Safer or more dangerous? Union v management.
Depends on how sophisticated it is. For example, how good is it at detecting raised spikes on the gage side which is an indication of a possible incipient rail rollover condition? Can it detect where a tie plate is sliding when the gage widens under load (only)? Can it detect a rotten tie under the tie plate? Can it tell what a varying longitudinal wear (or grease) pattern on the top of a rail means?
What would be interesting is to have both the machine and a human inspect the same track just a few minutes apart (no train between them) and compare the results. Didn’t see that addressed in the article.
Long-term, I think the track inspector’s jobs are in jeopardy. May still be a need for some to confirm the machine’s results, or look at anomalies. OTOH, might be hard to fill those positions anyway.
- PDN.
Shades of John Henry [(-D]
Can an inspector in a hirail vehicle detect them while rolling along at 5 to 10 mph detect these items? A walking inspection might but where do they walk track anymore?
You both raise good questions. I hope this pilot program is actually seeking answers rather than providing a rationale for more labor cost reduction.
You would be surprised… Main track walking is still a regular thing, especially where the TS’s territory was purposely shrunk because of the dispatcher situation. Automated anything has trouble with switch measurements, esp guard check and face gauges along with fit at the points (to the stock rail… esp when the “art” of determining if the point is condemnable… plus all the OTM issues in the switch and frog areas with braces, A&B Blocks, hook plates and the like. )
I’m not sure Batory gets the concept of the difference in maintenance inspections, as in how different railroads approach the issue (and who is qualified to do things to what level (least common denominator thing)- The BN foreman could not hold a candle to his ATSF counterpart, same with the D&RGW guy being way ahead of his SP and UP counterparts. Testing and training go only so far.). I do not see regionals and shortlines being able to afford this in the near term and for a good while out into the future.
Big buggaboo out there remains documentation of defects found and the&
It is the same argument that surrounds automatic trains. Unions say it will reduce safety because automation can fail to see something. Management says that it will increase safety because it will eliminate human error.
If there is a clear argument that the automated inspections will decrease safety, that point is not explained in the article in the first post.
I would suspect that such a machine would be made to find the most common problems - especially those which are known to be the most disruptive to the railroad.
It’s possible that some things might actually be better (and more consistently) measured this way.
The article doesn’t say how the automated inspections will be done. Perhaps it will be via some version of the automated test cars already in circulation on some railroads (can’t remember which). If that’s the case, the the sheer weight of the car may find defects a human in a hi-rail may not detect, or may overlook.
For ties, if they haven’t already developed something, perhaps some sort of ultrasonic technique could be used. That’s over my head, technologically. Don’t expect my name on the patent.
Guage, pulled spikes, and other such items that have expected parameters are easy. Switch gaps and the like, not so much.
Such instrumented cars could be added to the consists of numerous trains. Perhaps one cold be used as a buffer in hazmat trains, instead of a covered hopper full of gravel. It still wouldn’t be a revenue car, but at least it would be productive.
If other railroads adopted a similar technology, or perhaps simply contracted to receive the information, these cars could be used on run-through trains, saving switching at hand-over points.
[quote user=“tree68”]
I would suspect that such a machine would be made to find the most common problems - especially those which are known to be the most disruptive to the railroad.
It’s possible that some things might actually be better (and more consistently) measured this way.
The article doesn’t say how the automated inspections will be done. Perhaps it will be via some version of the automated test cars already in circulation on some railroads (can’t remember which). If that’s the case, the the sheer weight of the car may find defects a human in a hi-rail may not detect, or may overlook.
For ties, if they haven’t already developed something, perhaps some sort of ultrasonic technique could be used. That’s over my head, technologically. Don’t expect my name on the patent.
Guage, pulled spikes, and other such items that have expected parameters are easy. Switch gaps and the like, not so much.
Such instrumented cars could be added to the consists of numerous trains. Perhaps one cold be used as a buffer in hazmat trains, instead of a covered hopper full of gravel. It still wouldn’t be a revenue car, but at least it would be productive.
If other railroads adopted a similar technology, or perhaps simply contracted to receive the information, these cars could be used on r
And this points to another potential advantage - a busy stretch of track may get “inspected” several times per day, vs once per day, or less. And at operating speed.
Case in point - if you watch the Deshler Diamond cam for a while, you’ll see the locos and the cars westbound taking quite the bounce. Never mind that it’s been that way since they replaced the diamond and they still haven’t fixed it…
An inspector in a hi-rail will be slowing down there, to safely traverse the diamond, and may not even sense that dip.
Going back to my original point - and my previous post - if the instrumented equipment can sense the most common problems, and that equipment runs over the rails multiple times a day, that’s got to be a good thing.
Agreed … These arguments seemed to be based on a “or” instead or a “and”
They may use this to reduce eyeball inspections, but will get more data to focus on trouble areas.
Saw this today:
https://www.caranddriver.com/news/a30443516/bridgestone-connected-tire/
Not sure if cars sending pothole data to the DOT will speed things up …
Much of the current approach to autonomous vehicles includes enormous amounts of ‘big data’, much of which will be analyzed and put into similarly vast (just not quite as huge) repository and information ‘form’. Look for some of the techniques that develop ‘emergent’ relationships between disparate variables, as in many predictive-maintenance applications, to be fired up ‘in various clouds’ for good or ill.
