Freight car distribution


jim_mischke <jmischke@...>
 

It's been said in the model press that half of the cars on railroad
at any one time are its own. One fourth its direct connections. One
fourth all others. A rough rule of thumb. I would agree, plus or
minus 10%.

So much depends on region and traffic base.

And odd factors as below.

Example: Friend of mine has wheel reports for the B&O Buffalo
Division for a decade stretch. It shows many northbound empty
Canadian boxcars, only a few southbound wood and paper loads. In
the old days, lumber was shipped before sold and then brokered
enroute. Eastbound by a leisurely route until diverted. Return by
original route is hard to trace for each car, circuitous and
undesireable in any event. My speculation is that the B&O served as
a sump for expediting empty boxcars back to Canada for much of the
eastern region, many of which were not B&O loads. Only some boxcars
from a railroad that connected with B&O directly (only CN).

Example: The B&O Buffalo Division served the parafin-based petroleum
fields (great for lube oil) and connected with the other railroads
who also did. Much of this traffic was in tank cars, with each
refinery and chemical plant having its own lease contacts for tank
cars (SHPX, GATX, UTLX, etc.) One would have to know what tank cars
each plant used to have an accurate tank car mix.

Example: Most coal traffic from the B&O Indiana (Pa.) branch went
north of Punxsutawney to customers up north and in Canada. Very
little went south. Not only a shortage of customers south but bridge
limitations to heavy tonnage drags. So you can model this line and
exactly where you do makes a huge difference in traffic and car
distribution.

Example: To protect their car supply, certain railroads subscribed
to an interchange rule that mandated their empty cars could not be
loaded on the return trip without their express consent. This was
C&O, Virginian, N&W, and L&N, to a lesser extent C&EI, WM. You would
see these cars on railroad like B&O in due course but hardly ever as
empties delivered to B&O served mines. (Note that WM had trackage
rights to mines on B&O, plenty of WM hoppers in that service).

Example: In 1965 (out of list era, sorry), the J&L Pittsburgh Works
began taking iron ore deliveries from a new iron mine at North Adams,
Ontario, on the Ontario Northland. Routing was Ontario Northland -
CN - Canada Southern - P&LE. Each railroad contributed cars in
proportion to the mileage. This influenced car distribution seen
from trackside.


Researching this is a lot of fun. No pat answers.







--- In STMFC@yahoogroups.com, "armprem1" <armprem@...> wrote:

I would like to reopen the dialogue on freight car distribution.I
am
aware that to some, it may have been discussed ad nauseam.I firmly
believe that we tend see more freight cars within their own region
than by size.What is your opinion?Armand Premo


laramielarry <ostresh@...>
 

Hi Folks

Several people in this thread have commented about the nature of a
random distribution of cars, and how it might differ from a
proportional distribution. For those with access to Excel, it is a
simple matter to generate car lists chosen at random from a universe
of cars representative of the fleet at any desired time. (Actually,
they are chosen pseudo-randomly.)

For example, here are three randomly generated lists of 40 cars each,
in which the universe of cars is an approximation of the U.S. boxcar
fleet in 1949 (from Tim Gilbert's Excel file
"4060totalboxcarUSownerhip.xls", available in the files section of the
STMFC).

The first column is the railroad; the second column is the number of
cars Excel chose at random from that railroad; and the third column
is the expected number of cars if the choice had been proportional
rather than random.

Example 1:
RR, Random, Proportional
PRR, 5, 4
C&NW, 4, 1
NYC, 4, 4
C&O, 3, 1
CRI&P, 3, 1
AT&SF, 2, 2
GTW, 2, 0
MILW, 2, 2
ACL, 1, 1
CB&Q, 1, 2
CGA, 1, 0
DRG&W, 1, 0
GN, 1, 1
IC, 1, 1
L&N, 1, 1
LV, 1, 0
NH, 1, 0
NP, 1, 1
SAL, 1, 1
SLSF, 1, 1
SP-Pac, 1, 2
T&P, 1, 0
UP, 1, 1
In this example, 5 of the 40 randomly chosen boxcars were PRR,
whereas we would expect 4, had the choice been proportional. There
were 4 C&NW cars – only 1 was expected. We would not have expected
any cars from the GTW, but we got 2.

Example 2:
RR, Random, Proportional
PRR, 6, 4
ACL, 3, 1
C&NW, 3, 1
AT&SF, 2, 2
DRG&W, 2, 0
L&N, 2, 1
NP, 2, 1
P&LE, 2, 0
SOU, 2, 2
SP-Pac, 2, 2
B&O, 1, 2
C&O, 1, 1
CB&Q, 1, 2
CGW, 1, 0
DL&W, 1, 0
GN, 1, 1
LV, 1, 0
MILW, 1, 2
MKT, 1, 0
NYC, 1, 4
RDG, 1, 0
SAL, 1, 1
UP, 1, 1
WAB, 1, 1

Example 3:
RR, Random, Proportional
NYC, 6, 4
CB&Q, 5, 2
PRR, 5, 4
NP, 3, 1
SOU, 3, 2
C&O, 2, 1
IC, 2, 1
AT&SF, 1, 2
B&M, 1, 0
C&EI, 1, 0
C&NW, 1, 1
CNJ, 1, 0
DRG&W, 1, 0
ERIE, 1, 1
MP, 1, 1
N&W, 1, 1
NKP, 1, 1
RDG, 1, 0
SAL, 1, 1
SOO, 1, 1
SP-Pac, 1, 2

If the list is re-generated a sufficient number of times, presumably
Mike Brock's SP boxcars will constitute the entire train; some other
iteration will consist only of cars from the WP; yet another will be
in exact proportion to the U.S. fleet. (All such instances would be
rare, of course.)

