Question(s) re Scheduled Trains

[SALfan]

From what I've read, CN and NS (and probably others) have moved to a policy of scheduling as many trains as possible to maximize the efficiency of yards and minimize the number of crews needed (and probably other reasons). Are there other reasons to work this way? Is this working? What do the railroads do, have a cutoff time at a major terminal and at that time the train for Wherever leaves, regardless of the number of cars on it and any cars headed toward the terminal? What kind of research/planning/changes did it take to institute the practice? All info welcome - I'm having trouble understanding how well (if?) it works in the real world.

[Junctionfan]

I can't answer about NS but only CN in my area. They don't seem to follow any schedule anymore and seem to just sprout like mushrooms when ever they are ready and dispatch clears them. I partly blame (well not really blame) dispatch for CSX Buffalo because most of the CN trains originating from Buffalo, come from the CSX Frontier Yard-plus I would imagine the CN trains are subjected to any kind of delay by the train master.

Slow orders also have made the trains less accurate on the schedule and so has the St.Lawrence Seaway (Welland Canal) which CN crosses a set of 2 lift bridges.

CN actually has or still is having crew shortages and power shortages as well.

[Randy Stahl]

It's a marketing gimmic more than any thing. A good sales tool when you can predict for a shipper actual transit times.
Randy

[MP173]

Mark:

This sounds like a great topic for your monthly column, or better yet an article, or even better yet...a series of articles.

As I previously have posted, I became intrigued by the "scheduled railroad" terminology earlier this summer and began tracking times that certain trains passed thru my area.

I guess I am old school. To me, a schedule is a time plan. I recall as a youth collecting timetables and Official Guides and there were schedules that were published. While I dont expect freight to run on passenger schedule type accuracy, I would think within 4 hours would be normal, perhaps 2 hours.

Comments?

MP

[Junctionfan]

CN rarely follows the 4 hours or less schedule except for the Triple Crown roadrailers. Trains 148 and 149 tend to be on schedule but not always.

[oltmannd]

A "Scheduled Railroad" on the NS means operating all the trains in the plan from origin to destination, making all intermediate pickups and set offs, each and every day. And, these trains are to operate on the times published in the operating plan. SInce the beginning of the Thoroughbred Operating Plan (TOP), this has been the goal. There has been some minor backsliding from time to time trying to save crew starts, but each time it's been shown that the crew savings are more than offset by increased equipment costs and locomotive cycles being broken.

It also means that cars make their scheduled connections at intermediate yards. (this is why you have intermediate yards - to make connections - not to have low dwell. Handling a car once with 24 hrs dwell versus twice with 12 hrs dwell each seems like a wash excpect all that extra motion has cost and represents another potential point of failure. Avg dwell, in part, is a function of your train service plan and is a second order metric to connections)

Adherence to the train plan, train on time performance at origin and destination, and connection performance are all monitored very closely and in great detail.

The "bottom line" is overall shipment performance and shipment consistency versus plan and these are measured and tracked, as well.

It has been a rather large shift in how things done. A huge change from the way things were in 1998, but it's been rather successful.

[oltmannd]

QUOTE: Originally posted by MP173

Mark:

This sounds like a great topic for your monthly column, or better yet an article, or even better yet...a series of articles.

As I previously have posted, I became intrigued by the "scheduled railroad" terminology earlier this summer and began tracking times that certain trains passed thru my area.

I guess I am old school. To me, a schedule is a time plan. I recall as a youth collecting timetables and Official Guides and there were schedules that were published. While I dont expect freight to run on passenger schedule type accuracy, I would think within 4 hours would be normal, perhaps 2 hours.

Comments?

MP

How does 50% at destination +/- 2hrs from plan sound for manifest (non intermodal) road trains sound?

[SALfan]

Oltmannd-
Thanks for the info. Could you elaborate a bit on how it was a big change from the way things were done in 1998? Also, how were the decisions made on when to run each train?

This may show my ignorance, but I assume by making connections you mean arriving in a connecting RR's yard in time to make it on that RR's train toward the car's ultimate destination as well as getting to an intermediate yard on NS and being dropped off in time to get on an NS train heading in a different direction. Correct? If NS picks up a car in, say, Baltimore, headed for Atlanta, would that car be put on a westbound to be dropped in Hagerstown, MD, then picked up by a train headed down the Valley Line toward points south?

