L'Ambiance Plaza was to be a 16 story structure with 3 parking levels and 13 apartment levels. It was composed of two offset rectangular towers, separated by a construction joint at a central elevator lobby. Each tower was approximately 63 ft x 110 ft (19.2 m x 33.5 m) in plan. Typical floor to floor height was 8 ft - 8 in. (2.64 m). The structural frame had steel columns and 7-in. (178 mm) thick post-tensioned concrete flat plate floors.

The flat plate floors were designed to be constructed by the "lift slab system." The design used unbonded plastic-sheathed post-tensioning tendons in each direction. Tendons were banded within a strip along the column lines in the east-west direction and were distributed uniformly over the width of the bay in the north-south direction. Mild steel reinforcing was limited, consisting of small mats of #5 bars in the top of theslab at each column, 2 - #4 or #5 bars along slab openings and edges, and miscellaneous bars at locations of minor extent in, or adjacent to, pour strips and shear walls. Steel channel shear collars were provided at all column locations for load transfer and lifting operations.

Lateral resistance in the completed structure was provided by two shear walls in each direction in each tower. These extended to two levels below the roof. For the top two levels, lateral resistance relied on frame action with rigid joints between steel columns and post-tensioned slabs.

The building was founded on spread footings intended to be on rock. Ground level was up to 2 stories higher on the north side than on the south side. The unbalanced lateral earth pressures were supported by the building foundation walls and the shear walls.

The Contract Documents were prepared by an architectural firm engaged by the owner-developer of the project. The architect retained a structural engineering firm that developed the basic structural design for the project and prepared the contract structural drawings and specifications. The contract documents recognized that the structural frame would be constructed by the lift-slab method. They did not contain certain important structural requirements and details because the project specifications required the contractor who was responsible for lift-slab construction to develop the slab post-tensioning design and details and the column connection design and details.

The contract drawings show the type of system intended, namely steel columns and 7-in. (178 mm) thick flat plate floors with draped unbonded post-tensioning tendons, arranged in column strip bands east-west and uniformly distributed over the bay lengths north-south. No design was provided for the required post-tensioning reinforcement. The responsibility for this was given to the Contractor. The drawings also show the size and location of large slab openings for elevators and stairs and the arrangement of shear walls and other walls that are to be cast through floor slots after the slabs are lifted. Based on a schematic typical detail, the drawings indicate that east-west post-tensioning bands may be given horizontal sweep (offset) to accommodate openings, and that certain mild steel reinforcement is to be provided around the edges of slab openings. No information is given about the specific arrangement or design of post-tensioning reinforcement at the bays that contain large openings for elevators and stairs. Again this design responsibility was given to the contractor.

The structural drawings show design requirements for shear walls. A note on a drawing states that: "shear walls shall be advanced to within 3 floor levels of the slabs being lifted" during construction. A schedule of column sizes, to be reviewed by the lifting contractor and increased, if necessary for stability and strength during erection, is included on the drawings. Minimum requirements for moment transfer through lifting collars also are given.

Construction was by the lift-slab method. After the first three-story sections of the steel columns were erected, the entire set of 16 flat-plate floor slabs was constructed on the ground. Each floor was cast on top of the next lower floor with a bond breaker between each slab. Next, the slab lifting apparatus was installed. It consisted of a hydraulic jack at the top of every column and a pair of lifting rods extending down from the jacks to lifting collars cast into the slab at each column opening. The lifting collars, which also act as shear heads, are comprised of 5 in. (127 mm) steel channel boxes with two opposing slotted angles projecting inward between the column flanges to provide seats for the lifting rods. No studs or other embedded components are welded to the steel channel boxes to enhance anchorage of the shear collars to the concrete.

After all 16 slabs were cured, they were raised progressively in stacks (i.e. 2 or 3 slab packages) to their interim or final positions using the hydraulic lifting apparatus on the top of each column. When the top most slab package reached the top of the first 3-story section of column, it was "parked" temporarily. Additional 2-story column lengths were added and lifting proceeded in stages. In all, seven lifting stages were planned for L'Ambiance Plaza. The roof and 16th floor were lifted simultaneously as the top stack of two slabs. Other floors were lifted simultaneously in stacks of 3 slabs. Prior to reaching it's permanent elevation, each stack of slabs was supported temporarily on steel wedges inserted between the bottom of the lifting collar and the top of column flange plates. While in the temporary "parked" position, wedges were tack welded to the flange plates. Individual slabs were permanently attached to columns by direct bearing between lifting collars, wedges, and column flange plates and by welding between the underside of lifting collars and wedges, wedges and column flange plates, and the top of shear collars and seal plates on the columns. The spaces between the collars and the columns were then filled with concrete.

