<< Back to Construction Incidents Investigation Engineering Reports

For problems with accessibility in using figures and illustrations, please contact the DOC at 202-693-2020.

Investigation of the March 15, 2008 Fatal Tower
Crane Collapse at 303 East 51st Street, New York,
NY


U.S. Department of Labor
Occupational Safety and Health Administration
Directorate of Construction

September 2008

Investigation of the March 15, 2008 Fatal Tower Crane Collapse at 303 East 51st Street, New York, NY

Investigation of the March 15, 2008 Fatal Tower
Crane Collapse at 303 East 51st Street, New York,
NY

September 2008

Report Prepared by
Mohammad Ayub, P.E.
Directorate of Construction
Occupational Safety and Health Administration

Contributions to this report made by
Dinesh Shah, PE, Directorate of Construction, OSHA National Office
Manhattan OSHA Area Office, New York, NY


TABLE OF CONTENTS

  1. Introduction
  2. Analysis
  3. OSHA retained sling expert
  4. Lehigh University Fritz Laboratory tests
  5. Standard Erection Procedure as per the crane manufacturer
  6. Metro Wire Rope Sling and failure scenario
  7. Conclusions

REPORT

The Incident:

On March 15, 2008 at approximately 2:30 P.M. a tower crane approximately 250 ft. high collapsed in uptown Manhattan, NY, killing six construction employees. In addition, a civilian in a nearby apartment was also killed when part of the crane struck the apartment building. The crane was used in the construction of a 43-story concrete framed building located at 303 E. 51st Street. At the time of the incident, employees were placing lateral tie beams on the 18th floor of the building under construction to provide lateral support to the crane. The crane was "jumped" about an hour earlier by adding four additional sections to the tower mast. The jumping of the crane took place without any reported problems.

The building under construction, a 43-story condominium, was framed in poured-in- place columns and beams/slabs. The crane mast was located on the south side of the building under construction. The center of the mast was approximately 22 feet away from the exterior of the building, see Fig. 1 through 5. At the top of the crane mast was the crane platform with operator cab, machinery and a 120 ft. long boom.

The following were the key contractors:

  1. Joy Contractors Inc., (Joy) of Elizabeth, NJ, subcontractor for concrete structure
  2. Rapetti Rigging Services, Inc., (Rapetti) responsible for crane erection, jumping and rigging
  3. Reliance Construction Group, of Quebec, Canada, General Contractor

The tower crane, model M440E, involved in the incident was manufactured by Favelle Favco (Favco) of Harlingen, Texas, see Fig. 6. The crane was reported to be 1 year old and is owned by New York Crane of New Jersey. The foundation and the tie beams were designed by Stroh Engineering Services, PC, of West Babylon, NY. The total height of the crane at its final stage was to be 472 ft. high. At the time of the incident, the height of the crane was approximately 250 ft. The tower consisted of typical sections, 13'-1/2" (4 meters) high. The sections were bolted together at four leg sections by four 2" diameter bolts, specially manufactured by Favco, see Fig. 8 & 9. The mast consisted of four wide flange shape legs 6'-10" center to center on all four sides. All structural steel conformed to Australian standards. The equivalent U.S. steel shape for the legs is W12x106 (Grade 50). The mast legs were connected to each other by horizontal and diagonal members. On two opposite sides, the horizontal and diagonal members consisted of round steel pipes. On the other two opposite sides, the horizontal and diagonal members consisted of steel angles, see Fig. 8. The diagonal pipes were reversed in direction on opposite faces.

The crane was erected so that the crane column flanges were parallel to the building. The north face of the crane mast was closest to the building. The diagonal bracings on the east and west sides consisted of 6" steel pipes welded to the mast leg flanges. The horizontal members on the east and west sides were 5" steel pipes, also welded to mast leg flanges. The knee bracings on the north and south sides consisted of steel angles 6x6x3/4". The horizontal members on the north and south sides were 6x6x1/2". Both steel angles were welded to the mast leg flanges.

