occur for a beam con-
there. Restrainingforces should be used to tinuous over interior supports. The heating
produce bending moments about both the pattern is shown in fig. 3.17. Note the re- --
strong and weak axis as indicated in fig. versed direction of the vees to reflect the
3.14 tendingto straightenthe damage. Once multiple curvaturedamage. The restraining
the damageis corrected about one of the moments must also reflect the reverse cur- -
principal axes, the heating pattern should vature natureof the damageas shown in the
revert to one of the fundamental patterns figure.
until straightening is complete about the -
otherprincipal axis. Vee Depth.-In generalthe vee depth
should be equal to the width of the plate
As a second example, consider a being straightened. Partial depth vees pro- --
wide flange beam with weak axis bending vide no advantagesin reducingshorteningas
damagecombinedwith a local bulge in one somehavespeculated.The primary situation
flange. The heating pattern is shown in -
(c) Side
Figure 3.16. Heating pattern and sequence for Figure 3.17. Heating pattern for reverse curvature
combination of weak axis bending of lower flange bending.
and twisting.
Number of Simultaneous Vee
for half depth vees is in the repair of local Heats.-Simultaneousvee heatsmay be per-
damage. formed with proper spacing. It is recom-
Vee Angle.-The angle of the vee is mendedthat the veesbe spacedat least one
usually limited by practical considerations. plate width, W, apart. Also, if multiple
It shouldbe as large as practical for the spe- plastic hinges occur, each hinge may be
cific application. If the open end of the vee heatedsimultaneously.
is too wide, out-of-plane distortion often oc- 5. Developa Constraint Plan
curs. Likewise the vee areashouldbe small
enoughto heat quickly so that differential Sincejacking forces can expeditere-
pairs, it is recommendedthat such forces be
cooling is limited. A good rule of thumb is
to limit the open end of the vee to approxi- utilized. Jacksshouldbe locatedto produce
their maximum effect in the zonesof plastic
mately one-third to one-half the plate width
deformation. It is recommendedthat jacks
but not greaterthan 254 mm (10 in). These
always be gauged and calibrated prior to
limits translateroughly to 20-30” vee angles.
use. Also, jacks must be properly securedso
If the width of the open end of the vee, V, is
they will not fall out as pressure subsides
selected,the vee angleis
during cooling. The loads applied to the
structure should always be known and lim-
V iting valuesestablished.A jacking arrange-
8 = 2 tan-’ - (Eq. 3.10) ment for a compositegirder bridge is shown
2w
in fig. 3.18. Lateral forces are utilized on
the lower flanges, fig. 3.18a, while jacks
where W is the platewidth. between flangesare used for local damage,
fig. 3.18b.
.-
i -
(9 @I -
Figure 3.18. Jacking arrangements for global and local damage on a composite girder bridge.
For cases where residual moments use a jacking moment of only ?4 Iv$, during -
are small, the jacking moment, Mj, should be the first two cycles.
limited to On occasion, a hairline fracture will _._
occur or become visible during an interme-
diate cycle of heat-straightening repair. The
M,I- MP causes are believed to be: (1) excessive re- -
2 straining forces being applied during the
heating process; (2) repetitive repair of a re-
damaged element; and/or (3) the growth of -
where I$, is the plastic moment capacity of micro cracks initiated during the induction
the member. of damage. As item (1) is the primary cause,
For cases where residual moments restraining forces should always be specified -
exist, the jacking moment should be limited at safe limits and should be monitored dur-
to ing actual repair. For item (2) the repair of
previously heat straightened material should -
Mj ~~(M~ +M,) be limited to only two damage/repair cycles.
One problem associated with the -
computation of jacking forces is that for in-
where M, is the residual moment and is determinate members, the bracing, dia-
positive when tending to straighten the phragms or other attachments may be diffi- -
member. Residual moments will be relieved cult to model. In addition, it is sometimes
during the first few heats. Rather than com- necessary to make an estimate in the field as
puting residual moments, an alternative is to to the magnitude of jacking forces. The -
50.4ft* k pi
-=- 25.2
20.2 61.5
50.4ft.k Pj = 8.2Skips (36.8k-N)
(c) Moment diagram due to impact
--
.--
%
Cementite and pearllte $2 -
s
0 1 I I I
0 0.8 2 4 4.3
Percent Carbon
-L
ductility of steel allows for a moderating
redistribution of residual stresseswhen a
member is subjectedto large loads, and (2)
since residualstressesare self-equilibrating,
largecompressivestressesat one location on
a cross section are balanced by tensile
stressesat another location. As a conse-
quence,the stressesat a specific cross sec-
-6-4-2 0 2 4 -60-40 0 40 60 tion productedby applied loads is additive
Strain (in/in ~10~) Normalized Residual Stress to the residual stressesat some points and
(Percent of Yield)
are subtractiveat others. The result is that
the ultimate strengthof a member is usually
not affected by residual stresses. The ex-
Figure 4.2. Experimental strain and theoretical
residual stress distribution for 2/3 depth, 45’ vee ception is compressionmembers in which
heated plate subjected to 1000°F temperature high residual stressesmay reducethe buck-
(Roeder, 1985). ling strength. American design codes ac-
(Brockenbrough1970a)on heat-curvedplate count for residual stressesin compression
girders subjected to line heats. These members by assumingan averageresidual
stresses, determined by the “sectioning stressvalue of 50 percentof the yield stress.
method”, were reasonably consistent with This assumption may lead to somewhat
the theoretical values, Similar theoretical conservative designs for rolled shapes
methodswere used on vee-heatedplate ele- (which have smaller residual stresses)and
ments (Nicholls and Weerth 1972) and on slightly lessconservativedesignsfor welded
wide flange beams (Horton 1973). How- built-up shapes(which have larger residual
ever, the resultswere not supported by any stresses). Europeancodeshave adoptedthe
experimentaldata. multiple column curve approachin which
different formulas are used depending,on
Significant residual stressesoccur in the magnitudeof residualstresses.For these
most structural steel members. Such codesthe level of residual stressaffects the
stresses usually result from differential designcapacity.
shrinkageduring cooling in the manufacture
of both rolled and welded built-up shapes. Avent, Robinson, et. al. (1993) have
However, the cutting and punchingprocess conductedresearchto provide insight as to
during fabricationmay alsoproduceresidual whether heat straightening produces some
stresses. Residual stressesare quite high negative effects due to residual stresses.
andvaluesmay reach50 percentof yield for The study included: both plates and rolled
some rolled shapesand approachyield for shapes;variations in vee angle, vee depth
some welded built-up members. With one and level of externalrestraining forces; and
degree of initial damage. Residual stress
exception, residual stresseshave been ne-
glectedin coderequirementsgoverningsteel patterns were determined by using the
“sectioningmethod”, a well-established,but
300 -200 -100 0 100 200 300 -300 -200 -100 0 100 200 300
8 a
4
-
3
1
::
-50 40 -30 -20 -10 0 10 20 30 40 50
-50 -40 -30 -20 -10 0 10 20 30 40 50
Resaual Stress (ksi)
ReskJual Stress (ksi)
Figure 4.3. Measured residual stresses in a vee Figure 4.4. Average residual stress values for vee
heated plate prior to heating. heated plates which were originally undamaged.
destructive procedure. After taking initial (Avent and Wells 1982). A series of 100 x 6
distance measurements between two fixed x 610 mm (4 x */4 x 24 in) long, initially
points on the steel, a narrow strip containing straight A36 steel plates were vee heated
these points is cut out (by milling to avoid four times each. The heating parameters are
heating the steel). The distance is re- shown in table 4.1, Residual stresses were
measured and the change reflects the mag- measured across the vee heated zone for
nitude of residual stresses in that strip. each plate with the 102 mm (4 in) gauge
Practical considerations limit strips to ap- length centered on the vee for each strip. A
proximately % in width and changes in distinction can be made by classifying
length are quite small over the gauge lengths “small vee angles” as those being less than
required; typically 4 in. These considera- or equal to 60” and “large vee angles” as
tions limit the accuracy of the process. those greater than 60’. These two categories
However, the results provide a reasonable have significantly different magnitudes of
assessment of residual stress patterns after residual stresses, especially at the edges.
heat straightening. For all residual stress The averages of all plates within each cate-
values given in this chapter, a positive sign gory are shown in fig. 4.4. The smaller vees
denotes tension and negative denotes com- exhibited considerably higher compressive
pression stresses. stressesat the edges.
An unheated plate (Plate UH) was The residual stress patterns in all of
tested for residual stresses to provide the ba- the plates were similar in shape to Roeder’s
sis for determining changes resulting fi-om theoretical distribution (fig. 4.2), where
vee heats. Stresses found in each strip are normalized values were used. An evaluation
plotted in fig. 4.3. The values are fairly low of the individual results indicates that most
and the shape compares reasonably well vee heat parameters had little effect on re-
with standard residual stress assumptions sidual stresses. The exception was that the
and previous experimental measurements largest vee angle cases (82”) had maximum
-
Table 4.1. Heating parameters for undamaged plates.
I I I I I
Plate Vee Angle Jacking Ratio Depth Ratio
P-l 20 0.00 1.00
P-2 45 0.00 1.00
P-3 60 0.00 1.oo
P-4 20 0.00 1.00
P-5 45 0.50 0.75
P-6 45 0.00 0.75 I
P-7 82 0.50 0.75
P-8 82 0.00 0.75
Table 4.2. Heating conditions and degree of damage for deformed plates.
