Craig Hill

Track maintenance strategy on Burlington Northern Santa Fe

Long Version

Burlington Northern Santa Fe (BNSF) operates one of the largest rail networks in North America, with 33,000 route miles (53,000 route kilometers) of track covering 28 states and two Canadian provinces. BNSF is an industry leader in Web-enabling a wide variety of customer transactions. The railway moves more intermodal traffic than any other rail system in the world, is America's largest grain-hauling railroad, and hauls enough coal to generate more than 10 percent of the electricity produced in the United States.

BNSF's traffic mix is diverse, reflecting all sectors of the world economy. BNSF derives a substantial portion of its revenues from carload products (including chemicals, forest and building products, metals and minerals, and machinery), intermodal containers and trailers, and coal, as well as agricultural products, and automobiles and automobile parts. BNSF does not provide direct passenger service, although it does handle both intercity and commuter passenger trains under contract to federal, state and local government agencies.

BNSF's vision is to realize its tremendous potential by providing transportation services that consistently meet our customers' expectations. We will know we have succeeded when our customers find it easy to do business with us, receive 100 percent on-time, damage-free service, accurate and timely information regarding their shipments, and the best value for their transportation dollar. To provide service that meets customer expectations, we have invested heavily in our infrastructure in the years since BNSF was created by merger in 1995.

One of BNSF's major challenges is to tailor service to meet the needs of a wide variety of customers. Intermodal customers, for example, demand not only highly consistent, but also truck-competitive transit times. Coal customers demand highly efficient movement of large volumes on consistent schedules.

Equipment availability is important not only to our agricultural products customers but also to customers that move all types of products and commodities in carload lots. The two expectations that all our customers have are safety and reliability. An infrastructure capable of supporting safe, reliable operations day in and day out is essential to BNSF's realizing its customer-focused vision.

The range of physical characteristics and geographic locations comprising BNSF's infrastructure is just as diverse as the railway's traffic mix. For example, of 6410 main track curves, the degree of curvature of 459 of them is five degrees (r=350 meters) or sharper. Nearly 1000 route miles (1600 route km) of main track rests on grade that is one percent or steeper, with some over three percent. Nearly 32,000 streets and highways cross the track at grade. Fourteen thousand bridges span more than 340 miles (1640 km), and 35 of them have movable spans. There are 87 tunnels covering 35 miles (56 km), the longest being 7.8 miles (12.5 km) in length.

As a result of these challenges, coupled with some very high traffic levels, BNSF's 2002 maintenance program is expected to approach $800M (USD). Although financial imperatives have led to reduced capital spending in recent years, the amount of capital devoted to maintenance of our infrastructure has remained steady, reflecting our commitment to safe, reliable infrastructure.

World Class Maintenance
To insure that resources of this magnitude are distributed appropriately, BNSF utilizes innovative tools to analyze maintenance and engineering processes to reduce cost, to improve reliability, and to extend asset life through our World Class Maintenance (WCM) System. The goal of the WCM organization is to maximize the availability / reliability of BNSF's physical assets at the least possible cost, highest quality, and shortest possible cycle time, while providing the safest possible work environment for the railroad's employees, customers, and neighboring communities. A Work Order System, Lean Processes, and Six Sigma Techniques are a few of the tools employed to achieve this goal.

World Class Maintenance utilizes comprehensive planning and scheduling processes to drive improvement from a 30% proactive vs 70% reactive environment towards a goal of an 80% proactive regime. Ultimately, WCM supports a scheduled railroad, eliminates waste and idle time, and improves manpower and asset utilization, work process flow, best practices, material handling, and physical plant reliability.

The planning and scheduling of track, structures and signal maintenance activities utilize a work order system. Field maintenance personnel are linked to a central work order database by laptop computer. This central database compiles and stores work orders, provides critical data for planning, tracks labor and material usage, monitors repair history, and determines best practices for field use and analysis. Supervisors of the various work groups input and prioritize the work orders.

Another tool employed, The Lean Process, focuses on waste elimination and value creation through the use of lean analysis tools. A cross-functional team is assembled to analyze the business process in a workshop format. Initially, this team will determine the baseline for current total product cycle-time, throughput, utilization, productivity, safety, and quality. The team will then identify the existing process bottlenecks, and implement corrective action. A leadership driven, follow-up process ensures lean implementation and process improvements are communicated across the organization. Repetitive analysis using the Lean Process can further refine a given process.
Finally, Six Sigma problem-solving techniques are used when the problem to be resolved is variation, and the source of variation is unknown. The techniques are divided into a five-step process: D-M-A-I-C, which stands for Define, Measure, Analyze, Improve, and Control. Six Sigma techniques work with both attribute and variable data. Additionally, a Six-Sigma project consolidates the possible causes of variation into families, and then performs statistical experiments to eliminate entire families of variation. The BNSF has used Six Sigma methods to successfully address issues such as rail defects, joint bar defects, and rail lubrication.

