Capitol Limited & Lake Shore Limited Delays Summer/Fall 2014

[Barciur]

No. 29 (10) left WAS on time today at 4:05. The Amtrak desk at WAS assured my party that the train would not be stopped at Toledo and the passengers bused onward, but would go straight through to CHI.

We'll see what happens now.

No. 30 (9) is more than 4 hours late now.

That might be true given the earlier talk on Trainorders AND the fact that 29 is NOT showing a service disruption.

 

[afigg]

No. 29 (10) left WAS on time today at 4:05. The Amtrak desk at WAS assured my party that the train would not be stopped at Toledo and the passengers bused onward, but would go straight through to CHI.

We'll see what happens now.

That is consistent with the ability to buy tickets on the reservation system. The bustituted trains were shown as Sold Out through yesterday for the westbound CL & LSL and tonight for the eastbound CL & LSL. As we know, the trains were not canceledl I figure Amtrak did not to subject any more passengers to the bustitution beyond those who had already brought tickets.
So with #29 and #49(10/10) enroute to Chicago, tune in early Saturday morning to see how late they get to CHI!

 

[MikeM]

I just checked online tonight to see what I could see on the Capitol Limited -- everything shows a service disruption. Also, when I first signed on, it was showing the reservation system was down, but it came up over a few minutes. Hopefully we don't have a crash like last weekend...still waiting with baited breath to find out if there's a bustitution tomorrow or not.

 

[Barciur]

Well current 29 is not showing as a service disruption so I'm with you - hoping for no bustitution! My trip on the 29 is in 3 weeks exactly!

 

[Henry Kisor]

Sadly, at 7:06 am CDT No. 29 (10) is overdue, 5 hours 26 minutes late before Cleveland. So much for Amtrak's promises.

 

[Henry Kisor]

At 8:37 a.m CDT No. 29 (10) still not into Cleveland . . . running 6:43 late and counting. My party aboard says it's due to freight congestion plus a medical emergency.

 

[Barciur]

Oh wow. Certainly not what I expected. 49 seems to be doing a little better..ish.

Back to square one and busses then?

 

[antiguru]

I'm on the 29, we're still building up delays. It just passed Cleveland international airport.

And now we're getting new conductors because their service time is over.

 

[Barciur]

Has anyone told you why you're delayed so much? And are you going all the way to Chicago or being put on buses somewhere?

 

[afigg]

Oh wow. Certainly not what I expected. 49 seems to be doing a little better..ish.

Back to square one and busses then?

CL #29(10/10) left ALC 9 minutes late and then took 8 hours and 24 minutes to get to CLE; departed CLE a whopping 7 hours and 21 minutes late. Buses between CHI and TOL were not going to help here.
#49(10) departed TOL 4 hours and 38 minutes late. Has the NS meltdown shifted to east of TOL and CLE?

 

[Barciur]

No, but that's not what I was getting at - the need for more crew changes means there is more issues and I do not even want to try to imagine what time today's 30 will leave. Terminating at Toledo could possibly help get crews more rest time and more time to turn the train etc.

 

[antiguru]

I think the reason is quite simple: first we were blocked by freight trains before and after Cleveland. Then we needed a crew change which blocked all traffic between Cleveland and elyria because there is construction work going on leaving only a single track open. Now we seem to be blocked by another train, we just got out of elyria to stop again. It was just confirmed we will not reach any connecting trains in Chicago. But everyone seems happy

 

[benale]

Wonder how long this nightmare is going to last? We are connecting with 7 from 29 a few months away. We planned to park in Toledo which has free parking. If things don't improve we may have to drive to Chicago and avoid 29 and 49 altogether. Does anyone know of any Amtrak stations around Chicago that offer free long term parking? We can't afford to miss our connection. The Cardinal is an option,but the three day a week schedule could be a problem.

 

[neroden]

Has the NS meltdown shifted to east of TOL and CLE?

Good question. NS has said for years that Englewood was the primary bottleneck on their network. Well, Englewood Flyover is open now. Having eliminated that bottleneck, the question is where the next bottleneck is.

NS is apparently using a new, stupid, incompetent computer dispatching system. I know more than a little about how those sort of badly designed systems get written. The programmers are told to try to achieve maximum throughput (which is WRONG design, but that's typically what they get told to do). Because the program can't possibly be informed of all sources of unexpected delays, this overloads the network. The overload will show up at the biggest bottleneck. The result is that mindlessly following the computer system, while it is very poor at actually doing the job, is very excellent at demonstrating to you where your biggest bottleneck is.... we should see very quickly, within a couple of weeks, what the weak point of the NS system is now.

It could easily be that the second-biggest bottleneck is between Cleveland and Ravenna. Why shouldn't it be? It's another area where too many of the former parallel routes were ripped out or downgraded.

 

[Barciur]

Looks like Amtrak got an extra consist in for Capitol Limited as 30 has departed while 29 is still en route..

 

[afigg]

Looks like Amtrak got an extra consist in for Capitol Limited as 30 has departed while 29 is still en route..

Yes, resulting in a rare sight of 3 CLs shown on Amtrak Status maps at the same time, 2 of them near CHI.

 

[neroden]

To continue on the topic of how difficult automated dispatching is as a problem, I'll share some wisdom from an expert in control theory who's related to me.

The biggest problem in designing a control system is getting the model right.