The Contract Documents imposed on the contractor substantial structural design responsibility for the portions of the structural frame involved in the lifting process. The result is that responsibility for the structural design of the building was fragmented among a number of different organizations. The structural engineer-of-record developed the conceptual design of the structural system, including the type of prestressed concrete lift-slab system, column and major opening locations, shear wall system, and slab thickness. The drawings do not show specific post-tensioned reinforcement. The design, as well as the detailing, of the post-tensioning system is left to the contractor via his specialized sub-contractors. The design of the steel columns, when governed by the conditions during lifting, and the design of the shear head - lifting collars, together with their connections to the columns, also are the contractor's responsibility.

In the actual performance of the contract, Subcontractor A, a sub-contractor to the General Contractor, was responsible for the design and construction of all components of the structural frame that were involved in the lifting process. However, Subcontractor A subcontracted design and preparation of detailed tendon drawings for the post-tensioning system to Sub-subcontractor B. The shop drawings prepared by Sub-subcontractor B are the only drawings that show the number, arrangement and forces for all post-tensioning tendons in the project. The post tensioning calculations and shop drawings were reviewed by the engineer-of-record. The review stamp stated "checking is only for general conformance with the design concept of the project and general compliance with the information given in the contract documents."

Although the Contract Documents require the contractor's design to be performed by a registered professional engineer, who was to seal the drawings showing his design, only the contract drawings by the engineer-of-record were sealed with such a stamp.

Subcontractor A also subcontracted preparation of shop drawings for mild steel to Sub-subcontractor C, a subcontractor for furnishing of mild steel reinforcement. These drawings were reviewed by the engineer-of-record. This reinforcing appears to be based on the contract drawings, with some modifications made by the engineer-of-record during shop drawing review.

Subcontractor A subcontracted furnishing of structural steel, except for lifting collars, together with preparation of related shop drawings, to Sub-subcontractor D. These shop drawings were reviewed by the engineer-of-record.

Finally, Subcontractor A subcontracted lifting operations to Sub-subcontractor E, who also designed and furnished the lifting collars that were used as shearheads in the slabs. Shop drawings for lifting collars were not available during the writer's investigation.

A team of engineers from the National Bureau of Standards (NBS), investigated the collapse, as did many other engineers (including the writer) who were engaged by various principals and legal representatives of parties in the ensuing litigation. NBS published a report (NBS 1987) on its investigation that is summarized in (Scribner and Culver 1988); the following description of the status of construction at the time of the collapse is based on information given in (NBS 1987), as well as in (Scribner and Culver 1988).

At the time of the collapse, slab lifting had reached the top of stage IV (the fourth of the two and three story sections of columns), with the uppermost roof slab 83 ft-6 in. (25.5 m) above the foundation. The six lowest slabs already were positioned at their permanent locations in the west tower and the nine lowest slabs were at their permanent locations in the east tower. The remainder of the slabs were parked in temporary locations above the permanently fixed lower slabs. Also, the construction of shear walls was two to three levels behind the specified three level minimum limit below the top of the highest slab.

Based on various eyewitness accounts (NBS 1987), the most likely location of initial failure was within the levels 9, 10 and 11 slab stack near line E between lines 6 and 3.8 of the west tower. However, several eyewitnesses placed the start of the collapse high in the building at the west end of the west tower. This failure was followed very rapidly by total collapse of both towers. Neither the precise sequence of collapse nor the triggering mechanism can be ascertained positively from examination of the rubble or from information given by survivors.

The source of the post tensioning design in the elevator bay is not given by any information available to me. No special design information is shown in the contract structural drawings for this bay, probably because the Contract Documents require the contractor to design the post-tensioning system for the slabs. With the exception of the horizontal offsets or sweeps of the east-west tendon bands that conform to the column line offsets, the shop drawing show the same typical tendon size, spacing and layout in this bay as used in the typical bays which do not have large openings and column line offsets.

The banded arrangement used for east-west tendons for typical bays with approximately regular spacing of columns reflects recommendations given by the Post-tensioning Institute (1977). Each line represents one to five 1/2 in. (12.7 mm) diameter post-tensioning tendons. Note that the regular arrangement of columns is interrupted by a large opening to accommodate the elevators. The engineer-of-record specified that two columns be provided, one to the north and one to the south of the elevator opening, both offset from column line E. A shear wall was to be constructed at the west edge of the elevator opening but it was not in place at upper levels at the time of the collapse. To maintain the banded tendon concept for east-west post-tensioning, the shop drawings prepared by Sub-subcontractor B specify that the tendon band on Column Line E split and sweep in a horizontal plane around the elevator opening to the two nearby columns.

