By Charles F. Scribner,1 Member, ASCE, and Charles G. Culver,2
Downloaded from by Universidad Politecnica De Valencia on 06/09/15. Copyright ASCE. For personal use only; all rights reserved.

Fellow, ASCE

ABSTRACT: Results from an investigation of the collapse of the L'Am-
biance Plaza building on April 23, 1987, are presented. The building was
being constructed using the lift-slab method; collapse occurred during
construction. The investigation included examination of debris at the
site of the collapse, review of eyewitness accounts of the collapse,
review of project documentation, laboratory and field tests, and analy-
ses of the structure. Several potential failure mechanisms are investi-
gated. The most probable cause of the collapse is determined to be loss
of support at a lifting jack in the west tower during placement of a group
of three floor slabs. The loss of support was likely due to excessive
deformation of a shearhead lifting angle, which caused a lifting nut to
slip off its lifting angle. This failure mechanism was duplicated in
laboratory experiments. As loads were redistributed after the initial
failure, the remaining jack rod lifting nuts along column line E support-
ing the package of floor slabs slipped off their lifting angles and the slabs
failed in flexure and shear. These slabs fell causing the slabs below them
to fail.


At approximately 1:30 P.M. on April 23, 1987, the L'Ambiance Plaza, an
apartment building under construction in Bridgeport, Connecticut, col-
lapsed, killing 28 construction workers. This was the largest loss of life in
a construction accident in the United States since 51 workmen were killed
in the collapse of a reinforced-concrete cooling tower under construction
at Willow Island, West Virginia, in 1978 (Lew et al. 1979). A photograph of
the structure taken approximately two hours after the collapse is shown in
Fig. 1.
Representatives of the Occupational Safety and Health Administration
(OSHA) arrived on the building site shortly after the collapse occurred.
O S H A requested that the National Bureau of Standards' (NBS) Center for
Building Technology conduct an investigation to determine the most
probable cause of the collapse. Engineers from N B S arrived on the site and
began their investigation on April 24, 1987. In addition to information
collected during examination of the debris, the N B S team, in carrying out
its investigation, used:

1. Descriptions of the collapse given by survivors and eyewitnesses.
•Res. Struct. Engr., Structures Div., Ctr. for Bldg. Tech., Nat. Bureau of
Standards, Gaithersburg, MD 20899.
Chf., Structures Div., Ctr. for Bldg. Tech., Nat. Bureau of Standards, Gaith-
ersburg, MD 20899.
Note. Discussion open until October 1, 1988. To extend the closing date one
month, a written request must be filed with the ASCE Manager of Journals. The
manuscript for this paper was submitted for review and possible publication on
December 21, 1987. This paper is part of the Journal of Performance of Constructed
Facilities, Vol. 2, No. 2, May, 1988. ©ASCE, ISSN0887-3828/88/0002-0058/$1.00 +
$.15 per page. Paper No. 22453.


J. Perform. Constr. Facil. 1988.2:58-79.

Downloaded from by Universidad Politecnica De Valencia on 06/09/15. Copyright ASCE. For personal use only; all rights reserved.

FIG. 1. L'Ambiance Plaza Immediately after Collapse

2. Project documentation, including design specifications, plans, shop
drawings, construction records, testing laboratory reports, and project
3. Laboratory tests of samples of construction materials removed from
the collapsed structure and critical subassemblies.
4. Data obtained from an investigation of the foundations at the site after
the collapse.
5. Analytical studies, including computer analyses.

This paper summarizes the results of the investigation by NBS. Com-
plete details of all aspects of this investigation are available in a National
Bureau of Standards report (Culver et al. 1987).


L'Ambiance Plaza was to have been a 16-story structure consisting of 13
stories of apartments and 3 levels of parking. The structure consisted of
two rectangular towers, each 112 ft (34 m) by 62 ft (19 m) in plan,
designated in this report as the east and west towers. A five-story parking
garage planned for construction adjacent to the two towers had not been
erected and was not involved in the collapse. A plan view of the main
building is shown in Fig. 2.
Steel columns (W and HP shapes) supported the floor slabs. The floor
slabs in each tower were erected independently and were connected by

J. Perform. Constr. Facil. 1988.2:58-79.

org by Universidad Politecnica De Valencia on 06/09/15. . all rights reserved.11 x 1 H H -0) H H • w 5 -0 (8 N ra -0 2o I • X Hf I £ tn 2 0—1 * x H_J i » 8 5 0^- -0= is a: • c s| z i.9. Copyright ASCE. Facil.U . .9.9. . Constr.U .9.1 o) M (a) (x . Perform.i.t>3 X .9.2:58-79.Z9 £0) (Q Downloaded from ascelibrary. For personal use only.Zt J .B. .fr3 .9. 1988.929 60 J.

For personal use only. The elements of the lifting system are shown in Fig. Copyright ASCE. .Downloaded from ascelibrary. 3. Columns were erected and slabs were lifted in stages. threaded jack rods and attachments. The floor and roof slabs were two-way. and welded steel collars called shearheads placed in the concrete floor slab at each column." With the exception of the first and last packages of two floor slabs in each tower. The lower crossarm sat directly on the column top. In the east-west direction. 4. The tendons along column line E in the west tower were splayed as shown in Fig. 3. Slabs in temporary position m » » »i Slabs in permanent position FIG. Sequence of Erection of Columns and Floor Slabs cast-in-place reinforced-concrete pour strips at each floor level after the floor slabs had been secured in their final positions. After being post-tensioned. the tendons were banded along the column lines. Facil. Constr. all packages consisted of three floor slabs. 5. The location of the post-tensioning tendons is shown schematically in Fig. The tendons in the north-south direction were approximately uniformly spaced. all rights reserved. As this figure illustrates.2:58-79. Perform. 1988. Normal-weight concrete was used throughout the structure. the slabs were erected using a lifting system consisting of hydraulic jacks. The building was being constructed using the lift-slab method. A hydraulic jack was mounted on the top of each column. unbonded post-tensioned flat plates cast one on top of the other at ground level. up to three slabs were lifted at one time in a "package. by Universidad Politecnica De Valencia on 06/09/15. The centroids of the banded tendons did not coincide with the column centerlines. Slabs contained a nominal amount of bonded reinforcement in the vicinity of the columns and shearwalls. The upper 61 J. Each jack consisted of an upper and lower crossarm separated by a hydraulic cylinder. The sequence of lifting floor slabs and extending columns is shown schematically in Fig.

