Hello and welcome to Failurology; a podcast about engineering failures. I’m your host, Nicole, and I’m from Calgary, Alberta.
This week’s engineering failure is the Hartford Arena roof collapse; which is the first known engineering failure caused by a computer aided design. Now, my discipline is mechanical. Structural design is not in my wheelhouse. I can read the drawings to find beams and slab thickness, but I really had to dig deep to research this one. And I learned a lot. I think this is a really important failure due to being the first computer aided design to fail.
But before I get into all of that, the news.
This week in engineering news, the 11.2km Eysturoy tunnel in the Faroe Islands contains the world’s first subsea roundabout. The Faroe Islands are an autonomous territory within the Kingdom of Denmark located 320 kilometres NW of Scotland, halfway between Norway and Iceland. The islands are roughly 1,400 square kilometres with a population of over 52 thousand.
The tunnel opened on December 19th and connects the Streymoy (stray-moi) Island, where the capital city of Tórshavn (tor-shavn) is located, to two parts of the Eysturoy (estroi) Island. The tunnel travels from Streymoy (stray-moi) across the bay and then wyes at the roundabout to serve two arms of the Eysturoy (estroi) island.
There is a bridge further north that connects these two islands, but the tunnel reduces travel time by up to an hour and a half. This will allow services to be more available than before; as well as, significantly reduce commutes for residents.
Construction on the tunnel began in January 2017. The tunnel is 187m below sea level at its deepest, with the steepest grade no more than 5%. It is estimated that the tunnel will see 6000 vehicles per day. Costs for the tunnel construction will be repaid through toll fees.
I do love a good tunnel. I have run the Detroit marathon a couple times and the route runs through the Windsor-Detroit tunnel which is my favourite part of the race. And I think I’ll have to add Faroe Islands to my list of places I want to travel to, this tunnel sounds really cool.
If you want to read more on the tunnel, check out the website page, link in shown notes, for this episode. The article also contains a really informative video of the tunnel.
Now on to this week’s engineering failure; the Hartford Arena roof collapse in 1978; the first known engineering failure caused by a computer aided design.
Going into research for this failure, I had expected that the failure would have gone somewhat unrecognized by all parties until the collapse occurred. But as it turns out, several people pointed out excessive deflection throughout construction, and it went ignored. The arrogance that the new fancy computer model couldn’t be wrong really did the engineer in here. As budgets and deadlines tighten, we all rely more and more on technology to help us, but we have to double check the calculations. Also, when a contractor tells you there’s a problem, you should at least hear him out. And always double check your numbers.
Unfortunately, the legal outcome of this failure has not been made public, so information was a bit limited. But I was able to dig up enough to tell you about the cause, and of course, what we can learn to prevent future failures from happening.
Construction on the Hartford Arena started on April 2, 1971. The arena opened on Jan 9, 1975.
It cost $30 million USD, which is about $143 million USD today. It is owned by the City of Hartford.
The arena has over 10,000 seats and was built to house the Hartford Whalers who were part of the World Hockey Association from 1972-1979, and later in the NHL from 1979-1997. The Whalers started out in New England, moved to Hartford Connecticut in 1974 and then moved to Carolina in 1997 to become the Carolina Hurricanes.
Vincent Kling – architect
Fraoli, Blum, & Yesselman Engineers also known as FB&Y were the structural engineers
At 4:25 am Jan 18, 1978 following heavy snowfall caused by winter storm Igor, the roof collapsed. Over five thousand people were in the building 5 or 6 hours earlier for a basketball game. Luckily no one was in the building when it collapsed. Finally, a failure where no one was hurt!!
Computers were relatively uncommon in the 70s. not saying they didn’t exist, but its not like everyone had one in their house. I think we got our first computer in 1993. Good ole dos and floppy disks.
The engineers used computer analysis for the structural design calculations. Even though the design was susceptible to buckling, the engineers were overconfident in the computer's ability, which led to them ignoring evidence of a problem.
Typical space frame design
The design was a typical space frame design. Which is a rigid, lightweight truss-like structure constructed in an interlocking geometric pattern, specifically triangles, typically built from steel tube or “I” struts. Which is a really fancy way to say that the roof is supported by upside down pyramids, made from steel truss-like members, laid out in a grid pattern.
space frame designs allow for lighter but stronger structure
it's common in industrial buildings, warehouses, swimming pools, stadiums (Like the Rogers Centre where the Blue Jays play), airports, etc; wherever a large open space is required with minimal columns. Maybe you, like me, have seen this structure style hundreds of times, and just never knew it was called a space frame design. This one however was not implemented correctly
The Arena`s roof truss structure was 3m deep, 91x110m in plan, 25m above floor
Differences from traditional space frame
In a space frame design, it is important that all of the intersecting truss members meet at the same points so the load is transferred evenly. This is somewhat challenging to explain, please be sure to look at the photos on the episode webpage for more info. But let’s say you have your upside-down wireframe pyramid, in a typical space frame design. We have a big square at the top made of four structural members, a little square in the middle, a point at the bottom, and a bunch of other structural members connecting them all together. The points from several upside down pyramids are then connected together using additional structural members to form a cohesive roof system.
