Skip to content
Science, Maths & Technology

Tay Bridge Theories: Peter Lewis

Updated Wednesday 9th May 2007

Peter Lewis's theories about the Tay Bridge Disaster, part of the BBC/OU's programme website for Forensic Engineering

Dr Peter Lewis gives his opinion on why the Tay Bridge collapsed.

How did you come to investigate the Tay Bridge disaster?

"The Materials Engineering Department at the Open University were developing a course in forensic engineering and we thought that it would be useful, and educational, to look at one of the great disasters from the past; and the Tay Bridge is probably the most important structural failure within the last two centuries in Britain.

Back to top

Theory Summary

From the study of the pictures taken immediately after the accident, tied in with the witness testimony from the Board of Trade inquiry, we think that the disaster is more complicated than we’ve been led to believe by others. It wasn’t a static failure by wind pressure but rather there were almost certainly dynamic effects on the bridge itself well before the disaster. These led to gradual deterioration of the ironwork supporting the high girders. And come the night of the storm, the bridge piers were no longer capable of supporting the applied load. Within that we think that fatigue, crack propagation, probably played an important role.

Back to top

What do you think happened on the night of the disaster?

The most important event is the passage of the earlier six o’clock train. This train crossed the bridge despite the very strong winds and it was noted that sparks flew from the wheels as if it were being pushed over by the wind. The bridge itself might have been oscillating from side to side to cause the same effect or it might have been a combination of the two. Either way, it got to safety on the other side and we think that the combination of wind pressure plus oscillations on the bridge probably caused the failure of a large number of tie bars (which stabilise the bridge).

When the second train, the mail train from Edinburgh, came to the signal box at the start of the bridge the wind was equally strong as far as we know. It proceeded along the low girder section of the bridge satisfactorily, although again sparks were seen by one of the signalmen as he watched it disappear over the bridge. Then the train entered the high girder section and between piers four and five the bridge started to collapse.

We believe that the passage of the previous train had caused a lot of structural damage - in other words the bridge was in a critical state. The locomotive they used was very much heavier than normal. The six o’clock train was only pulled by a small tank engine whereas the mail train was pulled by a very heavy 35 tonne loco.

The extra load and the state of the bridge led to collapse of one of the first five piers. We don’t know which one exactly. So there followed a succession of pier failures, not just one pier. If the piers had been in a good structural state then it could have withstood the failure of one pier alone but in fact what happened was that all twelve of the high piers failed. And they failed spectacularly; most of the piers, ten of the twelve piers were actually left flush with their platforms. In other words, almost all traces of the cast iron piers had vanished. The girders were found by the side and they didn’t move very far; they’re actually quite close. At pier one they’re only 16 feet away, which means that the high girders just slipped off.

It’s the total state of collapse of all these piers which we think must point towards deterioration of the ironwork of all the piers.

Back to top

How does this theory match the actual evidence on the ground?

We base this picture of steady deterioration on eye witness evidence from the painters and fitters. I think there are about ten of them in all who gave evidence to the inquiry. They were working on all of those piers in the high girder section and they reported that when a train passed over, the piers oscillated from side to side as well as up and down and forward and back. So they were all oscillating in three different directions.

This oscillation hadn’t been seen at all a year previously when it was tested by the Board of Trade. So that leads us to the view that there was steady deterioration.

There’s a second piece of entirely independent evidence which supports this. The inspector of the bridge, Mr Noble, discovered that the joints were rattling when he stood on the platforms. This is a symptom of deterioration because the joints in the diagonal tie bar are meant to be very tight. But if all those joints were loose, then the structural integrity of the piers was in question. If a pier is not held and braced then it loses a lot of its integrity. And unfortunately Mr Noble didn’t tell the engineer, Bouch, of the problem. What he did do was he tried to fix it himself by knocking in shims (small pieces of metal). But these actually made the problem worse because he didn’t re-tighten the tie bars. He effectively left them loose but stopped them rattling. And we’re led to believe from the evidence that he did between a hundred and a hundred and fifty of these joints, which is a large number of loose and untrustworthy joints in a bridge of that size.

If you look at the photographs from the Board of Trade inquiry set of photographs then you can see which component parts were weakest. It’s immediately obvious that the lower lugs which held the tie bars in place have all fractured. And they must have fractured before the final collapse because they are lying on the platforms. You could think, ‘well this might have been a consequence of the accident’, but if you look there’s very little trace of any piers left on most of the pier platforms. Now, if those lugs had been breaking while the piers were still upright then they would naturally fall on to the platform. But of course if they’d broken as the piers were falling then you would see very little trace of any lugs on the platforms. And the number of lugs is such that they must have arisen from the failure of many tiers, there are just so many of them.