If anyone wants a copy of this "random train" Excel file (with some
rather minimalist instructions), contact me off-list. It is easily
modified to accommodate any universe one wishes, and any desired
number of cars for the train. For example, I used each box car
series from the July 1950 ORER as a universe from which to create
lists of 100 cars. The lists will be one of several factors I'll
consider while I construct my model railroad fleet.

Best wishes,
Larry Ostresh
Laramie, Wyoming


cvsne <cvsne@...>
 

Larry,

Interesting set of numbers.

All:

I finally managed to locate the car count of boxcars northbound out of White River Junction
on the Central Vermont for September, 1955.

In Shared Files section for anyone who cares enough to peruse.

Marty


Stokes John
 

All very interesting. It seems to me that unless you are modeling a specific day and time period, using actual data of consists, you are essentially chasing your tail. Yes, you can generalize about percentages and link with the roads with interchanges, but when it comes down to it anything you do otherwise is a guess and an approximation. As someone else pointed out, the sheer number of railroads, cars, types of cars etc., etc., makes it impossible to really know what happened on average or over time or anything else than hard facts on a sheet of paper, or good photographic evidence of the train in question. Yes there are trends, but to say that one would never see such and such a box car on this line, or that you should have 29% SP box cars (what kind, what series of boxes?), on the UP in Utah during the period modeled is just playing around with reality and not coming close. I guess I am reacting to the run-on posts that go over the same ground and parse out the same basic erroneous facts and try to make some absolute system of the mess. how me where I am off base, otherwise I think most folks will just throw up their hands and say, "I'll run what I think looks good and I like" and the small minority of a hobby within a hobby will spend their precious modeling time pursuing the Holy Grail that does not exist, but if they are having fun, that's fine with me. It is just boring and for the main, unproductive.

Just my observations,

John Stokes
Bellevue WA



To: STMFC@yahoogroups.comFrom: cvsne@yahoo.comDate: Wed, 13 Aug 2008 22:00:47 +0000Subject: [STMFC] Re: Freight car distribution




Larry, Interesting set of numbers. All:I finally managed to locate the car count of boxcars northbound out of White River Junction on the Central Vermont for September, 1955. In Shared Files section for anyone who cares enough to peruse. Marty


Anthony Thompson <thompson@...>
 

jim_mischke wrote:
It's been said in the model press that half of the cars on railroad at any one time are its own. One fourth its direct connections. One fourth all others. A rough rule of thumb. I would agree, plus or minus 10%.
To assert this, without regard to car type, is silly (unless you aren't reading Mr. Gilbert's results). It is entirely true, of course, that this "formula" has been touted in the model press, but unfortunately that doesn't make it true, even as a rough rule.
That's my two cents.

Tony Thompson Editor, Signature Press, Berkeley, CA
2906 Forest Ave., Berkeley, CA 94705 www.signaturepress.com
(510) 540-6538; fax, (510) 540-1937; e-mail, thompson@signaturepress.com
Publishers of books on railroad history


Tim O'Connor
 

Larry wrote

Several people in this thread have commented about the nature of a
random distribution of cars, and how it might differ from a
proportional distribution. For those with access to Excel, it is a
simple matter to generate car lists chosen at random from a universe
of cars representative of the fleet at any desired time. (Actually,
they are chosen pseudo-randomly.)

One must be very careful with computer "random numbers". I
once wrote a simple program to find good "seeds" to start a
random number generator -- which produces a value of 0 or 1.
My criteria was that in any sequence of 100, there had to be
at least 45 of each value (i.e. 46 out of 100 was ok, 39 was
not) and that in a sequence of 1000, there had to be 475 to
525 of either value, or for 10000, it had to be 4950 to 5050.
I ran this with thousands of seeds, and found about 10 that
met the criteria. Therefore I had a "generator" that would
over a long period of time, produce a nearly balanced number
of 0's and 1's. This is what many people consider "random".

My switchlist software did not using random numbers by the
way. There were simply more cars than available assignments,
so there was a natural and inherent "rotation" involved in
the car selection process. Also some cars were programmed to
go 'offline' for longer periods of time than others -- this
is how I dealt with oddball or specialty cars.

Tim O'Connor


Walter M. Clark
 

Once upon a time the archive for the Freight Car List was at
http://sunny16.photo.tntech.edu/~richard/Freightcars/archive/
However I just tried that and got the old "404 not found" routine.
So, does anyone know where it went to, if it even is still available?

Time stopped in November 1941
Walter M. Clark
Pullman, Washington, USA

--- In STMFC@yahoogroups.com, "Kurt Laughlin" <fleeta@...> wrote:

I think they are referring to a previous USENET-based mailing list
whose
archives is out in the ether somewhere. I occaisionally run across
individual threads when doing a google search, but I haven't
stumbled onto a
reliable way to reference it as a whole.