Many thanks.

[MP173]

Don:

Thanks for the response. I listen in on the NS and it is very easy to monitor their operations as their crews call out each signal, with the name of the train.

NS's performance on this line is pretty damned good, in my opinion. I dont know how close to being "on time" they are, but their consistancy is outstanding.

After a couple of months of writing down train times (Ok, so my life seems to be boring...not really my girlfriend is kinda into my statistical analysis) I can almost tell by what time it is when I hear a NS whistle. Coal trains are the exception, but their
intermodals are really tightly scheduled. The manifests are a little looser, but still pretty consistant.

I remember back in the 90's the NS train 412 would always go thru town at 830pm. Could set a watch to it...well maybe not a watch, but certainly a sundial.

I too would be interested to know how the TOPS program differs from earlier system.

MP

[oltmannd]

"Ad-hocing" is a long standing tradition that came about from the focus on train productivity, which is fairly easy to measure, rather than network efficiency, which is much harder to grasp. Getting managers to focus on "running the plan" instead of looking for combo/annul opportunities was difficult. It is particularly difficult when the budget isn't alligned with the plan, i.e. there aren't enough budgeted train starts to cover the operating plan.

The other issue is the fixed plan has to be an efficient plan. One way to do this is to use computer traffic flow models to help generate an optimized plan. Part of NS's TOP plan goal was to take the original post Conrail plan, which was largely created by hand and reaplace it with an optimized one.

Another issue is how to deal with seasonal or business related traffic flow changes. You need good computer modeling tools intergrated with your car reporting and waybilling systems to be able to analyze and make changes to the plan in a matter of weeks rather doing it blindly or taking months to generate and feed off-the-shelf traffic modeling tools.

[oltmannd]

QUOTE: Originally posted by JOdom

Oltmannd-
Thanks for the info. Could you elaborate a bit on how it was a big change from the way things were done in 1998? Also, how were the decisions made on when to run each train?

This may show my ignorance, but I assume by making connections you mean arriving in a connecting RR's yard in time to make it on that RR's train toward the car's ultimate destination as well as getting to an intermediate yard on NS and being dropped off in time to get on an NS train heading in a different direction. Correct? If NS picks up a car in, say, Baltimore, headed for Atlanta, would that car be put on a westbound to be dropped in Hagerstown, MD, then picked up by a train headed down the Valley Line toward points south?

Many thanks.

Any time a car moves from one train to another, it is a connection. For your example, the plan might work as follows.

Car blocked into Enola block at Baltimore Bayview. Car moves on train carrying Enola block to Enola. Car humped into Linwood block and picked up by train carrying Linwood block. For the example, lets assume this train goes to Roanoke, not Linwood, and there is a block swap at Hagerstown to a train that will take the Linwood block to Linwood. At Linwood, the car would be humped into an Atanta block and carried to Atlanta in a train carrying an Atlanta block. Each change of train is a connection and each time, NS measures what acutally happens versus the plan. For connections to be made, trains have to arrive on time and the yard has to have a work plan that will get the outbound blocks built on time to make the planned outbound train.

The difference between "then" and "now" on NS, is that "then", if there were only 40 cars cars for the Enola-Roanoke train today, they would have annulled the train or combined it with another one, so the cars would either lay another day at Enola or perhaps, wind up at Roanoke rather than Linwood. You save a train start, which is easy to count $$ but push volume bulges into the network, delay traffic, incur more car hire costs and wreak havoc on any central planning of power distribution - all of which are much harder to count.

Also, in the past, the yards and terminals may have had work plans that minimized cost and maximized convenience, such as putting several locals in on top of each other on a receiving track at a hump yard and then humping the whole thing at once. It's simple and efficient to do it this way, but many cars on the first-to-arrival local will miss planned outbound connections.

[SALfan]

oltmannd-
Thanks for the additional info. I think I'm finally beginning to understand (sometimes a good indication of danger ahead, LOL). Your explanation of what might happen in my example helped a lot, but in my ignorance it sure sounds like a lot of times to handle a car or a block. Assuming there is enough traffic to justify an Enola-Atlanta train every day, would a block be dispatched toward Atlanta as in your description every few hours or would Enola build a once-a-day train for Atlanta? Or, would Enola build a train or trains containing everything to be humped at Linwood?

