May 23, 2004

Paris Airport Collapse

A section of the new passenger terminal at Charles de Gaulle airport collapsed this morning. After the now obligatory denial that any terrorism was involved, officials declared, "It's the structure that gave way, the structure itself."

The article provides a photo of the collapse in which one can see an intact section of the structure next to the failed section. An elegant and futuristic web of steel and glass covers an internal structure that can best be described as a perforated oval concrete tube. Notwithstanding the questionable aesthetic appeal of walking through a giant sewer pipe, one is left wondering why concrete was used as the structural component. Modern polymer supplemented concrete is a remarkable building material, enabling clear-span bridges hundreds of feet in length. But its strength lies in its mass. The structural design feature of strength versus aesthetic design features of light and visibility are at odds with each other. So this design approach represents a challenge, though a forced one rather than an elegant and delicate balancing act, demonstrating the human ability to successfully design and construct unlikely structural elements.

The problem is that this failure gives engineering a bad name. And why? Because the final design uses window openings with square corners! And in an airport, no less. The problem with square corners is that they create what is known as a "stress concentration" point. (This mode of failure occured in early aircraft designs using square windows as well.) The stresses in the areas near the corners of the openings can be orders of magnitude (10, 100, 1000 times or more) higher than in the rest of the structure, while the simple incorporation of radiused corners can reduce these stresses to geometric multiples (2, 4, 8 times or more) the nominal stress. Look closely at the picture and you can see cracks intersecting the corners of the windows. These cracks may have been a result of the root cause failure rather than a cause of it, but they are failures nontheless.

Another pitfall of these stress concentrators is that seemingly minor variations in the fabrication process can create unpredictable weaknesses. These variations include things like the unfortunate location of a large piece of aggregate at the corner, or a slight undercut at the corner from ill-fitting forms. In other words, no experienced engineer worth his salt would agree to a designer's demand that the windows have square corners. The verdict must certainly be that the engineers were either inexperienced, incompetent, or politically subservient to the vanity of the designer. Any of these scenarios are a recipe for disaster, as we can clearly see.

UPDATE 24 May- Fox News reports "France may tear down terminal after roof collapse." That's exactly what they should do if, as I postulated yesterday, the failure is due to unsafe design. The FNC article even mentions the "hundreds of square windows" that "honeycomb" the roof, but with no suggestion that this is a problem.

French newspaper Le Parisien asks on this morning's front page, Comment est-ce possible (How is this possible?) Using Altavista's Web page translator I examined the possible explanations the paper gave: The building's opening was delayed one week pending approval of the "commission of safety;" a workman was killed on the construction site; a large lamp fell from the ceiling (during an inspection, I understand); the floods in France last March were a concern; construction union representatives denounced, at the time, the pressures to open on schedule; and finally, "the report of the low "relative" cost of the buiilding..., established by the Court of Auditors at the beginning of 2003, calls other questions today." And yet, still no mention that the problem may lie with the design.

"How is this possible?" As I explained yesterday, incompetence, inexperience or politics are likely explanations. These are classic traits of a collectivistic society such as France's. No individual takes responsibility for fear of being reassigned. Individual ability is not encouraged or nurtured in education establishments. Assignments are awarded based on "pull" or influence, and not on ability or merit. This is why it is imperative that we resist the growing influences of relativism and multiculturalism in our universities. That is what makes tragedies such as this "possible."

Posted by JohnGalt at May 23, 2004 10:08 AM
Comments

I hate to make light when lives were lost. My sincere condolences go out to all who lost loved ones. But perhaps if the engineers were to work 37, 38 or (GASP!) 40 hour weeks, better structures could be designed.

A society cannot stop innovating and advancing to seek comfort. I am in the middle of Michael Barone's "Hard America, Soft America" and I strongly believe you would not likely see this in America.

Posted by: jk at May 23, 2004 09:26 PM

Yes, France's government imposed restrictions on ambition are a bad idea, and have a negative impact on their economy (though the Germans manage to do more in 32 hours per week than the French did in 40.) Yet this is not the cause of the ills we see in this case. (See UPDATE above.)