Copyright ASCE. 1988. by Universidad Politecnica De Valencia on 06/09/15. ! O <5 o o u. 03 (A a. Downloaded from ascelibrary. all rights reserved. Constr. Facil. 62 J. I . For personal use only.2:58-79.

Floors were typically lifted at rates of approximately 5 ft (1. Once the slabs had been raised to approximately the desired elevation for "parking" (mechanical attachment to the columns).7-mm) increments. . Elements of Lifting System arm supported two jack rods. The jacks lifted the floors in 1/2-in. Perform. all rights reserved. 1/2" stroke limit Nut drive arm (In raised position) Holding nut Lower crossarm Lifting rod 1-3/4" dia.334 kN). A system of electrical interlocks prevented the floor slabs from being raised more than 1/2 in. G3. and G4 in the west tower). Facil. or lifting cycles. were used. Take-up nut Upper crossarm Hydraulic jack Downloaded from ascelibrary.2:58-79. and D10 in the east tower and columns E3. Constr. BIO. 1988. "super jacks. 5." with a capacity of 300 kips (1. The nominal load capacity of a small jack was 150 kips (667 kN). At the four columns with the heaviest sections in each tower (B9. Separate control consoles and engine-driven power units supplied hydraulic pressure to the jacks in each tower. (12. All jacks shared similar essential features and modes of operation.8. which were attached to the shearhead and supported the floor by Universidad Politecnica De Valencia on 06/09/15. D9. Copyright ASCE. Two sizes of jacks were used. the jacks at each column could be controlled individually to raise or lower the slab to the 63 J. differentially between any two jacks. Acme thread Column Rod coupler Extension rod Sleeve Lifting angle Shearhead Lifting nut FIG. For personal use only.52 m) per hour. All jacks connected to a given console operated at the same line pressure. E3.

org by Universidad Politecnica De Valencia on 06/09/15. STATUS OF CONSTRUCTION AT TIME OF COLLAPSE Several types of construction work were in progress in the building on the day of the collapse. the wedges were fully welded around their perimeter. 6.Downloaded from ascelibrary. The floor slabs were supported at the columns by steel wedges placed between the lower face of the shearheads and blocks welded to the flanges of the columns. in the cavities between the columns and floor slabs in each tower. all rights reserved. Detail of Slab-to-Column Connection desired elevation. Facil. Constr. electricians. Copyright ASCE. and in pour strips connecting the floors of the east and west towers at level 2. the top flanges of the shearhead channels were welded to the seal blocks. A typical connection of a slab to a column is shown in Fig. Masons were placing con- crete in shearwaU forms. Plumbers.2:58-79. 1988. and concrete was placed in the cavity between the shearhead and the column. . 10. 6. Slabs could be parked either temporarily or permanently. FIG. The group of slabs for floors 9. If the slab was to be parked temporarily. wedges would be held in position only by tack welds and friction. and carpenters were installing nonstructural items in both towers. Perform. Floor slabs were being lifted and attached to columns with wedges in both the east and west towers on April 23d. During that morning the fourth stage of erection of the east tower had been completed and workmen were welding wedges permanently in place when the collapse occurred. and 11 of the west tower had been raised to its temporary position in the fourth stage of lifting at approxi- 64 J. For personal use only. If the slab was to be parked permanently.

all rights reserved.. WEST TOWER EAST TOWER U ROOF Downloaded from by Universidad Politecnica De Valencia on 06/09/15. approximately 1:30 P. At the time of the collapse.2:58-79. J l Pour strips In place Slab-Column Joint Configuration in " ' | [ e between towers Wedges tack-welded 1 only TT Shear wall Wedges permanently welded Shear walls and pour strip Wedges permanently welded. DESCRIPTIONS OF COLLAPSE Forty-five individuals who were on the site or nearby provided descrip- tions of the collapse. Copyright ASCE. Most of the witnesses indicated the collapse started in the west tower. workmen were installing wedges to hold slabs 9.8. all witnesses agreed that the collapse was extremely rapid. 1988. Three other individuals who did not witness the collapse provided general information about construction procedures used at the job site. 7. 7. with most estimating its duration at between two and ten seconds. In general. A schematic represen- tation of the state of construction at the time of the collapse is shown in Fig. Fourteen of these individuals were in the structure when the collapse began. LEGEND: Jl. Status of Construction at Time of Collapse—South Elevation mately 11:30 A. cast on day of collapse = * * = Seal blocks welded and column II pockets filled FIG. . 10. All workmen who believed they saw the failure begin said the failure 65 J. Almost all witnesses first noticed the start of the collapse because of a loud snap or bang.M. Witnesses did not see the building moving prior to the collapse. Perform. For personal use only. None of the workmen in the building reported any swaying or unusual motion of the building. and 11 in position temporarily. Constr.M. Wedges had been installed along column lines G and H and ironworkers were installing wedges at column E4. which some likened to the slamming of a hinged tailgate of a dump truck or the sound of a piece of metal snapping under pressure. Facil.

