5 coal cars (fast dump) were made available. The intent was to measure vibration and temperature effects on bearing and ride quality to predict when maintenance would be necessary. Maintenance in the owner's facility is much less expensive than a breakdown on the road. Each bearing had a vertical channel capable of 100 KHz @ +/- 80 g acceleration. A tachometer integral to a Timken bearing generator provided angular velocity information. Frequencies for cone, cup, cage and rollers were extracted from modulation around the 11 KHz audio range. From that data, over 55 distinct types of bearing defects can be extracted. I did the FFT and data analysis onboard the car, the uploaded to my server in Fairfax, VA via a Kyocera 1XRTT cellular modem to a mySQL database. I had an onboard Garmin GPS to provide time/position information with the data. Local to each train, a Wifi 802.11b radio running in adhoc mode was augmented with an OLSR mesh network allowing a 255 car train to route non-line of sight around a mountain to the monitoring console in the locomotive. My team members at WiTronix built the UI into their console and data traveled in a specialized openDDS publisher/subscriber to allow them to collect the vibration data, temperature data on each bearing (inner/outer) for hot bearing detection, brake piston position and GPS data.
Subsequent to that basic system, I created a network of PIC18F micro-controllers running a CANopen network to modularize the temperature and vibration sensors. Active actuators for handbrake, anglecock and cutlever were added. Sharma and Associates added auto-couplers, controllable anglecocks, handbrakes and cutlevers. WiTronix extended the interfaces to allow remote control of each car without need of human to couple/decouple and set handbrakes. The aim was to reduce injury to brakeman and switchman assigned to the train.
The project was going well. I added a special project with DHS to support tracking containers with full accountability as they traversed truck to train to storage yard.
The trains with all the new capability were ready to hit the road in December 2008. About 3 weeks into January 2009, I had to call all 46 team members in 7 partner companies an announce the program was over. What happened? Obama. We abandoned $1 million in research cars in a yard near Joliet, IL. Much of the project hardware sits in my basement. No money was allocated to return it. A very disappointing end to a fairly successful program.
I did the hardware, software and firmware design. My colleague did the analysis algorithms in Matlab that I implmented in C. Timken did the bearings. Wilcoxon Research built the specialized accelerometers as well as fabricating the hardware. Sharma and Associates in Countryside, IL did the specialized anglecock, cutlever, handbrake design/fabrication. WiTronix was a spin-off of GM Electromotive and became a very successful standalone business. BNSF provided a research caboose for us to use on the road. The 5 cars were made available by an interest coal car owner working out of Roanoke, VA at BNSF. The program was sponsored by the Federal Railroad Administration.
I miss the program. My colleague died from melanoma in January 2010. He did the business development. I did the design/implementation. Lots of hard work with excellent results. Sadly, thrown out by the Obama administration. They didn't like the coal industry.
Great stuff! I can see how identifying those bearing failures early could pay huge dividends. Sorry it got cut short. Was any of it ever commercialized? Sounds like a no-brainer with high payback.
I’m learning Arduino right now as a hobby for several home automation projects. It is amazing the power you can buy today in a microcontroller with WiFi for $10.