Future Forward Interview: Ensuring Institutional Memory of Old Steel-Bridge Structures
Informed Infrastructure met Robert J. Connor, associate professor in the Lyles School of Civil Engineering at Purdue University, at the Commercial UAV Expo, where he gave a presentation on the use of UAVs for bridge inspection and, more importantly, the need to do a better job of inspecting bridges. In addition to his teaching duties, Connor also is director of the Center for Aging Infrastructure (CAI) and the Steel Bridge Research, Inspection, Training and Engineering Center (S-BRITE).
I2: What was your impression of the Commercial UAV Expo?
Connor: I think the area of inspecting infrastructure is all headed that way in the future. There were some great presentations on imaging capabilities and how you can use UAVs to deploy the camera or sensor. While I think it’s a great opportunity, we also have to move forward cautiously, in other words, we have to sure we don’t mess up. I think one thing that is critical is the idea of setting standards on approach and image quality, etc., I think that will go a long way to add legitimacy to the industry.
I2: Are there efforts that you’re aware of on approaches for setting those standards and instilling a rigor on accuracy?
Connor: There are a lot of people trying to figure this out. They’re working on getting approval from the FAA, and all of those regulatory issues with the UAV access, and they know drones are a great tool. However, I don’t think anybody is looking to set standards on the inspection quality, though. The producers that I spoke to, and those who came up to me after my talk, nobody was aware of any standards that you would have to meet in terms of image standards, lighting standards, access capabilities for say, inspecting steel girder bridges, etc.
I think the highway transportation industry isn’t that good at that side of the technology, that is when it comes to inspecting existing structures and validating what we can actually detect. We’re very good at making sure at ensuring material meet certain standards. For example, producer of high-strength bolts must meet a certain test specification; the producer must test x number of bolts and show they meet the specification. When it comes to materials or other structural components, I think we do a very good job. However on the inspection/detection side, I don’t think we have really spent enough time developing methods to ensure we can actually find defects, or spent effort understanding the variability between inspectors. For example, as related to visual inspection of bridges, we just sort of assume it works. That is when an inspector submits a report, we assume that is the truth and there is no variability or possibility something could have been missed. Even if someone understands there is variability, there are no sets of data that can be used to determine the likelihood that a certain type of defect will or will not be detected. Basically, there’s a lot of interest in detection techniques, but nobody ever has to demonstrate that it can do exactly what they say it can do, simply because there’s no standard.
I2: In your presentation, you discussed the failure of many inspectors to find defects even though they were there. Is that the first step toward doing a better job of sensing?
Connor: The message that Glenn Washer, my colleague at the University of Missouri, and I have been promoting is that we have to be smarter about how we inspect our aging infrastructure. We have to be more confident in the data that we get with our limited resources.
The visual inspection study I referenced tries to understand how we can get better and have more confidence in our inspection data. The only way to achieve that goal is to objectively quantify what inspectors can find and what they can miss. We set out to quantify how well bridges were being inspected now for corrosion, impact damage, scouring and other things. We’ve focused on this one area of looking for cracks in steel bridges. Inspection for fatigue (cracking) is always something that concerns owners of steel bridges and something our industry spends a lot of resources on.
We wanted to get a feel for what inspectors could find and what they miss simply through visual inspection. I want to be clear however, it wasn’t to conclude that inspectors are not trying, or that they are not any good. Rather, we want to evaluate how we can help them become better at it. Once we have some idea of what is missed and why, we can then develop procedures and training that will increase the probability of detecting a crack. As Glenn and I observed UAVs being used for inspection, it led to the awareness that UAV operators and other sensors were making claims that they were better than inspectors. However, there is no data that supports those claims, at least not on real bridge details and nothing we could find in the literature.
First of all, we worked to quantify the success of inspectors for finding cracks. Thus, we now have some data related to how likely it is a human will find a certain fatigue crack of a certain length. But how do we know that a given technology is any better? If it isn’t, then I don’t want it. But right now, we rely on a producer coming to an owner and trying to sell his or her technology based on claims that are not substantiated. Or they say that the technology found this or that, but what is found is not of concern, what was missed is what is important. We would never accept that from people making bolts or producing steel or making concrete. In other words, we would test the bolts and make the producer demonstrate it with a standard test.
