posted
I checked on the status for train #1 arriving Los Angeles on June 9:
Estimated arrival: 11:56 pm
Estimated arrival: 13 hours and 46 minutes late.
As of the last report at 8:18 pm at Deming, NM (DEM), it was running 52 minutes late.
Posts: 286 | From: Knee deep in the retention tank | Registered: Jan 2006
| IP: Logged |
posted
As you can tell, the internet status page is not so hot at extreme distances out. Figure out where the train is and get the most recent station departure status report. The same applies for Droolie.
Posts: 391 | From: Schenectady | Registered: Jan 2002
| IP: Logged |
posted
That is the first time I have seen it actually estimate delays...Usually, it estimates a train will make up time and it does not happen. It is much nicer to see some honesty.
Posts: 1082 | From: Los Angeles, CA. USA | Registered: Aug 2003
| IP: Logged |
posted
I think what the system does (normally) is keep track of arrivals for the past X trips, then average them. I've seen it predict the Chief as arriving early (even though it's running a tad late). However, I've seen it predict the Starlate and Sunset coming in 1-2 hours later than they currently are. e.g. if the train is running 2 hours late, it will predict it arriving 3-4 hours late.
Posts: 286 | From: Knee deep in the retention tank | Registered: Jan 2006
| IP: Logged |
posted
I believe the "train status" section of the AMTRAK web site is a bit more hi-tech than you think, Paul -- the way I understand it is that it is based on actual "running times" between station A and station B -- these are known quantities based on track speed limits. Sure, trains get delayed by freights, mechanical problems, and whatever else, but when the system reports a train departed station X and 10:40 AM, I believe the system knows how long it NORMALLY (with NO delays) takes to get from station X to whatever station you are asking about, calculates the running time, and posts that -- whether the system programs in other factors on a given segment of a line (i.e. freights, double-track, single-track, etc.,) I do not know.
But I believe there is more of a "method to the madness" than just basing it on "averaging the arrival times for the last X days."
Posts: 2428 | From: Grayling, MI | Registered: Mar 2002
| IP: Logged |
posted
What I meant was the "average arrival time" is included along with the "normal" running times. To calculate an estimated arrival, the system would compute:
This is why I think certain trains, such as the Southwest Chief, are predicted to arrive early, even when they are running a tad late 1000 miles away. The system includes some "record" of previous performance along with the actual running times.
Posts: 286 | From: Knee deep in the retention tank | Registered: Jan 2006
| IP: Logged |
posted
For trains which are late and are predicted to arrive less late, that can simply be explained by the known padding time in the schedules. This is certainly the case in most rail computer systems I've seen. Of course, the fatal flaw in this logic is that "a late train gets later" - often it's more likely to lose time than to make it up.
Predicting an on-time train arriving late is probably an input of predicted delays due to scheduled maintenance or congestion. The BNSF certainly had this system in trial a few years ago, where it would predict the meets several hours in advance, and thus come up with predicted arrival times for any train.
Geoff M.
-------------------- Geoff M. Posts: 2426 | From: Apple Valley, CA | Registered: Sep 2000
| IP: Logged |