In building structures, lamellar tearing is a weld-related failure that occurs in highly-restrained connections between rolled steel members, most commonly in plates. Rolled steel is not a truly isotropic material, and sulfurous defects incorporated during the rolling process create a susceptiblility to internal, brittle fractures between the rolled layers. Lamellar tearing usually occurs in thick steel plates in the "through" direction (perpendicular to the direction of rolling), as localized strains from thermal shrinkage act to fracture the brittle defects and pull the plate apart. These localized strains develop most acutely in highly restrained connections where the steel is not free to to shrink evenly as it cools. Lamellar tearing can be completely internal to the member, or at times can be visible on the surface.
In the 1970's as the AEC industry moved toward building larger, more complex steel structures, lamellar tearing became an important (and costly) issue for structural designers to face. Although lamellar tearing itself was not a new issue, the costly 1970's failures served as an impetus for industry changes to prevent future expensive or dangerous failures. This article contains brief discussions of the first major lamellar tearing failure cases in the country, including the El Paso Civic Center, Grand Coulee Transmission Line Towers, Atlantic Richfield Towers, and King Street Bridge.
Advances in material technology, current AISC connection standards, and conscientious design have helped to reduce and eliminate the threat of lamellar tearing in modern steel construction.
1 Failure Mode
1.1 Failure Mechanism
Figure 1: Lamellar Tear in Through-Thickness Direction, From “Commentary on Highly Restrained Welded Connections”, AISC Engineering Journal, 3rd Quarter, 1973. Reprinted with permission from AISC.
Lamellar tearing is characterized by a combination of brittle ruptures parallel, and ductile shear failures perpendicular (or nearly perpendicular) to the plane of rolling. The two conditions form stepped cracks similar to that shown in Figure 1. These cracks commonly occur within 5 mm of the welding heat affected zone (HAZ) where thermal strain differences are most acute, but can occur well into the base metal. (Ross 1984, p. 265) Interestingly, from a fracture mechanics perspective lamellar tearing is a "sub-critical growth phase," meaning that the cause of member failure is never "lamellar tearing", but rather "brittle fracture" resulting from a net reduced tensile area. (McEnerney 1978, 7)
Lamellar tearing occurs in three phases: terrace formation, terrace linkage, and ductile shear. The three phases occur after welding during the cooling process as weld shrinkage strains increase. (American, 1973, p. 63) They are shown graphically in Figure 2.
Terrace Formation occurs as singular, non-metallic impurities begin to fracture or decohere, forming voids between the steel matrix and the impurity. These voids are called "terraces" because they usually form in clusters parallel to the rolling plane. (Samuels 1999, pp. 108-109) Although this is the most common initiation mode, several others have been reported including:
splitting along ferrite bands
liquation in reheated HAZ
intergranular cracking
strain aging
Figure 2: The three phases of lamellar tearing. C. Brandmeier, adapted from McEnerney.
Terrace Linkage involves the growth of the terraces to form common levels. The void levels remain separate, but converge to the point of near vertical material boundaries. (Samuels 1999, pp. 108-109) The mechanisms of terrace linkage are widely varied and include:
Ductile Shear joins the varied terrace levels with nearly vertical, ductile shear failures. (McEnerney 1978, 6)
1.2 Physical Characteristics
Because the compressed impurities take on a grained, planar nature during the rolling process, the appearance of the failed surface has been described as "woody" or "fibrous" (TWI 2000). Lamellar tears are usually internal to the member, although they are occasionally visible at weld toes, plate edges, or on the welded surface. (McEnerney 1978, 5) Figures 3 and 4 show specimens from lamellar tearing failures.
Figure 3: "Fibrous" Failure Surface, Reproduced courtesy of TWI Ltd.
Figure 4: Failure Section, From “Commentary on Highly Restrained Welded Connections”, AISC Engineering Journal, 3rd Quarter, 1973. Reprinted with permission from AISC.
1.3 Important Distinction: Localized Strain vs. Global Stress
It is important to note that lamellar tearing occurs from localized shrinkage strains as a direct result of the high heat of welding, and not typically from global member loads. Very rarely do service forces reach a magnitude that can internally fracture steel. Historically, significant failures from lamellar tearing are the product of loads applied to members that have already experienced internal fractures at the time of fabrication. This idea is reinforced by the quote from an AISC journal article below:
"Properly designed restrained connections which transfer frame bending moments from one structural member to another do not generally induce localized strains responsible for lamellar tearing. The restraint that is of concern is internal restraint (within a connection made up of several joints) which inhibits small total--but large unit--localized strains resulting from weld shrinkage. ... There is often a lack of recognition that under certain conditions localized strain is significantly more important than stress." (American 1973, p. 62)
2 Causing Factors
The development of lamellar tearing at the time of welding is influenced by three major factors: percentage of sulfurous inclusions, presence of hydrogen sources, and method of welding.
2.1 Inclusions in the Hot Rolling Process
Small non-metallic inclusions such as silicates, sulfides, or oxides from deoxidation are unintentionally ingrained in steel during the hot rolling process. These materials have different properties than the surrounding steel and create clusters of low ductility susceptible to internal brittle failure or decohesion. If hot rolling is conducted at insufficient temperatures, sulfurous materials (typically manganese silicates) will exhibit higher plasticity than the surrounding steel and elongate in the direction of rolling. (Samuels 1999, p 108) Elongation of the inclusions creates a greater chance for clusters to form and develop a susceptible failure plane. Inclusions within layers of pearlite are particularly liable to elongation, although there is no established relationship directly linking pearlite banding and lamellar tearing (American, 1973, p 64). Often the rolling process will compress pearlite and ferrite into grained layers running parallel to the direction of rolling. Inclusions between the layers "pancake" and occupy a larger surface percentage, reducing the strength of the bonds. (Abyazi 2009, p 3)
Figure 5: Short Transverse Reduction in Area vs. Sulfur Content. Reproduced courtesy of TWI Ltd.
Short Transverse Reduction in Area (STRA) is a good indication of inclusion-based tearing susceptibility. Steels below 10-15% STRA are highly susceptible to lamellar tearing, while steels above 20% are almost completely resistant. Figure 4 shows the relationship between percentage of included sulfide content and STRA. Generally, steels with less than 0.005% sulfur content are desirable, and manufacturers can test plates to ensure an STRA value of greater than 20% for critical connections. (TWI 2000)
Three types of sulfide inclusions can occur. Known simply as Type I, II, and III, their presence is determined by the type of deoxidation process used during manufacturing.
Type I inclusions result from the addition of silicon for partial deoxidation in semikilled steels. The inclusions are randomly dispersed, globular, and deform minimally during rolling. They range in size from 2 to 200 µm and aspect ratios of about 2 to 20. Of the three types, Type I inclusions pose the least threat to lamellar tearing because of their random disbursement and minimal elongation. (Samuels 1999, pp. 109-110)
Type II inclusions form from the addition of aluminum or other strong deoxidizers in fully killed steels. They have a rod-like shape, and a higher plasticity at rolling temperatures than Type I inclusions. The rods are prone to align in the direction of rolling and elongate to maximum lengths of several millimeters. Aspect ratios range between 100 to 1000. Type II sulfides usually develop in high-strength, fracture-tough grades, which are fully killed. (Samuels 1999, pp. 109-110)
Type III inclusions also occur in fully killed steels when carbon, silicon, phosphorus, and aluminum are all present in appropriate ratios. The inclusions are octahedral, and have a higher plasticity at rolling temperatures than either Type I or Type II. They exhibit similar characteristics to Type II inclusions after rolling and are nearly indistinguishable without examining a parent sample. (Samuels 1999, pp.109-110)
2.2 Hydrogen Induced Cracking
Author: Seth Moyer, BAE/MAE 2012 Hydrogen-induced cracking in the heat-affected zone (HAZ) of the base or parent metal is another leading factor responsible for occurrences of lamellar tearing in modern steel construction...A major source of hydrogen during the welding process is the moisture that is contained within the electrode coating itself. Other sources of hydrogen in the weld material include moisture in the flux, hydrogenous compounds in the coating and/or flux and dirt, oil, grease and rust on the welding wires. For the weld base material, hydrogen sources include surface dirt, oil, grease and rust, degreasing fluids for cleaning the surface to be welded and hydrogen in the parent steel. Ambient atmosphere moisture can also be a minor source of hydrogen (Bailey 1973, pp. 5-6).
Hydrogen is absorbed into the weld bead from the arc atmosphere during application of the weld. A lot of this hydrogen diffuses out of the solidified weld and some that escapes diffuses into the base metal (Bailey 1973, p. 5). The diffused hydrogen acts in a similar manner to its behavior in a sour (H2S) environment. In a sour environment, hydrogen sulfide dissociates into atomic hydrogen and works into the weld and base materials. The hydrogen atoms reform into molecular hydrogen when they cross non-metallic inclusions. This process can eventually cause a buildup of the molecular hydrogen which builds pressure within the inclusion. This internal pressure forms where the shrinkage stresses have also concentrated at the individual laminations that result from material impurities, eventually leading to tearing (Still 2004).
