Different agendas - different statistics.
All I have ever been saying is - Know what the numbers REALLY represent, know who created the numbers and what they discounted in making the numbers, and know who is using them to bolster their position.
All numbers are not created and presented without bias.
Never too old to have a happy childhood!
Let me try to correct this. Not sure what Euclid is saying. I have never changed my view. Ed misread or misunderstood what I said and what the FRA threshold means, but quoted BaltACD accurately. My posts dealt with the use of FRA statistics, which have a railroad reporting threshold of about $9900 for all costs, to eliminate the expense of reporting the many minor derailments and to focus on identifying causes for possible prevention of the more serious derailments so that the railroads can rationally apply cost benefit analysis. That is what the FRA does. The problem seemed to be some folks on here seemed to have a problem with statistics used against their agenda. I only used the 20/80 ratio to show the absurdity of taking the sensible FRA derailment stats and replacing them with the actual number of derailments, ala BaltACD, and what poor PR that would become for the freight railroads.
C&NW, CA&E, MILW, CGW and IC fan
edblysard You [Schlimm] have re written and altered what BaltACD wrote…. You imply that what he meant was that only 20% of FRA reportable derailments are actually reported. What he wrote was… While I have not done an exact accounting over the years - my observations are that approximately 1 in 5 derailments ends up being FRA reportable. Discounting roughly 80% of actual derailments puts a serious skew to any reports generated from FRA reportable derailments - especially when one is looking at 'cause'.
You [Schlimm] have re written and altered what BaltACD wrote….
You imply that what he meant was that only 20% of FRA reportable derailments are actually reported.
What he wrote was…
Ed,
Actually, the point Schlimm is now making agrees with the point originally made by BaltACD.
It makes sense to me to discount the most minor derailments, mostly in yards, mostly very low speed, as you mention. In this case, the minor derailments that have been excluded are roughly 80% of the total number of derailments. As you say, including all the minor derailments would skew the statistics away from their objective purpose. BaltACD's point was that excluding the 80% skews the statistics (as I have red highlighted in his quote above).
I agree with you in that the most recent point made by Schlimm seems to indicate that he feels that excluding the 80% of minor derailments is an agenda by someone to understate the derailment problem. And yet, earlier, he was debating with BaltACD over the point of whether statistics lie or can be accepted at face value, and he took the position that the derailment statistics were fair and objective while BaltACD argued the opposite. So, Schlimm seems to have reversed his earlier position.
Actually, BaltACD, originally brought up the 80% figure as a way of claiming that that excluding 80% of derailments skews the statistics. That is exactly what Schlimm is saying now. So, at this point, Schlimm and BaltACD are in agreement.
So your position about how excluding 80% of the statistics is fair and objective disagrees with what BaltACD has said all along, and also disagrees with what Schilmm says now. So, Schlimm has not rewritten what BaltACD originally said. He has changed his position to now agree with what BaltACD has said all along.
You have re written and altered what BaltACD wrote….
What he is writing is that only one derailment in every 5 meet the threshold for reporting to the FRA, which is pretty close to my own observation…the other 4 are simple yard derailments, cars being cornered and bumped off the tracks, switches lined under cars, stuff like that, or stripped joint bars that dump a wheel or truck on the ground.
The reason the FRA sets a reportable threshold is that if you include every single derailment, including those that cause little damage and no injury, the results of any statistical analysis would be horrible skewed to the point of becoming useless.
And this highlights part of this discussion…presented with a statement, (see the above) you choose to interpret it and present it in such a manner that it alters what was intended, and used that statement to forward your agenda.
I highlighted the portion showing your written interpretation of his statement, what he wrote was “my observations are that approximately 1 in 5 derailments ends up being FRA reportable.
Both statements are in a way valid, but how they are being used again highlights the ability to use “statistics” to forward an agenda.
Both of you are using “numbers” to bolster and support your positions.
23 17 46 11
One can argue the Stats made from evidence all day, but in one way the AAR is correct. Much of what goes wrong CAN be prevented. On the Safety pyramid, at the BOTTOM, you have "At Risk behaviors". In some industries, these alone are broken down into sub categories to be addressed. These are the Foundation of ALL problems. In 99% of Industry HUMAN error IS the ERROR, even if there are 3 white hats at a accident or NOT.
