Blown up after 1 iterations
Warning: 20 voltage controlled buses whose voltages differ from their scheduled voltages by more than 0.00001 pu
Largest mismatch: 2646.40 MW1079173.40 Mvar1079176.62 MVA at bus 228 [TEZPUR 220.00] System total absolute mismatch: 2361363.55 MVA
Can anyone help me?
Hi I am writing this based on my experience. To solve a case ( or converge a blown up case) , you need to do following:
STEP1. You need to relax the solution parameters to following values. Using default values for solving a case might blow it up. Set solution parameters to following values:
i. Blow-up parameters to a large number (typical range is 90000 to 99999)
ii. TOLN (largest mismatch in MW and MVARs) to 1.1.
iii. VCTOLQ (controlled bus Q mismatch convergence tolerance) to 10.1
iv. VCTOLV (controlled bus voltage error convergence tolerance) to 10.00001
STEP2. I would suggest following selections for solving-use the “Fixed slope decoupled Newton-Raphson” method as this method performs better that other algorithms for difficult cases. (Full Newton or Decoupled). Enable FLAT start, Disable all other parameters, as shown below:
STEP3. Disconnect buses that have large angle changes that cause the solution to diverge. These buses might show up in the column DELTA/V/ column ( Highlighted) NOTE THESE BUS NUMBERS, AS YOU NEED TO RECONNECT THEM LATER.
STEP4. Now gradually enable the parameters. My personal experience says following this sequence:
i. VAR limits first, then Adjust DC taps, then TAP (stepping) then Adjust phase shifters, then Switch shunts ( Enable all)
ii. For very difficult case, the case might blow up if you enable any option mentioned above (including the FLAT start). In that case disconnect the buses mentioned in step 3'
iii. Lock any switched shunts that toggle while solving.
iv. Turn generators on or off if:
• Bus voltages are too high (≥1.2) or too low (≤0.7) for the case to converge
• There is not enough reactive support for the case to converge.
STEP5. Once you reach a solved case, you need to reconnect the buses which you disconnected while solving. Make sure that you manually match the voltage and angle of the disconnected bus to the voltage and angle of the Bus it is connected. After you have reconnected all the buses, GRADUALLY put the solution parameters to back to the default values.
i. Blow-up = 5.0
ii. TOLN =0.1
iii. VCTOLQ = to 0.1
iv. VCTOLV = to 0.00001
I wasn’t able to able to upload pics because of some restrictions imposed by this site. I hope this helps though Nihal
First step is to check the impedances of the branches (including transmission lines and 3-winding transformers) linked at the bus 228. Usually low impedance will cause this kind of blown up.
when i enable FLAT START all the bus voltage becomes 1..... although it seems as if case is solved but it leads to a state from where it cannot be reverted back
were u talking about flat start above adjusts dc taps and non divergent solution
Yes. I was talking about the Flat above non divergent solution. Remember - once the case blows up, it cannot be converged back unless you make changes in angle and voltages of those buses which are causing the case to blow up. In large systems, when the case diverges, it causes too many buses voltage and angle to arrive at a point from where the case cannot converge. There flat start comes handy. It resets the initial voltage and angles and of all buses at 1 and zero, then the solution algorithm chosen ( Newton-Raphson, Gauss etc. ) arrives at some final values. It is worth mentioning that since all these solution algorithms are non linear, there can be more than one possible set of solution, meaning- more than one value of voltage and angle for which the system will converge.
Asked: 2014-03-19 01:36:33 -0500
Seen: 7,711 times
Last updated: Mar 31 '14
whit loves you. Content on this site is licensed under a Creative Commons Attribution Share Alike 3.0 license.
Correcting a blown up case is more art than science. If possible, you should try to flat start the case. If that isn't possible, try to get the case back to some kind of state where it last worked and then use very wide Q limits on your gen to nudge the case to the correct solution.
i have trying my best to solve the case..... but the value of Q gen goes to abnormally high value whenever i try to solve the problem..... i have tried with Full Newton Raphson and gauss seidle method but the case is just not working.... if u want to have a look i can provide u the file
Terminated after 1 iterations--singular Jacobian at row 9 this is the problem creeping up now.... guess voltage is not getting transfered
Lower total load (load shedding, scale down load) near the bus where the Jacobian goes singular (see the iteration messages from the load flow) and solve at every step. total Pgen shall be just over total Pload, same for Q.
I would check the mismatch at bus 228 before attempting to solve the case. It could be the voltage magnitude and/or angle at bus 228 is far away from the buses one level away. If that is the case, try to change the voltage and angle at bus 228 to the same value as the buses around it.