recent brittle fractures in steel bridges - connor fisher

8/10/2019 Recent Brittle Fractures in Steel Bridges - Connor Fisher

1/17

1

Recent Brittle Fractures in Steel Bridges

and Preventative Mitigation Strategies

Topic Area III Fatigue

Robert J. Connor1

John W. Fisher2

William J. Wright3

Keywords: Fracture, Fatigue, Bridge Failure, Forensic Investigation

AbstractBrittle fracture results in unplanned loss of service, very costly repairs, concern regarding the future

safety of the structure, and potentially loss of life. These types of failures are most critical when there is no

evidence of fatigue cracking leading up to the fracture. Hence, the failure occurs without warning and the detailsare essentially non-inspectable. In these cases, it appears desirable to take a proactive approach and introducepreventative retrofits to reduce the potential for future crack development. These efforts will help ensure that the

likelihood of unexpected fractures is minimized.

This paper examines the behavior of two bridge structures in which brittle fractures have developed in

recent times, discusses the causes of the failures, and offers suggestions preventative mitigation techniques. In

situations where considerable uncertainty exists in the prediction of accumulated damage or in the ability toreliably inspect critical details, preemptive retrofit strategies appear to be highly desirable.

1Research Engineer ATLSS Engineering Research Center, Lehigh University, Bethlehem Pennsylvania, 18015, USA, Email:

[email protected]

2Professor Emeritus of Civil and Environmental Engineering, Lehigh University, Bethlehem Pennsylvania, 18015, USA,

Email:[email protected]

3Research Engineer FHWA Turner Fairbanks Research Center, 6300 Georgetown Pike, McLean, VA 22101, Email:

[email protected]

right ASCE 2 5 Structu

Structures Congress 2005
mailto:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]


2/17

2

IntroductionCompared to fatigue cracking, the number of fractures in highway bridges has been relatively small over the past

40 years. However, brittle fracture results in unplanned loss of service, very costly repairs, concern regarding the

future safety of the structure, and potentially loss of life. These types of failures are most critical when there is no

evidence of fatigue cracking leading up to the fracture. Hence, the failure occurs without warning and the details

are essentially non-inspectable.

Field instrumentation and in-service monitoring is a very useful tool in estimating the remaining fatigue life in a

bridge. However, for the fracture limit-state, many details are very difficult to inspect and instrumentation cannot

provide the needed information to make a sufficient evaluation. In these cases, it appears desirable to take a

proactive approach and introduce preventative retrofits to reduce the potential for future crack development.These efforts will help ensure that the likelihood of unexpected fractures is minimized.

This paper examines the behavior of two bridge structures in which brittle fractures have developed in recent

times, discusses the causes of the failures, and offers suggestions preventative mitigation techniques. In situations

where considerable uncertainty exists in the prediction of accumulated damage or in the ability to reliably inspect

critical details, preemptive retrofit strategies appear to be highly desirable.

Case Study 3 1 US 422 BridgeThe US 422 bridge over Schuylkill River is a six span two-girder steel structure, designed and built in 1965. The

bridge was originally built as a pin and hanger bridge but was retrofit in the early 1990s to remove the pin and

hangers. Full moment connections were installed by splicing a complete section of girder into the area where the

pin and hanger details were located. The concrete deck is supported by a stringer/floorbeam system and was

designed as noncomposite with the steel members. A lateral bracing system is located near the bottom flanges

and is connected to the web and floorbeam connection plate by a gusset plate.

On May 20, 2003, a fracture was found during a routine inspection on one of the two girders supporting the

eastbound lanes in the positive moment region of an interior span, as shown in Figures 1 and 2. Based on criteria

developed by FHWA and Lehigh University, the bridge had been identified as being susceptible to constraint

induced fracture (CIF). Retrofit plans to mitigate the condition had been developed. Ironically, the bridge was

scheduled to be rehabilitated within a matter of months when the fracture occurred.

