Discussion Forum

Hello everyone,

I am a student of Grenoble INP currently on my master's. We are trying to realize a project with my group and are using COMSOL to simulate the behavior of a micropump. Out micropump is somewhat irregular i'd say though. It consists of a lower chip on which channels are patterned but the channels themselves are not continuous. Every so often the channel is lost (not patterned ). right above this another plate of material is placed in perfect contact with the patterned substrate. On the upper surface of this upper chip we place 3 circular disks of PZT positioned exactly on top of the "channel holders". When voltage is applied on top of the PZT disks, they deform (upwards) thus creating a semisphere volume right below, and opening a path to the fluid to pass from one side of the channel to the other.

Simulation using the piezoelectric devices interface seems to work OK but when we try to introduce the fluid-structure interraction domain, things mess up. In this approach, in the pzt devices we only declare the piezoelectric disks and associated conditions to be involved. On the other interface we define everything else (as well as the pzt, for which i am not sure if it;s the correct way) as a linear elastic material as well as the volume of water inside the channels as fluid. Now the problem lies in the fact that by these definitions and BC included comsol should in theory understand the inner boundaries between channel walls and the upper chip as fluid structure interaction boundary condition. But in this field everything seems to be non applicable.

I should add that the overall geometry was designed as 2D extruded in 3D and an overall assembly was formed using mostly contact pairs (as the lower and upper chips must be allowed to move in general. In order to solve the problem I tried to use a partition operation in geometry to create inner boundaries on the surface of the upper plate which is in contact with my fluid the shape of whom will be exactly the same as the one of the chambers/ disks. But these boundaries are also set as non applicable.

Last, I made sure to define a free mesh so that we can track a fully coupled case. (as structure will affect the fluid but in turn the fluid response will limit the operation frequency of the reciprocating movement )

Has any1 faced this kind of problem before ? I suspect it is probably due to geometrical mismatching of the fluid/structure boundaries or maybe some parts of PZT anf FSI cancelling out each other but I could use some1 else's opinion.

Thank you for your time.