2.3 Welding Process
Studies have shown that the method of welding has a significant impact on development of lamellar tearing. Typical welding processes include flux cored arc (FCA), gas metal arc (GMA), shielded metal arc (SMA), and submerged arc welding. Of these, the GMA and submerged arc methods are the most resistant to lamellar tearing. The electrodes used in FCA and SMA welding contain hydrogen, which contributes to hydrogen cracking effects and reduces lamellar tearing resistance. Other factors that can affect tearing include heat input, preheat, filler material, and bead sequence. (McEnerney 1978, p 12)
The weld metal properties have a significant impact on tearing susceptibility. Weld metals with significantly higher strength than the base steel absorb a smaller percentage of the thermal contraction strains, putting more strain on the base. Softer weld metals reduce the risk of lamellar tearing. (McEnerney 1978, 12) Steps can be taken during welding to reduce the risk of lamellar tearing; these are discussed in the Prevention and Modern Detailing section. (Kaufmann 1981, p 49)
3 Detection Methods
Inspection and testing are an integral part of the welding process. Lamellar tearing often manifests below the material surface on a microscopic level, which means that beyond visual inspection, further forms of non-destructive and destructive testing are often necessary to verify weld integrity. The first line of defense against lamellar tearing is material susceptibility testing during the manufacturing process, as specified in ASTM 770. Destructive testing of sample specimens often accompanies the manufacturing or shop welding processes.
In the field, testing engineers employ non-destructive tests such as visual inspection as well as dye penetrant, magnetic particle, and ultrasonic testing. Ultrasonic testing is the only non-destructive test that can detect internal ruptures at any great distance below the material surface. This fact, and its ease of use has made it the most common field test for lamellar tearing. If a lamellar tearing failure is suspected to exist, surface methods such as dye penetrant testing or magnetic particle inspection are quick ways to locate cracks.
Figure 6: Standard test specimens per ASTM A770 testing procedures. Reprinted, with permission, from A770/A770M-03 Standard Specification for Through-Thickness Tension Testing of Steel Plates for Special Applications, copyright ASTM International,100 Barr Harbor Drive, West Conshohocken, PA 19428.
3.1 Destructive Testing Methods
Destructive testing is used either for material susceptibility testing or forensic testing. The most common destructive test is the through-thickness, short transverse reduction in area test governed by ASTM A770/A770M. In forensic applications, additional microscopic analysis of material composition and fracture surfaces are necessary.
ASTM A770/A770M provides a method for conducting a through-thickness tension test to measure a plate's resistance to lamellar tearing. The standard requires that two specimens be taken from different ends of a plate at the center of the plate width. "Prolongations," or extensions, can then be welded to either side of the plate specimen to facilitate the tension test as shown in Figure 6. To meet the standard, each specimen must exhibit sufficient ductility with a reduction of area greater than 20%. The failure surface in a transverse tensile test often exhibits lamellar tearing characteristics rather than the cup-cone failure expected in a longitudinal tensile test. (ASTM 2003, pp. 1-2, 4)
ASTM A770/A770M recognizes the potential for data scattering due to the non-uniform distribution of inclusions, and that two test specimens is insufficient data to characterize the behavior of the entire plate. (ASTM 2003, pp. 1-2, 4) It is important to realize that even a plate that meets the ASTM A770/A770M standard can experience lamellar tearing failures in the field.
3.2 Non-Destructive Testing Methods
Visual Inspection: If the design professional has specified a weld sequence, a visual inspector will often be present during welding to ensure conformance. Unfortunately, in the case of lamellar tearing, visual inspection is often insufficient to detect a failure except in extreme cases. Visual inspection is not a reliable detection method for lamellar tearing failures. (Cooper 1985, p 319)
Magnetic Particle Testing (MT): MP uses a magnetizing current in the welded connection to create a magnetic field within the element. Cracks and inclusions interrupt the field and the dry magnetic powder dusting applied to the surface gathers at the interruptions. This method can indicate surface cracks (which may have been too tight/small for dye to enter in penetrant testing) as well as internal cracks within 0.1 inches of the surface. (Moyer 2012)
Dye Penetrant Testing (PT): PT uses a liquid dye that "penetrates" into surface cracks by capillary action. After the surface is cleaned, the dye is either visible to the naked eye, visible under ultraviolet light, or visible after chemical development depending on the dye system used. PT is an inexpensive method for detecting surface cracks; however, the test can slow welding if conducted properly after every pass. (Cooper 1985, p 319)
Ultrasonic Testing (UT): UT uses high frequency sound waves to detect material interruptions. The waves are passed through the material at an angle, and an oscilliscope detects the resulting reflection patterns. A technician then compares the pattern results against those from a standard test specimen. UT is not as "black and white" as the surface detection methods, and skilled technicians are required to correctly interpret the test data. An unskilled ultrasonic technician can interpret false-positives or overlook important failures. (Cooper 1985, p 319)
Two types of ultrasonic measurements can be used to reveal inclusions; the tests measure either energy levels of waves reflected off inclusions or the material's absorptive properties using full-spectrum frequency analysis. (Cooper 1985, p 319) ASTM A435 testing standards govern the method. The test has become one of the most common in the industry.
4 Initial Cases (1970s)
Lamellar tearing failure cases that surfaced in the early 1970's are noteworthy for the industry response that they instigated, and not because they were the first, most costly, or most dangerous (in fact no lives have been lost due to lamellar tearing failures). This section provides details on four widely documented and published instances of lamellar tearing from that era, including El Paso Civic Center, Grand Coulee Transmission Line Towers, Atlantic Richfield Towers, and King Street Bridge.
During construction of the civic center theater of El Paso, Texas in 1972, hidden cracks developed in the welded joints of the roof compression ring (Ross 1984, p. 256). The 138-foot-diameter compression ring consisted of a 2' x 4.6' box girder and was located between the two intersecting crescents making up the theater (Concrete 1974). The girder was originally designed as twenty-five separate sections joined together with heavy, full-penetration welds connecting the flange plates, which ranged from 2 1/2 to 3 inches thick, to the 1 inch diaphragm plates. However, the design was changed to use fillet welds along the top and bottom of the ring girder. The use of these fillet welds, in turn, eliminated the full-penetration welds that were to be used in the side plates of the girder (Kaminetzky 1991, p. 244). The altered welding configuration created joints in which the flange plates were particularly susceptible to lamellar tearing.
In Figure 7 below, the sketch is showing an enlarged plan view of the 4.6' wide box girder, looking down at a joint joining two sections of the ring at a diaphragm plate. The diaphragm plate is shown extending beyond the flange plates and the fillet welds joining the diaphragm and flange plates are depicted with dark shading. The steel tears appeared beneath the fillet welds in the heavy flange plates. Figure 8 shows the repairs that were made at the girder connection joints. Part of the diaphragm plate and portions of the flange plates affected by the original fillet welds have been cut out and replaced with weld material fill, depicted with gray shading. Both of the following sketches were based on the photos of "Figure 6-4" on page 259 of Construction Disasters: Design Failures, Causes, and Prevention by Steven S. Ross.
The civic center was originally estimated to cost $5 million and was expected to be completed by March 1973. After discovery of the steel tears, El Paso's public works department had all of the welded joints in the cable roof compression girder inspected ultrasonically at a cost of $150,000. Robert E. McKee, Inc., the general contractor, also spent nearly $150,000 on additional quality control measures. In March 1975, with the project two years overdue, McKee sued the city of El Paso for $2.3 million. A total of sixteen subcontractors as well as the architects were also sued. In response to the initial suit, the city of El Paso filed against the structural engineer, the steel supplier and the American Institute of Steel Construction (AISC) for the $2.3 million and another $247,603 (Ross 1984, pp. 260, 264).
With just two weeks until the start of the trial in February 1980, the dispute was settled with all parties sharing the blame. Almost every party involved agreed to pay into a combined $1 million which went to McKee. A portion of that sum was also distributed to the subcontractors. The amounts of the individual payments were all at least $100,000 and were contributed by the city of El Paso, the AISC, Armco Steel Corp. (steel supplier), architects Garland & Hilles, Carroll, Daeuble, DuSang & Rand and Barnard Mulville and structural engineers A. B. Peinado & Son, Inc. and Severud-Perrone-Sturm-Bandel. In addition, some of the subcontractors withdrew or decreased the amount of their individual claims (Ross 1984, pp. 264-65).
4.2 Grand Coulee Transmission-Line Towers
Author: Seth Moyer, BAE/MAE 2012
Figure 9: Aerial View of the Grand Coulee Dam (Photo Credit: Wikimedia Commons)
Near the summer of 1973, the U.S. Bureau of Reclamation was forced to halt the construction of twenty power transmission-line towers at their Third Powerplant, located at the Grand Coulee Dam in Colorado (shown in Figure 9), due to lamellar tearing. At the time, only one tower had been erected and needed to be taken down. However, twelve of the other towers also experienced cracking in their welded steel frames (Ross 1984, p. 260). The cracking was not thought to be all that significant originally until magnetic-particle testing showed that the cracking was much more severe. These findings resulted in a total redesign of all towers and the costly deconstruction of the towers that had already been built (Kaminetzky 1991, p 244).
