This post continues some of the discussion I initiated in talking about the role of the local agent in my system, and though the agent has some leeway, car flow is not fundamentally determined by agent decisions either (see http://modelingthesp.blogspot.com/2011/10/operations-role-of-agent.html).
Obviously the combined car flow for the entire layout is merely the sum of the car flows for all the industries modeled (using “industry” broadly to include any place where cars may need to be spotted, including team tracks and depot house tracks). That means that the first step in constructing a car flow system is to examine each industry in turn. What kind of inbound loads will it receive, what kind of empties will it load for outbound movement (or both), and how often? This may sound simple, but can readily become quite complex and even beyond the reach of reasonable research.
Rather than pursue great complexity, I have resorted to simple estimates of the work that a model industry could do, and the cars accordingly needed. Most of our modeled industries are extremely compressed compared to their prototypes, but so are our sidings, so the one- or two- or three-car spots we model are often entirely suitable to the size of our modeled industry. The big exception is placing the industry on the backdrop or in the form of flats against the backdrop, in which case we only really model the set-out or unloading tracks. This is probably the best way to represent a truly big industry at all convincingly.
Let me give a couple of examples of industry work estimates. I have a model at Ballard of a wholesale grocer. This business receives canned goods and packaged foods, as well as meat, dairy products, and fresh produce (other than produce types produced locally). This means inbound box cars and reefers and nothing else; and there are no outbound loads. I have four door spots at this building, so can accommodate up to four cars at once. Anything perishable will be unloaded promptly, so those cars will be picked up empty no later than the following day. But non-perishable cars will not demand the same hurry, and might stand at the door spot for a second day in some cases.
Balancing the different types of inbound food products in proportion to reasonable needs (you can eyeball the proportion of total shelf-feet at your supermarket for each of the different food types, remembering of course that today’s food variety is not yesterday’s). I did this several years ago, and made up a list of relative amounts of food types to guide inbound load frequencies.Here is a photo of a meat reefer standing at Door 2 of this building (Peerless Foods), which handles meat and dairy products.
The structure adjoins the backdrop and accordingly is of somewhat ambiguous size, a useful arrangement for any business which ought to look fairly big.
Conversely, here’s an example of an industry with almost all outbound loads. I will eventually have four packing sheds on the layout, and for the part of coastal California that I model, at least some crops are being shipped in most months. But the sheds I plan to model handle different kinds of produce, so in any particular month, a different one might be most active. All will have two- or three-car spots, and in the peak season for each one, they would get at least one and probably two empty reefers a day. Real packing sheds load a lot more than that in peak season, but of course my modeled buildings are nowhere near the size of most real packing sheds.
Some other industries will receive loads, or load empty cars, much less frequently. This is especially true for team tracks and house tracks, which typically serve businesses which don’t handle enough shipments to justify their own siding. It is helpful to know something about actual businesses in the town or area you are modeling, in your era, to choose possible patrons of the team or house track. As I have mentioned before, library collections of period phone books can be very helpful for such information, but I never cease to be pleasantly surprised at the amount of this kind of history that I find using Google.
To some extent, you may also want to consider train size, so that you are not going to create a demand pattern which on average is bigger than your switching capability. But that will become more evident in the next step, when a schedule is created.
Once a set of inbound and outbound car types, and kinds of loads, is created for each industry modeled, along with some indication of how frequently those car movements would be needed, it is time to start making up a schedule of car movements. But I will describe that in an upcoming post.