MTC 470: Term Paper: The Collapse of the Hartford Civic Center Roof Name: Chris Pohorence Date: 04-08-2001

Introduction:

The Hartford Civic Center was planned in 1970 as a part of a downtown revitalization plan. The arena was constructed during a period from 1972 to 1973, when it opened. The arena hosted large events and sporting events like, the Hartford Whalers hockey team, and the University of Connecticut basketball and hockey games. On January 18, 1978, at around 4:15 A.M., the roof collapsed following a heavy snowstorm. Immediately, the original roof was demolished and a new roof was installed. The Civic Center was reopened in 1980.

The Construction of the Roof:

When the arena was planned in 1970, the designers, Fraoli, Blum and Yesselman, proposed a 300-foot by 360-foot space frame roof over the arena. This space frame roof was an innovative design for that time; also, this type of roof was less expensive to construct than other, more conventional designs. This design consisted of “two main layers arranged in 30-foot by 30-foot grids composed of horizontal steel bars 21 feet apart.” The two layers (top and bottom) are connected with a 30-foot diagonal beam and a layer of horizontal beams for support. The 30-foot bars were also braced with diagonal bar attached to the midpoint of the same bar.

Figure 1: Diagram of the space frame roof beam. (Figure from, “Design Flaws Collapse Steel Space Frame Roof”, Engineering News Record, April 6, 1978, p.9.)

Figure 2: Three-dimension of the space frame roof. Martin, Rachel, “Hartford Civic Center Arena Roof Collapse”, http://www.eng.uab.edu/cee/reu_nsf99/hartford.htm

The 1400-ton, 108,000 square foot roof was fabricated on site and raised into place by temporary lift towers. Then, the concrete support columns were built inside the lift towers. When the roof the in place and installed, the rest of the arena was built around it.

The Collapse of the Roof

In the winter 0f 1977-78, was one of the snowiest winters on record, capped off by the infamous blizzard of 1978, which left the Midwest and the Northeast with several feet of snow. On January 18, 1978, at around 4:19 AM, a beam located in the center of the roof buckled, causing the roof to crash 83 feet to the floor of the arena. The collapse of the roof happened, about two weeks after the blizzard of 1978. Luckily, the arena was empty and no one hurt or killed.

Causes of the Roof Collapse

There were two factors in the cause of the roof collapse, one was the structural and material failure in the beams, and the other was the heavy snow pack on the roof after the blizzard.

Of these two factors, the least important factor was the heavy snow pack onto the roof before the collapse. On the same night, roofs fell in other parts of the East due to the heavy snow.

Three days later, following another heavy snowstorm, another space frame roof collapsed. The auditorium at the C.W. Post College in Brookville, NY domed space frame roof fell in. After these collapses, both the Hartford Civic Center and the C.W. Post College Auditorium roofs, other space frame roof across the country were being investigated and inspected.

The snow and ice build up on the Civic Center roof was caused by numerous snowstorms during that winter season. The build up also increased by the heating and cooling cycle on the roof surface during the same time. These cycles are cause by air temperatures rising above freezing then falling below freezing. When air temperature is above freezing, the snow pack will begin to melt. The water will naturally run to lower areas on the roof, where drainage systems for the roof are located. When the temperature drops, the liquid water will freeze again. The once liquid water would refreeze and leave a thick layer of ice on the roof surface. The process would also cause the clogging the draining systems with ice. The layer(s) of ice would weigh more than the already existing snow pack, adding weight to the roof structure.

The major factor in the failure was the design of the roof itself. The roof was designed using an untested, non-conventional design and computer structural analysis. The results of the computerized analysis are shown in Figures 1 & 2. After the completion of the design stage and during the construction stage of the roof assembly is where the flaws of the design were first detected. Below are some of the many of the flaws in the design:

1. The configuration of the four steel angles did not provide good resistance to buckling. The cross-shaped built up section has a much smaller radius of gyration than an I-section or a tube section.

Figure 3: Comparison of the different cross sections. Martin, Rachel

2. The top horizontal intersected at a different point than diagonal bars rather than the same point, making the roof especially susceptible to buckling.