The towers were designed with welded tubular steel columns and rectangular crossarms. Where the tapered steel crossarms connected to the columns, a rectangular box girder tied the columns to one another. The lamellar tearing originated in the diaphragm plates which extended through the cross section of the box girders and between each of the tubular columns (Ross 1984, p 260). Engineering News Record reported that the lamellar tearing issues were unexpected and all the more shocking since the plates ranged in thickness from 3/8 to 1 1/16 inches. These plates were significantly thinner than the 1 1/2 inches and heavier sections that are typically associated with tearing susceptibility. Prior to this, tearing had not been recognized in plates so thin. The arrangement of the joints in the towers led to very heavily restrained conditions, eventually initiating the cracks (Kaminetzky 1991, p. 244).
In the redesign of the transmission towers, the connections were configured in a way that the plates were no longer running through the cross section of the box girders. The original towers, which were designed in 1970, cost the U.S. Bureau of Reclamation $950,000 and another $952,000 was spent to reconstruct them in accordance with the new specifications (Ross 1984, p. 260).
The 52-story twin towers in Los Angeles, California (see Figure 10) tallied $400,000 worth of repairs due, partly, to the effects of lamellar tearing (Ross 1984, p. 256). Delayed cracks were detected at the column spandrel connections and nearly ten percent of the spandrel/beam flange connections were damaged. The cracks developed in the plate base metal as well as within the welds themselves. Both the column and spandrel members that experienced material failures were made up of very heavy elements, all 1 1/2 inches or thicker. However, lamellar tearing was not the only kind of cracking failure that occurred in the steel members. In an extremely unusual manner, the ASTM A-36 mild steel that was used in the fabrication of the column and spandrel members behaved in a brittle, rather than ductile, way (Kaminetzky 1991, p. 244).
The all-welded steel girder King Street Bridge experienced similar brittle behavior to that of the Atlantic Richfield Towers in its BS 968 (British Steel Specification) members. The bridge, which spanned the Yarra River in Melbourne, Australia, collapsed on July 10, 1962, when its four girders failed due to brittle fracturing. The high tensile strength steel had a high carbon content, which led to lamellar tearing and overall weakness in and around the heat-affected zone. The low temperatures (down to 14 degrees F) exacerbated the brittle behavior and allowed for crack propagation once stresses had been concentrated at non-metallic inclusions and tearing was initiated. In the end, the high carbon steel (not appropriate for welding in this capacity), poor detailing and low temperatures combined to cause the failure of the bridge (Bridge 2011).
5 Industry Response
Section Source: (Ross 1984, pp. 255-266) unless otherwise noted
5.1 Industry Climate, 1970
The first article about the El Paso Civic Center failure, published in Engineering News-Record (ENR) on July 27, 1972, helped bring lamellar tearing issues to light and spurred an open dialogue between designers, owners, builders, and consultants. After the ENR publication, consultants and testing engineers began to bring forward further instances of confirmed failures, including Chicago's John Hancock Center and San Francisco's Alcoa Bank of California, Bechtel Building, Security Pacific Building, and Embarcadero Regency Hyatt House hotel, among others. A New York-based testing engineer reported that most structures in New York City built between 1960 and 1973 and containing large plates with welded connections had lamellar tearing problems. The majority of affected buildings were over 20 stories tall and required extensive, expensive testing and repair, precipitating legal actions.
The financial and legal implications of finding a solution to lamellar tearing caused finger pointing between designers, builders, and manufacturers, all seeking to avoid culpability. As Tom Diamond, an attorney in the El Paso Civic Center legal proceedings, recalls, "There are four different ways of looking at lamellar tearing and who is responsible for it. Is it a design problem, construction problem, materials problem, or nobody's specific problem that the owner should pay for?" Eventually, pressure for change came from the insurance industry, which was bearing the largest portion of the financial burden.
The American Institute of Steel Construction (AISC), American Iron and Steel Institute (AISI), and American Welding Society (AWS), as industry-leading professional organizations, became the major players working toward finding solutions to lamellar tearing. The American Society for Testing and Materials (ASTM) also became involved, since no standard methods existed at the time for testing steel's through-thickness strength. Academic institutions such as Lehigh University participated in the process by conducting metallurgical studies. Steel manufacturers began conducting in-house research on the issue.
5.2 AISC Takes a Careful Stance
At the AISC national conference in May, 1973, lamellar tearing became the topic of focus. The event was the first time AISC had recognized the lamellar tearing as an industry-wide problem. Previously, the issue had been minimized, and according to William Milek, AISC director of engineering and research, it was the organization's understanding that, "Designers should recognize, with their academic training, that differences in strength exist in steel."
In September, 1973 AISC published its first official report on lamellar tearing entitled "Commentary on Highly Restrained Welded Connections." The document was 9 pages long and contained a list of 13 preventative design recommendations. Understanding the legal gravity of publishing a report on such a volatile issue, AISC prequalified the information with the following disclaimer:
"While every precaution has been taken to insure the information is as accurate as possible, the American Institute of Steel Construction discaims responsibility for the authenticity of the information herein and does not guarantee that in specific applications any of the material contained in this paper will prevent lamellar tears."
In effect, the document placed much of the responsibility for prevention of lamellar tearing on the knowledge and expertise of the design professional. The 13 recommendations even outline the specification of inspections as a designer's role (bullet 13).
The 13 design recommendations are reproduced below:
Select electrodes which deposit weld metal with the lowest yield strength adequate to carry design loads.
Design connections to minimize accumulation and concentration of strains resulting from weld metal contraction in localized areas.
Where possible, arrange connections so as to avoid welded joints which induce through-thickness strains due to weld shrinkage.
Make connections with welds having the minimum throat dimension required to carry the stresses and having a minimum practical volume of weld metal.
Design corner joints with proper consideration of edge preparation
Consideration of the use of soft wire cushions or other means to permit contraction of weld metal without producing high concentration of stresses may be helpful in difficult situations.
Whenever practical, completely weld sub-assemblies prior to final assembly of the connection.
Do not arbitrarily use prequalified joints without considering restraints provided by the complete connection assembly.
The designer should fully research and utilize available experience and knowledge on specific design details that might be potential sources of lamellar tearing.
Do not use larger welds than are necessary to transfer calculated forces.
Do not specify stiffeners when they are not required by design calculations.
Before making repairs to highly restrained connections, determine whether the repair will be more detrimental than the original cause for repair.
The designer should selectively specify ultrasonic inspection after fabrication and/or erection of those specific highly restrained welded connections critical to structural integrity that he considers to be subject to lamellar tearing.
5.3 Designers Respond
Following the failure cases published in the early 1970's, it was widely acknowledged that steel design had outpaced steel research. Pushing the boundaries of steel capabilities, designers had specified unprecedented welded connections that resulted in lamellar failures. As Ralph Webb, a U.S. Steel metallurgist stated, "Designers have gotten ahead of materials research. Some of the complex welded connections came up unexpectedly."
At the time of the initial failures, many designers were unaware of the danger of lamellar tearing or the directional properties of steel. Hanskarl Bandel, a principal with consulting engineer Severud-Perrone-Sturm-Conlin-Bandel in New York, stated, "There are no publications by the steel institutes or in steel design booklets indicating the different strengths in steel's different directions." Similarly H. G. Arthur, director of design and construction with the Bureau of Reclamation responsible for the Grand Coulee transmission-line project, said he was not aware of lamellar tearing when designing the towers. Once the problem was more widely known, designers began calling for design tables with both longitudinal and transverse strength values.
With little information on the causes of lamellar tears, designers adapted in different ways: by avoiding welded connections, conducting tests on the structure, or specifying low-impurity steel. A primary solution was to avoid welded, through-thickness-loaded connections entirely. One designer, Joseph Colaco, the director of design and computer operations at Ellison Engineers Inc., specified bolted connections for an entire 32 story structure in Detroit. Another method was to conduct extensive testing of each welded connection. Over 775,000 tests were conducted on the 110 story Sears Tower (now Willis Tower) in Chicago. Leslie E. Robertson, then principal at Skilling, Helle, Christiansen and Robertson, found an economic balance between both methods, saying, "We try to organize plates so that their through thicknesses are used minimally. We also specify ultrasonic scanning of critical pieces at the shop." Some designers chose to continue using standard welded connections, but specified specially-manufactured, low-impurity steels. Designers of a 56 story office building in Dallas specified vacuum-degassed steel for critical welded connections. Their approach raised the cost of the project by only 3.5%.
5.4 Improvement in Material Technology
Studies conducted in the 1970s concluded that inclusion content was the major factor to be eliminated in order to control lamellar tearing. A low-impurity, low-sulfur steel became available in July, 1976 from Nippon Steel Corp (Ross 1984, p ). With the introduction of low sulfur steel, widespread lamellar tearing was all but eliminated. These "clean" steels have sulfur levels below 0.005% (Still 2004). Construction projects before the mid-1970s used commercially produced steel in which sulfur levels could be as high as 0.04% (Still 1995, p. 447) (Moyer 2012)
6 Prevention and Modern Detailing
Beyond the 13 recommendations for lamellar tearing prevention published by AISC (1973), designers and fabricators can take additional steps during the design and welding processes to minimize the risk of lamellar tearing.