Plate Angle of Damage Max Strain Vee Angle Jacking Ratio Vee Depth Avg. Plastic Rotation
(deg/millirad) (Multiple of Yield (degree) W&j Ratio per vee heat (millira-
Strain) dians)
-
Reddual stress (MPa)
Reddual Stress (MPa)
-300 -200 -100 0 100 200 300
-300 -200 -100 0 100 200 300
8
8
7
6
6
2 , ! ! 1 I -...I54 I I I 1 1
I! / / I I ! j -50 40 -30 -20 -10 0 10 20 30 40 50
-50 -40 -30 -20 -10 0 10 20 30 40 50
Residual Stress (ksi)
Residual Stress (ksi)
Figure 4.6. Residual stress distribution for dam- Figure 4.7. Residual stress distribution for dam-
aged and vee heated plates in region B (assumed aged and vee heated plates in region A and C (as-
modulus = 200,000 MPa or 29,000 ksi). sumed modulus = 200,000 MPa or 29,000 ksi).
69 MPa (10 ksi) and a maximum compres- locationsareshown in the figures alongwith
sive stressof 138MPa (20 ksi), the residual the locationof the vee and strip heats.
stressescan be approximatedby In the two originally undamagedan-
gles (figs. 4.8 and 4.9), the residual stress
patterns were quite similar. Somewhat
0, =10(1-12x2) (Eq. 4.1) higher compressivestresseswere found at
the edgesin Fig 4.9. The only difference
where x is the distancetiom the centerof the between these two specimenswas the vee
plate to the residual stress location divided angleused (20” and 45”, respectively). For
by the plate width and o, is the residual thesetwo casesthe apex of the vee was lo-
stressin ksi. cated at the toe of one leg and a strip heat
was usedon the oppositeleg.
Residual Stressesin Rolled Shapes
The residual stressesfor a 4x4~114
Residual stress patterns have been angle that was damaged and then heat
experimentally determined for some repre- straightenedis shownin fig. 4.10. An inter-
sentative samples of angles, channels and esting fact is that the damagedangle speci-
wide flange sections. The geometry of the men exhibited the same pattern of residual
shapes prevented measurementswith the stressesas the undamagedangles although
extensometeron both sidesof certain strips. the damaged angle had somewhat higher
However, the continuity and consistencyof values. For this case, the apex of the vee
the values indicate that by just measuring was locatedat the heel of one leg anda strip
one side, sufficient accuracywas obtained. heatwas not requiredon the oppositeleg. It
The residual stress values for angles are is apparentthat the heating/coolingprocess
shown in fig. 4.8 - 4.11. The strip number in the angles results in quite high (around
280 MPa or 40 ksi) compressivestresses
-_
Stress (ksi)
Stress (ksi)
L -40 -20 0 20 40 Heated Leg
-
. . ,
Heated Leg
-_
67654321 .- -
67654321 Strip Number
Strip Number
Figure 4.8. Stresses in angle VI-l (20” vee, apex at Figure 4.9. Stresses in angle VI-4 (45’ vee, apex at
toe, M&M, = 0.00, depth ratio = 1.00). toe, Mj/M, = 0.00, depth ratio = 1.00). -
Stress (ksi)
L -40 -20 0 20 40
8 Stress (ksi)
1
1 . . . 8
1 .‘. 8
I2146676 .: ::
Strip Number 1 *J.o*7*
Strip Number
Figure 4.10. Stresses in angle L4x4 (45’ vee, apex Figure 4.11. Stresses in angle L6x4x5/16 (45” vee, -
at heel, M,/M, = 0.50, depth ratio = 1.00). apex at heel, Mj/M, = 0.33, depth ratio = 1.00).
-_
.
I... 4 1 . ..8
Vee Heated We
Figure 4.13. Residual stress strip locations (Cate-
Strip Heats
Flange
3
-40
l...-20
020
40 4
Stress (kti)
gory S heat).
12SL
Strip
Number
1 . . . a
Figure 4.12. Stresses in channel IX-6 (45’ we, Figure 4.14. Residual stress strip locations (Cate-
M#$Q = 0.50, depth ratio = 1.00). gory W heat).
near the toes, regardlessof the location of straightenedangles are quite high and ap-
the vee apex, relative to the stiffening ele- proach yield stress at some points. While
ment. For each of these casesthe residual the distribution of these stressesmay vary,
stresseswere large compressivevalues at the magnitudesare similar to that of welded
edges and comers and somewhat smaller built-up shapes.
tensile forces over the central portion of The residualstressesfor a categoryS
eachcrosssectionelement. heating pattern on an originally undamaged
The residual stress pattern for an C 6x8.2 channelare shownin fig. 4.12. The
unequalleg angleis shown in fig. 4.11. The pattern is not as well defined as for angles.
angle was damagedand straightenedwith However, significant residual stresseswere
m
vee heatson the long leg. Sincethe apexof found with the magnitudesapproachingthe
the vee was at the heel,no strip heatwas re- yield stress.
quired on the stiffening leg. Here the pattern Residual stresseswere also experi-
varied from the equal leg angles,although mentally determinedin the heatedregion of
the maximum values were of similar mag- W 6x9 wide flange beams using the sec-
nitude (approximately equal to the yield tioning method. In all of the beams, eight
12345676 12345671
Strip Number Strip Number
Vee
L’ ::.i
*(D * _..j...g...i...:
If i :::;::j::“-.:::
“g 5 . ..i...i :._.i...
a :.
la
o-20 40
12545678
12345678
Strip Number Strip Number
Figure 4.16. Stresses in beam B-l (20” vt?e, Mj/M, Figure 4.17. Stresses in beam B-2 (30” vee, M,/M,
= 0.00, depth ratio = 1.00). = 0.00, depth ratio = 1.00).
--
-
Strip Number Strip Number
0 20 46
Stress (ksi) .-
Figure 4.18. Stresses in beam B-3 (45O vee, Mj/M, Figure 4.19. Stresses in beam B-4 (45’ vee, Mj/M,
= 0.00, depth ratio = 1.00). = 0.25, depth ratio = 1.00). -
-40~: ::::’
12345676
Vee Heated- 1 . . ’ 8
-20 0 (ksi)46
20 0 20 40
strip ;po”,‘k.$/ 1 . . 8 Stress Vee Heated/ 1 . - . 8 Stress (ksi)
Flange
12345678
Strip Number Strip Number
Figure 4.20. Stresses in beam B-5 (45” vee, Mj/M, Figure 4.21. Stresses in beam B-6 (20” vee, Mj/MP
= 0.50, depth ratio = 1.00). = 0.00, depth ratio = 1.00). _
Strip
Figure 4.22. Stresses in beam B-7 (20” vee, MI/M, Figure 4.23. Stresses in beam B-8 (45’ vee, Mj/M,
= 0.50, depth ratio = 1 .OO). = 0.00, depth ratio = 1.00).
A unique part of thesetests was that has some effect on the residual stressdistri-
all beams except the first were re-damaged bution. Values are shown, using an assumed
andrepairedseveraltimes. Residualstresses modulus of elasticity of 200,000 MPa
were obtained after the last repair cycle for (29,000ksi).
eachbeam. In eachcasethe degreeof dam-
It is interestingto note that the resid-
age was approximately 7O which required
ual stresspatternsin all of thesebeamswere
about 20 heats to complete the repair. Re-
exactly oppositein nature to that of the un-
sidual stress measurementswere made on
damagedbeams which had tension in the
beamsafter 1,2, 4, and 8 damage/repaircy-
flangesand compressionin the web. For the
cles. Measurementswere taken at the center
damage/repaircases, the large number of
of damage. Shown in fig. 4.24 are the aver-
vee heatstendedto shortenthe flangesmore
age residual stressesin the flanges of the
than the strip heatsshortenedthe web. Thus
specimensfor the different categoriesand
the flanges had tension stresseswhile the
locations (the shorteningof the beamspre-
web had compression stresses. The web
vented the measurementof residual stresses
compressionwas obvious by severe web
in the webs, except for the single dam-
buckling which occurred after a number of
age/repaircycle). The stresseswere fairly damage/repaircycles (A correction factor
consistent in the beams with one and two was applied to account for curvature when
damage/repaircyclesand fairly consistentin computingthe residual stressof eachstrip).
those with four and eight damage/repaircy-
Actual repairswould haverequiredthis local
cles. This behavior indicatesthat the num- buckling to be heat straightened.
ber of damagecycles
12345678
Strip Number
Figure 4.24. Residual stress distribution damaged, Category W side flange beams (assumed E = 200,000 MPa
or 29,000 ksi). -
Residual stresses were measured for cept, these members are all treated the same _
a single W 6x9 beam with Category S dam- and no capacity reduction would be as-
age which was repaired using the standard sumed. However, if multiple column curves
pattern. The residual stress patterns are are used (typical of many European coun- -
tries), then heat straightened columns would
shown in fig. 4.25. Both flanges were in
fall in a lower strength curve after heating
compression while the web was in tension.
The maximum compressive stresses in the due to residual stresses. Consequently, there -
flanges approached yield while those in the would be some loss of design strength.
web were somewhat less. A comparison of Second, high tensile residual stresses
the residual stresses for the undamaged and reduce the effectiveness of jacking forces by ‘-
damaged beams showed a reasonably good effectively canceling out the compressive
correlation for Category S. stresses in areas with externally applied __
force which causes compressive stresses.
The large residual stresses created
Movement could be reduced or even re-
during heat straightening have several impli-
versed, if the jacking force moment does not _
cations. First, if the member is a compres-
compensate for the residual stresses.
sion element, the high residual stresses are
similar to welded built-up members. Since
-_
U.S. codes use a single column curve con-
1 . . * 8 Stress (ksi)
Temperature, OF
12346676
Strip Number
Figure 4.25. Residual stresses in Category S dam- Figure 4.26. Variation of coefficient of thermal
aged wide flange beam (45” vee, Mm = 0.50, expansion versus temperature (Roeder, 1985).
depth ratio = 1.00).
