Trains.com

Oil Trains & Lag Screws

27526 views
426 replies
1 rating 2 rating 3 rating 4 rating 5 rating
  • Member since
    September 2003
  • 21,669 posts
Posted by Overmod on Wednesday, July 13, 2016 12:33 AM

Buslist
... a GRMS development vehicle detected a rash of broken hair pin spikes on Pandrol rolled plates around North America in the 90s.  Peed off Pandrol as the industry heard about it first (as it was an FRA funded test)rather than Pandrol.  I believe this led to the use of the coach screw (as it was then called) rather than the hair pin as standard kit in the rolled plate system.

I feel reasonably certain that we are to see reintroduction of one sort or other of hairpin spike in this plate fixation system, as I think while we may chuckle a bit at Moishe's name, he's conducted a reasonable analysis of some of the requirements. I am not sure how the 'hairpin' top of these spikes can be reshaped to give positive clamping pressure (and lateral 'centering' with elastic restoring force) through the shape of the 'eye' and the shape of the seats adjacent to the spike holes in the plates.  Relatively small refinements here may make for increased clamping and 'nonlifting' performance, even when comparatively great deferred maintenance occurs.

  • Member since
    November 2013
  • 1,097 posts
Posted by Buslist on Tuesday, July 12, 2016 10:02 AM

Overmod

 

 
Euclid
... there is critical visual evidence that the tie plate experiences cyclical sliding on the tie. There is bluish, polished wood, stained with the carbon in the tie plate steel that exposed next to the outer edge of the plate. This shows that the plate is worked back and forth with the passage of trains. That visual indication alone is clear evidence of the defective screw connection.

 

But note that the extent of this 'visual evidence' is likely very slight, in any case other than colossal deferred maintenance, for most of the early 'tension failure' effects -- it is only when all four lags are broken that horizontal excursion much more than shank clearance plus deflection will occur, and even then unlikely that one plate will deflect measurably more than its neighbors.  Much more likely in the latter case will be preferential 'shedding' of the deflection load into the perhaps-already-compromised lags in those neighbors, accelerating their own tendency to loosen, then deflect, then fail.  But note that the actual deflection needed to drop a wheel into the gauge, as presently claimed to be proximate cause at Mosier, is not much more than an inch.

This calls for much better machine vision systems for inspection and reporting, including proper three-axis positioning systems that will stand up to anticipated "inspection" running, storage, and maintenance conditions.  The hardware and at least some of the required control logic and software versions of such systems are, I think, already available from a number of industry suppliers: they just have to be adapted and reprogrammed for this use.

 

 
But it will be missed by inspectors riding in hi-rail vehicles.

 

Not if those vehicles have the machine-vision systems installed on them.  It might require some careful editing of existing rules to permit realtime monitoring screens to be used in rail-guided hi-rail vehicles (or permit their use by a second inspector riding in the vehicle) but high-speed data acquisition and automated pattern recognition can easily flag potential defects for near-instant review.

I also see little reason why some of the gage-restraint functions cannot be installed on a comparatively light vehicle (as they involve some mutual lateral outward displacement of paired wheels while the machine-vision system observes the effect).  Here the light weight of the vehicle may actually enhance the integrity of the measurements as there will be little tendency for the spring clips to deflect or the plates to friction-lock under the load of a typical geometry train or Sperry car.  (Note that it may be necessary to distinguish rail movement in the Pandrol spring clips from lateral movement of the plate due to loose or broken fasteners, but this should pose little technical difficulty for a properly-designed system.)

More to the point:  It should be cost-effective to adapt OTS machine-vision systems to multiple hi-rail vehicles, and then use them for streaming data acquisition and near-realtime analysis (to reduce the data-storage and transmission requirements) continuously, whenever a hi-rail vehicle runs over a given stretch of track.  What is then required is more prompt attention to the remediation of 'loose' lag screws ... at a minimum, gang-tightening them (at which time, of course, any broken lags or 'spindled' or internally-compromised ones will be discovered for remediation as well ... than existing track-geometry deviation calls for.  That was an expensive lesson for railroads to learn, but I think a far less expensive one than replacing miles of sprung track with dubious alternative technology...

 

 

The problem with your machine vision approach is it only detects gage widening if the plates are translating, there is at least one other cause, rail rotation As it only measures delta gage. The beauty of the GRMS system is that it will detect both and any other causes Of wide gage under load. In fact a GRMS development vehicle detected a rash of broken hair pin spikes on Pandrol rolled plates around North America in the 90s. Peed off Pandrol as the industry heard about it first (as it was an FRA funded test)rather than Pandrol. I believe this lead to the use of the coach screw (as it was then called) rather than the hair pin as standard kit in the rolled plate system. Several Engineering Dept. officials assure me that a GRMS inspection would have caught this if conducted more frequently than UP's 18 month interval.

  • Member since
    December 2007
  • From: Southeast Michigan
  • 2,983 posts
Posted by Norm48327 on Monday, July 11, 2016 2:33 PM

If, as the article in the Portland Mercury News says, those bolts were installed in 2013 it kinda confirms my suspicion they were made from "junk steel" and not of a satisfactory grade for the task.