Standard procedures are rarely double checked due to costs. Track problems, ARE the result of Human intervention at some point, design, or activity, or lack of upkeep. Its funny, while they are slow, the automated trains in yards do NOT have many, if any derails, while I have seen plenty otherwise. Accountability top to bottom is the key, but you have many different players at work, and the Unions... IF the FRA was like the NRC, the HUMAN part of the problem, Union or non union, would be minimal. Until the "powers that be" are all on the same page, and in CHECK.. the Stats will not change much..
tree68 schlimm...an average of 4046 per year, about 11 every day. Long way to go to ZERO. While zero is definitely a long way out, that computes to one derailment for each 12,750 miles of track.
schlimm...an average of 4046 per year, about 11 every day. Long way to go to ZERO.
While zero is definitely a long way out, that computes to one derailment for each 12,750 miles of track.
Thanks to Chris / CopCarSS for my avatar.
Larry Resident Microferroequinologist (at least at my house) Everyone goes home; Safety begins with you My Opinion. Standard Disclaimers Apply. No Expiration Date Come ride the rails with me! There's one thing about humility - the moment you think you've got it, you've lost it...
mudchicken Jackelopes! (really big ones)
Jackelopes! (really big ones)
Johnny
Irony: Dave Kepper's title of this thread was "Zero derailment railroading" The official FRA total derailments for the 10-year period 2001-2010 was 8,092. BaltACD suggests that only 20% of actual derailments are reported to the FRA. If that is the case, the actual number would have been 40,460, an average of 4046 per year, about 11 every day. Long way to go to ZERO.
Maybe the Gremlins are still around from years past.
Y6bs evergreen in my mind
The last white hat that arrives on scene usually loses. (and I had the experience of the SoCal trainmaster that could see "wide gage" from 30 miles away through the little holes in his Motorola microphone - the whole division heard that conversation over PBX)
zugmann dehusman I don't think you do. Nobody is "blaming" anybody and flat spots have relatively little to do with the service failures. When a roadmaster orders field weld kits and how they are stored probably has more to do with broken rails/failed welds than flat spots. There have been other studies done that have come up with similar conclusions. So you've never seen the triangle of blame at derailment sites?
dehusman I don't think you do. Nobody is "blaming" anybody and flat spots have relatively little to do with the service failures. When a roadmaster orders field weld kits and how they are stored probably has more to do with broken rails/failed welds than flat spots. There have been other studies done that have come up with similar conclusions.
So you've never seen the triangle of blame at derailment sites?
Would be a hit TV reality series - 'The Blame Game!'
It's been fun. But it isn't much fun anymore. Signing off for now.
The opinions expressed here represent my own and not those of my employer, any other railroad, company, or person.t fun any
zugmann I know how this plays out. Transportation blames track department for broken rails. Track department blames mechanical for allowing too many flat spots. Mechanical blames transportation for putting flat spots on wheels (sliding wheels around). Lather. Rinse. Repeat. I think we know which side came up with the quoted conclusion.
I know how this plays out.
Transportation blames track department for broken rails.
Track department blames mechanical for allowing too many flat spots.
Mechanical blames transportation for putting flat spots on wheels (sliding wheels around).
Lather. Rinse. Repeat.
I think we know which side came up with the quoted conclusion.
Dave H. Painted side goes up. My website : wnbranch.com
Would the use of slab-track where possible reduce the number of broken rails?
Conspiracy theories may be fun, but it is a lot more productive just to read the acknowledgement sections. In the paper in question, (Liu, et al.) grants were from the BNSF, NEXTRANS University Consulting Center and ABSG Consulting.
NEXTRANS is a consortium of eight midwest universities led by Purdue, primarily civil engineering.
ABSG Consulting is a worldwide company based in Houston "in the areas of probabilistic risk assessment, process safety management and hazard modeling...to create practical solutions to clients operating in high technical and regulated industries."
BNSF.
All three are part of a really suspicious, anti-rail cabal...
zugmannI think we know which side came up with the quoted conclusion.
Well maybe. But another clue to the conclusion of the study would be the interest that paid for the study.
Euclid This is their conclusion: “Prevention of broken rails or welds is expected to yield a larger percentage reduction in train and car derailment rates than other accident prevention strategies.”
This is their conclusion:
“Prevention of broken rails or welds is expected to yield a larger percentage reduction in train and car derailment rates than other accident prevention strategies.”
In order to judge the motive of statistics, you have to know the reason for them being compiled, and the conclusion they reach. With the derailment statistics in question here, this is the reason they were compiled:
“Analysis of the causes of train accidents is critical for rational allocation of resources to reduce accident occurrence in the most cost-effective manner possible.”
