Amtrak: Charlotte to Atlanta

 I turned my Train Performance Calculator to passenger trains.  The Crescent.  A way-too-slow, runs where the people are when they are sleeping, train. 5:40 minutes for 260 miles from Charlotte to Atlanta.  Very not useful. How can it be fixed?  Let's take a look...

The route is chocked full of curves.  54 miles of the 257 mile route are 2 degree or sharper.  About half of that is 3 degrees or sharper.  It seems that 3 degrees was likely the design standard for the route.  NS has maintained 5 inches of superelevation in most places to keep the track speed up to 60 mph or higher as much as possible. 

Miles of curves between Charlotte and Atlanta by curvature (degrees)

There are so many curves, so closely spaced, it's rather impractical to speed up much between them and gain much reduction in running time.  In fact, the timetable actually limits the majority of the route to 60 mph, the max for 3 degree curves with 3 inches of underbalance.  Amtrak equipment is generally allowed 3" of underbalance in curves - slight centripetal force greater than the superelvation.

Boosting the horsepower through additional locomotives or electrification improves running time only a couple of minutes.

What does help is being able to run through those three degree curves faster.  Let's try a tilt train.

Talgo equipment has been tested to 8 inches of underbalance.  Other countries run tilt equipment from 7-9 inches.  So, I chose 8 inches. For reference, 

Increasing the under balance to 8" allows 80 mph on three degree curves with 5" of superelevation knocks over an hour off the running time.

Increasing track speed to 90 mph helps running time a bit, but involves nearly constant acceleration and braking.  Not very practical.  What would make 90 mph practical would be easing all curves over 2.2 degrees.  Added with tilt equipment, another 20 minutes could be saved.

Here's a summary of scenarios.

Some detail from the scenarios:

This is just a piece of the route, but it is very typical.  This is "as constructed" in the 19th century when the south didn't have the wealth to build to higher standard and there was no money subsequently to modernize.  

The sheer number and spacing of the three degree curves make running much faster than 60 mph an impracticality.  If you use tilt equipment and allow 8" underbalance, you can get to 80 mph without too much trouble.  But if you want more than that, you get into easing curves.  There are more than 50 miles of curves that would need attention. That's a very expensive proposition and not one a handful of Amtrak trains could justify.  

Conclusion?  Tilt equipment.  Getting the running time from Atlanta to Charlotte down under 4 hours makes a New York to Atlanta day train possible.  If similar improvements can be made along the North Carolina RR the the Durham to Petersburg HrSR route created, multiple NY to Atlanta trains become practical.  

For freight?  Electrification.  It's the roller coaster grades of this route that kill the average speed.  Electrification brings much higher HP and the potential for much higher average speeds.  See previous blog post.

Ultimately, this route's engineering is just so bad, it might only be suitable for freight and a separate HSR rail route might be needed for passenger.  The state of GA wasted some big money on a study of alternatives about a decade ago (https://www.dot.ga.gov/GDOT/pages/AtlantaCharlotteRail.aspx.)  

However, we'll never see any HSR construction if we can't show conventional passenger trains are attractive and the current schedules are not competitive.

Tilt trains are the practical answer.

I built a train performance calculator

 Yes.  I did.  In Excel.  Piecewise in 1/10th mile increments.

I've always wanted to try to do it.  ...because I am a RR geek.  

But, also, because I've made a lot of noise on the interwebs about "railroads should electrify" among other things, and I need to put up or shut up.

The verdict?  Stay tuned.

But, first, some "how I did it" stuff.

For train resistance, I found a 1992 CN paper that had some Davis equation work that had been validated. So that went in.

For grade resistance, I just used the standard of 20 lbs per ton per percent.

I ignored curve resistance.  I didn't check to see if grades were curve compensated or not.  As curvy as this line is, I'm not sure it would make much of a difference, although it would be easy enough to add in.  Something to save for later.

For tractive force, I used the old HP = force x speed/308 equation with max of 35% adhesion.  Yeah, I know 308 isn't always the right number to use for all locomotives, but is close enough for my purposes.

I had it check to make sure the whole train was through a lower speed spot before speeding up.