Specimen/Strip Yield Stress Maximum Ten- Percent* Percent Red. Modulus of Elasticity
F
(Mpa) WI sile strength Elongation In Area
(MFa x 103) (ksi x 10’)
WW (W
P-9 UH 323 46.8 474 68.7 45 58 199 28.9
1 356 51.7 486 70.5 33 46 ---- ----
Strip No. Yield Stress* Maximum Stress* Percent Elongation** Percent Red. In Area**
land2 99.9 92.0 99.9 69.8
*Confidence level that heat straightening a deformed plate will cause an increase in the property shown over that of
the same specimen before the damage/straightening cycle
**Confidence level that heat straightening a deformed plate will cause an decrease in the property shown over that
of the same specimen before the damage/straightening cycle
Table 4.6. Comparison of material properties in heat straightened steel plates with unheated specimens and
ASTM standard values.
not been directly heated, it is suspectedto and tensile tests conductedon strips near
retain more strain hardeningeffectsthan if it the apex, center and open end of the vee.
were containedwithin the vee heatedarea The resulting properties are given in table -
(as other strip #Z’s are for full-depth vees). 4.7 where UH indicates an unheatedstrip
The minor restretching effect in the upper (seefig.4.14 for strip numbers). _-
portions of this plate (addressed by Roeder, Yield Stress.-A significant increase
1985)may have causedcyclic hardeningnot in yield stressand tensile strength occurred
experiencedif the materialwas heated. This near the apexof the vee and it was progres- --
specimenalone (among the plates) experi- sively larger in proportion to the number of
enced a significant increase in tensile damage/repaircycles. A plot of the varia-
strengthover the unheatedspecimenfor that tion is given in fig.4.28. The yield stresses -
plate (10 percent). It should also be noted at other locationsincreasedin the range of
that similarly elevated yield and tensile 9-21 percentand averageda 13 percent in- _
strengths were experienced(near the vee crease(similar to the damagedplate results).
apex) after the first damage cycle in the The dataconfirms that the apexof the vee is
study of repetitively damagedwide flange the most sensitivezone. Repetitive damage _
beams,describedin the next sectionwhere a and repair cyclesresult in large increasesin
depthratio of 0.75 was alsoused. yield stress,especiallyafter two or more cy-
Mechanical Properties of Heat-Straight- cles. Tensile strength followed a similar -
ened Wide Flange Beams pattern as shown in table 4.7 and fig. 4.29.
However, the tensile strength at the apex
Tensile tests were conducted on increasedat only half the rate of the yield -
strips taken from four W 6x9 beamsdam- stressas shownin table4.8. This narrowing
aged by bending about their minor axis of the normal gap between yield stressand
(Category W). The residual stressesfor tensile strength suggeststhat heat straight- -
thesebeamswere discussedearlier and the ening should be limited to no more than 2
resultsshown in fig. 4.24. In eachcase45”, damage/repaircycles.
3/4depthvees and a 50 percentjacking ratio -
were used. The standardCategory W pat- Modulus of Elasticity.-The modulus
tern of vee heatson both flangesand a strip of elasticity averaged8-23 percentlower for
heat on the web was employed. In addition memberswith one or two damage/repaircy- -
to evaluating material properties, the pur- cles. However an increasewas observedfor
pose of thesetests was to determinethe ef- the beam with four cycles. In general,the
fects of repetitive cycles of damageand re- level of variation was similar to that of dam- -
pair. Consequently,Beam B-l was dam- agedplates.
agedandrepairedonce,while BeamsB-2 to Ductility.-The elongation after one _
B-4 were damagedand repairedtwice, four or two damage/repaircycles(31-32 percent)
times and eight times, respectively. In each followed the trend of plates with about a
casethe degreeof damagewas in the range one-third reduction. However, for four or -
of 6-8” and required about 20 heat cyclesto eight cyclesthe elongationis proportionally
repair. After the last damage/repaircycle for reducedasshownin fig. 4.30 andtable 4.8.
eachbeam,one of the flangeswas sectioned --.
Specimen/Strip Yield Stress Maximum Stress Percent Percent Red. Modulus of Elasticity
WV (ksi) WW W) Elongation In Area (MPa x 10’) (ksi x 103)
zg 60 400 5u)
50 f4
s 300 g
w 40 w -
P
> 30 200 5s
-A- Middle of Vee Heat
20 l - -Bottom of Vee Heat
100
10
0 0
0 12 3 4 5 6 7 8 9
Number of Damage/Repair Cycles
-_
Figure 4.28. Yield stress versus number of damage/repair cycles.
110 -
100 700
90 600
g 80 i?
s 500 g
70
z2 60 400 z
G 2
50 3i
=Q)
t 40 300 g
g 30 200 2
E
20
100
IO
0 0
0 12 3 4 5 6 7 8 9
Number of Damage/Repair Cycles
-
Figure 4.29. Tensile stress versus number of damage/repair cycles.
--
50 , I I I I I I I I I
0 12 3 4 5 6 7 8 9
Number of Damage/Repair Cycles
Table 4.9. Material properties taken from an improperly heat straightened girder of an Iowa bridge re-
moved from service.
-
--
-
Note: Chemical analysis percent by weight: Carbon-0.289; Manganese-0.647; Silicon-0.032; Sulfur-O.03 1; Phos-
whorus-0.009 I
-
0.6
+ Experimental -
0.5 - - - - - -Theoretical Eq. 5.2
0.4 -
0.3
-
0.2
-
0.1
ducted on initially straight plates for each Plate Thickness and Geometry.-
caseand the results averaged. The results Researchershave generallyconsideredplate _
are shown in fig. 5.2 for a combination of thicknessto have a negligibleaffect on plas-
three depth ratios, three vee anglesand two tic rotation. The only reservationexpressed
jacking ratios. Thejacking ratios reflect that hasbeenthat the plate shouldbe thin enough _
a jacking force was usedto createa moment to allow a relatively uniform penetrationof
at the vee heat equalto either 25 percentor the heat throughthe thickness. The practical
50 percentof the ultimate bendingcapacity limiting value is on the order of 19-25 mm -
of the plate. As can be seenTom fig. 5.2, (3/4 to 1 in). Thicker plates can be heated
the depth ratios of 75 percent and 100 per- on both sides simultaneouslyto ensure a
cent track each other well. In fact the 75 uniform distributionthrough the thicknessor -
percentdepthratio resultedin slightly larger a rosebudtip can be used. The results from
plastic rotations in all but one of the six testsinvolving different plate thicknessesare
cases.The 50 percentdepthratio resultedin shown in fig. 5.4. Each bar representsthe -
an erratic behavior when comparedto the averageof at least three heats. No jacking
other two. In three of the six casesthe 50 forceswere usedin thesetests. --
percent depth ratio producedmuch smaller The results illustrate the level of
plastic rotations. In the other three cases, variability that may occur amonggroups of
the plasticrotationswere similar. heats. However, there is no discernable -
To fiu-therveri& this behavior, a se- pattern among the plate thicknessesfor the
ries of plateswas damagedand straightened. three different vee anglesused. The ran-
The degreeof damage was large enough domnessof theseresults indicatesthat plas- -
that at least20 heatswere required for most tic rotation is not a function of plate thick-
of theseplates. Therefore,more statistically ness. A similar plot is shown for data fi-om
significant average plastic rotations were Roeder’stests (Roeder, 1985)in fig. 5.5. In -
obtainedfi-om thesetests. Results are com- thesetests only one vee anglewas used(0 =
paredin fig. 5.3 for ajacking ratio of 0.5 and 60”) along with several different jacking -
two vee depth ratios: 0.75 and 1.0. Again forces for plate thicknessesof 6, 10, and 19
the patternof plastic rotations doesnot have mm (0.25, 0.375 and 0.75 in). Again, there
a direct correlationto the vee depthratios. is no discernablepattern among the plate -
Therefore, even though it would thicknesses.
seemintuitive that increasingthe vee depth Roederalso investigatedthe effect of
would increasethe plastic rotation, there is plate width on plastic rotations. Shown in
no experimentaljustification for sucha gen- fig. 5.6 is a comparison of the two plate
eral statement. It can be concludedthat the widths of 150and 200 mm (5.9 and 7.9 in).
variation of vee depth ratios between 0.75 Only one or two heats were used in each
and 1.0 has little influence on plastic rota- caseandseveralplate thicknesseswere used.
tion. However, a vee depth ratio of 50 per- Relatively little difference in plastic rota-
centmay reducethe plastic rotations. tions was observedfor thesetwo widths.
--
Depth Ratio 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
Figure 5.2. Influence of vee depth on plastic rotations of originally straight plates for various vee angles and
jacking ratios (heating temperature = 650°C or 1200°F9.
7
‘iii
.-5 6
74
.k 5
=.-
!E.4
s Y
‘i; 3
m
‘i
p! 2
0
*=
4 1
p 0 I 0 8
0 10 20 30 40 50 60 70
Figure 5.3. Vee angle versus average plastic rotation for damaged plates having different depth ratios (Jack-
ing ratio = 0.5 and Temperature = 65O’C or 1200°F).
0 l- -t-
20 45 ’ 60
Vee Angle (degrees)
--
Figure 5.4. Influence of plate thickness on plastic rotation (average of 3 heats with depth ratio = 1, jacking
ratio = 0, temperature = 650°C or 12007 -
Thickness (in) 0.25 0.38 0.75 0.25 0.38 0.75 0.25 0.38 0.75
(mm) 6.35 9.65 19.05 6.35 9.65 19.05 6.35 9.65 19.05
Figure 5.5. Influence of plate thickness on plastic rotations, Roeder, 1985 (single heats with 9 = 60°, depth
ratio = 0.67, heating temperature = 552-682°C or 1025-1260°F).
1
-
0 i
Plate Width (in) 7.g 5-g 7.9 5.9 7.9 5.9 7.9 5.9
(mm) 201150 201150 201150 201 150
Jacking Ratio 0% 0.09% 0.16% 0.23%
Figure 5.6. Comparison of plastic rotations for 150 mm (5.9 in) (average of 2 heats) and 200 mm (7.9 in) (sin-
gle heat) plate widths, Roeder, 1985 (9 = 60”, depth ratio = 0.67, and heating temperature was approximately
650°C or 12OOT).