********************************************************************************************
On the face of it, these statistics purport to help the railroad industry make more money by preventing derailments. However, if these statistics were intentionally skewed, it would be possible for them to help the rail and weld monitoring industry make more money. This objective could be accomplished by exaggerating the broken rail/broken weld problem in relation to the other problems. Another possible motive for skewing the statistics in this way would be to help the electronic monitoring scientific research and development industry make more money.
schlimm BaltACDThere are thousands of decisions that go into making the FRA reports and the statistics that get generated from those reports. Decisions one doesn't think of when someone says the FRA report says X. Would you prefer no statistics? What is your solution to the problem? The human errors for accidents are probably limited by defintion to engineer errors and dispatcher-type errors, not poor work on track repair. By your reductio ad absurdem, a faulty weld could be track error or human. Would you prefer every rail accident, no matter how minor, be reported? Or do you have some problem with the parameters the researchers set to analyze the data base? If the latter, the FRA database is open and you (or some staffers at your railroad) could make your own analysis according to your needs
BaltACDThere are thousands of decisions that go into making the FRA reports and the statistics that get generated from those reports. Decisions one doesn't think of when someone says the FRA report says X.
Would you prefer no statistics? What is your solution to the problem? The human errors for accidents are probably limited by defintion to engineer errors and dispatcher-type errors, not poor work on track repair. By your reductio ad absurdem, a faulty weld could be track error or human. Would you prefer every rail accident, no matter how minor, be reported? Or do you have some problem with the parameters the researchers set to analyze the data base? If the latter, the FRA database is open and you (or some staffers at your railroad) could make your own analysis according to your needs
What I have a problem with is people using 'the numbers' and thinking they know what the numbers represent when they don't have a clue. This isn't limited to FRA 'numbers' but all 'studies' that end up reporting 'numbers'. I'm guilty, your guilty, the world as we know it is guilty of using 'numbers' that we have no REAL idea of what the represent and how they ACTUALLY were derived.
When the going gets tough, get tough on the numbers and find out how they were derived and what they are actually representing. Everybody that publishes 'numbers' have some agenda - what that agenda is may not be obvious - that is our world and it is what we have to battle through to form 'our truth'. My truth and your truth won't necessarily be the same truth.
YMMV
daveklepper An evaluation worth reading from Progrressive Railroading: — by Toby Kolstad -- snipped -- There is one common denominator to all the tank-car derailment tragedies of late that has hardly been mentioned: Train speeds all exceeded 40 mph. High-speed derailments almost always involve dozens of cars piled into a very small space; in derailments at slower speeds, cars tend to stay coupled together and upright. Limiting the speed of trains with blocks of haz-mat tank cars to 25 mph would reduce the catastrophic consequences of derailments involving these cars by almost 100 percent. That would be far more effective than the 56.6 percent reduction estimated by the AAR's own research for the tank-car changes they have suggested, or the 37 percent reduction that would be achieved with RSI's recommendations. However, a 25 mph speed limit is as impractical to railroads as insulating and jacketing all existing tank cars in haz-mat service is for rail-car lessors.
An evaluation worth reading from Progrressive Railroading:
— by Toby Kolstad
-- snipped --
There is one common denominator to all the tank-car derailment tragedies of late that has hardly been mentioned: Train speeds all exceeded 40 mph. High-speed derailments almost always involve dozens of cars piled into a very small space; in derailments at slower speeds, cars tend to stay coupled together and upright. Limiting the speed of trains with blocks of haz-mat tank cars to 25 mph would reduce the catastrophic consequences of derailments involving these cars by almost 100 percent. That would be far more effective than the 56.6 percent reduction estimated by the AAR's own research for the tank-car changes they have suggested, or the 37 percent reduction that would be achieved with RSI's recommendations. However, a 25 mph speed limit is as impractical to railroads as insulating and jacketing all existing tank cars in haz-mat service is for rail-car lessors.
The suggested 25 MPH speed limit would have done nothing to minimize the consequences of the Lac Megantic disaster or any other runaway.
- Erik
Feel good article in today's paper about how the state is stepping up to ensure safer railroading here. They went through a couple of yards, found some wheels and brakes out of tolerance, and a few lose bolts. One broken rail.
Pretty much everyday railroading.
They inspected a total of six miles of track (in yards) and 44 switches...
That didn't stop the governor from saying, "This inspection blitz has resulted in immediate improvement to some of the state's busiest rail sites..."
A example of why I distrust 'statistics' when the input data has been categorized.
CSX had a derailement in Philadelphia at the end of January with a oil train on a bridge over the Schuykill Expressway.
