CSX Hump Closing Questions

[overall]

 

Hunter Harrison has stated his belief that hump yards are not economical if they had a through put of less than 1400 cars per day. I am going on memory with that figure of 1400. Someone may correct me on that.  Anyway, he has acted on that belief by closing humps all over the CSX system. NS has also closed their hump at Chattanooga, TN. Here is my question; How is the breakpoint of economical vs uneconomical through put arrived at? In other words, what is Harrison basing his belief on? Is there a formula or an article in an industry publication somewhere?

 

[MP173]

Great question...I have asked several times what the economic break point is for a hump yard. The magic number these days seems to be around 1200 - 1400.

I would like to know the costs involved with a hump yard...the only answer I have seen is that the electronics and the retarders are expensive to maintain.  

Obviously the freight flows and the volumes dictate the costs involved with sorting the cars (flat switching vs hump).  If a train can be built from origin to destination without adding or deleting blocks en route, that is obviously the most efficient method...but those types end points do not exist often (if at all).

How do railroads determine the functionability of a hump yard (vs flat)?

Ed

[tree68]

MP173

I would like to know the costs involved with a hump yard...the only answer I have seen is that the electronics and the retarders are expensive to maintain.  

I should think that time has a value as well.  The question being how much time a car spends in a hump yard vs a flat-switched yard.  This has to assume that the rest of the handling remains the same - ie, a car arrives in a train that needs to be sorted.  Dwell time is probably one measure of that, but there may be other factors involved with that measurement.  

There is the question of blocking - does a humped car sometimes require several trips over the hump in order to be blocked with other cars for the same destination(s)?   Clearly, the cars are sorted into some sort of "togetherness," but how far will that carry - the next yard, possibly several yards beyond?

And this would lead to another question, does being humped in one yard beget the need to be humped in the next?

[BaltACD]

Humping is a function of the overall Operating Plan.  That plan takes into consideration the switching abilities of car's origins as well as the switching requirements of car's destinations.  With tweaks to the Plan, a hump can be made to be busy or idle.  Where does 'the one' making The Plan want the 'savings' to show.

The railroad version of Three Card Monty!  A sucker every time!

[Saturnalia]

From what Harrison has written and done in the past, it is much more about reducing dwell than cost, at least from the PSR standpoint. Hump Yards can be dwell hogs. 

Hump yards basically build blocks the same way as any other yard, just with a slow consistent push over a hill instead of the more traditional kicking. Yes it can go faster, but it really doesn't cut many costs except that you'd need far more crews to flat-switch 2000 cars than hump 2000. 

And often forgotten function of switching yards is the car department. They're generally pretty quiet and us railfans don't notice them much. But they've gotta go around the bleed the air on every inbound car, and then somebody - either the car department or outbound crew - must Class I airtest the new trains. For those who don't know, a Class I airtest is a time-consuming process which requires a full set and release of the brakes, with each car visually inspected at both set and release to ensure proper functionality of the brakes. 

So those people are expensive: labor is hardly ever cheap. It also takes time. 

So Harrison's solution, is to cut down you car handling time to a minimum. So long as cars are kept on air (and ground-supplied air counts) for at least 4 hours and certain other rules are followed, such as the handoff of the air slip, then nobody needs to bleed the cars off or retest them. You just couldn't do this with individual cars, but you can with blocks. 

So the idea is that Train A brings in a bunch of blocks, and sets them out. Yard air is supplied to keep the train pumped. Then, when Train B arrives to pickup that cut. Since everything has been on air since its last Class I test, they just test to make sure the brakes set and release - verified usually by the already-active EOT (so long as the new block isn't going on the rear), and then they're off. 

Humping and kicking cars requires the air to be bled from the cars: meaning anytime that this is done requires a Class I test, before the car can leave the yard again. 

Cut down your handlings and time spent, and you should in theory save money by virtue of less labor required and less time required, saving your labor, facility and car-hire costs. 

Whether or not EHH was too aggressive in his implementation of this rationale is pretty clear, how it gets out of it remains to be seen with much clarity, but it is currently heading in the right direction. 

[MidlandMike]

My understanding is that a hump yard still takes a lot of personnel to operate.  There are the hump pusher engineers, hump operator, pin puller, retarder maintainers; and still in the bowl the cars are scattered up and down the tracks, so there are bowl trimmer crews and crews to gather up the tracks and build trains.  These are all fixed costs, and if the hump runs low on cars to sort, I guess the bean-counters go crazy.