,-
of plate geometry. Thus, plate geometry is
consideredto be a minor factor influencing
plasticrotation behavior.
2.5 Temperature.-One of the most im-
portant and yet difficult to control parame-
ters of heat straightening is the through-
2 thicknesstemperatureof the heatedmetal.
Factors affecting the temperature include:
1.5 size of torch orifice, intensity of the flame,
speedof torch movement, and thickness of
1 the plate. In his experimentsRoeder (1985)
made careful temperaturemeasurementsof
the heatsproducedby knowledgeableprac-
0.5 titioners. He found that these individuals,
when judging temperature by color, com-
0 monly misjudgedby 56°C (100°F) and, in
some cases, as much as 111°C (200°F).
Plate Width (in) 4 8 12
Thus, there are considerablevariations in
(mm) 102 203 305
temperaturecontrol, even with knowledge-
ableusers.
Figure 5.7 Comparison of average plastic rotation Assuming adequatecontrol is main-
(for three 20” vee heats) for plates of three widths tained over the applied temperature, the
(jacking ratio = 0, depth ratio = 0.75, and heating
questionarises as to what temperaturepro-
temperature = 650°C or 1200V).
duces the best results in heat-straightening
In another test series conductedby without altering the material properties.
Boudreaux(1987), three plate widths were Previous investigatorshave differed in an-
studiedas shownin fig. 5.7. The plastic ro- swering this question. For example, Sha-
tations are the averageof three heats. An nafelt andHorn (1984) statethat heatsabove
unusuallylow averagewas observedfor the 650°C (1200°F) on carbon and low alloy
102mm (4 in) width. However, little differ- steels will not increase plastic rotation.
encewas found betweenthe 8 in. (203 mm) Rothman andMonroe (1973) concludedthat
and 12 in. (302 mm) widths. The results of reheating areas where previous spot heats
these tests show no clear relationship be- were performedwill not produceany useful
tweenplastic rotation andplate width. movements. However, Roeder (1985) has
In summary, the parametersof plate shown that the resulting plastic rotation is
thickness and width show little definitive generally proportional to the heating tem-
influence on plastic rotations. The test re- peratureup to at least 870°C (1600°F). To
sults do illustrate the variability of response more clearly define the behavior suggested
typical of heat straightening. It is probable by a limited number of data points in Roe-
that the fluctuations shown here reflect this der’sstudy,a seriesof heatswere appliedto
variability characteristicrather than effects
--
-
0 IO 20 30 40 50 60 70 80 90
Figure 5.9. Influence of jacking ratio on average plastic rotation for % depth vee heats and 650°C (1200°F) --
heating temperatures (lines represent a least squares curve fit).
Figure 5.10. Influence of jacking ratio on average plastic rotation for full depth vee heats and 650°C (12OOOF)
heating temperature (lines represent a least squares curve fit).
, I I ,
1. Jackh# Ratlo-030 1 I 1 II
3i:
6
II I
5 /--
I
I
4 I I
-1
10 20 30 40 SO 60 70 60
0 5 10
Jacking Rotlo
15
(X)
20 25
Vee Angie (dogmas)
Figure 5.11. Average plastic rotation versus jack- Figure 5.12. Jacking ratio versus vee angle for
ing ratio for 60” vee heated plates from Roeder, straightening damaged plates (Temperature =
1985 (depth ratio = 2/3 and heating temperature = 1200°F or 650°C and depth ratio = 1)
650°C or 12OO’F).
reflected we
strain occurred in the vee zone, the strain specified maximum heating temperature,T,
4 was not constant in the longitudinal direc- after a heating/coolingcycle, then,
tion. Rather, he found that most of the per-
manent strain occurs over the middle two-
thirds of the vee. While it is recognizedthat sp’(T)= f m- 5.4)
the assumptionslisted above are approxi-
mate, the poorest is that of perfect single
axis confinement. This assumptioncan be or using trigonometric relations from fig.
improved using the experimentaldata as a 5.14 to eliminateV:
guide. fig. 5.14 illustratesthe geometryof a
vee, before and after heating, basedon the
first three assumptionslisted previously. (Eq. 5.5)
6 = 24%’ (T) tan;
The changein the width of the openend of
the vee,6, canbe written as:
Equating eqs. 5.3 and 5.5 and using the
trigonometric identity for the tangent fiurc-
tal++$)
1 (Eq. 5.3) tion:
where:
(Eq. 5.6) -
E,(T) = 5a(T)dT (Eq. 5.10)
-
Sincethe experimentaldata shows that both
s,(T)=-
f$m
‘pPand s;(T) are small, it is assumedthat
E(T)
tamp&2z (~42,E,’ (T) << 1 and <pPis in radi-
ans. Eq. 5.6 canthen be solvedfor ‘pP:
and F,,(T) is the yield stress at temperature -
T, E(T) is the modulus of elasticity at tem-
(pp= 2&,‘(T) sin B (Eq. 5.7) perature T, and a(T) is the coefficient of
2 thermal expansionat temperatureT. In or- -
der to obtain values for E, and E,, equations
However, Roeder (1985) carefully are neededfor F,, E, and a as a function of -
measuredstrain distributions over the vee temperature.
heat areasof plates. He found that most of Roeder (1985) used the equations
the permanent strains occurred within the shownin figs. 4.26 and 4.27 to approximate --
inner two-thirds of the vee. This observa- a (T), F, (T) andE (T) where T is in degrees
tion suggestsusing an effective vee angleof Fahrenheitand E is in ksi. Substituting a
two-thirds the actual angle. By incorporat- (T) form fig. 4.26 into eq. 5.10, assuming -
ing this assumption,the equationfor plastic that the ambient temperature is 21.1“C
rotation becomes: (70°F) andcarryingout the integrationgives .-
-
0 10 20 30 40 50 60 70 80 90
Vee Angle, 0 -
Figure 5.15. Plastic rotation versus vee angle for vee heated plates having a heating temperature of 650°C
(1200°F).
-
l A simple analytical formula for predict-
ing the plastic rotation in a vee heated
platewas developedas a function of: vee
angle, heating temperatureof the steel,
restrainingforce, and the plastic strain at
the elevatedtemperature. This plastic
strain can be related to yield stress,
modulus and coefficient of thermal ex-
pansionat the elevatedtemperature.
l The influence of yield stress on plastic
rotation is small for mild steelhaving an
F, between 230-345 MPa (33-50 ksi).
Thus, the analytical formula of eq. 5.20
is applicableover this range.
Primary strsss at a
plostlc hinge
plate
element
(a) Web element
Figure 6.1. Primary and Secondary plate elements Figure 6.2. Typical yield zone patterns in the plate
for a channel bent about its major axis (Category elements of the channel shown in fig. 6.1. -
S damage).
due to increasedinternal restraints. A good
flanges yield due to axial stress(eitherten- practice to minimize the heating time is by ^_
sion or compression)as illustrated in fig. using more than one torch for complexpat-
6.2b. In the third case,the secondaryele- terns.
ment is yielded in weak axis bending in
which a region of yield is forrned as shown In additionto the jacking load factor, -
in fig. 6.3. The stressdistribution is similar the various combinationsof plate elements
to that shownin fig. 6.2a. However, the re- found in structural steel shapesintroduces
sults are a narrow strip of flexural yielding two other parameters that may affect the -
often referredto as a yield line. member’s behavior during heat straighten-
ing. The first is a shapefactor and the sec-
As a consequenceof the various ond is a stressfactor. It is obvious that the -
patternsof yielding which occur in damaged shape may influence behavior. However,
rolled or built-up shapes,the heatingpattern the stressfactor requiresan explanation. -
for repair must be tailored to fit. While the
vee heat is generallyused on primary plate For casesin which jacking forces are
elementsof a sectionbent abouttheir major applied prior to heat straightening,the dis-
axes,the secondaryelementsmay require a tribution of stress over the heated section -
strip heat, line heat or no heat at all. These will vary accordingto the shapeof the cross
heatingpatternsintroduce an additional de- section and the magnitude of the moment.
gree of variability in that the time to com- As the torch moves over the section, the --
plete a heatmay be considerablylonger than steel temperaturerises and then falls in a
heatinga singleplate. Considerablecooling mannersomewhatanalogousmoving across
may have occurredat the initial heating lo- calm water. The heat variation produces -
cations before the last element is heated. continuous and complex changes in the
Movement may be retarded in such cases stressdistribution. As a consequence, stress _
0%. 6.1)
Edge view
10 20 30 40 50
Vee Angle (degrees)
Figure 6.5. Experimental theoretical plastic rotations for a C 6x8 channel with Category S damage.
F, = 1- 2Cl- 2 ’
(&‘I Mj
M (Eq. 6.5)
P
Figure 6.7. Heating patterns for channels bent about the weak axis (Direction of moment producing damage
indicated by MJ.
25
v-4
I L
A Jadting Ratio=&0
l Jacking Rattio=O.25
.r*
- n Jacking f?af0=0.50 Jacking Rati sO.50- /’
- - - Theoretical Eq. 6.1 /
W’
-Least squares curve ffi /
/ / .
/’
/’ 4.H-
0 10 20 30 40 50
Figure 6.8. Experimental and theoretical plastic rotations for Category W damage of a C 6x8.2 with the open
end of vees at flange-web-juncture as shown in fig. 6.7a.
I27
Federal Highway Administration
-
Y
Avg.b-i
I-
dm
----c--
x-- --x
v9.6
II
(0) Chonnel
equivalent
cross section
plate cross
and
section
where d is the vee depth and b, is the width
of the channelweb or stiffening element.