I2: I’d like to touch on how S-BRITE came about, it’s history and your approach of taking in older bridges in a way that you can inspect and monitor them.
Connor: There are a lot of different objectives that we hope we can meet. It really came from Indiana having some issues on the Sherman-Minton Bridge over the Ohio River. It’s a long span tied-arch bridge that takes Interstate 64 over the Ohio River. It made national news because it got closed back in 2011 due to the observed cracking. There were some cracks in the bridge that we worked together with INDOT on evaluating and repairing. Purdue has a great long-term collaborative relationship with the Indiana DOT, and they asked where we could collaborate to help improve some things. Basically, as an owner, they don’t want to be in that kind of situation again, particularly looking at steel bridges. After a lot of discussions, together we arrived at the concept of S-BRITE, which is focused on extending the safe life of aging steel bridges.
Every day our infrastructure gets older. However, when we’re building new bridges, we’re using new materials, new technologies and new techniques. But the bridge that was built in 1964 or whenever, most of the people that worked on that, unfortunately, aren’t with us any longer and their expertise has left us. All of the legacy knowledge is gone, but we still have that inventory with us.
We saw on the professional training side that people weren’t familiar with the problems that can and do occur with these older materials and designs. We know we produce good students at Purdue, but they are taught very little about older materials, fatigue, riveted joints, welding and other historic issues. When they go to work for a consultant or for a DOT, if the engineer with the legacy knowledge retired three days earlier, that knowledge may never get transferred to the new person.
Our role is to transfer some of that knowledge in short courses and other outreach that we’ve gained over the years into the people that have to take care and maintain bridges built many years ago. A big component of S-BRITE is the training to help graduate and undergraduate students understand fatigue and fracture, and very few civil engineering programs have dedicated classes in that topic anymore. We provide training on welding, bolting, and even discuss riveting.
The other aspect is on the research side, and in particular, researching deterioration and inspection. The challenge, of course, is having the space to perform long-term deterioration studies or housing full-scale components that contain defects that can be used for inspection training. Basically, it’s hard to find real estate where you can put something and spray saltwater on it.
After much discussion within the School of Civil Engineering and the College of Engineering, we presented our proposal to the Purdue administration, specifically requesting some land be dedicated to a facility where we could perform such research and training associated with aging infrastructure. We were successful and the University provided us with about 22 acres.
The actual bridge component gallery where we put the full-scale bridges and pieces of bridges, that was something I had in my head for 10 to 15 years. My thought was simple. Let’s say there’s a girder or steel bridge that has a problem. We normally take a bunch of pictures of it, and we put it in a PowerPoint and show students the problem. As for the bridge, it gets melted down for scrap. My thinking was that the bridge will eventually go out of service, and why don’t we keep that somewhere and use it for training or research, etc.
For example, one of the concerns with trusses these days is related to gusset plates. These are large complex connections. To try and convey to an undergrad or a grad student or a professional who’s never gotten their head inside a giant truss connection, how are you going to inspect this?, what are the challenges with inspection?, what are the challenges of analyzing it?, how do you make measurements?… is extremely difficult.
So, my thought was to get some big gusset-plated joints and keep them and bring students and professionals to examine them in person. That’s what S-BRITE is trying to do, to create this component gallery where we have these types of specimens that you can go up and lay your hands on.
When I bring my undergrad steel class down there, they’re just blown away. The conversations and discussions that get going are great because now they can look and touch at these things. The neat part is that we don’t have to worry about traffic control. You’re also not hanging on a bridge 100 feet in the air, and it avoids all the liability with that. We go down and walk up to the pieces. There are also unique collaborative opportunities. For example, I might use one for the bridges in a steel design class to illustrate a certain detail or connection. However, if I have a colleague teaching the structural analysis class, he may use the same bridge, but for teaching the students how to accurately model the bridge. Same bridge, drastically different uses. Basically, I’m using the truss to discuss corrosion, gusset plates and members, and one of my colleagues is looking to install some strain gauges and use it in our structural-analysis courses.
