I’ve pulled the R1 reports here but the language on Sch 755 is a bit odd.
Using CSX in 2017 for an example and it shows CSX had a total of 67,619,832 train miles (per line 2-04). Now the weird part is down below in sections 9-01, 9-02, and 10.
9.1: Train Hours - Road Service - 3,032,384
9.2: Train Hours - Train Switching - 939,862
10: Total Yard-Switching Hours - 2,057,406
The instructions says Train Switching hours in 9.2 are “Train switching is the time spent by the train while performing switching service at terminals and way stations where no regular yard service is maintained.” and that 9.2 hours are included in 9.1 hours. That makes sense - but the Total Yard-Switching Hours in section 10 says these are hours expended in yards that maintain regular switching services. Are sec 9.2 hours included in 10 hours?
If I divide train miles by 9.1 hours I get an average speed of 22.3 mph. If I divide train miles by 9.1 less 9.2 hours I get an average of 32.3 mph. And if I divide train miles by 9.1 hours less sec 10 hours I get average speed of 69.4 mph. Those are all for 2017 - in 2016 those numbers are 18.6, 25.5 and 45.6 mph which sounds more reasonable. The combined numbers for all Class 1 railroads are 20.0, 22.9, and 34.0 mph which, again, sound reasonable.
Anyway, apologies in advance for a very obscure question but since you folks are such an incredibly knowledable lot I thought I would toss this query into the ether. Could the CSX 2017 numbers just be an anomaly? Does 34.0 mph average speed for class 1 trains while moving between origine and destination, not counting terminal time, sound right? And then terminal time brings that down from 34.0 to 20.0 mph?
Bonus Question: CSX had 3,078 freight locomotives in service in 2017.
By implication from the changes CSX made to its metrics, the AAR figures don’t include “intermediate dwell” and therefore Hunter’s change “has the effect of reducing speed when compared to the prior methodology.”
For the week ending the 23rd, train velocity was a blistering 18.9mph, and terminal dwell was 10.1 hours, but the latter figure is a fairly meaningless average without standard-deviation information I’m not going to be bothered to calculate (as CSX provides the relevant dwell for particular terminals further down the page).
Minor pause for railfan nostalgia moment: Cumbo and Willard are taken off the top-ten list of terminals … and Jax and Rocky Mount substituted, reflecting “changes to the volumes handled at terminals following the implementation of [Precision Scheduled Railroading].”
Thanks to all for the link to the AAR data. I downloaded the CSV and looked at average overall train speed over the past 53 weeks for BNSF, CN, KCS, NS, and UP and then compared those numbers to the R-1 data. I was hoping to find a consistant matchup with one of the formulas I used but, alas, that’s not the case and I’m left more confused as ever. Per the AAR website these railroads should all be calculating the average speed the same way and obviously the Surface Transportation Board expects the railroads to fill out the R1 data in the same way. So where’s the disconnect?
Using 9.01 hours less 9.02 switching hours was closest to the mark with 3/5 railroads but that’s only 60% and what’s worse is the other two railroads each went the other direction. I highlighted the cell in green that was closest to the AAR stated average speed.
