23 June 2010

"Failed Bridges: Case Studies, Causes and Consequences"

If the news is any guide, bridges around the world seem to be falling down at an alarmingly frequency. In early June alone, there were collapses reported in Idaho, USA, Ohio, USA, Connecticut, USA (1 injury), India and Indonesia (12 dead). The casual onlooker might conclude that the bridge engineering industry was rife with incompetence.

That impression wouldn't be dispelled by reading "Failed Bridges: Case Studies, Causes and Consequences" by Joachim Scheer (Ernst and Sohn / Wiley, 2010, 307pp) [Amazon.co.uk]. This features an impressively long (but still far from comprehensive e.g. no Ynys-y-Gwas) list of bridge collapses guaranteed to strike alarm into the hearts of wary gephyrophobes. Of course, as the author emphasises, these statistically remain a tiny minority of structures. Nonetheless, it's hard to avoid the impression that there is a widespread problem.

"Failed Bridges" is the second edition, in English, of a book which first saw print in German. The decision to publish this expanded version in English recognised that engineers around the world often seem to fail to learn from the mistakes in their predecessors, and it was desirable to expand the book's audience much more widely. The literature of bridge failure is lengthy, but not always accessible to practicing engineers. The aim of this book is to bring as much data into one volume as possible and thus provide a single point of reference documenting the reasons for past failures, and the lessons which might be learned.

It contrasts with Björn Åkesson's book "Understanding Bridge Collapses", which I reviewed previously. Åkesson offered 20 case studies, with detailed technical explanations including calculations where appropriate. As such, it's of interest both to design engineers and to students. Scheer widens the net to capture at least 440 failures, with the inevitable consequence that much less detail is provided for each one. Indeed, for the better known failures, little if any detail on the cause of collapse is given, with the author assuming readers can access the information elsewhere.

Much of "Failed Bridges" comprises data on individual bridge collapses presented in the form of tables, indeed 115 pages are taken up with these tables, which are for reference rather than for reading. Some similar information is available online (e.g. at Wikipedia or BridgeForum), and neither the book nor the online efforts are comprehensive, with numerous bridge failures absent in each case. "Failed Bridges" seems particularly short on failures due to floods (probably the single biggest cause of collapse throughout history) and seismic activity, with the latter category being a new inclusion in this second edition.

Failures are categorised by physical situation or cause: during construction; during service under "normal" load including wind; ship collision; vehicular collision; flooding and ice floes; fire or explosion; seismic activity; and falsework failures.

The broader view of a system like SCOSS's "3 Ps" (people, process, product) only comes out in the book's final chapter, which summarises the lessons learned. Nor is the explanation of failure as thorough as the approach used by the RAIB, which identifies the immediate cause (often physical), causal factors, contributory factors and underlying factors, in an attempt to lay bare the complex web of human error which underlies most bridge failures. None of this is to criticise the author, as for most historic bridge failures, detailed information of this sort is simply not available.

Scheer makes a number of interesting points in his introduction. Recounting his own experiences of bridge failures (he has acted on several occasions as investigator and expert witness), he notes the high proportion which occur during construction: "My own experience reflects what everyone 'on site' knows: building work is often linked with failure; this has always been the case and always will be". However, I suspect very few of the people with the greatest responsibilities for construction of bridges (i.e. contractors) will ever read this book. The challenge for its main audience (academics and designers) is therefore how to bring its lessons to the attention of contractors, and clients (the other group who could benefit but won't read it), without appearing alarmist.

I'd suggest that the risk management processes now common throughout engineering design and construction offer one way forward: do designers completing their obligations under the CDM regulations always make clear the more extreme risks of failure? "Failed Bridges" offers countless examples of what can go wrong, and plenty of advice on mitigating the risks. Perhaps if more risk assessments clearly identified key mitigation actions such as the involvement of the designer in construction supervision, the situation would be improved.

Scheer also notes that: "when analyzing the causes of structural failures today, I find that there is hardly any case which could have been prevented by more detailed calculation". Instead, failures occur because certain possibilities were never even considered in calculation, or they are the result of what with hindsight can be seen as gross rashness. A similar point was made by D.W. Smith in a major survey of bridge failures (ICE Proceedings, 1976), who also warned of the dangers of reliance on complicated and sometimes ambiguous design standards. Reading this, it is tempting to wonder whether the introduction of Eurocodes offers any improvement at all on structural safety, especially where their implicit reliance on probability theory can hide the very real uncertainties that go beyond the boundaries assumed.

