August 5, 2007

A Primer On Non-Destructive Testing

Earlier today, I noted that one focus of the collapse of the St. Anthony Bridge will be the methods used to inspect the bridges in Minnesota and across the nation. The latest inspection, less than three months before the collapse, never indicated that the bridge was in imminent danger of collapse, but the Star Tribune reports that the inspections mainly rely on eyeballs and ears as the highest-tech devices to find structural deficiencies. The technology of bridge inspections remains largely what it has been for the last several decades.

My father, Ed Morrissey (Sr), spent 29 years working on the space program, most specifically on the kind of non-destructive testing technology that would replace those eyes and ears on the bridges. He held a high certification level -- high enough to have a successful post-retirement career as a consultant at the same company from which he retired. I asked him to give us a primer on non-destructive testing techniques, their applications, and their limitations -- and he responded quickly.

Non-Destructive Testing

There are problems with testing structures for cracks. Some of these problems are: surface rust, paint, build-up of grime and also the way the structure is designed. Assuming there are no inherent flaws, you would be testing for deep pockets of corrosion and cracking.

There are a number of methods available to search and locate flaws. First it would be necessary to identify the areas in which cracking or corrosion would cause catastrophic failure. Then you may use one or more of the non-destructive testing procedures such as:
penetrant inspection, magnetic particle inspection, ultrasonic inspection, radiographic inspection, eddy current inspection, acoustic emission inspection, and even thermal measuring techniques.

Each of the above methods has the ability to detect certain types of flaws:

  • Penetrant inspection will detect cracks that are open to the surface, but you must have a clean surface to apply it to get satisfactory results.

  • Magnetic particle inspection may be used to look for cracks which may be under a painted surface; however, the material which is inspected must be ferromagnetic (magnetizable steel).

  • Ultrasonic inspection will detect surface and subsurface flaws by sound reflecting off the surface of a crack and also detecting deep pockets of corrosion by sound absorption.
  • Radiographic inspection (x-ray and gamma ray) will detect large pockets of corrosion but no small surface cracks. The cracks have to be large and parallel the path of the x-ray or gamma ray when passing through the tested material.
  • Eddy current may be applied to test the metallic structure to detect small cracks. However, since bridges are constructed with magnetic steel for the most part, one has to eliminate the response of the magnetic field.
  • Acoustic emission testing measures the noise created by stress on the structure parts under test to identify those specific sounds emanating from a crack.
  • Thermal testing is performed by applying heat to the structural point and accurately mapping the heat dissipation for anomalies caused by crack or other flaws.
  • Each of the above methods has its own advantage in the detection of flaws, but they may only be of use under certain conditions. It requires a high degree of training and experience to apply these methods. Therefore most of this type of testing is performed by materials testing companies who specialize in these techniques.

    The Admiral Emeritus offered a few more salient points in a follow-up call. In order to do this kind of testing, one has to already have identified the structural elements that would have the most likely fatigue or failure points. None of these tests could be easily conducted on an entire bridge structure; they're costly and not intended for 100% review. Inspectors would have to develop procedures and standards for the tests, and they'd likely have to do that for every bridge that they inspect.

    Some of these tests do get performed for transportation system inspections. Railroad track inspections use NDT techniques, for instance, when they suspect a problem exists. The equipment could get positioned in place using cherry pickers and other lifts if used for bridges, but it would be slow and expensive to do.

    Keep a link to this post handy. When more comes out about the limitations of the current inspection regimes and NDT techniques get discussed, it will make a good reference. Perhaps my father will expand on this technology in between rounds of golf as the story develops.


    TrackBack URL for this entry:

    Comments (19)

    Posted by Weight of Glory | August 5, 2007 2:32 PM

    All of the structural dwgs, I work from require those methods of inspection (mainly ultrasonic) prior to the engineer passing the building. I only deal with commercial steel construction, but I assumed that these inspection practices were also required on bridges. Does your dad know if they are in fact required?

    Posted by Carol Herman | August 5, 2007 3:15 PM

    This failure was so catastrophic, it sheered through the structure and took it all down. (Just like in Chaos Theory, where you hear a butterfly could flap her wings in China, and there would be consequences thousands of miles away, with winds at hurricane force.)

