I would and still take the word of anything Dave Nelson and Tim Gilbert did as gospel. After that, I support Bruce Smith's reply to get further clarity.
Seeing Tony Wagner's post, yes it may well require a goodly amount of research on your part. How you look upon doing that will affect how deeply you want to dig.
Set some significant time aside, grab your favorite beverage and search this list for Nelson-Gilbert or Gilbert-Nelson, and the same letters without the dash or hyphen "-". Read all of those. They provide a good start without leaving your chair.
Brian Carlson did an excellent presentation years (a decade or two ago by now) at CCB. It inspired me to follow his example and later do a presentation at Naperville Sunshine, and the following CCB titled Closer to the Rail Head.
In a few words, taking the above research by Dave Nelson and Tim Gilbert, and using the reasoning Bruce put forth, that gets you a national balance. Then determine the geographic parameters, of the overall rail division you want to model. Search that area for major industries in that area or just beyond. And interchanges. That will get regional flavor.
And the time frame you model affects things. As has been stated here repeatedly, the closer you can get a year, season, month, as a setpoint, the more valid your data search as applied will be.
I model the Twin Cities - Twin Ports, late September of 1953 and what did I find? At first I thought it would be heavy northbound grain trains of 40 foot boxcars to the ports, and in some manner - it was that.
However, and here is where the local major industries had effect, there was heavier coal traffic coming south for domestic and commercial heating than the grain going north. Something that would taper off by the late 50s as homes switched to natural gas. (Hence the importance of timeframe.) Still that coal traffic south, via hopper, gon, and boxcars was the major tonnage in either direction and helped determine my car purchases.
A steel mill in the Duluth area that sent pig iron ingots south via gons, in addition to boxcar loads of wire and nails southbound in significant amounts entered in.
There was a pipeline running east to west through both the Twin Cities and Twin Ports and they balanced transfers between the pipelines via rail in tank cars, in significant numbers. I need to clarify, both pipelines had transfer points directly on or adjacent to my railroad. Also as I have posted here prior, some tank cars loaded from Montana and Wyoming to Minnesota, enough that it enters in as a traffic factor to me.
A surprise was meat, there were the St. Paul stockyards and several meat distribution offices in Duluth. Swift, Armour, and others. That in addition to a slaughter house in West Duluth which meant stock cars northbound. From the St. Paul stockyards went carloads of meat for export via ship. How or why did they do that? I don't need to know, all I needed was the carload numbers of meat reefers.
The connection from Canada via the DW&P into Duluth was significant, as I have posted to this list before, 44 cars per day on average on my railroad, CN cars, mostly of newsprint, going south loaded, and returning empty. I bought a whole bunch of CN TLT cars for representation there.
All of this affected my balance of the purchase of gons, hoppers, reefers, tank cars, in addition to boxcars.
And all found through research.
How deeply do you want to research, that depends and is up to you. I will add that there were several branch lines off the above Twin Ports -Twin Cities. The Grantsburg branch and the Stillwater branch were two of those. The Stillwater branch connected with the Omaha and Milwaukee at Stillwater, MN. There was pulpwood that came off the Grantsburg branch, then went south to the Stillwater branch, and was transported by prior named connections to papermills in central Wisconsin. Again, something that affects modeled traffic.
Of steel vs wood sided, I cannot give you a clear answer. Other have posted on wood vs steel car by year. You'll just have to search this list for those posts and that data.
However Mike Brock himself posted this in the files https://realstmfc.groups.io/g/main/files/DS-SS-Steel%20Split%201938%20to%201950%20Summary.xls
Perhaps if you named your region and date frame modeled, others could help. Jim Dick - Roseville, MN