I'm trying to model the unsaturated hydraulic behavior of a highly permeable (gravel) constructed wetland (CW) via a finite element approach. For the characteristics of the system, I need to start my simulation with an almost empty CW. This leads to a lot of trouble in the inflow portion, due to very steep pressure gradients, which obliges me to use a very very short time step (therefore, very long simulations). I've tried different mesh refinements (triangular, quadratic with diagonal, very coarse or very fine mesh densities) near the inlet region without good results.
So, if anyone has had some experiences with similar studies, any advice could be very useful.
Anacleto,
I do not have experience in wetland modeling, but in open channel flow I use to deal with similar situation. If I don't have initial condition for my simulation I use to start setup period from dry bed conditions. It was always very slow simulation, even the finite element scheme was quite efficient. Then I start to use some initial water level condition with no flow and this was much less time consuming. In general if the water level was close to the desired level the simulations was much faster and stabilized relatively quick.
Not sure if you really need to initiate the simulation with such condition when you do not have any water in your domain. If so, from my perspective the only way is to gradually increase water level at boundary condition. The time step need to be short in such a case. If not try to set the initial water level close to the one you want to start with and that will make your setup time much shorter.
Not sure if any of my suggestions will work in your case, but is always worth to discuss. Maybe you come out with something else.
Thank you very much Tomasz.
I need to start with empty domain, but the gradually increase in water level is feasble, so I hope that my future simulation would be better.
- Badges
- Science topic
- Similar topics
- Engineering
- Modeling