3. The top layer of this roof did not support the roofing panels; the short posts on the nodes of the top layer did. Not only were these posts meant to eliminate bending stresses on the top layer bars, but their varied heights also allowed for positive drainage.

4. Four pylon legs positioned 45 feet inside of the edges of the roof supported it instead of boundary columns or walls.

5. The space frame was not cambered. Computer analysis predicted a downward deflection of 13 inches at the midpoint of the roof and an upward deflection on 6 inches at the corners. These deflections were taken into account.Note: Reason 5 from: Martin, Rachel, “Hartford Civic Center Arena Roof Collapse”, http://www.eng.uab.edu/cee/reu_nsf99/hartford.htm

When the roof was being assembled on the ground, an inspector pointed out that some of the nodes holding the top and bottom layers of the roof together were bending excessively. After the roof was installed onto the concrete pillars, a report from another inspector using actual roof measurements showed the amount of deflection from the roof was twice the estimates from the computer analysis. These initial problems were greatly ignored by the engineering teams in charge of the project. The engineers stated that some discrepancies between the theoretical calculations and the actual measurements should exist. In other words, they stood by their design and computer analysis. Other problems soon popped up when the roof was being completed. A subcontractor had problem fitting supports for the fascia panels on the outside of the truss due to the extreme deflection of the frame. To correct this problem, many of the supports were cut and welded back together in order to make the panels fit. After the roof was complete, a citizen noted to the same team of engineers and inspectors that the completed roof had a noticeable bow that can be observed from above the arena. However, the engineers and contractors that worked on the project were so confidant of their design that the design and data seemed impermeable to these concerns brought up by ordinary citizens and subcontractors who worked on the project. To the defense of these “opponents”, there were many pictures of the roof taken before the collapse, showing evidence of bowing especially in the center of the arena.

Figure 6: Picture of corner of the arena, after the collapse.

Figure 7: Aerial view of the collapsed roof. Pictures from users.ipa.net/~frzy/frzyjobs.html

The City of Hartford hired Lev Zetlin Associates, Inc. to investigate the collapse of the roof. They drew the conclusion that roof was designed poorly and began failing as soon as it was completed. They also concluded that three major design errors cause the collapse:

  1. The top layer’s exterior compression members on the east and west faces were overloaded by 852%.
  2. The top layer’s exterior compression members on the north and south faces were overloaded by 213%.
  3. The top layer’s interior compression members in the east-west direction were overloaded by 72%.

A consultant found in the structure was missing braces for the midpoints of the rods that support the top layer of the beam. In addition, the exterior rods were braced every 30 feet, instead of a brace every 15 feet as it was called for in the initial plan. Interior rods were partially braced at their midpoints. The omission of the addition bracing and the change in bracing and fabrication made the structure susceptible to premature failure by reducing the amount of total load the roof could handle.

Original Design Connection A Connection B Connection C Connection D
Allowable force: 160,000-lb Allowable moment: 0 Allowable force: 185,000-lb Allowable force: 625,000-lb Allowable force: 565,000-lb
As-built Design Connection A Connection B Connection C Connection D
Allowable force: 15,440-lb Allowable moment: 9,490 lb-ft Allowable force: 59,000-lb Allowable force: 363,000- lb Allowable force: 565,000-lb

Since there were changes from the original design, which ultimately weaken the roof, this chart compares the changes from the original design. The amount of surface contact for the end of the member is reduced greatly. With the setup the way it was, the weld joints have the ability to flex in many directions. If the original design was followed, the member at the intersection would have more points of contact, which reduces the amount of movement in the joint to nil. Thus creating a strong, stabile structure, all of the movements in the structure would be the minor deflections in the member to accommodate applied loads on to the roof.

The roof was designed so weak could not bear it’s own weight, the top layer bowed in compression, shortening the member’s length, causing further deflection and distortion in the truss. When a load was applied to the roof, the top layer, already stressed due to design flaw, will even stress further to the point of buckling. The forces acting on the top layer of the loaded roof will transfer the forces on to the lower layer causing it to go from tension to compression also causing buckling.

