Getting Control of Electric Power
There is a temptation for control engineers to simply assume that facilities managers will magically provide electric power in sufficient quantity and of sufficient quality to feed the automated systems they create. These systems, however, have appetites large enough to disrupt finely balanced electric distribution systems, degrading power quality.
There is a temptation for control engineers to simply assume that facilities managers will magically provide electric power in sufficient quantity and of sufficient quality to feed the automated systems they create. These systems, however, have appetites large enough to disrupt finely balanced electric distribution systems, degrading power quality. At the same time, they are often finicky enough to get bellyaches when consuming power of less than perfect quality.
Therefore, it is important for control engineers to understand the basics of how to analyze electric power quality and how to remediate ailing distribution systems.
As independent power-system analyst Gerald Hajek says: “Power analysis involves almost anything and everything that might have to do with electricity, all the way from a generating station down to batteries inside of a DC-powered device. It compares what was expected to what we have, or what kind of a problem we have to what might cause it.”
Frank Healy, product marketing manager for power quality products at Fluke Corp., points out: “Harmonics do not cause the biggest complaints. When equipment goes wrong for some unknown reason, it’s usually things like voltage dips caused by motor startups, switching modes, faulty breakers, and all those kinds of things.”
“In addition,” says Dan Carnovale, power quality solutions manager for Eaton Corp., “issues include energy management and operational cost efficiencies. We like to break it down into symptoms, sources and solutions. That’s the way we do any kind of analysis.”
“Power chain management,” says Brandon Ekberg, senior business manager for the software and meters group at Eaton Corp., “is what we call controlling the complete power distribution system. The goal is to provide the tools, equipment and services to help our customers optimize the operation and use of that power chain. Basic power chain management includes a facilities audit to understand the current situation, what kind of risk your system might pose with respect to safety and its ability to withstand outages.”
“A lot of people can get the idea that harmonics are a huge issue,” Carnovale says. “While they’re important, there are really four main parts to power quality, and harmonics are just one of them. Another is transients. A third is voltage variations (sags, interruptions and swells). The last one is grounding.”
Being able to quantify such power anomalies makes it possible to understand their causes, which can provide clues about how to fix them. Quantification requires measurement. Simply capturing apocryphal observations, such as “so-and-so saw the lights flicker” is not enough. You don’t know how deep the event was or how long it endured. Without quantification, you really don’t know how to fix it.
Quantification requires measurement using specialized recording equipment. When we say “recording equipment,” we mean just that. It’s equipment capable of capturing electric-power anomalies over a significant period of time and archiving them for future analysis.
First step: measurement
“Metering,” says Ekberg, “is the ability to measure—in exact detail—the quality of the incoming power. You need baseline information as well as the anomalies to understand the situation before you pursue any kind of remediation.”
Use the "divide and conquer" strategy when siting power monitoring equipment.
If the nominal condition is a line voltage of 480 V, deviations from that nominal condition have two variables: how big they are and how long they last. These deviations affect downstream equipment in various ways. At some magnitude, a deviation may trip automated shutdowns or produce other unpleasant behavior. A long-standing over- or undervoltage condition can cause motors to overheat.
“More and more,” Healy reports, “we’re finding customers who take a proactive approach. They effectively do what we call benchmarking. They make measurements before they install equipment and then follow up with a study later to see what installing the new equipment may have done for (or to) them.”
Siting monitoring equipment follows Julius Caesar’s “divide and conquer” strategy. If you have a problem, place one or more monitors close to the misbehaving equipment to find out what’s going on. To monitor the overall health of the power-distribution network, however, you would first look at the main (incoming) service panel. Then you could install monitors closer to the various loads, moving downstream toward the loads.
“We have quite a few users employing quite a number of monitors,” Healy reports. “What they do is to take a whole bunch of monitors and put them on the system for three or four days, then move them. I was speaking to a customer last week who took at least 12 monitors to see exactly what was going on at the incoming parts of the building and then further down the chain closer to the loads and then right down the individual loads as well.”
