Closed RomanLF closed 2 years ago
Unless I am misunderstanding the problem, it seems like the problem is ill-posed since any constant value at the center would satisfy those conditions.
You should also keep in mind that derivatives are computed using the nearest neighboring nodes, regardless of any boundaries in the domain. So derivatives at the center of the domain may be computed using some nodes from outside of that center region, which I am guessing you would not want. I think this is why you are getting any solution at all rather than a singular matrix error.
Hello,
Thank you for your answer. As you said, these nodes seem to disturb the solution, as the results are different when the 2D-courtyard is not considered. I imagine that in 3D with a limited wall height the problem would disappear too as court-yard's nodes would be connected to the exterior by vertical axis.
Thank you a lot for your answer !
Hello again, I meet difficulties to procceed diffusion on a domain occupied by a sort of court-yard, with the model :
on the domain :
The source is located at the outside of the courtyard. A neuman zero-flux condition is set on the edges of the obstacle. However, the solution inside the courtyard is not equal to zero which is not physical as visible below.
Do you know how I could avoid this issue ?
Here is my code :
Best regards, Roman