It's a little math and a little "seat of the pants" experience. The wattage is pretty universally in the 2-3 watt range, and you're looking for 5-8 watts of power dissipation in the smoke unit. In the simplest form, the input to the smoke triac would be a pure sine wave.
Consider 18 VAC sine wave from a transformer. The P-P voltage is 1.414 times the RMS value of 18V. So the RMS value of the half wave computes to 12.726 volts. In general, I consider the smoke resistor to be receiving 12 volts and work from there. However, that was really only the starting point. My real limitation with TMCC is the smoke triac on the R2(4)LC board. I tuned my resistor calculations to the triac running at 75C or less after a few minutes of running. That ended up being 20 ohms at 18VAC input to the locomotive. If you use 12 volts and 20 ohms, you compute .6A through the smoke triac. I'm assuming the triac drops .7 volts in half-wave operation.
The problem is, most modern electronically controlled transformers don't output a pure sine wave, so the computations get a little more complex. That's where the seat of the pants comes in, but I'm still using 20 ohms as my self-imposed minimum smoke resistor value with TMCC smoke units.