The 2002 rail program calls for relaying 630 track miles (1010 track km) of rail, more than three-quarters of which is new. Ten percent of the new rail will be placed in track in conjunction with the installation of concrete ties.

The total depreciated value of BNSF's rail in track is $2 billion (USD), making it the railroad's highest-value physical asset. Rail is the point of the track structure pyramid and is subjected to extreme contact stresses. Dependable rail performance is key to the reliability of the physical plant. The strategy associated with this vital and expensive track component is to extend rail life as long as possible without jeopardizing the safety and reliability of rail operation.

Top initial product quality is essential for long rail life. BNSF pushes suppliers for continuous quality assurance and improvement, and collaborates globally with rail producers on new rail metallurgies and products. BNSF's specification for standard rail calls for minimum hardness of 300 Brinell. The premium rail spec requires minimum surface hardness of 370 Brinell, and minimum hardness of 350 Brinell at depth of 15 mm. Currently, 141 pound/yard (70 kg/m), head-hardened rail is used on all curves in tracks carrying more than 45 million metric tons (mmt) per year and on curves two degrees (r=870 m) and sharper on tracks with less than 45 mmt/year.

Rail grinding is necessary to remove - and even better, avoid - rolling contact fatigue conditions such as shelling, spalling, and head checking that shorten rail life. BNSF uses a preventive grinding strategy that shapes the rail head to control contact stresses and retard defect formation. Curves > 2.5 degrees (r=700 m) are currently ground every 14 mmt, curves < 2.5 degrees every 28 mmt, and tangent track every 56 mmt. With this frequency of grind, usually only a single pass at 10-16 kph is required. Minimizing rail grind off necessary to avoid defects and surface conditions optimizes rail life.

The third piece of BNSF's rail life extension strategy is rail lubrication, a difficult process for most railroads, including BNSF, to manage and measure. In 2001, six sigma techniques were applied to identify rail lubrication improvement opportunities. The "red Xs" identified were type of lubricant and output of lubricator. One brand of lubricant lacked the tackiness necessary to be effective in trackside lubricators, and that lubricant was removed from the approved list while the supplier works to improve the product. Studies also determined that required output for effective lubrication was site specific, and a methodology was developed to establish a site factor for each lubricator location and measure lubricator output.

Effective rail flaw detection enables BNSF to get full life from rail by finding internal defects before they break and jeopardize the safety and reliability of rail operations. A rail test scheduling system is in place that analyzes defect statistics and recommends test frequencies for each subdivision to keep internal defects below an established risk factor for the subdivision. Defect rates drive the model, so factors that influence defect rates, such as rail conditions, wheel loads, annual tonnage, and accumulated tonnage, automatically are built into the model. The system is based on Weibull statistical analysis that anticipates rail degradation. As defect rates change over time, the system automatically adjusts rail test frequencies to keep service failures below risk factors.

These rail life extension strategies are producing favorable trends. Rail life on curves is much shorter than on tangent, so longer rail life evidences on curves first. As the rail relay graph shows, from 1997-2002 the length of new rail necessary to replace worn-out rail on curves decreased by 405 kilometers, a 48% reduction.

The rail defect graph shows a reduction in rail defects on curves from 1997-2001 of 992, or 22%.

Since 1997, when BNSF first began to develop and phase in these strategies, both consumption of rail on curves and defects on curves have been reduced.

As previously alluded to, BNSF operates across a full range of climate differences from the ultra-dry Mojave Desert of California to the humid environment of Houston, Texas. The varying climate and terrain, as well as different traffic densities, necessitate a variety of tie choices across a rail system of the BNSF's magnitude.

BNSF installs approximately 2,000,000 creosote-treated, hardwood ties annually. These hardwood ties constitute the vast majority of the tie population, however other engineered ties are installed as well. More than one-fourth of the 2002 wood ties will be installed by a gang working in two shifts, 24 hours per day, four days per week.

BNSF has been installing concrete ties, primarily on high-tonnage track segments, since 1986. As part of the 2002 work program, 216,000 concrete ties are scheduled for installation. Accumulative through the 2002 work season, BNSF will have approximately 8,000,000 concrete ties in track on 3,000 miles (4,800 km) of track. This however is only 7.4% of the total BNSF trackage of 40,800 miles (65,700 km).

To supplement its conventional wood tie population, BNSF has tested and will continue to test a number of engineered ties including steel, plastic, concrete, and composite wood. However, creosote-treated, hardwood ties remain the backbone of the tie maintenance program.