The fracture initiated at the intersecting welds connecting the web, gusset plate, and transverse connection plate

approximately three inches above the bottom flange. The fracture propagated upward above the gusset plate

through the web about six inches and arrested at some discontinuities within the web plate. The fracture also

propagated downward and completely severed the bottom flange. Immediately after the crack was found, all

traffic was diverted off of the eastbound structure and weight restrictions were placed on the twin westbound

structure. The eastbound bridge remained closed to all traffic until a full bolted splice was installed and other

retrofits already planned were in place. The complete forensic investigation is fully reported in reference [1].

Examination of Fracture on SiteThe day after the fracture, an inspection of the failed girder was made using an under bridge inspection unit

placed on the westbound bridge. The inspection revealed that the fracture surface was still clean and rust free,indicating that it had recently occurred. In order to prevent the fracture from reinitiating and traveling further up

the web, a two inch diameter core was taken in the web (see Figure 3). This was done to effectively blunt thecrack tip and allow it to be removed for inspection.

Examination of Fracture in the LaboratoryA portion of the girder was removed and sent to Lehigh Universitys ATLSS Engineering Research Center for

examination. The region removed from the plate girder is shown in Figure 3 and consisted of the entire portion ofthe web and flange that fractured. Portions of the fractured lateral gusset plate were also attached to the specimen.




3/17

3

The two-inch core, previously mentioned was also included for examination. Figure 4 shows the generalappearance of the crack surface.

A light corrosion layer was evident over most of the surface from exposure to the weather prior to removal.

However, the fractured portion of the girder remained on the bridge for several weeks prior to removal.

Nevertheless, no evidence of heavy corrosion resulting from long-term exposure was seen anywhere on the crack

surface which precluded the existence of an earlier crack in the gusset-vertical stiffener weld or girder web.Chevron marks on the web fracture surface clearly pointed to the weld intersection of the lateral gusset plate and

the vertical stiffener as the origin of the girder fracture (see Figures 4 & 5). The brittle fracture propagated into

the bottom flange via the longitudinal web-to-flange fillet welds and also propagated upward in the web until it

arrested about nine inches from the bottom flange.

The origin of the brittle fracture in the fillet weld connecting the gusset plate to the vertical stiffener connection

plate appeared to be in the vicinity of the weld intersection (within 1 inch) of the web. However, the exact point

of origin was less clear since the weld quality was poor and multiple initiations occurred along the weld before

arresting in the gusset plate beyond the vertical stiffener. The cause of the poor weld quality appeared to be due to

excessive root opening between the gusset and the vertical stiffener from poor fit-up. It is apparent in the detailed

view of the web fracture origin shown in Figure 5 that the gusset-vertical stiffener weld crack and web crack were

offset about inch and not continuous. As a result, this required a separate but near simultaneous brittle fractureinitiation in each element.

Figure 5 shows a detailed view of the crack surface at the lateral gusset/vertical stiffener weld intersection after

cleaning corrosion and debris from the surface. A shear lip at the termination of the gusset-vertical stiffener weld

crack was also apparent and consistent with the two separate fractures noted above. It is also clear from

convergence of chevron marks that the origin of the web fracture was located in the weld metal at the corner

intersection of the gusset-web weld and the gusset-vertical stiffener weld. Incomplete fusion and entrapped slag

were also evident at the weld root and toe that is not uncommon at welds intersecting at corners.

SEM Examination of Fracture OriginThe web fracture origin region was examined microscopically with a scanning electron microscope (SEM) to

identify the presence and source of an initiating defect at the onset of crack instability. Other than several sub-millimeter sized weld porosity defects no evidence of an initial fabrication defect was observed in this region.

There was also no evidence of fatigue crack development in this region from surface or internal discontinuities

present. The only fracture mechanism observed in this area was cleavage fracture as shown in higher

magnification images in Figure 6.