The crane mast was laterally supported by steel beam ties connected to the building's structural slabs at the 3rd and the 9th floor. The ties at each floor consisted of three wide flange beams fastened at one end to the building floors and at the other end pinned to a square steel collar surrounding the mast of the crane. The tie beams on the 3rd and the 9th floors were installed using a mobile crane at the time of the initial installation of the crane. The employees were installing the tie beams on the 18th floor without using any mobile crane. This was the first time the employees were installing the tie beams in this manner by using the crane itself.

As the building construction progressed, the building increased in height. At the time of the incident, the 19th floor was already poured with forming in progress for the 20th floor. Under the 19th floor, there was one level of shoring and two levels of reshoring. Subsequently, on the day of the incident, the crane height was also increased by four sections to a total height of 250'. The crane was then placed back in operation. The next thing to accomplish was to connect the mast to the 18th floor slab through the tie beams to provide lateral support to the mast. The process to accomplish this task was first to erect a steel collar around the crane mast by suspending it from the mast steel members above the collar. At this time, the collar would not be physically connected to the mast but will have an approximate gap of 2" between the collar and the mast. Then the collar to the 18th floor was to be connected by three tie beams. One end of the tie beams would be fastened to the structural floor slab, and the other end placed in the collar pocket and pinned. The crane mast would then be re-plumbed, and the gap between the collar and the mast would then be eliminated by tightening the blocks to provide a tight fit. There would be no positive connection between the collar and the mast. The tie beams were to transfer lateral loads only, and not the gravity loads.

The collar weighed approximately 11,200 pounds and came in two halves. The collar was fabricated by the crane manufacturer in 2006 (Model: Building Ties BT110393, Serial No: 012, 018). Each half of the collar is made of a C shape, and when bolted together at its final location, it becomes a square shaped collar ready to receive tie beams. Each half of the collar essentially consists of approximately U.S. W12x87 oriented about the minor axis on all three sides. 2" thick plates were welded to the flange tips of the collar beam parallel to the web of the collar beam, see Fig. 10. The vertical space between the two plates provided the pocket for the tie beam connection to the collar. Through the two plates was a hole for a 3" diameter pin to fasten the tie beams.

After the crane was successfully "jumped", the employees were then ready to erect the collar and then to install tie beams on the 18th floor. Approximately one hour before the incident, the crane hoisted the first half of the collar, weighing approximately 5,600 pounds, and brought it near the 18th floor, see Fig. 13 & 14. Each half of the collar was equipped with six lifting lugs from which it could be supported, see Fig 11. For the sake of discussion and clarity for this report, see Fig. 12, each supporting lug has been marked as A through F, in one half of the collar and G through L in the other half. These alphabetical markings were made by the author of this report and not by the manufacturer. The crane hoisted the first half of the collar on the east side of the crane mast by using lugs marked as C and D. As the hoist approached the crane mast, the employees using the tag line positioned the collar by hanging it at northeast and southeast corners by two 2-inch wide polyester slings choked around the column flanges and the steel angles of the K braces. Each sling was attached to a come along which was, in turn, connected to chain fall fastened to the collar lifting lugs marked B and E. Both slings used at the southeast and northeast corners were manufactured by LiftAll.

In a similar manner, the other half of the collar was then brought by the crane on the west side of the crane mast, again lifting at lugs marked I and J, see Fig. 12. This half of the collar was also hung by two 2" wide polyester slings on the northwest and southwest corners using lugs marked H and K using the same arrangement described above. The sling at the northwest corner was manufactured by LiftAll Company. The sling at the southwest corner was manufactured by Metro Wire Rope of Union, NJ. When both halves of the collar were leveled and plumbed, the two halves were bolted together with four bolts on the north side and four bolts on the south side.

There were no reported problems to this point. The employees then began to maneuver to place tie beams into the collar. As stated earlier, there were three tie beams to be placed, for configuration of the tie beams, see Fig. 5 through 7. At the northwest end, there were two tie beams to be placed, and on the northeast side one tie beam was to be placed. At the time of the incident, only the east tie beam, still supported by the crane, was placed in the pocket of the collar but the pin was not yet placed when suddenly the employees heard a popping sound. Then the employees heard another popping sound followed by a third sound. Soon the collar was sliding down freely to the lower collar, nine stories below, near the 9th floor. It struck the collar at the 9th level and sheared the tie beams. Next, the two collars then fell together over the last collar near the 3rd floor and rested over it, see Fig. 15 through 21. The west and the middle tie beams remained on the 18th floor. The east and west tie beams of the 9th floor failed at the welded connection to the 9th floor slab and got sheared off at the collar. Only the middle tie beam on the 9th floor remained connected to the floor slab, see Fig. 22. All three tie beams on the 3rd floor remained connected.