    I'm not so sure you ever know when something gets to be this bad. Or if the jackhammering where roadway was being removed, didn't in some way add to the "harmony" that was going through the fatigued metal. No one as yet knows.

    But fall it did.

    And, it also presumes testing eliminates bad things happening in the future. When this doesn't happen, either.

    I remember seeing a film of the bridge that came down in Washington State. (Kurt Vonnegut's sister was on that bridge! With her husband. Her children were not in the car.) Again, it's the way random events sort themselves out.

    Later, in a wind tunnel testing at Caltech, Richard Feynman, I think, figured out how the wind shear took that bridge down.

    Who knows, IF one answer isn't to divert traffic for miles; and leave the structure exposed only to various testers? People won't be happy with that! Even repairing bridges tries to encompass "some traffic" on them. While roadways are worked one-by-one.

    At least this wasn't terrorism. But given our recent experiences, when the news first broke, that was what lots of people thought had happened.

    Now? We've got to become more aware of the cumulative affects of wear and tear.

    And, as we progess ahead, let me guess that this bridge will be replaced in less than two years' time.

    Winston Churchill once said it, after the Market Crash in 1929. Then? The subject was greed. Something that you cannot prevent. Humanly impossible.

    On the other hand, said Churchill, what America had, and no other country has, is that it didn't collapse. It came in and had the laws that made cleanup possible.

    Yup. We have those laws.

    Posted by cirby | August 5, 2007 3:24 PM

    The problem is that, even if you manage to test 100% of the actual metal bridge structure, there are still a lot of things that can go wrong that you WILL miss. Like the concrete supports, or the ground underneath those supports, or a bad batch of incorrectly-formulated metal in the bridge itself (which will show up as 100% in the tests mentioned above, but which could be only 1/3 to 1/4 the strength of what you designed the piece for).

    The easy way to handle many problems (as I mentioned in other posts) is to massively overdesign the structure to begin with.

    Look at Interstate highways in the middle of the upper Midwest. They tend to be thicker and more expensively-built than the highways in the South - because it's cheaper to make them much thicker and tougher than to have to go out every six months to resurface them (from the heavy truck traffic and hard winter weather).

    Posted by cirby | August 5, 2007 3:33 PM

    One of the big lessons in this particular disaster is certain to be "more redundancy in big structures."

    Generally, we don't build things like the I-35 bridge any more. It had, basically, no redundancy for a structural failure (one "foot" goes, and the whole thing comes down).

    A modern engineer would have had a much different structure, with more pilings and a less-critical set of failure modes. Instead of four pilings, it would have been six or eight, and the "box" itself would have been of a very different design - if they'd even gone with the same general structural idea to begin with. There are a lot of clever designs that have only become practical in the last ten to fifteen years...

    Posted by docjim505 | August 5, 2007 3:46 PM

    Cap'n Ed,

    Thanks to you and your father for a quick overview of NDT. Can you throw some light on how bulky the test instrumentation is? How easy is it to use on a large, outdoor structure like a bridge? How long would it take to inspect the entire structure?

    On another note....

    So... It might NOT be George Bush's or Tim Pawlenty's fault??? IMPOSSIBLE!!!


    We've reached a sorry place in this country when an accident occurs and the first reaction of some people is to start playing political games. I'm waiting for the conspiracy theorists to start coming out of the woodwork. Maybe Rosie O'Donnell can announce to the nation that steel and concrete bridges CAN'T collapse, just as fire CAN'T melt steel. Perhaps Calpyso Louie Farrakhan or Jesse "Pimp Daddy J" Jackson can put a racial angle on it and prove that George Bush blew up the bridge. I hear the Michael Moore has already started filming his next masterpiece.


    Posted by RBMN | August 5, 2007 4:41 PM

    Future bridges will probably have pressure and vibration sensors built right into the structure, and if any of the sensors goes outside design parameters, the bridge is automatically closed until the anomaly is explained. Compared to the cost of a bridge, a sensor system like that would be pocket change.

    I did hear one news report that said: workers resurfacing the deck thought the bridge had abnormally large sways, or vibrations. If that's accurate, I'm sure we'll hear more about that.