As for this not happening in America, don't be so sure. Right here in Denver we saw an incorrectly installed steel girder collapse onto a highway killing a family of three (the same death toll as in the French airport.) The assault on competence is a worldwide movement. The American spirit of creation and achievement IS the world's best hope for defeating it, however.

Posted by: johngalt at May 24, 2004 09:01 AM

Wow, an in depth engineering failure analysis from a single photograph, I am impressed. Is this really from the author of Truth and Legitimacy right here on this blog?

Posted by: Silence Dogood at May 24, 2004 01:17 PM

Yes, I do have extraordinary abilities. That's why I'm JohnGalt and you aren't. Seriously though, you might want to re-read the parts of my blog that include "must certainly be" instead of 'is,' and "postulated" instead of 'proved.'

Posted by: johngalt at May 24, 2004 11:41 PM

Sorry, but I just have to take you and JK to task for this one. For two who complain so mightly about bias in reporting you have taken a very small amount of data, added a heap of bias, jumped to unsupported conclusions, and even gone so far as to assess blame to the socialist system.

So, "must certainly be" is a caveat, not an indictment? I quote, "In other words, no experienced engineer worth his salt would agree to a designer's demand that the windows have square corners. The verdict must certainly be that the engineers were either inexperienced, incompetent, or politically subservient to the vanity of the designer." Now I will refrain from commenting on the design of the structure because I, and so too you, really have no idea how it is structurally supported or what may have caused it to fail. Note however that while your information on stress concentrations at square corners is technically correct it is for structures in tension. Concrete is used in compression, steel rebar or outside structural elements handle the tension loads. Now it is certainly possible that there was a design error here or that a loading put the concrete in tension and caused it to fail at the square corners, but it is at least equally possible that the concrete was in compression and formed a shear web between the steel rings such that the square corners were of no significance structurally. As to the assertion that no designer worth his salt would agree to square corners in a concrete structure I would have to direct your attention to about 99.99% of the concrete structures with windows. I cannot think of a building I have been in that didn't have square windows - obviously driven by the socialist cabal of window makers who refuse to put radii in their glass.

But wait, it gets better! "These cracks may have been a result of the root cause failure rather than a cause of it, but they are failures nontheless." A great statement that cannot be refuted which is at the same time completely meaningless. Yes cracks in a sturcture would have to be defined as failures, but if they were not the cause what does that mean? If I smash my keyboard with a hammer it will certainly be left with some cracks in the plastic housing so I can conclude that my keyboard structurally failed. It would be ridiculous however to conclude that it was not designed or manufactured correctly because of this failure.

I think it is a bit early to start tying the lynching rope for the designers and engineers, and I can't help but wonder how your report would have been different if this had been a new Iraqi school built by the Army Corps of Engineers.

Posted by: Silence Dogood at May 25, 2004 10:15 AM

Mama told me not to get in the middle of two mechanical engineers fighting -- this software guy will offer no exegesis of cracks or corners.

But Silence, the good people of France have taken a nation and people that were world leaders in science and innovation and have put them what Hayek called "The Road to Serfdom."

Yes, I should not go off on this like I did with insufficient information, but I remember 10,000 elderly citizens dying because of insufficient air conditioning, and a French Navy that boasts one aircraft carrier that cannot leave port.

So, the fallen building plays into the stereotype and I ran with it. Mea minima culpa.

Posted by: jk at May 25, 2004 12:57 PM

I thought it was clear from my dissertation that I was referring to square cornered windows in THIS structure... one that uses concrete as an arch to support the roof. And if all of the concrete were in compression then how would any of it be able to fall, even if it were riddled with cracks?

The point of saying that the corner cracks were failures even if not the root cause failure was to identify the window corners as a weakness in the design. In concrete pavement one makes saw cuts to intentionally create such weak points. In an arch structure such joints may also be employed, but not likely at every window opening.

