As others have pointed out, having the valves properly set on an engine is necessary for efficient operation and for the general health of the engine as a whole. A steam locomotive tends to try to tear itself to pieces just by its nature and the valve timing being off makes it even worse.
There is sometimes the impression that valve setting is some sort of a black art but it is really just mechanics and getting all the parts adjusted to the correct dimensions. That being said, there are some methods of valve setting that can be used when the entire process cannot be undertaken for some reason that do require making some educated guesses based on experience. The D&RGWs standard practices called for setting the valves square with the engine in forward gear at 50% cut off. Their practice seemed to be to just tweak whatever parts were easiest to adjust to achieve this which resulted in some pretty awful sounding engines when run at other than that particular notch, but it was close enough for the most part.
Kelly listed the basic process of valve setting. I will go into a some of the actual work as we are getting into it in the C&TS shop.
One of my goals since rejoining the C&TS a year and a half ago has been to improve the timing of the engines. To that end, I designed, and we built a set of timing rollers last winter to make the process easier. More or less a treadmill for a locomotive. These allow us to turn the main driver and run the engine through all the valve events with it sitting in one place in the shop.
The complete set of rollers. Note the radius rods and combination levers off the 488 in the background. They have been repaired and are awaiting installation.
489 on the "treadmill"
First of all, the valve motion has to be put in good condition, which requires a considerable amount of machine work. It would be a waste of time to attempt to set the valves if there is a lot of wear and lost motion in the valve gear. We have discovered that, over the years, some of the dimensions of the valve motion parts have not been maintained to the exact specifications of the prints so adjustments have to be made to bring them back into spec. The valve spools, rings, and cages must be checked for condition, proper dimensions, and adjusted if necessary. Once everything is known to be in good condition we assemble it all, with the exception of the side rods, and put the engine on the rollers and begin the process of setting the valves.
We started with the 489 last winter and have already done much of this work on 488 as well. One variation in the process is in adjusting the valve lead, step #4. If the radius rods are fairly close to the proper length to start with and you are working with an engine with spool valves, the valves can be shimmed on the valve stems rather than having to lengthen or shorten the radius rods themselves. Sadly, I am not aware of an easier way to adjust the lengths of the eccentric rods.
Eccentric rods after adjustment.
Although it is pretty close, the timing on the 489 did not come out quite perfect due to some issues with the valve cages and valve spool length on one side that we were unable to correct last spring. We hope to have time to make some more improvement this winter. Our plan is to rebuild the valve motion and set the valves on the other three K-36s this winter. If all goes well, we should be able to get them pretty well squared up for next season.