Simulating 60Hz as 6Hz System - Approximate Model for Real-Time Trainer Simulation

[digi001]

I wanted to get the forums thoughts on Simulating a typical 60Hz Power System as 6Hz. 10 times less than the actual system frequency.

The motivation for this is to be able to simulate at only 0.005 time step (200Hz) and run real-time on a typical Computer.

At the 0.005 timestep there would be about 33.3 samples of a 6Hz waveform which would be adequate, however if simulating 60Hz it would only be 3.3 samples oer waveform which is unacceptable.

The Power System consists of 4-5 (4-50MW) Generators, Induction Motors, DC Loads, Power Converters. Exact simulation detail is not required, only basic dynamic operation for Training/Learning Purposes.

(Phasor simulation was also considered, however the 6Hz solution would provide better real-time domain waveforms.)

-How would making this conversion effect the various simulation parameters when comparing it to a full 60Hz simulation at 10us time step?
-How would the impedance/controls change etc? (Controls are mostly done RMS)

 

[BradtheRad]

It's a sign of an inflexible simulator if you cannot adapt the timestep to your circuit's operating frequency.

If you simulate a 60Hz circuit as 6Hz, then it will throw off all time dependent behavior. Capacitors and inductors no longer charge/discharge at the intended rate. Etc.

 

[digi001]

It's a sign of an inflexible simulator if you cannot adapt the timestep to your circuit's operating frequency.

If you simulate a 60Hz circuit as 6Hz, then it will throw off all time dependent behavior. Capacitors and inductors no longer charge/discharge at the intended rate. Etc.

I can pick any time step I want in my simulator (using Matlab/Simulink/Simscape/SimpowerSystems).

However the limitation is regarding the Resources on the intended Real-Time target. I am not able to choose a fast FPGA or Embedded Processor solution like OPAL-RT or RTDS. Instead our target must be C code which will run on a Windows PC. The Overall Model is very large so a time-step of 0.005 (200Hz) is already stretching it.

I understand the Power System dynamics will be effected as the impedance at 6Hz are different of that at 60Hz. Is there someway I can convert these all to get somewhat close results?

 

[BradtheRad]

The Overall Model is very large

The Power System consists of 4-5 (4-50MW) Generators, Induction Motors, DC Loads, Power Converters. Exact simulation detail is not required, only basic dynamic operation for Training/Learning Purposes.

Humans interact with equipment on a long timescale, so 6Hz is a suitable frequency to model that size of environment.

However to model specific events in 60Hz equipment, it is practical to create small scenarios. Example, a 60Hz transformer can overload at 50 Hz not to mention 6Hz. Therefore you would need to alter component parameters everywhere in your system, to make it behave as normal in 6Hz.

 

[digi001]

Yes exactly. 6Hz is good enough to develop an RMS that is OK for a User to see.

So for example regarding the Synchronous Machine. I am using the Model from SimPowerSystems-Simscape Technology. I have turned off any saturation effects.

It already has various parameters that I would like to use from a verified model. However I need to change it from 60Hz to 6Hz. There is an impedances tab.

Impedances Tab
Stator resistance, Ra
Stator leakage reactance, Xl
Stator leakage reactance.
d-axis synchronous reactance, Xd
q-axis synchronous reactance, Xq
zero-sequence reactance, X0
d-axis transient reactance, Xd'
q-axis transient reactance, Xq'
d-axis subtransient reactance, Xd''
q-axis subtransient reactance, Xq''

But from my understanding of this particular machine model.....frequency should be taken out of the equation with the DQ Transformation onto the DQ reference frame. Therefore these shouldn't need to change at all.

There is a Main Tab
Rated apparent power
Rated voltage
Rated electrical frequency
Number of pole pairs
Specify field circuit input required to produce rated terminal voltage at no load by
Field circuit current
Field circuit voltage

Here if I put in the new Rated Frequency, the documenation says it will compute the correct Apparent Power based on this frequency. The rest should not need to change?

There is also a Time Constants tab, which I believe should be independant of Frequency as the inputs are all in seconds.

 

[D.A.(Tony)Stewart]

Impedance, saturation core effects, surge currents, motor stored energy, core stored energy all would invalidate your simulation at 10% of line rate.