I fully realize that my immediate assessment may ultimately prove partially, or even fully, incorrect. Please keep in mind that I didn't conclude the designer should be jailed on my analysis. It's only a prediction. I'm still convinced of my original judgments, however: 1) Square cornered windows should not be used in the walls or roof of a tubular concrete tunnel; 2) There are far more elegant (and efficient) structural materials to use in supporting a glass tunnel than concrete; and 3) The realities of France's statist society contributed to the eventuality of this disaster.

Posted by: johngalt at May 26, 2004 08:58 AM

Thousands of pardons JK for this lengthy (boring?) comment on engineering principles. It is just interesting to me that bias against the French has led you to run headlong down the path of blame, something for which you often rightly accuse the liberal press. To avoid the appearance of squabbling and just for my own experiment I would like to just toss out a couple of facts and watch the esteemed powers of critical reasoning take flight to banish ignorance, bias, and self delusion.

It is quite correct to think of the tubular section of the airport concourse as a arch, or two arches mirrored around a plane. It would in fact be quite proper to perform a structural analysis in this manner, reflecting the reaction loads at the imaginary plane. An arch is in compression, this is precisely the point of the structure and the reason it has been used so successfully for thousands of years. The world is quite literally filled with examples of ancient arches (and domes which are quite simply revovled arches) riddled with cracks that still stand. In fact it was the observation of this that showed more modern engineers the real capability of the arch, that much thinner cross sections and even cutouts were not only feasible, but in some cases desireable to lower the weight of and thus the stress on the structure.

Saw cuts in concrete pavement are primarily to allow expansion and contraction with temperature. Yes, cracks do often develop as the ground underneath may move in a non-uniform manner, one area more than another, thereby introducing tensile loads for which concrete is quite weak. This effect is mitigated by the use of steel reinforcing rod placed so that it coincides with the tensile section of the slab whereby the steel rod handles the tension load. Further improvements can be had by pre-stressing the concrete - applying a tension load on the rebar with plates that transfer this to a compression load on the concrete. Thus pre-compressed it has that much higher allowance for tension loads before cracking. Even further improvements can be had by casting concrete in a mold like structure rather than pouring it in place. This is very likely the method used to make the concrete sections of the airport concourse. In fact pre-cast concrete has much higher material properties than the poured in place variety familiar to anyone with a patio or driveway.

So this structure was most likely built with precast and prestressed concrete with internal and external reinforcement in an oval tubular shape that by its very geometry would place the structure in compression. Thus designed, stress concetrations from square corners would not be of structural significance. I won't comment on the elegance of the design, but in terms of cost efficiency concrete (and other forms of masonry throughout history) have proven to be tough to beat in the construction of large civil structures.

Side note on the relevance of the similarly shaped aircraft fuselage (perhaps the look the designer was going for?) that seems to have started your line of reasoning. A pressurized fuselage is in tension by fact of the pressure within it, it is for analysis' sake a pressure vessel. Long term studies have shown that the cycles of pressurization and depressurization are the leading cause of fatigue cracks in aircraft fuselages. Again, stress concentrations are crack initiators in tensile structures.

Posted by: Silence Dogood at May 27, 2004 10:14 AM

I still have to put the spurs to this horse a little longer. He's not quite dead yet but he's definitely gumming his feed.

I too find it interesting that you found my primary motive in this story to be France bashing. I assure you my reaction would be the same if this building had collapsed in America. This country has had its share of civil engineering failures too. The reports of state control in French civil industry tend to buttress my speculations about political considerations trumping engineering ones.

Now, if the vaunted French had performed their structural analysis in the manner you suggest, "two arches mirrored around a plane...reflecting the reaction loads at the imaginary plane" that could explain the failure. While the upper arch, if properly designed, could be completely in compression, the lower arch would be almost completely in tension to support the upper load. According to the article below, each 4m-long segment of the tube was formed from three precast elements that had to be "supported...on a rail-mounted frame until they were stitched together with reinforcing steel and concrete." You are correct that there is far more information needed to reach a DEFINITIVE conclusion, but the effects of square holes in non-vertical concrete structures may not be dismissed out of hand.

http://enr.construction.com/news/buildings/archives/040531-1.asp

Additionally, vibrations in a concrete structure from nearby vehicles, sound waves, and seismic activity are analogous to the pressure cycling you mentioned in aircraft, though with lower magnitude and higher frequency.