The lack of lateral ties transformed the crane mast into a free-standing structure with no lateral support above the 3rd floor. The crane mast leaned a little towards the north and then fell towards the south, pivoting near the base of the mast, see Fig. 29 & 30. The crane boom was facing towards the north and the counterweights were towards the south. The crane fell in one piece striking the building, known as 300-304 E. 51 St., across the street. The crane mast was sheared off at the roof of the building with the top portion of the mast including the crane super-structure separating from the lower portion of the mast. The top portion of the crane somersaulted and landed one block away over another building, see Fig. 32 & 33. The tie beam hoisted by the crane which was placed in the pocket of the collar but not yet pinned flew over 306 and 308 E. 50th Street buildings, damaging both structures.

At the time of the incident, there were six Joy employees on the crane mast assisting with the placement of the tie beams. Five were killed, and one was seriously injured. The crane operator, also a Joy employee, was also killed. In addition, a civilian located in the building where the crane superstructure landed was also killed. The president of Rapetti, located on the 18th floor assisting with the placement of the tie beams, sustained serious injuries.

Analysis:

Based upon review of available information, eyewitness statements, and all documents, the use of polyester slings and the manner they were rigged around the mast column, resting in the V shape made by the column legs and the steel angle braces, warranted further engineering evaluation by OSHA. It was readily acknowledged that none of the slings were protected against sharp edges of the column legs and the steel angle legs. OSHA proceeded to determine whether the slings placed in the V-shaped crotch could have a significantly reduced capacity to support the load.

There were four slings used to support the collar. All were 2" wide polyester slings, 6 ft. long. Three were manufactured by LiftAll Company, and one was manufactured by Metro Wire Rope. Post-collapse examination of the slings revealed that the Metro sling was used at the southwest corner of the collar, and had a rated capacity of 6,400 pounds, 5,100 pounds and 12,800 pounds under vertical, choker and V basket configurations, respectively, see Fig. 23. The Metro sling bore a serial # Sample-028. The Liftall slings had a similar rated capacity i.e., 6,400 pounds, 5,000 pounds and 12,800 pounds under vertical, choker and basket configurations, respectively, see Fig. 24. Two LiftAll slings, type EE2-802D, bore serial numbers 1049068. The other LiftAll sling had a serial number, 1020620. The come alongs, Series 653 hand operated load hoists, were manufactured by Columbus McKinnon Corporation of Amherst, NY, and had a rated capacity of 3 tons each. With a usual factor of safety of 4, each sling, if choked properly and in a good condition, would provide an ultimate failure capacity of approximately 20,000 pounds. Given the weight of the collar to be 11,200 pounds, four slings, if all are supporting the collar weight equally, would provide a factor of safety of approximately 7 or more. However, if the slings are not choked properly, and if one of the four slings fails, the capacity of the remaining slings would be greatly reduced.

All four slings failed in the incident, with each sling shearing in two pieces, see Fig. 25 through 28. Out of the eight broken pieces of the slings, only seven were retrieved. Of the seven retrieved pieces, two were still attached to the top most collar lugs at points E and K along with come alongs and chain falls. It was later concluded that the missing piece was the choked portion of the Metro sling located at the southwest corner of the collar.

The seven retrieved slings were marked as follows. The suffix A indicates that one end of the slings was attached to the come alongs. The serial numbers of come alongs are provided below.

1A 57" long Metro (Come along Serial Number R8120)
2A 58" long LiftAll (Come along Serial Number R8113)
4A 24" long LiftAll (Come along Serial Number Q8552)
7A 26" long LiftAll (Come along Serial Number xx474)
11 11" long LiftAll  
12 47" long LiftAll  
13 47" long LiftAll  

Close examination of the slings indicated that the matching pieces were:

2A with 11
4A with 12
7A with 13
1A's matching piece could not be found.

