On Daniel's suggestion, I have made some projections for how we can expect the interferometer range to improve if we double the power and/or fix the low-frequency power-law noise.

The first attachment shows various projected curves at the current power (23 W), and the second attachment shows projected curves at twice the power. The ranges are computed using Salvo and John M's new version of int73.

I've assumed the unknown noise has an ASD of 1×10⁻²⁰ m/rtHz at 100 Hz, with a shape of 1/f². This is just done by eye, with the magnitude chosen so that the subtraction does not force the remaining noise to dip significantly below the "total expected" trace.

It seems that either fixing the power-law noise or doubling the power gives similar results for both BNS range (for two 1.4M_⊙ stars) and BBH range (for two 30M_⊙ holes). Doing both gives us a bigger boost.

From the curve in which the power is doubled and the low-frequency noise is fixed, it is apparent (if you believe the subtraction) that there is still some irregular excess noise (perhaps scatter) between 60 and 100 Hz. This actually seems to limit the range significantly; looking at the red traces, the "total expected" trace has ranges of 110 Mpc / 2.1 Gpc, and the "total expected" trace at twice the power has range of 135 Mpc / 2.7 Mpc. Additionally, some of the noise below 50 Hz is LSC control noise, which can be reduced by implementing better subtraction.