KL

----- Original Message -----
From: Cyril and Lynn Durrenberger

You all keep saying to look at the past messages for some of this
data. I
found a number of them, but I did not locate any that had the work
that is
being sited. Could some one provide the message numbers to assist this
process.


Walter M. Clark
 

--- In STMFC@yahoogroups.com, Anthony Thompson <thompson@...> wrote:

Gene Green, perhaps tongue in cheek, wrote:
All this discussion of freight car distribution has what purpose? A
purpose does not seem self-evident to me.
Is there a model railroad, as opposed to model building, purpose?
Gene, I believe people are interested (perpetually) in knowing
the balance they need in freight car fleets--which I guess could be
called model building--but are also interested in accurate handling of
empty cars as well as of inbound loads from "off the layout."

Wouldn't the area one models and the industries on that model
railroad
play a greater role (than statistics) in determining which cars one
might model?
Of course. If you have a pickle plant on your layout, then
yes,
you should have pickle cars, regardless of their proportion of the
national car fleet. But most layouts model a certain amount of bridge
traffic, which of course isn't local, and moreover inbound loads need
to be in credible cars (car reporting marks). Tim Gilbert showed that
free-running cars, primarily box cars but also flat cars to a fair
extent, ARE in proportion to the national fleet. I would guess this is
true also for gondolas handling merchandise loads, but much less so for
bulk cargo--in the latter case, they are much more like hoppers and are
more nearly distributed by region.
But Gene's point is still vital to keep in mind, and one has
to know how one's prototype worked. SP lumber traffic was extremely
heavily on SP flat cars, because SP maintained a big fleet of those
cars for that traffic. Train photos show very few foreign flat cars
with lumber loads on the SP lines. That's an example of something a
modeler just needs to know.

Tony Thompson Editor, Signature Press, Berkeley, CA
2906 Forest Ave., Berkeley, CA 94705 www.signaturepress.com
(510) 540-6538; fax, (510) 540-1937; e-mail, thompson@...
Publishers of books on railroad history
And Tony's comment about SP flat cars means those of us who model the
SP pre-World War II are up the creek because there aren't any models
of pre-World War II SP flat cars (in HO scale that I know of, at least).

Time stopped in November 1941, which is my problem regarding SP flat
cars <g>
Walter M. Clark
Pullman, Washington, USA


Anthony Thompson <thompson@...>
 

Walter Clark wrote:
And Tony's comment about SP flat cars means those of us who model the SP pre-World War II are up the creek because there aren't any models of pre-World War II SP flat cars (in HO scale that I know of, at least).
So far, you're right. But straight-side-sill flats are not exactly the toughest scratchbuilding job there is.

Tony Thompson Editor, Signature Press, Berkeley, CA
2906 Forest Ave., Berkeley, CA 94705 www.signaturepress.com
(510) 540-6538; fax, (510) 540-1937; e-mail, thompson@signaturepress.com
Publishers of books on railroad history


laramielarry <ostresh@...>
 

Hi Folks

I ran my "random train" Excel program for 100,000 iterations to test
whether it was working OK. The average number of boxcars per train,
by railroad, seemed to properly mirror the national fleet as
represented by Tim Gilbert's data for 1949:

RR___Number___Nat'l %___Random %
ACL___13,831___1.99%___1.99%
AT&SF___35,234___5.08%___5.09%
B&M___2,755___0.40%___0.40%
B&O___28,683___4.13%___4.12%
C&EI___1,185___0.17%___0.17%
C&NW___23,284___3.36%___3.37%
C&O___21,699___3.13%___3.13%
CB&Q___31,634___4.56%___4.55%
CGA___5,153___0.74%___0.74%
CGW___3,622___0.52%___0.52%
CMO___3,422___0.49%___0.49%
CNJ___1,757___0.25%___0.26%
CNJ-CRP___1,247___0.18%___0.18%
CRI&P___17,160___2.47%___2.46%
D&H___2,469___0.36%___0.36%
DL&W___7,585___1.09%___1.09%
DM&IR___337___0.05%___0.05%
DRG&W___5,094___0.73%___0.74%
EJ&E___1,427___0.21%___0.21%
ERIE___12,070___1.74%___1.74%
GM&O___6,177___0.89%___0.89%
GN___22,399___3.23%___3.24%
GTW___8,556___1.23%___1.22%
IC___21,248___3.06%___3.08%
KCS___2,923___0.42%___0.42%
L&N___13,324___1.92%___1.92%
LV___5,584___0.80%___0.81%
MILW___32,076___4.62%___4.62%
MKT___5,235___0.75%___0.76%
MP___18,112___2.61%___2.61%
MP-IGN___2,572___0.37%___0.37%
N&W___8,986___1.30%___1.31%
NC&SL___3,603___0.52%___0.52%
NH___6,012___0.87%___0.86%
NKP___11,255___1.62%___1.63%
NP___19,317___2.78%___2.77%
NYC___66,867___9.64%___9.64%
P&LE___4,118___0.59%___0.59%
PRR___64,670___9.32%___9.33%
RDG___7,639___1.10%___1.11%
SAL___12,396___1.79%___1.79%
SLSF___13,255___1.91%___1.91%
SOO___8,728___1.26%___1.27%
SOU___26,331___3.80%___3.79%
SP-Pac___27,757___4.00%___4.01%
SP-T&NO___7,438___1.07%___1.07%
SSW___3,700___0.53%___0.54%
T&P___4,288___0.62%___0.61%
UP___23,758___3.42%___3.43%
WAB___11,572___1.67%___1.68%
WM___2,143___0.31%___0.31%
WP___1,995___0.29%___0.28%
Total___693,682___100.00%___100.00%

(The "Number" and "Nat'l %" columns above are from Tim's 1949 list of
boxcars. His data are at "4060totalboxcarsUSownership.xls" in the
files section of this list. The "Random %" column is the average
percentage of cars per train generated by my Excel program after
100,000 iterations. Each train consisted of 40 boxcars.)