The shop drawings contain a serious deficiency in the tendon drapes adjacent to the elevator opening. Tendons normally are placed high in the slab at support points and low in the slab at mid-spans between support points. This would require the east-west banded tendons to be high at columns and low at mid-span and the north-south distributed tendons to be high over the banded east-west tendons and low at mid-span. However, the north-south tendon drape that is specified for typical bays also is used for the region where the tendons on column line E are divided and splayed; the north-south tendons in the bay between column lines 5 and 6 are high on the continuation of column line E and low at mid-span between and C or G, in disregard for the actual location of the band beam support created by the divided east-west tendons. The effect of this arrangement is that at about the mid-span along the west edge of the elevator opening (at E-line), the post-tensioning stresses in the slab add to the stresses due to gravity loads, rather than compensate for them, as in a proper design.

In the vicinity of the elevator opening, the mild steel shop drawings show only two #5 bars at the perimeter of the opening. Therefore, for north-south flexure in the bay adjacent to the elevator opening, the tension zones at the bottom surface of the slab near midspan and the top surface over the supporting banded tendons effectively are unreinforced.

Tendon band layout in slabs. The banded post tensioning tendons in the east-west direction are centered on interior columns and are placed entirely on the interior side of some of the exterior columns. Several details of the post-tensioning layout raise questions about the adequacy of local shear and bending strength in the transfer region between slab and columns. One of these questionable details is the placement of the entire exterior tendon bands on the interior side of certain exterior columns combined with only a light mat of top mild steel reinforcing at each column (as per ACI Code minimum percentage) and a single top and bottom #5 bar along the slab edge. Because this band of tendons is placed inside the interior face of the exterior columns and the slab bending capacity perpendicular to the slab edge is limited, transfer of a large portion of the column load is by torsion and shear. The torsion produced by this type of eccentric banding creates high shear on the interior edge of the shear collars. Embedded studs or other anchoring devices would have increased the size of the effective shear cone and would have improved anchorage between the shear collars and the concrete, but none were provided. During the collapse, most lifting collar frames were stripped clean of concrete, contributing to the brittle nature of the failure.

Another defect in the tendon band arrangement is the large distance between column E4.8 and the adjacent splayed tendon bands. This arrangement produces large downward reactions from tendon drape at locations much further from the column than in typical bays. These tendons have so much sweep that they are located well beyond the shear zones on either side of column E-4.8. This seriously compromises the load transfer mechanism and the ductility of the system. This arrangement also violates the requirement in Section 18.12 of ACI 318, 1983 and 1989, that a minimum of two tendons pass through the shear zone of every column.

These defects in tendon layout reduce or negate the ability of the tendon system to support load by "netting" or "membrane action" in the event that the otherwise unreinforced slab becomes extensively cracked. This increases the likelihood of brittle and rapid progressive collapse should failure occur due to a localized defect.

The curvature associated with the sweep of the tendon bands near Col. E4.8 also produces tension in the slab transverse to the tendon bands. Bonded reinforcing was not provided to resist these significant tensile forces, as required by the ACI code.

No information is contained in the reports of investigations cited elsewhere herein, nor came to light in the writer's interviews with several key employees of the general contractor to indicate that changes were made in the tendon layouts shown on the shop drawings. The actual location of tendons in the slabs could not be determined from observations after the collapse because of the slabs were essentially reduced to rubble as a result of the collapse. Because of this, it is possible that the incorrect tendon drapes shown on the shop drawings were changed prior to placing concrete in the field. However, in the absence of information about any changes of this nature in the investigations cited herein and considering that the engineer of record did not regularly observe construction at the site, it seems highly unlikely that any tendon drape corrections were made, and it is reasonable to assume that the tendons were installed in accordance with the shop drawings.

The original basis of the shear design of the collars could not be determined. After the collapse, the collar design was evaluated by NBS in a series of tests of individual lifting collars subject to various restraint conditions that were intended to approximate the confinement provided by the concrete around the collars in the actual structure (NBS 1987). These NBS tests did not simulate certain structural actions such as the possible stiffening effects from collars and slab concrete in the upper two slabs in a three slab stack. The tests modeled several positions of the lifting rods within the slot at the rod-collar connection, providing different amounts of deviation from the design position. Based on these tests, NBS concluded that the connection between each lifting rod and its support angle on the lifting collars had insufficient strength and that the angle could deform sufficiently to allow a rod to slip out of the lifting angle slot, initiating a collapse. NBS concluded that the lifting collar design was improper and that failure at a lifting collar was the most probable initiating cause of the collapse.

