Voltage Oscillation Detection

[eRicC]

Hi there,

There is an application requiring the advanced PQ meter (i.e. ION7400 / 7650 / 9000) to detect the power quality of "voltage oscillation" on the distribution network. 

Such oscillation (in RMS voltage) is around 20Hz as shown in the attached photo.

Understood that currently there is no build-in module has such capability and thus would like to seek for the inspiration if anyone has similar application before (currently do think of using differentiator module with setpoint module for detection; or using averaging module to measure RMS voltage as baseline then use relative setpoint module for detection.  But not sure if any better way to build such framework).

Thanks.

[Tom_Stevenson]

Hi Eric - do you have more detail on the standard you are trying to accomplish?

Thanks

Tom

@DanL 

[eRicC]

Hey Tom,

Not a standard as required by the customer at this moment. 

Sorry maybe i should give a bit more background.

The customer encountered the reliability issue in equipment and found those voltage oscillation from the high speed log (in elspec meter).  Hence they want to have the ION meter to detect and flag them so they can handle it pro-actively.  

Given the voltage oscillation does happen and wonder how the other experts would leverage on the ION to monitor / detect it with the existing available module.

Thanks!

[DanL]

I would suggest using the RVC (Rapid Voltage Changes) and the Interharmonics magnitude features on the ION9000 (or legacy ION7650v416) to detect sub-second voltage amplitude oscillations.

1. The RVC feature can be configured to be quite sensitive to the start of these oscillations (but will also be sensitive to all rapid voltage changes - even the non-oscillatory ones). You may be able to BIN the RVC metrics to filter on characteristics you are interested in to improve triggering logic.

2. Assuming the oscillation are a true amplitude modulation of the fundamental power frequency (as opposed to just the modulation of a higher harmonic or interharmonic), then all modulation frequencies between 7.5Hz and Fundamental-7.5Hz (eg 52.5Hz for a 60Hz system) will appear as side bands in the first two interharmonic magnitude outputs of the HME (Harmonics Measurement) module (configured for interharmonics evaluation)

[sesa56307_bridg]

In addition to the ideas from @DanL, one other module I can think of that may be useful to consider is the Mains Signaling Evaluation module.

The Mains Signaling Evaluation module is intended to monitor a limited number of frequencies for signal content that exceeds a predefined limit. On each module, you configure the frequency of interest (down to 5 Hz minimum). Then, when you pulse the Evaluate input, it aggregates the values since the last pulse from the 4 5Hz bins closest to the configured frequency, and checks them against a defined limit (the algorithm is defined in the IEC 61000-4-30 standard). It outputs some statistical counters, as well as the actual magnitude. The statistical counters probably won't help you much, but maybe you could configure this module to monitor at 20 Hz on a given phase, then pulse it every second, and connect the actual magnitude output to a Setpoint module and trigger an action if the magnitude output exceeds a certain level? The oscillation would have to last at least one second for this to work.

You may need to experiment further to understand if this will fit your application. But at least it's another idea to try.

Best regards,

David

[eRicC]

Thanks for so much your advice and inspiration!

Yes did think of the RVC feature.

Based on my understanding, that's the compare the half cycle change against the 100/120 cycle average.  As shown in the attached pic, trust the RVC could detect the zone A, C, D (maybe), F & H.  The expected detection shall be zone D, E, F & G.

So it seems like eventually causing hundred of thousand of RVC event as the fluctuation of each half cycle could go beyond the RVC threshold during the voltage oscillation event.

Or maybe i get it wrong?  

Also will further study the use of HME.

[RobGraham]

I apologize, as this is as much for my information as a response to your question, but isn't a 20Hz periodic voltage distortion defined as flicker under IEC61000-4-15?  If so, the ION9000 and similar advanced PQ meters already have a Flicker Module available to measure its characteristics.

[DanL]

Flicker is somewhat sensitive to a 20Hz amplitude modulation of voltage, but also to many other frequencies (performance is validated from 0.008 Hz to 40 Hz). The reasons I decided not to mention flicker for this application are:

1. The flicker meter algorithms described in 61000-4-15 are weighted and tuned to the lamp-eye-brain model (60W tungsten incandescent lamp to human eye + brain response). As such modulation frequencies near 8.8Hz are given ~3x more "weight" than at 20Hz.