Back to top

So, I’m sure that they must have occurred before the final collapse, and as a result the tie bars were being weakened by the passage of previous trains.

There is also some supporting evidence from lugs in situ on the few columns that remained on some of the platforms. You get partial cracks, so-called sub-critical cracks, where a component breaks part way and they are characteristic of fatigue crack propagation in a large structure. You find fatigue cracks at stress concentrations at the holes and lugs which is where we see the cracks on the columns that remain. On several fractured surfaces from lugs we can see traces of crack arrest paths from hidden defects like blowholes - they’re very clear evidence of fatigue crack growth.

The bridge fell in a very regular way which is a bit puzzling. It fell in a wave form, as three arcs in fact, one for each of the big girders and it’s so uniform, there’s very little difference between the ends and the middle. That it’s just that the piers are in a very similar state of very poor structural integrity.

To have fallen in such a regular way, I think, shows that the piers must have been in a state of very poor structural integrity and that this state was very similar across all the piers in the high girder section.

If you read the proceedings of the inquiry the three commissioners and the barristers keep referring to the racking of the piers. The structure of these piers was such that they weren’t tied together at the top, there were two L-shaped girders which each connected three columns - essentially creating two sets of columns which were only held together by the tie bars.

It would have been much better, much safer, to have connected all of the columns with a single girder, which would have been hexagonal in shape. But they weren’t connected and it’s clear when the inquiry talks about a racking action that they mean the sets of columns swaying in parallel. So, all the stress is being taken by the tie bars which, I think, supports the idea that they were the first to fail, and caused the disaster.

Back to top

Does the speed of the wind play a factor in your theory?

The problem with the wind is that people couldn’t measure it exactly in those days. In fact, one of the outcomes of the Tay Bridge inquiry was that instruments like anemometers (used to measure wind speed) were developed.

Clearly the wind cannot have been that high if the train reached the high girders themselves, it had to cross a third of the bridge to actually get to the high girders across the open estuary. And although sparks were flying from the wheels, which indicates there was wind pressure against it, it was by no means critical or high, at that stage anyway.

But, of course, the wind played a role because you know it applied a moment, a levering effect, against the piers. But it wasn’t the only effect, the state of the pier remains leads us to believe they’re the major cause of the disaster and not the wind alone.

In fact, because we believe that the bridge was progressively getting weaker, it was going to collapse before long anyway. The storm on that night probably made it happen sooner, rather than later; cracks only grow in one direction - towards a critical failure.

Because all the evidence on the wind strength depends on indirect evidence it is difficult to state with any certainty what it actually was. Benjamin Baker used the state of walls and the state of the signal boxes in the area to estimate the wind strength. For example, there were three signal boxes very close to the start of the bridge. Although one had its chimney-pot knocked off there were no broken windows which you’d expect with the hurricane force winds that were suggested.

And remember that it wasn’t a closed structure with a very large surface area. This was an open lattice structure and presenting very little resistance to the wind.

Back to top

Do you have an ethos that guided your investigation?

In any accident you must look at the direct evidence and the fresher the evidence the better. It’s essentially the philosophy we used. You must tackle the direct evidence without looking to secondary or tertiary sources for opinion and come to your own view of what that evidence shows. That’s how I work as a forensic engineer.

Back to top

How do you rate the quality of the BOT inquiry, both against today’s standards and also those of the 1880s?

The inquiry sat within a few days of the disaster up in Dundee, which was I think an achievement in itself and it started interviewing eyewitnesses which was good because you’re getting direct testimony when it is fresh in the minds of the witnesses.

Then they moved to Westminster and received the expert evidence. Some of which was a bit mixed and looking in hindsight, they could have been more systematic in their approach and recorded more evidence. But they did take a lot of photographs and they did test some of the surviving materials so they were able to tie together quite a reasonable picture for the time.

However, although fatigue was beginning to be known about then, the expert engineers seem to have neglected to look at the fracture surfaces in any detail.

But they reached their conclusions remarkably swiftly compared to today, six months from start to finish. It was essential in that day and age because the public wanted a result quickly, especially as Bouch was planning to build another bridge over the Forth. And people wanted some reassurance that another whole train could not be lost in another such disaster. He’d also constructed quite a few other bridges to not dissimilar designs, so all those bridges had to be tested for their integrity. So yes, they did the right thing I think.

Overall, I think they did a pretty good job, although obviously today things would be examined in much greater detail, which would take much longer."

Back to top

Anatomy of a disaster

 

For further information, take a look at our frequently asked questions which may give you the support you need.

Have a question?