While running the program, I tallied the maximum number of cars for
each railroad over all the iterations. In a 40 boxcar train, the
average number of cars would be 40 times the national percentage
shown in the above table. For example, the average number of SP-Pac
cars would 40 * 4% = 1.6 cars – 1 or 2 cars per train. Any
particular 40 boxcar train may have more or less SP-Pac cars (0 to
40). After 100,000 trials, there were in fact trains with no SP-Pac
boxcars, but no train had more than 9 of them. Here are the results
for all the railroads:

RR___Nat'l Avg___Max Rnd
ACL___0.80___6
AT&SF___2.03___9
B&M___0.16___3
B&O___1.65___10
C&EI___0.07___3
C&NW___1.34___9
C&O___1.25___8
CB&Q___1.82___9
CGA___0.30___5
CGW___0.21___4
CMO___0.20___4
CNJ___0.10___4
CNJ-CRP___0.07___3
CRI&P___0.99___7
D&H___0.14___4
DL&W___0.44___5
DM&IR___0.02___2
DRG&W___0.29___4
EJ&E___0.08___3
ERIE___0.70___7
GM&O___0.36___5
GN___1.29___8
GTW___0.49___5
IC___1.23___8
KCS___0.17___4
L&N___0.77___6
LV___0.32___4
MILW___1.85___9
MKT___0.30___5
MP___1.04___7
MP-IGN___0.15___4
N&W___0.52___6
NC&SL___0.21___4
NH___0.35___5
NKP___0.65___6
NP___1.11___8
NYC___3.86___14
P&LE___0.24___4
PRR___3.73___13
RDG___0.44___6
SAL___0.71___7
SLSF___0.76___7
SOO___0.50___5
SOU___1.52___9
SP-Pac___1.60___9
SP-T&NO___0.43___5
SSW___0.21___5
T&P___0.25___4
UP___1.37___8
WAB___0.67___6
WM___0.12___4
WP___0.12___3
Total___40.00___NA

("Nat'l Avg" is the average number of boxcars in a 40 car train based
on the national average; "Max Rnd" is the maximum number of boxcars
for any 40 car train after 100,000 iterations of the Excel program.)

Best wishes,
Larry Ostresh
Laramie, Wyoming


Tim O'Connor
 

Good work, Larry! That's what I'm talkin' 'bout!
A 40 car train with 9 SP (not counting T&NO) box
cars.. More than 5 times the 'average' value of
1.6. And the peak CMO value at 20 times its average
of 0.2! And yet, the OVERALL averages are exactly
what we expect, representing the fact that each car
has an equal chance.

Model railroad operations would be much improved if
someone could devise an easy way to do this using
paper car cards and waybills. (Since most people seem
to hate using computer switchlists.)

Tim O'Connor

At 8/17/2008 02:54 PM Sunday, you wrote:
Hi Folks

I ran my "random train" Excel program for 100,000 iterations to test
whether it was working OK. The average number of boxcars per train,
by railroad, seemed to properly mirror the national fleet as
represented by Tim Gilbert's data for 1949:

RR___Number___Nat'l %___Random %
ACL___13,831___1.99%___1.99%
AT&SF___35,234___5.08%___5.09%
B&M___2,755___0.40%___0.40%
B&O___28,683___4.13%___4.12%
C&EI___1,185___0.17%___0.17%
C&NW___23,284___3.36%___3.37%
C&O___21,699___3.13%___3.13%
CB&Q___31,634___4.56%___4.55%
CGA___5,153___0.74%___0.74%
CGW___3,622___0.52%___0.52%
CMO___3,422___0.49%___0.49%
CNJ___1,757___0.25%___0.26%
CNJ-CRP___1,247___0.18%___0.18%
CRI&P___17,160___2.47%___2.46%
D&H___2,469___0.36%___0.36%
DL&W___7,585___1.09%___1.09%
DM&IR___337___0.05%___0.05%
DRG&W___5,094___0.73%___0.74%
EJ&E___1,427___0.21%___0.21%
ERIE___12,070___1.74%___1.74%
GM&O___6,177___0.89%___0.89%
GN___22,399___3.23%___3.24%
GTW___8,556___1.23%___1.22%
IC___21,248___3.06%___3.08%
KCS___2,923___0.42%___0.42%
L&N___13,324___1.92%___1.92%
LV___5,584___0.80%___0.81%
MILW___32,076___4.62%___4.62%
MKT___5,235___0.75%___0.76%
MP___18,112___2.61%___2.61%
MP-IGN___2,572___0.37%___0.37%
N&W___8,986___1.30%___1.31%
NC&SL___3,603___0.52%___0.52%
NH___6,012___0.87%___0.86%
NKP___11,255___1.62%___1.63%
NP___19,317___2.78%___2.77%
NYC___66,867___9.64%___9.64%
P&LE___4,118___0.59%___0.59%
PRR___64,670___9.32%___9.33%
RDG___7,639___1.10%___1.11%
SAL___12,396___1.79%___1.79%
SLSF___13,255___1.91%___1.91%
SOO___8,728___1.26%___1.27%
SOU___26,331___3.80%___3.79%
SP-Pac___27,757___4.00%___4.01%
SP-T&NO___7,438___1.07%___1.07%
SSW___3,700___0.53%___0.54%
T&P___4,288___0.62%___0.61%
UP___23,758___3.42%___3.43%
WAB___11,572___1.67%___1.68%
WM___2,143___0.31%___0.31%
WP___1,995___0.29%___0.28%
Total___693,682___100.00%___100.00%