Facil.8 underneath slabs 9. For personal use only. underlying soil and/or fractured rock. Standard penetration tests and pressuremeter tests were run at selected locations. DESCRIPTION OF INVESTIGATION Photographs were taken and videotape records were made immediately following the collapse and as debris was removed from the site over the following week. The material below the bottom of the foundations was explored by core borings. and (4) design and construction drawings. nuts. and the underlying bedrock. Copyright ASCE. Constr. all rights reserved. which he believed came from within 25 ft (7. Routine classification tests and direct shear tests on reconstituted soil samples were run in the laboratory. and in situ tests where feasible. and mechanical-electrical drawings.2:58-79. 10.8. Elevations at the column locations were measured to determine whether the footings had settled. (4) one large and one small hydraulic jack and jack rods. 1988." The floor slabs above him then fell. Wind-speed data and temperature records were obtained from the weather station at the Bridgeport airport. They indicated the west tower then fell slightly eastward into the east tower.6 m) of where he was working. and architectural. and (5) several shearheads and by Universidad Politecnica De Valencia on 06/09/15. and 11 stated that at the start of the collapse he heard a loud noise. He was carried down with the collapsing structure and later rescued. (3) project corre- spondence. test pits. The objective of this investigation was to determine whether the conditions of the footings or the backfill behind the basement wall could have contributed to the collapse. Standard tests included compression and 66 J. The core borings penetrated the slab on grade. (3) samples of post-tensioning strand. him or toward the center of the slab in the vicinity of column E3. . project specifi- cations. either directly above Downloaded from ascelibrary. The NBS staff also examined project documentation including: (1) The daily construction logs. Descriptions of the collapse given by three witnesses who were south of the building were essentially similar. (2) portions of structural steel columns. and attachments. and other fittings. column footings. He then noticed the floor slab directly over his head "was cracking just like ice breaking. (2) testing laboratory reports. Sketches were made of the pattern of deformation of the columns and the orientation of the collapsed floor slabs. The east tower then listed slightly eastward and collapsed completely. All three heard one or more loud noises and then saw the west tower beginning to collapse. Samples of the construction materials tested at the National Bureau of Standards in- cluded: (1) Concrete core samples from floor slabs and shear walls. An ironworker installing wedges at column E4. LABORATORY TESTS Tests performed in the laboratory included standard tests to evaluate properties of the construction materials and tests of components and assemblies of the lifting system. causing the top floors of the east tower to fall onto the floors below. appeared to start high in the structure in the west tower. structural. The in situ condition of the foundations and the backfill behind a basement wall on the north side of the building were examined after the majority of the debris had been removed from the building site. weldments. Perform.

The static yield stress. fracture. and tensile tests of the post-tensioning strand.3 ksi (547 MPa).000 MPa). splitting tensile tests of the concrete cores.4 ksi (265 MPa). Results of tests using the rebound hammer indicated the quality of the concrete was relatively uniform throughout the structure. 67 J.6 MPa) specified on the plans. Speci- mens were tested to failure in tension and their ultimate tensile strengths were recorded. Eight core samples were instrumented with a compressometer to measure axial deformations and elastic moduli of the cylinders as they were tested.5 MPa) for cores taken from shearwalls.570 psi (31. A group of unma- chined coupons was also prepared from these column segments.000 MPa).4 MPa) for cores taken from slabs and 4. The average static yield stress and ultimate tensile strength of coupons cut from the sample of A572 Grade 50 steel were 52. The strength of the concrete throughout the structure was also evaluated approximately using a rebound hammer. and shearhead). Perform. all rights reserved. head-to-column connection and the lifting assembly (jack. The average ultimate tensile strength of those coupons was by Universidad Politecnica De Valencia on 06/09/15. jack rods. . All mechanical and chemical properties measured satisfied the appropriate requirements.8 X 103 ksi (205. jack rods and attachments. Constr. Average compressive strength was 5. including the shear- Downloaded from ascelibrary. Project specifications required that steel in the columns conform to the requirements for either ASTM A36 or A572.2:58-79.420 psi (37. The average splitting tensile strength of all cores was 565 psi (3. ultimate tensile strength. and chemical analyses of samples of the steel columns and weldments.9 ksi (468 MPa) and the average value of the elastic modulus was 29. Steel Specimens Coupons taken from three column segments were tested to evaluate the mechanical properties of steel in the columns. tensile and hardness tests and metallographic. were tested to determine their behavior and capacity. and percentage of elongation were calculated for each coupon.000 psi (27. elastic modulus. depending on the location of the column in the structure. Facil.2 x 103 ksi (201. Five cores were tested in diametric compression to determine splitting tensile strength of the concrete. Critical subassemblies. The average value for the elastic modulus of that steel was 29. 1988.9 ksi (365 MPa) and 79. Grade 50 structural steel. No concrete was found that had a strength less than the 4. No evidence was found to indicate that any concrete had been damaged by cold weather. For personal use only. Specimens were tested to failure in tension according to American Society of Testing Materials (ASTM) procedures. The average static yield stress of coupons cut from the specimens of A36 steel was 38. Concrete Specimens The strength of concrete was determined from 3-3/4-in. Weldments Fifteen coupons of weldments were machined in accordance with ASTM E8-85b from field splices in three column segments.90 MPa). Forty-three core samples were taken. Copyright ASCE. and attachments. diameter cores cut from floor slabs and shear walls following the collapse.