6.1 Design-Based Prevention
During the design phase, the design professional has the opportunity to improve joint detailing or to specify high-performance steels.
Figure 11: Susceptible and Improved Weld Details, From “Commentary on Highly Restrained Welded Connections”, AISC Engineering Journal, 3rd Quarter, 1973. Reprinted with permission from AISC.
Joint detailing has a significant impact on a connection's resistance to lamellar tearing. The standard details shown in Figure 11 depict susceptible and improved welding details. The improved details utilize the weld groove to reduce the susceptible through-thickness dimension. Note that the improved details have weld metal spanning the susceptible layers rather than acting perpendicular to them. A similar figure can be found in the Steel Construction Manual (14th Edition) on page 8-22. (Steel 2011)
Steel in suspect connections should meet the ASTM A770 standard with a STRA of greater than 20%, and have sulfur levels below 0.005%. (TWI 2000) The design professional should be aware that such steels, although resistant to tearing, are not fail-proof; and that other factors, such as proper welding technique, are just as important to the integrity of the connection.
6.2 Fabrication-Based Prevention
The welding process has a large impact on lamellar tearing susceptibility. To reduce the risk of hydrogen cracking, vacuum packed electrodes should be used, and the electrodes should be kept in the manufacturer's packaging until use. Rebaked or old electrodes should not be used for critical connections. (Still 1995, p. 448)
In addition to good welding practices, several methods have been shown to decrease the risk of lamellar tearing, including buttering, peening, and pre-heating.
Buttering is the process of replacing a portion of the base metal with weld material or adding a buffer-layer to the material surface. The buffer weld material acts as the HAZ during welding and reduces thermal strains within the base material. Replacement and surface buttering have been shown to be equally effective at suppressing lamellar tearing. (Kaufmann 1981, p. 46) The buttering layer dimensions are recommended to extend at least 15 to 25 mm beyond each weld toe and have 5 to 10 mm of depth. The strength of the buttering metal should not exceed that of the base material.
Occasionally, surface buttering has resulted directly in lamellar tearing; therefore it is important to use proper technique and appropriate layer thickness to achieve effective suppression. (McEnerney 1978, p.14)
Peening is the use of hammers or metal shot to mechanically impact the weld material and reduce residual weld stresses. Peening is mainly used on the weld passes closest to the suspect base material, as shown in Figure 12.b. Fabricators report mixed results on the effectiveness of peening to prevent lamellar tearing. In one survey, only 1 out of 11 fabricators reported improved results after peening; however, the method has not been shown to negatively impact lamellar tearing. (McEnerney 1978, p.15)
Figure 12: Buttering and Peening details, From “Commentary on Highly Restrained Welded Connections”, AISC Engineering Journal, 3rd Quarter, 1973. Reprinted with permission from AISC.
Pre-heating, the process of raising the base metal's temperature prior to welding, has also shown mixed results in preventing lamellar tearing. In large components where pre-heating should be ideal, the method has been reported to increase overall contraction strains and cause further decohesions in the base metal. Although small, the decohesions can interfere with ultrasonic testing. (McEnerney 1978, p. 15) Preheating is beneficial in reducing the effects of hydrogen during welding, but the method must be conducted in a way that does not increase overall strains. Peening can be used in combination with pre-heating to relieve the additional contraction strains. (Kaufmann 1981, p. 49)
7 Citations
Abyazi, A., & Ebrahimi, A. R. Characterization of inclusions causing lamellar tearing in S355N. (2009) Sahand University of Technology, Tabriz, Iran. A discussion of inclusion properties and characteristics of inclusion elongation in rolled steel.
American Institute of Steel Construction. (1973). Commentary on Highly Restrained Welded Connections, Engineering Journal, Vol. 10, No. 3, (3rd Qtr.), AISC, Chicago, IL. An example of AISC's response to lamellar failure in welded connections. A resource for figures, causes, and prevention of lamellar failure.
ASTM A770/A770M - 03. (2012). "Standard Specification for Through-Thickness Tension Testing of Steel Plates for Special Applications." ASTM International. pp. 1-2, 4. This is the standard specification for through-thickness tension testing of steel plate elements published by the American Society for Testing and Materials.
Bailey, N. (1973). Welding Steels without Hydrogen Cracking. 2nd Ed. Cambridge, England: Abington Pub. pp. 3-14. This book covers, in detail, hydrogen cracking, which is the most common problem in welding steel structures..
"Bridge Collapse Cases/King Street Bridge." (September 5, 2011). MatDL: Failure Cases Wiki. <http://matdl.org/failurecases/Bridge_Collapse_Cases/King_Street_Bridge>(December 18, 2012). This page of the failure case studies site provides an overview of the King Street Bridge collapse in Melbourne, Australia.
Concrete Construction Staff. (April 1, 1974). "A Problem in Structural Steel." Concrete Construction. <http://www.concreteconstruction.net/structural-steel/a-problem-in-structural-steel.aspx> (October 4, 2012). A brief article about the issue of lamellar tearing and its construction and cost impacts on both the El Paso Civic Center and the Atlantic Richfield Towers.
Cooper, S. E., & Chen, A. C. (1985). Designing steel structures: Methods and cases. Englewood Cliffs, N.J: Prentice-Hall. pp. 316-20 A short discussion of restraint in welded connections and types of detection methods for lamellar tearing failures.
Kaminetzky, Dov. (1991). Design and Construction Failures: Lessons from Forensic Investigations. New York: McGraw-Hill. pp. 243-44. Various lamellar tearing case studies, with details particularly on the El Paso and Richfield Tower occurrences.
Kaufmann, E.J, Pense, A. W., & Stout, R. D. An Evaluation of Factors Significant to Lamellar Tearing (1981). Supplement to the Welding Journal, March 1981. A look at various factors that affect lamellar tearing, including inclusions, weld type, and heat.
McEnerney, J. W. (March 1978). "Assessment of Lamellar Tearing." Oak Ridge National Laboratory. [Wiki author's recommendation as a first-source for further research.] A report prepared for the U.S. Nuclear Regulatory Commission that contains everything from descriptions of failure mechanisms to ASME code requirements. An excellent resource for all things lamellar tearing.
Moyer, Seth M. (2012) "Lamellar Tearing Overview and Failures Cases." The Pennsylvania State University. Original wiki-site posting. Author of several sections in this wiki-article, most notably the failures case overviews, as well as discussions on hydrogen cracking, magnetic particle testing, and improvements in material technology.
Ross, Steven S. (1984). Construction Disasters: Design Failures, Causes, and Prevention. New York: McGraw-Hill. pp. 255-66. A detailed description of lamellar tearing causes and solutions as well as a sampling of cases from the early 1970s. This resource also contains dialogue on the political climate surrounding lamellar tearing at the point of its discovery and the AISC response.
Steel Construction Manual. (2011). 14th Ed. Chicago: American Institute of Steel Construction. pp. 8.4-8.7, 8.22. A reference for current design standards, figures, and commentary for mitigation of lamellar tearing causes in steel structures.
Still, J. R. (January 2004). "Understanding Hydrogen Failures of Ferritic Welds." Welding Journal, American Welding Society, Vol. 83, No. 1, pp. 26-29. This article addresses how to reduce the effects of hydrogen-induced cracking and corrosion as well as the common failures associated with hydrogen, including lamellar tearing. It can be found in electronic format here: http://www.aws.org/wj/jan04/still_feature.html.
Still, J. R. (November-December 1995). "Lamellar Tearing - A Ghost from the Past." Welding and Metal Fabrication, Vol. 63, No. 10, pp. 445-48. This article covers the causes of the tearing that are common with steel used in the 1970s while outlining various tests used to assess the likelihood of material failure.
ASTM A435/A435M - 90 (2012). "Standard Specification for Straight-Beam Ultrasonic Examination of Steel Plates." ASTM International. This standard governs ultrasonic testing of steel plates for lamellar tearing.
Evrard, M., Dubresson, J., & Le Penven, Y. (December 1975). "Ultrasonics in the determination of susceptibility of rolled products to lamellar tearing." Non-Destructive Testing, pp 299-303. An article on the refinement of ultrasonic testing to better detect lamellar tearing specifically.
Farrar, J. C. M. (1974). "Inclusions and Susceptibility to Lamellar Tearing of Welded Structural Steels." Welding Research Supplement, Welding Journal. August, 1974. pp 321-331. An investigation of the impact of inclusions on lamellar tearing susceptibility.
Holby, E., and Smith, J. F. (1980). "Lamellar tearing - the problem nobody seems to want to talk about." Welding Journal, 59, 37-44. A description of lamellar tearing and commentary on the political factors surrounding lamellar failures in the mid-1900s.
Maranian, Peter. (2009). Examples of Major Historical Events. Reducing Brittle and Fatigue Failures in Steel Structures: pp. 23. A short description of the Atlantic Richfield tower case, and discussion on the Alexander Keilland offshore platform failure (which was determined to have had lamellar tearing involved as partially responsible for the failure along with several other factors).
Steel Construction Manual. (2011). 14th Ed. Chicago: American Institute of Steel Construction. pp. 2-42. Section 2-42 in the Steel Manual discusses lamellar tearing prevention and lists several studies that help shaped the current specifications.