Monitoring equipment captures voltage inputs and current inputs. On the voltage side, monitoring three-phase supplies requires five inputs: ground, neutral and three voltage phases. Ideally, you want to do the same with the currents. Healy recommends measuring at least four currents. You should always measure the neutral current because that shows the effects of harmonics.
To disturb the distribution system as little as possible, power monitors use high-impedance analog voltage amplifiers to collect voltage data. To measure current, they use clip-on sensors that respond to the magnetic field around the wire carrying current for each phase, ground and neutral.
Typically, monitors use current transformers (CTs) to pick up the alternating magnetic fields around wires carrying alternating current. CTs respond to the current in the wire through Faraday’s law of induction:
where E is the electromotive force (EMF) induced in the CT secondary, k is a constant based on the CT construction details, I is the current threading the transformer’s core, and t is time.
Clearly, CTs only respond to alternating current, so they cannot measure dc current. Faraday’s law also ensures that the EMF amplitude varies directly with frequency, so CTs distort harmonic-bearing waveforms.
To sense dc currents or dc components appearing on ac wiring, use a Hall-effect current clamp. Hall-effect sensors use a semiconductor chip called a Hall-effect magnetometer, which responds directly to magnetic fields. It can, therefore, measure dc currents as well as ac and faithfully reproduces harmonic-bearing waveforms.
“The simplest product we have,” Healy says, “is a little plug-in logger which just goes into a standard socket to look for dips and also measures RMS voltage. As loads switch in and out, you can see the effects on your electrical system and whether they affect control equipment, PCs or whatever other loads you have.”
At the opposite end of the spectrum, there are three-phase monitors, which capture a more comprehensive range of information. In addition to voltage events, they can measure harmonics and quantify flicker events, phase shifts and other important quantities characterizing the distribution network’s quality.
“The important thing is, it’s not just taking snapshots,” Healy points out. “The data should cover at least 24 continuous hours (or one work day), and ideally the study should extend over at least seven days as well. And in many cases, it would go beyond that.”
“If we’re doing a harmonic study,” he continues, “we take an average of harmonics and we look for the minima and maxima over the period—a typical period being 10 minutes. This provides a set of measurements showing what is happening now. Then we follow up with a later study to see what changes installation of new equipment have made. We’d record for a few days, or even a couple of weeks, then study trends of harmonics over that period.”
Raw data to useful information
Captured data is of no use until it is analyzed and reported. There are three ways engineers have of turning raw data into useful information:
On-board—Most power monitors have some on-board memory for data storage. Many have built-in analysis software as well. These units also have at least rudimentary human-machine interfaces to aid user analysis and report results quickly.
Local upload—Monitors often have some means of uploading archived data to a host computer for subsequent analysis. Often this takes place over built-in USB, serial or other short-range connections, or even “sneaker net” via memory sticks.
Network connection—More and more monitors are capable of uploading data via local area networks (LANs) such as Ethernet. This strategy allows users to permanently mount power monitoring equipment in cabinets behind service panels and in control enclosures. Engineers then run software packages on their personal workstations, which regularly upload the data over the facility’s LAN, save it and generate regular reports.
Especially in complex systems involving many individual monitors scattered about the electric-distribution system, reports of anomalies funnel into the engineer’s system for more complex analysis. Software running on the host computer can discover related effects manifesting at different points in the system and even automatically speculate on what might cause them.
Second step: analysis
“The monitor itself can help you analyze the symptoms that you see by picking up the waveforms, capturing trends in current and power,” Eaton’s Carnovale says. Sophisticated triggering methods allow monitors to capture tell-tale waveforms, such as voltage sags, inrush from motor starting, transient activity from events like lightning, as well as steady state anomalies like harmonics.
For example, if you had a voltage sag, the symptoms that you would see might include light flicker, variable frequency drives might drop out, computers might lock up, etc. You would then try to correllate these qualitative behaviors to the events captured in the monitors.
Let’s say you had one event in which the input dropped to 80% of normal voltage, and it lasted 10 cycles or 1/6 of a second. That might cause the lights to flicker, but it won’t necessarily cause a variable speed drive to drop out. Some loads may drop out and others may not.
