Droop control Problem Simulation

General
1

Hi everyone,

I am building a three-phase grid-forming (GFM) inverter with an LCL filter in PLECS Standalone 5.0 for my undergraduate thesis. Despite strictly following standard cascading control rules and bandwidth separations, the system completely fails to stabilize.

I would really appreciate it if someone could point out what I might be missing in my setup or PLECS model.

System Parameters

  • Rated Power: 10 kVA (P = 8 kW, Q = 6 kVAr)

  • Grid Voltage: 400 V (L-L), 50 Hz

  • DC Link: 800 V

  • Switching Frequency: 10 kHz (SPWM, continuous)

LCL Filter Design

Designed for 15% current ripple and a resonance frequency of 1500 Hz.

  • Li (Inverter side): 4.35 mH

  • Cf (Capacitor): 9.95 uF

  • Lg (Grid side): 1.53 mH

  • Damping: Currently none (or very small parasitic R = 0.001 ohm).

Control Scheme Topology

I am using a standard Droop + Cascaded PI (dq-frame) control. Sensor Placements:

  • Inner Current Loop: Feedback taken at L .

  • Outer Voltage Loop: Feedback taken at Cf (capacitor voltage).

  • Droop Loop (P/Q Calc): Power calculated using V and I at the PCC (grid side, after Lg), passed through a Low-Pass Filter.

Tuning (Decade separation applied):

  • Inner Current Loop (500 Hz bandwidth)

  • Outer Voltage Loop (100 Hz bandwidth)

Cross-coupling decoupling terms (w*L and w*Cf) are included.

The Issue

Everything worked well during the GFM formation, but when connected to the ideal grid as an initial test, the droop control, which should have made the experiment successful, ended in failure for every test parameter. I have followed several journals and books to fix this, but everything I have done has failed. Did I miss something or is there an additional step in the PLECS simulation? I ask for your help and guidance. Below is the image of the initial simulation results.

image
image1919×1079 91.6 KB

image
image1919×1078 91.8 KB

My Questions:

  1. Is a passive/active damping strategy strictly mandatory for an LCL GFM inverter in PLECS, or should the 500Hz/100Hz cascaded PI be able to stabilize it without explicit damping?

  2. Are there any specific PLECS solver settings (variable step, tolerances) or digital delay blocks (ZOH) that are strictly required for this topology to avoid numerical instability?

  3. Is connecting this directly to an Ideal Voltage Source (with only Lg in between) causing a stiff-grid conflict, even with Q-V droop enabled, or is the droop equation I’m using incorrect?

here is my plces simulation file

voltage_gfm_droop2.plecs (206.2 KB)