Friday, January 25, 2013

Detector systems and the modern railway

Written by  Ryan McWilliams, Contributing Editor
It's very impressive when one takes the time to consider all of the fault detection systems in the railway industry today.

To give an idea of how automated the modern railway has become, here is a list of two dozen systems (not a complete list and in no specific order) that have been in use for at least three years (most much longer) to improve the safety and the productivity of our industry.

 

 

  • Vehicle Component Monitoring Solutions.
  • Wheel Profile System.
  • Wheel Back to Back Spacing Monitor.
  • Wheel Impact Load Detector.
  • Acoustic Bearing Monitor.
  • Brake Shoe Vision System.
  • Cracked Axle Monitor.
  • Cracked Wheel Detector.
  • Stuck Handbrake Detector.
  • Hot/Cold Wheel Detector.
  • Truck Performance Detector.
  • Hunting Truck Detector.
  • Hot Bearing Detector.
  • Track Component Monitoring Solutions.
  • Tie Integrity Vision System.
  • Fastener Position Vision System.
  • Joint Bar Integrity Vision System.
  • Ballast Integrity Vision System.
  • Subgrade Monitoring (GPR).
  • Rail Profile Monitor.
  • Rail Flaw Detection.
  • Gauge Corner Crack Detection.
  • Coefficient of Friction Monitoring.
  • Track Geometry Systems.
  • Lubrication System Monitoring.
  • Ballast/Subgrade Movement Detector.

This is quite a catalog of hardware and a lot to handle if you happen to be a railroad on the leading edge.

Let's address the reason for such devices and the obvious advantages first. Recall how arduous and costly it is to have a team of specifically trained people crawling around a vehicle or track 365 days a year in all sorts of crazy weather conditions. However even the best trained person can make mistakes and there is also a natural variability between inspection personnel; people simply do not report the same results each time. So it is a fairly safe bet that there are missed defects when human eyes are the only detection systems in use. There are also missed defects from detection devices, often caused by various environmental factors or poor maintenance, but after decades of improvements, I would venture to say that we are much better off than previous generations. Where human error is fairly constant, railway detection systems have continued to improve. Today, we can actually measure our railway system and then monitor our changes, successes, and failures. The industry is now able to prioritize maintenance programs, by categorizing detector data, and assigning costs to various mitigation techniques.

A couple years ago this writer performed a cost analysis for Indian Railways to evaluate whether or not it would be cost effective for them to purchase a specific track component detector network. My initial thought was "not in a million years can we compete on a cost basis with the extremely low labor rates in India." Well, much to my surprise, that was absolutely wrong. This specific analysis showed that even with Indian labor factored in working 16 hours a day, the break-even point of implementing this detector network was less than two years. This also changed the detector network from a manual 16-hour inspection day to an automated 20-hour inspection day (considering an extreme four hours of undefined system down time). This simply demonstrates that there is often a significant financial benefit, in addition to the safety benefit, of a good detector network.

What is interesting now is the fact that to the best of this writer's knowledge, no firm or railway operation has performed a cost analysis to determine if it is worth putting boots on the ground to evaluate the current North American concrete tie issue. We simply know that there is a problem and we send people out to evaluate – end of story. Conversely, many railways want to perform a cost/benefit analysis for any detection system they buy that performs a similar task. The metrics have not been collected or analyzed. We all understand that the operations budget (manpower) is very different from the capital budget (detector hardware) but are we not working on the same railroad with the same problem and the same goal in mind? I bring up this point because I am very focused on value-added and of the opinion that while most cost/benefit analysis of this type may be required, due to organizational structure, we should be careful not to waste a lot of time and effort on this task. The fact remains that these problems must be evaluated and solved no matter what approach is taken. Regretfully the numbers that are used during such analyses are often marginal at best because without empirical data from an actual new system, all assumptions are based on speculation.