Posted by: johngalt at June 1, 2004 08:04 AM

Don't get me wrong, I am definitely not defending socialism (French or otherwise) and I am with you and JK on the detrimental effect it has on technological innovation. I am just trying to keep it factual, critical reasoning and all. If you are looking for weak points in the design I would suggest looking at the pylon structure that holds the tube off the ground. Ground movement (local expansion and contraction) is the single leading cause of structural failure, more than all other causes combined, including all the natural disasters such as earthquakes, floods, and storms.

You are still looking at an arch as beam I think, which it is not. A horizontal beam under a vertical load has tension and compression elements. An arch under vertical load converts that load into a vertical and horizontal component. The vector sum of these is tangent to the arch at any point along its length. This is true for an upturned arch as well. You can in fact build an arch or a tube with blocks cut in the correct wedge shape with no mortar or adhesive at all. Compression holds the structure together.

Great that you found more info on the design. Precast was the likely manufacturing method. That fact that the sections had to be supported until they were stitched together is quite normal. In fact most bridges see their highest loading during consturction. Think of the three sections as each being a "keystone" whereby upon fitting them together each balances the weight of the others.

I agree with the seismic concept, see above for possible failure areas with the support structure tying the tube to the ground. The other vibrations you mentioned are not likely candidates becasue of the low magnitude.

Hi JK! Taking all this down and drawing vector free body diagrams no doubt.... I like your example of the aircraft carrier Charles de Gaulle much better as an argument against state control. No doubt some of the problems could have surfaced with a free market enterprise, but I suspect most of them would have been avoided with a competitive bid process. Interesting that the French government has cancelled the order for the second carrier of that design and is seriously considering just buying one of British design.

Posted by: Silence Dogood at June 1, 2004 12:42 PM

Yes, the French carrier uses an inverted-Reinmann-toraul drive propeller which diverts the torque normal to the hull's downwash. I think they'd be better off with a...

Hey, maybe there's a big demand for an engineering blog. This is starting to push the comments-per-post records normally reserved for thongs and gay marriage. Don't stop on my account!

Posted by: jk at June 1, 2004 01:50 PM

Well.... OK.

I'll stipulate to the specific terminology. An arch is completely in compression, and an arch with any element in tension is an arched beam.

Now, can you cite any examples of an inverted arch, i.e. no tension elements?

Posted by: johngalt at June 2, 2004 09:39 AM

Jazz, Guitars, and Right Wing Politics gives way to Gays, Arches, and Thongs? Surely would increase the hits on the website.

JohnGalt I think you've got it. Inverted arches with no tension elements are not that common but are used occasionally just where you might think, underground, tying together sets of piers or posts when deeper tiered footings can not be used. The more common use does involve a tension member running across between the legs, in the case of the airport concourse this is the floor upon which the passengers walk. Regular arches are often supported this way as well, as in any truss arch bridge. An arch is supported either by abutments outside the legs taking a thrust load as in the classic stacked block arch or by a tension member running between the legs. A more common example of an inverted arch is the hull of a ship in classic ribbed design where the deck is the tension member. Even in these cases though the tension element is outside the arch and takes all the tension load of the structure. The load in the arch itself is pure compression. Note also that the classic arch is actually considered an inverted version of the catenary arc, the shape formed by a slack string or cable held only at the ends.

Ok, I think I am done here. Back to some good Liberal vs. Conservative arguments.

Posted by: Silence Dogood at June 2, 2004 01:03 PM

My mistake for not asking for a STRUCTURAL inverted arch with no tension elements. How interesting is an arch completely supported by earth, after all?

Posted by: johngalt at June 3, 2004 10:54 PM
| What do you think? [14]