A Purpose For Frt Car Distribution Studies. Was: Re: Re:Fwd: Re: Freight car distribution


Mike Brock <brockm@...>
 

Gene Green writes:

"All this discussion of freight car distribution has what purpose? A
purpose does not seem self-evident to me."

A very legitimate and timely question.

OK...here's an opinion...and it seems obvious. The data developed by Dave Nelson and Tim Gilbert have resulted in a theory that the % of a foreign RR's box cars on a given RR will match the % of the foreign RR's % of the national box car fleet. Tim noted:

"The freight car data which Dave Nelson & I have parsed from wheel
reports and other car reports will not solve Mike Brock's "California
Zephyr" problems. All they represent is a pool of cars which were
reported by a railroad over a course of time."

Tim was responding to a comment that I made that my op sessions always replicate May 14, 1953. While I did say that, it was tongue in cheek...I actually simulate the entire month of May, 1953. But his point was, I think, that the theory could predict and, therefore, project the foreign box car population over a relatively long period of time...I think a year was suggested. The problem with this concept is that frt train consists, at least on the UP in Wyoming, were very variable. IOW, 30% of one train might be UP box cars while another might have no UP box cars although 50% of the train was box cars. About 8% of the box cars should be UP [ home road % of about twice the national % ]. As I pointed out, major RRs interchanging with a RR...at least in UP's case...seem to be more represented than that of the national %...as in the case of SP. The problem is, even with compression of model train lengths and compression of the number of trains present during an op session, the need to have enough cars to match a particular train...like the lumber trains with 31 SP box cars in 1949...means that a rather large car base would be required...in this case, at least 1000 box cars IF that base reflected the national per centages. So...what does one do? Well, I for one recognize that the value of the theory and statistical data is that it shows that box cars went everywhere. The numbers of different foreign box cars on a given layout probably reflect the RR being modeled. IOW, there might be a "local" factor at work as in the case of UP and Milw, CB&Q, C&NW and certainly SP and I would have more of them than the theory calls for....probably calling for 1.5 times the national %. Why? I don't care. It's just the way it was. At the same time if I wanted to simulate a particular train...compressed...I would simply have enough of those cars available as well and I wouldn't be concerned that the total box car fleet would not reflect the national per centages. IOW, my box car fleet probably won't exceed 150 cars[ hmmm...guess I'll have to count ] .

I realize that we have a great deal more information about UP in Wyoming than might be available for other RRs and what applies for the UP trunk line across WY might not apply to other areas. After all, everyone knows that at least one of every class of frt car went over Sherman Hill at some time. Heck, even General Tojo went over Sherman Hill.

Mike Brock


Stokes John
 

Mike,

When you get to the end of your commentary on freight car percentages and what UP might have been running in a consist in a given time period, you end up with virtually nothing concrete but using the national averages to play around with it, and vary from day to day, and let it rip. This reminds me of the old saw about if you let a group of monkeys play with typewriters long enough they will end up writing the great American novel, by pure random chance. Build up a fleet of over a 1000 cars from all roads and then mix them up endlessly over time and maybe, just maybe, on one day you will have replicated an actual consist. Otherwise you are guessing and trying to create some semblance of reality, but reality is always waiting for the next roll of the dice.


The other factor that seems to be evident is that the UP is not typical of many railroads, given its national significance and overland route that channeled a lot of freight across the West, and the same time period for the CB&Q would produce a very different mix of cars, given the nature of the Q and the types of traffic it generated, as an example comparison. All of this is speculation piled on infinite possibilities based on very small snapshots of what happened in a really great big country over thousands of days, every one of them different and reflective of an ever changing rail scene and traffic and commodity trends. If spending inordinate amounts of time agonizing over this and trying to match real consists on your favorite railroad on a given day or month or year floats your boat, then go to it, but I think Gene has a cogent question, and I have seen no good answers yet.John Stokes
Bellevue, WA



