Discussion Forum

Hi,

I am trying to simulate a parallel plate capacitor consisting of multiple layers. This is for a project where we try to create a capacitive pressure sensor. The order of magnitude (regarding pressure) is about 100MPa.

Basically, we've got multiple steel films, with some kapton films in between.
The idea is that we apply pressure on the capacitor, as a result the distance between the layers decrease so the capacity increases. Using this increased capacitance, it should be possible to 'calculate' the pressure applied on the capacitor.

First, I tried simulating this with two sheets of metal, with a kapton layer in between.
I created the geometry using a 2D-rectange, and the extrude option. I applied -10V on the bottom layer, and +10V on the top layer (using a terminal). I also a boundary load on the top layer, and a roller boundary condition on the bottom layer. I tried to apply symmetry by simulating a quarter of the capacitor and applying symmetry boundary conditions on the two symmetry edges.

It seems to work alright, but the capacitance value is way off (at least with my calculated values). My calculated value is 50 times as large as the value COMSOL found. (~2nF vs 39pF). My measured value is about 1nF, so that is still a factor 25 difference with COMSOL's answer. (Also, our physical model is a rough test (as proof of concept), so it is fabricated rather poorly. This deviation in the physical model is to be expected)
I could explain a bit of this deviation due to the fact that I only simulated a quarter and then applied symmetry; I have no idea how COMSOL handles this. But even then, this should only account for a factor of 4.

I also tried to simulate this with four sheets of metal and three layers of Kapton. It can calculate the capacitance just fine, but when I try to apply pressure on it, COMSOL either gives an out of memory error, or tells me my result is not convergent. Obviously, I have to optimize the model.

I used a swept mesh. I tried free tetrahedral, but the computer has issues handling this mesh.

Anyway.
- Does anyone know what I am doing wrong with the capacitance. I can expect a deviation with the calculated value, but a factor 30 is a bit extreme.
- Does anyone know how I can optimize the problem any further? Ideally I want a capacitor consisting of multiple layers of Kapton, then I want to do a parametric sweep over the applied pressure and see how the capacitance changes over pressure. (It should be linearly, so a single calculation would be acceptable too if a parametric sweep is too much).

My files are a bit too big for attachements (8MB) , so I uploaded them to Google Drive. They can be found here:
Simple Parallel Plate: drive.google.com/file/d/0B3wLLdCX07-fVXdKT3NhOENybms/view?usp=sharing
Multiple Parallel Plates: drive.google.com/file/d/0B3wLLdCX07-faC11aFhCVzcxSTA/view?usp=sharing

Thanks in advance.

The parameters I used for my geometry are measured from my physical model. Except for the thickness of the steel plates, which are way smaller in reality, but are temporarily made bigger in order to make the calculations easier. (Steel thickness should not matter in the case of two steel plates with a dielectric in between)
I could not attach a file