Norm


  • Member since
    September 2003
  • 21,669 posts
Posted by Overmod on Monday, July 11, 2016 10:29 AM

Euclid
... there is critical visual evidence that the tie plate experiences cyclical sliding on the tie. There is bluish, polished wood, stained with the carbon in the tie plate steel that exposed next to the outer edge of the plate. This shows that the plate is worked back and forth with the passage of trains. That visual indication alone is clear evidence of the defective screw connection.

But note that the extent of this 'visual evidence' is likely very slight, in any case other than colossal deferred maintenance, for most of the early 'tension failure' effects -- it is only when all four lags are broken that horizontal excursion much more than shank clearance plus deflection will occur, and even then unlikely that one plate will deflect measurably more than its neighbors.  Much more likely in the latter case will be preferential 'shedding' of the deflection load into the perhaps-already-compromised lags in those neighbors, accelerating their own tendency to loosen, then deflect, then fail.  But note that the actual deflection needed to drop a wheel into the gauge, as presently claimed to be proximate cause at Mosier, is not much more than an inch.

This calls for much better machine vision systems for inspection and reporting, including proper three-axis positioning systems that will stand up to anticipated "inspection" running, storage, and maintenance conditions.  The hardware and at least some of the required control logic and software versions of such systems are, I think, already available from a number of industry suppliers: they just have to be adapted and reprogrammed for this use.

But it will be missed by inspectors riding in hi-rail vehicles.

Not if those vehicles have the machine-vision systems installed on them.  It might require some careful editing of existing rules to permit realtime monitoring screens to be used in rail-guided hi-rail vehicles (or permit their use by a second inspector riding in the vehicle) but high-speed data acquisition and automated pattern recognition can easily flag potential defects for near-instant review.

I also see little reason why some of the gage-restraint functions cannot be installed on a comparatively light vehicle (as they involve some mutual lateral outward displacement of paired wheels while the machine-vision system observes the effect).  Here the light weight of the vehicle may actually enhance the integrity of the measurements as there will be little tendency for the spring clips to deflect or the plates to friction-lock under the load of a typical geometry train or Sperry car.  (Note that it may be necessary to distinguish rail movement in the Pandrol spring clips from lateral movement of the plate due to loose or broken fasteners, but this should pose little technical difficulty for a properly-designed system.)

More to the point:  It should be cost-effective to adapt OTS machine-vision systems to multiple hi-rail vehicles, and then use them for streaming data acquisition and near-realtime analysis (to reduce the data-storage and transmission requirements) continuously, whenever a hi-rail vehicle runs over a given stretch of track.  What is then required is more prompt attention to the remediation of 'loose' lag screws ... at a minimum, gang-tightening them (at which time, of course, any broken lags or 'spindled' or internally-compromised ones will be discovered for remediation as well ... than existing track-geometry deviation calls for.  That was an expensive lesson for railroads to learn, but I think a far less expensive one than replacing miles of sprung track with dubious alternative technology...

  • Member since
    January 2014
  • 8,221 posts
Posted by Euclid on Monday, July 11, 2016 9:58 AM

U.P. routinely inspected the track at Mosier every few days, but completely failed to detect the lag screw failure that had been in progress for many years.  Detecting the lag screw problem requires walking the track and pulling up on the heads of the screws.  The broken ones will lift right out of the hole, thus revealing the fact that they are broken.  In addition, every so many screws, one is actually unscrewed, removed, and visually inspected for corrosion, cracks, and damage.  Of particular importance is to check the condition of the screw shank right under the head because this damage precedes the actual breaking failure further down on the screw shank. 

In addition, the walking inspection includes looking for damage to the tie plate.  This link has a photo of the track at Mosier, and shows what can be found without actually lifting or unscrewing lag screws:

http://www.portlandmercury.com/blogtown/2016/06/16/18240394/state-defect-that-caused-mosier-oil-train-crash-wasnt-detectable

Even without the presence of broken and dislodged screws (as shown in the photo), there is critical visual evidence that the tie plate experiences cyclical sliding on the tie.  There is bluish, polished wood, stained with the carbon in the tie plate steel that exposed next to the outer edge of the plate.  This shows that the plate is worked back and forth with the passage of trains.  That visual indication alone is clear evidence of the defective screw connection.  But it will be missed by inspectors riding in hi-rail vehicles. 

These walking inspections need not be done every few days.  I suspect that every several months would be sufficient.  Another way to inspect in less detail, but sufficient find evidence of lag screw failure is by the use of a gage restraint vehicle.

Broken lag screws lead to gage widening, but simply checking the gage is not enough. This is because, as the problem develops, the gage on track with no passing train can be normal.  But when a train passes, the rails spread as the trucks pass, but spring back to normal gage after the trucks pass. 

So, to check if they will spread under a passing train, the gage restraint vehicle places lateral pressure on the rails as would happen with a passing train. It measures the track’s ability to hold the gage under the lateral force of a passing train:

https://www.fra.dot.gov/Page/P0291 

  • Member since
    January 2014
  • 8,221 posts
Posted by Euclid on Sunday, July 10, 2016 10:27 PM

tdmidget

[Euclid in blue text] "What this does is allow the screw shank to bend over some distance of the unthreaded shank, so the bending stress does not concentrate at the start of the threads and reduced shank"

"But bear in mind, that there will also be less bending with the use of the bushing because it spreads out the lateral force to a much larger area of wood surface."

Trying to have it both ways ,Bucky?

