Performance monitoring devices provide railroads a targeted, educated approach to maintenance or safety issues.
2015 has been a very busy year for the rail industry so far. We’ve had tank car regulation “discussions”, PTC implementation “challenges” and tremendous activity in infrastructure and vehicle regulation, construction and maintenance. All of these topics have the intention of making the industry safer and/or more efficient even though the approach and implementation may not be very desirable according to many industry experts. A common catchphrase these days is “show me the data” to prove a specific point or condition. In the past, such data was not readily available or did not exist at all. However, the industry has been doing a great job of implementing more wayside monitoring systems along with advancing the forward-looking subject knowledge on various performance-related topics when it comes to both vehicle and infrastructure monitoring.
Most people agree that there usually is not a single cause of any one accident, incident or performance related situation. It is almost without exception that there are a combination of conditions (infrastructure, operations and/or equipment related) that have coalesced to create a less-than-positive outcome. Performance monitoring devices are now allowing the railroads to take a more targeted and more educated approach when it comes to prioritizing maintenance and addressing performance or safety related issues. As an industry, we are in the process of implementing tools that allow us to better identify undesirable conditions and then focus our repair and maintenance efforts in that direction. As long as the regulators, the politicians, and the public allow us to stay the course, many industry insiders see a very bright and very positive future in this arena. However, if this goes the way of tank car or PTC regulation, this progress will likely be side-tracked in a less than optimal direction.
Suppliers and railroads are currently pulling together in the right direction when it comes to new technologies that help to ensure a safer and more proficient rail system year after year. As an example, wayside monitoring systems such as thermal hot box detectors (one of the most mature detectors in operation) are a great asset and are complemented by more advanced systems such as the acoustic based technologies. The acoustic systems do a very good job in predicting failures instead of just informing that a bearing failure is in process. The obvious path forward is to continue to complement and support the current detector network with more advanced technologies to ensure total coverage of key assets.
The data from a Truck Performance Detector in Figure 1 shows the load environment a decade ago and compares it to that same locations load environment today.
Although this is a complex scenario to evaluate and there are a lot of variables to consider (speed, geometry, load, maintenance, environment, etc.), this comparison shows that we are improving our stress state in this location, especially in the presence of heavier loads and increased train speeds. The lateral load environment from this heavy-haul curve shows a slight reduction in lateral load observed over a 90-day period in 2015 as compared to the same period in 2005. This is the result we are striving for by improving grinding programs, advancing friction management, and improving freight car maintenance and components. The advancement of many different technologies has been proven to reduce system stress, which reduces wear and fuel consumption and promotes a safer and more efficient railway. There are interesting artifacts hidden in this distribution curve that point to the fact that we are decreasing our overall stress state in many different ways. This is highlighted by the reduction of occurrences observed on the positive tail of the curve as well as from the shift toward lower values at the peak of the distribution. This is a relatively straightforward improvement in maintenance allocation whereby we are targeting the worst performers first and correcting those outliers so that they do not skew the fleet’s performance. Additionally, we are also very likely improving our wheel profile as is identified by the reduction in negative lateral values. Those values are simply denoting the direction in which the lateral load is applied. Positive values are pushing toward the field side of the rail while negative values are pulling in toward the gage side. By reducing the amplitude of the negative values, we assume that the hollow wheel population of the fleet is better managed and we are therefore seeing a reduction in negative (false flanging) wheels. Today, with more integral and more proficient wayside detectors, we have even more accurate tools to determine if our efforts are of value and to what degree.
Machine vision systems are the fastest-developing detectors within the industry today. These devices are now coming down in price and increasing in performance which makes the adoption of this newer technology that much more palatable. Not only are these systems being utilized to monitor for safety and performance improvements simultaneously, today’s machine vision systems can also perform inspections at a much faster rate than is possible with current manual methods.
CSX Director of Advanced Engineering Kim Bowling states that CSX began installing machine vision systems in 2009 and currently has three hump-based systems, eight line-of road systems (with three more on the way), three wheel profile detectors on the hump, and five wheel profile detectors on line of road (with one on the way).
