Prev: Reading and modifying XML file using MATLAB
Next: Change size of Principal Component Visualization Tool (mapcaplot)
From: Kamel on 19 Apr 2010 06:41 Thank you Tortsen for your help. By the way, I found also an other way to resolve this problem using one variable, u. I derived manually the term dsig/dY to obtain an expression depending only on DuDz to form the expression of 1/H. And for the left boundary condition, I wrote Y=Y-1^(sig0) as 1-Y-1^(sig0)+DuDz=0, Y-1^(sig0) being resolved beforehand, and it works. c = 1/H; f = DuDx; s = 0; BC: pl = ul; ql = 0; pr = 1-Y-1^(sig0); qr = 1; Regards, Kamel Torsten Hennig <Torsten.Hennig(a)umsicht.fhg.de> wrote in message <1191033208.35186.1271658944531.JavaMail.root(a)gallium.mathforum.org>... > > Dear Torsten, > > > > Resolution of the nonlinear equation is now easy to > > do. > > I have just a small trouble with one of the boundary > > cdt using pdepe. > > > > I wrote the equations like this for the resolution > > using pdepe: > > (u1=u, u2=Y) > > > > (1/H, 0)*d/dt(u1, u2)=d/dz*(du1/dz, u1)+(0, 1-u2) > > and the pdepe coefficients are normally: > > c = [1/H; 0]; > > f = [DuDz; u] ; > > s = [0; 1-u2]; > > > > What I am suggesting for the boundary conditions is > > left: > > (u1, u2-1) + (0, -du1/dz*(du2/dz)^-1)*(du1/dz,du2/dz) > > = (0, 0) > > right: > > (u2-u1-1, u2-u2^-1(sig0)) + (0, 0)*(du1/dz,du2/dz) = > > (0, 0) > > > > which should lead to pdex1bc coefficients in the > > following form: > > pl = [ul(1); ul(2)-1]; > > ql = [0; ??]; > > pr = [ur(2)-ur(1)-1; ur(2)-ur2^-1(sig0)]; > > qr = [0; 0]; > > > > My question is: how can I write the second ql > > condition (in ??) corresponding to the term > > -du1/dz*(du2/dz)^-1 ? > > > > Thank you in advance for your precious help. > > > > Best regards, > > > > Kamel > > > > > > I thought to take > > c = [1/H; 0]; > f = [DuDz(1); 0] ; > s = [0; Dudz(1)+1-u2]; > > pl = [1-ul(2);ul(1)]; > ql = [1 ; 0]; > pr = [1-ur(2);ur(2)-u(2)^(-1)(sig0)]; > qr = [1 ; 0]; > > A problem may arise at the right boundary because > at t=0, du/dz=0 at z=h, but Y=Y^(-1)(sig0) <> 1 in > general. > Maybe you can adjust the initial value for u > at z=h such that du/dz|z=h = Y^(-1)(sig0) - 1. > > Best wishes > Torsten. |