It is imperative that each of the intersecting structural members meet at the same spot, whether that be at corners or at the middle of spans.
But that didn’t happen on the Hartord Arena, the top chords intersected at different points than diagonal members – which resulted in an increasing risk of buckling.
In addition to this major problem, there were four other issues that contributed to the failure. the structural members were cross shaped, which has a lower resistance to buckling than square or round tube or “I” shaped members
The top chord didn’t directly support the roof panels, small posts were installed between the top square chord corners and midpoints and roof panel – this was done to allow sloping, meant to eliminate bending stress
As I mentioned early, there were only 4 columns 13.7m inside the roof edges to support the entire roof
And lastly the, frame was not cambered or curved, which increases strength of members
If you didn’t quite follow all of that, it’s ok, the important part to know is the roof was under designed and not constructed properly.
Construction was divided into five contracts, managed by a construction manager; this setup isn’t uncommon, but unless you have a construction manager who is on the ball, it's easy for things to slip through the cracks, creating confusion over who was responsible. In addition to gaps in construction, there were also gaps in the design and inspections
The architect recommended that a structural engineer oversee construction of the steel. This is very common; I’ve seen 3rd party steel inspections carried on lots of projects. But the construction manager refused and said he would do it to save money. That said, the design engineer should have done inspections, even a general overview.
Peer reviews were commonly required in Hartford for projects of this magnitude with new design techniques – but not for this project. Connecticut is one of the few states that requires peer reviews today. I have seen clients hire other consultants for peer reviews of various designs over the years; sometimes they just want a second opinion on a critical system. But I can’t recall a time when the authority having jurisdiction, or the City, required a third party review, although I may not have been involved.
Deflection - Throughout the process, engineers assured the city that the deflection was acceptable. Engineers were negligent for not checking design and again for ignoring deflection.
The structure started failing as soon it was completed
The design underestimated both the live and dead load by about 20%. The dead load is the static load or weight of the structure itself
Live load is also known as dynamic or variable loads, which in this case was the snow from winter storm Igor.
Some members were overloaded by 852%.
The structural details omitted midpoint braces for the exterior rods of the top layers. Interior rods were partially or improperly braced at the midpoint
Slenderness ratio of members violated the AISC (American institute of steel construction) code provisions
Members with bolt holes exceeding 85% of total area violated AISC code
Some steel didn’t meet spec
Computer model only included the main members for the structure If it had included intermediate members, the instabilities and primary bending moments would have been detected
The ultimate cause was the top layer, which buckled, causing other layers to buckle. The lower layer, originally in tension, was now in compression and also buckled.
Computers can run calculations very fast, much faster than humans, and it allows them to be run at minimum loads to provide more economical designs. But this limits the safety factors, which limits room for error in calculations, manufacturing and installation. Should safety factors be increased for high occupancy structures?
Who is ultimately liable for faulty designs where software is used? In Canada, the engineer is liable. There is a disclaimer from software manufacturers which shifts liability to the user and provincial engineering associations agree on this.
6 years after the collapse, an out of court settlement was reached. This did not allow for legal precedent to be set for future cases. The outcome of this settlement was also not published.
The arena re-opened after collapse Jan 17 1980; two years less a day after the collapse.
What are the takeaways from this failure, the lessons learned?
It is important to outline and quantify responsibility in contract documents
double check computer calculations, even if its against previous projects, rules of thumb, or simple calculations
when someone says the design has a problem, hear them out, double check the numbers
review the installation of your designs during construction
read literature related to your profession to stay relevant.
Engineering failures are an integral part of engineering. Engineers must educate themselves on failures from peers before them in order to prevent history from repeating itself. This is the main objective of this podcast, to educate engineers on past failures and prevent future ones from happening.
I hope you found this failure as interesting as I did. I’m starting to see a theme that engineers are told there is a problem and just ignore it until it’s too late. So please don’t be one of those engineers.
Check out the podcast page, link in show notes, for photos from this week’s episode. And if you want to chat with me, my twitter handle is @failurology or you can email me at firstname.lastname@example.org. Thanks everyone for listening. And tune in next week to hear about the Malahide Viaduct; a collapse that stopped the only train route between Dublin and Belfast for 3 months in 2009. But more on that next week. Bye everyone, talk to you next year!