Felix, in addition to what has been suggested before, think also about the fact that you don't know which of the two solutions you have is correct. Before you draw any conclusions, you need to find a way to establish which one is the one that is correct and which one is not.
Best W. On 7/19/19 2:37 AM, Félix Bunel wrote: > Hi guys ! > > I'm curently working on a project which consist in solving on a circle the > Leslie Eriksen problem. I'm facing an issue with the stokes part of the > problem : > > a, phi_s, phi_lin, X are constant u, v, P, phi are variables. > > I'm solving the two different equation (the one with \phi, and the stokes > problem) separately. For each time-step I solve phi and inject it in the > stokes problem to get a suitable u,v. Then I re-inject u, v in the first > equation to get a better \phi. Then I inject it in stokes, and so on... You > get the idea. This procedure converges towards a solution. > > I have another code (Mathematica) that do exactly the same, but only for the > equilibrium state (t->infinite, all time derivative disappear) > > I wanted to compare the results of the two codes in this case to check for > the > viability of my solution. (I used X=4, a=2, phi_s = phi_lin = 5) > > Unfortunately the two codes do not output the same results. Although the > solutions looks very similar they are still a bit different as you can see > below. > > To understand better where the problem is coming from, I'm comparing only the > first iteration : > - I solve Phi in the first equation with u,v=0, which gives me a solution (it > can be calculated analytically in this case, the solution is \phi = > \phi_lin*(1-x^2 -y^2) > - I solve u, v, P in the stokes problem using this phi > Then I compare the results of both codes. > > Here are the results : > > 1 - The first equation in phi is very well solved and both codes gives the > same results. They both gives a solution that corresponds to the one > calculated analytically. > 2 - The two codes do not give the same results for the stokes problem. The > two > solutions looks almost exactly the same, but the one given by my dealii code > give higher values for the velocity field and smaller values for the pressure. > > Here, I plot the velocity component u, along the the y axis in both codes : > > Mathematica : (maximum is 14.33) > Dealii : (maximum is 18.44) > > Here are what the stokes solution looks like : > <about:invalid#zClosurez> > > > > > > > > > > > Here are the idea I thought could be the reason : > > 1 - Mesh : I tried different mesh refinement in both code and they always > gives the same values > 2 - Fe-element : I know for the stokes problem that it needs to be at a > higher > order for u and v than for P. In the dealii code I use : > stokes_fe(FE_Q<2>(2), > 2, FE_Q<2>(1), 1) and I also enforce this in the mathematica code. I also > tried different fe order and it did not change the results > 3 - My error : I have re-read my equations in Mathematica again and again. > Re-calculated my weak formulation again and again and checked my assembly > process in dealii again and again... I did not find a single mistake... > > > If you want more detail about what i'm doing, I'm joining the pdf where I > have > written the equations and my weak formulation (from page 3 to page 5). > > Thanks in advance > > -- > The deal.II project is located at http://www.dealii.org/ > For mailing list/forum options, see > https://groups.google.com/d/forum/dealii?hl=en > --- > You received this message because you are subscribed to the Google Groups > "deal.II User Group" group. > To unsubscribe from this group and stop receiving emails from it, send an > email to [email protected] > <mailto:[email protected]>. > To view this discussion on the web visit > https://groups.google.com/d/msgid/dealii/cd5ad853-5fbb-4daa-8a3e-a2f4d626760d%40googlegroups.com > > <https://groups.google.com/d/msgid/dealii/cd5ad853-5fbb-4daa-8a3e-a2f4d626760d%40googlegroups.com?utm_medium=email&utm_source=footer>. -- ------------------------------------------------------------------------ Wolfgang Bangerth email: [email protected] www: http://www.math.colostate.edu/~bangerth/ -- The deal.II project is located at http://www.dealii.org/ For mailing list/forum options, see https://groups.google.com/d/forum/dealii?hl=en --- You received this message because you are subscribed to the Google Groups "deal.II User Group" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. To view this discussion on the web visit https://groups.google.com/d/msgid/dealii/d7501170-2162-be57-4e79-2d32d1afa9f8%40colostate.edu.