I guess I'm trying to get a handle on the thinking that goes into deciding which yards will build which trains for which destinations. I assume that would depend on how much traffic goes thru a yard or hump, where those cars are going and what yards/humps are in the general direction of the car's destination, correct?

[oltmannd]

QUOTE: Originally posted by JOdom

oltmannd-
Thanks for the additional info. I think I'm finally beginning to understand (sometimes a good indication of danger ahead, LOL). Your explanation of what might happen in my example helped a lot, but in my ignorance it sure sounds like a lot of times to handle a car or a block. Assuming there is enough traffic to justify an Enola-Atlanta train every day, would a block be dispatched toward Atlanta as in your description every few hours or would Enola build a once-a-day train for Atlanta? Or, would Enola build a train or trains containing everything to be humped at Linwood?

I guess I'm trying to get a handle on the thinking that goes into deciding which yards will build which trains for which destinations. I assume that would depend on how much traffic goes thru a yard or hump, where those cars are going and what yards/humps are in the general direction of the car's destination, correct?

You are thinking along the right lines. In general, you configure your train and blocking plan to minimize the number of handlings per shipment and to concentrate handlings as lower cost facilities. Less activity = less cost. The traffic flow models help you figure out the details and determine which scenarios are most advantageous. If there was sufficient volume, there probably would be an Enola to Atlanta train, but it would likely carry blocks to intermediate locations like Linwood, as well.

Generally, there is only one departures a day for "road train" blocks from the major terminals. At some locations, there are two departures a day, but this is less common.

[Overmod]

Don --

This is an article, perhaps a series of articles, for Trains Magazine.

50% deliveries plus or minus 2 hours is, to put it bluntly, amazing. Is this percentage mostly intermodals (and if so, is it to final destination or pre-arragned pickup) or does it contain switched-out car deliveries>

Is there a graph or histogram that indicates how close to the 2-hour window the remaining 50% is, or was -- and how much of that traffic was NOT time-critical-window deliveries?

Is there a way to quantify the amount of "slack" (in Hirschman's sense) that is kept in the plan to allow for logical slow-orders, emergency requirements, etc.?

[dldance]

Since my consulting clients in another industry often ask scheduling related questions, I have found this thread very informative. Any comments about the traffic flow analysis software?

dd

[jeaton]

Don,

Your information on the TOP system has been very enlightening.

There are many stories of the railroads of earlier times running "schedules", i.e., trains on a timetable, specific plans for handling blocks and more precise instructions for handling hot freight. For the last, auto parts and perishables come to mind. However, it would be my guess that even those railroads given high marks for their service still handled a substantial portion of their loads on an "ad hoc" basis.

In light of the size and complexity of the today's Class 1 railroads, just producing an operating plan that goes to the "car" level is a remarkable acheivment. And once that is done, there is still the need for operating people to buy into the program even if certain parts might counter local intuition and then the process of monitoring actual results against plan. That is some job.

Jay

[oltmannd]

QUOTE: Originally posted by Overmod

Don --

This is an article, perhaps a series of articles, for Trains Magazine.

50% deliveries plus or minus 2 hours is, to put it bluntly, amazing. Is this percentage mostly intermodals (and if so, is it to final destination or pre-arragned pickup) or does it contain switched-out car deliveries>

Is there a graph or histogram that indicates how close to the 2-hour window the remaining 50% is, or was -- and how much of that traffic was NOT time-critical-window deliveries?

Is there a way to quantify the amount of "slack" (in Hirschman's sense) that is kept in the plan to allow for logical slow-orders, emergency requirements, etc.?

No. The 50% number is merchandise road train arrivals at train destination - and does not include intermodal trains, which perform much better, in general. Shipment performance is different. That gets measured against the ETA generated when the RR pulls or receives the car. That's more like 80% +/- a day.

Train schedule slack is still more of an art than a science. 15% of the "unopposed running time" is a rough number. Actual practice is to measure actual vs sched for each piece of the route and then adjust the worst offenders accordingly.

[oltmannd]

QUOTE: Originally posted by dldance

Since my consulting clients in another industry often ask scheduling related questions, I have found this thread very informative. Any comments about the traffic flow analysis software?

dd

Multirail is sort of the industry's defacto standard. TOP was originally developed with Multirail but is now supported by an in-house system that is fed directly with NS data without having to massage and clean it up which allows much faster analysis cycles.