·Table of Contents
NDE of Delayed Cracking in Pivot Steel Box GirdersM.G. Lozev
Edison Welding Institute,
1250 Arthur E. Adams Dr., Columbus, OH 43214, USA
(Formerly with VTRC, Charlottesville, VA, USA)
G.A. Washer, W.J. Wright
Federal Highway Administration,
Office of Infrastructure R&D, 300 Georgetown Pike,
McLean, VA 22101, USA
The second-largest double-swing-span bridge in the world, was replaced in record time and was the first bridge replacement "floated" into position with everything ready to carry traffic. The six new sections were preassembled on temporary supports and then floated on barges to the bridge site. ASTM A852, Grade 70W, a C-Mn low-alloy high-strength steel, up to 75 mm thick, was used to fabricate the box girders. To produce high-quality welds with good penetration an automated submerged arc welding (SAW) was used. Several cracks were observed in the first pivot box girder after it was shipped by the fabricator. The cracking probably occurred in the fabrication shop, during handling and shipping, during preassembly, and after on-site erection. Questions have been raised about the safety of these fracture-critical members and the entire superstructure because the interaction between welding-induced defects and fatigue stresses during service is extremely damaging. A comprehensive field nondestructive evaluation (NDE) of the box girders before and after on-site erection and a failure analysis of the steel cores were performed to assess the potential for crack growth.
Visual testing (VT), magnetic particle testing (MT), radiographic testing (RT), and eddy current testing (ET) of the fillet welds and heat-affected zone (HAZ) were performed after the repairs, before and after the swing trusses were floated from preassembly site, and after erection. The depth of the cracks was measured using the drop potential testing (DPT) method and an alternating current field measurement (ACFM) device. Acoustic emission (AE) monitoring after erection was also performed.
The first delayed cracks were originally found in the web-to-stiffener fillet welds in the north girder upon visual inspection following shipment to preassembly site in March 1995. The cracks were typically transverse to the welding direction in the weld crown. Some cracks propagated into the web base metal. A few cracks were discovered at the crown of the fillet weld in the longitudinal direction.
Magnetic Particle Testing.
More than 100 cracks were discovered in the north girder by the MT conducted in March and April 1995. The Parker probe for longitudinal magnetization was used, and dry red particles were applied. MT was performed in accordance with ASTM E709, and the standard of acceptance was in accordance with Section 9.21 of the Bridge Welding Code. The girder had a coating system, made up only of a zinc primer. The primer was removed at the electrodes to provide direct electrical contact with the metal surface. The second MT conducted just before the preassemby was finished detected 7 new transverse cracks in the far girder side in the web-to-stiffener No. 8 weld. The cracks were in the web plate and were approximately 6 mm long and 3 mm deep. Two additional MTs were performed in June 1996 and April 1997. The prod MT method with dry red particles was used with a 100-mm prod spacing, with continuous magnetization at 500 A DC.
The most severe cracks were five transverse cracks detected at the web-to-top flange groove weld in the midspan, otside the south (far) side. The potential of top flange fracture initiating from the existing transverse cracks in the midspan should be the subject of a great deal of concern. MT showed the crack length indications to be from 6 to 31 mm long. The actual crack length observed during the repair work was 6 to 12 mm longer. To excavate the cracks, the weld metal was completely removed in the crack areas. This was an indication that some of these cracks were very deep and probably through-weld cracks.
The distribution of the number and length of cracks in the north girder vs. time after fabrication is shown in Figures 1 and 2. Increased delayed cracking was observed in the web-to-top flange groove weld in June 1996 and in the web-to-bottom flange weld in April 1997. A steady decline of cold cracking was noticed in the web and flange-to-stiffener and diaphragm welds. Most of the transverse cracks were less than 50 mm long. The longest longitudinal crack was more than 0.5 m long and was detected in the web-to-stiffener and flange-to-diaphragm welds. Two longitudinal cracks longer than 375 mm were detected in the web-to-top flange welds. MT reinspection after field repairs detected five transverse cracks 13 mm long and 13 to 25 mm outside the repair areas and only one transverse crack 13 mm long in the repair areas.
|Fig 1: Distribution of Number of Cracks vs. Time after fabrication||Fig 2: Distribution of Length of Cracks vs. Time after fabrication|
Eddy Current Testing.
Detecting transverse cracks in the weld crown once the final bridge coating is applied is still a difficult MT inspection problem. The possible solution is the use of ET. The preliminary field test showed that ET is an alternative to MT that will not adversely affect the coating system on the bridge. If ET were adopted for inspecting the south girder, MT or grinding could be performed to confirm the results, and the number of times the coating system had to be penetrated to make electrical contact would be significantly reduced. This would extend the life of the coating system and reduce the time required to inspect the welds.