I set max deceleration based on dynamic braking with 20% adhesion, but I can adjust it to simulate train braking or coasting or anything in-between. 

I piece-wise integrated the train resistance to calculate the energy in motoring and braking.  Simulating dynamic braking as the maximum, I can assume electric locomotives can regenerate/recover.

One tricky part was using distance increments.  That meant the time calculation was a quadratic equation. (ew!)

The tough part was manually entering grade and curve and timetable speed.  The ETT and track chart data came from multimodalways.org.  Want to do 260 miles of RR?  That's 2600 lines in Excel.

I did it so that I could play with all sorts of variables.  Locos per train, Electric locomotives.  Recovering braking energy, Easing curves.  Flattening grades.  Increasing speeds. Tilting passenger equipment.

For my first - and so far only - territory I chose Charlotte to Atlanta.  Why?  Because it's terrible.  For a non-mountain railroad, it has a terrible alignment.  Roller coaster profile,  Three degree curves as apparent design standard.  Used to be double track, now single with "sidings".  Used to have a lot of spring switch equilaterals at double to single transitions, but most are now regular #20s.  I assume no diverging routes with speed restrictions.

Atlanta to Greenville is a crew district. 150 miles.  Sounds okay, but this route hosts some "hot" intermodal traffic.  That's not very far compared to districts on the former NYC and PRR.

Here's a sample. 

What did I find out?

Well, for starters, IT WORKS!

It produces reasonable speed profile for the route profile.  This is for a 100 car train with 95 gross tons per car, and 9500 total tons.  NS's allowable tonnage for this route in this direction with this power is 9800 tons.

Some typical trains in the past decade or so.

Norcross, GA

Buford, GA

Duluth, GA

So, for some comparisons.

(Diesel fuel at $3/gallon, Electricity at 11 cents/KWHR.  35% efficiency for diesels, 90% for electrics) 

Some interesting results.  

First, DPUs save time, and fuel!  Why?  Because the longer trains effectively smooth out the roller coaster grades, with fewer times the whole train is entirely going up grade or down grade. The shorter trains use more braking per car going down grade to limit to track speed, only to have to use more on the next up grade.  The longer train has more time where the rear of the train is helping push the front up the next grade.

Here's a zoom in on part of the route, comparing the 100 car train with the 300 car train.  Arrows show approximate train length.

There are a few practical considerations for long DPU trains on the is territory.  One, is the time and location to build such a monster.  The yards on NS aren't generally configured to build such long trains. They'd have to hang out on the main tracks while doubling and tripling out.  The second is the time to "commission" the DPU system, linking three - or more- DPU sets.  Third, is radio continuity around hills and curve on the route.  Would stationary repeaters be needed?   Fourth, what would be the train consist and train handling considerations need to be?

Now, let's look at electrification.

This route is not all double track, but I doubt that would change the $4M/mile cost much.  A lot has to be infrastructure to feed catenary.

I assumed that the costs for locomotives and locomotive maintenance would wash, in the long term.  

"Savings" here include the indirect (for now) cost of CO2.  There are economic estimates that $75 a ton is what's needed to get to carbon neutral by the 2050 deadline.  It's a real cost, but whether it's a direct tax or provided as cost saving incentive for construction matter little in the end. (for comparison, a $75/ton tax would equate to 70 cents a gallon for gasoline)  

Obviously, the more traffic you have, the faster the payback for construction.  This is not nearly as busy a mainline as some on NS.  I assumed 30 trains a day, which is probably a bit high.  This route is part of NS's route from Harrisburg to Atlanta (and beyond) that parallels I-81 and I-85.  These are the two of the largest untapped intermodal conversion markets left in the country.  

If the entire route was electrified, the running times would decrease and crew districts could expand.  If electrification saves over 2 hours on 259 miles, it could save 6-7 hours between Harrisburg and Atlanta.  Atlanta to Charlotte becomes a practical long pool crew and the route could be covered by three or four crews.   This makes rail intermodal more cost and time competitive, increasing the traffic and justifying the investment.

So, should railroads hang catenary on the heavy mainlines?  Yes.

Next up, Amtrak.  Is there any hope for Amtrak on this route?