Field Thermite Rail Welding Productivity Studies

In-situ thermit welding of continuously welded railWe performed an analysis of work assignments, processes, and equipment utilized by welding gangs in different districts on a major suburban railway to identify opportunities for productivity improvements using a six-sigma type methodology.  The prevailing wisdom at the time was that certain welding foreman were more “productive” than others, which led to dramatically different weld count outputs per period.  Through field observation and data analysis, we discovered that in fact there were many factors contributing to the differences in welding outputs: (1) certain gangs were assigned unreliable hirail equipment; (2) due to nature of their assignments, some gangs found it easier and quicker to obtain necessary track time; (3) gang productivity were much higher when multiple rail joints needed to be welded within the same signal block; (4) no formal process existed for providing replacement welding foreman or qualified welder in case of vacation or absence; (5) differences in weather between geographic regions contributed to number of days when welding gang can safely work.  The minor differences in methods, equipment type, and worker habits did not materially contribute to the differences in observed rates of output*.  Based on this study, we recommended that a systemwide “extra board” of welders and foremen be established to provide cover for necessary absences (which could be utilized on spare equipment as an additional gang when there were no absences), that assignments should prioritize those temporary joints within the same block and on the same track, that new hirail equipment be procured for three gangs with least reliable equipment (keeping one best set for spare), and a project be initiated for testing field electric-arc welding equipment in place of chemical “pot” welding.  The new hirail equipment improved welding productivity significantly as soon as they were commissioned.

* Note: Quality of resulting welds was outside of the scope of this study.

Track Density (Tonnage) Estimation for Commuter Railroad

Freight train in commuter rail territoryCommuter railroads do not typically consider their output in terms of million gross tons (MGTs) carried, and typically have negligible freight activity.  However, tonnage estimations can be important for assessing turnout and curve rail replacement, and resurfacing needs.  On behalf of an infrastructure owner of mixed-traffic trackage in a commuter district, we estimated the track density on a track and block level.  Starting with electronic train movement records, we designed a heuristic to automatically filter out questionable data, and provide reasonable estimates for use in their place.  This was then combined with planned train-level consist data and seated-load gross equipment weights to determine MGTs attributable to passenger trains and EMU equipment.  For road freight trains, we used one year’s worth of freight train manifests to determine tonnage carried, marrying this with train movement data to determine their routing.  For unit trains (that load or unload within the territory), and local trains (that pick-up and set-out loads within the territory—negligible volumes), we tracked the tonnage changes en-route where this was deemed a significant factor in the total MGT estimate.  On several mainline track segments, freight MGT was found to be between 30%~35% of total MGT and a significant contributor to track wear, despite this being a very busy commuter district that only saw 4~7 road freight trains per day (compared to upwards of 800 daily commuter train-starts).  This finding triggered additional management interest in freight activity.

Freight Tenant WILD and Manifest Data Processing

For a passenger railway that hosted a number of freight track-rights tenant operators, we served as the business representative on a multilateral project to install wayside Wheel Impact Load Detectors (WILD) and utilize that data for operational purposes (alerts, audits, billing, etc.)  The project involved providing the real-time information to the operations control centre (OCC) such that decisions can be made about whether the freight train is permitted to enter the passenger rail territory, and also matching real-time WILD data to a nightly freight manifest export for audit and billing purposes.  We were able to design a fuzzy algorithm that matches the WILD data (containing only car IDs, and only some of the time) to the Manifest data (having train IDs, but not always accurate, and often contains ‘complications’ such as cars being dropped off or picked-up en-route, on both scheduled and unscheduled bases).  This project improved the accuracy of the Manifest data being transmitted by tenant railroads, and reduced the instances of overweight cars, which indirectly contributed to improved host-tenant relationship.