OSHA retained sling expert:

OSHA retained a sling expert to examine the sling remnants and opine on the failure characteristics of the slings; see Appendix A for his full report. The sling expert conducted microscopic examination of the fractured surfaces of the slings. The expert was further asked to conduct actual tests of similar new slings at Southwest Research Institute (SRI), San Antonio, TX to determine failure loads of the slings when subjected to sharp edges of the supporting wide flange shapes. The conclusions of his report are contained in Appendix A. The tests at SRI were conducted by rigging the slings around 4" deep wide flange shapes. The tests indicated that the slings failed at loads significantly lower than their ultimate capacities due to contact with edges of the wide flanges. The tests conducted at SRI, however, did not replicate the actual manner the slings were rigged at the site. OSHA was most interested to determine the load carrying capacity of the slings when trapped in a V-shaped notch with sharp edges of the crane mast legs and the steel angles of the braces.

Lehigh University Fritz Laboratory tests:

OSHA contracted with Advanced Technology for Large Structural Systems Research Center (ATLSS) of Lehigh University, Bethlehem, PA to determine the load carrying capacities of the slings under conditions replicating the actual manner they were used. The tests were performed at Fritz Engineering Laboratory (Fritz) of Lehigh University, PA. OSHA rented a section of the crane mast similar to the one used at the site and transported to the Fritz. 12 slings (9 manufactured by LiftAll, and 3 manufactured by Metro) were tested choked around the column flange and trapped in the V shape, as discussed above. It was concluded that under sustained load, the slings failed at approximately 7,100 pounds, significantly lower than 20,000 pounds (5,000 pounds x factor of safety of 4.0 = 20,000 pounds). If loads are quickly applied, the failure loads were approximately 9,000 to 10,000 pounds, see Fritz data in Appendix B. Most interestingly, the failure was preceded by popping sounds similar to what the employees had described to have heard before the incident.

The entire crane mast section was supported on a steel dunnage on the Fritz's concrete floor. Slings were tested one at a time. A tension load was created on the slings. The tensile test fixture included a cylindrical base, a pull bar and a forcing member. The pull bar included a limiting member, a specimen-fixing member and a shaft member. Air was used in the cylinder to produce the tensile force.

The sling was choked around the wide flange of the crane mast leg passing through a V- shaped notch between the angle of vertical knee bracing and the tip of the flange of the mast leg. The sling at the other end was connected to the hook. A chain was passing through the eye bar of the hook and was connected to the cylinder shaft. Air pressure was introduced in the cylinder to induce tension on the slings and the magnitude of tension was increased in segments until the sling failed. The elongation of the slings under load was also measured until the failure of the sling. The graph was drawn for the elongation of the sling against the induced tension load. The sling was choked clockwise as well as counter-clockwise around the column during the testing. Similar testing was carried out by chocking the slings at other end of the column where round diagonal bracing was located. It was discovered that the sling was invariably trapped in the V-shaped notch created by the leg flange and steel angle brace regardless of whether there was a diagonal pipe brace or not. However, testing was also conducted by forcing the sling to touch the round pipe brace.

In summary, it was found that the slings failed under a tensile load of 7,100 to 10,100 lbs. The sling under a sustained load failed at 7,100 lbs. Initially a tension of approximately 6,000 pounds was placed on the sling. This did not produce any fracture. Then a higher tension of 7,000 pounds was placed. After a few minutes, the sling began to fracture and consequently the tension on the cylinder was partially relieved. The tension was brought back up to 7,000 pounds, after a few minutes the sling suffered further fracture, relieving the cylinder of the load. This cycle continued until the sling completely failed at approximately 7,100 pounds. All slings failed in an angular direction by being cut between the edges of the wide flange of column and the angle leg of knee bracing. This was similar to the fracture pattern observed in the slings involved in the incident. Manual observation of the failed slings indicated that there was heat generated in the fibers of the sling at a failure point, resulting in melting of the fibers. Further, it was observed that there were three angular cut marks or impressions where the sling touched the other (remaining three) flanges of the wide flange legs.

