[FLASH-USERS] "jet-by-hand" in unsplit

[dongwook at flash.uchicago.edu]

Hi Hal,

What kind of Riemann solver in the unsplit solver are you using for this
simulation?
It seems to me that you have some low density regions in your simulation,
which
could be vulnerable for unphysical negative states in density (or pressure).

To alleviate such situation, you may want to use HLL type of Riemann solver,
rather than a linearized Roe solver. HLL families of Riemann solvers preserve
solution positivity, which is not guaranteed in the Roe solver.

There are two HLL solvers in the unsplit hydro solver in FLASH 3.2: HLL
and HLLC.
HLL solver is quite diffusive and I suggest you use HLLC for accuracy and
stability.

Please let me know how it goes with HLLC solver, if you haven't tried yet.

Also, it will be very useful for me to give you more suggestions if you
provide
flash.par and log file you used.

Best,
Dongwook


>    Hi fellow flash-users,
>
>    I have been using FLASH 3.2, and the PPM split solver,
> to simulate the evolution of the bound remains of a
> tidally-disrupted star around a black hole (in cartesian
> 3D).  I'm currently only interested in the aftermath, so
> the stellar remains enter my simulation via source term :
> a rectangular yz-plane (1 cell thick), of dens., temp.
> (20K)*, and x-velocity values (~ -1e9).  This is set
> inside a frame of buffer cells given ambient values to
> prevent spurious backflow.  Maximum source--ambient
> density contrast is typically 12 orders of magnitude in my
> simulations (10^-20 vs 10^-12)...
>    So far this source term has worked nicely using the
> gamma-law or even the Helmholtz (with some complaints)
> EOS.
>    However, when I switch to the unsplit solver (with
> gamma-law), I notice that after a few iterations, cells
> with smalle values and opposing velocities form at the
> leading edge of the jet and then spread everywhere...
> *I have tried using the unsplit "shock fixes" but saw no
> effect
> *I have tried reducing the density contrast (in hopes of
> smoothly ramping it up) -- which didn't work by itself,
> nor with increased or decreased values of the eintswitch
> toggle.  Just to be thorough, removing the point mass (the
> only gravity here) does not remove the error.
> *I also found that inputting density at the maximum
> contrast with no velocity seems to cause no problems
>    I tried to be comprehensive yet brief, so let me know
> if you'd like more detail.  Thanks for your time,
>
> Hal
> hcambier at ucsc.edu
>
> * such low T was just chosen to keep the problem as
> initially 'ballistic' as possible
>
> (Figures of upper xyz corner of source & buffer cells)
>
> /|\+y
>   |
> o o o o
> o o o o
> o o o o
> o o o o
> x o o o
> x o o o ----->+x
> ...etc...
>
> /|\ +z
>   |
> o o o o o o o o
> o o o o o o o o
> o o o o o o o o...etc...
> o o o o o o o o
> o o o o x x x x
> o o o o x x x x ---->-y
> ... etc ...
>