The absence of evidence of stable crack growth at the fracture origin is similar to the Hoan Bridge girder fracturesat similar details [2]. In this case fracture was attributed to constraint induced fracture (CIF) from high levels of

tri-axial constraint at the crack-like geometrical condition at the weld intersection of the gusset plate and vertical

stiffener elevating stresses well beyond the yield point in this region and stress intensities exceeding the fracture

toughness of the web material. Although direct evidence of defect sharpening by cyclic stresses was also not

observed in this case, it was concluded that this had likely occurred at the microscopic level prior to the fracture.

The similarity of the detail and absence of stable crack extension in the current fracture is suggestive of a similarfracture mechanism.

Material Properties

Chemical CompositionA chemical analysis of the flange, web, and lateral gusset plate was performed to verify the steel grade of each

plate material. The results of the analyses indicated that all three plate materials conformed to the chemical

requirements of ASTM A36 steel. The carbon, manganese, and residual element (P, S) content were typical of

this grade of steel.




4/17

4

Mechanical PropertiesThe tensile and Charpy-v-notch (CVN) properties of the plate materials were measured to determine their strength

and fracture toughness properties. The web and flange plate conformed to the strength requirements of ASTM

A36 providing average yield points of 38 ksi and 35 ksi, and average tensile strengths of 62 ksi and 76 ksi,

respectively. Although specimens were not tested from the gusset plate the composition indicated it was grade 36

steel and was assumed to possess strength consistent with this grade. Although only the flange satisfied thecurrent AASHTO Zone 2 fracture critical toughness requirement for Grade 36 steel of 25 ft-lbs @ 40 F, the web

and gusset plate were marginal. It is important to note that these requirements were not in place when the bridge

was fabricated.

Conclusions Case Study #1

The material properties of the web, flange, and gusset plates were found to conform to ASTM A36 steel. The

fracture toughness of the flange plate elements satisfied or exceeded the current AASHTO Zone 2 fracture

critical toughness requirements for this grade of steel. The web and gusset plate were marginal.

Fractographic examination indicated that brittle fracture developed at two separate locations at the lateral

gusset connection detail nearly simultaneously. A brittle fracture of the gusset plate vertical stiffener fillet

weld and a brittle fracture of the web developed at the intersection of the gusset plate and vertical stiffener

weld. A fracture mechanics analysis of the crack-like conditions existing at both fracture origins indicated that the

gusset plate fillet weld likely fractured under an unusual loading at moderately low ambient temperatures

(50F) and imparted a dynamic load to the crack-like condition at the intersection of the gusset plate and

vertical stiffener which exceeded the dynamic fracture resistance of the web plate material at that temperature.

Case Study #2 - Hoan Bridge FailureOn December 13, 2000, cracks were detected in all three girders of the south approach span o the Hoan Bridge

adjacent to the tied arch spans. Two of the three girders had full depth fractures, as illustrated in Figure 7. The

span had sagged about a meter, leaving the span near collapse and the entire roadway was closed to traffic. On

December 28, 2000, the critically damaged section of the northbound road north of the field splice adjacent to

Pier 3S, was removed by explosive demolition. Figure 8 shows the dropped end span girders that fell about 30 m

onto a gravel pile. The southbound roadway was reopened to two-way traffic on February 27, 2001, with weightand speed restrictions after short term retrofit measures. A more comprehensive discussion of the forensic

investigation can be found in references 2 and 3.

Design Details and Material PropertiesThe three span continuous plate girder bridge, had three hybrid girders that were 3.2 m deep. The 12.5 mm web

plates were A36 steel with a yield stress of 270 MPa. The girder flange plates were A588 grade 50W steel. The

center girder flange plates were 76 mm thick with a yield stress of 353 MPa. Outside girder flanges were 57 mm

thick. Girder F was observed to have a yield stress of 362 MPa and Girder D had a yield stress of 322 MPa.

The original structure was designed as a noncomposite system. The calculated dead load stress in the girder web

at the lateral gusset plate was about 105 MPa. Although designed without composite action between the steel

girders and the concrete slab, strain measurements verified that full composite action from friction was presentbetween the girder and the deck for all live load conditions. The lateral K-bracing system was connected to the

girder web plates with the shelf or gusset plates using a bolted connection.

