Monday, November 7, 2011

Operations: demand-based car flow-2

My approach to layout operations begins with realistic paperwork (as I’ve discussed in several posts with the series title “Waybills”). That paperwork is used as part of a system of realistic and flexible car movements. This post continues description of my car movement system. In my previous post on car flow, available at: , I described the creation of a list of inbound and outbound car movements for each industry. In the present post, I describe the combining of information from this list into a schedule. Such schedules can be used as a shipper-demand system of waybill selection for operation.
     The goals of this process are as follows. First, I want to represent regular, periodic car movements. For example, a manufacturer may load an empty box car every day of the work week. Second, I want to represent periodic movements which are not quite so regular. For example, a fuel dealer may get a tank car of LPG at longer intervals in the summer than in the winter. Or a wholesale grocer may get reefer loads of dairy products at irregular intervals. Third, there are cases in which a car delivery may be at either regular or irregular intervals, but it is usually the same car. This can happen, for example, with a bulk oil facility which repeatedly receives the same exact car, unloads it, and sends it back for another load of, say, diesel fuel. Fourth and finally, there are rare or unusual movements. Though these may be a small fraction of total car movements, they provide important variety and realism.
     I begin with a simple example of a regular movement. Imagine a packing shed which needs an empty refrigerator car to load, every other day. (This need may have been developed as part of the list of car movements at this industry.) Then I would make a schedule line that looks like this. The “open” letters denote empty cars; “filled” or solid letters would denote loaded cars.

 The 10-day span of this example schedule line was selected as a simple example, not for any particular significance of 10 days.
     It is reasonable to expect these empties to be loaded promptly with a perishable type of load, so the loads would need to be picked up late the same day, or the next day. If the next day, it might be expected that the table would then look like this:

Again, the filled letters denote loaded cars.
     But in fact, my system would not show these outbound loads, because the waybill assignment to these cars will “automatically” result in their outbound movement. In other words, the role of the schedule is to initiate car flow through waybill selection, and once initiated, the flow continues under the guidance of waybills.
     Thus for each industry, only the cars directed to that industry are in the schedule, whether loads or empties, and subsequent movement is via waybill or Empty Car Bill information. So if I were to have four locations of car spots--the same produce shed which was just shown, plus a team track, an oil dealer, and a warehouse of some kind--my town schedule might look like this:

The frequencies of delivery of loads or empties are those developed in the prior exercise of listing car movement needs at each industry. Note that the team track receives only a loaded box car and a loaded flat car in these ten days (these are irregular car flows), the oil dealer a pair of loaded tank cars, and the warehouse two empty box cars to be loaded with outbound cargoes. The latter two cases are periodic but less frequent than the produce shed’s needs, every fourth or fifth day instead of every other day.
     Already in this example, it can be seen that no two days of the ten are alike, yet each car spot is receiving a plausible amount of traffic (and remember, that this schedule only shows half the traffic, because the subsequent movement of cars, after they are delivered according to this schedule, need not be shown but occurs entirely by waybill or Empty Car Bill).
     Let’s call the town just outlined “Shumala” and then let’s add a second town, again with four industries for illustration purposes. The second town might be called “Ballard,” having a wholesale grocer (Peerless Foods), a winery (Zaca Mesa), a team track, and a produce shed (Guadalupe Fruit). The distribution schedule might look like this:

The four “Shumala” industries are the same as shown in the prior figure. Now with eight car spot locations, traffic is fairly diverse, and it remains true that no two days are alike (and again, this shows only car deliveries, omitting pickups). Note also that both regular and irregular car deliveries are shown here.
     These two towns are of course towns on my layout, and the industries are in fact ones I already have or plan to add in those towns, but the illustration could be constructed for any combination of industries and towns that might be desired. For more about my actual Ballard, for example, see the description in an earlier post ( ).
     This kind of scheduling works well for fairly frequent car arrivals, or ones which are predictably periodic. What it does not do well is include the rare or truly occasional car flows. Here is how I have added those car flows.
     My first step, listing all car movements at an industry, even the rather uncommon ones, along with a few much smaller consignees which might use the team track or freight house in a town, resulted in a considerable list of such car movements for each town.
     I then made up waybills for all these movements, and filed them by town, in a fully shuffled or “randomized” pack for each town. I select these by simply drawing the appropriate number of bills from the front of the pack, and when movement is completed, return them to the back of the pack. Occasionally shuffling the pack can be done if desired. How do I know the “appropriate number” to draw? It is shown in a line on the schedule, called “random,” like this:

Here the initials refer to each town: B = Ballard, S = Shumala, SR = Santa Rosalia. For each operating day, I would draw the number of bills shown. Given the large number of total “occasional” bills for each town, repetition is again not an issue. More can be said about what kinds of waybills represent these “rare” or infrequent flows, but that discussion is sufficiently complex that I will defer it to a future post.
     Now my total schedule for the two towns, three or four industries each, has an added line for these infrequent flows:

This is of course only illustrative, and as mentioned, need not be restricted to a 10-day span but could extend as far as desired. Obviously longer time spans can be used for more complex patterns, and will increase diversity of operating days still further.
     When my layout was in Pittsburgh, and had a somewhat different operating rationale, my version of this car flow system had a 31-day schedule. Here it is, with somewhat different industry names and locations from the present arrangement (Jalama is now called Shumala, and Los Olivos is now Santa Rosalia). You can click on the image to enlarge it.

My idea had been that whatever calendar day happened to be the day we operated, I would just use that day in the schedule. If we operated on October 18, we would use Day 18 in this schedule. But I quickly discovered that the periodic cycles fell apart this way, as operating days were scattered through the month in no particular order. I then simply used the 31 days in the schedule as sequential operating days. So if the last time we operated was Day 18, the next session would be Day 19. This worked fine.
     Even in the 10-day version shown earlier in this post for illustration, there is no repetition of identical days within the ten-day span, nor can there be with the presence of the “random” draws of unusual flows. That statement can be made even more strongly for a longer span of days, such as the 31-day schedule shown immediately above. Thus this approach does accomplish the goals I stated at the outset, of reproducing regular, irregular, repeating, and rare car flows, without significant repetition of operating cycles.
     I am occasionally asked how to implement a system like this on a large layout such as a club. To that end, I have experimented with a computer approach to generate the schedule. I have also tried a “permanent waybill” approach. But those both require enough explanation, that I will need to postpone description of them to a future post.
Tony Thompson


  1. Tony,
    Could you please direct me to the followup post for this reference "More can be said about what kinds of waybills represent these “rare” or infrequent flows, but that discussion is sufficiently complex that I will defer it to a future post."

  2. a question - how do you handle industries with more than 1 spot and/or car type - by adding lines under the industry?
    Alan Lagocki

  3. Good question, Alan. I rely on the physical arrangements at an industry, for example, a loading dock alongside the building, and a separate tank car loading platform. I assume (not always correctly) that switch crews can recognize which spot takes the tank car. (That example is for the Zaca Mesa winery.) In more complex cases, it might make sense to have a second line in the schedule.
    Tony Thompson

    1. so would the demand chart have two car types in the line for that industry?

  4. No, it would not need to. Look at the team track, which shows box (XM)) and flat (FM) cars in the same line. But as I said, if the industry has complex inbound or outbound movements, a second or third line could be helpful.
    Tony Thompson

    1. OK, thanks. i'm looking forward to implementing this demand-based system to my rr. it simplifies the logic of car movements. terrific series!