Transcript

Introduction to Merlin™ Strip Chart Analysis

Afternoon, everyone, and welcome to today’s white paper webinar. Today, we’re here with Vinny, and we’re talking about Merlin and specifically how Merlin looks at strip charts.

Merlin is our power quality AI assistant in PQ Canvass, and Merlin looks at strip charts, looks at waveforms, looks at compliance reports, and applies a whole slew of heuristics that we programmed in, and applies IEEE standards to what it sees. It’s doing a visual analysis in many of these cases with waveforms and strip charts, and that’s the ground-level evidence that it’s looking at, and then there are layers that interpret this.

Here, we’re going to talk more on the strip chart analysis. We’re gonna talk about how it works and show this in PQ Canvass, and also talk about how it relates to the top-level analysis that Merlin gives you. When you’re using Merlin, you don’t always dive into this level of detail, but if you need to or there’s evidence in here that you want to access, we’ll talk about how to get to that and how Merlin did this.

Why Strip Chart Analysis Matters

Hi, everyone. My name is Vinny. So today, we’re gonna be going over the strip chart aspect of Merlin. In many investigations, the strip chart is the first place that you go to look at, and to understand it, you have to go through them, and it tells the entire story of the whole recording. But having to go through them is a lot of work. What Merlin really does is it gives you what it finds as some of the most interesting areas.

Manual review was traditionally inefficient. Any seven-day recording can contain thousands of integral data points, and when you look at it from a high level, you don’t really see much, and so you really have to dig down to it. For a human, reviewing the data is a constant battle between that volume and attention. It’s tedious to scroll through them all.

What Merlin really does is it helps alleviate a lot of this. It shifts it by applying an AI-driven pattern for the entire strip chart recording. Rather than treating each channel in isolation, it evaluates the whole context of the data.

Six Analysis Modules

What it finds in the strip chart is organized into six distinct analysis modules:

• Voltage sags
• Voltage swells
• Loose neutrals
• Voltage regulation
• Flicker
• Harmonics

Navigating the Strip Chart Section in PQ Canvass

When we go into PQ Canvass and look at any Merlin analysis, this is a recording from a residential customer, you’ll notice there’s a strip chart section over here. Here, you can see the distinct breakdown of what it finds inside the strip charts. You have each of the six sections, and any of these numbers, you can look at the legend over here to look at what the severity means, as it correlates between them.

Each time that you click down, it’ll have a detailed analysis, where you can actually go through and see what it finds for this specific area. One, like voltage swells, it sees as a high severity, which means that this is probably something that we’re gonna wanna take a look at. Something like loose neutrals that has a severity of one means that it’s not really affecting any of the recordings. It doesn’t really find anything of interest to us. All of this helps us by really pointing us in the right direction of this analysis.

Voltage Sags: Context and Attribution

The first one is gonna be voltage sags, context, and attribution. Its primary goal here is attribution, distinguishing between voltage sags caused by a customer’s own equipment and those originating from the utility grid. For each of the sag that it identifies, there’s a clear determination, customer source or utility source, along with supporting evidence.

One of the things that’s cool is when you actually look at it, Merlin will support what it finds with evidence. You can go in and investigate it further. You receive a definitive answer that they can present to a customer or use to prioritize field of work.

In this one, we have a severity of five, which is about moderate. You can see it says that it detects that there’s minor voltage dips with load spikes, rare deeper utility dip, and it gives you the whole analysis of everything that it finds.

Here, for example, like Vinny said, it’s kind of a mixed bag here. There are voltage sags that are caused by the load, by the customer, along with utility-caused deeper sag, which is probably an upstream fault. This is a one-line summary, and based on the size and frequency of both of those, that gives it a five and severity moderate.

Then here, it’s telling exactly what it saw, when it saw. Here’s where it saw the deepest sag. Here, it’s not accompanied by a significant current change, so it knows it’s from the utility side. And then here is what it saw with customer loads, so we see current spikes in the 130-140 amp range that are causing two to four-volt dips. Merlin does this analysis for you.

Viewing Raw Strip Chart Data

You could see this yourself in the strip charts, and if we go to the raw data, here is the RMS voltage and current in strip charts. If you’re doing this by hand, you would zoom in and see these current spikes and current sags. Some of them align up with the current and some of them don’t. The ones that do line up are caused by the customer. The ones that don’t are caused by the utility.

For example, these little voltage changes, those are caused by the customer. You got nine amps here, and you got a couple of volt sag. Others are not, but Merlin does this kind of grunt work for you, and that’s where it shows up here in the strip charts.

Top-Level Analysis and Voltage Regulation

As the paper talks about, you can start at the top with the overall power quality report, and here is the summary of the entire recording from a high level, each of the issues that it saw in the recording. Voltage regulation is really the biggest part of the story here in this recording, with over-voltage, some voltage swells, sags. And generally, is it utility-leaning or customer-leaning in terms of cause and the evidence, the snapshots.