To: STMFC@yahoogroups.comFrom: brockm@brevard.netDate: Thu, 14 Aug 2008 13:18:54 -0400Subject: A Purpose For Frt Car Distribution Studies. Was: Re: [STMFC] Re:Fwd: Re: Freight car distribution




Gene Green writes:"All this discussion of freight car distribution has what purpose? Apurpose does not seem self-evident to me."A very legitimate and timely question.OK...here's an opinion...and it seems obvious. The data developed by Dave Nelson and Tim Gilbert have resulted in a theory that the % of a foreign RR's box cars on a given RR will match the % of the foreign RR's % of the national box car fleet. Tim noted:"The freight car data which Dave Nelson & I have parsed from wheelreports and other car reports will not solve Mike Brock's "CaliforniaZephyr" problems. All they represent is a pool of cars which werereported by a railroad over a course of time."Tim was responding to a comment that I made that my op sessions always replicate May 14, 1953. While I did say that, it was tongue in cheek...I actually simulate the entire month of May, 1953. But his point was, I think, that the theory could predict and, therefore, project the foreign box car population over a relatively long period of time...I think a year was suggested. The problem with this concept is that frt train consists, at least on the UP in Wyoming, were very variable. IOW, 30% of one train might be UP box cars while another might have no UP box cars although 50% of the train was box cars. About 8% of the box cars should be UP [ home road % of about twice the national % ]. As I pointed out, major RRs interchanging with a RR...at least in UP's case...seem to be more represented than that of the national %...as in the case of SP. The problem is, even with compression of model train lengths and compression of the number of trains present during an op session, the need to have enough cars to match a particular train...like the lumber trains with 31 SP box cars in 1949...means that a rather large car base would be required...in this case, at least 1000 box cars IF that base reflected the national per centages. So...what does one do? Well, I for one recognize that the value of the theory and statistical data is that it shows that box cars went everywhere. The numbers of different foreign box cars on a given layout probably reflect the RR being modeled. IOW, there might be a "local" factor at work as in the case of UP and Milw, CB&Q, C&NW and certainly SP and I would have more of them than the theory calls for....probably calling for 1.5 times the national %. Why? I don't care. It's just the way it was. At the same time if I wanted to simulate a particular train...compressed...I would simply have enough of those cars available as well and I wouldn't be concerned that the total box car fleet would not reflect the national per centages. IOW, my box car fleet probably won't exceed 150 cars[ hmmm...guess I'll have to count ] .I realize that we have a great deal more information about UP in Wyoming than might be available for other RRs and what applies for the UP trunk line across WY might not apply to other areas. After all, everyone knows that at least one of every class of frt car went over Sherman Hill at some time. Heck, even General Tojo went over Sherman Hill.Mike Brock


boyds1949 <E27ca@...>
 

The discussions on this list should serve two purposes: accurate and
complete (so far as is possible) knowledge of the prototype freight
car fleet and how to accurately model them.

Determining as best we can how the cars were actually distributed
should be just as important as knowing the type of underframe on a
particular class of freight cars.

How modelers use this information may vary. I may not have the
skills or patience to accurately model every aspect of an FGEX
refrigerator car as well as many others on this list, but my starting
point still has to be an accurate set of data.

None of us may be able to develop a fleet which exactly reflects
percentages of the national ownership but we need to have the data to
provide an accurate starting point. If nothing else, having access
to accurate data on the distribution of freight cars can help us keep
from going overboard when some exciting new model comes out.

John King







--- In STMFC@yahoogroups.com, John Stokes <ggstokes@...> wrote:

Mike,

When you get to the end of your commentary on freight car
percentages and what UP might have been running in a consist in a
given time period, you end up with virtually nothing concrete but
using the national averages to play around with it, and vary from day
to day, and let it rip. This reminds me of the old saw about if you
let a group of monkeys play with typewriters long enough they will
end up writing the great American novel, by pure random chance. Build
up a fleet of over a 1000 cars from all roads and then mix them up
endlessly over time and maybe, just maybe, on one day you will have
replicated an actual consist. Otherwise you are guessing and trying
to create some semblance of reality, but reality is always waiting
for the next roll of the dice.