Only 1 of the 15 machined coupon specimens failed in the base metal. and one of the specimens necked down near the center of the rod. (3) analysis of lateral stability. (2) analysis of stability of individual columns. seven-wire. Four segments of 1-3/4-in. Copyright ASCE. specimens. Perform. The other 14 coupons failed in the weld metal. The ultimate tensile strength of 10 of the 15 coupons was less than the minimum ultimate tensile strength specified for ASTM A572. (44.5 times the rated load of 75 kips (334 kN). all rights reserved. Jack Rods Jack rods used with the small jacks were tested to determine their tensile strength and pattern of fracture. the incomplete portion of the weld comprised between 15 and 60% of the cross-sectional area of the by Universidad Politecnica De Valencia on 06/09/15. The average breaking strength for the eight specimens of strand was less than 1% less than the specified minimum. AH four of the specimens of used strand had strengths in excess of the specified minimum. a value 2. Columns were modeled as point supports at the column centers. Porosity and incomplete penetration were seen in the fracture surfaces of three specimens. The fracture surfaces indicated a ductile failure. (4) calculation of the capacity of the floor slabs in shear and flexure. Facil. The analyses included: (1) Estimation of reactions at all column locations under a variety of conditions that could have existed at the time of the collapse. Constr. . Results of tests of the three unmachined coupons were consistent with those obtained from the machined coupons.5 kips (825 kN). Yield strengths of all eight specimens exceeded the strength specified by ASTM A416. Elongations of approximately 8% in a 40-in. (5) analysis of possible vertical or lateral deformations of the structure due to foundation conditions. The tensile strengths for these welds were less than the strengths expected for the weld metal (near 75 ksi).7-mm). Grade 50 steel.5-mm) diameter jack rods were loaded monotonically to failure in tension. In these Downloaded from ascelibrary. Only one of the specimens of unused strand had a breaking strength above the specified minimum strength. (1.000-mm) gage length were recorded. STRUCTURAL ANALYSES Structural analyses were performed to determine forces present in various parts of the structure under a variety of loading conditions that might have been encountered at the time the collapse occurred and to evaluate the likelihood of various modes of failure. an amount greater than that considered acceptable by the American Welding Society specification ("Structural welding" 1983). The average failure load was 185. and (6) analysis of the structure under the wind loads that existed at the time of the collapse. (12. The effects of post-tensioning were included in the model as 68 J. All splices of columns were designed to be complete penetration welds. Floor Slabs A two-dimensional finite element computer model was used to analyze the floor slabs. For personal use only. 1988. Grade 270. Low-Lax post-tensioning strands were tested. Post-Tensioning Strand Four specimens of both unused and used 1/2-in.

For column E4. The reactions near the center of the tower.8 increased 48% when the support at E4.2 kips E o -2. Note. however. The reactions at the majority of columns due to the dead load of three slabs were below the capacity of the jacks used at those columns. The reaction at column E3. The reactions and stresses resulting from loss of support at various columns and support reactions due to raising the slab at one location were also determined. Support Reactions for Single Slab in West Tower compressive stress in the plane of the slab and secondary moments induced by the drape of the post-tensioning cables. The majority of the analyses dealt with the west tower. and G4). 69 J. .88x103ksi FIG. 1988. failure of a weld block or loss of a wedge. and E. The support reactions for a single slab in the west tower in the "as-cast" position are shown in Fig. Copyright ASCE. Constr. all rights reserved.5. These reactions can be multiplied by three to obtain the reactions for a package of three slabs. E4. For the east tower. Such loss of support could have resulted from failure of a jack rod. For personal use only. E3. G2.8. Loss of support at column E4. were higher than the reactions around the perimeter. G3. Facil. with considerable increases in reactions at the surrounding columns. by Universidad Politecnica De Valencia on 06/09/15. and D10. Loads on the lifting jacks in the east tower were also calculated. E3. where super jacks were used. with the reaction at E4. where super jacks were used (columns E3. failure of a shearhead.8 caused a large change in reactions at other supports.5. however.8 being the largest in the west tower. G5.8. the largest reactions occur at columns B9.8 was removed. D9. BIO. Support reactions and rotations at each column location and stresses and displace- ments throughout the slab were determined for the slab subjected to its own dead weight. that the reactions at columns E2. C. C4. and G5 are also relatively large. * kips SLAB: Unit weight -145 lbs/ft 3 Total weight-601. Perform.8.2:58-79. the three-slab load was equal to the nominal capacity of the jack used at that column.8. 8.8. The computer model was also used to calculate column reactions resulting from loss of support of the floor slab at selected column locations.Downloaded from ascelibrary.

inelastic analy- sis of each column line. or 12.1.8 kips (675 kN) to 201. Columns were modeled as line elements. Results of a lateral load analysis indicated that it was unlikely that the lateral force or displacement due to this jack could have initiated the collapse of the west tower. Computed ratios of capacity to load were between 4.1 kips (894 kN) to raise the slab one-half inch at column E4.8.3 m). however. Rotational restraint of the columns by temporary slab-to-column connections was neglected. The effective length factor was conservatively assumed to be K = 1. For personal use only.8 would have to be increased from by Universidad Politecnica De Valencia on 06/09/15. Shear Capacity of Slabs The capacity of the slab to resist punching shear failure at the column locations was computed according to procedures recommended by the American Concrete Institute (ACI) ("Building code" 1983).2:58-79.7 mm) was considered. 1988. This model indicated the sidesway buckling capacity-to-load ratio for east-west motion of the west tower was approximately 5. Constr. Copyright ASCE. A 12-ton (107-kN) hydraulic jack was being used on the day of the collapse to plumb the west tower. Several analyses were performed. Buckling loads were between 2. the maximum unbraced length in the west tower at the time of the collapse. as was commonly done during placement of Downloaded from ascelibrary. Other analyses indicated. 70 J. Facil. The safety factor against lateral buckling of the west tower for this case was calculated to be 1. This analysis was designed to determine the changes in support reactions that could result from raising the slab at a single column location. The support reactions resulting from raising the slab at a single column location were also calculated. The analysis showed that all columns were stable for these conditions. that neglecting rotational restraint due to placement of the wedges on the column flanges was a conservative assumption. On the basis of these calculations. Raising of the slab at column E4. The length of each column was assumed to be 30 ft 5 in.07.2 and 7 times as large as the actual loads on the columns.6. wedges at a column. it was considered unlikely that the collapse of the structure could have been due to failure of the slabs in shear. The shear walls in place at the time of the collapse were assumed to provide lateral restraint. The concrete slabs were assumed to be rigid in their plane and pin-connected to the columns. (9. Inelastic stability in the east-west direction was determined by a two-dimensional. . This analysis showed that the force applied by the jack at column E4. Instability of Individual Columns Instability of the columns in the west tower was examined to determine whether this could have caused collapse of the building. An analysis was conducted with temporary slab-to-column connections being modeled as linear elastic-plastic rotational restraints whose characteristics were based on the actual dimensions of the connec- tion.0 and 23.8 by one full stroke of the jack (1/2 in. Perform. all rights reserved. Lateral Instability The possibility of sidesway buckling of the west tower was also considered.