    Posted by chris edwards | August 5, 2007 5:27 PM

    There is a a human/greed flaw in any testing, I knew north sea divers in the 1980s, one job was ndt on the legs of rigs, often they do a thorough video of one leg and label various scenes as bits of all legs, then a leg sheared and killed a bunch of men. Things were tightened for a while then slip back, reason-- cost, if a repair is needed the divers go home and rig dry docked, all down the line money is lost and because there had been no recent failures then it must be OK, there are some rather old bridges that have lasted the years, some pushed the envelope (Brunels Saltash train bridge is one and mr Eifels is another) but a lot of crap has been built since.

    Posted by H. Tietz | August 5, 2007 5:31 PM

    I am also an old hand in the NDT (NDI) business (over 40 years nuclear/aerospace). NDI is a probabalistic science. As was stated, the probability of detection (pod) is dependent on technique, surface condition, flaw size and orientation, and inspection environment, such as hanging from a rope 200 feet inthe air over a river in the winter. Pod is normally stated as a % probability that a % of flaws of a specific minimum size can be detected. As the admiral stated, pod is extremely poor for ferritic, corroded, aged structures such as bridges. In aerospace an ndi technique is not nromally accepted unless there is a 90/95 pod. On bridges, it is probably very much lower. Don't blame the inspector; the techniques are not adequate for the applications. You use the best you have. Tapping structures with a hammer in experienced hands can produce very good results, if that is all you have to go with. We manually "coin tap" aircraft every day, and sometimes we use an electro-mechanical tapper such as mechanical impedance analysis (MIA)

    Because of this, there are charlatans and snake oil salesmen, some with impressive credentials, that will sell reporters and congressmen the answer to all their prayers. They will bring a known, impressive flaw with them to demonstrate thair equipment and it will find it. Therefore, ayone that doesn't buy that equipment is placing babies in danger and is criminally negligant. Unfortunately, that equipment often has a pod less than 50/50 (chance).

    I have spent many years white hat testing, sorting through, culling, and transitioning NDI equipment. Maybe one in ten are developable to an acceptable detectability, but that is generally for a specific application. The Militaty has a cooperative program among the services to find and help develop these applications. DOT also cooperates in this development. Maybe they need some separate money to target their applications.

    I (from Oklahoma) know impressively well qualified companies and technicians in MSP. Some of the best contract with the local airline.

    Although under continual reconstruction, Minnesota has some of the best roads with the worst traffic I have ever seen. I hope you get everything fixed, the walleyes are awaitin'.

    Posted by Ray | August 5, 2007 7:34 PM

    Ed, thank you and your father for this information.

    "Tapping structures with a hammer in experienced hands can produce very good results, if that is all you have to go with."

    I used this very procedure when I was a Cobra mechanic. It was the easiest and most reliable way to discover voids in the laminated aluminum skin of the helicopter. It is still being used to discover voids in metal structures like the beams and struts in bridges.

    Of course, it seems more art than science but anyone with proper training can discover even small voids in metal (and other materials) using this procedure. The human ear can distinguish small subtleties in reflected sounds that other methods may miss when testing large areas. Since there is little perpetration required, it allows one to test large areas quickly and effectively. Once an area is marked as suspicious, other methods can be used to determine the actual extent of voiding and repairs can be made to those areas that actually need them.

    Posted by patrick neid | August 5, 2007 8:55 PM

    and then there are the "unspoken" actuarial tables.

    most of the bridges now fretted about will still be standing in 50 years. yes there will probably be a few that collapse but the cost to replace all bridges to prevent a few failures won't pass the green eye shade test.

    in layman's terms---do we spend 200+ billion dollars to possibly save a hundred lives and whatever the surrounding area's economic impact will be? thankfully, in hushed tones, we won't. Bridges? hell, we still get on planes--more of those crash than bridges collapse. in the meantime we have to put up with the people who see a parade and run out in front and act like they are leading it.

    life's a bitch. it led hindus to believe in karma.

    Posted by Robert I. Eachus | August 5, 2007 11:32 PM

    Watching the 4 frame/second footage of the collapse about 50 times, and reading some of the preliminary accounts of the collapse, I suspect that the cause will be found in or under the northern footing.

    This should be taken as a wild SWAG for now, but an intelligent one. The truss did not fail, it fell as a unit. If you watch the footage long enough you can figure that the truss got pulled off the northern supports then fell as a unit, also unsupported at the southern end.