Other design and construction errors that contribute to the failure are the following:

  1. The slenderness ration of the loaded members violated American Institute of Steel Construction (AISC) standards.
  2. Member with bolt holes exceeding 85% of the total area of the member. This violates another AISC standard.
  3. Spacer plates were placed too far apart in some of the four-angle members allowing individual angle to buckle
  4. Some of the steel used in the structure failed to meet specifications.
They also found that dead load (zero added load) for the roof was underestimated. The actual dead load for the roof was 23 psf (pounds per square foot) instead of the calculated 18 psf. The designed deflection for just the frame only, minus the roof material, would have been 7.35 inches, but the actual deflection was 8.4 inches. The roof material, which was composed a cementous decking and 3-inch thick wood fiber composition roof, would have added 20 pounds (I think that should be pounds per square foot) to the frame. The design did allow for settling of the roof after installation. The settling measurement for the roof deflection was calculated to be 8.5 inches. However, a measurement of the roof deflection in April 1975, showed the empty roof deflected 12 to 13 inches, about 125% over the calculated measurements. The designers of the roof calculated a deflection of 13 inches in the middle and 6 inches at the ends for all live and dead loads. The night when the roof collapse, the load on the roof from snow and ice was approximated to be 66-73 psf (includes snows loads of 12 to 19 psf). This estimate was figured from snow loads on neighboring buildings around the Civic Center. The design states that the roof should have a load capacity of 140 psf. However, the structure failed at around one-half the maximum load capacity. A consultant suggests in a conclusion that maximum design load should have been 84 psf with a factor of safety of 1.67 to have a capacity of 140 psf.

Conclusion:

There are many lessons that can be learned from this. First, computers are as good as their programmers. Computers and structural analysis software was used exclusively to develop and test the roof design for feasibly. The engineers and contractors in charge of the project became total reliant on the computer’s results for answers solving basic engineering structure analysis. Secondly, a radical, untested design involving “cruciform shape of rods” should not be used, but a more conventional design like the railroad truss type roof that replaced the fallen roof in 1980. Thirdly, the members in the beam would have been attached the way it was planned; disaster would have been avoided. Finally, the entire roof should have not collapse if a few members would have failed. The rest of the roof should have been stable enough to support itself, if a small part should have failed.

On a side note not to do with the structural failure, the construction of the arena was done a number of contractors coordinated by a manager. The manager did not want to hire a structural engineer, due to the expense, and decided to inspect the project himself, although he lack the training. Errors like the bowing member and the poorly supported joints would have been picked up if a structural engineer was present doing inspections. In addition, the construction manager set him up for total responsibility if the project failed. Ironically, when the roof failed, he denied all responsibility.

There were a number of instances where the project was retooled due to cost cutting measures. Cost-cutting measures in this project did the following:
  1. Prevented proper inspection of the structure during construction
  2. Incorporated an untested, non conventional roof design
  3. Prevented the proper installation of the roof (the roof was built on the ground and lifted up to be installed, instead of build it from the supports outward.)
  4. The reduction the amount of beams and members that would have increased the stability and the strength of the structure.
  5. Incorporate relativity new (1973) computer modeling software to design, troubleshoot and solve construction problems.
Cost cutting is a nice idea, especially in a large project such as this. However, there is a limit to how much cost cutting could be done until it affects the integrity of the project. The collapse should have not happened if the designer and the construction firms were not worried about the financial aspect of the project but the quality of the construction in the project.

Bibliography

1. Martin, Rachel, “Hartford Civic Center Arena Roof Collapse”, http://www.eng.uab.edu/cee/reu_nsf99/hartford.htm.

2. “Design Flaws Collapsed Steel Space Frame Roof”, ENR, April 6, 1978, pp. 9-11.

3. “Space Frame Roofs Collapse Following Heavy Snowfall”, ENR, January 26, 1978, pp. 8-9.

4. “Collapsed Space Truss Roof had a Combination of Flaws”, ENR, June 22, 1978, pp. 36-37.

5. “Snow and Ice Loads”, http://www.uuic.com/autospec/lossprev/snowload.html

6. Swift, Mike, “In memory, Civic Center Collapse Proven to be Hartford’s Finest Hour”, Danbury, Connecticut The News-Times, February 2, 1998,http://www.newstimes.com/archive98/feb0298/rgg.htm.
Last Updated: 03-24-2002 Copyright 2000, 2002 Christopher Pohorence