Combined Design and Construction Deficiencies
Often in accidents it is found that design flaws and construction compromises combine to result in collapse although separately they might not have been that much of a problem.
Hyatt-Regency Walkway Collapse
A classical example of a combination of design and construction error is the Hyatt Regency Walkway failure at Kansas City, Missouri, USA, on 17 July 1981, [15].
On that fateful evening, during a 'tea dance', at 7.05 pm, the fourth floor walkway of the 4-storey high atrium in the 40-storey Hotel collapsed with excited foot-tapping couples, on to the second floor walkway, dragging it and its occupants to the bottom dance floors, leaving 114 dead and more than 200 injured. (Figure 10, left two parts.)
Fig. 10. Left pair - Hyatt Regency Walkway, before and after collapse; Middle - Nut at upper walkway; Right - Static analysis of single vs. double hanger rods
The hotel was opened in July 1980, with three walkways suspended from the ceiling by hanger rods, one set of rods on one side of the atrium for floors 2 and 4, and a second set on the other side for floor 3, as shown in Fig. 10 (Left). Both floors 2 and 4 were to hang from a single rod, with each floor load being supported by a washer and nut under the respective floor, as indicated by the arrows in the first and third parts of Fig. 10.
But the contractors felt that to place a nut in the middle of the single rod to support the fourth floor was too much trouble, and proposed two rods, one going from the ceiling to the fourth floor and the other going from the fourth to the second floor, as shown in the third part of the figure. Whether the contractors got approval for the change from the designers was not quite clear, but the contractors went ahead and implemented their change.
The problem was that while the single rod, inconvenient to erect though it was, would have been loaded with only a single floor load at each of the second and fourth levels, the two-rod solution, while simpler, transferred the second floor load also on to the same washer and nut holding up the fourth floor, thus in effect doubling the design load on it, which became the direct cause of the collapse during the dance.
Although this could be referred to as a design flaw in that nobody checked the design capacity of the revised design, it was triggered more by the contractors wanting (and rushing) to change the design without proper re-analysis.
The error that killed 114 people would not have occurred if the two-rod problem had been given to a first-year engineering statics student as a home assignment - so simple that veteran engineers did not think to check! Moral of the story: Nothing is too simple to avoid checking!
Rebar-cage Collapse
I was invited by the authorities to investigate for the prosecution, the collapse of a rebar-cage used for 3m and 5m R.C. base slabs, which killed two workers and injured 29 others. I investigated the accident and discovered a number of design and erection deficiencies, [16].
Bar-Chairs:
The main bone of contention were the supports for the upper layer of slab reinforcement by means of 'bar-chairs' made from 32mm steel rebar bent into a hat shape. (Figure 11, left.)
Fig. 11. Left - Rebar cage collapse (Inset: Bar-chair and rebars); Right - Bar-chairs
Bar-chairs are in common use in many parts of the world, but they are generally less than half a metre tall, with legs that are longer than the height (Fig. 11, right, top), so that there is no problem with stability.
However, the bar-chair system that collapsed was for 3m and 5m thick slabs with 2.49m and 4.35m chair heights, and their foot projections were only 500mm, much smaller than the heights, as shown in Fig. 11, right bottom. The stand-alone stability of both chairs was greatly in question.
Erection drawing consisted of free-hand sketches as in right bottom inset of the figure. Another design consultant had suggested bracings to ensure stability, but these suggestions were not carried through as complete designs, did not have Safe Work Procedures, and were not implemented consistently throughout the system.
I analysed the chairs from fundamental structural analyses principles. As there were only six joints and five members, simple moment distribution would have given the answers. Computer analysis with the stiffness method was used only for speed and convenience.
Asymmetry Problems:
The asymmetry in the bar feet led to asymmetrical stress and deformation behaviour. Moreover, the 3m bar chair had a mid-span support which was not vertical but inclined because the contractor used longer props he already had but did not want to cut up. This led to further asymmetry, and tendency to move in the direction of the slope, or alternatively, if such longitudinal displacement was restrained, a reaction force at the restraint. (Figure 12.)
Fig. 12. Effect of the inclined leg of the bar-chair
Design Error:
Another of my main concerns was the buckling capacity of the bar-chair legs used to support the upper rebar layer in the cage.