[BaltACD]

EHH is gum numbing with double talk, as long as his oxygen tank holds out.  When you devise new ways to count things you come up with 'new' ideas.  Ideas that aren't worth the new ways of counting.

[zugmann]

MidlandMike

My understanding is that a hump yard still takes a lot of personnel to operate. There are the hump pusher engineers, hump operator, pin puller, retarder maintainers; and still in the bowl the cars are scattered up and down the tracks, so there are bowl trimmer crews and crews to gather up the tracks and build trains. These are all fixed costs, and if the hump runs low on cars to sort, I guess the bean-counters go crazy.

 

But if you want to block swap, that just means you need more crews further up the pipeline to sort cars into blocks to begin with.  Save $5 here, spend $5 there. 

[Saturnalia]

zugmann

 

 

MidlandMike

My understanding is that a hump yard still takes a lot of personnel to operate. There are the hump pusher engineers, hump operator, pin puller, retarder maintainers; and still in the bowl the cars are scattered up and down the tracks, so there are bowl trimmer crews and crews to gather up the tracks and build trains. These are all fixed costs, and if the hump runs low on cars to sort, I guess the bean-counters go crazy.

 

 

 

But if you want to block swap, that just means you need more crews further up the pipeline to sort cars into blocks to begin with.  Save $5 here, spend $5 there. 

 

Maybe, but it's when you bypass yards entirely that's where the savings come from. Look at Willard: they aren't closing the hump because they think they can kick it all...they're closing it because they've made huge cuts in the amount of traffic demanding to be humped/switched there by blocking at the origins. 

So if you cut the number of terminal handlings - wholesale classification, flat or hump - from 5 to 3 over a car's trip, that is real savings, in theory. 

[oltmannd]

Saturnalia

Maybe, but it's when you bypass yards entirely that's where the savings come from.

Yes, but this has been Railroading 101 for quite a while now.  I'm certain a hard push for more block swapping can move the needle a bit, but most of the low hanging fruit has been harvested quite a while ago.

 

[zugmann]

Saturnalia

So if you cut the number of terminal handlings - wholesale classification, flat or hump - from 5 to 3 over a car's trip, that is real savings, in theory.

You're still classifiying cars.  Just at differnet points in the journey.  If you have cars from 6 terminals, that's 6 more terminals doing extra blocking.  May require more crews and a few more engines at those places, as opposed to one hump crew and a pusher engine (and maybe some trimmers and pullers) here.  I'm sure in some situations there's savings.   But other times, I'm sure there isn't.   Numbers just moved to a different budget sheet. 

[Saturnalia]

zugmann

 

 

Saturnalia

So if you cut the number of terminal handlings - wholesale classification, flat or hump - from 5 to 3 over a car's trip, that is real savings, in theory.

 

 

You're still classifiying cars.  Just at differnet points in the journey.  If you have cars from 6 terminals, that's 6 more terminals doing extra blocking.  May require more crews and a few more engines at those places, as opposed to one hump crew and a pusher engine (and maybe some trimmers and pullers) here.  I'm sure in some situations there's savings.   But other times, I'm sure there isn't.   Numbers just moved to a different budget sheet. 

 

Are you sure about that? Obviously there are cases where that is true, but what of, on Harrison's scale at CSX, you're halving the number of in-trip classifications? 

Let's take three railcars, all from Grand Rapids, MI (GRP), a medium-sized flat yard. One is going to Baltimore, one to Selkirk, NY and the other to the Union Pacific by way of the BRC. All three cars are coming off of Marquette Rail, who drops their cars in CSX's yard. 

Prior to Harrison's arrival, pretty much every car went over every hump it encountered. Blocking just wasn't done. 

So Car1, going to Baltimore, is brought into GRP, where somebody bleeds off the air before it is flat-switched. Before Harrison, pretty much it was a question of "to Chicago, or to Toledo". So it goes to Toledo. Then it gets its Class I and is taken to Stanley Yard in Toledo, where it is again bled, humped, tested, and then sent to Willard, where it is done again. And then probably again at Cumberland. There's a chance there were more efficient blocks out of Stanley, but from what I've heard there's a decent chance said car would be pushed over three humps, two of them in Ohio. 