Verdeja, J. I., Asensio, J. and Pero-Sanz, J. A. (January 2003). "Texture, Formability, Lamellar Tearing and HIC Susceptibility of Ferritic and Low-Carbon HSLA Steels." Materials Characterization, Vol. 50, No. 1, pp. 81-86. A discussion of the influence of the texture and grain orientation of ferritic and low-carbon steels on the mechanical material properties, such as ductility and toughness. <http://www.sciencedirect.com/science/article/pii/S1044580303001062>
Keywords: lamellar, lamellar tearing, steel plates, crack, cracking, welds, welding, metallurgy, brittle fracture, inclusions, El Paso Civic Center, Atlantic Richfield, King Street, through thickness, transverse thickness, hot rolled, HAZ, heat affected zone, magnetic particle testing, dye penetrant testing, ultrasonic testing, material susceptibility, AISC, hot potato, sulfur, sulfide, oxide, silicate, detailing
Christopher J. Brandmeier, BAE/MAE, The Pennsylvania State University, 2014
Table of Contents
In the 1970's as the AEC industry moved toward building larger, more complex steel structures, lamellar tearing became an important (and costly) issue for structural designers to face. Although lamellar tearing itself was not a new issue, the costly 1970's failures served as an impetus for industry changes to prevent future expensive or dangerous failures. This article contains brief discussions of the first major lamellar tearing failure cases in the country, including the El Paso Civic Center, Grand Coulee Transmission Line Towers, Atlantic Richfield Towers, and King Street Bridge.
Advances in material technology, current AISC connection standards, and conscientious design have helped to reduce and eliminate the threat of lamellar tearing in modern steel construction.
1 Failure Mode
1.1 Failure Mechanism
Lamellar tearing is characterized by a combination of brittle ruptures parallel, and ductile shear failures perpendicular (or nearly perpendicular) to the plane of rolling. The two conditions form stepped cracks similar to that shown in Figure 1. These cracks commonly occur within 5 mm of the welding heat affected zone (HAZ) where thermal strain differences are most acute, but can occur well into the base metal. (Ross 1984, p. 265) Interestingly, from a fracture mechanics perspective lamellar tearing is a "sub-critical growth phase," meaning that the cause of member failure is never "lamellar tearing", but rather "brittle fracture" resulting from a net reduced tensile area. (McEnerney 1978, 7)
Lamellar tearing occurs in three phases: terrace formation, terrace linkage, and ductile shear. The three phases occur after welding during the cooling process as weld shrinkage strains increase. (American, 1973, p. 63) They are shown graphically in Figure 2.
Terrace Formation occurs as singular, non-metallic impurities begin to fracture or decohere, forming voids between the steel matrix and the impurity. These voids are called "terraces" because they usually form in clusters parallel to the rolling plane. (Samuels 1999, pp. 108-109) Although this is the most common initiation mode, several others have been reported including:
- splitting along ferrite bands
- liquation in reheated HAZ
- intergranular cracking
- strain aging
Figure 2: The three phases of lamellar tearing. C. Brandmeier, adapted from McEnerney.
(McEnerney 1978, 6)Terrace Linkage involves the growth of the terraces to form common levels. The void levels remain separate, but converge to the point of near vertical material boundaries. (Samuels 1999, pp. 108-109) The mechanisms of terrace linkage are widely varied and include:
- necking (common)
- microvoid coalescence (common)
- "zig-zag" tearing
- intergranular cracking
- quasi-cleavage
- cleavage
(McEnerney 1978, 6)Ductile Shear joins the varied terrace levels with nearly vertical, ductile shear failures. (McEnerney 1978, 6)
1.2 Physical Characteristics
Because the compressed impurities take on a grained, planar nature during the rolling process, the appearance of the failed surface has been described as "woody" or "fibrous" (TWI 2000). Lamellar tears are usually internal to the member, although they are occasionally visible at weld toes, plate edges, or on the welded surface. (McEnerney 1978, 5) Figures 3 and 4 show specimens from lamellar tearing failures.1.3 Important Distinction: Localized Strain vs. Global Stress
It is important to note that lamellar tearing occurs from localized shrinkage strains as a direct result of the high heat of welding, and not typically from global member loads. Very rarely do service forces reach a magnitude that can internally fracture steel. Historically, significant failures from lamellar tearing are the product of loads applied to members that have already experienced internal fractures at the time of fabrication. This idea is reinforced by the quote from an AISC journal article below:"Properly designed restrained connections which transfer frame bending moments from one structural member to another do not generally induce localized strains responsible for lamellar tearing. The restraint that is of concern is internal restraint (within a connection made up of several joints) which inhibits small total--but large unit--localized strains resulting from weld shrinkage. ... There is often a lack of recognition that under certain conditions localized strain is significantly more important than stress." (American 1973, p. 62)
2 Causing Factors
The development of lamellar tearing at the time of welding is influenced by three major factors: percentage of sulfurous inclusions, presence of hydrogen sources, and method of welding.
2.1 Inclusions in the Hot Rolling Process
Small non-metallic inclusions such as silicates, sulfides, or oxides from deoxidation are unintentionally ingrained in steel during the hot rolling process. These materials have different properties than the surrounding steel and create clusters of low ductility susceptible to internal brittle failure or decohesion. If hot rolling is conducted at insufficient temperatures, sulfurous materials (typically manganese silicates) will exhibit higher plasticity than the surrounding steel and elongate in the direction of rolling. (Samuels 1999, p 108) Elongation of the inclusions creates a greater chance for clusters to form and develop a susceptible failure plane. Inclusions within layers of pearlite are particularly liable to elongation, although there is no established relationship directly linking pearlite banding and lamellar tearing (American, 1973, p 64). Often the rolling process will compress pearlite and ferrite into grained layers running parallel to the direction of rolling. Inclusions between the layers "pancake" and occupy a larger surface percentage, reducing the strength of the bonds. (Abyazi 2009, p 3)Short Transverse Reduction in Area (STRA) is a good indication of inclusion-based tearing susceptibility. Steels below 10-15% STRA are highly susceptible to lamellar tearing, while steels above 20% are almost completely resistant. Figure 4 shows the relationship between percentage of included sulfide content and STRA. Generally, steels with less than 0.005% sulfur content are desirable, and manufacturers can test plates to ensure an STRA value of greater than 20% for critical connections. (TWI 2000)
Three types of sulfide inclusions can occur. Known simply as Type I, II, and III, their presence is determined by the type of deoxidation process used during manufacturing.
Type I inclusions result from the addition of silicon for partial deoxidation in semikilled steels. The inclusions are randomly dispersed, globular, and deform minimally during rolling. They range in size from 2 to 200 µm and aspect ratios of about 2 to 20. Of the three types, Type I inclusions pose the least threat to lamellar tearing because of their random disbursement and minimal elongation. (Samuels 1999, pp. 109-110)
Type II inclusions form from the addition of aluminum or other strong deoxidizers in fully killed steels. They have a rod-like shape, and a higher plasticity at rolling temperatures than Type I inclusions. The rods are prone to align in the direction of rolling and elongate to maximum lengths of several millimeters. Aspect ratios range between 100 to 1000. Type II sulfides usually develop in high-strength, fracture-tough grades, which are fully killed. (Samuels 1999, pp. 109-110)
Type III inclusions also occur in fully killed steels when carbon, silicon, phosphorus, and aluminum are all present in appropriate ratios. The inclusions are octahedral, and have a higher plasticity at rolling temperatures than either Type I or Type II. They exhibit similar characteristics to Type II inclusions after rolling and are nearly indistinguishable without examining a parent sample. (Samuels 1999, pp.109-110)
2.2 Hydrogen Induced Cracking
Author: Seth Moyer, BAE/MAE 2012Hydrogen-induced cracking in the heat-affected zone (HAZ) of the base or parent metal is another leading factor responsible for occurrences of lamellar tearing in modern steel construction...A major source of hydrogen during the welding process is the moisture that is contained within the electrode coating itself. Other sources of hydrogen in the weld material include moisture in the flux, hydrogenous compounds in the coating and/or flux and dirt, oil, grease and rust on the welding wires. For the weld base material, hydrogen sources include surface dirt, oil, grease and rust, degreasing fluids for cleaning the surface to be welded and hydrogen in the parent steel. Ambient atmosphere moisture can also be a minor source of hydrogen (Bailey 1973, pp. 5-6).
Hydrogen is absorbed into the weld bead from the arc atmosphere during application of the weld. A lot of this hydrogen diffuses out of the solidified weld and some that escapes diffuses into the base metal (Bailey 1973, p. 5). The diffused hydrogen acts in a similar manner to its behavior in a sour (H2S) environment. In a sour environment, hydrogen sulfide dissociates into atomic hydrogen and works into the weld and base materials. The hydrogen atoms reform into molecular hydrogen when they cross non-metallic inclusions. This process can eventually cause a buildup of the molecular hydrogen which builds pressure within the inclusion. This internal pressure forms where the shrinkage stresses have also concentrated at the individual laminations that result from material impurities, eventually leading to tearing (Still 2004).