The following formula was found used in the original design for the critical column load:
Pcr = 2π2EI/L2
which is the well-known Euler Formula, with an extra '2' attached, doubling the result. (E is Young's modulus, I and L are the moment of inertia of the bar and its unbraced length.)
For '2' to reflect increased capacity, it would imply end-restraint factor k of √(1/2), i.e. 0.707, which would correspond to Code case of both ends ‘fixed’, and theoretical case of one end pinned and other fixed in position and direction, as marked by asterisk in Fig. 13.
If it was indeed the designer's intent to provide supports at the two ends which would enable this doubling of capacity, it was not mentioned in the design, and there was no method statement (SWP) or recommendation of any special device for this purpose.
Then, pinned ends being the most basic and most realistic assumption for temporary structures at site, the ‘2’ was in error, which over-estimated column capacity by 100%.
As the design load factor was 1.5 (at that time, since then revised to 2.0) obviously, the column was under-designed by 25%.
When the testifying designer was pressed to explain what the significance of the '2' was, he admitted it was an "oversight".
Fig. 13. Buckling of bar-chair
It was thus a design error, and by itself, might also have contributed to the collapse.
Apart from all these, as part of my forensic investigation, I had conducted numerous finite element analyses of the entire grillage and come up with a lot of interesting leads pointing to other design flaws, with many factors of safety in the design working out to less than 1.5.
Additional Analyses requested by the judge:
After my first day's expert witness testimony in this case, the judge wanted me to come up with information on the effects of the loss of one or more such support struts, particularly whether and how such losses would result in complete collapse.
Needless to say, I worked most of the night with the computer, and came up with sufficient results to demonstrate that loss of one or a few of the supports would not have resulted in collapse - provided the system was suitably anchored in the lateral direction.
My presentation in court next day included (of course in colour PowerPoint) of a fairly exhaustive analysis of loss of the central 1, 5 and 9 legs (Fig. 14), and (alternatively) the corner 1, 3, and 6 legs. The reactions simply re-distributed themselves at the still active legs - in what I called the 'ripple effect', surviving at least for the first two rounds of losses.
The spans for horizontal rebars would have increased corresponding to the support losses, but the rebars simply sagged more without failing, although any other spanning material might have failed under the increased span.
Again the following day, after this presentation and some further testimony, the judge asked me to explain (next morning) the effects of in-line supports versus staggered supports.
Again, I went to work with my computer software, to get usable results, as in Fig. 15. The judge recorded appreciation of my contributions both times.
But it would be wrong (and unwise) to think that such extra requests (more like commands when coming from a judge) are an indication of your high professional stature or that you are doing a favour to the court. Far from it. It may simply mean that while examining an expert the Court is claiming its right for more information – or even that you were not clear enough or complete enough in the first place.
Fig. 14. Local failure and ripple effect
In fact, the more you say, the deeper you may be getting into trouble! So, it was no feather in my cap that I was able to come up with useful results at short order. It would have been egg on my face and a real tragedy if I had not come up with the answers.
But it would be wrong (and unwise) to think that such extra requests (more like commands when coming from a judge) are an indication of your high prmso-bidi-font-family: 'Times New Roman'; mso-ansi-language: EN-GB; mso-fareast-language: EN-US; mso-bidi-language: AR-SA;" lang="EN-GB" Professional stature or that you are doing a favour to the court. Far from it. It may simply mean that while examining an expert the Court is claiming its right for more information - or even that you were not clear enough or complete enough in the first place.
Fig. 15. Collinear vs. staggered supports
In fact, the more you say, the deeper you may be getting into trouble! So, it was no feather in my cap that I was able to come up with useful results at short order. It would have been egg on my face and a real tragedy if I had not come up with the answers.
Disposition of Case:
As expert witness, I testified to all the above and other deficiencies in design and construction. However, there were other erection considerations such as placement of heavy loads on the cage, on which there was no definitive information, but which also might have led to the collapse. In the first portion of the case, the main contractor pleaded guilty to "failing to ensure that the worksite was properly built and safely maintained", and was fined S$160,000.
None of my preceding technical testimony was contradicted. However, some of my recommendations were for future improvement, and violation of those norms could not be held against the management personnel on the convention that a case must be judged according to the laws and practice prevailing at the time and place the accident happened. The rest of the charges were dismissed on certain legal inconsistencies, and hence the details of design and construction deficiencies never became a cause for individual penalties.