" The entire bolt would be encased in the threaded insert, so it cannot bend"

How can a bolt, or any other fastener, hold anything if it is entirely encased in something? A threaded fastener works by being in tension between the head of the fastener and what it is screwed in to, be it a nut or threaded object.

"So its ultimate failure would be shearing or breaking off right under the shoulder or flush with the top of the tie"

No, the weakest part of any threaded fastener is the minor diameter, which is usually the gullet of the threads. You can't change the laws of physics. The weakest point is where the least material is.

Curious, Buckey, What do you do for a living? Apparently nothing to do with machines or materials or you would know better.

Have it both ways?  You’ve got to be kidding.  I thought you of all people would readily understand it.  It is indeed a matter of having it both ways if you stop to think about it. 

The application of rail securement is full of constraints that prevent it from being one bit more reliable than what it takes to get by.  I don’t expect my suggestions to be adopted.  Any proposal that even requires an added part will be rejected out of hand.  At the very most, a slight change in screw metallurgy might be considered.  So there can be no ideal design solution.  That calls for compromise, or as you call it: “Having it both ways.”

As you must know, the lateral force on the screw causes it to crush the wood of the tie that surrounds the screw.   To do that, the screw must bend. 

So, the overall resistance to the lateral shifting of the tie plate is from two factors:

  1. The screw’s resistance to bending.

  2. The surrounding wood’s resistance to crushing. 

The wood’s resistance to crushing depends on the surface contact area between the screw and the wood.  The bushing adds to the screw diameter, thereby increasing the contact area.

Thus, the greater resistance to the crushing of the wood also adds to the screw’s resistance to bending. 

There is also a third element that contributes to the resistance of the lateral shifting of the tie plate.  That is that the bushing also directly stiffens the screw shank by effectively increasing its diameter.  That is what I mean by the screw being “encased” in the bushing.  Obviously, by “encased” I do not mean that the screw and bushing are bonded together as one; as you have seized upon as a point of criticism.  Obviously, it is a sliding fit between the screw and the bushing.   

HOWEVER: There will still be crushing of the wood and bending of the screw, even though they are greatly reduced by the application of the bushing.  The bending and crushing can never be zero.  And since ultimate failure always culminates over time with repeated load cycles, even a greatly reduced amount of bending and crushing will inevitably lead to failure if given enough time of use. 

The effect of the solid bushing around the unthreaded portion of the screw will nearly prevent bending there because for the screw to bend, the bushing would also have to bend.  But as you point out, the crushing of the wood will allow the bushing to tip; and that will allow bending of the screw shank; and that bending will simply be shifted to the part of the screw shank that is not surrounded by the bushing. And that part of the screw will be at the bottom of the bushing where the screw threads begin, which is the weakest point of the screw in terms of resisting bending, as you mention.  So, in that regard, while the bushing solves one problem, it creates another problem.

So this is where it calls for compromise, or “having it both ways” as you say.  The compromise is to open up the bushing bore for some distance above the start of the screw threads so the bending can spread out and not be concentrated at the weakest point of the screw shank.  Although the bore is opened up to clear the screw shank, the full length of the bushing O.D. is still tightly fitted to the corresponding bore in the wood tie.  That is the design compromise; the “having it both ways” with just the right amount of each way to achieve the “sweet spot.”  It is not perfect, but will nevertheless vastly improve the resistance to failure over the present system without bushings. 

  • Member since
    December 2001
  • From: Northern New York
  • 25,022 posts
Posted by tree68 on Sunday, July 10, 2016 9:23 PM

Barring some other technique (assuming such screws actually still get used after all this), I would suggest that perhaps the only inspection that will work will be the removal of some percentage of the lag bolts for an actual physical inspection.

As I recall, it's already been established that current inspection methods (I presume a visual in-place look at them) are not enough.

Should such bolts be considered for future use, I'd think that the screws pulled from Mosier would be carefully scrutinized for how they failed, and even when.  Some of the bolts in the picture look to have been broken off for quite some time, and others in the image seem to show a wide variety of modes of failure.

LarryWhistling
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...

  • Member since
    January 2014
  • 8,221 posts
Posted by Euclid on Sunday, July 10, 2016 3:04 PM

Both the U.P. and the FRA seem to be in the hot seat over the fact that the lag screw mass failure at Mosier was not seen as it developed over many years.  U.P. said they did the inspections required by the FRA, and the FRA says the inspections they require are only the minimum.  I would say that inspecting the lag screws should have been a part of that “minimum.”

The TSB of Canada clearly defined the lag screw problem after investigating the CN Fabyan Bridge derailment in 2012.  That should have been a red alert wakeup call to the FRA.  But I guess the FRA was not required to consider it.   

http://www.opb.org/news/series/oil-trains/how-inspectors-missed-broken-bolts-mosier/

 

  • Member since
    December 2007
  • From: Southeast Michigan
  • 2,983 posts
Posted by Norm48327 on Sunday, July 10, 2016 2:39 PM

tree68

Let's go back and take a look at the bolts actually used.  This image is of bolts removed at the Mosier site.

Lag Bolts

AP Photo

 

Take particular note of the bolt in the  lower left. Threads are almost nonexistent. Corrosion appears to be a major problem. Without knowing how long they've been in service I'd hazard a guess they were made from junk steel with no consideration given to corrosion resistance or service life. On the SAE scale I'd say less than grade five.