We use some software for that now, but we can do full-scale experiments where we apply some dead load to the bridge and take some measurements and figure out why the model doesn’t match reality.
There’s a lot of cross collaboration that we hope to gain. My colleagues in geomatics are doing things with photogrammetry for inspection, and they’re thinking about how to measure corrosion on a gusset plate. You don’t want to go out to a real bridge to take many pictures and try different sensors. We have real bridges and components in a safe but realistic environment that they can use for that aspect.
As I mentioned, the idea grew out of conversations with INDOT, basically about getting better at inspection. I laid this out as an idea, and they liked it. We approached the Federal Highway Administration, and they also thought it was a good idea. We ended up putting together what is called a pooled fund study (TPF-5(281) is the number), and we now have five states involved. Several railroads have also donated components.
It really feels like we’re gaining momentum to try and turn this into something. My long-term hope is that we develop a certification program for a yet to be defined level of bridge inspector. Maybe that would be for a higher-level person that actually has to demonstrate that they can find faults.
The idea is that when someone goes through NHI training, which presents a very good foundation for bridge inspection, at the end of the day, we still don’t know if they can actually find anything. They just take the training and pass an exam, and they go out and do bridge inspection.
Nobody ever checks to see if they need reading glasses or if they use the right flashlight properly or if they’re even capable of finding a fatigue crack. It’s like qualified vs. capable. You might be qualified to do something, but you may not be capable of doing it.
I’d like to see, down the road, S-BRITE becomes a national center where to be say, a Level 1 steel-bridge inspector, you’d have to come to a facility like ours, and we can ensure a certain level of competency. We need to know that our inspectors can find fatigue cracks and that they properly measure corrosion. If not, we have to come up with a plan to help improve that person.
I2: Does this tie into ISO or other types of auditing standards?
Connor: I think it does in a way, as other industries do this. They do performance testing of people. We need to evaluate what we can find, regardless of whether they’re a Level 3 or Level 2 ASNT technician.
Glenn and I did this on the Sherman-Minton bridge, where we inspected the bridge to find cracks as well as internal weld defects. We had to make serious decisions about whether to keep the bridge open or close the bridge.
We actually had plates made up that had cracks and internal weld defects, where we knew where they were. We put them on the side of the bridge, and anyone who wanted to work on this bridge had to demonstrate some level of competency in terms of finding these defects. There were people that didn’t pass. These were people with ANST Level 2 certifications. While the majority of them could find surface-breaking cracks using magnetic particle testing, there was one that couldn’t. He didn’t pass, and we were happy to know that ahead of time. So, while the person was qualified to some degree, we literally don’t have data on what people can and cannot find.
The other aspect of this that I think is very important to understand is that when we send inspectors out, if there’s just a 5-percent probability they can find fatigue tracks greater than one inch or so, and there’s a 95-percent chance that I’ll miss it, you need to rethink what your inspection is getting you, and you need to rethink what your structures are able to tolerate.
We presume that everything is working fine, but as the infrastructure ages, our inspections become harder, as there are fewer resources and funds. Doing many frequent, bad inspections is not as good as doing less-frequent and more-detailed inspections. For example, say you go to the dentist, and they ask you how your teeth are, but they ask you every two weeks. Then finally one day you say that you have one that really hurts, and it fell out this morning.
In a way, that is a form of inspection, but if the dentist looked in your mouth and did an x-ray once in a while they would have likely found a cavity very early that could have been repaired well before it turned into this big problem. So rather than spending a lot of time and resources to do a lot of inspections that don’t have a lot of value, maybe we could do a really good inspection every five years that we can have high confidence in. Extending the inspection interval on the structures that are in good shape allows us to focus our efforts and more frequent inspections on those that need more attention.
I2: I wonder about your experience instrumenting bridges with sensors. Are you seeing that trend growing, and have you had any good successes?
