SAC Phase 1 Analytical Studies of Building Performance

Project Title:

Lessons from Inspection, Evaluation, Repair and Construction of Welded Steel Moment Frames following the Northridge Earthquake

Sub-contractors:

William E. Gates, S.E.; Dames & Moore, Inc.

Manual Morden, S.E.; Brandow & Johnston Associates

Project Summary:

Under the SAC Task 2, interviews were conducted to document the significant experiences of key participants involved in the discovery, inspection, evaluation, design, and construction repair of the steel moment resisting frame buildings. Structural engineers, testing and inspection agencies, contractors and building officials were selected as participants in the interview process. The interviews were designed to systematically gather, synthesize and analyze perishable data, such as impressions and unusual experiences that may have been encountered during the process, identify key issues or concerns, and lessons learned.

These interviews produced several key findings, including:

None of the engineers interviewed anticipated that a brittle rather than a ductile mode of failure would occur in welded steel moment frame (WSMF) construction prior to the Northridge earthquake experience. In hindsight, a few of the engineers admitted that they were somewhat skeptical about the ability of the welded beam-column connection to fully develop a plastic hinge before some form of failure occured in the highly stressed weld zone. However, these engineers felt they had no basis on which to reject the building code premise that ductile yielding of the connection could be achieved in an earthquake.

All of the engineers interviewed, without exception, felt that their education or knowledge of metallurgy and the behavior of steel welding processes was lacking the necessary elements for them to appreciate the limitations fo teh materials they were working with.

The Northridge experience has reduced the engineers' confidence in the earthquake performance of WSMF's. from the standpoint of life safety, a well designed reinforced concrete shear wall building is now considered safer than a ductile WSMF constructed to the 1988 UBC (pre-Northridge) standards. About half the engineers consider the steel braced frame and ductile reinforced concrete frame to be safer than a WSMF and the other half consider it to be equally safe. The majority of the engineers still consider the WSMF to be safer than the following structural systems: non-ductile reinforced concrete moment frames with or without unreinforced masonry infill, ordinary steel frames with unreinforced masonry infill, URM's upgraded or not upgraded, concrete tilt-up (post-1976) and precast concrete without adequate connections.

Almost all of the failures, observed by those interviewed, were a brittle form of failure in the welded joint area between beam and column flanges. Only two or three joints in the thousands surveyed were observed to suffer significant plastic deformation. In some cases, plastic deformation occured as a consequence of an initial brittle failure.

The observed damage to the welded beam-column connection tended to be concentrated in the bottom beam flange at the welded joint to the column. If the damage was ultrasonically identified as a weld crack, the ratio of bottom flange to top flange damage was on the average of 15:1. If the damage fractured the weld or base metals and was visually identifiable, the estimated ratio of bottom to top flange damage was 30:1.

The degree of observed weld damage reported in the interview varied significantly from engineering office to engineering office. In general, when weld damage occured, minor cracks, having a depth less than 1/4 inch, were found in approximately 40 to 60% of the damaged cases. Significant cracks with greater than 1/4 inch depth were found in 20 to 40% of the damaged cases, and severe cracks in 10 to 20% of the connections.

The distribution of damage within buildings was reported to be relatively random in the low rise buildings with the greatest amount of damage in the first two stories. Damage in high rise buildings was found to be located in the upper one-half or two-thirds of the building. Directionality of the earthquake ground motion played a significant role in the damaged WSMF's. Those buildings located in the San Fernando Valley tended to have more damage in the north-south frames, while those located in West Los Angeles tended to have more damage in the east-west frames oriented parallel to Santa Monica Boulevard.

The geographic distribution of damage, as reported, seemed to be related to the location of the building relative to the earhquake epicenter or center or energy release. Damaged buildings were located within a 20-mile radius of the epicenter. This included West Los Angeles, Santa Monica, Burbank, Santa Clarita Valley, and much of the San Fernando Valley. No damage was reported in Mid-Wilshire, Hollywood, or downtown Los Angeles.

There was little consensus among the engineers interviewed as to the major factors leading to the brittle failure in the welded moment connections. However, most now believe that the welded moment connection was a flawed design due to the high triaxial state of stress that limits yielding and plastic deformation and due to stress risers, such as the backup bar, and stress concentration factors, such as the interaction between the column web and beam flange in the weld zone. Furthermore, most of the engineers interviewed now believe that the qualification tests for the prescriptive connection failed to represent the evolving field conditions. Prior to the Northridge earthquake, the engineers were overly optimistic about the test results and tended to forget poor results. Other key factors named were the welding practices and materials (both base steel and weld metal) used in construction. The engineers were in general agreement that the rapid impulsive energy release from the Northridge earthquake may have been a key factor in the brittle failures observed. Few of the engineers felt that the amplitude of the ground motion or the high component of vertical ground thrust contributed significantly to the damage.

