Closed imitator19 closed 2 years ago
Thanks for reporting the issue. I was able to replicate the issue but don't have a solution yet.
The error is due to not being able to find solutions to algebraic equations. It might be caused by the control parameters of the renewable model. One way to troubleshoot is to keep only the REGCA1 model and test it again. Adjust the parameters until it converges, then add back REECA1, and tune the parameters.
I'm tied up recently and will be able to work on it late this month.
Thank you for your fast response and suggestions. I will try to troubleshoot it on these two days, and report the result here.
I have removed REECA1 and REPCA1, only leaving the REGCA1 for renewable modeling, in the case of ieee14_solar.xlsx. I have tried a variety of parameter combinations, the simulation continues to fail, however. I think it may be necessary to inspect the equations of REGCA1 that do not converge. Thank you.
Thank you for your time and update.
If you are interested, another direction for troubleshooting is to increase the fault impedance xf
to a large value, say 0.1
, to reduce the severity of the fault. If the system converges, then slowly decrease xf
and see if any of REGCA1's variable sees sharp changes. All variables of REGCA1 can be viewed using ss.TDS.plt.panoview(ss.REGCA1)
. Related parameters of sharply changing variables may offer some clue.
I will work on it later this month.
Hi Cui, I‘m Payne, from CSG, China, I've been studying Andes recently and would like to ask a question about why andes can finish a 20-second transient simulation of a 2000-bus system in a few seconds. while other software or toolbox takes tens of seconds to calculate. What is the secret to speed up?
@Payne1992 Please start a new thread in Discussions and do not reply to an unrelated issue in the future :)
It's mostly owing to the implicit integration method, vectorization, and numba compilation.
@imitator19 :
The convergence issue you experienced is likely a no-solution case. A fault is implemented by a shunt with a small impedance, which effectively draws reactive power from the fault bus. As a result, the voltage will be reduced to almost zero, but the drawn reactive power, i.e., V^2/X
, can still be too large for the network.
I created the case with REGCA1 only in PSS/E and got the same non-convergence issue. A fault is applied to Bus 8 at t=0.1
and cleared at t=0.11
with the default impedance parameters. PSS/E complains that the network does not converge immediately after the fault clearing. Note that in PSS/E, the fault is cleared at t=0.11 but will take place at the next step.
The PSS/E output is in the below, and I attached the case files for your reference.
Ordering network...
Diagonals = 13 Off-diagonals = 22 Maximum size = 36
ITER DELTAP BUS DELTAQ BUS DELTA/V/ BUS DELTAANG BUS
0.0 0.0001( 2 ) 0.0001( 9 )
0.00000( ) 0.00000( 2 )
0.5 0.0000( 2 ) 0.0001( 2 )
0.00001( 2 ) 0.00000( )
1.0 0.0000( 2 ) 0.0000( 2 )
Reached tolerance in 1 iterations
Largest mismatch: -0.00 MW 0.00 Mvar 0.00 MVA at bus 2 [BUS2 69.000]
System total absolute mismatch: 0.01 MVA
SWING BUS SUMMARY:
BUS#-SCT X-- NAME --X BASKV PGEN PMAX PMIN QGEN QMAX QMIN
1 BUS1 69.000 81.5 200.0 50.0 2.0 100.0 -50.0
Generator conversion completed using ZSORCE
11 loads converted during this step
11 of 11 loads converted
5 Output channels have been added
Next available addresses are:
CHANNEL VAR ICON
6 23 12
5 Output channels have been added
Next available addresses are:
CHANNEL VAR ICON
11 23 12
5 Output channels have been added
Next available addresses are:
CHANNEL VAR ICON
16 23 12
5 Output channels have been added
Next available addresses are:
CHANNEL VAR ICON
21 23 12
5 Output channels have been added
Next available addresses are:
CHANNEL VAR ICON
26 23 12
5 Output channels have been added
Next available addresses are:
CHANNEL VAR ICON
31 23 12
5 Output channels have been added
Next available addresses are:
CHANNEL VAR ICON
36 23 12
14 Output channels have been added
Next available addresses are:
CHANNEL VAR ICON
50 37 26
80 Output channels have been added
Next available addresses are:
CHANNEL VAR ICON
130 117 146
Ordering network...