If you had a deeper event, say dropping to 50% voltage for that same 10 cycles, you might lose motor contactors and drives and high-intensity lights. Even more severe events might knock out power to an entire facility.
Another thing to look at is your demand profile. Understanding the peak periods of usage can help you find ways to smooth it out to minimize impact from an energy usage standpoint or cost saving standpoint, as well as avoiding system overloads.
From a reliability standpoint, if harmonics ramp up because you are adding more and more harmonic loads, you might start causing overheating and transformer damage. Those kinds of things can be flagged on a routine basis by software running on the engineer’s workstation.
“Harmonic analysis is the first thing that I do,” consultant Hajek points out, “but what other methods are there? I always like to look at the voltage signal that occurs between the neutral conductor and the ground conductor at the point-of-equipment use. I also like to know how many linear feet away from the actual source of power the load is located. In other words, are we talking about a problem 400 to 500 feet away from a service entrance or are we talking about a problem 50 feet away from a service entrance?”
Third step: remediation
The advice control engineers get most often regarding electric power remediation is to add filters to their drives. This is good advice only when power problems are caused by the drives themselves. And it only helps remove harmonic distortion.
As Fluke’s Healy pointed out at the top of this article, harmonics do not cause the biggest complaints. More often, problems arise from things like voltage dips, transients or grounding problems.
“The very first thing I would tell anyone to do is minimize inductive and capacitive reactance in the entire circuit,” Hajek says.
You can also reduce reactive impedances by carefully selecting transformers and generators. Lower impedance units may cost more, but can pay for themselves by reducing power-quality headaches.
Monitors scattered around the power-distribution system funnel data to a central database where power-monitoring and analysis software helps engineers understand the big picture or mine dow to tlocal details.
Use higher voltages and lower currents whenever possible. Reducing currents allows you to use smaller-diameter conductors. Smaller diameter conductors have lower reactive impedance per foot. For a given wire run, smaller diameter wires present lower reactive impedances.
Of course, you can reduce the total reactance of a wire run by making it as short as possible. In other words, carefully lay out wiring routes to minimize distance. Minimizing wire lengths makes good economic sense as well.
This strategy can conflict, however, with the need to provide a neat, workmanlike installation. It is important to properly bundle wires to ease route tracing and avoid safety hazards. As with most design choices, it is a matter of finding the most advantageous compromise. So minimize wire-run lengths consistent with other design requirements.
Hajek also insists on using a separate neutral for each feeder. “Everyone starts moaning and complaining when I say that because it costs more money for wires,” he says, “but it costs less in the end.”
Finally, there are no pat solutions that will work in every situation. “Usually, the more difficult problems in this industry are not problems that you can solve in, say, an hour and a half some morning,” Hajek says. 'It’s going to take some gumshoeing detective work to ferret out the solution.”
To be successful, your analysis must be based on understanding of the basic physical principles underlying the situation, whether it’s electromagnetism telling you to reduce reactance with thinner wire or electrochemistry telling you about grounding problems affecting your pipes. You have to be similarly creative in devising solutions.
“Say a machine is just occasionally getting hung, and you find the PLC is resetting,” Eaton’s Ekberg hypothesizes. “Monitoring the incoming power to that PLC may show surges or voltage dropouts that the PLC can’t handle. A remediation might be to put a small uninterruptible power supply in the control cabinet so the PLC always gets very, very clean power, even when the mains power may drop out for a few seconds.”
“If power analysis is required,” Hajek says, “something is not working as intended. One must carefully compare what was intended to that which resulted. The difference is the magnitude of the fix.”
“If it’s a machine shop having an issue with power glitches,” Ekberg points out, “and you lose five or 10 minutes of production time, it’s not that big of a deal. That’s very different from, say, a silicon wafer foundry where a large power spike can bring all their equipment down and lose millions of dollars. Understanding the impact of the problem is critical whenever approaching power distribution issues.”
C.G. Masi is a senior editor at Control Engineering. Reach him at firstname.lastname@example.org .