[FLASH-USERS] 2D Spherical and Cartesian Geometry

[Michiel Bustraan]

First of all, thank you so much for the quick response.

Secondly, I did read the section in the userguide on geometries, and I understand that different geometries handle block volumes differently.
But even so, shouldn't it be possible to simulate a 2D slice of a spherical shock using Cartesian geometry (say, a 2D simulation of the equator)?
Or does the nature of the Cartesian geometry make it fundamentally inappropriate for this kind of simulation?

Also, I looked at the the SodSpherical example, as you suggested, and it runs fine.
So clearly whatever's wrong with my simulation is related to the setup or the runtime parameters.

-Kind Regards,
Michiel
________________________________________
From: Klaus Weide <klaus at flash.uchicago.edu>
Sent: Friday, September 23, 2016 3:49 PM
To: Mordecai-Mark Mac Low
Cc: Michiel Bustraan; flash-users at flash.uchicago.edu
Subject: Re: [FLASH-USERS] 2D Spherical and Cartesian Geometry

On Fri, 23 Sep 2016, Mordecai-Mark Mac Low wrote:

> Dear Michiel,
>
> On your second question: are you sure that you’re not setting up a
> cylindrical blast wave, rather than a spherical one?  A 2D, slab
> symmetric, point explosion is effectively that of an infinite line
> charge, rather than a spherical, point explosion.  The line charge
> indeed expands faster.  There is, of course, an exact analytic solution
> for this. For an analogous case with a constant energy input, see eqs.
> 5-7 of Mac Low, McCray, & Norman, ApJ (1989). Therefore, you may be fine
> sticking with your Cartesian grid, so perhaps you don’t need to solve
> your first problem.

Yes indeed, to the above.

There is also a subsection in the FLASH Users Guide with some discussion
related to the meaning of 2D simulations, which might be useful to some:

  8.11 Grid Geometry
  8.11.1 Understanding Curvilinear Coordinates


Michiel, you may also want to refer to the provided SodSpherical example.
1) Note that Y ranges only from 0 to 90 degrees, with reflecting
boundaries in Y direction.
2) Note that the *physical* geometry set up in the SodSpherical problem
is cylindrical - comparable to a line charge - but the coordinates used by
FLASH are spherical, thus in the parfile:

  # Grid geometry
  geometry = "spherical"



Klaus