Symmetric boundary condition not satisfied

Dear all! I’m solving a chemical problem like this.

And this is my edp code:
test.edp (1.8 KB)
I try to compute half of the domain. dx(ci)=0 should be satisfied at left boundary, but it seems to be wrong.

I’m not sure what causes this problem, because when I compute the whole domain, it turns out to be right. Dose anyone have some ideas? Thank you!

I guess I’ve solved this problem. Just use P2 finite element. Everything is fine now.

Your boundary conditions are on 1 and 2 AFAICT are the bottom and right
sides respectively although I may have gotten something confused.
This system apparently describes X catalytiucally removing M
but also able to react with O2 and itself. I guess the light based generation
rate however is not dependent on any reactant or is a const included in the
light term.
Your solve seems to include the boundary or equillibrium values
(Xo ) that would try to impose those values everywhere instead of letting
the diffusion term model that.

Thank you for your reply!
As you already know, this is a UV curing process including oxygen inhibition. There are three constituents in the simulation, M(monomer), MR(free radical species) and O2(oxygen). The light field is given fixed over time. Only the diffusion of O2 is considered, so other two equations lose their diffusion terms and become dci/dt=ri. Adding BC to it may causes some problems. So I add a small diffusion coefficient to them, but it doesn’t work. I guess it’s because the boundary effect is too strong that nothing changed after adding small diffusion.
Finally, I try P2 element which may improve the accuracy and capture changes in second order. Whatever, it works well now.

I guess your beam profile is an extend source gaussian and it tapers
at each edge but is flat in the middle. If you cut the domain in 1/2
then what happens to your light profile? If you are going to try
to compare simulation to expected experiment you may
need some changes in the model although comparing FF output
to expectations to the model is a different issue.

I know what you mean. But after using P2 element, the result tend to be right. So I think there is no problem in boundary settings including light.