Elaboration: If your model is missing key pieces of information, you can write a computer program which will accurately do the right thing *for the model*, which could be catastrophically wrong *in the real world*. However, most people commissioning these systems do not check their model carefully enough. If you are employed to write such a system, you are well advised to take the model you are given and spend several weeks checking it to see where they've screwed up the modelling, before even trying to write the control program.

Likely places for screwup on the model for an automated dispatching system include: failure to account for the incidence of trespassing by cars bypassing gates and getting hit (and the complex knock-on effects this has), failure to account for grade crossing equipment failures, poor modeling of the other railroads with which you interchange (for instance, not realizing that you're big enough to congest *them*), poor modeling of shippers, failure to model hours-of-service requirements (and the knock-on effects of running out of hours), et cetera.

 

[Just-Thinking-51]

Might be off topic, still a good read.

 

[MikeM]

Well, tonight (10/11) Capitol Limited shows being late 30 minutes. Be interesting to see when it actually shows up. LSL seems to be about an hour late, but does not display service status on Track a train.

Today, saw the LSL coming through Holland OH at 11:01AM; normally it would be through here in the wee morning hours. Train had two private varnish, a dark red heavyweight, and a dome observation car. Wasn't close enough to get car information. One thing that would be a pain is trying to move a private car with all these bustitutions and delays in the system.

 

[Barciur]

Looking at 29 and 30 this morning it seems obvious that indeed the bottleneck has been moved to the east of Toledo, rather than the west. So bustitution will probably not be happening any longer as if you want to avoid the delays, you would have to run buses from, well, probably Pittsburgh for 29.

 

[neroden]

The reports I've been seeing are that the NS system is now bottlenecking between Cleveland and Alliance, particularly between Cleveland and Ravenna.

...so does this mean the LSL is running more-or-less on time again? That would be a major help for Amtrak, even if the CL is still suffering.

 

[afigg]

Looking at 29 and 30 this morning it seems obvious that indeed the bottleneck has been moved to the east of Toledo, rather than the west. So bustitution will probably not be happening any longer as if you want to avoid the delays, you would have to run buses from, well, probably Pittsburgh for 29.

Based on last night and this morning, the delays between TOL and CHI may be down to 1 to 2 hours. Not good, but not the extreme delays of late last week and last weekend that led to the busitutions. However, yes, the delays west of TOL can be split to between TOL to CLE (affecting both trains) and then for the CL, one more delay zone to get through, CLE - ALC.

Fortunately, Amtrak had a new CL consist ready to send eastward on Saturday night when #29 (10/10) arrived at CHI a staggering 11 hours and 1 minute late (for a 17 and half hour scheduled trip time). Gives the CL another day to hold off on the late arrival and later departure cycle at CHI. For #30(10/11), we are getting a snapshot of the current eastbound delays for a CL that departs CHI on-time (or 9 minutes late as in this case) as #30(10/11) arrived at PGH 5 hours and 44 minutes late. At least it is not 10 hours late. :wacko:

 

[Just-Thinking-51]

With back to the wall. Amtrak push out two complete train sets. One for the Empire Builder, now one for the Capital. Glad to see all that federal tiger rebuild funds in action.

 

[afigg]

With back to the wall. Amtrak push out two complete train sets. One for the Empire Builder, now one for the Capital. Glad to see all that federal tiger rebuild funds in action.

The extra #30 CL consist sent out on Saturday night may have been a 1 time deal with days to assemble a new consist at CHI so the inbound equipment could be sent off for maintenance. We may find out tonight (Sunday) if there is late departure of #30.

Anyway, the westbound trains made it to CHI, but still pretty late with the LSL ending up right behind the CL (again) and both losing less than an hour TOL to CHI.:

CL #29(10/11) departed WAS on-time, PGH 4 minutes late, ALC 25 minutes late, TOL 3 hours and 45 minutes late, and arrived CHI 4 hours and 28 minutes late at 1:13 PM CT.

LSL #49(10/11) departed NYP and ALB on-time, BUF 1 hour and 22 minutes late, TOL 3 hours and 16 minutes late, arrived CHI 3 hours and 37 minutes late at 1:22 PM CT.

So what is the next step in this sequence for the departures from CHI? Stay tuned to the hit season series: the Norfolk Southern meltdown!

 

[PaulM]

Likely places for screwup on the model for an automated dispatching system include: failure to account for the incidence of trespassing by cars bypassing gates and getting hit (and the complex knock-on effects this has), failure to account for grade crossing equipment failures, poor modeling of the other railroads with which you interchange (for instance, not realizing that you're big enough to congest *them*), poor modeling of shippers, failure to model hours-of-service requirements (and the knock-on effects of running out of hours), et cetera.

I'm not following here. It seems to me that most of the factors you mention would affect capacity decisions, not real time dispatching. Wouldn't a dispatching system be determanistic, i.e., use the current status of the system, i.e., where each train is, current speed limits, train speed capability, hours of service remaining, etc., plus priorities for each train? If train X is 5 hours late, does it really matter if it was caused by a trespasser?.

When you say "automated" dispatching, do you mean the computer sets the lights and switches (e.g., for a siding), or it tells the human dispatcher what to do, or it shows the effects of what a dispatcher wants to do?

I can't picture a real time dispatching system that would take into account the probability of a grade crossing equipment failure or a locomotive failure down the line. Hopefully these events are extremely rare.