The 19 mm lateral gusset plates were slotted to fit around the transverse connection plates. These plates had a 13

mm beveled partial penetration weld joint between the girder web and the gusset plate. Figure 9 shows a view of

a typical gusset plate fitted around the transverse connection plate. One side of the slot is in contact with the

transverse connection plate creating a crack-like geometry. High strength bolts provided a connection between

the gusset plate and the transverse connection plate.




5/17

5

Charpy V-Notch tests were carried out on the girder web and flange plates at the fracture locations by the Turner-

Fairbank Highway Research Laboratory. Full transition curves were developed to define the variation in material

toughness with temperature. The testing revealed that all three web locations satisfied the AASHTO requirements

for Zone 2 fracture critical applications where the lowest anticipated service temperature is -34C.

The CVN test results were used to develop dynamic and bridge loading rate estimates of fracture toughness, K,

based on the correlation equation [4] KId = )(64.0 CVNE , where KId is the dynamic fracture toughness

MPa m , CVN is the Charpy energy in J, and E the elastic modulus MPa. The bridge loading rate was estimated

from the temperature shift Ts= 0.75 (119 0.12 Fy) C. These results indicate that the crack initiation resistance

of the web plates would be on the upper shelf at the service temperatures encountered at the Hoan Bridge. The

dynamic toughness is in the lower transition region, providing limited resistance to dynamic crack propagation.

Similar tests on the flange plates yielded test results and fracture toughness estimates. The fracture toughness of

the flange plates was also obtained using 2 in. thick compact tension tests on Girders D and E at -34C and 21C

at static and intermediate load rates. Only the low temperature tests on the 3 in. thick Girder E flange plate

provided valid KICvalues for true plain strain conditions. Nonetheless, all of the test results provided reasonable

agreement between the compact tension test results and the KICtransition curves calculated from the CVN data.The results suggest that the estimated dynamic fracture toughness of the 57 mm flange plates from Girders D and

F was about 66 MPa m at 23C, the probable temperature at failure. For the 76 mm Girder E flange plate,

the estimated dynamic toughness was only 27.5 MPa m .

Fractographic ExaminationSelected areas of the fractures in Girders D, E, and F at P.P. 28 were examined both visually and microscopically

using a scanning electron microscope (SEM) to characterize the fractures and gain additional information about

the fracture mechanism(s). Examination focused primarily at the fracture origin areas in the girder web at the

shelf plate weld terminating adjacent to the vertical connection plate in each of the girders. Similar areas in other

girders where cracks were also found were also examined.

In general, all of the fracture surfaces examined were covered with a layer of corrosion of varying thickness

resulting from exposure to the weather and salt prior to removal from the structure. Evidence of abraded areas on

some crack surfaces, which occurred during demolition of the span, was also observed. Prior to examining, the

as-received fractures were documented photographically. Areas for detailed examination were extracted by saw-

cutting followed by ultrasonic cleaning in Alconox detergent to remove as much of the corrosion product as

possible. Removal of the corrosion layer to examine the underlying fracture surface was moderately successful,

however, the tenacity of the corrosion product in some areas, often at fracture origin areas, frequently prevented

microscopic information from being obtained in these areas.

The fracture origin in Girder E had been traced to the girder web at the lateral bracing connection where the shelf

plate partial penetration weld joint to the web terminated adjacent to the vertical connection plate. The opposing

side of the web fracture with the vertical connection plate still attached is shown in Figure 10. Cleavage chevronmarks on the web fracture surface were observed to point from both directions to the shelf plate weld termination

thus confirming this area to be the fracture origin. The web fracture origin at the weld termination is shown in

Figure 10 and clearly shows a fan-shaped cleavage chevron pattern emanating from the vicinity of the weld root

lack of fusion. A similar radial cleavage chevron pattern at the shelf plate weld termination was also observed on

both sides of the fracture coinciding with a origin point in the vicinity of the weld root lack-of-fusion.