Norm


  • Member since
    December 2001
  • From: Northern New York
  • 25,022 posts
Posted by tree68 on Sunday, July 10, 2016 2:03 PM

Let's go back and take a look at the bolts actually used.  This image is of bolts removed at the Mosier site.

Lag Bolts

AP Photo

LarryWhistling
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...

  • Member since
    May 2003
  • From: US
  • 25,292 posts
Posted by BaltACD on Sunday, July 10, 2016 1:39 PM

rdamon
 I think he may be referring to a design like this. Seems like it would add additional complexity. 

eff

Effectivelty a Helicoil that gets used routinely in the automotive world to 'rethread' bolt holes that have had their thread stripped.

Never too old to have a happy childhood!

              

  • Member since
    June 2004
  • From: roundhouse
  • 2,747 posts
Posted by Randy Stahl on Sunday, July 10, 2016 1:04 PM

I think the tie plates are too slippery.. We use them here and I don't think much of them.

 

Randy

  • Member since
    January 2014
  • 8,221 posts
Posted by Euclid on Sunday, July 10, 2016 12:16 PM

rdamon,

I was talking about two different design approaches, one using a threaded insert and one using a bushing.  What you show is the threaded insert approach.  Either approach would add complexity and cost.  In lieu of such improvment would be new inspection methods, and the time it takes to do them.  That too would add cost.  If nothing is done, then you have the cost of the train wrecks that tell when the system has worn out to the point of failure. 

Union Pacific has decided that the most cost effective approach is to revert back to the conventional driven cut spike and accept the relatively higher cost of its maintenance. 

As to the threaded insert, the one you show is primarily intended to increase pullout resistance.  The threaded insert I have proposed would be primarily intended to resist the lateral force on the screw.  So it would have a relatively larger outside diameter to increase the contact area with the wood tie. 

  • Member since
    July 2008
  • 2,325 posts
Posted by rdamon on Sunday, July 10, 2016 11:23 AM
 I think he may be referring to a design like this. Seems like it would add additional complexity. 

eff

  • Member since
    December 2004
  • 707 posts
Posted by tdmidget on Saturday, July 9, 2016 11:56 PM

"What this does is allow the screw shank to bend over some distance of the unthreaded shank, so the bending stress does not concentrate at the start of the threads and reduced shank"

"But bear in mind, that there will also be less bending with the use of the bushing because it spreads out the lateral force to a much larger area of wood surface."

Trying to have it both ways ,Bucky?

" The entire bolt would be encased in the threaded insert, so it cannot bend"

How can a bolt, or any other fastener, hold anything if it is entirely encased in something? A threaded fastener works by being in tension between the head of the fastener and what it is screwed in to, be it a nut or threaded object.

"So its ultimate failure would be shearing or breaking off right under the shoulder or flush with the top of the tie"

No, the weakest part of any threaded fastener is the minor diameter, which is usually the gullet of the threads. You can't change the laws of physics. The weakest point is where the least material is.

Curious, Buckey, What do you do for a living? Apparently nothing to do with machines or materials or you would know better.

 

  • Member since
    January 2014
  • 8,221 posts
Posted by Euclid on Saturday, July 9, 2016 9:42 PM

tdmidget

Absolutely wrong Buckey. The only reason they last as long as they do is that the shank can bend. If the shank can't bend that entire moment is transferred to the minor diameter at the first thread not reinforced by your bushing. You might snap quite a few in the first year. It is inescapable that the minor diameter is the weak point of the fastener and a steel screw in a hardwood tie cannot produced enough tension to prevent lateral movement. Even if you used a bolt into a flanged bushing inserted from the bottom the wood will compress and tension will be lost in a matter of months, at the most.

The lag screw is like cocaine, immediate gratification but no long term benefits. If the problem is lateral movement then a threaded fastener in a wood tie is NOT the answer.

Tdmidget,

I agree with your conclusion that the typical lag screw threaded fastener in a wood tie is not the answer.  I do not believe that entire system is robust enough for what it is expected to do.  I believe that its vulnerability has been under estimated. 

I also believe, but cannot prove, that screw tension or even the wheel loading on the tie plate will not produce adequate, reliable, long term friction sufficient to prevent lateral movement of the tie plate.  I am not convinced that such restraint by compression of the tie plate producing friction has ever been intended.

And finally, I agree somewhat with your first comment.  I must point out, however, that when you start by saying I am wrong, you are disagreeing with points I never asserted. 

I agree with your point that if the shank is prevented from bending over the length encased in a snug bushing; and if the bushing crushes into the wood laterally from the shear force of the tie plate against the bolt; then that force will tend to bend the screw shank immediately below the bushing, which is exactly where the shank reduces diameter for the start of the threads, as you say.  But, I was completely aware of that, and described it when I wrote the previous post.  If you read what I said, you will see that I even included a special feature to offset that problem. 

In that regard, I said this:

“I would make the inside diameter of the bushing fit snug to the lag screw only for the first inch or so below the screw head.  Then below that, the bore of the bushing would increase by about 1/16 inch so it has a clearance gap between it and the screw shank.  With this gap feature, any slight bending of the screw that does occur will not concentrate stress on the point where the screw threads begin. 

Otherwise, if the bushing bore were snug all the way to the threads, the bending force on the screw would not be able to bend the screw where it is confined by the bushing.  In that case, all of the bending force would be concentrated right where the screw shank enters the bottom of the bushing.”