2.3 Welding Process
Studies have shown that the method of welding has a significant impact on development of lamellar tearing. Typical welding processes include flux cored arc (FCA), gas metal arc (GMA), shielded metal arc (SMA), and submerged arc welding. Of these, the GMA and submerged arc methods are the most resistant to lamellar tearing. The electrodes used in FCA and SMA welding contain hydrogen, which contributes to hydrogen cracking effects and reduces lamellar tearing resistance. Other factors that can affect tearing include heat input, preheat, filler material, and bead sequence. (McEnerney 1978, p 12)The weld metal properties have a significant impact on tearing susceptibility. Weld metals with significantly higher strength than the base steel absorb a smaller percentage of the thermal contraction strains, putting more strain on the base. Softer weld metals reduce the risk of lamellar tearing. (McEnerney 1978, 12) Steps can be taken during welding to reduce the risk of lamellar tearing; these are discussed in the Prevention and Modern Detailing section. (Kaufmann 1981, p 49)
3 Detection Methods
Inspection and testing are an integral part of the welding process. Lamellar tearing often manifests below the material surface on a microscopic level, which means that beyond visual inspection, further forms of non-destructive and destructive testing are often necessary to verify weld integrity. The first line of defense against lamellar tearing is material susceptibility testing during the manufacturing process, as specified in ASTM 770. Destructive testing of sample specimens often accompanies the manufacturing or shop welding processes.
In the field, testing engineers employ non-destructive tests such as visual inspection as well as dye penetrant, magnetic particle, and ultrasonic testing. Ultrasonic testing is the only non-destructive test that can detect internal ruptures at any great distance below the material surface. This fact, and its ease of use has made it the most common field test for lamellar tearing. If a lamellar tearing failure is suspected to exist, surface methods such as dye penetrant testing or magnetic particle inspection are quick ways to locate cracks.
3.1 Destructive Testing Methods
Destructive testing is used either for material susceptibility testing or forensic testing. The most common destructive test is the through-thickness, short transverse reduction in area test governed by ASTM A770/A770M. In forensic applications, additional microscopic analysis of material composition and fracture surfaces are necessary.ASTM A770/A770M provides a method for conducting a through-thickness tension test to measure a plate's resistance to lamellar tearing. The standard requires that two specimens be taken from different ends of a plate at the center of the plate width. "Prolongations," or extensions, can then be welded to either side of the plate specimen to facilitate the tension test as shown in Figure 6. To meet the standard, each specimen must exhibit sufficient ductility with a reduction of area greater than 20%. The failure surface in a transverse tensile test often exhibits lamellar tearing characteristics rather than the cup-cone failure expected in a longitudinal tensile test. (ASTM 2003, pp. 1-2, 4)
ASTM A770/A770M recognizes the potential for data scattering due to the non-uniform distribution of inclusions, and that two test specimens is insufficient data to characterize the behavior of the entire plate. (ASTM 2003, pp. 1-2, 4) It is important to realize that even a plate that meets the ASTM A770/A770M standard can experience lamellar tearing failures in the field.
3.2 Non-Destructive Testing Methods
Visual Inspection: If the design professional has specified a weld sequence, a visual inspector will often be present during welding to ensure conformance. Unfortunately, in the case of lamellar tearing, visual inspection is often insufficient to detect a failure except in extreme cases. Visual inspection is not a reliable detection method for lamellar tearing failures. (Cooper 1985, p 319)Magnetic Particle Testing (MT): MP uses a magnetizing current in the welded connection to create a magnetic field within the element. Cracks and inclusions interrupt the field and the dry magnetic powder dusting applied to the surface gathers at the interruptions. This method can indicate surface cracks (which may have been too tight/small for dye to enter in penetrant testing) as well as internal cracks within 0.1 inches of the surface. (Moyer 2012)
Dye Penetrant Testing (PT): PT uses a liquid dye that "penetrates" into surface cracks by capillary action. After the surface is cleaned, the dye is either visible to the naked eye, visible under ultraviolet light, or visible after chemical development depending on the dye system used. PT is an inexpensive method for detecting surface cracks; however, the test can slow welding if conducted properly after every pass. (Cooper 1985, p 319)
Ultrasonic Testing (UT): UT uses high frequency sound waves to detect material interruptions. The waves are passed through the material at an angle, and an oscilliscope detects the resulting reflection patterns. A technician then compares the pattern results against those from a standard test specimen. UT is not as "black and white" as the surface detection methods, and skilled technicians are required to correctly interpret the test data. An unskilled ultrasonic technician can interpret false-positives or overlook important failures. (Cooper 1985, p 319)
Two types of ultrasonic measurements can be used to reveal inclusions; the tests measure either energy levels of waves reflected off inclusions or the material's absorptive properties using full-spectrum frequency analysis. (Cooper 1985, p 319) ASTM A435 testing standards govern the method. The test has become one of the most common in the industry.
4 Initial Cases (1970s)
Lamellar tearing failure cases that surfaced in the early 1970's are noteworthy for the industry response that they instigated, and not because they were the first, most costly, or most dangerous (in fact no lives have been lost due to lamellar tearing failures). This section provides details on four widely documented and published instances of lamellar tearing from that era, including El Paso Civic Center, Grand Coulee Transmission Line Towers, Atlantic Richfield Towers, and King Street Bridge.
4.1 El Paso Civic Center
Author: Seth Moyer, BAE/MAE 2012Further Information: El Paso Civic Center Lamellar Tearing Failure (Jiaodong Jiang, BAE/MAE 2014)
During construction of the civic center theater of El Paso, Texas in 1972, hidden cracks developed in the welded joints of the roof compression ring (Ross 1984, p. 256). The 138-foot-diameter compression ring consisted of a 2' x 4.6' box girder and was located between the two intersecting crescents making up the theater (Concrete 1974). The girder was originally designed as twenty-five separate sections joined together with heavy, full-penetration welds connecting the flange plates, which ranged from 2 1/2 to 3 inches thick, to the 1 inch diaphragm plates. However, the design was changed to use fillet welds along the top and bottom of the ring girder. The use of these fillet welds, in turn, eliminated the full-penetration welds that were to be used in the side plates of the girder (Kaminetzky 1991, p. 244). The altered welding configuration created joints in which the flange plates were particularly susceptible to lamellar tearing.
In Figure 7 below, the sketch is showing an enlarged plan view of the 4.6' wide box girder, looking down at a joint joining two sections of the ring at a diaphragm plate. The diaphragm plate is shown extending beyond the flange plates and the fillet welds joining the diaphragm and flange plates are depicted with dark shading. The steel tears appeared beneath the fillet welds in the heavy flange plates. Figure 8 shows the repairs that were made at the girder connection joints. Part of the diaphragm plate and portions of the flange plates affected by the original fillet welds have been cut out and replaced with weld material fill, depicted with gray shading. Both of the following sketches were based on the photos of "Figure 6-4" on page 259 of Construction Disasters: Design Failures, Causes, and Prevention by Steven S. Ross.
The civic center was originally estimated to cost $5 million and was expected to be completed by March 1973. After discovery of the steel tears, El Paso's public works department had all of the welded joints in the cable roof compression girder inspected ultrasonically at a cost of $150,000. Robert E. McKee, Inc., the general contractor, also spent nearly $150,000 on additional quality control measures. In March 1975, with the project two years overdue, McKee sued the city of El Paso for $2.3 million. A total of sixteen subcontractors as well as the architects were also sued. In response to the initial suit, the city of El Paso filed against the structural engineer, the steel supplier and the American Institute of Steel Construction (AISC) for the $2.3 million and another $247,603 (Ross 1984, pp. 260, 264).
With just two weeks until the start of the trial in February 1980, the dispute was settled with all parties sharing the blame. Almost every party involved agreed to pay into a combined $1 million which went to McKee. A portion of that sum was also distributed to the subcontractors. The amounts of the individual payments were all at least $100,000 and were contributed by the city of El Paso, the AISC, Armco Steel Corp. (steel supplier), architects Garland & Hilles, Carroll, Daeuble, DuSang & Rand and Barnard Mulville and structural engineers A. B. Peinado & Son, Inc. and Severud-Perrone-Sturm-Bandel. In addition, some of the subcontractors withdrew or decreased the amount of their individual claims (Ross 1984, pp. 264-65).
4.2 Grand Coulee Transmission-Line Towers
Author: Seth Moyer, BAE/MAE 2012Near the summer of 1973, the U.S. Bureau of Reclamation was forced to halt the construction of twenty power transmission-line towers at their Third Powerplant, located at the Grand Coulee Dam in Colorado (shown in Figure 9), due to lamellar tearing. At the time, only one tower had been erected and needed to be taken down. However, twelve of the other towers also experienced cracking in their welded steel frames (Ross 1984, p. 260). The cracking was not thought to be all that significant originally until magnetic-particle testing showed that the cracking was much more severe. These findings resulted in a total redesign of all towers and the costly deconstruction of the towers that had already been built (Kaminetzky 1991, p 244).