Visual examination of the fracture origin area of both sides of the fracture showed no indication of a base metal or

weld defect in this area. Microscopic examination with the SEM (see Figure 11) showed the presence of cleavage




6/17

6

fracture as close to the fracture origin as could be discerned. Corrosion product not removed during cleaning ofthe surfaces obscured some areas at the origin, however, no indication of stable crack growth processes such as

fatigue or ductile tearing was observed in corrosion free areas within 1 mm of the origin. Without the presence of

a macroscopic defect cleavage likely initiated from a microscopic discontinuity, which may have sharpened by

fatigue over time. Fracture initiation from a microscopic defect is not inconsistent with the high triaxial stress

state believed to exist in this area.

The fracture origin in Girder D traced back to the same location as in Girder E. The section of Girder D

containing the crack origin at the shelf plate weld joint termination is shown in Figure 12. The opposing crack

surfaces at the fracture origin removed from the corner of the shelf plate and from the web along the vertical

connection plate as for Girder E. One distinct cleavage fracture origins occurring in offset planes resulting in astepped appearance of the fracture. One origin appeared to be in the shelf plate weld near the weld root although

a well defined original as determined by cleavage chevron markings were not as clear as in Girder E. The other

origin was clearly traceable to the shelf plate weld toe area. In both cases no visual indication of an initiating

defect was observed in the origin areas. A weld metal gas pore about 1 mm in diameter was observed in one of

the origin areas, however, it is not certain if this defect was associated with the cleavage initiation.

The fracture origin in Girder F as well as two other web cracks that were detected at other lateral bracing

connections in other spans of the structure provided identical results when examined.

The crack arrest in the bottom flange region of Girder D was examined to determine the location of crack arrest

and characterize the fracture mechanism. The web crack clearly propagated through both web-flange fillet welds

and arrested in the fine grain heat affected zone of the weld.

Conclusions

All three girder web fractures initiated from the crack-like geometric condition that resulted from the

intersecting shelf or gusset plate and transverse connection plate welded connections with intersecting and

overlapping welds.

The existing geometric configuration caused extreme high levels of constraint and stress to develop in the

web plate gap.

Brittle fractures (cleavage) were found to develop at every web crack examined without any detectable fatiguecrack extension or ductile tearing at the crack origin.

Once the web fractured, it was found that the bottom flange plates of Girders E-28 and F-28 were not capable

of arresting the propagating crack.

All of the flange and web steels were found to have mechanical tensile properties and Charpy V-Notch

toughness that satisfied the AASHTO requirements at the time the structure was built. The Charpy V-Notch

toughness was also found to satisfy current (2001) requirements for Zone II.

The web plates were found to have sufficient toughness to tolerate through plate thickness cracks under

normal conditions without the high constrain conditions that were imposed on the girder web plate by theshelf plate welds and the intersecting transverse connection plate welds.

The nature of the web crack development results in a detail that is not inspectable. The small critical crack

size cannot be detected.

Identifying Details Susceptible to CIFThe most common details susceptible to CIF are located at intersection of gusset plates and transverse connection

plates (stiffeners). In addition, intersections between vertical stiffeners and longitudinal stiffeners can also present

a problem. A recently completed in-depth finite element study conducted at Lehigh University, has confirmed

that gaps greater than inch sufficiently reduce the triaxial stress condition within the gap to acceptable levels.

Figure 13 is an example of the detailed finite element model developed to analyze this complex connection.




7/17

7

As a result of increasing the gap, the material in the web is permitted to yield rather than being subjected to higherand higher stresses until the limit state becomes brittle fracture. In this study, various geometries of gusset plate

connection were evaluated for different gap lengths. The values presented in Table 1 are from one of these

studies and clearly show that a web gap size of inch is effective in reducing the stresses in the X-direction ( 1)

and in the Y-direction (2) by 26.2% and 36% respectively, as well as totally eliminating the stresses in the Z-

direction (3) (through thickness). The table also indicates there is a significant decrease in the triaxiality factor

(T) to within acceptable limits.

In the presents of a inch gap, the level of constraint is reduced and brittle fracture should not occur. It is critical

to point out that the inch gap is the distance between weld toes of the components and not the actual distance

between the plates.

