OK, I finally had a little time to play with WinISD so I'm going to take a stab at answering my own question.
I'm going to use the CSS SDX15 as an example using 3 different models: the default 7cuft ported tuned to 19.31Hz (Yellow) which is Fs, 9 cuft tuned to 17Hz (grey) and 11cuft tuned to 15Hz (blue).
My understanding is that this is where I would look at the tradeoff of headroom vs extension. Looking at these three graphs, I have little or no headroom loss in upper bass (above 50Hz??). I loose ~1db between 25Hz and 35Hz with the 17Hz tuning, and loose 1-2db between 21Hz and 46Hz.
In return, I get an extension of flat room response from say 17Hz down to about 13Hz (assuming 5db room gain). 2db of headroom for 4db of extension seems to be a good tradeoff.
Now lets look at excursion, and by extension distortion (maybe??). I am going to make an asumption here that I can get linear excursion to about 80% of Xmax. I don't know if that is realistic or not.
So, 80% of 30mm = 24mm and look at that, we get clean excursion all the way down to 16.6Hz with the tuning at Fs. With a tuning of 17Hz, we exceed maximum linear extension at 28Hz down to 21Hz. Thats definately in audible territory. Similarly, we exceed linear excursion between 30 and 17Hz with the 15Hz tuning. The amount of excursion for a given frequency (and thus distortion) also increases.
So, how did I do? Is it E for effort of B+ for [ahem] brilliant deduction?
If I got it more or less right, the question is, how much distortion does it really add and is it audible?