The slings elongated considerably under loads, and the elongations were inconsistent. The following are the elongations of the slings as recorded by Fritz at 2,800 pounds, 1/4 of the collar weight:

Sling test # Elongation

  1. 5 1/2"
  2. 4 1/2"
  3. 3 3/8"
  4. 3"
  5. Not recorded
  6. 4 3/4"
  7. 3 3/8"
  8. 3"
  9. 5 1/2"
  10. 2 1/4"
  11. 6"

As is obvious, the elongations of the slings are not consistent. The magnitude of elongations indicates that the collar must be leveled by using come alongs. Under sustained loads, there could be additional elongations after the collar has been leveled. Therefore, there could be a need for constant monitoring of the level of the collar. If the leveling is not undertaken at all four corners, and if the collar is permitted to dip at one corner greater than at other corners, then the load of the collar might be taken by only two slings instead of the four due to the rigidity of the collar. This would double the load on the supporting slings.

Standard Erection Procedure as per the crane manufacturer:

The standard erection procedure of the "ext.climbing/tie erection sequence" is contained in Favelle Favco drawing No. A1-1100.123, issue B, of January 27, 2006. The procedure consisted of five stages with details provided for each stage, see Fig. 11. As described above, each half of the collar is provided with six points where it could be supported or suspended. The drawing A1-1100.123 in Section B-B describes points B and E, and corresponding points H and K as lifting points. The other points (i.e., A, C, D and F, and the corresponding G, I, J and L) are identified as chain block points. From the ground, the collar was supposed to have been hoisted by the crane at the lifting points, (i.e., B and E,) and the other corresponding lifting points H and K to transport it to the final location. Unfortunately, the collar was hoisted at points C and D, and at corresponding points I and J. When the collar was positioned around the crane mast, the employees had no alternative but to suspend the collar from points B and E, and corresponding points H and K. Therefore, because of location of points, B, E, H and K, the slings were choked around the column, thus landing the slings in the V-shaped groove. This resulted in a drastic reduction in the load carrying capacity. If the collar had been supported at points A and F, and corresponding points G and L, and later after the crane hoist was detached from the collar, at points C and D, and corresponding points I and J, the slings would have had adequate capacity because the slings would have been supported from steel members directly above the collar. This would have completely eliminated the need to choke the slings in the manner it was done. Furthermore, the drawing No. A1-1100.123 required that each half of the collar be supported at four points instead of two. If the instructions contained in the drawing were followed, the collar would have been supported at eight locations, until the two halves were bolted together.

Metro Wire Rope Sling and failure scenario:

Post-incident examination of the Metro sling revealed that the sling was already frayed and deteriorated even before it was used to support the collar, (see expert's report in Appendix A). The situation worsened when the sling was choked around the column, landing it in the V-shaped groove. The degradation and damage to the slings were so extensive that the Metro sling should have been discarded and not used. If, indeed, the Metro sling failed first, then the load of the collar could have been supported by the two opposite diagonal slings, each supporting approximately 5,600 pounds under static loading. With the impact factor due to dynamic loading, and the contributing load of the tie beam, the slings could have reached the failure loads.

Conclusions:

  1. The choice of using polyester slings to suspend the collar at four points was questionable as they are subject to large elongations under tensile loads, thus creating a need to constantly monitor and level the collar.
  2. The collar was rigged improperly in that the slings used to suspend the collar were choked around the vertical legs of the crane mast and was seated in the V-shaped groove between the angle bracing and the flange of the crane mast leg. This significantly reduced the load carrying capacity of the slings.
  3. The slings were not protected against sharp edges for cuts and abrasions.
  4. A deteriorated sling, which should have been discarded if proper inspection of the sling was done prior to its use, was used to suspend the collar.
  5. The crane raised the collar from the ground hoisting it at locations different from the crane manufacturer's recommendations. This led the employees to suspend the collar from locations above which there were no horizontal members. This resulted in choking the slings around the legs of the crane mast.
  6. Each collar half was suspended at two points instead of at four points as recommended by the crane manufacturer.

Accessibility Assistance: Contact the OSHA Directorate of Construction at (202) 693-2020 for assistance accessing PDF materials.