One of the major limitations in tie maintenance management is the lack of adequate data to define tie condition. In order to overcome this deficiency, BNSF has introduced the use of Tielnspect i\e condition data acquisition systems in order to obtain a tie by tie and mile by mile condition report for tie conditions. Using the Tielnspect units, tie inspectors are able to "map" the tie condition of thousands of miles of track each year. BNSF mapped over 6000 miles (9650 km) of track in 2001 utilizing 23 Tielnspect units. This map identifies 100% of all ties within an inspected mile as being "bad", "marginal", or "good".

Using this tie condition data, together with specially developed offline analysis software, BNSF is able to identify specific ties to be replaced based on cluster size, number of adjacent good and/or marginal ties, proximity to a switch, bridge or crossing and other similar parameters based on BNSF system standards. Logic settings vary for class of track and degree of curvature. The identified replacement ties are then uploaded back into the Tielnspect unit for field identification and marking for programmed replacement.

Good drainage and clean ballast conditions are major components in maximizing tie life. To attain these qualities, BNSF accomplishes a significant level of both track undercutting and shoulder ballast cleaning in conjunction with an aggressive track surfacing program.

In 2002, a total of 300 miles (480 km) of track is scheduled to be undercut. In addition, over 600 miles (1,000 km) of track will have the ballast shoulders cleaned. Approximately 12,000 miles (19,300 km) of track are scheduled in the surfacing program. Twenty percent of this work will be accomplished by large production surfacing sets and the remainder by local surfacing sets. Four million tons (3.6 mmt) of ballast is shipped annually to system-wide projects. A fleet of 2,000 ballast cars is used to transport the products.

BNSF and Union Pacific (UPRR) standardized on ballast specifications in 2001. These specifications were not actually all that different before they were standardized. With the standardized specifications, it is hoped ballast may be exchanged to reduce our combined transportation expenses.

BNSF is currently negotiating with potential ballast suppliers. Existing as well as new potential suppliers are submitting proposals that will result in long term contracts with BNSF. Fifty to sixty percent of the contract's volume will be awarded in mid-year of 2003, and the balance will commence in 2005. It is hoped that BNSF will be able to reduce ballast expenses by 20%.

To augment its 2,000 car fleet, in 2001 BNSF acquired the services of a Herzog high-speed ballast train consisting of 54 cars. The high speed ballast train unloading is a concept that BNSF brought to Herzog in 2000. Herzog developed the concept, and is able to unload ballast by mapping and programming accomplished through GPS (global positioning satellites). With the program determined, Herzog is able, from any locomotive with a portable computer, to control all the door opening and closing. With that control and plan, the actual volume is placed where required any time of day. The only restriction naturally, is freezing weather conditions. The BNSF / Herzog high speed ballast train has an average cycle of 2.7 days between loads. The advantage of continuous unloading at 5-10 miles per hour (8-16 kph) and 24 hours per day unloading windows has given BNSF the ability to reduce the ballast car fleet by 150 cars. BNSF and Herzog are considering adding an additional high speed consist in 2002, and further reducing the fleet.

In 1999, BNSF began reconditioning 70 and 100 ton (64 and 91 mt) air dump cars by installing hydraulic cylinders that are powered through air pressure. The upgraded cars are providing more reliable service for this important emergency fleet.

Emphasis in signal technology is placed on automation that reduces switching labor and improves operational reliability through the elimination of train delay. BNSF's standard interlocking installation is a microprocessor with a radio communication link to microwave or fiber optic base stations. This combination has improved reliability to the extent that up time now stands at 99.98%. Electronic coded track is primarily used for track occupancy detection and as the communication medium between wayside signals. Recently, fiber optics installations have been added to communicate aspects between signals on a very busy commuter route west of Chicago, Illinois.

Operation of BNSF hump yards, or marshaling yards as they are referred to in Europe, involves traditional equipment for the most part. The use of spread spectrum radios to communicate from the control house to the switches eliminates the need to install and maintain hardwire cable.

The railroad is currently developing diagnostic devices that will perform and record various tests at interlockings, highway grade crossings, and wayside detectors to reduce maintainer workload. A round-the-clock help desk will be placed in service later this year. The help desk will be manned by "signal experts" that will be linked to the field maintainer via the BNSF intranet. The maintainer has the ability to call up the troubleshooting guide on each piece of equipment. If the maintainer is unable to solve a problem, he or she will contact the help desk manager who can look at the same guide and talk the maintainer through the problem. Simply knowing where to go for help will greatly reduce the amount of time required to correct an equipment problem. This same medium will be utilized for individual computer-based training, eliminating the need to send field personnel into a classroom for every session.