(b)
I+i
Heating
symmetrical
pattern
about
and geometry
the x-x axis
is
The equation can be modified to include
both positive and negativemoment damage,
if the term d in the numerator of the second
term is replacedby d, which is definedas the
distance from the apex of the vee to the
stiffening element (the web in this case),
thus:
1 bd
F, =1+-(-=-) (Eq. 6.12)
2 d*
6.11. Vee angle versus plastic rotation for wide flange beams using the Category W heating pattern, Horton
(1973).
10 20 30 40 50
Figure 6.12. Plastic rotation versus we angle for W 6x9 using the Category W heating pattern (Temperature
= 650°C or l~OO°F).
I
l-B-1 ‘- -
Figure 6.13. Spreading of yield zone in subsequent cycles of damage and repair for W 6x9 wide flange beam
specimens. -
Table 6.1. Summary of damaged beam data for W 6x9 beams with Category W damage.
.-
Beam Length of yield zone (in) after Bend #: (Top row = A, bottom row = B)
1 2 3 4 5 6 7 8
1 -I
---
2 --- 8.20
--- 8.40
3 5.90 8.05 9.75 10.30
6.00 8.65 10.40 11.65
c11
around its weak axis to a degreeof about Sincethere were over 200 heatsper-
sevendegrees. All beamswere repairedus- formed, statistical analysis can be used to
ing the heating pattern of fig. 6.10 and 45” evaluatetrends in the data. DeBajar, etal.
veeswith a depthratio of 0.75 and a jacking (1992) addressedtwo questions: (1) Does
ratio of 0.5. The number of damage/repair the responseto heat straighteningvary if the
cycles varied for each of the beams. Each member is re-damagedand repaired more
repair cycle consistedof approximately 20 than once; and (2) Is the responseduring a
heats, with the average plastic rotation single repair cycle dependenton the heat
shownfor eachrepair cycle in table 6.1. number? An independent sample t-test
As the beams straightened,a thick- (Hicks, 1982) was used to evaluatethe re-
ening developedin the middle region, caus- sponsefrom different damage/repaircycles.
ing a spreadingof the yield zone in each The probability that the averageplastic ro-
subsequentbend as seenin fig. 6.13 (values tation in successiverepair cycles was less
are shown in table 6.2). The thickening re- than the averagevalue of the first repair
sulted in a smoother distribution of curva- rangedfi-om 71-87percent. The results sug-
ture (due to thinner portions further fi-om the gest that there may be a small trend toward
centerline tending to yield earlier than be- larger plastic rotations in the first repair cy-
fore), although the total angle of damage cle.
was kept as consistentas possible for each To look at whether the amount of
bend. Due to the larger yield zone, the vee movementvaried from heat to heat within a
heats were distributed over a longer length single repair cycle, a dependentsamplest-
to accommodatethe spread. test (Hicks, 1982)was used. Theprobability
values were high enough to conclude that improper heating procedure. The values
the first heat does produce greater move- for the shapesheatedwith a 50” vee angle _
ment than successiveheats, although there also appearlow, again suggestingthat the
was no trend that anotherheat had preemi- steel temperature was not adequate. The
nenceover any other. This conclusioncon- valuesfor the W 6x9 with 20” and 30° vee _
fums observationsin the field that the first angles appear too high suggesting either
heat producesmore movement. The prob- over-heatingor lack of lateralrestraint. The
able causeis that residual stressesare often only values which appear consistent with -
created during the damageprocess. These other categoriesof damageare those with a
stressesoften act in a direction tending to jacking ratio of 50 percent. To evaluate
reduce the damage thus magnifying the these inconsistencies,a W 12x14 was vee -
jacking force. The initial heat tendsto relax heated. The temperatureand jacking force
these residuals so that successiveheats are were carefully controlled. The results are
-
not affected. Becauseof this behavior it is plotted in fig. 6.15.
recommendedthat the jacking force during A model to predict this behaviorcan
the first two heats be limited to approxi- be developedsimilarly to the channel. The -.
mately 25 percentof the membercapacityso flange of the beam is the stiffening element
as not to risk over-stressingthe member. for the category S damageheatingpattern.
The averagevalue of plastic rotation The flange at the open end of the vee pro- -
for all damage/repaircycles (240 heat cy- vides a magnificationeffect similar to that of
cles) was 6.78 milliradians. The average the channel. Using the folded/unfolded
value, considering the first damage/repair flange concept (fig. 6.16) similar to that of .--
cyclesonly (75 heats) was 5 percenthigher the channel along with the same assump-
at 6.64 milliradians. These values were tions, a shapefactor can be derivedidentical
within 20 percent of the value predictedby to the channelof eq. 6.4 or 6.12 where for
the theoreticalmodel of eq. 6.1. the wide flangebeam ds= beam depth,b, =
Behavior of Wide Flange Beams with flangewidth and d = beamdepth. --
Strong axis (Category S) Damage The equationfor plastic rotation, Q,,
The heating pattern and yield zone is given by eqs.6.1 and 6.2, F e is definedby
for Category S damageis shownin fig. 6.14. eq. 6.7, F, is given by eq. 6.12, andF, by eq. -
Severalresearchers(Avent, 1992, Moberg, 6.5. Note that ZJS, is typically around 1.12
1979and Boudreaux,1987)have conducted for wide flange beams. Using this value in
tests on Category S damagedand undam- eq. 6.5, the stress factor can be approxi-
agedwide flange beams. The averageplas- matedby
tic rotation for eachcaseis shown in table
6.3 and valuesfor the W 6x9 are plotted in
fig. 6.15. F, =1-0.5- M.i Eq. 6.13
-
The resultsshown in table 6.3 follow MP
the trend that plastic rotation is proportional
to vee angle but are inconsistent. The W
10x33valuesare far too low and suggestan
-
Figure 6.14. Yield zone and heating pattern for Category S damage to wide flange beams.
Table 6.3. Plastic rotations for Category S damage to wide flange beams (heating temperature = 650” or
1200°F9.
0
0 10 20 30 40 50
Vee Angle (degrees)
Figure 6.15. Influence of vee angle and jacking ratio on plastic rotation for W 6x9 (Category S damage pat-
tern).
.-
axis. In addition, the jacking force applied
during heat straightening often causes flex-
ure about a non-principal axis. In one sense -
the angle is the most simple of rolled shapes
which can be considered as a plate with a
single fold. The heating pattern consists of a -
vee on one plate element and a strip on the
other element as shown in fig. 6.17a or sim-
Figure 6.16. Geometric relationship between ply a vee heat if the apex of the vee is lo- -
Category S wide flanges and plates. cated at the stiffening element (fig. 6.17b).
The results of tests on originally straight an-
The theoretical equation for the W 12x14 is
gles using the heating pattern of fig. 6.17a -
plotted in fig. 6.15. The theory agrees rea-
are shown in fig. 6.18. Other parameters
sonably well with the data.
included jacking ratios, Mj/Mp, of 0, 25 and __
Angles 50 percent; vee angles of 20” and 45’; a
Angles differ from other rolled shapes in depth ratio = 1.0; and a heating temperature
that the plate elements of the cross section 650°C (1200°F). In general the plastic rota- __
are not orthogonal to the principal axes. As tions were significantly larger than for
a result, the damage to angles is often due to equivalent plates. The reasons for this
flexure about an axis other than a principal -
-
moment Compression
causing
damage damage loading
Compression
tons from J
damage loading
Figure 6.17. Yield zone and heating patterns for flexural damage o f a typical angle.
~1
-Experimental least squares curve fti
10 20 30 40 50
Vee Angle (degrees)
Figure 6.18. Influence of vee angle and jacking ratio on plastic rotation for L 4x4~114 angles with the heating
pattern of fig. 6.17a.
Table 6.5. Comparison of out-of-plane plastic rotations to plastic rotations in the in-plane direction of move-
ment for initially straight L 4x4x% angles (in-plane movement shown in fig.6.18).
comparedto the plastic rotations in the de- C 12x20.7 (strong axis bending), (c) L
sired direction, the zero jacking ratio cases 5x3~518(bendingabout the axis of the short _
exhibitedmore relative movement. In fact, leg with tensiondamageto short leg.
the out-of-plane movements were greater Solution.-The first step is to com-
than the in-plane ones. The large out-of- pute the expectedmovementper heat cycle -
planemovementsareprobablya result of the using eq. 6.1. For all cases(Piis computed
vee heatedleg of the angle already being from eq. 6.2 as:
.-.
heated and offering less resistanceto rota-
tion. The lower ratios of out-of-planeto in-
plane movement encounteredat larger jack- q)b= O.O147sin(y) = 0.00255rud = 2.55miZZirad _
ing ratios are probably due to an ineffective
distribution of stressesin the outstanding
leg. andFe is obtainedfrom eq. 6.7 as -
Depending on whether or not the
out-of-planemovementsarebeneficialto the
overall repair of the given specimen,either F, = 0.6 + 2(.5) = 1.6 .-
bracing againstthis movement or using al-
ternative heating patternsmay be necessary
to prevent or reduce them. The vee heat The other factors are dependenton the cross
would be applied as normal, and the stiff- sectionandarecalculatedasfollows:
ening elementwould still needto be heated (a) W 36x170(CategoryW damage)
to allow rotation is the desired direction.
d, = 12.03 in/2 = 6.02 in
However, instead of a rectangular heating
pattern, it is likely a reversevee heat (con- b, =36 in
tinuing fkom the open end of the original d = 12.03 ipt
vee, and tapering down to a point) would
encouragethe desired rotation, while re- Z/S = 83.8 in3 153.2 in3 = 1.58
ducing the out-of-plane movement. In its From eq. 6.5
“unfolded” position, the heating pattern
would resemblea diamond shape,a pattern F = 1+ = 1 75
usedby pipe welders to straightenpipe dis- s 1[(36)(6.02)]
2 12.033 -
tortions. Heating in the proper fashion is From eq. 6.12
essentialfor obtainingmovement. -
The heating pattern for rolled shapes stressfactor, F,, and can be generalized
typically consists of vee heats on the for rolled shapesas
primary elementsand strip heatson the
secondaryelements.
