Transcript
Introduction to Voltage Sags
Good afternoon, everyone, and welcome to today’s white paper webinar. Today we’re talking about voltage sags. This is the most common power quality problem that occurs in distribution. This is really an unavoidable PQ problem. It’s really impossible for utilities to completely eliminate voltage sags, and it’s a very costly power quality problem.
It’s the most costly problem related to power quality, and that’s especially the case for industrial commercial customers. Voltage sags are commonly responsible for equipment dropouts, contactors that drop out, VFDs that don’t have a lot of ride-through ability, and even cause flicker and other consumer complaints.
IEEE 1159 Definition of Voltage Sags
So let’s start with what a voltage sag is. Here is the official definition of a voltage sag from IEEE 1159, a really crucial standard for anyone involved in power quality. As you can see here, it’s a decrease in RMS voltage to between .1 and .9 per unit at the power frequency, for durations of a half cycle to a minute.
That’s a pretty wide time range. Between a half cycle, which is eight milliseconds, all the way to a minute, that’s a long span. Anything beyond a minute is considered a steady state issue by the IEEE, by 1159. Anything less than a half cycle is considered a transient. So this is the range between steady state under voltage and something that’s so short that it doesn’t affect the RMS value much.
The other range that we have here is the retained voltage. This .1 to .9 per unit is the percent of retained voltage. So .1 per unit would be 12 volts on a 120 volt nominal. Most customers would consider 12 volts to be more of an outage than a sag.
Sags Are Caused by Current Flow
The distinguishing factor between an outage and a really deep voltage sag isn’t really the specific retained voltage. Rather, what separates those two is that a sag is caused by current flow, either a fault current that’s flowing or large inrush currents from loads where, for example, a hard start motor whose inrush current is very high during the motor spin-up. So sags are caused by current flow resulting in voltage drop.
If we have, for example, a customer that has a high inrush current motor, when that motor starts, the current it draws could be eight, even ten times the running current, and that current flows through the system impedance, through the distribution wiring, through the service drop, the customer, through their transformer. All those elements have resistance. And from Ohm’s Law, if you have current flowing through that resistance, you’re gonna have a voltage drop across it. So the voltage at the load is gonna be smaller than the voltage at the source, and hence a sag.
If this current is steady state, then the voltage drop will be steady, and then you’ll just have low voltage here at the customer. If the current has a spike to it, if it’s an inrush current from a hard start motor or a fault current that flows very briefly, then you’ll have a voltage sag, where the voltage drops for the time that that high current was flowing.
In an interruption or an outage, on the other hand, you have an open circuit. You’ve got a break in the circuit, and there’s no current flow at all. So really the distinguishing factor between a sag, a deep sag, and an interruption is the fact that with a sag there’s current flow somewhere. There’s a complete circuit. It’s just that it’s so much current that you’re having low voltage. Whereas with an interruption, there’s an open circuit.
IEEE 1668 Standard
Jumping back to the white paper, it talks about the 1668 standard. This is a sag standard by IEEE that’s very specific to sags, and it’s really geared towards industrial customers who have equipment that could be adversely affected by voltage sags. In many cases, these are motor controllers or contactors that drop the contacts or the power to the load if the voltage sags to a certain level. These could also be sensitive electronic equipment and variable frequency drives, other types of process equipment.
So the standard is really written towards industrial customers and how they can test for sag withstanding ability in their equipment. But it’s also an important standard for utilities or anyone else affected because it has a really good primer on voltage sags at the beginning.
Section IV is a very comprehensive and well-written section on what causes voltage sags. How do you measure voltage sags? Why are sags virtually impossible for utilities to completely get rid of? Given that impossibility, we have to be able to deal with these sags. How is the voltage change related to the fault current or inrush current? And that’s a really good background for both utilities and utility customers.
The rest of the standard is geared more towards the fine details of measuring voltage sags and how to test equipment for sag withstanding ability. How rugged are they towards a drop in voltage for a certain amount of time? That is more geared towards the industrial customers, and hence that’s why this standard is by the IAS, the Industrial Society of IEEE, rather than the Power Engineering Society, which is more focused on electric utilities. But there’s a lot of overlap. And this primer on voltage sags is really good information for folks on both sides of the revenue meter.
Retained Voltage and Duration
As the standard points out, you can talk about sags in terms of the retained voltage and the duration. Those two numbers are really all you need for simple sags to quantify how severe they were. The duration is how long the voltage was low. We talk about the beginning of a sag as when the voltage drops to 90% retained voltage. And the voltage is considered returned to normal when it reaches 90% of the original voltage. So the duration of the sag is at the 90% points, where you have 90% or .9 per unit.
RMS Capture Graph
Here I’m annotating the RMS version of this voltage sag. When you have a waveform that’s captured, you’ll normally start with the raw sine waves. We can see all three phases of voltage, and we see that one phase has sagged. The other two have actually gone up in voltage. That’s not uncommon.
