HFSS: Strange results when simulating a capacitor (RLC boundary)

I tried to make a S11 based measurement of a chip capacitor in HFSS (simulated measurements)

I made a 50 Ohm microstrip line of layered impedance rectangles, put a rectangular patch at the end and applied 0.55 pF capacitive boundary to it. Other side of the capacitor is shunt with another vertical rectangle with perfect conductor layered impedance. The substrate is a box of 2.6 permittivity perfect dielectric. There's a vacuum box present. The port integration line and capacitor current line are defined.

Although, C=0.55 pF,

neither
im(1/(Freq*2*pi*( Zo(1)*(1+S(1,1))/(1-S(1,1)) )))
nor any formulas based on Zcap=sqrt(B/C) or Zcap=Z11-Z22 give a sensible results.

Tried to use port deembedding with HFSS, or via impedance transform on my own.

All the same, the capacitance is not constant and it is not 0.55 pF.

When measuring chip capacitors with network analyzer, I was getting realistic results at least at MHz range.

I can't understand, what am I doing wrong in HFSS.


My goal is to simulate capacitor with frequency profile measured in experiment. Therefore I am trying to compare the frequency dependent capacitance formula with "ideal constant" capacitance in simulated measurement model. But I can not get the reference to be "ideal constant", when extracted