[Mesa-users] Issues with mass loss from 20Msun star

Hannah Brinkman brinkmanhe at gmail.com
Sat Mar 14 07:40:48 EDT 2020

Dear Bill,

There are quite a few changes between the two versions of MESA and it is
complicated to get the inlist working in both version, also because a few
changes seem to involve the physics, such as the atmosphere, of the models.
So far I have not been successful in making an inlist that works for both
versions. Since it was also pointed out to me that the models I sent
earlier might not be fully resolved, I will continue down that line first
and hope a better resolution will solve the issue with the mass loss.

Thank you for your responses and advice.
With kind regards,

Op ma 9 mrt. 2020 om 22:02 schreef Bill Paxton <paxton at kitp.ucsb.edu>:

> Are you willing to do the work to get the same case running in both 10398
> and 12115 so we have something to compare?
> If not, then have fun with your new stuff from Frank.
> If so, then i’m happy to work on this with you.
> -b
> On Mar 9, 2020, at 2:57 PM, Hannah Brinkman <brinkmanhe at gmail.com> wrote:
> Dear Bill, dear Josiah,
> Thank you for your responses and suggestions.
> First of all, the difference of a factor of about 4 was because I was
> comparing two models that should not have been compared to each other, one
> with exponential overshoot and one with step overshoot. I did another run
> with the stepovershoot and now the difference is only a factor of 2, the
> model in version10398 losing about 10Msun and the one in version 12115
> losing about 5Msun. The  differences in the HRD, see attachment, are larger
> than I recalled this morning. However, it still seems to me that the
> difference in mass lost is too large for the change in the settings.
> Secondly, currently I am using an inlist given by Frank Timmes so I can
> get preSN models in which I changed the overshoot, thermohaline, and
> semiconvection parameter, as well as the wind prescription to what I had
> before. With the inlist I have used for version 10398 I could not get to
> the preSN phase. I think the inlist from Frank can be downgraded to version
> 10398, but I am not sure. The most important differences between the two
> inlist is that the gold_tolerances have been switched off and the
> use_eosELM as well. These were switched on in my earlier run. The other
> differences are mostly in the timestep controls, which I realise now, might
> have more impact than I previously thought. Another change is the size of
> the network, from 63 down to 21 because I was mostly running tests to see
> if I could get preSN. This definitely has some influence on the mass loss
> too, but the large factor still puzzles me. The original run done with
> version 10398 could not go to the preSN phase, also because my interest was
> only up to the onset of carbon burning. The differences in the effective
> temperature on the MS are due to a change from NACRE (10398) to JINA rates
> (12115).
>  I have attached the HRD showing three runs, the original done with
> version 10398 and two done with version 12115, one with exponential
> overshoot and one with stepovershoot. A plot of the change in mass is also
> attached. If any other plots are useful, please let me know, but I might
> not have all the information in my current history files and will have to
> rerun the simulations.
> I hope this helps with figuring out what is going on with the mass loss.
> With kind regards,
> Hannah
> Op ma 9 mrt. 2020 om 16:45 schreef Bill Paxton <paxton at kitp.ucsb.edu>:
>> here are some suggestions for items to add to your history_columns.list
>> if they aren’t already there.
>>       log_star_age
>>       log_dt ! log10 time_step in years
>>       star_mass ! in Msun units
>>       log_abs_mdot ! log10(abs(star_mdot)) (in msolar per year)
>>       log_L ! log10 luminosity in Lsun units
>>       log_R ! log10 radius in Rsun units
>>       log_surface h1
>>       log_surface fe56
>> -b
>> > On Mar 9, 2020, at 9:34 AM, Bill Paxton <paxton at kitp.ucsb.edu> wrote:
>> >
>> > the inlist you sent runs to core collapse, but for the current
>> question, we should change that to stop when the mass loss is (mostly)
>> complete, just long enough to show the problem.  any convergence runs you
>> do for this can stop at that point and save some time.
>> >
>> > b
>> >
>> >
>> >> On Mar 9, 2020, at 9:15 AM, Bill Paxton <paxton at kitp.ucsb.edu> wrote:
>> >>
>> >> Hi Hannah,
>> >>
>> >> I’m curious about this too.  So i’ll checkout those versions and try
>> running your inlist and run_star_extras.  Will they work in both mesa
>> versions?   if not please send me something that does.
>> >>
>> >> You could also help clarify this question by checking convergence for
>> your runs on both systems - is the change in mass loss relatively small
>> when you make ‘large’ changes in the time or space resolution (large
>> meaning ~50% or so)?  as part of this, please make plots of mdot, R, L, M,
>> number of zones, and number of timesteps vs. log age.  are they converged?
>> do they look similar for both mesa versions?  also, include any other
>> surface properties that your mass loss routines consider - e.g.,
>> Fe_div_Fe_solar for the outermost cell.   i’d be interested in seeing all
>> of those plots.   they will go a long way toward telling us what has
>> changed for the new mesa version.
>> >>
>> >> my main interest will be on changes in converged results, since for
>> unconverged results the details of the numerics play a large role.  so
>> please send along your results showing convergence so we know we are
>> comparing “physics" rather than implementation details.    if the converged
>> results have changed by such a large amount then we really have something
>> interesting to consider!
>> >>
>> >> let me know what you find.
>> >>
>> >> cheers,
>> >> bill
>> >>
>> >>
>> >>
>> >>> On Mar 6, 2020, at 7:29 AM, Hannah Brinkman via Mesa-users <
>> mesa-users at lists.mesastar.org> wrote:
>> >>>
>> >>> Dear all,
>> >>>
>> >>> Recently I have been trying to make pre-SN models for a 20Msun star,
>> using the inlist given by  Frank Timmes a couple of weeks ago. The inlist
>> works fine when using the settings given in the inlist. However, I have
>> encountered two issues around the mass loss. The first issue is, that when
>> using the mass loss as given in the inlist, about 11Msun for this star,
>> which is a bit much. And when switching from the exponential overshoot
>> given in the inlist to the step-overshoot I have been using before, the
>> winds become so strong that the star loses 13Msun, which is the entire
>> hydrogen envelope of this star, which should not happen yet for this mass.
>> >>> The second issue that I have encountered is that the wind scheme in
>> the run_star_extras.f has the opposite problem. Since updating from version
>> 10398 to version 12115, the custom wind scheme, following Schootemeijer et
>> al. 2018, is no longer working right, making the star lose only 2.5Msun
>> instead of the expected ~10Msun which it did in version 10398. However, I
>> cannot find what might be the problem here.
>> >>>
>> >>> Does anyone have any idea where these issues might come from?
>> Attached is the inlist I have used and the run_star_extras.f.
>> >>>
>> >>> Thanks in advance for any suggestions.
>> >>> Cheers,
>> >>> Hannah
>> >>>
>> <inlist_project><run_star_extras.f>_______________________________________________
>> >>> mesa-users at lists.mesastar.org
>> >>> https://lists.mesastar.org/mailman/listinfo/mesa-users
>> >>>
>> >>
>> >
>> <MassEvol.png><MassLossHRD.png>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.mesastar.org/pipermail/mesa-users/attachments/20200314/088372eb/attachment.htm>

More information about the Mesa-users mailing list