(The "Number" and "Nat'l %" columns above are from Tim's 1949 list of
boxcars. His data are at "4060totalboxcarsUSownership.xls" in the
files section of this list. The "Random %" column is the average
percentage of cars per train generated by my Excel program after
100,000 iterations. Each train consisted of 40 boxcars.)

While running the program, I tallied the maximum number of cars for
each railroad over all the iterations. In a 40 boxcar train, the
average number of cars would be 40 times the national percentage
shown in the above table. For example, the average number of SP-Pac
cars would 40 * 4% = 1.6 cars ­ 1 or 2 cars per train. Any
particular 40 boxcar train may have more or less SP-Pac cars (0 to
40). After 100,000 trials, there were in fact trains with no SP-Pac
boxcars, but no train had more than 9 of them. Here are the results
for all the railroads:

RR___Nat'l Avg___Max Rnd
ACL___0.80___6
AT&SF___2.03___9
B&M___0.16___3
B&O___1.65___10
C&EI___0.07___3
C&NW___1.34___9
C&O___1.25___8
CB&Q___1.82___9
CGA___0.30___5
CGW___0.21___4
CMO___0.20___4
CNJ___0.10___4
CNJ-CRP___0.07___3
CRI&P___0.99___7
D&H___0.14___4
DL&W___0.44___5
DM&IR___0.02___2
DRG&W___0.29___4
EJ&E___0.08___3
ERIE___0.70___7
GM&O___0.36___5
GN___1.29___8
GTW___0.49___5
IC___1.23___8
KCS___0.17___4
L&N___0.77___6
LV___0.32___4
MILW___1.85___9
MKT___0.30___5
MP___1.04___7
MP-IGN___0.15___4
N&W___0.52___6
NC&SL___0.21___4
NH___0.35___5
NKP___0.65___6
NP___1.11___8
NYC___3.86___14
P&LE___0.24___4
PRR___3.73___13
RDG___0.44___6
SAL___0.71___7
SLSF___0.76___7
SOO___0.50___5
SOU___1.52___9
SP-Pac___1.60___9
SP-T&NO___0.43___5
SSW___0.21___5
T&P___0.25___4
UP___1.37___8
WAB___0.67___6
WM___0.12___4
WP___0.12___3
Total___40.00___NA

("Nat'l Avg" is the average number of boxcars in a 40 car train based
on the national average; "Max Rnd" is the maximum number of boxcars
for any 40 car train after 100,000 iterations of the Excel program.)

Best wishes,
Larry Ostresh
Laramie, Wyoming


armprem
 

Unfortunately there are not enough different models available to even
begin to come close to "National Averages".Armand Premo

----- Original Message -----
From: "Tim O'Connor" <timboconnor@comcast.net>
To: <STMFC@yahoogroups.com>
Sent: Sunday, August 17, 2008 4:54 PM
Subject: [STMFC] Re: Freight car distribution



Good work, Larry! That's what I'm talkin' 'bout!
A 40 car train with 9 SP (not counting T&NO) box
cars.. More than 5 times the 'average' value of
1.6. And the peak CMO value at 20 times its average
of 0.2! And yet, the OVERALL averages are exactly
what we expect, representing the fact that each car
has an equal chance.

Model railroad operations would be much improved if
someone could devise an easy way to do this using
paper car cards and waybills. (Since most people seem
to hate using computer switchlists.)

Tim O'Connor


At 8/17/2008 02:54 PM Sunday, you wrote:
Hi Folks

I ran my "random train" Excel program for 100,000 iterations to test
whether it was working OK. The average number of boxcars per train,
by railroad, seemed to properly mirror the national fleet as
represented by Tim Gilbert's data for 1949:

RR___Number___Nat'l %___Random %
ACL___13,831___1.99%___1.99%
AT&SF___35,234___5.08%___5.09%
B&M___2,755___0.40%___0.40%
B&O___28,683___4.13%___4.12%
C&EI___1,185___0.17%___0.17%
C&NW___23,284___3.36%___3.37%
C&O___21,699___3.13%___3.13%
CB&Q___31,634___4.56%___4.55%
CGA___5,153___0.74%___0.74%
CGW___3,622___0.52%___0.52%
CMO___3,422___0.49%___0.49%
CNJ___1,757___0.25%___0.26%
CNJ-CRP___1,247___0.18%___0.18%
CRI&P___17,160___2.47%___2.46%
D&H___2,469___0.36%___0.36%
DL&W___7,585___1.09%___1.09%
DM&IR___337___0.05%___0.05%
DRG&W___5,094___0.73%___0.74%
EJ&E___1,427___0.21%___0.21%
ERIE___12,070___1.74%___1.74%
GM&O___6,177___0.89%___0.89%
GN___22,399___3.23%___3.24%
GTW___8,556___1.23%___1.22%
IC___21,248___3.06%___3.08%
KCS___2,923___0.42%___0.42%
L&N___13,324___1.92%___1.92%
LV___5,584___0.80%___0.81%
MILW___32,076___4.62%___4.62%
MKT___5,235___0.75%___0.76%
MP___18,112___2.61%___2.61%
MP-IGN___2,572___0.37%___0.37%
N&W___8,986___1.30%___1.31%
NC&SL___3,603___0.52%___0.52%
NH___6,012___0.87%___0.86%
NKP___11,255___1.62%___1.63%
NP___19,317___2.78%___2.77%
NYC___66,867___9.64%___9.64%
P&LE___4,118___0.59%___0.59%
PRR___64,670___9.32%___9.33%
RDG___7,639___1.10%___1.11%
SAL___12,396___1.79%___1.79%
SLSF___13,255___1.91%___1.91%
SOO___8,728___1.26%___1.27%
SOU___26,331___3.80%___3.79%
SP-Pac___27,757___4.00%___4.01%
SP-T&NO___7,438___1.07%___1.07%
SSW___3,700___0.53%___0.54%
T&P___4,288___0.62%___0.61%
UP___23,758___3.42%___3.43%
WAB___11,572___1.67%___1.68%
WM___2,143___0.31%___0.31%
WP___1,995___0.29%___0.28%
Total___693,682___100.00%___100.00%

(The "Number" and "Nat'l %" columns above are from Tim's 1949 list of
boxcars. His data are at "4060totalboxcarsUSownership.xls" in the
files section of this list. The "Random %" column is the average
percentage of cars per train generated by my Excel program after
100,000 iterations. Each train consisted of 40 boxcars.)

While running the program, I tallied the maximum number of cars for
each railroad over all the iterations. In a 40 boxcar train, the
average number of cars would be 40 times the national percentage
shown in the above table. For example, the average number of SP-Pac
cars would 40 * 4% = 1.6 cars � 1 or 2 cars per train. Any
particular 40 boxcar train may have more or less SP-Pac cars (0 to
40). After 100,000 trials, there were in fact trains with no SP-Pac
boxcars, but no train had more than 9 of them. Here are the results
for all the railroads:

RR___Nat'l Avg___Max Rnd
ACL___0.80___6
AT&SF___2.03___9
B&M___0.16___3
B&O___1.65___10
C&EI___0.07___3
C&NW___1.34___9
C&O___1.25___8
CB&Q___1.82___9
CGA___0.30___5
CGW___0.21___4
CMO___0.20___4
CNJ___0.10___4
CNJ-CRP___0.07___3
CRI&P___0.99___7
D&H___0.14___4
DL&W___0.44___5
DM&IR___0.02___2
DRG&W___0.29___4
EJ&E___0.08___3
ERIE___0.70___7
GM&O___0.36___5
GN___1.29___8
GTW___0.49___5
IC___1.23___8
KCS___0.17___4
L&N___0.77___6
LV___0.32___4
MILW___1.85___9
MKT___0.30___5
MP___1.04___7
MP-IGN___0.15___4
N&W___0.52___6
NC&SL___0.21___4
NH___0.35___5
NKP___0.65___6
NP___1.11___8
NYC___3.86___14
P&LE___0.24___4
PRR___3.73___13
RDG___0.44___6
SAL___0.71___7
SLSF___0.76___7
SOO___0.50___5
SOU___1.52___9
SP-Pac___1.60___9
SP-T&NO___0.43___5
SSW___0.21___5
T&P___0.25___4
UP___1.37___8
WAB___0.67___6
WM___0.12___4
WP___0.12___3
Total___40.00___NA

("Nat'l Avg" is the average number of boxcars in a 40 car train based
on the national average; "Max Rnd" is the maximum number of boxcars
for any 40 car train after 100,000 iterations of the Excel program.)

Best wishes,
Larry Ostresh
Laramie, Wyoming

------------------------------------

Yahoo! Groups Links





--
Internal Virus Database is out-of-date.
Checked by AVG Free Edition.
Version: 7.5.516 / Virus Database: 269.15.10/1091 - Release Date: 10/24/07
2:31 PM


Mike Brock <brockm@...>
 

Larry Ostresh writes:

"I ran my "random train" Excel program for 100,000 iterations to test
whether it was working OK. The average number of boxcars per train,
by railroad, seemed to properly mirror the national fleet as
represented by Tim Gilbert's data for 1949:"

Larry, could you explain what you are talking about? You show RR [ i.e., SP excluding T&NO ], number [ number of SP box cars { 27757 } ] Nat'l % [ you show 4.0% but Tim shoed 3.9% ] and Random % [ 4.01% ]. What is the derivation of the random %?