jack rods and attachments. Three tests of this assembly showed that the connection could withstand all static vertical loads present during normal construction operations. Photographs showed that no significant displacement of the wall had taken place immediately after the collapse occurred. and a section of column that included weld blocks and seal blocks. It was considered unlikely that settlements of the magnitude of those calculated could have caused failure of the structure. Differential Foundation Settlements The behavior of the foundations of the structure was estimated on the basis of the soil properties and bedrock characteristics determined during the NBS investigation of the conditions of the footings and fill used on the north side of the retaining wall.7. Field observations confirmed these calculations. (1.1 and the wind-speed data obtained from the National Weather Service station at the Bridgeport airport. Copyright ASCE.8. it was considered unlikely that shearwall failure or excessive displacements caused by lateral soil pressure against the basement wall contributed to collapse of the building. backfill behind the basement wall north of the building prior to the collapse were resisted in part by the floor slabs at levels C and D. On the basis of these analyses. sliding and/or overturning of shearwall footings was determined to be unlikely. TESTS OF SUBASSEMBLIES Tests were conducted to evaluate the performance and strength of two subassemblies critical to the erection of floor slabs. Perform.05 in. The ultimate shear resistance of the most critical shearwall was calculated to be 40% greater than forces transmitted to it by horizontal soil pressure. with approximately 90% of these settlements occur- ring before any slab was fixed in its permanent position. The first subassembly consisted of a shearhead. The test setup 71 J. It was concluded that wind effects did not play any significant role in the collapse. all rights reserved.3 mm) or less. The second subassembly consisted of a 150-kip (667-kN) lifting jack. It was estimated that differential foun- Downloaded from ascelibrary. The forces acting on the slabs were determined on the basis of soil properties obtained from the subsurface investigation. This assembly made up the temporary slab-to-column connection. by Universidad Politecnica De Valencia on 06/09/15. dation settlements not exceeding 3/8 in. Facil. This lifting assembly was tested to evaluate the behavior and capacity of the mechanism used to lift the slabs into place. The minimum reserve capacity against failure was approximately 2. and a shearhead of the type used at column E4. In addition. Wind Loading The maximum wind loads acting on the structure at the time of collapse were calculated using the procedure described in ANSI A58. The lateral loads calculated were an order of magnitude smaller than the calculated lateral capacity of the structure. Lateral Earth Pressure The lateral pressures exerted by the. which carried the forces to the foundations. . For personal use only. These forces were transmitted by the slabs to the columns and shearwall. The maximum settlement that could have occurred after initial lifting of the slab was estimated to have been 0. 1988. (10 mm) could have occurred during construction.2:58-79. two wedges.

9. The shearhead used for the third test was stiffened by two steel bars welded between the tops of the vertical legs of the lifting angles. This 72 J. 10. Facil. 150 kip lift slab jack (inoperable) Downloaded from ascelibrary. The lifting jack supported the jack rods as had been done in the structure. Load was applied by hydraulic rams positioned between the base of the lifting jack and a spacer fabricated from structural tubing. Perform. Five tests of the lifting assembly were conducted. 9. 1988. The shearheads were loaded uniformly around their top perimeter to simulate approximately the conditions at a shearhead in the structure. The major variable in the tests was the condition of confinement of the by Universidad Politecnica De Valencia on 06/09/15. and 11 in the west tower at the time of the collapse. Constr. -1"x5'x14" steel plate 2-100 ton hydraulic rams Threaded rod 1"x9'x16' steel plate 6x12 structural tubing 2* steel plate 2*x2*bar Mark P50 shearhead Lifting nut FIG. . all rights reserved. For personal use only. Bare shearheads (not confined by concrete) were used in three of these tests. Copyright ASCE. The distance from the underside of the hydraulic jack to the bottom of the shearhead corresponded to the distance between the top of the column and the bottom of slabs 9. Lifting Assembly Test Setup is shown schematically in Fig.