Today, Car1 will be blocked for Willard, where it will be reclassed, probably for Baltimore, bypassing Stanley and Cumberland. Now it is classed twice, not 4 times. 

Car2, same story. Based on routing, it woulda been humped at Stanley, Willard and probably Selkirk, too. Now GRP has a direct train to Selkirk, so it'd stay in its block the whole way, getting classed once in GRP and then again at Selkirk. Four handlings to two, once more

Car3 would have previously gone to Barr Yard mixed in with everything else, most likely, as blocking separetely between Barr and the BRC was reportedly not done much if at all, before Harrison. Now that traffic is blocked, so in theory when it gets to Barr they can take that BRC block directly to the Belt, instead of kicking them out in Barr. Two classifications to one. 

So I just just don't see how this is cost-shifting. For the additional price of having GRP class into about a dozen blocks instead of two (which doesn't require that much more time, because you're already kicking out the same cuts to begin with), you get to skip about half of all down-line classifications on the system. It appears as though the plan is to get every car on CSX blocked at the origin or on-lining point, and then touch it once, maybe twice, before its destination terminal, for a savings of half to a third of total classifications. 

Correct me if I'm wrong, but to me it seems clear that if this sort of scheme is followed, it isn't cost shifting, since you're eliminating the costs of all of those enroute classifications, not shifting them somewhere else. 

[BaltACD]

Saturnalia

 

zugmann

 

Saturnalia

So if you cut the number of terminal handlings - wholesale classification, flat or hump - from 5 to 3 over a car's trip, that is real savings, in theory. 

You're still classifiying cars.  Just at differnet points in the journey.  If you have cars from 6 terminals, that's 6 more terminals doing extra blocking.  May require more crews and a few more engines at those places, as opposed to one hump crew and a pusher engine (and maybe some trimmers and pullers) here.  I'm sure in some situations there's savings.   But other times, I'm sure there isn't.   Numbers just moved to a different budget sheet.  

Are you sure about that? Obviously there are cases where that is true, but what of, on Harrison's scale at CSX, you're halving the number of in-trip classifications? 

Let's take three railcars, all from Grand Rapids, MI (GRP), a medium-sized flat yard. One is going to Baltimore, one to Selkirk, NY and the other to the Union Pacific by way of the BRC. All three cars are coming off of Marquette Rail, who drops their cars in CSX's yard. 

Prior to Harrison's arrival, pretty much every car went over every hump it encountered. Blocking just wasn't done. 

So Car1, going to Baltimore, is brought into GRP, where somebody bleeds off the air before it is flat-switched. Before Harrison, pretty much it was a question of "to Chicago, or to Toledo". So it goes to Toledo. Then it gets its Class I and is taken to Stanley Yard in Toledo, where it is again bled, humped, tested, and then sent to Willard, where it is done again. And then probably again at Cumberland. There's a chance there were more efficient blocks out of Stanley, but from what I've heard there's a decent chance said car would be pushed over three humps, two of them in Ohio. 

Today, Car1 will be blocked for Willard, where it will be reclassed, probably for Baltimore, bypassing Stanley and Cumberland. Now it is classed twice, not 4 times. 

Car2, same story. Based on routing, it woulda been humped at Stanley, Willard and probably Selkirk, too. Now GRP has a direct train to Selkirk, so it'd stay in its block the whole way, getting classed once in GRP and then again at Selkirk. Four handlings to two, once more

Car3 would have previously gone to Barr Yard mixed in with everything else, most likely, as blocking separetely between Barr and the BRC was reportedly not done much if at all, before Harrison. Now that traffic is blocked, so in theory when it gets to Barr they can take that BRC block directly to the Belt, instead of kicking them out in Barr. Two classifications to one. 

So I just just don't see how this is cost-shifting. For the additional price of having GRP class into about a dozen blocks instead of two (which doesn't require that much more time, because you're already kicking out the same cuts to begin with), you get to skip about half of all down-line classifications on the system. It appears as though the plan is to get every car on CSX blocked at the origin or on-lining point, and then touch it once, maybe twice, before its destination terminal, for a savings of half to a third of total classifications. 

Correct me if I'm wrong, but to me it seems clear that if this sort of scheme is followed, it isn't cost shifting, since you're eliminating the costs of all of those enroute classifications, not shifting them somewhere else. 

The Operating Plan defines car handling - Flat Switching or Humping.

