[Mesa-users] Problems with white dwarf accretion

[Francis Timmes]

experiment with use_gold_tolerances = .false.
and/or decreasing varcontrol_target from 1.0d-3 to 1.0d-4.

fxt





> On May 23, 2019, at 1:40 PM, Liam Kroll via Mesa-users <mesa-users at lists.mesastar.org> wrote:
> 
> Hello Mesa-users,
> I'm currently attempting to simulate accretion of brown-dwarf-like material onto a series of CO white-dwarfs and a single ensuing thermonuclear runaway using mesa-r11701 with the latest MESA-SDK (20190503) on a machine running Ubuntu 18.04.2 LTS. The ultimate goal of this project is to obtain full temperature/density profile trajectories from the beginning of thermonuclear runaway to a (for now) arbitrary time when nuclear burning in the expanding envelope has subsided. So far, I have been using the make_co_wd test suite inlists to create the WD model I am starting with. I have been using this test_suite example completely unchanged save for the initial mass (2.75 Msun) of all inlists and the stopping condition (log_center_temp_lower_limit = 7.18 (~15MK)) of inlist_part3, which outputs the final WD model. To model the accretion and nova I am using inlists which are mostly based off of test_suite examples which model accretion onto a WD. Inlist_init loads the WD model created by make_co_wd and changes to a custom reaction network. Inlist_nova then begins accreting material onto the model. However I am running into some issues when trying to simulate a nova outburst. When the WD starts accreting material the timesteps have trouble increasing over roughly 2<lg_dt_yr<3 because MESA seems to be stuck in a vicious cycle of max_increase/retry. I have tried a few things which I thought might help, some of which did help but not eliminate the problem completely:
> -Increasing grid resolution, especially where there is burning occurring.
> -Turning on the report_ierr control, I received a message very similar to the one reported in the MESA-users archives: https://lists.mesastar.org/pipermail/mesa-users/2019-April/010002.html Following the suggestions therein and playing with which eos/form of the energy equation is being used.
> -relaxing the tau factor to both low and high values (0.0001-300). Many of the test_suite examples modelling novae do this, and as far as I can tell MESA proceeds smoothest for my models at a value of ~0.01. As I understand it, this value means more of the atmosphere is included in the model when compared to the default value of 1.
> I have included the starting model and inlists as well as some plots of the output from inlist_nova. To me it looks like MESA runs into trouble around model number 125, where both the number of retries and newton iterations spike. Looking at the entropy profile we can see that there is a discontinuity around this time in the outer layers of the model. I took a look at the mass fraction of h1 and indeed this discontinuity in the entropy coincides with the region where material is being accreted. Perhaps the composition gradient is an issue here? I was wondering if anyone might have any input or experience with similar problems? I apologise if similar issues have been discussed on the archives, but I am unable to find them. I am relatively inexperienced with MESA and certainly a novice when it comes to the theory of stellar modelling, so I'm hoping the hurdle here is just my ignorance and not a fundamental issue with the problem I am trying to model.
> Thanks for any and all feedback,
> Liam Kroll
> P.S. I modified run_star_extras but I don't think that is the issue here, as the problem persists even when I do not call the added routines.
> <h1_v_zone.png><entropy_v_zone.png><newton_iter_v_model.png><retries_v_model.png><dt_v_model.png><part3.mod><inlist><run_star_extras.f><CK_post.net><CK_pre.net><inlist_init><inlist_nova>_______________________________________________
> mesa-users at lists.mesastar.org
> https://lists.mesastar.org/mailman/listinfo/mesa-users
>