There are a number of minor issues I noted while reading the book. Chief amongst these is that while its translation is generally excellent, the opportunity has been missed to introduce references to material available in English. So for example, the failure of the Tay, Dee and Cleddau bridges is discussed with reference to papers in German, rather than the widely available and often informative material in English. For example, for the Tay Bridge (pictured), Peter Lewis's very thorough book "Beautiful Railway Bridge of the Silvery Tay" isn't mentioned, nor the essential technical papers by Lewis, Martin and MacLeod. Given that the author assumes readers will seek more details elsewhere, it's unfortunate that the bibliography remains largely aimed at German-speakers.

The level of detail given for individual failures is often uneven, with some, such as the 1940 collapse of the Frankenthal Rhine Bridge accompanied by levels of detail which seem excessive in comparison to others. However, this is as much a strength as a failing, as often the bridges covered in detail are German and little known to the wider world. Scheer also offers more detail where he sees it as necessary to provide a view other than that which has been widely reported. For example, the 1907 collapse of the Lawrence River bridge in Quebec (pictured) was widely stated to have been due to the underestimation of dead loads in calculation. Scheer draws attention instead to the non-compactness of the critical steel section, and hence its inability to redistribute internal constraint stresses in a plastic manner.

For some of the causes of collapse listed in the book, the technical lessons seem to be straightforward, and are largely now a routine feature of design standards. This is the case for failure due to ship and vehicular collision, where events like the Eschede railway disaster (pictured) have led to considerable conservatism in the design codes. The hazards from flooding and earthquakes are also to a certain extent predictable. The chapters devoted to these effects are correspondingly shorter. However, one lesson from history is that new hazards continue to become apparent as technology advances. One example not covered in this book is the phenomenon of ballast instability due to bridge resonance, observed when high-speed trains were introduced to the Paris/Lyon rail line, causing damage to bridges and increased risk of derailment.

Amongst the myriad of other case studies, one that stood out for me was the 1990 collapse of arch falsework for the Lake Street - Marshall Avenue Bridge at St Paul, Minnesota. This was a case where different elements of the falsework were designed by different firms, although the overall strength of the system was highly dependent on the stiffness relationships of each element, including the stiffness of parts of the bridge arch already cast. The result was that scaffolding props carried greater loads than assumed by their designer, and their support beams lacked sufficiently strong stiffeners. When the scaffold collapsed, one person was killed.

This case stood out because it's an example of something I've seen several times in my own work, especially beyond the field of bridge engineering. In building structures, it's entirely normal for responsibility for "details" to be divorced from the main designer (e.g. pile design, or bolted steelwork connections), and there's an accident waiting to happen wherever those details prove more significant to the global design than is commonly assumed (e.g. where the stiffness of a bolted connection is important, or a pile's stiffness against lateral loading).

The final chapter of the bridge seeks to summarise advice on how to avoid bridge failures, both from the author's own perspective and by surveying others who have attempted the same. These are mostly aimed at engineers, although they do touch on lessons for those involved in procurement, such as the need to incentivise the appointment of professionals who are competent (while seemingly obvious, this also seems to be frequently ignored in the delusion that "low cost" is the same thing as "high value").

Scheer also touches on difficulties with regulators, who by promoting particular procurement arrangements often hinder best practice (e.g. the lack of involvement of experienced designers in site supervision which has become increasingly widespread in recent years, resulting from a preference for design-build procurement). He notes that pressure to drive down costs and meet deadlines must have an adverse effect on time available to optimise design and coordinate work correctly. I particularly liked his statement that "designers are forced to commit themselves to a single concept at a much too early stage and to stick with it, at times, against their better judgment", a conclusion which should be read by anyone who is overly enamoured of the bridge design competition as a procurement route.

Scheer's advice extends to every phase of bridge design and construction. On conceptual design, the need for robustness and simplicity of structural form are discussed. In calculation, the key concern relates to unsafe extrapolation - the failure to realise when rules well-understood at one scale can become unsafe at larger scales, principally because effects which were once negligible become dominant. This includes buckling issues as well as the aerodynamic issues that were encountered on the Tacoma Narrows bridge. Scheer notes, quite correctly, that engineers in the modern age are seldom afforded the time, funding, or control required to undertake experiments which might render extrapolation safe, with the example of Stephenson and Fairbairn's experimental work on the Britannia Bridge box girders (pictured) being offered as an example of how extrapolation can be carried out sensibly.