Other details common to other bridge types like box girders for example are also are susceptible to CIF. These

details need to be evaluated on a case-by-case basis to determine if a retrofit is needed. The same gap

limitation will likely still apply.

As part of a research project currently underway at Lehigh University, a detailed survey related to CIF was

conducted. The survey revealed that brittle fracture as a failure mode is not as well recognized or understood by

most bridge engineers. For example, in some cases, small to moderately sized cracks were observed at gussetplates and the problem was dismissed as a fatigue issue. This conclusion was drawn although no forensic

investigations were conducted. In these cases, the corrective measures were taken to prevent future fatigue cracks

from growing into the web. Interestingly, these retrofits relied on hole drilling which also removed enough

material to decrease constraint effects (and hence triaxiality) and the stress concentration to acceptable limits. If

the holes were drilled in the web, a brittle fracture, if initiated, would enter the hole and likely be arrested.

Another interesting point revealed in the survey is that there are no standard procedures for investigating,

documenting, and archiving such failures. In summary, due to the lack of awareness of the phenomena as well as

the lack of defined procedures that should be implemented when a failure occurs has led to several cases of CIF

not being identified as such and obviously being documented at all. More importantly, it is not clear if the bridges

which have exhibited these fractures in the past were subsequently retrofit to correct the problem at the other

similar details or simply repaired and considered rogue failures.

Retrofit StrategiesSince the fracture of the Hoan Bridge in Milwaukee WI and the more recent fracture of the US 422 bridge, there

has been considerable interest in identifying effective methods to retrofit similar details. It is extremely important

to note that the failures observed in both bridges gave no warning. The details can not be inspected unless they

fracture. Hence, it is of utmost importance to identify details that may be susceptible to CIF and implement a

retrofit program. It is worth noting that the state of Pennsylvania has been very aggressive in this effort.

There are several retrofit options to prevent constraint-induced fracture in details that have been determined to be

vulnerable. Each case has to be evaluated on an individual basis. Geometry constraints, access limitations,

contractor capabilities, and long-term performance are factors that need to be considered when selecting options.

An engineer may determine that other retrofit options are the best choice for a particular case. Retrofit details andprocedures should be individually evaluated by an engineer to determine: 1) if the risk of constraint-induced

fracture is eliminated; 2) if the strength of the resulting joint is adequate to carry all applicable loads; and 3) if the

retrofit will provide adequate fatigue performance over the intended life of the structure

The problem is relatively easy to correct for most details and relies primarily on hole drilling techniques. For

gusset plate details, a preferred method is to drill holes in the gusset plate near the intersecting welds in order to

increase the gap and reduce constraint as well as the tri-axial stress condition. Difficulties arise in this retrofit

primarily due to the limited room for access at most of the details. In addition, quality control must be maintained




8/17

8

to ensure that the web plate is not nicked or gouged. After the holes are in place, the web gap area should beground smooth using appropriate die grinding equipment and flapper wheels to provide a smooth surface finish.

If the gusset plate or longitudinal stiffener is fillet welded to the web, care should be taken to blunt the transition.

Smooth tapering of fillet welds result in hair-thin cross sections of weld metal that can be vulnerable to fatigue

cracking. Poor quality work can result in fatigue cracking at the retrofit or other problems. Typical sketches of

effective retrofits for gusset plates and longitudinal stiffeners are shown in Figure 14.

Summary1. It is critical to identify and address details that are susceptible to CIF to prevent similar failures on other

bridges as there is no warning to these fractures. Hence, it is of utmost importance to identify details that may

be susceptible to CIF and implement a retrofit program.2. Standardized procedures should be developed to document, investigate and archive structural failures to

ensure problems are understood and prevent similar failures in the future.

3. A gap width of inch or greater reduces triaxiality and constraint to acceptable levels thereby allowing the

web material to yield rather than fail in brittle fracture.