In the final analysis, detector accuracy, reliability, availability and user payback are often much higher than they are given credit for. These systems typically have life cycles of a decade or better and in the long run users really do get what they pay for. Automated systems look even better in the light of rising labor rates over the life of these systems. So it is crucial that railways ensure that their purchasing departments don't simply base their acquisition decision solely on cost. Value is the most important metric and proposition that should be considered. The solution is simple: promote good technology and use common sense. The world may lack these traits and values but from a railway engineering perspective, this writer believes that railways must have a strong foundation of both.

When a new technology hits the market, there is a serious need for four items; proponency because any new technology needs a champion, money to implement the technology, testing to prove and fine-tune the detector, and education to instruct all involved about the system details and challenges. It often takes many iterations to get a system to function at a reasonable technical level and we can't overlook the fact that the railroad environment is extraordinarily harsh and at times can severely impact reliability.

Along that line, this writer recalls a statement from a very wise mentor during the early days of my engineering career. John Elkins is possibly one of the best dynamicists the industry has ever seen. He said that the railway environment might just be the most complex and harsh mechanical system on this planet. History has proven his point valid by showing that 100% detector up-time (availability) is not an easy target to hit. Remember that "reliability is directly coupled to cost" and automated detection systems are directly impacted by the financial resources railways are willing to allocate for such devices. Also note that detector systems must be maintained and will require repair, upgrade and calibration; so railways must be prepared to allocate resources for this too. The education of all who are impacted by the technology is also a challenge as railways do not change course easily. The old adage, "because we have always done it this way before" is a definite issue, and old habits die hard. Coupled with adoption of automated technology is the question "Who benefits and who pays?" It is not always straightforward.

In the case of impact detection technology, while the systems assess rolling stock, the people who pushed the technology came from the maintenance-of-way community. Why? It is simple, bad rolling stock damages good track. But this is a time for bold action by managers and to embrace shared benefits. Too often it often takes an act of Congress (perhaps a poor example considering their failure to mandate a reasonable PTC program) to get our industry moving in a different and progressive direction.

Implementation and integration

This brings us to detector implementation and integration. There are a lot of things to consider when implementing a detector network of any kind. Issues such as communications, detector placement, data handling, information dissemination, data formats, data access, error correction, alarm monitoring, repair management, and rule changes, just to name a few.

Our biggest problems as we expand our automated measurement systems are that they generate a lot of data. Data that has to be moved, stored, and managed and that demand high speed communications and significant amounts of bandwidths. To be blunt, information overload is the greatest challenge we encounter when implementing detector networks. There is so much data and so many new concepts and reporting with a myriad of caveats that the users can easily become overwhelmed. So we need to add another layer of automation by turning all of this data into actionable information. This is where a basic understanding of the key issues becomes vital. Railroaders have numerous issues to address every day. The goal of a detector network is to prioritize those issues so that they can pick off the biggest problems first and then work their way down the list.

While today's network of detectors does attempt to prioritize problems, detector-specific issues have been the focus. Cross-detector integration is very much needed by our industry, but it is also a very complex and controversial topic. If, for example, three of 24 detectors report a vehicle problem, and a different three of the 24 detectors report that a different vehicle has a problem, how does the user know which vehicle repair warrants a higher priority? We have started to advance these concepts but we still have a lot of work in front of us. Other scenarios are where multiple detectors (e.g. truck hunting, flat wheel, and acoustic bearing) all report marginal problems, but have not yet reached an AAR or FRA limit enabling them to be pulled. The issues are that these cars, which may be only 1% or 2% of the entire North American fleet, pose a much higher risk.

To address issues such as cross-detector integration, the industry must dedicate resources to learn and understand the big picture as well as the minutia details of their systems. Two of our colleagues have used a term which is key to detector system success. That term is "first principles." Mark Dembosky (TTCI ret.) and Harold Harrison (Salient Systems ret.) both drilled this concept into this writer's head and will always have my gratitude. This meant that if a person or organization does not understand the core issue of a subject, then they do not understand the "first principles." As an example, one cannot learn algebra if he or she does not learn simple addition and subtraction first. We all need a solid foundation of subject knowledge but unfortunately railway measurement systems and detector technology are neither taught in college or trade schools, so one either gains experience on the job or they don't. As a result, a broad range of railway people simply don't get it because they have never been exposed to this field. The dangerous scenario is for those companies (and people) that think they understand the field due to the fact that they understand the general concept, but unfortunately do not have any real experience or knowledge.