The other factor that seems to be evident is that the UP is not
typical of many railroads, given its national significance and
overland route that channeled a lot of freight across the West, and
the same time period for the CB&Q would produce a very different mix
of cars, given the nature of the Q and the types of traffic it
generated, as an example comparison. All of this is speculation piled
on infinite possibilities based on very small snapshots of what
happened in a really great big country over thousands of days, every
one of them different and reflective of an ever changing rail scene
and traffic and commodity trends. If spending inordinate amounts of
time agonizing over this and trying to match real consists on your
favorite railroad on a given day or month or year floats your boat,
then go to it, but I think Gene has a cogent question, and I have
seen no good answers yet.John Stokes
Bellevue, WA



To: STMFC@...: brockm@...: Thu, 14 Aug 2008 13:18:54 -0400Subject:
A Purpose For Frt Car Distribution Studies. Was: Re: [STMFC] Re:Fwd:
Re: Freight car distribution




Gene Green writes:"All this discussion of freight car distribution
has what purpose? Apurpose does not seem self-evident to me."A very
legitimate and timely question.OK...here's an opinion...and it seems
obvious. The data developed by Dave Nelson and Tim Gilbert have
resulted in a theory that the % of a foreign RR's box cars on a given
RR will match the % of the foreign RR's % of the national box car
fleet. Tim noted:"The freight car data which Dave Nelson & I have
parsed from wheelreports and other car reports will not solve Mike
Brock's "CaliforniaZephyr" problems. All they represent is a pool of
cars which werereported by a railroad over a course of time."Tim was
responding to a comment that I made that my op sessions always
replicate May 14, 1953. While I did say that, it was tongue in
cheek...I actually simulate the entire month of May, 1953. But his
point was, I think, that the theory could predict and, therefore,
project the foreign box car population over a relatively long period
of time...I think a year was suggested. The problem with this concept
is that frt train consists, at least on the UP in Wyoming, were very
variable. IOW, 30% of one train might be UP box cars while another
might have no UP box cars although 50% of the train was box cars.
About 8% of the box cars should be UP [ home road % of about twice
the national % ]. As I pointed out, major RRs interchanging with a
RR...at least in UP's case...seem to be more represented than that of
the national %...as in the case of SP. The problem is, even with
compression of model train lengths and compression of the number of
trains present during an op session, the need to have enough cars to
match a particular train...like the lumber trains with 31 SP box cars
in 1949...means that a rather large car base would be required...in
this case, at least 1000 box cars IF that base reflected the national
per centages. So...what does one do? Well, I for one recognize that
the value of the theory and statistical data is that it shows that
box cars went everywhere. The numbers of different foreign box cars
on a given layout probably reflect the RR being modeled. IOW, there
might be a "local" factor at work as in the case of UP and Milw,
CB&Q, C&NW and certainly SP and I would have more of them than the
theory calls for....probably calling for 1.5 times the national %.
Why? I don't care. It's just the way it was. At the same time if I
wanted to simulate a particular train...compressed...I would simply
have enough of those cars available as well and I wouldn't be
concerned that the total box car fleet would not reflect the national
per centages. IOW, my box car fleet probably won't exceed 150 cars[
hmmm...guess I'll have to count ] .I realize that we have a great
deal more information about UP in Wyoming than might be available for
other RRs and what applies for the UP trunk line across WY might not
apply to other areas. After all, everyone knows that at least one of
every class of frt car went over Sherman Hill at some time. Heck,
even General Tojo went over Sherman Hill.Mike Brock






[Non-text portions of this message have been removed]


Mike Brock <brockm@...>
 

John Stokes writes:

"When you get to the end of your commentary on freight car percentages and what UP might have been running in a consist in a given time period, you end up with virtually nothing concrete but using the national averages to play around with it, and vary from day to day, and let it rip."

