Ralph P. Johnson, "The steam locomotive" is a design book. It would be your best source of information. Secondhand available, I have the fourth reprint of 2002 myself.
Kind regards
Jos Koopmans

If you read last year's article in Trains, The Amazing Loco Valve Pilot, you can also learn a lot about how operating practice affected fuel consumption.

By the way, this is still true today. How an engineer handles the throttle affects fuel or electricity consumption even with modern power.

You are absolutely correct about the skill of the engineer for fuel consumption. In the late 80's I was involved with a fuel efficiency test with Conrail. This was the Lube Train that ran the B&A from Selkirk to Framingham 42 times. We were testing the Willy Vogel flange lubricators. There were 3 series of tests. The first was 14 trips for the base case and the next was 14 trips with on-board lube and wayside activated. The last was with the wayside turned off.

I had divided the line into segments so that we would have many more data points and we could have tests uphill, flat, and through curves. The 14 tests were determined by calculating a confidence level so the statistics would support the results.

Obviously, to avoid introducing errors, we used the same engine (SD-50 6703), CR 20, plus 18 hoppers, and a caboose. The 18 hoppers were stored in the local yard AND NOT TOUCHED during the tests. This meant that we had the same weight. We also measured wheel wear. The test car and the locomotive instrumentation that we used consisted of a FARR work and Power meter (a massive shunt) to measure kilowatt hours and we also had our own DAY TANK recirculating fuel oil measuring system. The measurement was with the "make-up fuel" to went to the "day tank". (Named for the inventor and not that it held a days worth of fuel". In run 8- the pump would be on constantly and the positive displacement meter would be running. I had to pick spots for the test where we took measurements based on when the throttle would be in idle and the tank would have been filled. We also took time measurements to the second and had markers on the route to mark the test locations. (A white triangle with a T.)

Now to the essence of the post. We used Elmo and Mike for the engineers. I can tell you without a doubt that Elmo was so consistent on the first leg that he was within 1 kilowatt hour every time. This was such an important part of the test as it reduced the variance for testing and reduced the number of trips we needed to make.

As a result, we published our results and had a number of other railroads install the same equipment. We saved 5% fuel and also significantly reduced wheel wear on the locomotive wheels. Amtrak also adopted. I made a presentation to the 1986 IIE society meeting. After this initial test, our test lab continued with other tribology testing.

Could I convince you to provide a link from which to download a copy of those results in detail?