Advice covers how the design process should be coordinated, as well as risks related to modelling and the misuse of computer analysis. Many of these issues have been well rehearsed elsewhere. Scheer records recommendations made by others: Sir Alfred Pugsley, W. Plagemann and D. Kaminetsky. The last of these includes one guideline which might act as a motto for this entire book: "The best way to generate a failure on your job is to disregard the lessons to be learned from someone else's failures".

Amongst Scheer's own advice I particularly like: "Always bear in mind that your model of a load-bearing structure is defective".

He concludes the book with suggestions for how the history of failure might hold lessons for the teaching of structural engineering, concluding that students learn too much about analysis, and too little about design, a view I very much agree with. As Scheer notes, too much maths and science can hinder a strong intuitive feeling for structural behaviour, rather than assist it.

Overall, "Failed Bridges" is an excellent contribution to the bridge engineering literature. It's singlemindedness doesn't make it an easy book to read right through, but the information it contains should be thought-provoking for younger engineers, and likely to cause many grimaces of recognition for their older colleagues.

I do wonder how the material can be more widely disseminated - the reality is that a very small minority of design engineers read books like this, and an even smaller proportion of the clients and contractors who can play an even larger role in preventing failure. So, if you're in any of these groups, please consider what you're missing!

It's a book that's certainly worth reading for those concerned with the education of engineers, and I'd particularly commend it to those involved in the development of standards and codes, procurement strategies, or with overseeing roles in the process of design management.

7 comments:

Paper Project said...

Lots to absorb in that post! "Understanding Bridge Collapses" remains one of my favourite engineering reference books - on one occasion I recalled one of the 'lessons' in it whilst peer reviewing a calculation pack...and saved an engineer from a potential mistake.

The subset of bridge disasters where new effects are found (Tay Bridge and wind effects, Tacoma Narrows and excitation) are, I think an almost unavoidable consequence of progress - as you mention, if you don't know it exists, you can't design for it.

The worrying bridge disasters are the ones where known effects are not considered, whether because that knowledge has dropped out of the industry, or because poor supervision of inexperienced\ambitious designers leads to errors. In small teams with wide ranging project portfolios there might be a propensity for good engineers to miss specialist design issues.

Something akin to CROSS, but for designers might be interesting - an anonymous forum to collate 'near-miss' design failures where designers can learn the limits of our ignorance!

The Happy Pontist said...

There is advice, in Scheer's book and elsewhere, on how unknown effects can be better anticipated. These include better knowledge-sharing with other fields of engineering and science (the issues at Tacoma might have been predicted by aeronautical engineers rather than bridge engineers); proceeding where possible by small steps rather than large steps (including greater use of experimental testing); provision of robustness in designs (in more than just the strength sense, e.g. building in damping to designs which are dynamically sensitive, even where it isn't apparent what the excitation method may be).

The Happy Pontist said...

It's also worth noting that the aerodynamic behaviour of Tacoma was perhaps a forgotten rather than an unknown effect - many early suspension bridges swayed in a fairly similar manner, and Scheer notes that this precise point has been made by a number of engineers.

Pedestrian said...

What a timely post, given the following story is making the rounds in western Canada:
http://www.globaltvcalgary.com/Fallen+girder+causes+traffic+chaos/3195881/story.html

It reinforces the point that the majority of failures occur during construction, and often due to a very minor temporary support condition that is often designed by a contractor.

Does the book address the more run-of-the-mill structures that us mere mortals design? i.e. multi-span girder bridges, or is it more in the realm of long-span structures in which significant environmental factors (wind, seismic) can really generate problems?

The Happy Pontist said...

I'd say it's aimed at run-of-the-mill structures more than anything else, especially with the amount of space it devotes to construction and falsework related failures.

Anna said...

Hi,

I was hoping you could provide me with details of the photograph of Tay bridge featured in this article.

Your advice on this will be greatly appreciated.

The Happy Pontist said...

I don't recall where I found the image, but it's the cover image from this book: http://happypontist.blogspot.co.uk/2010/08/beautiful-railway-bridge-of-silvery-tay.html?_sm_au_=ijVV478fjsZpNPDF