Well...not quite. I think I said that I would modify the theory to increase the % of closely associated RRs...Milw, C&NW, CB&Q and SP...and basically use the other per centages as projected by the theory. I would add the necessary cars to simulate in a compressed mode a few unusual trains like the case of a lumber train with about 30% of it in the form of SP box cars.

"The other factor that seems to be evident is that the UP is not typical of many railroads, given its national significance and overland route that channeled a lot of freight across the West, and the same time period for the CB&Q would produce a very different mix of cars, given the nature of the Q and the types of traffic it generated, as an example comparison."

Possibly so...although I have not analyzed CB&Q consists. My guess is that Santa Fe would be closer to the UP, absent, of course, an increase in Milw, CB&Q, SP and UP cars.

"All of this is speculation piled on infinite possibilities based on very small snapshots of what happened in a really great big country over thousands of days, every one of them different and reflective of an ever changing rail scene and traffic and commodity trends."

True enough. I have frequently noted the tiny size of the data. Still...it IS interesting that the data does support the theory fairly well.

"If spending inordinate amounts of time agonizing over this and trying to match real consists on your favorite railroad on a given day or month or year floats your boat, then go to it, but I think Gene has a cogent question, and I have seen no good answers yet."

Well, anyone who knows me knows that I don't agonize over anything associated with this hobby. You might note that I changed the subject heading to "A purpose..." My message explains why I think the theory is useful and the way I would and do apply it. If others find it to not be worthwhile...that's up to them.

Another aspect of studying frt conductor books, BTW, is that, in the UP case, some RRs are not represented to the extent of their %. NP boxcars, for example, are not as common in the 35 UP trains as the theory projects...in direct conflict with Brock's Fifth Rule of Frt Cars.

Mike Brock


Dave Nelson
 

-----Original Message-----
From: STMFC@yahoogroups.com [mailto:STMFC@yahoogroups.com] On Behalf Of Mike
Brock:

Gene Green writes:
"All this discussion of freight car distribution has what purpose? A
purpose does not seem self-evident to me."

A very legitimate and timely question.

OK...here's an opinion...and it seems obvious. The data developed by Dave
Nelson and Tim Gilbert have resulted in a theory that the % of a foreign
RR's box cars on a given RR will match the % of the foreign RR's % of the
national box car fleet. Tim noted:

"The freight car data which Dave Nelson & I have parsed from wheel reports
and other car reports will not solve Mike Brock's "California Zephyr"
problems. All they represent is a pool of cars which were reported by a
railroad over a course of time."

---------- begin reply--------
I don't recall Tim ever mentioning why he was interested in the topic but I
do recall why I was: I had changed scale and as that fact had put me in the
market for new rolling stock I wanted some guidelines that would serve me
when making purchases.

As to why the topic returns to this list with some regularity I can only
hazzard a guess: other people are in the same boat today as I was then.

VMMV.

As for Mike's later comment that perhaps a years time is necessary to get
sightings that conform to the hypothesis, I disagree. It appears around
800-1000 cars will come fairly close. More is obviously going to be better.
Whether any one location will see that many cars in a day, a week, or a year
is discussing, IMO, the unit of measure, not the sample size. Or if you
want, translate the unit of measure it into how many operating sessions,
with the sample size being some portion of your total foreign road boxcar &
flatcar roster.

Dave Nelson


Mike Brock <brockm@...>
 

Dave Nelson writes:

"As for Mike's later comment that perhaps a years time is necessary to get
sightings that conform to the hypothesis, I disagree. It appears around
800-1000 cars will come fairly close."

First, note that my data is about 1/35 of the train consists for only about a month and a half. What I was trying to say is that during this period of time, the number of appearances by SP box cars is about 2.6 times what the theory predicts. I think you and/or Tim suggested that my data may contain a few unusual trains and that, over a longer period of time, say a year, the total number of SP box cars would fit better with the theory's prediction. I don't, of course, know that to be true and I don't really care because I'm not interested in train operations over that long a time. What we do know is that those three UP frt trains with way over the predicted number of SP box cars did occur and being a prototype modeler, I want the opportunity to model them...without having to have 1000 box cars in the car base. For all we know, during the next month and a half even more SP box cars appeared in Wyoming.