    If, and note that it is a big if, I am right, all the King's horses and all the King's men could not have detected that fault in advance. Especially if the failure was not due to undermining by the river--I assume the pylon was built on bedrock. The fault mode I suspect is a sheer fault in the concrete. Reinforcing bars (rebar) and aggregate are added to concrete to prevent such faults from spreading. Minor cracking in the concrete is impossible to avoid, but it should be confined to the surface. If cracks get deep, especially in Minnesota with lots of freezing, you can have freeze thaw cycles forcing cracks open. Not that I think that was the primary, or even a contributing cause of failure in mid-summer. If the weather was a contributing factor, it would be extra stress from the heat expansion of the bridge on a hot summer day. Late afternoon is about the right time for that.

    What I am talking about is ordinary aging of the support resulting in a stress plane through the base of the support where the forces along the opposite faces of the stress plane are transverse to the plane. In such a situation, the sheer stress tries to force the fault open, the compressive stress holds it closed. Shake the support, and eventually it will pop open at the stress plane. But until it fails, no method of non-destructive testing will find it before failure. As long as compression dominates, there will be no visible crack.

    Well, that's not the best way to explain it. Using techniques like x-rays and ultrasound, you still won't see the stress plane, even in the lab, until it opens. Of course, there is little or no hope of getting good x-rays (or ultrasound images) of a bridge support.

    What hope is there of avoiding this type of failure? Good design. But at the time this bridge was built, the type of analysis that would be necessary to avoid this problem would have been beyond the state of the art. Even today, doing a large computer FEA (finite element analysis) of a bridge design is difficult. The cost today of not doing a full-up FEA model of a new bridge is much greater than subjecting the design to a pretty thorough simulation test, including fires, floods, earthquakes, and truck crashes. So whatever replacement bridge is built, you can expect it to be much safer than even the best 40 year old bridge.

    On the other hand, ignore all the talk you will hear about how the design without a center support was inherently less safe than the actual bridge design. Two, supports, three, or twenty, if one fails catastrophically, part of the bridge will go into the water. More supports may make a bridge easier to maintain, and may make it possible for the bridge to survive moderate damage to one support. But catastrophic failure of a bridge support is going to be, well catastrophic. So to some extent more supports can make a bridge more likely to fail. If you don't understand, look at the history of barges crashing into bridge supports on the Mississippi River. Fewer stronger supports makes such an accident more survivable for the bridge. (And also eliminating some of the supports makes such a collision less likely. ;-) That probably was considered when this bridge was originally built.

    Posted by edncda | August 6, 2007 12:36 AM

    A government inquiry into a sudden bridge collapse in Laval Quebec last year (5 people died) has been finding that the so-called experts who we trust to keep us safe seem to be assuming that this sort of tragedy is impossible - so they can afford to be lazy, presumptuous and perhaps even incompetent. As a direct result of this inquiry, the City of Montreal has gone from saying first: "All the other bridges are safe" to: "We have ordered extra inspections of certain bridges" then to: "Weight restrictions on certain bridges are deemed prudent "(No heavy trucks - on the honour system!) to: "We are closing certain lanes of certain bridges as a safety precaution only" to finally: "This bridge will be demolished as soon as possible".
    The conclusion to be drawn seems to be that the people we trust to protect us (in this case at least) are incompetent, and/or negligent and/or not properly trained for the the job. Visual inspections are considered adequate, even when broken, rusting rebar is poking through the cement!
    The head of the government commission has been so surprised by the shocking nature of the testimony that he has ordered the proceedings be published in their entirety on the internet:

    The majority of this information is in French only but there is a wealth of (heretofore unpublished information) on this subject and is catching the attention of structural engineers worldwide.

    Information on the actual bridge collapse and the aftermath can be found here:

    Posted by rml | August 6, 2007 1:30 AM

    I believe you're thinking of the Tacoma Narrows ("Galloping Gertie") failure. The wind tunnel failure analysis was led by Theodore von Karmen.