Perform. Failure of the lifting assembly took place in one of two ways. additional yielding of the lifting angles and arm channels of the shearheads allowed the ends of the jack rods to move toward the center of the shearhead. Load was applied slowly in 10-kip (44-kN) increments.2:58-79. FIG. 10. Facil. Load was transferred from the spacer to the concrete slab through wooden timbers positioned around the perimeter of the shearhead opening. In each test deformation of the shearhead consisted of twisting of the lifting angles and the arm channels as well as local deformations of the lifting angles near the! point at which the lifting nuts applied loads. In the second type of failure. When load was increased. In four tests transducers were used to measure the lateral displacement of the flanges of the arm channels. The first type of failure occurred when the lifting angle rotated sufficiently to allow the lifting nut to overcome friction forces and slip out from under the lifting angle. This condition represented an effective upper bound of lateral restraint. The behavior of the lifting angle of the shearhead and the lifting nut was recorded on videotape for all tests. Lifting angles deformed in the manner seen in this specimen were observed on several shearheads in the collapsed structure. all rights reserved. Constr. Copyright ASCE. For personal use only. The manner in which the shearheads deformed is shown in Fig. each of which occurred at the connection of the jack rod to the shearhead. 1988. No post-tensioning was used in these specimens. shearheads were confined by a concrete slab containing a nominal amount of mild steel reinforcement.Downloaded from ascelibrary. rotation of the lifting angle and arm channel of the shearhead caused the jack rod to fracture due to combined flexure and axial load. In tests 4 and 5. Deformation and yielding of the lifting angles and arm channels of the shearhead were first observed when the load on the system reached approximately 120 kips (534 kN). Deformed Shearhead Number 2 after Lifting Assembly Test prevented lateral movement of the top of the lifting angles and the top flanges of the shearhead arm channels. 10. .org by Universidad Politecnica De Valencia on 06/09/15. All failures occurred with very little warning and were accompanied by a loud bang as energy was released rapidly and 73 J.

All structural steel was found to be acceptable in both chemical composition and strength.2:58-79. 74 J. construction materials were found to conform to specifications established for the project. 1988. Other tests performed during the investigation (Culver et al. Kips kN Failure type and location (D (2) (4) (5) Downloaded from ascelibrary. west side. Facil. defined the bounds of strength of the lifting assembly. 1987) showed that lifting angles on shearheads designated as X51 (the type of shearheads used with the super jack at column E3. 2 TJnconfined 196 872 Jack rod broke above lifting nut due to flexure and axial load. east side. The shearheads used in these tests were designated as type P50. COMPARISON OF RESULTS OF TESTS WITH PROJECT SPECIFICATIONS No attempt was made to evaluate the design of the completed structure. as lateral restraint provided by the bars welded to the lifting angles was much stiffer than that which could be provided by confinement by concrete. The specimens confined by concrete (4 and 5) had capacities between these two extremes. The capacities of specimens 4 and 5 were considered good approximations of the capacities of the lifting assemblies in the actual structure. 227 1. The capacity of specimen 1 given in Table 1 represented a lower bound for capacity of the lifting assembly. No concrete was found that did not meet strength requirements set forth in project specifications. The ca- pacity of specimen 3 represented an upper bound of the capacity of the lifting assembly.8) performed in a manner similar to the lifting angles on the P50 shearheads. within the limits of the parameters of loading used in the tests. all rights reserved. concrete components of the subassembly slammed together. (3) 1 Unconfined 165 734 Bottom nut slipped off lifting angle. concrete 5 Confined 198 881 Bottom nut slipped off lifting by angle. the capacity of a lifting assembly could have been expected to be at least 170 kips (756 kN) but not much larger than 200 kips (934 kN) in the L'Ambiance Plaza structure. Constr. TABLE 1. Summary of Lifting Assembly Test Results Failure Load Test Condition of number shearhead . A summary of the results of tests of lifting assemblies is given in Table 1. The support conditions for the shearheads tested did not duplicate exactly the support conditions of the shearheads in the actual structure. by Universidad Politecnica De Valencia on 06/09/15. but rather. . 3 Unconfined. regardless of the size of the jack used to lift the shearhead. In general. 4 Confined 201 894 Bottom nut slipped off lifting by angle.010 Jack rod broke above lifting nut stiffened due to flexure and axial load. On the basis of these tests. as the shearhead was essentially unconfined. For personal use only. These tests indicated the flexural and torsional flexibility of the arm channels and lifting angles was the primary factor influencing the strength of the lifting assembly. Copyright ASCE.

In the vicinity of column E4. . Perform. The horizontal curvature of cables near column E4.3 m) and 2 ft (0. details of placement of post- tensioning tendons did not satisfy requirements established by the Ameri- can Concrete Institute. On April 23. east-west tendons are splayed in such a way that they are separated from the column by over 5 ft (1. Constr. Details of Placement of Post-Tensioning Tendons At several locations in the structure. Some welds did not meet the requirements of the AWS Dl. The yield strength of all specimens of strand tested exceeded the design value by Universidad Politecnica De Valencia on 06/09/15.000 psf (670 kPa). This tendon layout does not satisfy the assumption that banded tendons pass within the shear perimeter of the support. Information indicated the shop welds were visually inspected and the column splices were ultrasonically inspected.1987. Inspection require- ments for field welds were the same as those for shop welds. Fillet welds were to have been tested by liquid penetrant inspection or magnetic particle inspection.52 m). at the testing agency's option. all rights reserved. Types and locations of defects and work required and performed to correct deficiencies were to be recorded. Facil. Foundations Most footings rested on a layer of disintegrated rock or local material between 1 ft (0. However.8 is also questionable. All samples of post-tensioning strands recovered from the collapsed structure had yield and ultimate strengths that satisfied requirements for the type of strand specified for use on the project. No evidence was found during the investigation to indicate that the condition of the footings contributed to the collapse. there is no indication that this deviation from approved construction procedures contributed to the collapse. The erection of slabs had proceeded to five equivalent story heights (level 1 to level 6) above the cast top of the shearwall at level 1." This requirement was not met at the time of the collapse of the structure. Weldments Projects specifications required that all welds be visually inspected. Complete penetration welds were to have been tested by radiographic inspection or ultrasonic inspection. the shearwalls on the south side of the building between the ground level and level 1 were placed.8. This fill material was certified by an independent testing laboratory to be capable of supporting 14. It is not possible to determine the frequency at which this occurred throughout the structure based on the limited amount of weldment testing done during this investigation. Copyright ASCE. 1988. ACI Committee 423 ("Recommendations" 1983) specifically discourages the 75 J. Downloaded from ascelibrary.6 m) thick. or that the footings would have settled excessively under design loadings. The results from tensile coupon tests and fractographic analyses raised questions about the quality of some welds in the structure. The structural drawings note that the shearwalls were required to "be cast so that no more than three equivalent floors of height of lift slab structure shall be advanced above cast top of shear wall. Lateral Bracing Shearwalls constructed as the slabs were lifted provided lateral support after the concrete achieved a certain strength.1-83 ("Structural welding" 1983).2:58-79. For personal use only.

