Load Flow Model Limitation

[DistCoop]

All,

I work for a rural distribution cooperative and am playing with some station loss contingency scenarios in a commercially available model. I am running into what I believe to be a limitation of the model itself, but would like some opinions.

I have a scenario where I must serve a substation from a neighboring substation 10 miles away. The distance of the feeder and size of the load is bringing my voltage down to 80-90V. I think this is because I am modelling the loads as constant power. If the voltage goes down, the current must go up, increasing losses, further bringing my voltage down, and so on, in order to maintain my demand.

What would really happen in this scenario? Resistive loads would not force more current to flow. Motors maintain a constant power for a certain voltage range... but would a motor maintain constant power if a 0.7pu voltage was applied to it, for example?

I have a station that serves 4MW of load... under this contingency I get 80V, 4MW of load, and 2MW of loss. I wonder if anyone has seen this and can explain what really happens.

Thank you

 

[davidbeach]

Load characterization is a difficult problem, but a good load characterization is essential for what you're trying to do. You've probably got a mix of constant impedance loads ("good") and constant power loads ("bad"). Some loads will self remove if the voltage falls too far, but others do really nasty things. Consider home air conditioning equipment (including heat pumps) where if the voltage goes below some value the motor will stall. At the point the motor stalls it goes from being a constant power load to being a constant impedance load, but with a much lower impedance.

There is a place in the load mix that you can come close to having a constant current characteristic, try that and see how it works.

With good metering available, you can get a feel for the load response by forcing a two or three step change in the LTC and see how the load responds to the change in voltage.

 

[dpc]

You might do a Google search on their "conservation voltage reduction". There are some research papers where testing as described by David Beach has already been done. Load will go down as voltage goes down - not 1 to 1, but there is a definite reduction indicating a good percentage of typical distribution load is constant impedance or constant current as opposed to constant power.

This might be a start:

http://ieeexplore.ieee.org/xpl/logi...ore.ieee.org/xpls/abs_all.jsp?arnumber=119238

 

[pwrtran]

2MW system loss? I would check the conductor impedance values against manufacture's specification first in the model first, I found in our system there were errors from messing up per km and kft values, or messing up pu and real values.

 

[desrod]

Distribution system losses level should be less than 10% at peak (typical average 3-4% in North America). How is your network behaving in its normal configuration? What is your MV level?
Be aware that some software will switch load model when voltage reaches a certain threshold. We recently looked at a network automation project aimed at power restoration and we also achieved voltage drop in the same range for some load transfer configurations.
I agree with pwrtran: check you data. To quickly spot problem areas, look at the voltage profile on all your nodes. Steep voltage drop will indicate area worth checking.
You can also contact your software provider. They usually have experienced people that can take a look at your model and spot rapidly if something is wrong. Also check your LTC and regulator controls. I have seen a lot of wrong setting resulting in the device lowering the voltage instead of correcting it.

Desrod