What this does is allow the screw shank to bend over some distance of the unthreaded shank, so the bending stress does not concentrate at the start of the threads and reduced shank diameter. 

But bear in mind, that there will also be less bending with the use of the bushing because it spreads out the lateral force to a much larger area of wood surface. 

I go back and forth on my preference of the bushing versus the insert as the most favorable solution.  In my previous post, I was leaning toward the bushing due to thinking that there is no way to relieve the fit of the screw within the insert as I propose for the bushing.  But that type of relief for slight bending is simply not needed with the insert.  The entire thread body would be within the insert, so there is no weak point below the insert that would break due to stress concentration as there is with the bushing.  With the insert, you would use a structural bolt with an un-threaded shoulder feature that would fill the thickness of the tie plate.  The entire bolt would be encased in the threaded insert, so it cannot bend.  So its ultimate failure would be shearing or breaking off right under the shoulder or flush with the top of the tie.  And like the bushing, this insert would spread out the force over a much larger area of the wood tie. 

So I now think the threaded insert and the bolt is the way to go.  It would not break within the tie, losing all of its holding power, where the break could not be detected.  And it would leave the tie to be far less susceptible to damage from the shearing force against the bolt under its head.  Although the bushing approach with the conventional lag screw might also solve the problem do the greater contact area receiving the lateral pressure. 

  • Member since
    December 2004
  • 707 posts
Posted by tdmidget on Saturday, July 9, 2016 8:36 PM

Absolutely wrong Buckey. The only reason they last as long as they do is that the shank can bend. If the shank can't bend that entire moment is transferred to the minor diameter at the first thread not reinforced by your bushing. You might snap quite a few in the first year. It is inescapable that the minor diameter is the weak point of the fastener and a steel screw in a hardwood tie cannot produced enough tension to prevent lateral movement. Even if you used a bolt into a flanged bushing inserted from the bottom the wood will compress and tension will be lost in a matter of months, at the most.

The lag screw is like cocaine, immediate gratification but no long term benefits. If the problem is lateral movement then a threaded fastener in a wood tie is NOT the answer.

  • Member since
    January 2014
  • 8,221 posts
Posted by Euclid on Saturday, July 9, 2016 4:50 PM

Norm48327
Euclid

I do not believe that sufficient friction between the wood tie and the tie plate can be created by the clamping pressure of the lag screws to the extent that it prohibits the shear movement of the tie plate. 

Perhaps you could use your engineering experience to find a cure for that.

I can see two potential improvements:

  1. The use of a threaded insert and straight thread bolt instead of a lag screw.

  2. The use of a loose bushing with the lag screw.

As I mentioned, I don’t think clamping force causing friction between the tie and tie plate is sufficient for the task.  I am surprised that it was ever intended to perform that function.  I would have assumed that was never intended as part of the design, except that the TSB report cites it as though it is part of the design intent.  They say this:

“If the screws are tight on the rolled plates, the clamping force produces friction between the plate and the tie, providing lateral resistance to gauge-widening forces in the curve. If the screws are loose, they act as pins, providing the sole lateral resistance.”

Another thing to consider as was mentioned by others earlier, is that the wheel loading itself would provide a lot of clamping pressure and tie plate friction just at the moment that it is needed to resist the lateral wheel force to toward the outside of the curve.

While it would be nice to have the screws tight, I don’t think they can be relied on to provide a critical range of clamping force.  Overall, I don’t think tie-to-tie plate friction is the answer to resisting the lateral shift of the tie plate.  So (for wood ties), that leaves only the pin effect of the screws to hold the tie plate from shifting laterally.

The problem is that, as pins in shear, the lag screws driven into the wood ties are not adequate to hold the tie plates from shifting laterally. So, what to do?

What would solve the problem is bigger screws, stronger screws, or some type of threaded steel insert driven into a bore in the tie.  With the insert approach, you could reduce the thread height and use a standard UNC bolt thread.  This would allow a larger shank diameter of the bolt and reduce the stress concentration where the unthreaded shank meets the threads.  The threaded insert might be up to 1.5” outside diameter, and be driven into a bore in the wood of nearly that size.  So this would provide much more resistance to bending the bolt and crushing its shank into the wood compared to the present system using just a .94 dia. lag screw.  The insert would have an internal threaded contact with the screw, so this would prevent the screw from bending from the force of the laterally shifting tie plate. 

So, pressing in threaded inserts into hole bored into the tie would be one approach to increasing the robustness of the bolted connection between the rail and the ties.     

Another approach would be to use the lag screw threaded into its pre-bored hole as is typical, but with an added loose bushing.  This bushing would be added around the screw shank between the bottom of the screw head and the start of the threads.  Such a bushing would be a steel cylinder with a .94” nominal bore to fit the screw shank, and an outside diameter of 1.50”, leaving a wall thickness of .28”. 

The length would be around 2.00” or whatever is needed to cover the unthreaded portion of the screw shank.  The tie plate would have 1.50” dia. Holes to fit the bushing. 

Like the threaded insert mentioned above, the loose bushing would spread out the bending load of the bolt onto a larger area of the wood tie, which would then be more able to resist the bending without crushing.  The overall effect would be to greatly reduce potential bending of the screw shank.