Welds 211 m long at the south girder were tested using SmartEDDY 3.0 and DEFECTOMETER 2.837. An ET reference standard, fabricated from A709 steel, was used for the calibration. Three electrode discharge machine (EDM) notches were machined into the standard. The notches were approximately 0.15 mm wide and 25 mm long with depths of 0.5, 1, and 2 mm. The reference standard was unpainted. The measurements of the coating system indicated an average thickness of 0.4 mm. Lift-off changes, attributable to the coating system, were compensated during the calibration process. Lift-off attributable to the coating was assumed to reduce signal amplitudes approximately 40 percent. Indications were determined by a simple reactance threshold equivalent to approximately 20 percent of the full screen height.
Eighteen crack indications were revealed during ET. Six indications were noted in the welds inside the girder, and 12 were noted in the outer welds. The surface in these areas was removed by grinding to allow verification of the ET results. ET reinspection of these areas was performed to verify that the cause of the indications had been removed. Visual observation during the grinding process confirmed one visible transverse crack 6 mm long, two visible longitudinal cracks 6 mm long, eight slag inclusions, six undercuts, and one weld rollover. The probability of detecting discontinuities under the paint was very high, with some probability of making an incorrect determination (overestimation) of cracklike indications. Much research remains to be done in ET of high-strength steels. However, the evaluation of the welds in the south girder demonstrated the viability of using ET to test painted ferromagnetic steels.
Alternating Current Field Measurements.
Another advanced technology used at the bridge was ACFM. The operation of ACFM instrument is similar in theory to that of MT in that a magnetic field is induced in the material and the flux leakage caused by a crack is measured. The significant difference is that the flux leakage is measured by coils rather than by placement of particles on the material surface. The output signal of these coils can be analyzed quantitatively to determine the dimensions of a surface-breaking crack.
The evaluation of the longitudinal crack at location H was performed using the ACFM instrument. The results indicated that the crack was actually two crack segments. The first crack was 30 mm long and 3.8 mm deep, and the second was 11 mm long and 2.4 mm deep. These results were consistent with the MT results and were confirmed later during the metallographic and fractographic examinations.
The U9 ACFM instrument was shown to be effective at detecting and measuring longitudinal cracks in the weld. The ability of this instrument to detect cracks and measure crack depth makes it a valuable tool that resolves a long-standing problem. The benefits of this type of instrumentation far outweigh the inconvenience of the instrument's size and weight. At present, FHWA is providing strong leadership in the bridge inspection field by bringing this innovative and well-developed technology to the United States.
Drop Potential Testing.
DPT, or the four-point probe technique, has been around for years, and its possible use for evaluating crack depth is still being investigated. If four electrodes are placed in contact with an electrically conducting object and current is passed between two outer electrodes, a potential will be produced between the other two inside electrodes (usually placed above surface-breaking cracks). For a fixed electrode arrangement and one material, an observed change in the current/potential ratio will represent a change in the depth of a crack. The CC-800B instrument with an SP1-B probe was used to measure crack depth in different locations during this investigation. The crack depth varied from microns to 25 mm. The accuracy was ± 10 percent. The measurements were performed at the weld face only on surface-breaking cracks. The depth of the cracks in several locations were confirmed later during the metallographic and fractographic examinations.
Isotope Ir-192 with 100 Curies activities was used to conduct RT. The detectability of the longitudinal cracks at location H was not very good. At location A, two transverse cracks were detected. On the radiograph, the visible part of the longer crack (A1) was 25 mm long and the second crack (A2) was 10 mm long. In the plane of the radiographic film, the distance between the two cracks is approximately 1 mm. Only the longer crack was detected during MT in June 1996. The shorter crack was probably a root crack with a crack tip below the weld surface and was not detectable with MT. This assumption was confirmed later during the metallographic and fractographic examination.
Acoustic Emission Testing.
AE monitoring for 12 hours during afternoon rush hours and one very cold night was performed on the north girder in January 1997. A prototype of a portable, wireless, six-channel AE bridge monitoring system developed cooperatively by Physical Acoustic Co., FHWA, and VTRC was used. Very low crack-related AE activity was recorded on the web-to-top flange weld. The location was in the far side, similar to the locations of cores 4 and K. Twelve hours was too short a period to detect very slow and steady delayed crack growth. Continuous AE monitoring was needed.
The authors thank VTRC & VDOT's and FHWA personnel for their great cooperation and assistance. The materials reproduced or quoted verbatim in this paper were taken from VTRC 97-R18 report, Copyright by Virginia Transportation Research Council, a division of Virginia Department of Transportation, and are reprinted with the permission of Virginia Transportation Research Council. All rights reserved.
|© AIPnD , created by NDT.net|||Home| |Top||