<html><head>
<meta http-equiv="Content-Type" content="text/html; charset=utf-8">
  </head>
  <body>
    <tt><font size="+1">Dear Antoine -<br>
        <br>
        You do not want to change the CFl factor, but you want to
        control the source term actions better. In the case of
        catastrophic cooling, the local cooling time becomes shorter
        than the dynamical timescale. This is conceptually identical to
        the nuclear burning/heating timescale issue.<br>
        <br>
        If there is no timestep limiter due to cooling, implement one
        following the approach used in Burn_computeDt (or whatever the
        module name is). You should be able to stabilize calculations by
        limiting the source term contribution to about 10% of the
        internal energy.<br>
        <br>
        It would be nice to hear if that helped.<br>
        <br>
        (BTW, I do not recommend using artificial viscosity. This
        approach is largely obsolete these days and helpful only in very
        particular situations.)<br>
        <br>
        Tomek<br>
        --</font></tt><br>
    <div class="moz-cite-prefix">On 2020-10-22 10:50, Antoine Gintrand
      wrote:<br>
    </div>
    <blockquote type="cite" cite="mid:CAO_-DzROTMPdj9tOG352FREH_83YX_QpKejLOOP-5t9i-V-arA@mail.gmail.com">
      
      <div dir="ltr">
        <div>Dear flash users,</div>
        <div><br>
        </div>
        <div>I have an issue that has been troubling me for a long time.
          <br>
        </div>
        <div>I do the simulation of a strong shock in the interstellar
          medium that cools <br>
        </div>
        <div>with a source term according to <span class="gmail-aCOpRe"><span>the
              cooling tables of <em>Sutherland</em> & <em>Dopita</em>
              (1993).</span></span></div>
        <div><span class="gmail-aCOpRe"><span><br>
            </span></span></div>
        <div><span class="gmail-aCOpRe"><span>the problem comes when the
              cooling becomes strong and I receive the following message
              :</span></span></div>
        <div><span class="gmail-aCOpRe"><span><br>
            </span></span></div>
        <div><span class="gmail-aCOpRe"><span><span class="gmail-aCOpRe"><span> 
                       n          t                      dt
                                  (         x,          y,          z) |
                                     dt_hydro  dt_Cool   CFL     </span></span></span></span></div>
        <div><span class="gmail-aCOpRe"><span>                   <br>
            </span></span></div>
        <div><span class="gmail-aCOpRe"><span> 253369 8.7315E+09
              2.1942E+03  ( 1.773E+16, -8.164E+15,   0.00    ) |
               2.194E+03 4.696E+04 0.1000000<br>
                253370 8.7315E+09 2.1958E+03  ( 1.773E+16, -8.164E+15,  
              0.00    ) |  2.196E+03 4.687E+04 0.1000000<br>
                253371 8.7315E+09 3.0850E+01  ( 1.864E+16, -7.383E+15,  
              0.00    ) |  3.085E+01 4.678E+04 0.1000000<br>
                253372 8.7315E+09 2.4375E-01  ( 1.864E+16, -7.383E+15,  
              0.00    ) |  2.438E-01 4.678E+04 0.1000000<br>
                253373 8.7315E+09 2.4375E-01  ( 1.864E+16, -7.383E+15,  
              0.00    ) |  2.438E-01 4.678E+04 0.1000000<br>
                253374 8.7315E+09 2.4375E-01  ( 1.864E+16, -7.383E+15,  
              0.00    ) |  2.438E-01 4.678E+04 0.1000000<br>
                253375 8.7315E+09 1.0000E-10  ( 1.864E+16, -7.383E+15,  
              0.00    ) |  1.745E-16 2.569E+02 0.1000000<br>
                253376 8.7315E+09 1.0000E-10  ( 1.864E+16, -7.383E+15,  
              0.00    ) |  1.289E-60 2.569E+02 0.1000000<br>
               dtCheck=   0.0000000000000000     <br>
               DRIVER_ABORT: [Hydro]: Computed dt is not positive!
              Aborting!</span></span></div>
        <div><span class="gmail-aCOpRe"><span><br>
            </span></span></div>
        <div><span class="gmail-aCOpRe"><span>and I can see that the
              pressure drops dramatically so the movement of matter
              accelerates and the compression increases drastically.</span></span></div>
        <div><span class="gmail-aCOpRe"><span> I believe that this is
              maybe due to the catastrophic instability of Falle (1981).</span></span></div>
        <div><span class="gmail-aCOpRe"><span>Indeed, the cooling time
              becomes smaller than the time needed for the sound wave to
              travel the cooling region <br>
            </span></span></div>
        <div><span class="gmail-aCOpRe"><span>and the pressure could not
              be stable anymore so it begins to fall inside this region.
              <br>
            </span></span></div>
        <div><span class="gmail-aCOpRe"><span>The problem is that during
              the process, the region becomes poorly resolved and the
              gradient of pressure density and temperature are very
              high.<br>
            </span></span></div>
        <div><span class="gmail-aCOpRe"><span>Also, the problem is that
              when I increase the resolution, the code fails at earlier
              times because the cooling and the compression are even
              more efficient. <br>
            </span></span></div>
        <div><span class="gmail-aCOpRe"><span>When I chose a lower cfl
              coefficient it would just delay the fail message.<br>
            </span></span></div>
        <div><span class="gmail-aCOpRe"><span><br>
            </span></span></div>
        <div><span class="gmail-aCOpRe"><span>One idea that seems to
              work would be to use some artificial viscosity but I do
              not know if this would be consistent and the physical
              meaning of this.<br>
            </span></span></div>
        <div><span class="gmail-aCOpRe"><span>Indeed when I increase the
              coefficient of artificial viscosity (cvisc),  at some
              value of the parameter (cvisc=1.5), the simulation will
              continue to run but the time step drops a lot.<br>
            </span></span></div>
        <div><span class="gmail-aCOpRe"><span><br>
            </span></span></div>
        <div><span class="gmail-aCOpRe"><span>Is anyone have some ideas
              to solve this issue,</span></span></div>
        <div><span class="gmail-aCOpRe"><span><br>
            </span></span></div>
        <div><span class="gmail-aCOpRe"><span>Thank you for your help,</span></span></div>
        <div><span class="gmail-aCOpRe"><span><br>
            </span></span></div>
        <div><span class="gmail-aCOpRe"><span>Best regards,</span></span></div>
        <div><span class="gmail-aCOpRe"><span><br>
            </span></span></div>
        <div><span class="gmail-aCOpRe"><span>Antoine Gintrand.<br>
            </span></span></div>
      </div>
    </blockquote>
    <br>
  </body>
</html>