Electrical impedance of a piezoelectric device

The electrical impedance of a piezo device, piezoelectric transformer e.g. around a resonance mode, can be electrical model with the as a LCR brance in parallel with a capacitor: "LCR||C". Equivalent to the Mason model of a piezoelectric transformer with either secondary or primary side short-circuited. The amplitude plot of the impedance around resonance will first have a resonance-peak followed by a anti-resonance-peak.

To narrow the problem down the, we look at a 2D model consisting of an simple rectangle shape.

-Using COMSOL 3.5 or 3.5a the electrical impedance of the 2D model matches the electrical "LCR||C" equivalent and a real-life measurement confirms the result.

-Using COMSOL 4.xx to simulate the same model dose not fully match the "LCR||C" equivalent. The resonance-peak is fine. The anti-resonance-peak seems to have the right frequency, however the impedance is fare from high enough. In the electrical domain a low anti-resonance-peak is equivalent to a material with to high conductivity. The material parameter nor the physics uses conductivity or similar as I'm aware of!?

What have changed in the physics from COMSOL 3.5a to COMSOL 4.xx?? Are there something that I miss in my model??

PS I have treble checked the material parameters and they are identical from 3.5a to 4.xx

Attached is plots that illustrate the problem.