When you have fault current flowing, the faulted phase will pull the neutral away from the ground, and that actually causes the other two phases to increase in voltage slightly. So you’ll see a voltage increase or a swell on the other two phases and a voltage sag on the faulted phase. Phases A and B are increasing in voltage, black, red, blue for A, B, and C. And phase C, the blue trace, is dropping in voltage.
But it’s hard to see the voltage change when you’re looking at peak-to-peak voltage waveforms. If you graph the raw sine waves, you’ve gotta graph the negative peak and the positive peak, and that can really obscure the level of detail. A better way to look at these sags is with the RMS capture graph. Here the software, in this case we’re looking at PQ Canvass, will apply a one-cycle window to the data and move that along one sample at a time, producing the RMS value of the traces. So what we have on the bottom graph is the RMS value on a continuous basis from these raw waveforms. And here it’s much easier to see what the retained voltage was and what actually happened, and also to measure the 90% points.
When to Use the RMS Capture Graph
When you have a waveform capture, you don’t know if it’s an oscillatory transient or a sag or some other issue. Look at those sine waves, and if it’s sinusoidal all the way through, like these are still sine waves just at different amplitudes, well, that’s an RMS event. If it was an oscillatory transient, then you would not use the RMS capture graph.
This is what a sag looks like. The amplitude drops, but you still have a sine wave shape. If you have a waveform capture that looks like this, well, that’s an oscillatory transient. The shape of the waveform has changed. This isn’t a voltage sag, so don’t use the RMS capture graph. That’s gonna obscure these waveshape changes. But if it truly is all sine waves all the way through, just an amplitude change, then it’s appropriate to use that RMS capture graph.
Viewing Sags in PQ Canvass
You can do this in PQ Canvass. Here I have a recording from a Seeker™ that’s actually installed on a substation. This is measuring on one feeder. It’s installed in the breaker cabinet, measuring the current and voltage going out from that feeder.
We can look at the RMS voltage and current. Here I’m using PQ Canvass, so this is in a web browser rather than using ProVision™. We can see the voltage and current. We see current spikes here in purple. We see some voltage sags here in green. And if we zoom in, we’ll have situations where we see a deep voltage sag. That’s fault current flowing somewhere on that bus, but not necessarily from the feeder that we’re monitoring, because this bus has several feeders.
Once we see that we have sags in the strip chart, we wanna dive in and see a little more detail. Because with these strip charts, you get one cycle min and max, but you don’t know how long the event lasted. For that, you need to go to the waveform capture.
If we go to waveform capture, we see the list here. We see that we have some periodic captures, and we see one here where we have an RMS voltage change. So I’m gonna launch this one. Now we’re looking at the sine waves, and we see voltage in red, current in blue for all three phases at this substation on this feeder. We can see that phase C is the phase that we have a fault. We have two cycles of fault current flowing on phase C. These are at five amp metering secondary, so when we see roughly 10 amps peak, that’s with a five amp metering secondary. I believe this is 120 to one. That’s about 1,200 amps of fault current flowing on that feeder.
Again, we can see that the waveshape hasn’t really changed dramatically. This is basically a fault causing a voltage sag. So let’s take a look at the RMS values. We simply click RMS up here at the top left, and now PQ Canvass is redrawing this waveform capture, showing us the RMS values with a sliding window. It’s easy to see if we hover over any point. The voltage before is 116 volts. The retained voltage during the sag is 65 volts, and we can see what that is per unit. We can look at the duration. It’s roughly 30 milliseconds when it started and about 72 milliseconds when it stops. That’s about 40 milliseconds of sag time. This is an easy way to quantify what’s happening with PQ Canvass.
Again, the RMS capture graph is a better way of looking at these sags once you’ve confirmed that it is indeed a sag and not some other type of event, like a transient of some sort.
Configuring the Recorder to Capture Sags
You can also configure the recorder to capture sags with certain parameters. The white paper shows this in PQ Canvass. Just for completeness, I’ll jump over to ProVision™ and show how that’s done. If we want to initialize a recorder, we go to Advanced and look at the waveform capture setup. The fault is usually a good starting point. This captures most voltage sags.
You can capture them with a percent voltage change. That’s the best way to do this. And this 3% means that if the RMS value changes by more than 3% from one cycle to the very next, it will trigger a waveform capture. So that doesn’t mean the voltage is just 3% low. It has to change by that amount in one single cycle, from one cycle to the next, which indicates a voltage sag, because the RMS voltage is changing rapidly. If the voltage is just slowly creeping up or down, this waveform won’t trigger. That’s not a waveform capture, that’s a steady state issue. So the percent change voltage threshold is really the most effective one for voltage sags. And the default of 3% will capture most sags.
Another adjustment you may wanna make is the number of cycles to capture pre and post. The default here is two before the trigger, six after. You can adjust this. In fact, you can set it to over 1,000 cycles pre or post, and you may wanna do that if you’re doing, say, a motor start study. If you’re with a customer and they’re starting a very large motor and you want to capture exactly what happens, you can actually capture the entire waveform for many, many seconds if it takes several seconds for that motor to spin up. And then you can look at the sine wave waveforms or you can use the RMS capture graph in ProVision™ to see the same thing you would see in PQ Canvass.
