why laplacian() failed
Hi guys im a newbie, i got a problem wired well..
i have 2 equations: dT/dt=1/a*(div(b*grad(T)+div(c*grad(wl))) (1) dwl/dt=1/d*(div(e*grad(T)+div(f*grad(wl))) (2) a--f are functions of T and wl so i give my solver here fvm::ddt(T) (1) == 1/a*fvm::laplacian(b,T)+1/a*fvc::lapalacian(c,wl) fvm::ddt(wl) (2) == 1/d*fvc::laplacian(e,T)+1/d*fvm::lapalacian(f,wl) my ideas here is to separate the T and wl, so in (1) "wl" is considered as known(fvc::) and in (2) "T" is considered as known(calculated from 1). problem: the solver is compiled without errors and it works but the results staye as initial conditions, the "T" ad "wl" i got didnt change, they are around the initial conditions and no convergence at all even run longtime. it seems laplacian failed when reading bundary conditions. note: scheme of laplacian is linear corrected scheme of ddt(T) backward mesh:1D initial condition: uniform bundary condition:timeVaryingUniformFixedValue thanks for all of your advices welcom~~ |
i tried another solver here
fvm::ddt(T) (1) == b/a*fvm::laplacian(T)+c/a*fvc::lapalacian(wl) fvm::ddt(wl) (2) == e/d*fvc::laplacian(T)+f/d*fvm::lapalacian(wl) this time the T and wl are resolved but the results of T and wl are not good well, coz we couldnt get the coeff "bcef" out of laplacian() if the "bcef" depend on the T and wl. so any1 have some ideas ? |
All times are GMT -4. The time now is 08:32. |