kslong
commented
6 years ago I was worried that the problem was obscure in the sense that there were a lot of contributing factors, but here is a clue to what is going on. This is for a temperatures 30,000 K for the star --> the system should have a radiated luminosity of 2.8e32.
This model is well converged. Notice that the heating and cooling balance pretty well, but the absorbed flux and luminosity, do not. The luminosity is higher. (Note that the two quantities differ, because absorbed flux and luminosity include recombination edge energies, whereas heating and cooling, which refer to the heating and cooling of the electrons, do not.)
Specifically, here the star has a luminosity of 2.8e32 and the system has a luminosity of 4.2e32 or an excess of 1.4e32, this should be compared to the difference in wind luminosity (at photon creation) of 1.38e33 vs the absorbed flux of 1.22e33 or 1.6e32, which is slightly greater (and which would partially be lost photons escape the wind).
To me this suggests this may have something to do with how we are treating photoionization and recombination. I recall that we are using recombination rates that are derived from literature rather than detailed balance. Perhaps this is the root of the problem, since for simple atoms, recombination photons are all made to the ground state, whereas for the purpose of the rates we are using all of the levels. If we are not making some kind of correction to account for this, we may have a problem. @Higginbottom you are the expert on this.
Note that if you run the above model at other temperatures it is less obvious that recombination is the root cause of the problem.
Higginbottom
jhmatthews
ssim
The funny spikes below 1e15 Hz are the excited level recombination - as you can see they escape the wind.
lazygun37
Reducing the density by an OOM means that the plasma is all ionised, so despite the still high densities, it is transparent to ionising photons and the model behaves itself.
With macro atoms, photons are created from external sources and escape the system with the same weights as initially. This assures that (in the absence of errors) that the luminosity of the system is that of the external sources.
With simple atoms, photons change their weights as they pass through the system through free-free, free-bound process, etc. To compensate for this, photons are created in the wind. The idea is that by balancing heating and cooling in the wind, we will make up the energy lost by creating additional photons in the wind that just compensate for absorption.
However, that is not currently happening. Instead in the cases I have been experimenting with, e.g.
simple.pf.txt
The radiated luminosity exceeds the luminosity of the (only) external source, the star, by a considerable amount. I have tried various permutations on this model, higher stellar temperature, more wind cells, etc. and though the percentage varies, the sense of the difference remains the same. This seems to me a fairly fundamental issue @ssim @Higginbottom @jhmatthews and I think we need to understand it. (I'm not sure why we have not noticed this previously, but I could imagine it has something to do with some of the problems we had with cells near the disk surface in the X-ray binary work. It came up now, because James and I have been looking more carefully at the effects of adiabatic cooling and extra heating on wind emission).