RR___Number___Nat'l %___Random %

SP-Pac___27,757___4.00%___4.01%

"(The "Number" and "Nat'l %" columns above are from Tim's 1949 list of
boxcars."

Yep.

"His data are at "4060totalboxcarsUSownership.xls" in the
files section of this list."

Yep.

"The "Random %" column is the average
percentage of cars per train generated by my Excel program after
100,000 iterations. Each train consisted of 40 boxcars.)"

Hmmm. Well, 40 box cars is OK...actually 39.

"While running the program, I tallied the maximum number of cars for
each railroad over all the iterations. In a 40 boxcar train, the
average number of cars would be 40 times the national percentage
shown in the above table."

Why do you say that?

"For example, the average number of SP-Pac
cars would 40 * 4% = 1.6 cars - 1 or 2 cars per train."

Well...actually that's not so at all. The correct number is 4. 136/34= If you use real...rather than theoretical data. IOW, the 4% [ actually 3.9% ] fails to produce the actual number of cars...136.

"Any
particular 40 boxcar train may have more or less SP-Pac cars (0 to
40)."

"After 100,000 trials, there were in fact trains with no SP-Pac
boxcars, but no train had more than 9 of them."

We know that there were 4 of 34 trains...11.76%...of the trains with far more than 9 SP bx cars.

The 1953 data shows the "infamous" train with 36 or more SP bx cars [ not T&NO ].

Mike Brock


Mike Brock <brockm@...>
 

Tim O'Connor writes:

"Good work, Larry! That's what I'm talkin' 'bout!
A 40 car train with 9 SP (not counting T&NO) box
cars.."

I assume you mean a train with 40 box cars. Actually the 1949 trains averaged 82.3 cars.

"More than 5 times the 'average' value of
1.6."

I'm not certain what the 1.6 represents. The actual number is 4.

Mike Brock...I never thought statistics could be fun.


Tim O'Connor
 

Mike Brock wrote

Larry, could you explain what you are talking about? You show RR [ i.e., SP
excluding T&NO ], number [ number of SP box cars { 27757 } ] Nat'l % [ you
show 4.0% but Tim shoed 3.9% ] and Random % [ 4.01% ]. What is the
derivation of the random %?
Mike, he ran 100,000 iterations of a 40 car box car train, composed
of cars chosen randomly from every possible prototype. The National
% is obvious, and then he finds the AVERAGE % found for that railroad
in the random consists. What is not obvious about this?


I tallied the maximum number of cars for each railroad over all the
iterations. In a 40 boxcar train, the average number of cars would be
40 times the national percentage shown in the above table."
Why do you say that?
Because he described it obscurely. He simply kept track of the largest
number of cars seen for each reporting mark in the 100,000 trains. Is
this not also obvious?


Well...actually that's not so at all. The correct number is 4. 136/34
If you use real...rather than theoretical data.
Real what? 136 box cars? 136 trains? 136 days? 136 train consists?

Larry's example was PURELY theoretical, and he made an excellent point.
I guess some people just don't get it.


We know that there were 4 of 34 trains...11.76%...of the trains with
far more than 9 SP bx cars.
Mike, you are having trouble grasping the idea. Larry has presented
100,00 theoretical trains, of fixed length, of random composition.
You persist with your miniscule sample of 34 actual trains of
varying length and composition of different sorts of cars (not just
box cars).


The 1953 data shows the "infamous" train with 36 or more SP box cars
[not T&NO ].
Yes, but I will argue that the Overland Route for the UP and SP was
a 'home road extension' in that each road was highly dependent on the
other for traffic. Therefore we can discount the presence of SP as
well as UP cars in all trains over Sherman Hill, for the purposes of
this discussion. This is exactly the "skew" that each layout has to
consider.

Tim O'Connor


Stokes John
 

This is the other factor in this essentially impossible and largely meaningless exercise in trying to figure out how many cars of each type would typically be seen on a train in the time period selected and then extrapolate that information to one's model railroad. Who has 1000 + cars? Who has the full mix of car types within box car classes alone and who has a layout on which to run full length freight trains? Not very many of us at all. So the averages, which are a rule of thumb perhaps but otherwise are just a mental exercise, don't really apply to any actual person's layout. Far too many, many variables in a national system, with the almost infinite perambulations of individual railroads, prototype and model, for these figures to have real meaning to a modeler. Again, the best and perhaps only lesson is to try to avoid the "iconic" cars when they were actually few in number, and balance your cars, with a good dose of home road cars, depending on what the history of your prototype was at any given time, when they had or didn't have a large fleet or kept it close to home, as in grain shipment time or on dedicated mineral runs.

Reality is not the same as a computer simulation under the rather limited and primitive conditions we are working in here.

It doesn't take a long time looking at photos, for example, of Northern Pacific or Great Northern freight trains and yards to see that at least on the days the pictures were taken, home road equipment often dominated, but the stats say otherwise. Go figure. Glad you are having fun with all this, Mike.

John Stokes
Bellevue, WA



To: STMFC@yahoogroups.comFrom: brockm@brevard.netDate: Sun, 17 Aug 2008 20:18:51 -0400Subject: Re: [STMFC] Re: Freight car distribution




Tim O'Connor writes:"Good work, Larry! That's what I'm talkin' 'bout!A 40 car train with 9 SP (not counting T&NO) boxcars.."I assume you mean a train with 40 box cars. Actually the 1949 trains averaged 82.3 cars."More than 5 times the 'average' value of1.6."I'm not certain what the 1.6 represents. The actual number is 4.Mike Brock...I never thought statistics could be fun.


