What would you do if a structure was behaving very differently from the way your state-of-the-art computer software package was predicting, asks forensic engineer and analyst Sean Brady.

You'd stop and question your software package, wouldn't you?

Or at least you'd like to think you would...

Hartford Stadium Center Stadium

The roof of the Hartford Civic Center Stadium, Connecticut, was an ambitious design – a 91.4m × 110m space frame suspended 25.3m above a 10,000-seat arena. This intricate space frame was a spiderweb of 2,300 individual steel members and was supported on four pylons.

But even before construction was complete there were problems. When the space frame was assembled at ground level, prior to being lifted to the top of the pylons, it was found to be deflecting more than expected.

It was like the designers had missed something.

But the issue doesn't appear to have been addressed, the roof was lifted into position, and then it was discovered that the deflection problem had gotten worse – the roof was now sagging more than twice what was anticipated.

To this news the designers replied that such discrepancies were to be expected in view of the simplifying assumptions used in their theoretical calculations.

Then the contractor installing the panels covering the space frame discovered yet another issue. The space frame was significantly distorted – probably as a result of the large deflections.

This distortion meant that the support brackets for the panels were not lining up with the space frame. But when the designers were told, they again expressed no concern, so the contractor either cut or extended the brackets to work around the distortion.

Then multiple complaints were received from members of the public – the roof deflections were so pronounced they were unsettling. So the city of Hartford, who were concerned, approached the designers, who again defended the accuracy of their design.

Now fast-forward five years to the night of January 18, 1978. The Civic Center is subject to its heaviest snow load since construction. But this snow load is still only half the design load, so the structure should have dealt with it comfortably.

But it didn't.

At 4.19am that morning, with the arena empty, the 1,270 tonne space frame collapsed. Only six hours earlier more than 5,000 people had been sitting beneath it, watching a basketball game.

Cause of failure

So what went wrong?

It would turn out that the design had been completely inadequate. The structure had begun failing almost as soon as it was completed – some of the members of the space frame were overloaded by more than 800%.

So how did the designers get it so wrong?

It would transpire that the problem was over-reliance on computer software.

The design firm had convinced the city of Hartford to purchase a state-of-the-art software analysis package. They argued that using this package for the design would save half a million dollars in construction costs.

But this software would introduce a number of serious deficiencies into the structure, which led to the inadequate design. We will not get into them here, but references discussing these deficiencies are provided below.

The bottom line is that the software relied on a number of assumptions that were simply not applicable to the structure in practice, and these assumptions led to big differences between the predicted and actual behaviour of the roof, and ultimately led to its collapse.

'Computers have enabled people to make more mistakes faster than almost any invention in history, with the possible exception of tequila and handguns.' Ratcliffe

Human factors

But, to me, this is not the most intriguing aspect of the failure.

It doesn't explain why the designers didn't revisit their software results when it became obvious that the structure was behaving in a different manner and showing distress.

Because you're probably going to argue that that's exactly what you'd do. And maybe you would. But the reality is that we all know engineers who insist on the validity of their computer software results, regardless of how these results compare with a structure's performance in practice.

Because one of the most insidious issues with computer software is it can breed overconfidence, which in turn can lead to over-reliance. This is what happened at Hartford.

As the designers specifically argued that they needed the state-of-the-art package to complete the project, why would they then doubt the outputs of that package? To do so was tantamount to doubting the very decision to purchase the software in the first place.

Now I'm not suggesting the use of computer software is all bad. It's not, it automates many boring, routine, and sometimes complex tasks that can be very prone to human error.

Over-reliant on computer software?

The real question is not whether we should rely on the use of computer software, but rather whether we have become over-reliant on it. Many feel we are now training a generation of computer programmers instead of engineers.

The reality is that software tools have made an immense contribution to our engineering profession, but they are only as good as the engineer driving them.

So as we take our steps into a new era in the construction industry – what we can call the 'BIM era' - we should be excited about the benefits such technology can bring, but we must never forget that computer software should never be more than a tool to be used wisely – a tool to make us better engineers.

It should never be something that replaces us thinking like engineers, as it did in Hartford.

Author: Sean Brady is managing director of Brady Heywood, a forensic engineering firm that specialises in investigating the cause of defects and failures in construction and engineering disputes.

Detailed discussion of the deficiencies in the structure can be found in this Structural Engineer article, as well as Beyond Failure and Why Buildings Fall Down.

Listen to the Brady Heywood Podcast on Desktop or iTunes for more discussion of engineering failures.