I would make the inside diameter of the bushing fit snug to the lag screw only for the first inch or so below the screw head.  Then below that, the bore of the bushing would increase by about 1/16 inch so it has a clearance gap between it and the screw shank.  With this gap feature, any slight bending of the screw that does occur will not concentrate stress on the point where the screw threads begin. 

Otherwise, if the bushing bore were snug all the way to the threads, the bending force on the screw would not be able to bend the screw where it is confined by the bushing.  In that case, all of the bending force would be concentrated right where the screw shank enters the bottom of the bushing.  This protection against stress concentration is the main reason why I prefer this method of using a bushing over the method of using a threaded insert. 

Overall, there is one basic problem with transferring shear force of the tie plate to the screws.  The principle requires a tight fit to be most reliable, and to equalize the force on all four screws.  Both the tie plate and the tie are manufactured with four screw holes.  For a tight fit, these two patterns of four holes must match perfectly or very close to it.  If they don’t quite match, one or more of the screws will not engage their threads.  So to make sure these holes line up well enough, they must be oversize to the screw shank which compromises the mechanical integrity of the securement. 

  • Member since
    October 2006
  • From: Allentown, PA
  • 9,810 posts
Posted by Paul_D_North_Jr on Saturday, July 9, 2016 3:52 PM

Kinda sorta related is an article in today's (Sat. 09 July) Wall Street Journal (pg. B-3) about many failures of large bolts in undersea oil rig equipment.  Pretty dramatic photo of one broken off about 1/3 of the way down (of about a dozen in a circle of a flange or similar). The main factors being looked at are the metallurgy of the alloys in the corrosive undersea environment, the coatings on the bolts, and over-torquing when installed.

- Paul North. 

"This Fascinating Railroad Business" (title of 1943 book by Robert Selph Henry of the AAR)
  • Member since
    December 2007
  • From: Southeast Michigan
  • 2,983 posts
Posted by Norm48327 on Friday, July 8, 2016 5:19 PM

Euclid

Overmod,

I agree that the Fabyan Bridge wreck was due to deferred maintenance, but it was deferred maintenance of the very lag screw problem that I am addressing.  And the reason for the deferred maintenance was that the lag screw problem was unknown. The way I see it, it was the very same development as that which led to the Mosier wreck.  So all that I am suggesting is a new method to address the maintenance of the previously unknown lag screw problem that has now been revealed by the two recent wrecks.

I do not believe that sufficient friction between the wood tie and the tie plate can be created by the clamping pressure of the lag screws to the extent that it prohibits the shear movement of the tie plate.  I understand your point about a lag screw preload being lost by a slight relaxation of its connection to the tie. Probably just the shrinking and swelling of the tie from moisture changes could disrupt that preload.  So, from a practical standpoint, I doubt that any lag screws in use today have much if any preload on the tie plate (unless they use some type of compression spring).

In practice, it is the pin effect of the lag screw shanks that resists the shear force from the tie plates.  Photos clearly show the wear and upset of the screw shank right under the head, which is caused by the shear force of the tie plate. 

If this shear force wear and upset under the head were to continue, it would eventually shear or break right there.  But it seems that the shear force also bends the screw shank, eventually causing fatigue sufficient to cause the screw to break several inches below the application of the shear force. 

So it is the bending of the screw, caused by the shear force under the head, which has to be resisted.  Then the screw must be replaced before it breaks off under the head where the shear force is directly applied.    

The resisting of the bending of the screw can be achieved by increasing its contact area with the tie along the shank of the screw.  Perhaps this could be achieved by adding a steel bushing to the screw, extending down an inch or two below the head of the screw.  Then the screw hole pre-drilled in the tie would also require a counter-bore for the bushing. 

 

Perhaps you could use your engineering experience to find a cure for that.

Norm


  • Member since
    January 2014
  • 8,221 posts
Posted by Euclid on Friday, July 8, 2016 1:01 PM

Overmod,

I agree that the Fabyan Bridge wreck was due to deferred maintenance, but it was deferred maintenance of the very lag screw problem that I am addressing.  And the reason for the deferred maintenance was that the lag screw problem was unknown. The way I see it, it was the very same development as that which led to the Mosier wreck.  So all that I am suggesting is a new method to address the maintenance of the previously unknown lag screw problem that has now been revealed by the two recent wrecks.

I do not believe that sufficient friction between the wood tie and the tie plate can be created by the clamping pressure of the lag screws to the extent that it prohibits the shear movement of the tie plate.  I understand your point about a lag screw preload being lost by a slight relaxation of its connection to the tie. Probably just the shrinking and swelling of the tie from moisture changes could disrupt that preload.  So, from a practical standpoint, I doubt that any lag screws in use today have much if any preload on the tie plate (unless they use some type of compression spring).

In practice, it is the pin effect of the lag screw shanks that resists the shear force from the tie plates.  Photos clearly show the wear and upset of the screw shank right under the head, which is caused by the shear force of the tie plate. 

If this shear force wear and upset under the head were to continue, it would eventually shear or break right there.  But it seems that the shear force also bends the screw shank, eventually causing fatigue sufficient to cause the screw to break several inches below the application of the shear force. 

So it is the bending of the screw, caused by the shear force under the head, which has to be resisted.  Then the screw must be replaced before it breaks off under the head where the shear force is directly applied.    

The resisting of the bending of the screw can be achieved by increasing its contact area with the tie along the shank of the screw.  Perhaps this could be achieved by adding a steel bushing to the screw, extending down an inch or two below the head of the screw.  Then the screw hole pre-drilled in the tie would also require a counter-bore for the bushing. 