Mike Brock <brockm@...>
 

Tim O'Connor writes:

"Mike, he ran 100,000 iterations of a 40 car box car train, composed
of cars chosen randomly from every possible prototype."

OK.

"The National
% is obvious,"

OK. 3.9% for SP in 1949.

"and then he finds the AVERAGE % found for that railroad
in the random consists. What is not obvious about this?"

Nothing. What's the value...IOW, so what?

I tallied the maximum number of cars for each railroad over all the
iterations. In a 40 boxcar train, the average number of cars would be
40 times the national percentage shown in the above table."
Why do you say that?
"Because he described it obscurely. He simply kept track of the largest
number of cars seen for each reporting mark in the 100,000 trains. Is
this not also obvious?

Well...actually that's not so at all. The correct number is 4. 136/34
If you use real...rather than theoretical data.
Real what? 136 box cars?"

Yep.

"Larry's example was PURELY theoretical, and he made an excellent point.
I guess some people just don't get it."

What's there to get?

We know that there were 4 of 34 trains...11.76%...of the trains with
far more than 9 SP bx cars.
Mike, you are having trouble grasping the idea. Larry has presented
100,00 theoretical trains, of fixed length, of random composition.
You persist with your miniscule sample of 34 actual trains of
varying length and composition of different sorts of cars (not just
box cars)."

Yep.

The 1953 data shows the "infamous" train with 36 or more SP box cars
[not T&NO ].
Yes, but I will argue that the Overland Route for the UP and SP was
a 'home road extension' in that each road was highly dependent on the
other for traffic. Therefore we can discount the presence of SP as
well as UP cars in all trains over Sherman Hill, for the purposes of
this discussion. This is exactly the "skew" that each layout has to
consider."

My point exactly. The theory doesn't seem to apply to the Overland Route for SP, CB&Q, C&NW, and Milw...although SP is the most significant violator. Other RRs might have similar situations...i.e, SP and SSW...ATSF and...well...ATSF. NP and GN with respect to CB&Q might also apply. Others might be MP and D&RGW and WP. The problem is, we don't have known data regarding those routes. However, the Overland Route should serve as a warning with regard to similar situations.

Mike Brock


Tim O'Connor
 

We've been down this road too -- most people take photos of
stuff they happen to like, or that catches their interest.
Almost no one took photos (and expended film) in order to
document "typical" stuff. Ergo: photos are not admissible
as evidence of anything other than what is depicted in them.

By the way, being an GN and NP fan, I've seen many photos
on each railroad where virtually no home road cars can be
seen. Including the infamous NP Montana branch line train
with a bunch of PRR empties headed to a grain elevator...

Tim O'

It doesn't take a long time looking at photos, for example, of Northern Pacific or Great Northern freight trains and yards to see that at least on the days the pictures were taken, home road equipment often dominated, but the stats say otherwise. Go figure. Glad you are having fun with all this, Mike.

John Stokes
Bellevue, WA


Tim O'Connor
 

Mike Brock wrote

My point exactly. The theory doesn't seem to apply to the
Overland Route for SP, CB&Q, C&NW, and Milw...
Actually, it applies exactly to the Overland Route. Your
samples are just too few to be statistically meaningful.
It's not even one day's worth of trains on the UP. The
data do not support ANYONE's theory. In the absence of
data, logic must prevail. Give me a logical explanation
for your consists, please! All you've said basically is
that Tim was wrong, based on a minute sliver of data. If
you have a better theory, I'm all ears.

Tim O'Connor


Mike Brock <brockm@...>
 

John Stokes writes:

"This is the other factor in this essentially impossible and largely meaningless exercise in trying to figure out how many cars of each type would typically be seen on a train in the time period selected and then extrapolate that information to one's model railroad. Who has 1000 + cars? Who has the full mix of car types within box car classes alone and who has a layout on which to run full length freight trains? Not very many of us at all."

Well...certainly not me. UP trains in my time period weere about 82 cars in length. Much of this was based on tonnage ratings. The line in question...between Laramie and Green River has...I believe...a max of 0.82 grade and UP generally liked to run relatively fast frt trains. Layout compression drops frt train lengths for me by at least 50%. 3 of 35 UP frt trains that we know of had 27-36 SP box cars in their consist. Assuming an 82 car length...I ain't gonna look it up...means the 3 trains had from 32% to 43% SP box cars. If I were to run a 40 car frt it could contain as many as 17 SP box cars. Well. Guess what. I don't have 17 SP box cars.

"Again, the best and perhaps only lesson is to try to avoid the "iconic" cars when they were actually few in number, and balance your cars, with a good dose of home road cars, depending on what the history of your prototype was at any given time, when they had or didn't have a large fleet or kept it close to home, as in grain shipment time or on dedicated mineral runs."

I agree. I will admit...as I have said...that it is sorta interesting to see a photo of cars where they "shouldn't be".

"Go figure. Glad you are having fun with all this, Mike."

Well, John, it's like this. Why else would I wade through this <G>.

Mike Brock