Diagonals = 14 Off-diagonals = 24 Maximum size = 40
14 diagonal and 24 off-diagonal elements
Model GENROU Bus 2 [BUS2 69.000] Machine "1 " :
X''D = 0.2800 and Imaginary part of complex machine impedance (ZSORCE) = 0.1300 differ by > 0.000005
Model GENROU Bus 3 [BUS3 69.000] Machine "1 " :
X''D = 0.3400 and Imaginary part of complex machine impedance (ZSORCE) = 0.1300 differ by > 0.000005
Model GENROU Bus 6 [BUS6 138.00] Machine "1 " :
X''D = 0.2800 and Imaginary part of complex machine impedance (ZSORCE) = 0.1200 differ by > 0.000005
Model TGOV1 Bus 6 [BUS6 138.00] Machine "1 " :
PMECH Initialized out of limit: PMECH (pu on Trate)= 0.3000 VMIN= 0.3000
PTI INTERACTIVE POWER SYSTEM SIMULATOR--PSS(R)E SAT, JUN 25 2022 12:38
IEEE 14 BUS TEST CASE
03/06/14 CONTO 100.0 1962 W
INITIAL CONDITION LOAD FLOW USED 1 ITERATIONS
----------------------------- MACHINE INITIAL CONDITIONS -----------------------------
BUS#-SCT X-- NAME --X BASKV ID ETERM EFD POWER VARS P.F. ANGLE ID IQ
1 BUS1 69.000 1 1.0300 1.8936 81.46 1.96 0.9997 49.33 0.6123 0.5010
2 BUS2 69.000 1 1.0197 1.5480 40.00 15.00 0.9363 22.90 0.2966 0.2959
3 BUS3 69.000 1 1.0004 1.4984 40.00 15.00 0.9363 21.38 0.3040 0.2999
6 BUS6 138.00 1 0.9987 1.3599 30.00 10.00 0.9487 14.63 0.2008 0.2448
8 BUS8 69.000 1 1.0189 0.0000 35.00 10.00 0.9615 -1.40 0.0981 0.3435
INITIAL CONDITIONS SUSPECT:
I DSTATE(I) STATE(I) MODEL STATE BUS#-SCT X-- NAME --X BASKV ID
2 -0.67697E-02 0.38037 GENROU K+1 1 BUS1 69.000 1
3 -0.50616E-01 0.76314 GENROU K+2 1 BUS1 69.000 1
4 -0.15055 0.70604 GENROU K+3 1 BUS1 69.000 1
8 -0.13354E-01 0.23064 GENROU K+1 2 BUS2 69.000 1
9 -0.15590 0.92783 GENROU K+2 2 BUS2 69.000 1
10 -0.18274 0.42298 GENROU K+3 2 BUS2 69.000 1
14 -0.15811E-01 0.24313 GENROU K+1 3 BUS3 69.000 1
15 -0.14114 0.88971 GENROU K+2 3 BUS3 69.000 1
16 -0.14804 0.43807 GENROU K+3 3 BUS3 69.000 1
20 -0.12451E-01 0.20013 GENROU K+1 6 BUS6 138.00 1
21 -0.23300 0.93079 GENROU K+2 6 BUS6 138.00 1
22 -0.17038 0.35925 GENROU K+3 6 BUS6 138.00 1
42 -0.36176E-02 0.30000 IEEEST K+5 3 BUS3 69.000 1
44 -0.15159 1.0300 ESST3A K 1 BUS1 69.000 1
47 0.48817E-01 0.49620 ESST3A K+3 1 BUS1 69.000 1
48 -0.10815E-01 1.0197 EXST1 K 2 BUS2 69.000 1
49 0.42509E-02 0.30959E-01 EXST1 K+1 2 BUS2 69.000 1
50 0.42200E-01 1.5480 EXST1 K+2 2 BUS2 69.000 1
52 -0.14209 1.0004 ESST3A K 3 BUS3 69.000 1
56 -0.20219 0.99871 ESST3A K 6 BUS6 138.00 1
59 -0.64465E-02 0.36608 ESST3A K+3 6 BUS6 138.00 1
78 -0.34852 1.0189 REGCA1 K+2 8 BUS8 69.000 1
Channel output file is "C:\Users\cuiha\Dropbox\notebooks\regc_fault\out.out"
Channel output file is "C:\Users\cuiha\Dropbox\notebooks\regc_fault\out.out"
*** Shunt " 1" at bus 8 [BUS8 69.000] ( 0.0000,-0.20000E+10) added ***
Channel output file is "C:\Users\cuiha\Dropbox\notebooks\regc_fault\out.out"
14 diagonal and 24 off-diagonal elements
*** Shunt " 1" at bus 8 [BUS8 69.000] ( 0.0000,-0.20000E+10) deleted ***
Channel output file is "C:\Users\cuiha\Dropbox\notebooks\regc_fault\out.out"
14 diagonal and 24 off-diagonal elements
Network not converged at TIME = 0.1167
Despite that nonconvergence, PSS/E is able to finish the simulation. I'm still troubleshooting the model.
Thanks a lot!
You are welcome! Thanks again for reporting!
For andes systems with wind or PV, time domain simulation fails at the time of fault: "Time step reduced to zero. Convergence is not likely."
Steps to reproduce the behavior: