Loadflow Study - Where do instantaneously required VARs come from ?

[EddyWirbelstrom]

I am using SKM's DAPPER loadflow software to calculate loadflow voltages when a 1.5MW motor is DOL started.
The plant has local steam turbine generators ( I have modelled as PQ ) and an interconnection to a Utility ( modelled as Swing Bus ).
The plant does not have pf correction capacitors or static VAR compensator.
Loadflow results show MVAr export/import to Utility as -
Before Motor Start - Export 1153 MVAr
At Instant of Motor Start - Import 7803 MVAr.
These imported MVAr's must instantaneously be available at the instant of motor start for the loadflow to converge.
The kinetic energy of the Utility's generators will supply the MW before the Utility generator governor responds, however the Utility's AVR can not respond instantaneously to produce the VARs required at the instant of motor start.
Where do instantaneous VARs required by the motor at the instant of starting come from before AVR response ?
How do loadflow software programs deal with it ?

 

[Skogsgurra]

The problem is more semantic than electrical.

The var (yes, SI says var, it is a unit of its own) is nothing but current multiplied with voltage. If both are in phase, then you have watts, if they are 90 degrees out of phase, then you have vars.

The phase angle is not determined by the utilities generator or its AVR. It is determined by the load. So, if - during the start - the motor draws a lot of current and most of that current is consumed by a reactive (inductive) motor winding, you will have a high var consumption.

The current needed is always there, ready to be consumed. No matter what the load's phase angle is.

So, the answer to your question is: The vars are delivered by the utility. Also during the start. There is nothing magic or confusing about that.

 

[EddyWirbelstrom]

Thanks for your lucid explanation Skogsgurra.
The angle of the load current is determined only by the load.
The angle of the current determines VAR = VA sin angle.
The generator AVR would eventually correct the voltage drop at its reference bus due to the increased current.
However the software corrects the utility's voltage instantaeously to the per unit setting.

 

[Skogsgurra]

Yes. That seems to be correct.

 

[Bung]

Load flows deal in the steady state. You will get results that deal with the situation immediately before you switch, or immediately after. The transition (the dynamic changes resulting from the switching) can only be seen by doing a transient analysis. The transient analysis will show how the current magnitudes and phases move around, and how the machine rotor angles react etc. The load flow will only tell you what the stroy is at the end of all the dynamic stuff.


Bung
Life is non-linear...

 

[Mstrvb19]

You mentioned "At the instant of motor start"- if this is not a motor starting analysis (i.e. a transient analysis) then, is your motor modeled at locked rotor in a steady state load flow or did you mean 'after the motor has accelerated'? How is a 1.5MW motor, even at locked rotor, pulling 8596 Mvar? It sounds like something funny is going on. Can you shed light on this? There is no way logistically that your motor can be the reason for going from an export of 1153Mvar to an import of 7803 Mvar unless one of the numbers is wrong (could you have meant kvar?). What voltage is this at?

 

[Vegemite]

From similar studies done using PTW it seems that the larger than expected VAR demand accompanying the locked rotor start condition is due to the depressed system voltage (at the instant of start) forcing the remaining, sometimes massive, induction motor population to draw more VARs to maintain their power output.

 

[timohearn]

I recommend that you read the manual for the software in the Dapper Reference Manual page 46 that describes the solution methods used for the load flow analysis.If you have questions I would recommend that you contact SKM Systems Inc. (they will endeavor to answer your questions and I have found them helpful in the past).

If you are analyzing your system for a motor impact study to determine the minimum voltage while starting (locked rotor) a large motor; I would recommend that you determine the minimum power system configuration and model it using a Thevenin's equivalent using the system's equivalent transient reactance. I would also recommend that the "voltage behind the transient reactance" should be the load flow voltage established prior to starting the motor.

 

[Mstrvb19]

Johnarcher-Yes- I would expect to see a larger number of VARS consumed during motor starting, but a 1.5MW motor couldn't draw 8000Mvar, even if it were to effectively become a dead short that was 100% reactive- the MVAR requirements are around 5000 times the motor rated power draw. Even a dead short would be around 20 to 30 times the current draw-not even in the ballpark. That was what prompted my question.

 

[EddyWirbelstrom]

The 1.5MW, 3.3kV, 2 pole motor has a FLC/LRC ratio of 0.1859, and a starting power factor of 0.1
The motor loadflow power on starting is 667kW + j 6632 kVAr.
Yes Mstrvb19, my original posting erroneously stated plant import of 7803 MVAr instead of 7803 kVAr.
Timohearn, why do you recommend using the generator transient reactance and not sub-transient reactance ?
The first instant would be voltage behind sub-transient reactance.

 

[timohearn]

mnewman, in my opinion when calculating the voltage profiles on power systems when starting large motors the generator's transient reatance is recommended because the sub-transient reactance will decay quickly and therefore the transient reactance is a more conservative representation of the power system's response.

 

[EddyWirbelstrom]

Timohearn,do you agree with the following ?
When using steady state loadflow software for motor starting studies,and when the motor starting kVA is significant in comparison with the generator kVA, using the generator transient reactance is appropriate because it presents a larger reactance for a longer time than the sub-transient reactance. eg. 0.1-0.3 seconds before full AVR response.

 

[timohearn]

mnewman, in general I agree with the statement. However, for systems that involve two machines (a generator and a large motor) a comparison of the machine's inertia should be done. If the motor's inertia is not significant with respect to the generator's inertia (1/10th of the generator's inertia or less) then steady state loadflow software can successfully model the system. Otherwise it may be advisable to use transient analysis methods to model the system.

 

[Bung]

But the loadflow is only telling you what happens at one instant in time. When you use it try to model a motor start, you are assuming a state for the motor at some point in the start up, and also that the system will settle to the calculated state before the motor's state changes. Doing a few time slices along the run-up curve may give you an approximation to how the start-up affects the system, but that is all it is. Of course, it may still be good enough for the studies you are doing. Only a transient study will give you a better model of the dynamic case.

I think that the assumed power factor of 0.1 is way too low - something between 0.3 and 0.4 might be better. The 0.1 pf might be OK for the first cycle or two, but once the field is established and the motor begins its run-up 0.3 to 0.4 is better.

Bung
Life is non-linear...

 

[mc5w]

The increased VAR flow causes the voltage at the generator terminals to dip because the generator supplies the instantaneous change in VAR flow. The voltage regulator then increases the rotor current to compensate.

On smaller generators 1.5 MW or less the electronic voltage regulator monitors all 3 phases and can sense a voltage dip in 1/6th of a cycle. On better 60 Hertz generators the voltage regulator gets its power from a 300 Hertz single phase generator and the brushless exciter that supplies rotor current generally runs at 120 Hertz 3 phase. The bandwidth of the voltage regulator is high enough that voltage recovery could occur in 1/2 cycle but in actuality is deliberately slower for stability reasons. The high bandwidth does make it easier to tune out instabilities much like how electronic servo drives are definitely easier to tune than amplidyne servo drives.

On old fashioned generators the voltage regulator is electromechanical which slow things down a lot.

Larger generators have voltage regulation that is deliberately slower that for smaller generators. One reason why electrical utilities still keep smaller generators on line with larger generators is that the voltage regulators for the smaller generators tend to be more nimble - the faster response helps with small rapid changes allowing the larger units to cope with slower changes to which they are more suited.

Mike Cole

 

[Bung]

If the utility source is much stronger than the local generation, or the local generation is electrically "far away" (relatively, depending on size and impedance between gens and load), then the generator control characteristics may not have much effect on the outcome of the motor start.

But this is all the more reason why a loadflow study is inappropriate for this task. None of the possible generator / system dynamics effects (and possible problems) will show unless a transient study is done.


Bung
Life is non-linear...