000 psi (20. The jack at column E4. (1. A jack force of 190 kips (845 kN) acting on the type X51 shearheads at column E3. made it the most heavily loaded column in the west tower.8 could have developed a lifting force of approximately 298 kips (1. and 11 along column line E in the west tower.330 kN). even when operating at the maximum line pressure. during erection of the slabs.'although their use is discouraged.2 MPa). all rights reserved. For personal use only. which the jack easily could have produced. The indentations of the top of column E4.500 psi (17. Whether loss of support of the slab at column E4.8. it is possible the 300-kip (1. G3. It was determined that increasing the elevation of the slab package at column E4. the jack at column E3. Downloaded from ascelibrary.8 could provide. and G4 carried less load than did E4. Assuming an operating pressure of 2.8 would have required a force of about 190 kips (845 kN). 10. and that the maximum line pressure attainable was approximately 3.8 required a force larger than the jack at column E4.8. which.2:58-79.50 in.8 by one full jack stroke of 0.8 or E4. Facil.8 due to slabs 9.8 could have been used to effect this adjustment.334-kN) jack at column E3. this could have resulted either from fracture of a jack rod or from a lifting nut slipping out from under a lifting angle. At the time of the collapse temporary wedges were being installed under floor slab 9 at column E4. Employees of the lifting subcontractor testified that the line pressure normally supplied by the lifting system power unit ranged from 2. A full-stroke upward displacement at column E3. Lifting nut impact marks were found on each side of the web of this column at approximately 52 in. .8 would produce significant rotations of the 76 J. This trans- lates to a jack force normally between 137 and 149 kips (609 to 663 kN) and a maximum available force of 179 kips (796 kN) for a small jack. observations of debris from the collapse.8 was the cause of the collapse or simply one of its effects cannot be established with certainty. Perform. At least one other sequence of events could have produced a similar result. specifically at column E3. If an upward adjustment of the slab at E4. 1988.9 to 17. As tests of the lifting assembly indicated.2 MPa) while lifting a package of three slabs.7 mm) from its as-cast position would require a jack load of 201 kips (894 kN). (12. Copyright ASCE.8 suggest that one of the jack rods lost its load. The investigation did not indicate that any of these details contributed to initiation of the collapse. Constr.7 MPa).300 to 2. It is probable that installation of these wedges required some raising or lowering of the slab package at that column.8 could not have developed the force required for an upward adjustment of 0. (12. and 11 was 152 kips (676 kN).32 m) below the top of the column. use of horizontally curved. SCENARIO OF COLLAPSE Based on the testimony of witnesses. 10. E3.50 in. indicating that at this column the lifting nuts "kicked out" from under the lifting angles of the shearhead at slab 9. the most probable cause of the collapse of the L'Ambiance Plaza building was failure of the lifting assembly to support the package of floor slabs 9. banded monostrand tendons unless special reinforcement transverse to the axis of the tendons is provided.8. Columns E3.7 mm) from the as-cast position. causing the jack to roll off the column top and chamfer the edge of the by Universidad Politecnica De Valencia on 06/09/15.500 psi (15. The nominal reaction at E4.8. and review of possible failure mechanisms.

11. 10. Marks on the bottom surfaces of the lifting angles of the shearhead of slab 9 at column E3. The absence of scrapes and gouges on the column tops along these two lines and the extensive damage to the tops of columns along column line C 77 by Universidad Politecnica De Valencia on 06/09/15. Probable Sequence of Collapse lifting angles. and 11.2:58-79. loss of a jack at either E3. all rights reserved.8 Loss of support of slab and reduction of ®. At approximately the same time. For personal use only. Copyright ASCE. post-tensloning along column line E » Shearwall <^ Flexural failure of slab . Racking forces produced by movement of the slabs could have then caused expulsion of wedges under floor slab 12 along column lines G and H.negative moment n\ \ Pour strip <§> Initiation of failure of roof and level 12 <§> Initiation of failure of East tower FIG. it is likely that the remaining jacks along column line E shed their load as one or more lifting nuts kicked out when loads were redistributed. The probable sequence of collapse is indicated in Fig. Q^-ir- n ^ <t> Loss of support of slab at E3. the two slabs at the top of the building (12/roof) remained intact. Facil.8 or E4. A yield line analysis showed that floor slabs would remain intact after loss of support at either column E3. 10. 10. During this initial phase of failure of slabs 9. The loss of support of floor slabs 9. 11. Analysis indicated the slab would lift off one wedge at each column under small lateral displacements. as indicated by the curvature at the very top of columns along column line C. Therefore. slabs had been temporarily wedged and the lifting nuts had been removed from the jack rods at the time of the collapse. with the exception of the shearheads in column lines 1 and 6. An estimate of how the initial failure propagated through the structure can be made based on eyewitness accounts and observations of the debris. Constr. Along column lines G and H.8.8 showed that lifting nuts did slip off the lifting angles of the shearheads at that location. .8 would have led to failure of the lifting assembly at the other column as loads were redistributed. whichever occurred first.8 or E4.8. and 11 along column line E is in agreement with observed marks on the stage IV column extensions and on the shearhead lifting angles. 1988. Perform. the shearheads released cleanly and slid straight down. After loss of support of slabs 9.8 or E4. The shearheads in those two column lines scraped the columns very little during the collapse and.8 and E4. IJLXAJk Downloaded from ascelibrary. and 11 at column E3. the slabs failed in flexure along the length of the west tower just to the north of column line G. possibly leading to kick out of one or both of the lifting nuts.

