firstly, i think you'd have to stipulate there would be no IMD for ease of analysis. and that Kms and Le are flat -- thus only Bl related distortion. further, distortion is only related to F(x) and not F(x,t), because that would be more difficult to analyze.
it seems you should be able to do this with basic calculus. find average Bl in a range to find the theoretical power without distortion in that range. then make a curve of (BL(x) - avg) and use that to find the power in distortion.
if BL is a linear function for output -- eg if you could hold everything else equal and double BL to get double acoustic intensity then:
distortion = 100 * integral ( (BL(x) - avg)^2 ) / (Xmax * (avg)^2)
avg = integral (BL(x)) / Xmax
both times integrateing over -xmax/2 to +xmax/2
i think that works out correctly.
of course Kms will also add distortion.