<html>
<head>
<meta http-equiv="Content-Type" content="text/html; charset=iso-8859-1">
<style type="text/css" style="display:none"><!-- p { margin-top: 0px; margin-bottom: 0px; }--></style>
</head>
<body dir="ltr" style="font-size:12pt;color:#000000;background-color:#FFFFFF;font-family:Calibri,Arial,Helvetica,sans-serif;">
<p>Dear all,<br>
</p>
<p><br>
</p>
<p>I recently had some trouble with numerical effects in a spherical shock wave simulation.<br>
</p>
<p>I was getting some strange numerical fluctuations, and was uncertain of the cause.<br>
</p>
<p><br>
</p>
<p>It was suggested to me to use FLASH's Grid_MarkRefineDerefine to force an initial refinement as opposed to introducing random temperature fluctuations.<br>
</p>
<p><br>
</p>
<p>After some experimentation with this, I seem to have found a reasonable solution, as well as a better idea of the problem.<br>
</p>
<p>Still, there is something strange about some of my results.<br>
</p>
<p><br>
</p>
<p>I added an INRADIUS refinement criteria at 5 levels of refinement, out to a radius of 1e15 cm.<br>
</p>
<p>With just 5 AMR levels in the simulations, this meant that the simulation effectively did not refine at all.<br>
</p>
<p>In this case, the results were fine (image: 5layer_15d), though there were some small bumps due to the low resolution.<br>
</p>
<p><br>
</p>
<p>Now, I wanted to add additional levels of refinement, which is where things get interesting.<br>
</p>
<p>I ran two simulations: the first was just with an extra level of refinement on top of the 5 layer INRADIUS, while the second also added a 6 layer INRADIUS up to 2e14 cm.<br>
</p>
<p>This meant that the second simulation would not have any additional refinement (or derefinement) until the shock reached past 2e14 cm.<br>
</p>
<p><br>
</p>
<p>The results are shown as 6layer_15d and 6layer_INRADIUS_15d, respectively.<br>
</p>
<p>While the 6 layer simulation with the additional INRADIUS refinement looks good, even beyond the boundary of 2e14, the simulation where I just added an extra level of refinement looks terrible.<br>
</p>
<p><br>
</p>
<p>Based on this, I think refining and derefining in early stages of the simulation causes the problem. The density waves in the image occur at early times and then just propagate.<br>
</p>
<p>Any simulation that does not allow refinement at early times shows good results.<br>
</p>
<p><br>
</p>
<p>While I consider my problem largely solved, I would still wonder why the refinement causes these large pressure waves to occur.<br>
</p>
<p>The outward motion of the matter leaves a minimum temperature region behind, as well as a steep pressure gradient, so that might be part of it.<br>
</p>
<p><br>
</p>
<p>Kind Regards,<br>
</p>
<p>Michiel Bustraan<br>
</p>
</body>
</html>