Fourth. With loss of anchorage of post-tensioning strands in one or more floor slabs. The failure most probably began at column E4. CONCLUSIONS The following conclusions are based on the laboratory and field tests. First. The east tower collapsed due to some combination of forces transmitted to it by the west tower collapse. debris from the west tower could have struck the columns along column line 7. The reserve capacity against lateral instability was small. 4. The most probable cause of the initiation of the collapse of the east Downloaded from ascelibrary. forces exerted by debris falling from the west tower could have caused lateral displacement and instability of the east tower. Counterclockwise twisting of several of the columns in the lower levels of the east tower and cracks in the shear wall on the south side of the east tower pointed to this failure mode. 1988. For personal use only. Loss of wedges supporting the upper floor slabs would follow. Third. causing all slabs beneath them in the west tower to fail. the progression of failure could have been similar to that described for column line E in the west tower.8 as excessive deformations of the lifting angles of the shearheads allowed the jack rods to slip off the lifting angles in the shearheads supporting the package of three slabs. damage to the unbonded post-tensioning tendons caused by falling debris from the west tower. These slabs fell. Four mechanisms appear possible. suggest that failure of the upper two floor slabs (12/roof) initiated in the south sector of the west tower. There were a number of deviations from the project plans and specifications in the structure as built. but the investigation indicated these deviations did not play a significant role in initiating the collapse. Perform. tower is less clear. The most probable cause of the collapse was failure of the lifting system in the west tower during placement of a package of three floor slabs. witness interviews. falling debris near the west end of column line D could have damaged the tendon anchorages in the east tower floor slabs at this location. all rights reserved. and review of the project docu- mentation: 1. This failure sequence is consistent with accounts given by several eyewitnesses. computer analyses.2:58-79. that lateral instability was the initial cause of the 78 J. Constr. . This failure mechanism was duplicated in labora- tory experiments. 3.8 or column E3. The failure propagated as the remaining jack rods along column line E became overloaded and broke or slipped off the lifting angles and the package of three slabs failed in flexure and shear. This loss of support could propagate through the floor slabs in the east tower in a manner similar to that discussed for the slabs at levels 9. the pour strips between the floor slabs of the two towers could have transmitted destruc- tive forces from the west tower into the east tower. 2. Second. Facil. by Universidad Politecnica De Valencia on 06/09/15. Copyright ASCE. or lateral instability caused by falling debris from the west tower. 10. and 11 in the west tower. causing the east tower to move laterally. It does not appear. thereby allowing temporarily welded wedges to fall out and initiate collapse of the east tower. The quality of materials in the structure was generally in accordance with the project plans and specifications and did not play a significant role in initiating the collapse.

ACI Committee 318. Miami. Downloaded from ascelibrary. W. Lew. Detroit. and state of Connecticut Building Inspector Leo Belval are acknowledged for their invaluable support. American Concrete Institute." NBSIR 87-3640.C. The Department of Public Safety of the State of Connecticut." NBSIR 78-1578. Schupack. G. all rights reserved. It is unlikely that differential foundation settlements caused the building to collapse. Thomas by Universidad Politecnica De Valencia on 06/09/15. Suarez Engineers Inc. H." (1983). John Stanton. Washington. For personal use only. "Investigation of construction failure of reinforced concrete cooling tower at Willow Island. ACI318-83. Romas Bossone. ANSI/AWS Dl. 5." (1983). Sep. Connecticut Test Borings. L. the Neil S. et al. Schnabel Engineering Associates. Marshall. Copyright ASCE. .1-83. 16 pp. Inc. "Structural welding code-Steel. collapse. 79 J. E. ACKNOWLEDGMENTS Staff of the Occupational Safety and Health Administration provided support throughout the investigation. It is believed that the majority of any settlements would have occurred early in the erection process. Fla.. Mich. (1979).. West Virginia. Detroit.. and J. Joseph Normand. D. American Welding Society.C.. C. Barrett and Sons provided valuable field service and assistance. (19 mm) could have occurred during construction. Facil. American Concrete Institute. Y. J. (1987). and Kang Yi are gratefully acknowl- edged. APPENDIX I. and NBS technicians are especially acknowledged. D. National Bureau of Standards. F. "Recommendations for concrete members prestressed with unbonded tendons. Gross. Ronald Morin. Hendrickson.. It is unlikely that lateral earth pressure acting against the basement wall on the north side of the structure caused the building to collapse. It is unlikely that the horizontal jack used to plumb the structure initiated the collapse. C.3R-83. The efforts of R. Constr. Washington. Lieutenant Edmund Dunstone. ACI-ASCE Committee 423.2:58-79. "Investigation of L'Ambiance Plaza building collapse in Bridgeport. Perform. Pat Clark. Yokel. 1988. ACI423. Mich. S. Many members of the NBS staff worked on this investigation." (1983). Yancey. National Bureau of Standards. REFERENCES "Building code requirements for reinforced concrete. Nov. Culver.. Data from the subsurface explorations indicated differential foundation settlements not exceeding 3/4 in. Moreton Company.. William Freeman. D. M. John Miles. and E. et al. Edward Eagan. Connecticut. 6.