Theories are fine but I prefer them to be empirically developed to match real data. Hence, I am convinced, barring data to the contrary, that "closely associated" RRs will be represented by a higher % than their national %...say somewhere between 1.5 to 2 times. Why? I don't really care.

Mike Brock


Dave Nelson
 

I feel the need to respond to my own post here, for clarification. Full
response at end.

-----Original Message-----

As for Mike's later comment that perhaps a years time is necessary to get
sightings that conform to the hypothesis, I disagree. It appears around
800-1000 cars will come fairly close. More is obviously going to be better.
Whether any one location will see that many cars in a day, a week, or a year
is discussing, IMO, the unit of measure, not the sample size. Or if you
want, translate the unit of measure it into how many operating sessions,
with the sample size being some portion of your total foreign road boxcar &
flatcar roster.

Dave Nelson
------------------------------------

To be clear, just as the distribution hypothesis requires a large number of
cars, observed over a period of time -- the length of which will vary by
traffic volume -- so must the modeler expect to use a large number of cars,
varying them over time, the length of which will vary by total roster size.
IOW, you should not plan your inventory of cars visible ON THE LAYOUT
according to the distribution hypothesis, but your COMPLETE roster. And you
must vary what's on the layout over time. Just like a very busy trunk line
will see a larger sample, a larger, complete roster will allow you to "hit
the numbers" sooner. And conversly, like any secondary route w/ fewer
sightings, a small roster will require more time before you "hit the
numbers".

This says nothing whatsoever about the composition of any one train. Or the
composition of the ON THE LAYOUT roster in any given session. It is the
variation over time that allows you to cycle in and out cars that in total
reflect the national distribution of boxcars.

This obviously presents some very real problems: folks often don't like to /
or shouldn't handle cars. The solution to this lies w/ each person and
their layout. Staging is probably part of the answer. Having decent off
layout storage and figuring out best way to remove in and out of that
off-layout storage has to be another.

With this understanding, you can expand your roster of boxcars to well more
than what's on the layout (we all do that anyway), keeping the more prolific
road names in place (perhaps changing cars) and then cycling in and out lots
of the smaller roads so that over a period of time, what's seen on the
layout is a fairly uniform distribution of boxcars per the national fleet.

Dave Nelson


Mike Brock <brockm@...>
 

Dave Nelson notes:

"To be clear, just as the distribution hypothesis requires a large number of
cars, observed over a period of time -- the length of which will vary by
traffic volume -- so must the modeler expect to use a large number of cars,
varying them over time, the length of which will vary by total roster size."

I suppose we can assume that the larger the data sample, the more likelihood that the data will fit the hypothesis...IF the hypothesis is correct. But it works both ways. If a much larger data sample does not fit the hypothesis, the hypothesis needs to be modified. The problem is, we don't really have the means to expand the data to adequately test the hypothesis. It just ain't there...at least in the perhaps somewhat unique case of the UP trunk in Wyoming. My frt conductor's book only covers about 1/35 of the trains operating in the month and a half covered, a pretty small sample. Are the consists of the trains we have typical...i.e., would another 1/35 sample show similar trains with much higher than expected numbers of Milw, CB&Q, C&NW and SP box cars? I have no clue. We do know that a similar situation occurred with regard to SP box cars 4 yrs after the 1949 data.

"IOW, you should not plan your inventory of cars visible ON THE LAYOUT
according to the distribution hypothesis, but your COMPLETE roster. And you
must vary what's on the layout over time. Just like a very busy trunk line
will see a larger sample, a larger, complete roster will allow you to "hit
the numbers" sooner."

No argument with that.

Mike Brock