    Posted by amr | August 6, 2007 7:16 AM

    My experience with non-destructive testing involved minimal direct testing but decades of observation and decision making based on the results. To me NDT is, to a large degree, an “art” performed by those with experience and knowledge. The technical tools are there to be used but as has been stated previously one has to know where to look, what to look for, which method is best and how to interpret the results. I expect the inspectors to be crucified by the media and politicians, but as I have experienced, the testing results had to be reviewed by engineering and management and a determination made as to the best course of action. Everyone needs to keep in mind that decisions are made on “a what if” thought process taking into account all of the possible scenarios of the situation, such as closing down the bridge and impacting thousands of businesses and workers, when no one can say for sure that there is a danger that the bridge will fail in any manner or collapse. So if you were one who believed that the bridge was unsafe and needed to be closed down having no absolute demonstrable facts backing up your position, imagine how you justify to the public and politicians that they need to be extremely inconvenienced, spend 100’s of millions of dollars on a replacement bridge and possibly cut into other projects’ funds that have strong and influential advocates. Remember that the three engineers at the Challenger launch could not overcome the need to make up for the schedule slippage in the shuttle program armed with their facts that the launch temperature was outside of the design test data. It takes an extremely strong person or group to advocate an unpopular position and then to have a management team that will decide to do what is most likely correct rather than to go the way that most outside of the decision making process would want; even when lives are potentially at stake. To take the unpopular position in such an instance can cost you your job and possibly your career or worse.

    Posted by agesilaus | August 6, 2007 9:28 AM

    THe Federal DOT should consider doing something the NASA way. Perhaps they should offer substantial prize, maybe $10,000,000, for an innovative and effective method for doing bridge inspections. And then stand back and let private enterprise and the universities work on it for a year.

    And perhaps a another prize for improved ways to access the underside of bridges for the inpection party.

    Posted by Wayne | August 6, 2007 9:49 AM

    I keep hoping that someone finds signs of explosive residue on the base structure of the bridge.

    Posted by physics geek | August 6, 2007 10:47 AM

    NDT is widely used in the nuclear industry, especially since many/most of the older operating plants have had their original licensing periods extended by another 20 years. Aging components need to be monitored and repaired/replaced as needed.

    Sure it's expensive to institute and maintain such a program. However, it's a lot cheaper than the alternative.

    Posted by deadrody | August 6, 2007 12:08 PM

    Yes, geek, you are correct. I don't know how much of the DOT bridge inspection program mirrors a nuclear plant, but I can lay out a nuclear plant's NDE program fairly quickly...

    There are a number of areas in pressure vessels, nozzles specifically, that are more prone to cracking and failure. The NDE program is setup so that all of those areas are inspected once per 10 year interval. If cracks are found, weld overlays are performed to repair them, and the inspection sample size is increased.

    I can only assume that the inspections of this or any bridge are handled in the same way. During original construction, however, every structural member and weld probably would have been tested and signed off as acceptable.

    It's interesting that someone mentioned barge crashes. I recall doing a case study on the fatal Amtrak crash in the south that was a result of a barge colliding with a bridge support. I would not be the least bit surprised if that were at least a contributing cause in this case.

    Often times in significant events like this, numerous factors align to allow seemingly small errors to result in catastrophic results. Nobody had to have committed an egregious, obvious, or grossly negligent error for this to occur, as counter-intuitive as that may seem. It could be as simple as someone failing to carry the 1 during the original design that resulted in a high stress member being overlooked in the subsequent NDE inspection program.

    I would also point out that most structural designs employ factors of safety on the order of 10. Meaning that if the failure point of a steel member would occur when subjected to, for instance, a 20 ton load, that the bridge would be designed to carry a 200 ton load. Certainly, cracks and flaws in either the steel or the concrete would undermine that strength, but still, the point is that it takes quite a failure to overcome such a high factor of safety. It is likely that an external event contributed (possible barge / support collision).

    Posted by Conrad Wareham | August 6, 2007 1:12 PM

    Ultra-sonic testing also reveals the thickness of the metal remaining after corrosion is removed from the surface. What effect could the de-icing agent have on the bridge structure if it was salt based? It seems to me that if you have a metal structure you have to maintain it from corrosion from the elements. Metal water piping will eventually corrode and fail if left unprotected from the elements - what is different about a metal bridge? It will probably be found out that this bridge had many structural weakpoints. In my experience, (we maintain private and municipal water systems) we have sometimes been called out to make an assessment on a system that had failed, and in this assessment we find multiple problems; mainly due to a lack of preventive maintenance. I suspect that this bridge will reveal many problems that contributed to it's demise.

    Post a comment