CSX also has a progressive plan for additional detector integration that includes implementing the best possible technology to improve early detection and safety along with the selection of various systems that survive in the harsh railroad operating environment. CSX relies on the system software to do the “heavy lifting” of selecting and inspecting each image and then allows the in-house rules engine to assign alarms and route cars to the shop.
Bowling further states that “the future of the industry’s machine vision program could include the inspection of open-top load securement conditions (pipes, logs, etc.), the integration of Optical Character Recognition to identify trucks and castings, forward looking infrared cameras, inward facing cameras with gesture based interfaces, and whole-car imaging to support damage-related claims.”
Suppliers are not resting on their laurels these days. KLD Labs has been accelerating the development of train scanning machine vision systems by testing in revenue service as well as at TTCI.
According to Dan Magnus, KLD Labs Vice President and company co-founder, “TTCI’s outstanding facilities, equipment and staff have enabled KLD’s Engineering group to enhance and validate their machine vision algorithms and solutions for what our customers really want and need. We are extracting more actionable information from the captured imagery than ever before and we are continuing to find innovative new ways to use that data.” Efforts to date include automatically scanning the entire train for defects such as broken or missing components. They have supported further development of machine vision algorithms to locate and detect asymmetrical conditions that reveal subtle or non-detectable defects such as bolster issues, interior broken springs or imbalanced loads. “What we are now finding is truly impressive and railroaders are getting more involved trying to find new ways to apply this technology. Machine vision has entered a new era in railways and it is truly an exhilarating time for our industry,” said Magnus.
Even with all of the recent advancements in machine vision systems, there are still a few challenges being aggressively addressed. With any new technology, there are issues that need to be overcome. For example, since car construction isn’t standard, camera placement limitations can hinder the ability to see every component from every angle on every car type. Finding small defects such as a missing cotter key can be a challenging task for machine vision systems just as it is for experienced human inspectors. However, machine vision systems are generally better at these type of tedious and repetitive tasks, but they are still not perfect and will be subject to very similar natural obstacles that will improve with time and experience. Since machine vision systems incorporate relatively new technology in the rail space, and since every algorithm is not perfect right out of the box, human intervention and oversight is still a smart notion when it comes to quality control and accuracy assurance in these early days.
Massive amounts of data are now being generated by all types of detectors and there are questions about how best to manage it. The industry is currently addressing the questions of data storage location, size limitations and the amount of time to archive historical information. Industry standards are being discussed by all stakeholders and interchange data management is also a serious topic of debate.
As we deploy more and more machine vision systems, we also need to handle the large amount of data and develop “Big Data” analytical tools. Currently, traditional defect detection data storage is manageable, but there is a growing interest in predictive maintenance and comprehensive fleet assessment. Additionally, more-advanced detectors are creating larger data sets with imaging systems coming in at the top of the charts. Dan Magnus states that “KLD Labs has responded to the big data issue by developing TrainBase, a web based data management application tool. This application generates user-defined equipment condition reports and can predict when maintenance is to occur based on the railroad’s specific priorities. Implementation of this system provides maintenance personnel the information needed so that they can better manage and plan maintenance activities. TrainBase also incorporates alerting functions that create messages to be sent to smart devices so key managers are quickly informed about more serious conditions tha need to be addressed immediately.The management of “Big Data” goes even deeper by providing comprehensive fleet statistics at a touch of the mouse and allows users to drill down from measurement values to component image data.”
Investing in tools to improve system capabilities, safety and performance is something that almost every railroad is undertaking. Many are in pursuit of zero accidents, which is an admirably ambitious goal. Supply companies and railroads are working together more now than ever before to advance wayside monitoring technologies and many other supporting technologies that benefit all stakeholders. If we can keep this momentum in the right direction, there is a very positive outcome for suppliers and railroads which also gets passed along to the public and to the shippers in the form of a safer and more cost effective transportation system.