2. The flicker meter algorithms and the ION module outputs take quite a long time to register a flicker signal. The design is focused on 10 minute measurement and reporting intervals, which is far too late of a delay to trigger an alarm. Even with the ION Flicker module configured for it's minimum reporting interval of 60s, my testing with this scenario shows typical "detection" of a change in flicker output is 2 or 3 minutes (events are asynchronous to measurement intervals).

[DanL]

I do not see any attached picture with "zones" designated... did you intend to attach a new picture? I have found that the "insert photo" feature doesn't work reliably for replies....

RVC has a built-in hold off period where at most one RVC event can be triggered per second.

Here I have simulated approx 0.4s of a 20Hz voltage oscillation (amplitude modulation) with the RVC feature in the SagSwell module:

[sesa56307_bridg]

Just a small correction to my earlier post. As Dan said, the signal is in fact a 20 Hz modulation of the fundamental (and not a 20 Hz component in the signal itself). That means it will show up as side bands from the fundamental. So if you do explore the idea of using the Mains Signaling Evaluation module, you would need to monitor one of those side band frequencies, and not the 20 Hz frequency itself. For example, for a 60 Hz fundamental, the side bands would be at 40 Hz and 80 Hz.

I think you will probably find more success using one of the ideas from Dan, either the interharmonics from the Harmonics Measurement module (updates at most once every three seconds) or a solution based on Rapid Voltage Changes.

[eRicC]

Hi Dan,

Thanks for your reminder and the photo was not uploaded successfully.

With diagram attached I hope that would illustrate a bit better.  RVC could be triggered by comparing the recent half cycle against the previous average of 100/120 cycle, so under Zone D & E (which is the customer's concern and they request immediate notification as well as high speed logging during the event period).  So the RVC could be either triggered every half cycle and meaning that a separate sag/swell module (with very low event priority to avoid the overwhelmed alarms) together with other logic & timer module to identify such voltage oscillation.  If my understanding is correct?

[eRicC]

Thanks David for your inspiration!

Will run a test on the use of this module too given customer has a bunch of ION7650 which also want this feature be functioning on them while RVC is not available.

[eRicC]

Hi Dan,

Re the discussion of Flicker module, yes that do think the ultimate issue could be flicker related but as what you said the issue is that module is designed as per standard so it can only provide a statistical result but not designed for an instant flagging - while which is required by the customer so as to trigger a short period high-speed logging for analysis.

[DanL · Accepted solution]

No, as I said before the RVC cannot trigger more than once per second.

I simulated this signal with various thresholds and found that setting the RVC threshold between 0.3% and 1.0% worked great for this signal (perhaps start with 0.7%). Setting the threshold > 1.1% missed regions D & E, while setting < 0.3% triggered on every region.

Yes, you could link the RVC Start Event pulse (Convert #8 in the default framework of the ION9000 and 7650V410+) to additional logic including highspeed logging.

[DanL]

NOTE: The latest release of ION7650 firmware for v4xx firmware does support RVC.

If the meters are v3xx version, then they are not able to upgrade to v4xx (hardware compatibility break). Some very old versions of V3xx do not support line-to-line flicker nor line-to-line interharmonics (while in "WYE" volts modes).

[DanL]

As I said above, Flicker could be used but it will only register the flicker several minutes late so it will not be useful for capturing events long since passed. It is also sensitive to all step changes (RVC) and so will have many false triggers from even slow moving changes in voltage. For example, regions A, B, and C (without D, E, F, G, H) are likely to register a similar level of flicker as just regions D and E.

[eRicC]

Thanks Dan for the simulation and further explanation. That's extremely helpful!

And realize the application running on old ION7650 is unlikely possible given the capability and lack of high speed module to monitor and calculate.  Will stick on ION9000 (and new ION7650 v4.1+). 

Thanks again!