Do not make the mistake of thinking that any ‘autonomous’ method of railroad inspection will, or can, replace skilled observation both ‘on the ground’ and commanding and using various forms of sensor-fused data. It may be easy to extrapolate from PSR and other follies like zero-crewed trains that such a thing is the end goal; it would be no more sensible than to have robots replace Sperry rail inspection as a means of enhanced track defect detection.
What I expect, in part, is a much quicker and ‘inexpensively granular’ method of determining track state and damage. When we get to ‘intelligent trucks’ for equipment maintenance, the information streams from these can be converged for even better ‘early warning’ (or autonomous flagging of emergent concealed damage, rapidly-evolving issues like the recent New Haven kink derailment, or problems ‘as they happen’ like the switch shenanigans at Cayce with the ‘signal system’ itself shut down).
Of course there will no more be “humans” looking over the raw data than we have “humans” looking at high-resolution satellite data. What will be among the most interesting things to see is the extent to which interworkable systems with artificial intelligence are structured and evolved to assist in the desirable purposes…
[quote user=“BaltACD”]
tree68
I would suspect that such a machine would be made to find the most common problems - especially those which are known to be the most disruptive to the railroad.It’s possible that some things might actually be better (and more consistently) measured this way.
The article doesn’t say how the automated inspections will be done. Perhaps it will be via some version of the automated test cars already in circulation on some railroads (can’t remember which). If that’s the case, the the sheer weight of the car may find defects a human in a hi-rail may not detect, or may overlook.
For ties, if they haven’t already developed something, perhaps some sort of ultrasonic technique could be used. That’s over my head, technologically. Don’t expect my name on the patent.
Guage, pulled spikes, and other such items that have expected parameters are easy. Switch gaps and the like, not so much.
Such instrumented cars could be added to the consists of numerous trains. Perhaps one cold be used as a buffer in hazmat trains, instead of a covered hopper full of gravel. It still wouldn’t be a revenue car, but at least it would be productive.
If other railroads a
[quote user=“Paul Milenkovic”]
BaltACD
tree68
I would suspect that such a machine would be made to find the most common problems - especially those which are known to be the most disruptive to the railroad.It’s possible that some things might actually be better (and more consistently) measured this way.
The article doesn’t say how the automated inspections will be done. Perhaps it will be via some version of the automated test cars already in circulation on some railroads (can’t remember which). If that’s the case, the the sheer weight of the car may find defects a human in a hi-rail may not detect, or may overlook.
For ties, if they haven’t already developed something, perhaps some sort of ultrasonic technique could be used. That’s over my head, technologically. Don’t expect my name on the patent.
Guage, pulled spikes, and other such items that have expected parameters are easy. Switch gaps and the like, not so much.
Such instrumented cars could be added to the consists of numerous trains. Perhaps one cold be used as a buffer in hazmat trains, instead of a covered hopper full of gravel. It still wouldn’t be a revenue car, but at least
Reasonable question. One answer is if multiple reports about a certain section of track causing a rough ride, then it would be safe to say that the track was the problem.
A very recent development with fiber optics is making an unused fiber in a buried fiber “cable” act as a distributed seismometer. Results from experiments on Verizon cables was that the fiber could pick vibrati
I am involved with a pipeline company that is putting a fiber cable on top of a new line to measure for defects using that principle.
I gather that the new testing will be done with a track geometry car, and what BNSF wants to test at this point is a variety of testing programs, as opposed to the actual testing for the sole purpose of finding track defects. However, the union focuses on replacing manual inspections with automated inspections, which will come only in the future, and it is not yet known what that change will look like. The union’s complaint is that automatic inspections may be less capable of finding defects than human inspections, and so automatic inspections may increase danger. Yet they offer no evidence that will actually happen. BNSF points out that they expect automatic inspections will add more safety than the manual inspections that are replaced.
To support their supposition, the union cites the fiery Casselton, ND oil train wreck. However, as far as I know, that accident cause had nothing to do with track defects.
(and it isn’t dawning on anybody here yet or at the unions that you have less than 24 hours to remediate/fix/slow-order and document whatever the geometized, remote liteslice scanner in a plain looking boxcar finds as a defect - you are not laying people off, you are adding qualified people to chase after that roaming digital troublemaker… or your railroad slows to a crawl…and then is there really a defect at all[?])
[sigh][sigh][sigh]
It’s cute that you think that.
Zugs: they (FRA) didn’t change the rules to accomodate the thing. There have been two AREMA presentations about the two prototypes wandering the country in plain brown (really blue) boxcar wrappers. Dispatchers and operating supervisors are not allowed to remove those slow orders or face an FRA Code 1.(but they certainly do howl, whine and cry)… The more you crank up the sensitivity, the more problems with “flyers” you get Something’s gonna give.
Ever noticed what happens when a regular g-car finds more than the section can fix the day the g-car slides by? (your train orders come with additional slow orders and your PSR theory goes to crap that much faster because nothing gets over the road on time)…The FRA geometry cars and their contract operators are well known for false defect reporting - you still have to check every one found.
It takes time just to determine if the defect is real or not. Qualified people to do that checking do not grow on trees…and FRA is obsessed with documented remedial actions.