So I will throw the BS Flag on your examples.

[Saturnalia]

BaltACD

The Operating Plan defines car handling - Flat Switching or Humping.

So I will throw the BS Flag on your examples.

Hmmm...not sure how that statement allows you to throw the "BS flag" here. 

What if the Operating Plan defines car handling as blocking out of GRP, with block-swapping along the way, bypassing classification at the humps mentioned in the examples? 

I can promise you that traffic from GRP goes directly to either Detroit, Willard, Selkirk, Columbus, Cincinatti, Walbridge (and maybe a couple more) going east, as blocks. They've completely cut-out at *least* Stanley from every single route going east of Grand Rapids, so only cars local to Toledo need to be classed in Toledo, versus literally everything before. 

Then everything going at least as far as Cinci, Columbus, Selkirk bypasses Stanley and also potentially Willard. Not 100% sure on the routings before all of the changes began. Certainly, if there was a direct Baltimore-Toledo freight before Harrison, then that car probably wasn't humped at Willard and Cumberland, and a GRP-Selkirk car could bypass Willard on a direct train, if one existed.

But, the examples prove how terribly inefficient some of CSX's car routings were before Harrison, in terms of leaving a fair amount to cut by blocking aggressively. 

I'll admit that the examples were worst-case-senarios. But it is still clear that the new operating creed reduces the number of classifications per car on the network, on average, and probably almost never increasing it. 

[BaltACD]

Saturnalia

 

BaltACD

The Operating Plan defines car handling - Flat Switching or Humping.

So I will throw the BS Flag on your examples. 

I'll admit that the examples were worst-case-senarios. But it is still clear that the new operating creed reduces the number of classifications per car on the network, on average, and probably almost never increasing it. 

Your example can be perform with or with hump yard use.  Any thing you can do in a flat yard can be done more efficiently in a hump yard.  Even block swapping can be done in hump yards - not every car MUST go over the hump.  The operating plan defines how much traffic each yard will handle.  The efficiency of each terminal defines its throughput.  There are efficient humps as well as humps that are less than efficient.  There are flat yards that are more efficient than other flat yards.  

Management by decree rarely works on the railroad.

[Saturnalia]

BaltACD

 

Your example can be perform with or with hump yard use.  Any thing you can do in a flat yard can be done more efficiently in a hump yard.  Even block swapping can be done in hump yards - not every car MUST go over the hump.  The operating plan defines how much traffic each yard will handle.  The efficiency of each terminal defines its throughput.  There are efficient humps as well as humps that are less than efficient.  There are flat yards that are more efficient than other flat yards.  

Management by decree rarely works on the railroad.

I see your point. It doesn't seem to have much bearing on my point however, since I was discussing the number of classification moves in total. Yes, the classifications at Stanley, Willard, Cumberland and Selkirk in my examples were humps, but Barr was not. 

Certainly as you say here is correct, I think we're trying to have the same discussion about different points...which just doesn't work! haha

But yes, I agree with your sentiments about hump vs flat. 

[BaltACD]

Saturnalia

 

BaltACD 

Your example can be perform with or with hump yard use.  Any thing you can do in a flat yard can be done more efficiently in a hump yard.  Even block swapping can be done in hump yards - not every car MUST go over the hump.  The operating plan defines how much traffic each yard will handle.  The efficiency of each terminal defines its throughput.  There are efficient humps as well as humps that are less than efficient.  There are flat yards that are more efficient than other flat yards.  

Management by decree rarely works on the railroad. 

I see your point. It doesn't seem to have much bearing on my point however, since I was discussing the number of classification moves in total. Yes, the classifications at Stanley, Willard, Cumberland and Selkirk in my examples were humps, but Barr was not. 

Certainly as you say here is correct, I think we're trying to have the same discussion about different points...which just doesn't work! haha

But yes, I agree with your sentiments about hump vs flat. 

The amount of switching is set by the operating plan - not the number of hump yards.  Anyone that gears their Operating Plan for more the minimum amount of intermediate switching has a fool creating their plan.

[oltmannd]

BaltACD

The amount of switching is set by the operating plan - not the number of hump yards.  Anyone that gears their Operating Plan for more the minimum amount of intermediate switching has a fool creating their plan.

 

+1.  It's a key metric analyzed every time to operating plan is tweaked or reworked...using the network modeling tools.