The unfortunate reality here is that the system benefit has a significantly diminished return due to lack of understanding, sometimes for not dedicating resources to manage and leverage the data and sometimes for simply not maintaining the detector networks. I'm relatively convinced that nearly every railway on the planet would buy a laser-counter just because the device has "laser" in the title and the entire railway community seems infatuated with lasers. Purchase solutions, not lasers. Many organizations have spent good money on great detectors only to have them sit on the trackside and suck up resources because they did not put full effort into understanding and utilizing their tool. If you don't understand the technology, let someone who does manage it for you. The better vendors in the market can help by providing training programs, which is money well spent.

Once the detector systems are in place and sending data, the challenge then becomes information merging into the existing data management and maintenance systems. Everything becomes even more difficult when IT departments are involved. Every detector is sending different information, and that information has to get to the correct people usually through the black hole that is IT. Keep in mind that the goal is to prioritize the information being sent so as not to overwhelm the maintenance personnel. If the detectors do their job correctly, it will make the maintenance workers job that much easier.

History can often cloud our judgment and prevent us from moving on to a better way of doing business. Our railway industry is a difficult ship to steer with a lot of inertia behind it, and this can be detrimental when it comes to performance monitoring devices.

Let me give a specific example of how history is holding us back. The Wheel Impact Load Detector (WILD) has more than 28 years of history in North America. The previous approach for determining wheel defects was to measure the length of the "flat spot" on the wheel using a hand operated gauge. Inspectors would measure the length of the flat spot and report that length. The longer the defect, the worse the wheel was considered.

What about the depth of the defect? The depth was most often ignored, and who was to say if a seven-inch-long, halfd-inch-deep defect was better or worse than a nine-inch-long quarter-inch defect. So in 1984 the WILD detector was implemented. It measured the actual force that the wheel was imparting into the rail.

So, we have a solution to the real issue at hand—we now know if the impact is actually damaging anything. We now have a method for measuring the force of the impact so that we can pull out those wheels that cause the greatest damage, regardless of the length or depth of the flat. We have resolved a century old problem and now we can place detectors around the network and prioritize our repairs.

Today, an extensive network of detectors is in place on all Class I's, and we have helped the entire industry only to have this recent statement conveyed to me. Someone in our industry actually said, "We are using the WILD and trying to determine the length of the defect based on the force level we are reading . . . " Common sense should dictate that the force is what causes the damage. We want to prevent damage ("first principles"). We desperately need to educate the next generation of railroaders. Mentoring and education is a necessity aspect to detector integration and advancement of the technology.

For too long, our industry has been geared around reaction to maintenance issues. Prediction is where we ultimately want to be. If we can move to predictive maintenance programs, the benefits are tremendous in both financial and safety terms. This is obviously easier said than done. Progress is being made but, there are a lot of issues to address prior to putting our eggs in the prediction basket. I understand that this is the latest buzzword but would caution anyone who thinks we are there. It will take a lot of time, effort and collaboration to get to a true prediction based system and I can't wait to be a part of what happens next.

Railroads and suppliers must remember that they cannot provide all of the money, all of the talent, all of the testing, and all of the education for new technologies on their own. Partnerships must exist and good faith gestures must be present. There are some very smart people in various organizations that think they can move the entire industry all on their own. I cannot applaud them enough and wish we could give them each a billion dollar budget to do so. However, it would be much more realistic and productive to call on organizations around the globe to put forth these people to help advance our industry by collaborating with other organizations and colleagues. Working in partnership will advance the state-of-the-art and benefit all.

The statement has been made many times that we want to change our industry's "finders" to our industry's "fixers." By moving our resources from finding problems to fixing problems, we will improve our industry's position and improve the future for the next generation of railroaders. Let us focus on our core competencies, collaborate, promote good technologies, hire good leadership, and use our common sense to advance the modern railway.