If a secondaryelementlies at the apexof where the basic plastic rotation factor,
the veeheat on a primary element,a strip R, is definedas
heatis not applied.
cpb=0.0147 sine -
The amount of movement when heat 3
straighteningrolled shapesis a function andthe modification factors are:
of: the vee angle, heating temperature,
magnitude of the externally applied F, =0.6+2- Mj
Perhaps the most typical type of of internal and external constraints? What
damagefound on steel bridge membersre- are the important parametersfor straighten-
sults Corn impact of vehicles or freight on ing compositemembers? What is the basic
the beams or girders of composite deck- methodology? Why do the memberssome-
girder bridges. Heat straighteningis an at- times crack during repair? How much
tractive repair alternativebecauseof its low jacking restraint canbe safely applied? And
cost and minimal disruption of traffic. can movements be accurately predicted
However, little information has been avail- analytically?
ableto guide either the Engineeror the heat- One fundamentalparameterthat has
straighteningtechnicianon safe and accept- been overlookedin previous studies is the
able implementationprocedures. The pur- internal redundancyof the structure. Often,
pose of this chapter is to provide such a the damagedsteel member in the field dis-
guide. plays an inherent redundancy due to its
The available literature shows little structural configuration, imposing an inter-
quantitative research related to heat- nal constraint on the potential heat-
straightening repair of composite deck- straighteningmechanism for the member.
girder bridges. Moberg (1979) describeda An understandingof this behaviorcombined
limited field investigation of the heat- with knowledgeof the role of jacking forces
straightening behavior of damagedbridge in heat-straighteningis needed to answer
members. Shanafelt and Horn (1984) ad- thesequestions.
dressedthe general damageassessmentof First, a detailedand controlledset of
structures and suggestedan approach for field experimentsis summarizedwhich were
using heat straightening as one of several designedto provide a more comprehensive
repair alternatives. A comprehensivesum- data baseon the field behavior of damaged
mary of the state-of-the-artof heat straight- composite bridge members. Second, the
eninggivenby Avent (1989) emphasizedthe heat-straig,hteningbehavior of those mem-
lack of quantified engineeringdata on heat- bers is quantified both analytically and ex-
straightening repair procedures and ad- perimentally. Emphasisis placedon study-
dressedmyths about heat-straighteningre- ing the interaction of the internal redun-
pair. To provide quantified heat- dancy with both the external restraining
straighteningdata, Avent and Fadous(1989) forces and heatpatterns. A number of heat-
initiated a study of the behaviorof compos- straighteningtests have been conductedus-
ite girders followed with additionalresearch ing two 6 m (20-R) long, A-36 steelbeams
conductedby Avent, et. al., (1993). of different geometries(WlO x 39 and W24
Severalimportant questionsarise in x 76). Each beam was damagedand then
conjunction with heat straightening com- repaired using the heat-straightening
posite deck-girderbridges: What is the role method. The level of jacking forces was
varied in the courseof repair to evaluatethe 1972; and Roeder 1986). However, three
effect of external restraining forces. Based additionalparametershave also beenshown I -
Figure 7.1. Typical deformed shape and yield zones in damaged composite girders.
(about the strong axis of the flange plate) Five primary heating paterns/jack-
over a short length and there was a local ing combinations were used in the repair,
bulge in this region. A yield line also with a sufficient number of repetitions to --
formed in the web approximately57 mm (2- establishthe pattern of behavior. The basic
1/4 in) below the top flange, extendingover heatingpattern is shown in fig. 7.3. In two I
the middle two-thirds of its length. A view cases,a singlevee heat on the bottom flange
of the damagedbeamin the HEAT facility is was used and the restraining force, applied
shownin fig. 7.2. laterally at the center of the bottom flange, -
producedan apparentjacking ratio w/MJ
Typically, a lateral jacking force is of either 56 percent or 112 percent. The
applied to the lower flange during heat- other two caseswere identical, exceptthat a -
straighteningrepair. However, the determi- line heat was also appliedto the centertwo-
nation of the jacking ratio is complicatedfor thirds of the web along the length of the
compositegirders due to the internal redun- girder at the damage-inducedyield line. In -
dancy of the system. First, when a lateral all cases,the vee heatswere identical: a vee
jacking force is appliedto the lower flange, depthof three-quartersof the flangewidth, a
only a portion of that force producesa mo- 30” vee angle, and a constant heating tern- -
ment in the flange. Part of the force follows peraturein the range of 650°C (1,200”F) (as
a load path through the web into the upper measuredwith contactpyrometer and multi- _
compositeflange and is resistedby the con- ple temperaturecrayons). The vee heat was
crete deck. The determinationof the actual always appliedwithin the yield zone of the
moment in the lower damagedflange is re- flange at the point of maximum curvature. _
quired to prevent over-stressduring jacking A single-orificeoxyacetylenetorch was used
andto predict the expectedmovement. Sec- for the heating. Graduatestudentstrainedin
ond, the moment capacity due to a laterally heat straighteningperformedthe heats. -
applied load is also influenced by the load
path transfer making it difficult to compute After the completion of the repair
the plasticmomentcapacity,q. process,the same W 10x39 girder was re-
damagedby exerting a concentratedstatic -
The procedureadaptedhere is to de- force with a midspanjacking device. That
fine the apparentmoment due to jacking, M+ force causeda permanentdeflection in the _
as the moment carried by the lower flange bottom flange of 57 mm (2.23 in) in magni-
assumingall the jacking force was trans- tude. A load-deflectioncurve was obtained
ferred to the lower flange alone. Likewise, during the process. Consideringthe elastic _
the lateral capacity of the beam, &, is de- rebound,the initial yield load could be esti-
fined as the plastic moment of the lower mated as 71kN (16,000 lb). Basedon this
flange aloneabout its strong axis. The ratio yield value, the externaljacking force was .-
of Mj to q will be referred to as the appar- limited to 35 percent of initial yielding.
ent jacking ratio. The actualmoment in the Since a 650°C (1,200”F) temperature can
lower flangedue to the jacking force will be reducethe yield stressto approximatelyone- -
referred to as M, and will be expressedas a half to one-third its original value, this limit
percentageOf Mj. was intendedto ensurethat hot mechanical
.-
Table 7.1. Summary of plastic rotations for a damaged composite WlO x 39 beam after heat
straightening with various heating patterns.
‘Only a partial line heat was used omitting the center one-third of the web yield line sincethis portion was essentially straight
3The first heating cycle had only a single vee heat
Vhe movement from the first heat of each sequenceafter damage inducement was not included in the average
I I
4
3
s 3.5
a n Heating Sequence 2 -
EE 3 A Heating Sequence 3
.-
E x Heating Sequence 4
; 2.5 -
‘ZI
2Q 2
-
.o 1.5
IiJ
m
a1
%
=?
F 0.5
4
0
0.2 0.4 0.6 0.8 1 1.2
Jacking Ratio (M#l,)
-
Figure 7.4. Comparison of average plastic rotation for various patterns and jacking ratios.
--
subsequentcycles were applied using two for resistance to gravity loads, and dia-
vee heats simultaneously. The decision to phragms at the ends for resistanceto lateral
apply two vee heatsin the sameheatingcy- loads. The ram was a 15kN (3,400 lb) steel
cle was madebecausethe middle portion of section which was suspendedfrom a 12 m
the bottom flange contained two sharp re- (40 ft) craneboom. Damagewas inducedby
gions of curvature. In addition to the vee swinging the ram in pendulumfashion such
heats, each heating cycle involved a line that the bearing plate impacted the lower
heat alongthe yield line in the web. A total flangeof the beam. The type of damagewas
of 13 cycleswas requiredprior to removing similar to that of the previous beam. The
the local flange buckles. At the end or the maximum deformation, which occurred at
repair process,the “dish-like” distortion in point of impact, was 110mm (4.33 in).
the bottom flange was also removed using The repair processwas similar to the
line andveeheats. most efficient of the previous cases. The
The valuesof the averageplastic ro- horizontal component of the jacking force
tation per vee heat for sequenceNo. 6 is applied as an external constraint produced
shown in table 7.1. Fig. 7.6 shows a plot of an apparentMj/Mp = 0.33. For eachheating
the progressiveeffect of heats on the be- cycle, a single 650°C (1,2OO”F), three-
havior of the girder. After the applicationof quarter-depth,30” vee heat was used to re-
the thirteenth heating cycle, the lateral dis- pair the damage. During eachheatingcycle,
placement of the girder was eliminated. a line heat was used along the yield line.
However, the “dish-like” distortion was still The location of the flange vee heat was var-
present in the bottom flange. A final se- ied fi-om cycle to cycle over the 610 mm (2
quenceof line and spot heatswas appliedto ft) yield zone. The straighteningprocess
remove the “dish-like” distortion which ex- (sequenceNo.7) required 28 heating cycles
isted in the bottom flange. This processalso for completion. The averagevalue of the
produceda reverse lateral deflection of 4.6 plastic rotation achieved during this se-
mm (0.18 in). One additional heat cycle of quenceshown in table 7.2. Again, signifi-
sequence6 was appliedto remove the slight cantly larger plastic rotation angles were
out-of-plane curvature which developedin obtainedduring the first cycle.
the bottom flange during removal of the The samegirder was damageda sec-
bulge. This last heatingcycle removedmost ond time and then repaired(sequenceNo.8).
of the lateral deflection resulting from the In this case a gradually applied static load
removalof the bulges. was used to inducethe damageand a lateral
Heat-Straightening repair of W 24x76 load-deflection curve was obtained. The
Composite Beams resultingdamagewas similar to the previous
casewith a bottom flange deflection of 53
Rather than use gradually applied
static loads to induce damage, a swinging mm (2.1 in). The girder was repairedin an
identical manner, except that two vee heats
ram was usedin the first caseof damageto a
were applied simultaneously during each
W 24x76 beam. The girder was supported
similarly to the W 10x39: composite con- heatingcycle.
nection to the slab, simple supportedends
0.5
0 -
0 2 4 6 8 IO 12 14 16 18 20
Length (ft)
-
Figure 7.6. Heat straightening progression for a damaged W 10x39 beam using heating sequence No. 6 (25.4
mm = 1 in).