Additional tests were performed by the lifting collar manufacturer. These tests used different conditions of restraint and also used full seating of lifting rods. Their results show that the collars had significantly greater strength than the probable load being lifted at the time of the collapse, and that failure at a lifting collar was not a probable cause of the collapse (Wonder 1988). Also, the lifting Sub-subcontractor contends that this collar design has been in successful use for over 100,000 lifting operations over a period of 17 years (Wonder 1988). Finally, both sets of tests showed that the lifting assembly did not have a minimum factor of safety of 2.5, as required in ANSI A10.9 Standard for Construction and Demolition Operations Concrete and Masonry Work Safety Requirements. This standard also prohibits anyone from being under the slab during jacking operations.

Prior to installation of the cast-in-place shear walls, the structural frame had to provide adequate lateral strength and stiffness to avoid any danger of collapse from lateral wind load, and from the effects of out-of-plumb columns under vertical load (i.e. frame instability under vertical load or combined vertical and lateral loads.) No guys or other temporary bracing were used to provide additional strength and stability for the temporary conditions that occurred during construction. When a new lift of columns was installed, the first step of the next stage of lifting was to raise the roof and top level slab package to a point near the top of the new sections of columns. During this operation, the columns resisted lateral forces and provided stability as vertical cantilevers for their full unsupported height, which was as much as three stories above the top stack of slabs temporarily supported on the previous column section. After the top-most slab stack of slabs was parked temporarily near the top of the upper lift of columns, the slabs and columns above the highest level of completed shear walls acted as a frame. Some of the slab-column connections in the frame were temporary, comprised of the slab supported on wedges that were tack welded in place. Others were welded in their final configurations. The number of final connections that were complete at any point in time increased as the work progressed.

At the time of the collapse, the uppermost stack of slabs on the west tower was about three levels above the highest level of welded connections. The top of the highest shear wall at one end of this tower was four levels below the highest lifted slabs and six levels below the highest slabs at the other end. Also, Stage IV columns in the west tower had a height of about 22 ft (6.7 m) without lateral support between levels of parked slabs.

During the time immediately before the collapse when shear wall installation was incomplete, the stability of the structure above the incomplete shear walls depends on the strength and stiffness of the steel column/concrete slab system acting as a frame, with unwelded connections at all temporarily supported slabs. However, because there is no positive connection between column and slab to develop frame action, and moment transfer for frame stability could be provided only by moments developed by the weight of the supported slab acting eccentrically at the support seats on opposite flanges of a supporting column, the stability of the frame was tenuous. Also, the degree of strength and stability obtained from these restoring effects diminishes as the frame deflects laterally. The stiffness of the slab-column connection becomes zero when the frame has deflected sufficiently to cause the shear collars to lift off one of the wedges (i.e. with sufficiently high lateral load). In view of this, the stabilizing mechanism was unreliable and should not have been used for design purposes, even for temporary construction conditions.

The potential for slippage of wedges prior to permanent welding produced additional uncertainty about the safety of the temporary connections, particularly under reduced vertical load associated with frame action. Slippage is prevented by proper tack welding at the time the wedges are installed, an operation that requires careful on-the-spot checks for quality assurance during construction. Also, tack welds may break due to load changes associated with large frame distortions.

The actual moment resistance and connection stiffness provided by the completed column-slab connection with final welding is difficult to determine by calculation. Tests have been performed that show substantial capacity for moment transfer in this type of joint (Hawkins 1981).

The contractor who erected the structural frame did not follow the contract requirement to limit the advance of the uppermost lifted slabs to not more than three levels above the level of completed shear walls. Also, he did not provide guys for lateral support, an alternative allowed in the Contract Documents if shear wall installation was to be below the required minimum height.

Various investigators identified additional construction deficiencies after the collapse. These included (NBS 1987):

• Use of broken rock fill under certain footings (instead of direct bearing on rock) without approval of the structural engineer-of-record.
• Placement of backfill against the north foundation wall prior to completion of the lateral force resisting system.
• Plumbing of the west tower frame with a hand jack placed against the east tower.
• Variations in control of lifting operations, particularly the failure to engage one of the slabs at one column during one of the slab lifts in the east tower.
• Defective welding of permanent column-to-slab connections and column-to-column connections, including both field and shop welds.

Although concrete had been placed in slabs during sub-freezing weather, tests and examinations of cores extracted from slabs revealed no deficient concrete strength, no evidence that concrete had been frozen when fresh, and no other problems with concrete quality.