  • Member since
    July 2006
  • 9,610 posts
Posted by schlimm on Friday, July 8, 2016 11:29 AM

Overmod
The 'failure' of the system at Fabyan Bridge is as part of an almost inconceivable amount of deferred maintenance, not just of the screws themselves but of the ties and other elements of the trackwork.

Symptomatic of the need for systematic change.

C&NW, CA&E, MILW, CGW and IC fan

  • Member since
    September 2003
  • 21,669 posts
Posted by Overmod on Friday, July 8, 2016 10:30 AM

Euclid
Therefore, the key to living with lag screws is to monitor life cycle of these lag screw systems. The entire lag screw system must be replaced at specific intervals before it reaches its failure point.

I am proceeding on very different assumptions.  The 'failure' of the system at Fabyan Bridge is as part of an almost inconceivable amount of deferred maintenance, not just of the screws themselves but of the ties and other elements of the trackwork.

As I see it, the principal issue of the screws is twofold, with one of the more significant implications arising, initially peripherally but with severe delayed impact, from those initial problems.  I also think that most of the actual 'failure' is catastrophic, resulting from large forces not calculated in the design development acting on a stretch of track that might only slightly (but, as it turns out, critically) be compromised.

One is that, as noted, even a slight relaxing of effective clamping tension produces an ability for the plate to shift laterally, imposing bending moment and perhaps other forms of strain on the bolts.  The other is that, once the clamping force has been compromised by many means, there is no assured method of restoring it before damage to the bolts commences.

The secondary problems are stress raising and corrosion of the bolts, breakage (in the zones noted), and the inability of current inspections to determine any of the conditions -- lack of effective clamping, wastage or corrosion in the installed bolts, or even the presence of broken bolts or 'bolt killing' in the upper part of the tie hole. 

Many of these can be addressed with better approaches, most of which in my opinion can be automated 'enough' to incorporate them effectively into a workable system of periodic inspection and remediation.  For example, it should not be that difficult to adapt the existing equipment that picks up pulled spikes for "recycling" (phased electromagnets around the rim of large wheels) to exert sufficient vertical force on the heads of the bolts to extract the 'broken' ones, and then use machine vision to coordinate the 'empty holes' with following attempts to remove clip and plate, remove the remaining broken section, repair and rebore the tie holes, and retension the replacement screws. 

I think the correct method for initial tensioning is to preload the screw in the last couple of inches of driving, before turning it to contact.  It follows that a maintenance approach might be to repeat that preload again for 'installed' screws, monitoring torque, acoustics, etc. to get a sense of the condition of screws that may be wasted, working in killed holes, etc.

Much of the approach derives from methods I already worked out for more permanent track structure, including concrete ties with plugs for lag screws.

Spikes, even 5 and 6 spikes per plate, were expensively removed in order to install the spring-clip lag-screw system.  It follows that re-installing them is a poor option if a means of remediating the lag-screw issues becomes practical.

There is a problem with 'recording working load that the track is subjected to over time' -- it is prohibitive to do proactively, and already understood if conducted via instrumented test.  Note that the factual equivalent of this -- the use of the geometry train with high lateral loading to measure both lateral excursion and other geometric movements -- is already being conducted at regular intervals (and I suspect building a higher-speed version of this testing, or even arranging enough 'windows' in traffic to permit it more frequently, would be far less expensive than rebuilding All That Track With Crummy Spikes and then doing the more frequent maintenance that such an approach would entail.  I was interested to find that some versions of the Pandrol-clip plates were being used with spike fixation; I can't really understand how such a thing would work long-term even with fairly frequent periodic redriving.

  • Member since
    January 2014
  • 8,221 posts
Posted by Euclid on Friday, July 8, 2016 9:15 AM

I believe that the key to solving this lag screw problem is not to revert back to conventional cut spikes.  Nor is it to conduct frequent inspections a few days apart.  The proper inspections are complex, so performing them at such frequent intervals would be cost prohibitive. But the larger point is that frequent inspections are not necessary.  The issue here is not the occasional failure of a lag screw which might call for frequent inspections to catch. 

Instead, the issue is the single mass failure of an entire installation of a lag screw system installed at specific high stress sites such as curves.   The failure happens when the entire installation wears out.  That was the experience at both the CN wreck at Fabyan Bridge and the U.P. wreck at Mosier, OR. 

Therefore, the key to living with lag screws is to monitor life cycle of these lag screw systems.  The entire lag screw system must be replaced at specific intervals before it reaches its failure point.

A mass failure of a lag screw system ought to be accurately predictable based on the time span of its service and the force it has been subjected to.  It would require the recording of working load that the track is subjected to over time.  I don’t know to what extent this is currently done, or to what level of accuracy.  At the very least, it would require a record of the number of trains, their speed, and their tonnage. 

This would be coupled with forensic inspection and testing of the track involved in the Mosier derailment.  This would include the plotting of chronology of lag screw breakage, and the location of the individual screws in the entire system installed at that site.  It would model the total failure in as much detail as possible.  It would help predict the total life cycle and how it progressed toward failure. 

With all of this information, an accurate model of the life cycle of such screw installations could be developed.  Then an installation would be inspected in detail say every two years, to see if it is following its life cycle prediction. 