Table 7.2 Summary of plastic rotations for a damaged W24 x 76 beams after heat straightening with various
patterns.
Damage Cy- Heating se- No. of No. of simulta- No. of web line Apparent Average plastic
cle quence Cycles neous vee heats heats Jacking rotation per vee -
ratio heat (milliradian)
1 7 28 1 1 0.33 2.22*
2 8 12 2 1 0.33 2.48* -
3 9 11 1 1 0 2.99*
3 10 9 1 1 0.5 3.89
3 11’ 10 1 1 0.5 3.75 -
4 12 8 1 1 0.33 4.33*
1 13 7 1 1 0.75 9.99*
‘A half depth web strip heat was also used in this sequence
-
*The movement from the first heat of each sequence was not included in the average
-
Figure 7.8. Crack in bottom flange of beam after heating sequence 13.
repair plate elements with plastic bending depth strip, the web can deform and relieve
about the major axis, while line heats are some of these stresses. The applicationof
appliedto repair plate elementswith flexural strip heats on the web in SequenceNo. 11
damageabout the minor axis. Hence, a vee did not influence the averageplastic rota-
heat on the bottom flange in conjunction tions appreciablybut, it did tend to reduce
with a line heat on the web, appliedto their the buckling of the web near the center of
respective plastically yielded portions, are damage.
the proper heat patterns to repair composite Residual Moments.-A characteristic
beams impacted by high loads. Care must of eachdamagedgirder was the presenceof
be taken to continually adjustthe spanof the residual moments. When damage is in-
line heats, so that only thoseportions of the duced,the web acts as a spring resistingthe
web are heated that show plastic curvature movement. While a yield line typically oc-
after the last heating cycle. Similarly, the curs near the top of the web, there is also an
vee heats are confined to the portion of the elastic component of stored energy. This
bottom flangewith plastic deformations. characteristicis often referred to as internal
An important modification intro- redundancy. The initial plastic rotation re-
duced in this study was the inclusion of a lieves the majority of this stored force. So,
half-depth web strip heat during one se- successiveheats are not influenced signifi-
quence. The purposeof this heat was to re- cantly. During the first heat cycle, this re-
ducethe differential shorteningbetweenweb storing force acts as an additional jacking
and flange. By heatingthe web with a half- force tending to straightenthe girder. Un-
lessthe externaljacking ratio is reduced,the
plastic rotation during the first heat cycle is Fig. 7.9 shows the effect of the ap-
magnified. For example,the plasticrotation plied apparentjacking ratios on the average
after the first heat cycle for all sevengirders plastic rotations for various heat sequences --
testedaveragedover two and one-halftimes conducted. The average values for each
greater than the averageof the succeeding jacking ratio were connectedwith a straight _
values with the samejacking ratio. If the line. As shown in fig. 7.9, the jacking ratio-
girder is externally indeterminate,residual versus-plasticrotation curvesexhibit a sharp
momentsare also createdduring the damage discontinuityat higher loads for the W 24 x -
phase. For either casethis behavior should 76 beam. This behavior will be explained
be consideredwhen developinga constraint later.
plan. It is recommendedthat a reduced Internal redundancyaffects the heat- -
jacking force be used during the first two straighteningresponseof a compositegirder.
heats to minimize this effect and minimize Caused by the interaction of the bottom _
the possibility of cracking. In order to flange and the web, redundant forces are
eliminate this factor when computing the produced at the web-flange interface and
averageplastic rotation for a specific set of impede the plastic rotations by acting as a _
parameters,the first heat after damagein- negativeinternal constraintto the vee action
ducement was not consideredwhen com- in the bottom flange. Comparingthe plastic
puting averages. rotations of the WlOx39 and the W 24x76, -
Restraining Forces.- The simplest the trend is apparentthat plasticrotations are
way of providing restraining forces is to al- directly relatedto the jacking ratio, although
low the unheatedmetal within the member not necessarilyin linear proportion. How- -
to restrict thermal expansionby usinga suit- ever,there is somethingakin to a rigid body
able heat pattern (as in the case of a vee shift of the data with the W 24x76 showing
heatedplate). This is a form of an internal consistentlylarger valuesthan the W 10x39. -
constraint. Internal constraint may also be It can be concluded that the redundant
imposedby the self-weight, axial loading,or forces producedby the web-flange interac-
statical indeterminacyof the member. Fre- tion in composite members inhibit the -
quently, externalrestraining forces are used straighteningeffect more for a shallowbeam
to complementor even substitutefor the in- than for a deep beam. The damageto the _
ternal constraints required for the heat- compositemembers produce plastic curva-
straighteningphenomenon. The importance turesin the web abouttheir minor axis along
of restrainingforcesin the heat-straightening the yield line. This plastic deformation re- _
processhas been recognizedin the field for sistsany elasticbendingof the web aboutits
many years. Hence,jacking forces have of- minor axis during the straighteningprocess
ten been used to enhance the heat- of the bottom flange. Hence,the presenceof -
straighteningrepair of a wide rangeof dam- the yield line in the web tends to magnify
aged structural steel members. The tests the counterproductiveredundantforces and
herewere designedto evaluatethe effect of further inhibits the straighteningeffect of the -
external restraining forces on the heat- bottom flangevee.
straighteningbehaviorof damagedcompos- -
ite bridgegirders.
0
0 0.2 0.4 0.6 0.8 1 1.2 1.4
Apparent Jacking Ratio, RI,/&
Figure 7.9. Apparent jacking ratio versus plastic rotation for composite girders.
By using a line heat on the web along with girder would be expected to have greater
the vee heat in the bottom flange, thesein- stiffness and thus smaller lower flange mo-
hibiting forces are mitigated. This conclu- ments than that of the deeper girder sub-
sion was verified in an appraisalof the in- jected to the same lateral load. The data
fluence of the web line heat on the heat- tend to verify these observationssince the
straighteningresponseof the W 10x39com- jacking ratio hasless effect on the shallower
positegirder. Referring to table 7.1, the first girder. By adopting a jacking ratio defini-
two heatingsequences1 and 2 (without web tion with the plastic moment of only the
line heats) result in very small plastic rota- bottom flange in the denominator,the impli-
tions in comparisonto sequences3 and 4 cation is that most of the lateral stiffhess is
with web line heats. provided by the bottom flange. With the W
10 x 39 compositegirder showingrelatively
When applying a lateral load to the small plastic rotations at jacking ratios
bottom flange near the center of the com- greater that 100 percent, it is obvious that
posite girder, the moment producedis trans- the web carries a significant portion of the
ferred to the end reactions by two mecha- jacking force. A primary questionthen be-
nisms: the bottom flange acts as a flexural comes: how is the distribution of jacking
c beamsupportedat the ends,andthe web acts forcesbetweenthe flange andweb to be de-
as a flexural plate (and, at large deforma- termined?
tions, a membraneplate) supportedby the
deck and end diaphragms. A shallower
157
Federal Highway Administration
-
WV 2
(Eq. 7.14)
F, = (46) p)b= 0.0147sinz
The measuredexperimentalvaluesof
161
Federal Highway Administration
-
. W24x76 /
--
“‘$&oretical(PIate Eq. I
0 10 20 30 40
Figure 7.10. Actual load ratio versus plastic rotation for composite girders.
1.2 I
1 .o
0.8 ,
0.6
0.4
0.2
0 L
20 30 40 50 60 70 80
F’, = 0.6+2y- Mj
MP
M,I- MP (Eq. 7.21)
3
where q is the plastic moment capacity of The apparentjacking ratio can be obtained
the lower flange about its strong axis. The usingeq. 7.13 as
theoreticalplastic rotations, eq. 7.17, for the
W 10x39 and W 24x76 composite girders
were superimposed on the experimental Mj < l
--
plots in fig. 7.10 in which effective jacking (Eq. 7.22)
MP - 3Y
ratio was the ordinate. Excluding the high
jacking ratio data point associatedwith the
hot mechanicalstraightening phenomenon, Modeling Statically Indeterminate
the analyticalformula provides a reasonable Spans Due to Intermediate Diaphragms.-
approximation. Practicallyall spansover roadwayshave in-
The experimentalevidenceindicates termediate diaphragms. When the lower
that the degree of plastic rotation per heat flange is impacted, its behavior resembles
cycle is proportionalto the magnitudeof the that of a beam continuousover severalsup-
restraining force up to a certain limit. For ports with the diaphragms acting as these
higher forces, the behavior becomesnonlin- supports, fig. 7.12a. The impact usually
ear with increasedplastic rotations, as illus- produces a plastic hinge mechanism as
trated in fig. 7.9. This phenomenonis at- shown in fig. 7.12b. The three plastic
tributed to a combinationof (1) the jacking hinges produce reverse curvature bending
forces creating stressesgreater than the re- andyield zonesat the impact point and adja-
ducedyield stressin portions of the flange cent supportsa shownin fig. 7.12~. The vee
heatedzones (often referred to as hot me- heat pattern is also shown in fig. 7.12~.
chanical straightening); and (2) the spread- Both the positive and negative curvature
ing of the yield zone and the associatedre- sectionsshould be heated either simultane-
distribution of moments. Since little evi- ously or in quick succession. Plastic rota-
dence exists as to the safety of such high tion will occur at all three locations with
jacking forces,a load limit is necessary. relatively little restraint from adjacentplastic
Figure 7.12. Lower flange of composite girder The apparentjacking moment is found by
modeled as a continuous beam.