All of the engineers interviewed agreed that standards and procedures for welding need to be revised and "tightened up" and that the design of the welded moment connection needs to be revised. There is also strong opinion on the part of some of the engineers interviewed that redundancy in the ductile steel moment frame needs to be increased.

The testing laboratory personnel offered the following statements or opinions:

a. The AWS standards for welded conection acceptance are too lax for this type of joint and permit too much poor fusion and imperfections in the welds.

b. The certification criteria for welders is inadequate. For new work, welders should be required to qualify by welding a real joint in the same position as it will be constructed, incuding penetration welds with continuity through an access hole (i.e., the "rat hole" at the bottom beam flange to column flange). For repair work, the welders should also be required to perform the qualification welds with restricted access, similar to the conditions commonly encountered in the field.

c. Many buildings investigated after the earthquake exhibited "atrocious" fitup, joint preparation and weld quality. This may be related to the degree of damage found in these buildings located furthest away from the epicenter, or in area of lower amplitude ground motion.

d. In some cases, there was evidence of falsified or totally inadequate previous post-earthquake inspection (i.e., chalk marks on columns with arrows pointing up and marked "OK" where fireproofing had never been removed from the beam-to-column connection).

Field inspection and testing procedures for weld damage varied from firm to firm. Most of the firms relied on engineering judgement to select the connections for inspection. Some firms used elastic dynamic analyses after the damage had been found to identify where to look further for potential damage. A few performed dynamic analyses, first, before going into the field to inspect the building. They assumed that this information provided a means of finding those connections that were more likely to be damaged. Visual and ultrasonic testing were the most commonly used methods for inspection.

Inspection costs for typical commercial buildings ranged from $800 to $1,200 per connection. This cost did not include cases where asbestos had to be removed. In such instances, the cost could double or triple.

When fracture develops through the welds or flanges of the moment connections, related damage was found about half the time in the non-seismic (gravity) frame connections. The damage ranged from partially torn shear tabs to cases in which all of the bolts at both ends of the beam's gravity connections had failed, leaving the beam resting on the shear tab or supported from the floor slab above by the shear studs. The ratio of observed moment connection to gravity connecion damage ranged from 3:2 to 20:1.

There is one documented case of progressive crack propagation with time in an 11-story high rise building that was designed to 1.5 times UBC Zone 4 requirements. Over the period from July 1994 through December 1994, the cracks have propagated from the welds into the bases metal of the column or beam flanges. The engineers involved postulate that the crack propagation may be due to continuing relaxation of original residual stresses and readjustment of strains in the frames induced by the earthquake.

There is no common definition or repairable damage vs. damage that requires retrofit strengthening. The decision varies from engineering office to engineering office. There appears to be no real guidance from either the building department in the local community, the engineering profession, or the welding society on this matter. (Similar significant ramifications are implied in future earthquakes if ductile flange buckling and yielding develops in the moment resisting frames due to seismic overload. When should the connection be repaired to compensate for damaging plastic distortions or replaced to restore the structure its original elastic strength?) Guidelines and standards are needed by the engineers to identify the acceptable level and extent of damage before repair procedures are converted to retrofit strengthening.

Engineers have proceeded with repair and retrofit strengthening based on judgement and common sense. They feel they are operation on their own, without specific guidelines as to scope of required repair and/or retrofit and without sufficient test data to substantiate the repair and/or retrofit schemes that they are using. The repair schemes typically consist of putting back a welded joint that more closely conforms to what the engineers thought they were specifying in the first place. Retrofit strengthening generally consists of adding plates or tees to the bottom and top flanges of the beam at the column joint to increase the connection capacity. There is little attempt to balance the design out by adding similar stiffness and strength to the undamaged connections of the WSMF. However, there is widespread concern is the engineering community that such an unbalance in stiffness and strength could result in less favorable performance in future earthquakes.

Few owners are going beyond the simple repair process (i.e., putting the damaged connection back as it was before the NOrthridge earthquake). Oly where FEMA or some other agency is picking up the cost of repair and retrofit are the owners requesting seismic upgrade.

Repair costs for damaged connections in typical commercial buildings range from $3,000 to $20,000 per connection. The typical costs are in the $5,000 to $8,000 range. If asbestos has to be removed, the costs may increase by an additional $2,000 to $3,000 per connection. The owner's hidden costs associated with tenant expenses an lost rent may range from $0 to $45,000 per connection.

Costs for retrofit stregthening may be comparable to those for repair in typical commercial construction. In residential and institutional construction, these costs may double or triple to $15,000 to $30,000 per connection for construction and $15,000 to $60,000 per connection for owner hidden costs if lost rental income is included.

SAC Home

SAC Steel Project
c/o Earthquake Engineering Research Center
1301 South 46th Street
Richmond, CA 94804
(510) 231-9477
FAX: (510) 231-5664
sacsteel@sacsteel.org