It was just recently published that the track had just ungone tie replacement maintenance. The proximate cause was the rail turned over because it wasn't properly secured to the ties in the area that had been worked on, because the personnel responsible for the tie replacement did not follow company standards before they released the track back for service.
Do we have a track failure derailment? Do we have a man failure derailment? The TRUE cause is man failure in not following company procedure. How do you report this for FRA statistics?
I have no idea if this incident will be reported as a track failure or a man failure. No matter how it gets reported it will skew the data. This is just a single incident. The reporting of hundreds, if not thousands, of incidents make up the FRA statistics and in many of the incidents decisions must be made on how the cause is actually reported. If both causes were to be reported then the incident would be double reported.
There are thousands of decisions that go into making the FRA reports and the statistics that get generated from those reports. Decisions one doesn't think of when someone says the FRA report says X.
There are two railroads in my town. One is owned by the government. It operates only on city property.
The other is privately owned. The privately owned railroad runs on two tracks through town and hundreds of miles beyond.
Over the last 50 years, the private railroad has had several derailments within the city limits. One derailment was caused by a run away string of loaded coal cars. Luckily, nobody has been killed, nor injured as a result.
The government owned railroad's safety record is spotless.In over 50 years, that old steam locomotive down by the depot become a park has never once derailed.
Comparing safety statistics, there is no comparison. Cost to benefit ratios, excluding esthetics, are a different matter.
OvermodOr is the wheelset containing that wheel automatically flagged, pulled and checked ... as I expect it would be.
That's what I've always seen happen. Even the smallest derailment sends the car to the shop.
There is rarely anything sinister about the point of view behind statistics, and particularly not in this case. But they are always developed for a very specific purpose which may or may not be suitable for other studies and conclusions.
How the categories are defined by the statistician will make a great deal of difference to the apparent results. In this case, including or excluding derailments under a dollar threshold makes a huge difference. Changing the threshold or merely adding certain specific types that are otherwise under the threshold will also change the numbers.
Statistics provide useful information but should be used only as a guide, not taken as fundamental truth. The railroad industry and the pipeline companies each produce accurate statistics about the safe transportation of oil, but the apparent conclusions are somewhat different. That is why knowing the point of view is important.
John
dehusman To answer the question of whether you would get a different answer if you included all the derailment data, the answer is probably yes. Would you get a "better" insight to the problem, it depends on what the problem is. If the problem is reducing deaths in children, knowing that the most common injury to kids is a skinned knee won't lead you to a soution that solves the problem. You would want to ignore the data on the minor injuries and concentrate on only those that could cause a fatality. Not including all the data is not necessarily mean the the analysis is flawed.
To answer the question of whether you would get a different answer if you included all the derailment data, the answer is probably yes. Would you get a "better" insight to the problem, it depends on what the problem is. If the problem is reducing deaths in children, knowing that the most common injury to kids is a skinned knee won't lead you to a soution that solves the problem. You would want to ignore the data on the minor injuries and concentrate on only those that could cause a fatality.
Not including all the data is not necessarily mean the the analysis is flawed.
Amen, and well-stated.
I would add that adding information about 'most' low-speed derailments would provide "data" that can be actively manipulated by people who actually want to 'lie with statistics'.
Are there aspects of low-speed derailments that might have implications for safety going forward? -- for example, would a low-speed derailment damage a wheel or flange and make it susceptible to accelerated failure in service? Or is the wheelset containing that wheel automatically flagged, pulled and checked ... as I expect it would be. I think you would need to establish what categories of lower-speed/lower-energy derailment were to be included, based on actual relevance to causing or inducing the higher-speed/higher-consequence ones. But certainly no criterion strictly based on derailment 'events' at any speed tells you much that is useful in our context...
BaltACD The principles that apply to the creation of the Safety Pyramid also apply to derailment causation. When you only look near the top of the pyramid you have no real understanding of everything that went into something getting near the top. People don't publish study statistics that don't support their point of view - whatever that point of view may be - and EVERYONE has a point of view.
The principles that apply to the creation of the Safety Pyramid
also apply to derailment causation. When you only look near the top of the pyramid you have no real understanding of everything that went into something getting near the top.
People don't publish study statistics that don't support their point of view - whatever that point of view may be - and EVERYONE has a point of view.
But why do you seem to assume that the point of view behind the derailment statistics is sinister? What is the point of view that you see in them? Why do you not address that question? Why do you keep shooting the messenger without any mention of the message?
Our community is FREE to join. To participate you must either login or register for an account.