The towers were designed with welded tubular steel columns and rectangular crossarms. Where the tapered steel crossarms connected to the columns, a rectangular box girder tied the columns to one another. The lamellar tearing originated in the diaphragm plates which extended through the cross section of the box girders and between each of the tubular columns (Ross 1984, p 260). Engineering News Record reported that the lamellar tearing issues were unexpected and all the more shocking since the plates ranged in thickness from 3/8 to 1 1/16 inches. These plates were significantly thinner than the 1 1/2 inches and heavier sections that are typically associated with tearing susceptibility. Prior to this, tearing had not been recognized in plates so thin. The arrangement of the joints in the towers led to very heavily restrained conditions, eventually initiating the cracks (Kaminetzky 1991, p. 244).
In the redesign of the transmission towers, the connections were configured in a way that the plates were no longer running through the cross section of the box girders. The original towers, which were designed in 1970, cost the U.S. Bureau of Reclamation $950,000 and another $952,000 was spent to reconstruct them in accordance with the new specifications (Ross 1984, p. 260).
4.3 Atlantic Richfield Towers
Author: Seth Moyer, BAE/MAE 2012The 52-story twin towers in Los Angeles, California (see Figure 10) tallied $400,000 worth of repairs due, partly, to the effects of lamellar tearing (Ross 1984, p. 256). Delayed cracks were detected at the column spandrel connections and nearly ten percent of the spandrel/beam flange connections were damaged. The cracks developed in the plate base metal as well as within the welds themselves. Both the column and spandrel members that experienced material failures were made up of very heavy elements, all 1 1/2 inches or thicker. However, lamellar tearing was not the only kind of cracking failure that occurred in the steel members. In an extremely unusual manner, the ASTM A-36 mild steel that was used in the fabrication of the column and spandrel members behaved in a brittle, rather than ductile, way (Kaminetzky 1991, p. 244).
4.4 King Street Bridge
Author: Seth Moyer, BAE/MAE 2012Further Information: MatDL King Street Bridge Collapse Case Study
The all-welded steel girder King Street Bridge experienced similar brittle behavior to that of the Atlantic Richfield Towers in its BS 968 (British Steel Specification) members. The bridge, which spanned the Yarra River in Melbourne, Australia, collapsed on July 10, 1962, when its four girders failed due to brittle fracturing. The high tensile strength steel had a high carbon content, which led to lamellar tearing and overall weakness in and around the heat-affected zone. The low temperatures (down to 14 degrees F) exacerbated the brittle behavior and allowed for crack propagation once stresses had been concentrated at non-metallic inclusions and tearing was initiated. In the end, the high carbon steel (not appropriate for welding in this capacity), poor detailing and low temperatures combined to cause the failure of the bridge (Bridge 2011).
5 Industry Response
Section Source: (Ross 1984, pp. 255-266) unless otherwise noted
5.1 Industry Climate, 1970
The first article about the El Paso Civic Center failure, published in Engineering News-Record (ENR) on July 27, 1972, helped bring lamellar tearing issues to light and spurred an open dialogue between designers, owners, builders, and consultants. After the ENR publication, consultants and testing engineers began to bring forward further instances of confirmed failures, including Chicago's John Hancock Center and San Francisco's Alcoa Bank of California, Bechtel Building, Security Pacific Building, and Embarcadero Regency Hyatt House hotel, among others. A New York-based testing engineer reported that most structures in New York City built between 1960 and 1973 and containing large plates with welded connections had lamellar tearing problems. The majority of affected buildings were over 20 stories tall and required extensive, expensive testing and repair, precipitating legal actions.The financial and legal implications of finding a solution to lamellar tearing caused finger pointing between designers, builders, and manufacturers, all seeking to avoid culpability. As Tom Diamond, an attorney in the El Paso Civic Center legal proceedings, recalls, "There are four different ways of looking at lamellar tearing and who is responsible for it. Is it a design problem, construction problem, materials problem, or nobody's specific problem that the owner should pay for?" Eventually, pressure for change came from the insurance industry, which was bearing the largest portion of the financial burden.
The American Institute of Steel Construction (AISC), American Iron and Steel Institute (AISI), and American Welding Society (AWS), as industry-leading professional organizations, became the major players working toward finding solutions to lamellar tearing. The American Society for Testing and Materials (ASTM) also became involved, since no standard methods existed at the time for testing steel's through-thickness strength. Academic institutions such as Lehigh University participated in the process by conducting metallurgical studies. Steel manufacturers began conducting in-house research on the issue.
5.2 AISC Takes a Careful Stance
At the AISC national conference in May, 1973, lamellar tearing became the topic of focus. The event was the first time AISC had recognized the lamellar tearing as an industry-wide problem. Previously, the issue had been minimized, and according to William Milek, AISC director of engineering and research, it was the organization's understanding that, "Designers should recognize, with their academic training, that differences in strength exist in steel."In September, 1973 AISC published its first official report on lamellar tearing entitled "Commentary on Highly Restrained Welded Connections." The document was 9 pages long and contained a list of 13 preventative design recommendations. Understanding the legal gravity of publishing a report on such a volatile issue, AISC prequalified the information with the following disclaimer:
"While every precaution has been taken to insure the information is as accurate as possible, the American Institute of Steel Construction discaims responsibility for the authenticity of the information herein and does not guarantee that in specific applications any of the material contained in this paper will prevent lamellar tears."
In effect, the document placed much of the responsibility for prevention of lamellar tearing on the knowledge and expertise of the design professional. The 13 recommendations even outline the specification of inspections as a designer's role (bullet 13).
The 13 design recommendations are reproduced below:
5.3 Designers Respond
Following the failure cases published in the early 1970's, it was widely acknowledged that steel design had outpaced steel research. Pushing the boundaries of steel capabilities, designers had specified unprecedented welded connections that resulted in lamellar failures. As Ralph Webb, a U.S. Steel metallurgist stated, "Designers have gotten ahead of materials research. Some of the complex welded connections came up unexpectedly."At the time of the initial failures, many designers were unaware of the danger of lamellar tearing or the directional properties of steel. Hanskarl Bandel, a principal with consulting engineer Severud-Perrone-Sturm-Conlin-Bandel in New York, stated, "There are no publications by the steel institutes or in steel design booklets indicating the different strengths in steel's different directions." Similarly H. G. Arthur, director of design and construction with the Bureau of Reclamation responsible for the Grand Coulee transmission-line project, said he was not aware of lamellar tearing when designing the towers. Once the problem was more widely known, designers began calling for design tables with both longitudinal and transverse strength values.
With little information on the causes of lamellar tears, designers adapted in different ways: by avoiding welded connections, conducting tests on the structure, or specifying low-impurity steel. A primary solution was to avoid welded, through-thickness-loaded connections entirely. One designer, Joseph Colaco, the director of design and computer operations at Ellison Engineers Inc., specified bolted connections for an entire 32 story structure in Detroit. Another method was to conduct extensive testing of each welded connection. Over 775,000 tests were conducted on the 110 story Sears Tower (now Willis Tower) in Chicago. Leslie E. Robertson, then principal at Skilling, Helle, Christiansen and Robertson, found an economic balance between both methods, saying, "We try to organize plates so that their through thicknesses are used minimally. We also specify ultrasonic scanning of critical pieces at the shop." Some designers chose to continue using standard welded connections, but specified specially-manufactured, low-impurity steels. Designers of a 56 story office building in Dallas specified vacuum-degassed steel for critical welded connections. Their approach raised the cost of the project by only 3.5%.
5.4 Improvement in Material Technology
Studies conducted in the 1970s concluded that inclusion content was the major factor to be eliminated in order to control lamellar tearing. A low-impurity, low-sulfur steel became available in July, 1976 from Nippon Steel Corp (Ross 1984, p ). With the introduction of low sulfur steel, widespread lamellar tearing was all but eliminated. These "clean" steels have sulfur levels below 0.005% (Still 2004). Construction projects before the mid-1970s used commercially produced steel in which sulfur levels could be as high as 0.04% (Still 1995, p. 447) (Moyer 2012)6 Prevention and Modern Detailing
Beyond the 13 recommendations for lamellar tearing prevention published by AISC (1973), designers and fabricators can take additional steps during the design and welding processes to minimize the risk of lamellar tearing.
6.1 Design-Based Prevention
During the design phase, the design professional has the opportunity to improve joint detailing or to specify high-performance steels.Joint detailing has a significant impact on a connection's resistance to lamellar tearing. The standard details shown in Figure 11 depict susceptible and improved welding details. The improved details utilize the weld groove to reduce the susceptible through-thickness dimension. Note that the improved details have weld metal spanning the susceptible layers rather than acting perpendicular to them. A similar figure can be found in the Steel Construction Manual (14th Edition) on page 8-22. (Steel 2011)
Steel in suspect connections should meet the ASTM A770 standard with a STRA of greater than 20%, and have sulfur levels below 0.005%. (TWI 2000) The design professional should be aware that such steels, although resistant to tearing, are not fail-proof; and that other factors, such as proper welding technique, are just as important to the integrity of the connection.
6.2 Fabrication-Based Prevention
The welding process has a large impact on lamellar tearing susceptibility. To reduce the risk of hydrogen cracking, vacuum packed electrodes should be used, and the electrodes should be kept in the manufacturer's packaging until use. Rebaked or old electrodes should not be used for critical connections. (Still 1995, p. 448)In addition to good welding practices, several methods have been shown to decrease the risk of lamellar tearing, including buttering, peening, and pre-heating.