Determining the Root Cause of Voltage Sags
One common issue with sags is the customer that’s having problems isn’t always the one that’s drawing the current causing the problem. You may have customer one with this hard start motor that’s producing lots of voltage drop, and anyone behind him on the feeder is gonna see that voltage sag also. So it may be customer two over here downstream that has the sensitive equipment or is experiencing a lot of light flicker or other problems from the sags.
One of the first things to determine is: you’ll get the complaint from this customer. This customer doesn’t really care about the problems it’s causing. It’s not affecting him. So you’ll set a PQ recorder here, and one of the first steps in the investigation is to determine, well, is the customer causing the voltage sags or is it on the primary side and some other customer is causing it?
If you have voltage sags, that means there’s current flowing somewhere causing that voltage drop. The question is where is it? Because mitigation on the utility side generally means either lowering your system impedance and upsizing the transformer or wire or having the customer lower their current spikes. But you need to know where that current flow is occurring before that mitigation can really be applied.
Correlating Voltage Sags and Current Spikes
One of the first ways of doing this is to look at the RMS voltage and current strip chart and look at the relationship between the voltage sags and current spikes. Here we have one channel of RMS voltage min, max, and average graphed on the left axis, and then the same phase of current with min, max, and average current. These are one cycle minimums and maximums. And then the average over the averaging interval, whether it’s one second or one minute. No matter how slow the averaging interval is, you’re still getting single cycle resolution on the voltage and the current.
Here we’re not concerned so much about sag duration as we are about what’s the root cause. Is this customer causing the voltage sag or not? What we wanna do is look for correlations. We have current spikes at the same time as voltage sags. That’s a pretty strong one-to-one relationship. But further cementing that is the fact that when we have a lower current spike here at the last one, we have a proportionally lower voltage sag.
Remember that change in voltage versus change in current is based on Ohm’s law, V equals IR. So if you have a certain delta I, that’s gonna produce a certain delta V providing the R is constant. And that R is the resistance, the source impedance, and that’s based on the transformer size and the wire size, and that’s not changing during the PQ recording. So if you have a certain amount of change in current, that should always produce roughly the same change in voltage. If you have a recording with a whole array of voltage sags, if the customer is causing those, those current spikes should be proportionally sized, and they are here in this example.
We have other examples where the customer is not causing the current. We have inrush current causing a voltage sag, and we have other sags that are present in the recording. They don’t have any change in current. So these sags are not caused by the customer. Some other customer on that feeder is drawing current to produce these sags, but not the one we’re monitoring.
Then you’ll have situations where it’s very clear. Here we have a zoom in of a 40 volt motor start. High inrush current, a couple hundred amps, causing a couple of second long voltage sag. If you look at the strip chart interval, this is a two second interval, and this motor start is so long it takes four seconds to start up. We’re dropping the voltage 50 volts. That’s roughly under 90% retained voltage during this 200 amp inrush current in the motor.
Sag Classification and Standards
To see more detail, we’d need to look at the waveform capture and RMS capture to get a precise time if we want to see where this falls with the 1668 standard or with the CEBEMRA curve or the CEMA F47 standards. Both of which judge voltage sags based on a duration and a magnitude. The 1668 standard gives guidelines for equipment manufacturers as to what sort of duration and what sort of depth voltage sag is acceptable.
It goes into more detail as we describe in the paper on things like, is it a phase-to-phase or phase-to-neutral sag? Is it a type I, II, or III sag? That becomes important especially to equipment manufacturers and talking about their ability to withstand certain types of events. This is information that’s found in the 1668 standard. What type of sag is this? Is it a single phase sag? Is it both phases, if it’s a phase-to-phase fault, or do all three phases drop?
Angle of Inception
Another interesting metric that is more commonly used on the industrial side is the angle of inception. This is more specifically when in the waveform the sag starts. Does the sag happen at the peak of the waveform or somewhere else? That doesn’t really matter in terms of quantifying it in the magnitude or duration. The point of inception on the waveform isn’t one of those two numbers. But it is important to equipment manufacturers.
Contactors and capacitors and other types of devices have more or less stored energy depending on where in the point on waveform this sag hits. So if you’re in the fine details of compatibility with an equipment manufacturer or with a customer, it may matter where this sag starts. Most sags from faults, for example from arcing, tend to happen at the peak of the waveform. That’s when the voltage is highest, and that’s when insulation is most likely to fail or you’re likely to have flashover. Sags that are caused by tree branches or animal contact may be more randomly distributed.
Contact Information
That covers the material here in the whitepaper. If you have any questions about voltage sags, give us a call anytime at 1-800-296-4120 or send an email to support@powermonitors.com and we’ll be happy to talk.