BNSF wayside detector strategy is to provide a hot bearing and dragging equipment detector every 25 miles (40 km) on main line subdivisions. In addition, stand-alone dragging equipment detectors are located every five miles (8 km) in concrete tie locations to minimize tie damage should a wheel get off the track.

Signal-Related Statistics
Dispatcher Control Points 2,729
Automatic Switch Locations 55
Power Switches 4,768
Major Hump Yards 9
Hot Bearing Detectors 1,032
Dragging Equipment Detectors 1,586
With the widespread use of new measurement technology for determining lateral and vertical forces exerted on the track structure by rolling stock, railroads now can work objectively to reduce the highest stresses. Alarms identify those cars creating the greatest forces. Inspection processes create factual information as to equipment condition. Analysis of the information allows for equipment maintenance practices to reduce the created stresses. Reduction of the stress state in the track structure will reduce track maintenance requirements. The design of the track structure may be affected as well by reducing the stress level. The attached diagram shows the distribution of forces exerted on the track and the track strength distribution or its ability to withstand the stress without failure. The area of overlap between the curves is where the track is subject to failure due to excessive stress. This has significant implications for both track design and maintenance.

One of the most useful new measurement systems is the Truck Performance Detector. This is a series of strain gages applied to the rail at a location where there are reverse curves. Level track is better than severe grades. Curves of four degrees are appropriate. The rail is strain-gaged at multiple locations in the curves and the tangent section between curves. The strain gages allow measurement of both the lateral and vertical forces. Alarms can be generated at various levels of the lateral forces; the ratio of lateral and vertical forces LA/; or the output of an algorithm which calculates truck side information, proximity of other trucks, and patterns of forces of leading and trailing axles. The information on the forces generated by the trucks is a good measurement of the curving performance and an indicator of the propensity of the equipment to cause or contribute to wheel climb, wide gage, or rail rollover derailments.

A second new measurement system is the Wheel Impact Load Detector which is a set of strain gages to determine dynamic wheel loads on the rail. The systems alarms can be used for such things as immediate setout, routing to an inspection facility when empty, or a flag for whenever the equipment arrives at a repair facility. Railroads believe that the data should be used to trigger wheel removal based on a loading measurement and any visible defect, rather than the current set of defect criteria based on out-of-round or size of defects. This technology, when it leads to greater removal of wheels imparting high loads, will result in longer life for that rail which does not wear out first.

To more accurately forecast maintenance and apply our maintenance resources more effectively, we need to develop better objective measurements of the track strength. This is particularly necessary in the areas of avoiding wide gage due to translation or rotation of the rail on wood ties. Current gage restraint measurement systems operate at relatively slow speeds and have been most utilized in yards. Today, this is our most objective measurement of tie performance and it remains too slow for widespread use of our main lines. Other measurements indicative of tie condition such as wood density may have considerable utility if the information can be demonstrated to be predictive of performance as well as time to failure.

Track Predictive Indices
One current challenge with track maintenance is being able to effectively determine when and where it will be needed. Past efforts have used measurement tools that evaluate the current state of the track. The Track Predictive Indices, or TPI, use successive historical data sets to accurately predict when and where maintenance is required.

TPI includes predictions for track geometry parameters, such as gage, cross level and rail wear, as well as predictions for rail defects. User friendly windows can list given track segments by predicted maintenance date, or can allow the user to view past parameter history for a given segment of track.

For the indices to accurately predict the future state of the track, two obstacles must be overcome: data synchronization from test to test, and appropriate application of statistical methodology. To address the data synchronization problem, TPI synchronizes the data to within 1.5 meters using Glo-cal Signal Matching and Reference Data Reconstruction techniques. Track geometry parameters are analyzed using a second order polynomial in a rules-based environment. Weibull methodology, along with defect rate smoothing for homogeneous track segments, provides a prediction of rail defects.

The TPI will allow BNSF RR to maintain its track with surgical precision, resulting in a well-maintained track structure using the least amount of assets.

BNSF is convinced that a track maintenance program founded on World Class Maintenance will support a scheduled railroad, reduce reactive maintenance activities, and maximize efficiencies. Accordingly, Burlington Northern Santa Fe has employed several innovative techniques including Six Sigma, Lean Analysis, and Predictive Indices to provide a World Class Maintenance organization. Today, World Class Maintenance techniques and processes provide the foundation for a safe and efficient infrastructure to meet the expectations of our customers. For tomorrow, continuous improvement and the development of new analytical techniques will be the cornerstone for Burlington Northern Santa Fe's World Class Maintenance Organization.

(click to enlarge)