substitutingeq. 7.23 into eq. 7.13 and solv- -
hinges. Consequently,the model for the ing for M& thus
singlespancaseshouldprovide a reasonable
approximationof this more complex situa-
tion. An important considerationfor com- 1 MP
posite girder repair is the residual stresses Mj S -(- -“r) (Eq. 7.25)
Y 3 -
inducedduring both the damageand the re-
pair phase. For diaphragm-braced members,
the result of the damageinducementprocess and the apparentjacking force can be found _
is the creation of residualmoments. These from eq. 7.1. Note that the movement asso-
moments can be computed by first per- ciatedwith the first severalheatswill elimi-
forming a plastic analysison the continuous nate the residual moment. Care must be _
beamto determinethe ultimate load (impact taken not to over-jackon the first few heats.
force), P,, and the plastic moment diagram, If the residualmoments are not computed,
Mu. A modified stiffnessof Ki = l/r Kr can than the maximum jacking ratio should be -
be usedfor thesecomputations. The second reducedduring the first two heats. The rec-
step is to take the computedP,, apply it in ommended limit is one-half the standard
the oppositedirection, and computean elas- limitation or 16 percent. After two heatsthe -
tic moment diagram,Me, evenif the stresses 33 percentlimit canbe used.
are greater than Fr. Finally, the superposi- -
tion of these two moment diagrams yields
165
Federal Highway Administration
-
-
Chapter 8. Heat Straightening Trussesand Other Axially Loaded Members
H
Section AA
Figure 8.1. Dead load conditions on a simply sup- Figure 8.2. PA effect on an axially loaded column.
ported beam.
both beamswere subjectedto axial compres- Category S casesthe standardpattern was a
sion of 0.35 times the AISC (1989) allow- full depth, 30’ vee on the web followed by a
able axial load. A total of 70 heatingcycles strip heat on the flange at the open end of
were performed. Jackingratios rangedfrom the vee (seefig. 2.11).
zero to 50 percent. A zero jacking ratio se-
Heat Straightening Response of Columns
quencewith an axial load of 0.175times the
with Category W Damage
allowable axial load was also conductedin
the caseof the CategoryW damage. With the axial load applied, a mo-
ment in the member is createddueto the PA
The specimensusedwere 6 m (20 ft) effect. This moment tends to impede the
long, A-36 steelmembers. In four casesthe
heat straighteningprocessas it acts to mag-
member was damagedabout the minor axis
nify the damage. The approachused here
while in three casesthe damagewas about
was to cancelout this moment with the ap-
the major axis. The damagewas statically plication of the lateral jacking force. After
induced at the center of the member using each cycle, the moment resulting from the __
hydraulicjacks. The memberswere simply
PA effect was computed. The jacking force
supportedand diaphragmswere placed only was adjustedto includethe specifiedjacking
at the ends. No axial loads were applied _
ratio plus a moment to cancel out the PA
during the damageprocess. Prior to heating,
moment at the center of damage. Thus, the
the axial force was appliedby meansof the
jacking ratio for thesetests is definedas the
tension rods attachedto end plates on the -
ratio of the moment at the center of damage
members (fig. 8.3). Lateral jacking forces
due to both the PA effect and the jacking
were applied by meansof hydraulic jacks.
force, M, divided by the plastic moment ca-
One heating pattern was % depth, 45” vees -
pacity of the cross-sectionabout the axis of
on the flangesand a strip heat on the web. damage,q. The HP 12X53 had an axial
This pattern is the standardpattern for Cate- load, P, of 362 kN (8 1.4 kips), producing a
gory W damage (see fig. 2.12). For the compressivestress of 36.2 MPa (5.25 ksi), -
Figure 83. Jacking force applied to axially loaded column in test frame (Tension rods are visible on right side
-
of column).
+
-
Axial PA Stress
compression
stress Total stress
-
(a) Stress distribution after damage
+ + + =
ii i
Axial PA stress PA stress Stress due
compression subtracted to part of
stress cs port of jacking Total stress
jacking force used
force to create
the load
ratio
Figure 8.4. Stress distribution in axially loaded column (A = cross section area and S = section modulus).
-.
measuredand the averageis shown in the the first heat to be exceptionally large. In
table. In generalthe plastic rotations of both order to comparethe data to theoreticalval-
were similar. The responseof beamNo. 1 to ues, the averageplastic rotation was also
heat straighteningis shown in fig. 8.5 and computedexcludingthe first heat. The re-
was typical in form to all category W col- sults are plotted in fig. 8.6. Also shown is
umns tested. For the casesin which the the theoretical curve for the beam without
jacking ratio was zero (thejacking force was axial load based on the same parameters.
only used to cancel the PA effect), there The plastic rotations varied linearly with the
were several heats where the resulting jacking ratio. However, they tended to be
movement was in the opposite direction to smaller than those predicted for the same
that expected. However, successiveheats beamwithout axial load (eq. 6.1). It appears
producedpositive movement. This behavior that the axial force tends to reduce the ex-
illustrates a facet that occasionallyoccurs pectedvaluesover thosewithout axial loads.
during heat straightening. Small or reverse While more researchis needed,the limited
movements sometime occur when low or data suggestthat the previously developed
zerojacking forces are used. The probable equationsfor plastic rotations is applicable
causeis the.build-up of unfavorableresidual here with a modification factor applied. If
stresseswhich are relieved in successive the movement is proportionally reduced
heats. A similar behavior was observedin with higher axial loads,the stressfactor, F,,
the category S tests for a zerojacking ratio, of eq. 6.5 canbe modified as
which is describedin the next section.
As found in other applications the 2 z Mj
residual moments resulting Corn damage F, ={l-2[1-(T)(+-)(1-+ (Eq. 8.1)
P a
inducementcausedthe plastic rotation Corn
8
-I
heating cycles for compression member No. 1 (45” Jacking Ratio (%) -
vee heats, 650°C (12OOT) temperature, and modi-
fied jacking ratio of 50 percent). Figure 8.6. Plastic rotation versus jacking ratio
where f, = the axial compressionstress in for axially loaded Category W columns (f.J F, = -
the member and F, = the allowable axial 0.35 except as noted, vee angles = 45” and heating
temperature = 650°C or 1200’39.
stress. -
The equationfor plastic rotation is in
the form of eq. 6.1, that is buckling to occur due to the large load re-
quired to induce major axis damageon a _
full-scalebeam. The bracingswere removed
during the repair sequenceso asnot to influ-
ence the plastic rotations. The three se- _
quencesare representedasbeamNos. 5-7 in
where F, is given by eq. 6.7, F, is given by table 8.2. A load of 129 kN (29 kips) was
eq. 6.12, F, by eq. 8.1 and <pbby eq. 6.2. A requiredto induce a deflection of 43.2 mrn -
plot of this equationfor the HP12 x 53 also (1.70 in) on the initially straight column.
is shown in fig. 8.6 and showsgood agree- Thirteen heating cycles were performed on
ment with the measureddata. Beam 5 using a jacking ratio of zero. A vee -
angle of 30” was used for all cases. The
Response of Columns to Heat Straight- standardcategoryS heatingpatternwas used
ening for Category S Damage consistingof a web vee heat followed by a -
A W 10X39 column,which hadbeen flange strip heat at the open end of the vee
damagedabout the strong axis, was dam- (seefig. 2.11). The columnswere not com-
agedand repairedthree times usingjacking pletely straightened in each test series. -
ratios of 0.00, 0.25 and 0.50. In order to Rather, ten heating cycles (13 on No. 5)
prevent lateral torsionalbuckling of the col- were conductedin order to define the re-
umn during the damagephase,the beamwas sponse. The averageplastic rotations are
laterally braced at the third points. There shown in table 8.2 and plotted in fig. 8.7.
was a tendencyfor lateraltorsional Becausethe first heat after re-damageoften _
Beam No. of Heats Jacking Ra- Ratio of Ax- Max. Stress? Min. Stress’ Average
tio ial Stress to (MPa ksi) W*i) Plastic Rota-
Allowable tions (milli-
stress rad)
5 10 0 0.35 -MO/-6.23 -43-O/-6.23 1.15
6 9 0.25 0.35 -139.01-20.2 53.617.77 2.74
7 9 0.50 0.35 -206.Ob29.9 , 121.Oh7.5 4.93
axially loadedcompressionmemberswhere
Fe is given by eq. 6.7, F, by eq. 6.12, F, by
eq. 8.1, and(pbby eq. 6.2.
Summary
A seriesof full scaletests were con-
ducted on both Category W and S damage
axially loadedcolumns. Sevenstraightening
procedureswere conductedwith at least 10
heatsused for eachset of parameters. The
axial load imposedduring the straightening
0 10 20 30 40 50 60
processwas typically 35 percent of the de-
Jacking Ratio (‘n)
sign allowablevalue. Thejacking force was
modified to cancelthe PA moment effect in
Figure 8.7. Plastic rotation versus jacking ratio addition to providing the specifiedjacking
for Category S columns (30” vee angle and 650°C ratio. The responsefor both caseswas sig-
or 1200°F heating temperature). nificantly lessthan that predictedfor a wide
results in larger than usual movements,this flange beam without axial load. A suggest
heat was not included in the averagevalues theoreticalmodel agreedwell with the data.
shown. Key Points to Remember
As with the CategoryW damage,the l Heat straighteningis effective for axially
CategoryS damageplasticrotations fell well loadedcolumns using the samepatterns
below those predicted by the theoretical as for caseswithout axial loads.
equationwithout axial load. The use of eq.
8.2 is shownin fig. 8.7 and agreeswell with l The movementsafter heatingwill tend to
the data. Thus, eq. 8.2 can be used to pre- be smaller than the case of zero axial
dict plasticrotations for repair of Category S loadson the samemember.