4. Retrofit techniques have been developed which are economical and effective in mitigating CIF for various

types of details. These methods primarily rely on hole-drilling techniques and require superior quality control

standards be maintained to ensure other problems are not created.

AcknowledgmentsThis forensic investigation into the failure of the US 422 bridge was sponsored by the Pennsylvania Department

of Transportation and the ATLSS Engineering Research Center at Lehigh University. This forensic investigation

into the failure of the Hoan Bridge was sponsored by the Wisconsin Department of Transportation and the Federal

Highway Administration. The authors are indebted to Bala Sivakumar and William Edberg, Lichtenstein

Consulting Engineers for providing support during the Hoan Bridge failure investigation. Thanks are also due to

Zhan Xi and Hernando Tjiang for their assistance with material tests and modeling at the Turner-Fairbank

Research Center, FHWA. The authors are also grateful to Mr. Hussam Mahmoud, who performed the detailed

linear and non-linear finite element analysis used to develop the data related to the triaxaility factor. Dr. Eric

Kaufmann performed the fractographic analysis for both studies.

References1. Kaufman E.J., Connor R.J., Fisher, J.W. Failure Analysis of the US 422 Girder Fracture Final Report,

ATLSS Report No. 04-21, Center for Advanced Technology for Large Structural Systems, Lehigh University,

Bethlehem PA, 18015, October 2004

2. Wright, W., Fisher, J. W., Sivakumar, B., Kaufmann, E. J., Xi, Z., Edberg, W. And Tjiang, H., Hoan Bridge

Forensic Investigation Failure Analysis Final Report, FHWA, June 2001.

3. Wright, W. J., Strock, T., Hartmann, J., Hoan Bridge Forensic Investigation - Documentation of Condition

Prior to Demolition, FHWA Research Report, March 2001.

4. Barsom, J. M. and Rolfe S. T., Fracture and Fatigue Control in Structures, 3rdEdition,ASTM Stock

Number: MNL141,ASTM, West Conshohocken, PA, Nov. 1999.




9/17

9

Web Gap Size

Zero inches 1/4 inch % Change1, (ksi) 57.2 42.2 - 26.2

2, (ksi) 18.9 12.1 - 36.0

3, (ksi) 18.3 0 -100.0

T 1.48 1.12 -24.3

Table 1 - Effect of web gap on principal stresses and triaxiality factor




10/17

10

Figure 1 Fractured girder exterior face

Figure 2 Fractured girder - interior face

Web Crack




11/17

11

Figure 3 Portion of girder removed for evaluation

Figure 4 - Crack surfaces with chevrons pointing back to fracture origin

at the weld intersection of the lateral gusset and vertical stiffener

Figure 5 - Fracture origin at the intersection of the lateral gusset and vertical stiffener welds

Gusset PlateCrack

Web CrackFlangeCrack

Core Removed atWeb Crack Tip

Shear Lip atFracture Offset

Web Fracture Origin

Web

LateralGussetPlate




12/17

12

Figure 6 - SEM images of the fracture origin region and cleavage fracture

[Mag. 10.1x & 600x]

Figure 7 - Fractured girders at floorbeam 28




13/17

13

Figure 8 - View after demolition of span

Figure 9 - View of lateral gusset insertion

with transverse connection plate




14/17

14

Crack

Origin

CrackOrigin

Figure 10 - Girder E crack surfaces at gusset plate

- transverse connection plate intersection




15/17

15

Figure 11 - SEM micrographs of crack surface at

fracture origin showing cleavage




16/17

16

Crack

Origins

Figure 12 - Section of girder D with crack origin

and arrested crack in bottom flange with view of

crack surface




17/17

17

Figure 13 - Typical finite element model use to investigate potential for CIF

Figure 14 Typical retrofits for gusset plates and longitudinal stiffeners to address CIF

Web

Gusset Plate

Transverse Stiffener

2" Dia. Hole(typical)

TOP VIEW

LongitudinalStiffener

Grind Transitionto Web

Cut Back Stiffener or Increase Web Gap

recent brittle fractures in steel bridges - connor fisher

Documents