  • Member since
    January 2014
  • 8,221 posts
Posted by Euclid on Wednesday, July 6, 2016 6:41 PM

I wonder if CN removed all their lag screws on their lines, and replaced them with cut spikes after the 2012 Fabyan Bridge wreck; or did they adopt the inspection routine recommended by the TSB instead?

  • Member since
    December 2007
  • From: Southeast Michigan
  • 2,983 posts
Posted by Norm48327 on Wednesday, July 6, 2016 6:36 PM

Euclid
Stronger lag screws may require stronger ties.

Law of unintended consequences: Fix one problem, create two more.

Norm


  • Member since
    January 2014
  • 8,221 posts
Posted by Euclid on Wednesday, July 6, 2016 3:44 PM

It may be a small price to pay now after finding out about the problem, but still I wonder if it is really a small price.  In the cost/benefit analysis, the decision to use lag screws in the first place was to save money.  The TSB of Canada laid out a process for the inspection of lag screws after the CN discovery that they could break inside of the tie and give no outward visual indication.  A lot of that inspection process is conducted on foot, so I am not sure about the limits of track time.  In any case, the inspection will add cost.  Reverting back to the shorter-lived spikes will save money by eliminating the more complex inspections needed with lag screws; but it will also add cost in for form of more frequent track rebuilding.  A stronger lag screw may be the answer, but that will also add cost.  Stronger lag screws may require stronger ties. 

  • Member since
    May 2003
  • From: US
  • 25,292 posts
Posted by BaltACD on Wednesday, July 6, 2016 1:01 PM

Euclid

U.P. implies that they did not anticipate this lag screw danger, and that they did not inspect for it, nor see any evidence to indicate that inspection was necessary.  They do refer to the CN wreck at Fabyan Bridge as being the only known example of similar lag screw failure.

Given this sudden realization on the part of U.P., I am somewhat surprised that their immediate reaction is to replace all lag screws on curves with conventional cut spikes. 

That decision seems pre-mature to me.  I would expect their first action would be to closely inspect those lag screws, consider their history of time, tonnage, and speed, and develop a better picture of their performance over time.  They do say that the replacement will be carried out with their track renewal program, so maybe close inspections will be done immediately with actual replacement spread out over a long time. 

Incidentally, according to my calculation, this entails replacing 6,360,000 lag screws.  At say 2 lbs. per screw, that is 6,360 tons of lag screws. 

It is an interesting proposal considering that lag screws were chosen for curves because their holding power exceeded that of spikes.  The reason they give for the replacement, however, is that despite the compromise in holding power, spike failure gives immediate visual indication in the form of spikes having risen from their fully driven position. 

 

To my knowledge (and mudchicken is a much better source) there is no efficient method of inspecting lag screws.  Track time is always at a premium for MofW activities - any inspection activities must deal with the constraints of allowed track time.  The track time necessary to properly inspect lag screws to determine if they are still in proper working order isn't there on a continuing basis. 

Replacing 6400 tons of lag screw is a small price to pay to install a device that can be inspected in a efficient manner, a manner that has been and is being done today and since cut spikes were invented.

Never too old to have a happy childhood!

              

  • Member since
    January 2014
  • 8,221 posts
Posted by Euclid on Wednesday, July 6, 2016 10:46 AM

U.P. implies that they did not anticipate this lag screw danger, and that they did not inspect for it, nor see any evidence to indicate that inspection was necessary.  They do refer to the CN wreck at Fabyan Bridge as being the only known example of similar lag screw failure.

Given this sudden realization on the part of U.P., I am somewhat surprised that their immediate reaction is to replace all lag screws on curves with conventional cut spikes. 

That decision seems pre-mature to me.  I would expect their first action would be to closely inspect those lag screws, consider their history of time, tonnage, and speed, and develop a better picture of their performance over time.  They do say that the replacement will be carried out with their track renewal program, so maybe close inspections will be done immediately with actual replacement spread out over a long time. 

Incidentally, according to my calculation, this entails replacing 6,360,000 lag screws.  At say 2 lbs. per screw, that is 6,360 tons of lag screws. 

It is an interesting proposal considering that lag screws were chosen for curves because their holding power exceeded that of spikes.  The reason they give for the replacement, however, is that despite the compromise in holding power, spike failure gives immediate visual indication in the form of spikes having risen from their fully driven position. 

  • Member since
    January 2014
  • 8,221 posts
Posted by Euclid on Thursday, June 30, 2016 1:20 PM

tdmidget,

I can assure you that I know what shear force is, and it is applied exactly where I said it applies.  That is where the tie plate pushes against the shank right under the head, and the opposing resistance is the top of the hole in the tie. 

However, the bolt cannot shear there because the opposing resistance of the wood is not sufficient.  The wood is crushed by the bolt rather than providing enough resistance for the shifting tie plate to shear the bolt.  So the force of the tie plate bends the bolt rather than shearing it off under the head.   

There is no shearing action 2-3 inches deep in the wood tie where the bolt breaks.  And the bolt is not separated by shearing.  It is separating by breaking in the range of bending. 

Definition of Bow:  bend into the shape of a bow.

 
 

Join our Community!

Our community is FREE to join. To participate you must either login or register for an account.

Search the Community

Newsletter Sign-Up

By signing up you may also receive occasional reader surveys and special offers from Trains magazine.Please view our privacy policy