Often, you would just stay at the high level, at the Merlin top-level analysis. Here, voltage swells and steady state voltage regulation are the highest priority ones, so we’ll jump to voltage regulation. Here, it’s telling you what it found, and this is a combination of strip chart evidence and the compliance reports for ANSI C84, and even waveform captures.

But if we want to go deeper, if we want to see more evidence or understand more of why it claims what it did, we could go to the strip charts and if we’re talking about voltage regulation, you would click on the voltage regulation strip chart. Here, as Vinny pointed out, you have these severities are higher for voltage swells and regulation, they’re seven, that’s closer to the high. This is where the real problem is.

We can click on that and see what Merlin saw out of these strip charts. When did it see it? What did it see? It’s talking about the fact that it’s going to almost a 7% high, and here’s the voltage ranges, and how it relates to ANSI C84. At the bottom, we have these regions of interest. These are particularly interesting sections of the strip chart that Merlin has seen. This shows you what it was and the start and stop time, so you can look in the strip chart to see this for yourself.

If we go into the strip chart with a manual analysis, we’ll see that, well, yes, this is a whole block here where the voltage was high for almost an entire day. Merlin obviously picked up on this. This is a very straightforward one, and gave a compliance-related analysis of it and arranged this by severity.

How Merlin Puts It All Together

In general, that’s how Merlin works. It’s looking at waveforms, it’s looking at strip charts, and looking at compliance. Here, voltage regulation is a 10, severe, because of this big fail on ANSI C84.1, and then putting all those pieces together in the top-level report.

This paper is focused on the strip chart layer. This is one of the ground layers of Merlin actually looking at data. This is one of the hands-on pieces of Merlin. We can see some of these that are low severity, like loose neutrals, as Vinny said. That’s severity one, basically not a problem at all. There was no loose neutral here, so this part of Merlin is focused on purely is it a loose neutral problem. It’s looking for those signs of one leg going up, one leg going down, but it didn’t find that at all because this is not a loose neutral issue.

These are low severity and then here are the high severities. The strip chart section of Merlin is one of those hands-on Merlin looking at raw data layers. If you wanna look with Merlin, just click on one and it’ll tell you exactly what it saw and exactly where it saw it. But again, start at the top level, because this ground level evidence layer feeds all the way up to the top. And of course, you can always go back to the raw data yourself and look at it.

Chatting with Merlin

Now another aspect that’s new since the paper was released is the chat. You can ask Merlin about the recording. When you’re chatting with Merlin, Merlin knows everything about that recording in its memory, so you’re chatting with a PQ assistant that is familiar with all the IEEE standards and exactly what happened in this recording itself.

I was asking about the voltage regulation of this recording. Well, it’s the main problem, and here is what it saw. We were sitting at 122 to 128 volts most of the time, and then there was a major overvoltage episode on the evening of the 13th. We were at 133 to 136 volts, so we were at over 10% high. That’s an overvoltage condition that’s a big problem, and Merlin knows about that.

We can also say, well, apart from regulation, what else is a problem here? And you can brainstorm with Merlin. You can ask it, say, “What if we upsize the transformer? What would that do to the flicker? Would that bring us into compliance?” You can have it walk through some impedance calculations with you. If you give it the transformer size and the service connector size, it will take that into account. You can brainstorm solutions with Merlin. It’s a very good engineering partner.

Here, we’re talking besides overvoltage, we have harmonic distortion is moderate, five to 6%, in some cases approaching eight to 10%. That’s kind of a high. Pointing out that we didn’t record voltage distortion, but it’s pulling this from the waveform captures. So it’s looking at everything and gleaning everything it possibly can from this. Some rapid voltage changes, some flicker. Not too bad, but noticeable. It’ll give you a good summary of the recording, but you can also drill down and ask it questions directly from that.

Getting Started with Merlin

If you have any questions about the paper, feel free to type them in the questions box. Or if you wanna try Merlin, simply go to our website. The link is right here to request demo access. With this, you could actually look at these files and play with Merlin yourself to see what it can do.

Merlin and Data Compatibility

“Does the data stay within PQ Canvass? Is Merlin native to PQ Canvass?” Yes, Merlin lives inside of PQ Canvass. Now, Merlin will soon be able to ingest other data besides power quality files from our recorders. Merlin will be able to import PQ diff or Comtrade format data. So if you have data from relays in a substation or permanent monitors, maybe smart meters that have PQ ability, Merlin will be able to ingest that by putting it inside PQ Canvass and letting Merlin analyze it.

Merlin will be able to analyze more than just our power quality recorder. If your device can produce waveforms or strip charts, voltage or current or other power data, Merlin can analyze that.

That’s all the questions for now. Again, if you have a question later, give us a call anytime at 1-800-296-4120 or send an email to support@powermonitors.com. Thanks for attending everyone. Everyone have a great afternoon.