Connor: We’ve done a lot of field testing around the country. In all honesty, I’m not a huge fan of this health-monitoring movement. There are a couple of reasons for it. Most of the people doing it disregard that we don’t have the money to put sensors on all the bridges, they don’t have the resources to maintain those systems or look at all the data. In all honesty, most of the structures don’t have problems.
There are people trying to sell monitoring systems as a silver bullet. But, if you build a new bridge, the sensors better be working 50 years from now, because that’s when the problems will be. Does the DOT have the time, resources, and knowledge to replace or fine tune that sensing system? The system better work, because we don’t have the time or funds to run out and fix it.
When they load up a bridge with hundreds of sensors, we don’t have the resources to maintain or even vet the system. I don’t believe that there is such a thing as a smart structure. There is an intelligently instrumented structure.
People say they’re doing health monitoring, and my first question is define health. Somebody will talk about putting accelerometers on the bridge, but what does that tell you about the health of the bridge? I can spend a lot of time monitoring things that don’t relate to the potential problem. So it seems that we should be looking at structures from a risk point of view and targeting what we think are going to be the problems. That’s a wild card too because a lot of times there is a problem that appears 40 years after the bridge was built that we didn’t realize was going to be the problem.
I also don’t think there is a good connection between a lot of researchers promoting structural health monitoring and DOTs. A lot of the work that I’ve done when we instrument bridges is targeted to the problem. We instrument for the problem, and we come up with a fix.
There are a lot of bridges and a lot of owners that have an interest in bridge health monitoring. As they’re taking out the bridge that lasted 75 years, I ask them to go back 75 years and ask them what data would you have today that you wished you would have collected over the last 75 years that would have extended the life of the structure? The key is what data would have actually extended the life or allowed more economical maintenance. Everybody I’ve ever talked to about this admits that they don’t know.
The idea of festooning a structure with sensors because you can is not the way to go. I don’t know what we’re going to learn, who will maintain the sensors and system, and who is going to vet the data.
The other thing to think about is forward and backward compatibility. If you went back 50 years, how would you have instrumented a bridge? It would be based on a completely different technology. In 2015, it might be a little overconfident of us to say that we have the latest and greatest, because by 2025 how we would monitor a bridge will almost certainly be wildly different
Although there are pilot studies to do and sensor studies to do, I’m not sure that it’s ready for prime time yet. The DOTs just don’t have the funds to have people looking at terabytes of data. Something else has to do with the response when an owner receives false positives or worse, false negatives, how do they deal with it?
I2: What are some of the biggest problems, and how do we address those?
Connor: Honestly, I find it amusing when the structural health monitoring community continues to talk about how steel bridges are susceptible to fatigue. Realistically, it is extremely unlikely any steel bridge designed since the mid-1980s will have any fatigue issues as we’ve engineered for that, and fatigue and fracture aren’t an issue. It isn’t the 1950s anymore.
The biggest problem with bridge deterioration generally has to do with the longevity of the concrete deck. It seems like no matter what we do with materials, we get 20 to 30 years out of a deck. If I’m monitoring the corrosion in the deck, maybe that will help me know when the deck will fail. But is it going to help me any more than normal inspections or will it be any more reliably than normal inspections? Remember that your sensor in the deck needs to be working in the deck on an active bridge for 15 to 20 years before there’s even a problem to measure. Again, it might work, but I don’t know if an owner wants 40 sensors in a bridge deck that they have to look at and maintain. Also, how will the owner know they are all still working and are calibrated.
The structures that I’ve monitored have had sensors on them for six to 24 months. Often, you’re running into problems with some of the sensors or lighting hits the structure and damages something. There are a lot of issues that need to be addressed.
It is also imperative that an owner has a plan regarding responding to changes in the response or the parameter that is being measure. For example, if one is monitoring acceleration, what are you going to do when the natural frequency changes by 0.1 percent? If you don’t have that question answered, you shouldn’t put a single sensor on a structure—until you know what you’re going to do with the information and how that impacts your management of the structure.
I2: What are some of your other thoughts about the entry of UAVs into the bridge-inspection market?
