> do I use the plate resistor for the load
If the tube were a dead-short, this would be valid.
If the tube were a dead-short, you could replace it with a much cheaper wire and get the same result (no signal). That aint right.
> the plate resister added to the internal plate resistance of the tube?
No. This is closer, but the data-sheet Rp is not given for the same conditions you will work at, varies a lot, and is small-signal which gives a modest error on triodes but is way-wrong for pentodes.
You want a rough analysis of your tube stage. Will the plate sit all the way at B+? No. Will it be stick to ground? No. The most general assumption is that it will be "halfway". i.e., if B+ is 400V, the plate will be around 200V. And if plate resistor is 100K, and has 200V across it, then clearly the current is 200V/100K= 2mA. And if resistor+tube drop 400V and the same 2mA flows in both, then the whole stage is 400V 2mA and therefore "acts like" a 200K resistor.
And.... in self-bias stages, this approximate equivalent resistance will stay nearly-the-same over a _wide_ range of B+. With same bias and load resistors, at 200V total the plate will sit somewhat higher than "halfway" and the equivalent may be nearer 240K.
Looking at many-many designs, plate is usually in the range from 40% to 70% of B+. The canonical Fender 12AX7+1K5+100K stage sets its plate near 69% of supply voltage. The stage acts like 320K. Lower-Mu tubes are sometimes biased with plate below half. Effective resistance is 1.5X to 2X the load resistor.
And all tube-work is approximate.
Take some rough assumption and run with it.
When you don't know better, double the load resistor.
> And if there is more than one tube
Take each stage and parallel them. One Fender-stage, 320K. One fat triode cooking 22K, say 40K. 320K||40K= 35K. That's "R2" in PSD.
Note the "R2" designation changes as you add filter stages.
Note also that the 320K || 40K calculation is not very different from assuming just the 40K. When one stage is much bigger than the others, it is often sufficient to estimate that and then round-down to cover the small crumbs.