A Prototype Waybill

Introduction:

This waybill looked really familiar when I got the scan from Andy. As I sorted through the previous posts I confirmed it is a near duplicate of a waybill that we looked at back in November (http://cnwmodeling.blogspot.com/2012/11/a-detailed-look-at-prototype-waybill-8.html). And then by going back to Mark Mathu's GBW site and its index of waybills that Andy compiled, I found that this shipment was a recurring theme in the A&W Collection (http://www.greenbayroute.com/1962ahwwaybills.htm; see the table at the bottom of the text). In Andy's listing, there are 14 of these shipments, and this waybill (which is not included in Andy's index) is a 15th example. The OLHC for all of these shipments is the Raritan River Rail Road Company, the shipper is the Weldrock Division of the U.S. Plywood Corp., the consignee is Algoma Plywood and Veneer Division [of U.S. Plywood], the routes are all identical, and the commodity is calcium silicate and asbestos in flat pieces in every case. With 15 of these shipments received in 58 days, this was clearly an important traffic lane for Algoma Plywood and the Ahnapee & Western.

Our goals for this post are a little different than most of the others so far:

We will look at this waybill and compare it to the waybill from the 8th post in the series and see if there are any interesting differences in the form itself.
We will look at the time series of arrivals at Algoma Plywood and see if we can construct a demand curve.
We will look at the empty car selection practices of the Raritan River agents in this area at this time.

Waybill Comparison:

The waybill is from the A&W Collection and Andy provided the scan:

The form itself is identical to the form we showed in post 8 of this series. The use of the form is almost identical; tonight's waybill has special instructions spelled out to weigh the shipment at Parlin N. J. which wasn't typed on the earlier waybill, however this must have been a pretty standard practice because both shipments actually were weighed at Parlin. The typeface is very similar if not identical. The use of stamps is similar. This suggests to me that at least in this case, over a relatively short period of time, waybills within a traffic lane may show only very minor variations. This is nice for the waybill modeler because they are amenable to mass production or reuse.

On the left-hand side of the waybill you can see traces of a C&O Empty Car Bill and Home Route Card showing through the front page of this waybill. The C&O attached this form to the waybill somewhere between Buffalo and Kewaunee to facilitate return of the empty car via the service route. Andy provided a separate scan of that item (which was a happy find for me as the CNW and C&O interchange at Milwaukee and I need this form for my layout) which is reproduced here:

Time Series of Arrival:

If we were modeling the receiving end of the shipment (i.e., the A&W in Algoma), we would want to model the time series of arrival of the shipments. Each of the waybills is stamped "RECEIVED" by the Destination Agent and this is the date we will use in this analysis to construct the time series. It turns out that at this time, the A&W was only running on Monday, Wednesday, and Friday, with occasional extras on Saturday. This table show the distribution of the day of the week that shipments showed up in Algoma:

Day	Shipments
Sunday
Monday	7
Tuesday
Wednesday	5
Thursday
Friday	2
Saturday	1

A way to simulate this particular time series would be to establish a week-day schedule with high probabilities for Monday and Wednesday a somewhat lower probabilities for Friday, and an occasional extra thrown in on Saturday.

We could also look at this from the Raritan River's (shipper's) point of view and see how often they shipped to fulfill the demand. Unfortunately we don't have the dates that the waybills were written to support this analysis, but is is a reasonable guess that they shipped to Algoma on a pretty regular schedule to meet an established demand. The transit time from the shipper to the consignee was on the order of 7 days (judging by the two examples we have). So the vagaries of transit and the M-W-F schedule of the A&W result in a less even distribution at the consumer end of the logistic chain (i.e., Algoma Plywood).

Another way to think about this is the lag time between shipments - how long has it been since this industry has been served with this commodity? This table shows the distribution of lag times (in days) between shipments:

Lag	Shipments
0	2
1
2	2
3	1
4	2
5	4
6
7	3

The average lag between shipments is about 4 days, and this suggests you could set up a schedule that would deliver one of these shipments every four days (with some variation). A Poisson process would probably provide a good descriptor of the variance.