Connor: I think there are great opportunities. Hopefully, they can get a lot of the regulation issues straightened out because that’s going to hinder progress. The more that regulation can get straightened out for researchers as well as commercial practice, that’s going to move things forward. If there’s too much regulation that will hinder progress, although I understand the need to make sure it’s done safely.
There is a lot of opportunity, but my concern is that if we go too fast, and the technology isn’t fully vetted, then all transportation people will say that UAVs don’t work. They won’t single out the vendor that sold a bad solution; they just won’t trust anybody, because there’s no way to evaluate them objectively.
It’s like high-strength bolts. If someone gets into the market and sells a bolt that isn’t high strength, then people will be gun-shy of all high-strength bolts if there is no standard test by which to measure or compare different bolt producers’ products.
I2: One area we like to cover for all our “Future Forward” profiles is motivation. You talked a lot about the motivation for starting S-BRITE, but how about personally: what motivates you to push for better answers?
Connor: I’m very fortunate, I enjoy what I do, and we have very good students here at Purdue that keep me on my toes. I think the part I like the most, and that I learned from the people who trained me (John Fisher at Lehigh University was my advisor; the late Robert Dexter of the University of Minnesota, and Karl Frank at the University of Texas-Austin, and many others), is that it’s great to be able to see the work that you do influence the practice. In other words, it’s great to see the DOT engineers come in and interact and take what we’ve learned and move it into their daily practice.
I find that transition really satisfying. We’re engineers, not scientists, and we’re not in an ivory tower. Being able to get knowledge into students’ hands is important. S-BRITE is really exciting to me to see the students’ reactions to seeing the bridges up close.
Being able to use the work on multiple levels, I find challenging and rewarding. Being able to help when we get a phone call about a fracture in a bridge, taking the knowledge and helping someone with it, is something that keeps me interested and trying to move forward.
The other part that ties into that is the fact that most of our infrastructure in this country was built during the interstate building era. There was so much building going on. The way that we could manage that infrastructure when it was brand new was with routine inspections, but the type of inspection, when they were brand new, is entirely different as is now much older.
Our infrastructure and the inspection program that was put in place in the 1970s was developed and put in place for the younger and newer infrastructure. Now that it’s much older, how we manage that has to be done differently. I’m hopeful that the work that Glenn Washer and I are doing related to inspecting the older and newer structures will change how we inspect bridges so we prioritize what needs more attention. I think this allow us to be more effective and improve reliability and safety.
I2: In addition to the gallery, you also test components. How is that additional hands-on work accomplished?
Connor: The Bowen Laboratory here at Purdue University is an incredible structural testing and research facility. We’re doing full-scale tests of riveted built-up girders, and my team learned how to do hot riveting. You know, at Purdue we are known as the boilermakers, and so my team now knows how to do hot riveting. Although nobody does that anymore. We had Mr. Vern Mesler from Lansing County Community College in Michigan show us how to do it, and we’re making riveted built-up members and looking at how they behave if a component breaks. There is member-level redundancy with an I-shaped beam that is built up from many components. If any one component breaks, the beam doesn’t fail. Everyone knows that, but nobody has quantified it, so we’re doing some research to quantify it. I think that will have some major impacts on how we treat older, riveted structures.
In the past, when we had more resources for inspection or those older riveted structures were newer, we could treat all bridges the same. We can’t do that anymore. If you think you’re going to find a half-inch crack in a riveted member, that’s not going to happen. We have POD data now that shows that. Our full-scale tests indicate that rather than try and find a small crack at any one of 10,000 rivets, all you really need to find a completely cracked component broken, such as a cover plate. Almost anyone can find a completely crack component. The research shows how the beam will behave after a component has broken and how the strength is affected. Based on the data, we can now calculate how long you can expect that piece to stay in service before there’s really a problem. We’re also doing research on high-performance steels and how to exploit that. We’re fracturing big girders to show how these modern steels are superior and trying to push forward with new materials at the same time we’re working on legacy testing.
Basically, we are really trying to develop an integrated fracture control plan, linking the design of the member, with inspection capabilities and the performance of the member. I find this work to be very interesting and challenging. Some of it may even get implemented in my lifetime!