 

[BradtheRad]

Your parameters are numerous. I thought it would be a big job simply to change Henry and Farad values throughout the system. I suspected it would require several days of experimentation, because it probably is not sufficient just to multiply all your present values by 10.

This is a time when you would like for a simulation to duplicate both a macro world and a micro world, together, simultaneously. It is possible but it requires sophisticated software and hardware.

Because you have limitations in software and hardware, you wish to reduce demands by slowing down the pace. However it may pay for you to consider going the whole distance and convert your macro model to DC. There should be nominal DC values which correspond to AC values.

Your equipment has volts and Amperes and ohm values, while in stable operation. I suspect that those are the units you want to display at the macro level, instead of parameters such as leakage reactance and synchronous reactance.

 

[digi001]

I am starting to think that maybe there is no clear systematic approach.....?

I spent a few hours trying to get the Synchronous Machine to match when changing from 60Hz to 6Hz but it is quite tricky.

I know in general that the impedance of an inductor should be 10x more and the impedance of a capacitor should be 10x less so I could compensate for this to get the same system RMS dynamics by multiplying by 10x or dividing by 10. But things break down after that.

Yes representing things as DC or DQ is a possibility, (simulating that at a lower rate) then pulling out the actual waveforms at 60Hz (simulating at faster rate) only where I need them via inverse Park/Clark.

 

[BradtheRad]

impedance of an inductor should be 10x more and the impedance of a capacitor should be 10x less

Both L and C values should be increased, because you want longer time constants.

 

[digi001]

Ok I see.

Looking at possibly getting 1kHz sample rate in order to get true 60Hz simulation. Only have about 16 samples per cycle but should be enough. Worried about it getting unstable. Controls probably will still run at 200Hz timestep. Any experience here?

(How can I move this to a different forum? I think I posted in the wrong one)

 

[D.A.(Tony)Stewart]

You must have > 2 samples per Hz of loop bandwidth for frequency accuracy and much more for phase stability.

Try 960 samples per cycle but only compute 60 times per second using algorithm to process incremental delayed samples.

or get a fast Linux OS real time simulator.

Which Uni are you at?

- - - Updated - - -

Is this any better?

New Features of PSCAD v4.6
https://youtu.be/3OJZosjUQ7o?t=1190

**broken link removed**
install https://www.youtube.com/watch?v=z7hs66I13_E

This was developed by MB Hydro in my home town.

Training: Power Flow Analysis and Voltage Control by PowerWorldCorp 26,299 views
✔7 Things you didn't know existed ▶19 by Freeze HD Recommended for you
How to blackbox modules by PSCAD1 1,966 views
Module 6: High Voltage DC HVDC Transmission Systems Part 4 by mekocbey 1,096 views
Multi-EMTDC using PSCAD X4 by PSCADDEVEL 4,437 views
Webinar 31.10.2013 PSCAD pt.2 by Svetozar Pindikov 1,496 views

 

[digi001]

Well we are already pretty invested in Matlab/Simulink tools and Windows....but I have used PSCAD at my last job. (I've also been to your hometown Manitoba for some RTDS training a few Januarys back. Wow that was cold!)

I'm reading a book now called 'Multicore Simulation of Power System Transients', some interesting info on how to split things up.

"Try 960 samples per cycle but only compute 60 times per second using algorithm to process incremental delayed samples."
I dont understand this?

 

[BradtheRad]

Only have about 16 samples per cycle but should be enough. Worried about it getting unstable. Controls probably will still run at 200Hz timestep. Any experience here?

I programmed my own homebrew simulator. I found that 8 samples per cycle is sufficient to display capacitors charging and discharging.

Most of the time now I use Falstad's simulator. I experimented with the timesteps. I find it requires at least 2 mSec timestep with 60Hz. This is 500 divisions per second. Or about 8 samples per cycle.

 

[D.A.(Tony)Stewart]

An electro-mechanical system with vector impedance, inertia and back EMF will need a lo0t more than 8 samples per cycle unless the system is inherently stable with more than 60 deg phase margin or 1/6th of a cycle. 60 deg phase is what you expect from a 1st order integrator in a closed loop Op Amp at unity gain.

So your simulation platform is seriously under powered. The Chinese use Racks of servers to run that simulator software.