Buttering is the process of replacing a portion of the base metal with weld material or adding a buffer-layer to the material surface. The buffer weld material acts as the HAZ during welding and reduces thermal strains within the base material. Replacement and surface buttering have been shown to be equally effective at suppressing lamellar tearing. (Kaufmann 1981, p. 46) The buttering layer dimensions are recommended to extend at least 15 to 25 mm beyond each weld toe and have 5 to 10 mm of depth. The strength of the buttering metal should not exceed that of the base material.
Occasionally, surface buttering has resulted directly in lamellar tearing; therefore it is important to use proper technique and appropriate layer thickness to achieve effective suppression. (McEnerney 1978, p.14)
Peening is the use of hammers or metal shot to mechanically impact the weld material and reduce residual weld stresses. Peening is mainly used on the weld passes closest to the suspect base material, as shown in Figure 12.b. Fabricators report mixed results on the effectiveness of peening to prevent lamellar tearing. In one survey, only 1 out of 11 fabricators reported improved results after peening; however, the method has not been shown to negatively impact lamellar tearing. (McEnerney 1978, p.15)
Pre-heating, the process of raising the base metal's temperature prior to welding, has also shown mixed results in preventing lamellar tearing. In large components where pre-heating should be ideal, the method has been reported to increase overall contraction strains and cause further decohesions in the base metal. Although small, the decohesions can interfere with ultrasonic testing. (McEnerney 1978, p. 15) Preheating is beneficial in reducing the effects of hydrogen during welding, but the method must be conducted in a way that does not increase overall strains. Peening can be used in combination with pre-heating to relieve the additional contraction strains. (Kaufmann 1981, p. 49)
7 Citations
Abyazi, A., & Ebrahimi, A. R. Characterization of inclusions causing lamellar tearing in S355N. (2009) Sahand University of Technology, Tabriz, Iran.
A discussion of inclusion properties and characteristics of inclusion elongation in rolled steel.
American Institute of Steel Construction. (1973). Commentary on Highly Restrained Welded Connections, Engineering Journal, Vol. 10, No. 3, (3rd Qtr.), AISC, Chicago, IL.
An example of AISC's response to lamellar failure in welded connections. A resource for figures, causes, and prevention of lamellar failure.
ASTM A770/A770M - 03. (2012). "Standard Specification for Through-Thickness Tension Testing of Steel Plates for Special Applications." ASTM International. pp. 1-2, 4.
This is the standard specification for through-thickness tension testing of steel plate elements published by the American Society for Testing and Materials.
Bailey, N. (1973). Welding Steels without Hydrogen Cracking. 2nd Ed. Cambridge, England: Abington Pub. pp. 3-14.
This book covers, in detail, hydrogen cracking, which is the most common problem in welding steel structures..
"Bridge Collapse Cases/King Street Bridge." (September 5, 2011). MatDL: Failure Cases Wiki. <http://matdl.org/failurecases/Bridge_Collapse_Cases/King_Street_Bridge> (December 18, 2012).
This page of the failure case studies site provides an overview of the King Street Bridge collapse in Melbourne, Australia.
Concrete Construction Staff. (April 1, 1974). "A Problem in Structural Steel." Concrete Construction. <http://www.concreteconstruction.net/structural-steel/a-problem-in-structural-steel.aspx> (October 4, 2012).
A brief article about the issue of lamellar tearing and its construction and cost impacts on both the El Paso Civic Center and the Atlantic Richfield Towers.
Cooper, S. E., & Chen, A. C. (1985). Designing steel structures: Methods and cases. Englewood Cliffs, N.J: Prentice-Hall. pp. 316-20
A short discussion of restraint in welded connections and types of detection methods for lamellar tearing failures.
Kaminetzky, Dov. (1991). Design and Construction Failures: Lessons from Forensic Investigations. New York: McGraw-Hill. pp. 243-44.
Various lamellar tearing case studies, with details particularly on the El Paso and Richfield Tower occurrences.
Kaufmann, E.J, Pense, A. W., & Stout, R. D. An Evaluation of Factors Significant to Lamellar Tearing (1981). Supplement to the Welding Journal, March 1981.
A look at various factors that affect lamellar tearing, including inclusions, weld type, and heat.
McEnerney, J. W. (March 1978). "Assessment of Lamellar Tearing." Oak Ridge National Laboratory.
[Wiki author's recommendation as a first-source for further research.] A report prepared for the U.S. Nuclear Regulatory Commission that contains everything from descriptions of failure mechanisms to ASME code requirements. An excellent resource for all things lamellar tearing.
Moyer, Seth M. (2012) "Lamellar Tearing Overview and Failures Cases." The Pennsylvania State University.
Original wiki-site posting. Author of several sections in this wiki-article, most notably the failures case overviews, as well as discussions on hydrogen cracking, magnetic particle testing, and improvements in material technology.
Ross, Steven S. (1984). Construction Disasters: Design Failures, Causes, and Prevention. New York: McGraw-Hill. pp. 255-66.
A detailed description of lamellar tearing causes and solutions as well as a sampling of cases from the early 1970s. This resource also contains dialogue on the political climate surrounding lamellar tearing at the point of its discovery and the AISC response.
Samuels, Leonard Ernest. (1999). Light Microscopy of Carbon Steels. ASM International, USA. p 103-110.
A perspective on lamellar tearing from a microscopic, metallurgical side. A resource for images of microscopic tears and tear samples. <http://books.google.com/books/about/Light_Microscopy_of_Carbon_Steels.html?id=rsdbs92yGs4C>
Steel Construction Manual. (2011). 14th Ed. Chicago: American Institute of Steel Construction. pp. 8.4-8.7, 8.22.
A reference for current design standards, figures, and commentary for mitigation of lamellar tearing causes in steel structures.
Still, J. R. (January 2004). "Understanding Hydrogen Failures of Ferritic Welds." Welding Journal, American Welding Society, Vol. 83, No. 1, pp. 26-29.
This article addresses how to reduce the effects of hydrogen-induced cracking and corrosion as well as the common failures associated with hydrogen, including lamellar tearing. It can be found in electronic format here: http://www.aws.org/wj/jan04/still_feature.html.
Still, J. R. (November-December 1995). "Lamellar Tearing - A Ghost from the Past." Welding and Metal Fabrication, Vol. 63, No. 10, pp. 445-48.
This article covers the causes of the tearing that are common with steel used in the 1970s while outlining various tests used to assess the likelihood of material failure.
TWI. (2000). Defects - Lamellar Tearing.
An brief overview of lamellar tearing, and good resource for figures. Contains a discussion of sulfur content's effect on tear susceptibility. <http://www.twi-global.com/technical-knowledge/job-knowledge/defects-lamellar-tearing-047/>
Additional Resources:
ASTM A435/A435M - 90 (2012). "Standard Specification for Straight-Beam Ultrasonic Examination of Steel Plates." ASTM International.This standard governs ultrasonic testing of steel plates for lamellar tearing.
Evrard, M., Dubresson, J., & Le Penven, Y. (December 1975). "Ultrasonics in the determination of susceptibility of rolled products to lamellar tearing." Non-Destructive Testing, pp 299-303.
An article on the refinement of ultrasonic testing to better detect lamellar tearing specifically.
Farrar, J. C. M. (1974). "Inclusions and Susceptibility to Lamellar Tearing of Welded Structural Steels." Welding Research Supplement, Welding Journal. August, 1974. pp 321-331.
An investigation of the impact of inclusions on lamellar tearing susceptibility.
Holby, E., and Smith, J. F. (1980). "Lamellar tearing - the problem nobody seems to want to talk about." Welding Journal, 59, 37-44.
A description of lamellar tearing and commentary on the political factors surrounding lamellar failures in the mid-1900s.
Maranian, Peter. (2009). Examples of Major Historical Events. Reducing Brittle and Fatigue Failures in Steel Structures: pp. 23.
A short description of the Atlantic Richfield tower case, and discussion on the Alexander Keilland offshore platform failure (which was determined to have had lamellar tearing involved as partially responsible for the failure along with several other factors).
Steel Construction Manual. (2011). 14th Ed. Chicago: American Institute of Steel Construction. pp. 2-42.
Section 2-42 in the Steel Manual discusses lamellar tearing prevention and lists several studies that help shaped the current specifications.
Verdeja, J. I., Asensio, J. and Pero-Sanz, J. A. (January 2003). "Texture, Formability, Lamellar Tearing and HIC Susceptibility of Ferritic and Low-Carbon HSLA Steels." Materials Characterization, Vol. 50, No. 1, pp. 81-86.
A discussion of the influence of the texture and grain orientation of ferritic and low-carbon steels on the mechanical material properties, such as ductility and toughness. <http://www.sciencedirect.com/science/article/pii/S1044580303001062>
Keywords: lamellar, lamellar tearing, steel plates, crack, cracking, welds, welding, metallurgy, brittle fracture, inclusions, El Paso Civic Center, Atlantic Richfield, King Street, through thickness, transverse thickness, hot rolled, HAZ, heat affected zone, magnetic particle testing, dye penetrant testing, ultrasonic testing, material susceptibility, AISC, hot potato, sulfur, sulfide, oxide, silicate, detailing