Empty Car Selection Practices:

Understanding empty car selection practices is difficult. The factors influencing empty car selection include the intent of the agent/clerk performing the selection and the supply of empty cars from which to choose. Typically, in retrospect, we don't have a good way to evaluate either of those factors. Andy and I decided after several of these waybill posts to describe the individual outcomes, and periodically evaluate the distribution of the cumulative outcomes.

We developed a set of empty car selection codes that are purely descriptive. They are:

N/A - Not Applicable; I used this code for waybills for empty cars that were requested by and delivered to the A&W for loading
1A - a home car loaded to a home road destination
1B - a home car loaded to a foreign destination
2A - a foreign car loaded via owner's road to any destination
2B - a foreign car loaded to its home district
2C - a foreign car loaded to an intermediate district (toward home, but not all the way)
2D - a foreign car loaded to an adjacent district (toward home but beyond 1 district)
2E - a foreign car loaded to any other destination

The January 1958 ORER tells us that the Raritan River Rail Road (RR) had no freight cars in commercial service, so the first three outcomes in the list above are not applicable to this discussion. If there were an infinite pool of empty cars from which the agent could make a selection for each shipment, and if the intent of the agent was to follow the AAR Car Selection Practices, we would expect all of the outcomes to be coded as 2A. In practice, there certainly was a restricted pool of empty cars from which the selection was made, and probably following the AAR Practices wasn't the top priority each time a selection was made.

14 of the 15 selections made by the RR were categorized and listed in the cumulative post Andy and I prepared several months ago (http://cnwmodeling.blogspot.com/2013/01/the-100-carload-waybill-study-2.html). Adding the selection from this waybill makes a total of 15 selections by the RR over a two month period that look like this:

2A - 0 selections
2B - 7 selections
2C - 2 selections
2D - 5 selections
2E - 1 selection

These results show that in this case, there was no use of cars from carriers that participated in the line haul (2A, which would include the CNJ, DLW, ERIE, CO, and KGB), that most of the selections resulted in cars moving to (2B) or near (2C and 2D) their home districts, and only one car that wasn't forwarded generally toward home (a TNO car). If I were modeling these selection outcomes I would probably pick cars from carriers whose home districts were 7, 8, 11, 12, 14, and 15 according to their frequencies in the national fleet. This would be in marked contract to the car selection choices made by the SP at Raceland LA or at several Oregon locations that tended to select home road empties for foreign shipments.

We can also code the outcomes in terms of a simple delta. We can easily compute:

Delta = (#Districts Away from Home District at Shipper's location) - (#Districts Away from Home District at Consignee's location)

Deltas that are positive represent moves that send cars closer to home districts; deltas that are negative represent moves that send cars away from their home districts. For the 15 shipments from the RR:

Delta = -1; 2 shipments
Delta = 0; 1 shipment
Delta = 1; 3 shipments
Delta = 2; 7 shipments
Delta = 3; 2 shipments

On the whole, this is a pattern of midwestern and western cars (that happened to be empty on the RR in New Jersey) moving west with loads, which is consistent with the discussion above. This table contains the data that supports the discussion above:

MK	#	#Int	Cat	Delta
SP	102242	5	2C	3
GN	6419	5	2B	2
TNO	60833	5	2E	1
NJII	3481	5	2D	0
NYC	41162	5	2D	-1
IC	15804	5	2B	2
MILW	51635	5	2B	2
MILW	51014	5	2B	2
IC	42460	5	2B	2
UP	165056	5	2D	3
ATSF	16620	5	2D	1
MILW	50382	5	2B	2
ATSF	13761	5	2C	1
MILW	51601	5	2B	2
NKP	85228	5	2D	-1

Points to Ponder:

The total traffic onto the A&W was made from a combination of a few strong traffic lanes and less frequent sporadic shipments.
Each traffic lane probably has its own characteristics in terms of consistency of waybill appearance, selection of routes, time series of arrival, and empty car selection.
If we can tease out the behaviors of the individual traffic lanes and develop models for them, we can get a combined behavior for a layout as a whole that is likely to be interesting and varied.
The shipments of calcium silicate and asbestos in box cars from the USP Weldrok Division tended to be consistent in route and waybill appearance, the time series of arrival was reasonably regular, and the empty car selections tended to be from carriers in and around District 11.

Andy Laurent and Charles Hostetler