Based on the StringFast prescription detailed here, we are forecasting on the fraction of power in strings at l=10 (in the same manner as the Planck collaboration)
The additional string spectrum parameters (wiggliness, string velocity, correlation length) are kept fixed for now but will be varied in future runs.
The relevant string model values are chosen to be:
gmu = 3.5e-6 (string tension)
alphastring = 1.05 ("wiggliness" parameter which encodes small-scale behaviour of the string network)
velstring = 0.4 (rms string velocity)
clstring = 0.5 (correlation length)
The string tension is chosen to normalise the contribution to have f10=1 at l=10, and then deviations from this are parameterised through the fraction. There is a mild dependence on the optical depth in this model which needs a little more attention, and checking the relative important of the scalar, vector and tensor contributions to the overall spectra.
The important feature of this model is that it predicts contributions to the polarisation power spectra as well as the temperature power spectra, and so these are added to the TT, EE and BB spectra.
The Planck bounds agree reasonably with predictions in the above paper, f10=0.02 which yields a string tension of Gmu < 6.8e-7 at 2 sigma, and the nominal S4 constraints are a factor of 10 better (again consistent with the projections above).
The variation in constraints with the resolution and sensitivity is here File:String tension sensitivity resolution.pdf
The next step with these constraints is including/comparing these constraints against other string spectra (particularly for the temperature-only case).