Has your utility ever been challenged by an industrial customer when measuring voltage fluctuation and light flicker? How did your utility handle it? Were your utility's standards up to date and technically valid? When Rocky Mountain Power (RMP; Salt Lake City, Utah, U.S.) faced a problem with light flicker a few years ago, the utility had to change its approach. Although many utilities still use the old method for dealing with flicker, RMP is finding the new method to be much more accurate.

WHAT IS FLICKER?

Light flicker can be caused by a poor electrical connection or a bad fluorescent ballast. However, the cause that gets the utility involved is flicker that results from voltage fluctuation on the system. It is important to draw a distinction between flicker and fluctuation. Voltage does not flicker. But voltage can fluctuate, and this sometimes causes lights to flicker.

Load fluctuations caused by arc furnaces, X-ray machines, welders and so forth can be the source of voltage fluctuations on the utility system. Motor-starting currents are another class of load fluctuations. When a flickering light becomes irritating, to people or the equipment, it becomes a problem. The difficulty that arises is that voltage does not need to fluctuate much to cause lights to flicker perceptibly — sometimes less than 1%, which is far less than the steady-state ANSI C84.1 voltage-range limits of ±5%. At this low level, equipment is usually not bothered, only people.

THE OLD WAY TO MODEL

Utilities model flicker at the planning/design stage and measure it at the time of verification or troubleshooting. The old way of modeling flicker at the planning stage is to calculate the voltage fluctuation in percent and then estimate the rate of fluctuation. This is then compared against the old GE flicker curve, or some form of it (Fig. 1). Any calculation that falls below the line is acceptable, and the customer can be given the green light to install the load. Loads above the line require some sort of change or mitigation.

The Fig. 1 diagram is based on test data that indicates people are more sensitive to flicker the faster the flicker occurs, up to about six to eight times per second. The voltage fluctuations driving this flicker are usually applied as two voltage changes per fluctuation.

The biggest problem with the old way is it is based on sudden voltage fluctuations following a regular pattern (50% duty-cycle square wave), and loads seldom fluctuate on that schedule. Also, load fluctuations often do not change suddenly, but have shapes to them that are not square. Voltage change caused by a motor start is one example of this.

When measuring fluctuating loads that are already in place, the old method is easy when the fluctuation rate is slow enough to measure with a voltmeter. But for fast fluctuations or fluctuations that are of small magnitude, the voltmeter is inadequate. And for an arc furnace or other load that is random in magnitude, shape and rate of fluctuation, the old way of measuring totally breaks down, as it did for RMP.

A WAKE-UP CALL

Several years ago, a steel company that wanted to convert its gas furnace to an electric arc furnace approached PacifiCorp, RMP's parent company. Instead of polluting the air, the steel company would be polluting RMP's power system. So the utility started to ask a few questions: What was the voltage magnitude change? What was the rate of fluctuation? How abruptly did the voltage change?

There were no clear answers to these questions. And it wasn't long before RMP learned that the old way of looking at flicker would not work. Fortunately, the IEC and the IEEE had been looking at these issues for quite some time and had come up with methods that worked with arc furnaces and other pesky loads that didn't fit the old mold.

THE NEW WAY TO MODEL

The new approach to modeling and measuring flicker first became a standard in Europe as IEC 61000-3-7 (modeling and limits) and IEC 61000-4-15 (flickermeter). Later, the flickermeter standard and some flicker limits were adopted in North America as IEEE 1453-2005. As shown in Fig. 2, the new method is based on a sophisticated model of the voltage fluctuation, a common incandescent lamp, and the human sensing and perception of the light emitted by the lamp.

This last stage includes the formation of a single number representing the human perception of flicker during a given interval of time. This perception number is called P_ST for a standard short-term (10-minute) interval, and P_LT for a standard long-term (two-hour) interval. Despite the seeming complexity, the output of this process is simple: if P_ST is less than 1.0 at the lamp socket, flicker is not a problem; if P_ST is greater than 1.0, it is a problem for the majority of people.

Finally, the new flicker standards include guidance on how much of the time P_ST flicker can exceed limits. This always had been a weakness in the old way. Using the old method, what if a customer exceeded limits just one time in the measurement session? Did that mean the load didn't pass the test? The old way was silent on this point, but the new standards give guidance in this area. Further, the new meters and methods are friendly toward such analysis, giving some leeway to occasional limit violations.

THE REAL WORLD

After being introduced to the new look at flicker, RMP has found that there are still a few problems with this method, mostly because it is new.

For one thing, unlike voltmeters, there are not that many digital flickermeters of proven accuracy. The problem is not so much with the meter manufacturers as it is with the existing 61000-4-15 testing protocol, which is too vague. When looking for a flickermeter, or a power-quality monitor that has a flicker-measurement module, be sure to ask if the meter has been tested against the developing CigrÉ C4.1.6 testing protocol. Verifiable flickermeters are now available that have been tested under this protocol.

A second challenge is that people generally do not understand P_ST. It is too new, and there is a learning curve. People are accustomed to measuring voltage with a voltmeter. They can relate to that. But P_ST, though more accurate for flicker, is still pretty foreign to most. This is something that only time and training will correct.

A third challenge is that an accurate flickermeter might reveal uncomfortable pockets in a power system where traditional methods had previously never shown a problem. Such cases must be handled carefully. Often, in an industrial area, these pockets of higher flicker are not a problem because the lighting is nearly all fluorescent and, hence, much less sensitive to voltage fluctuation. Also, there might not be residential load in the area.

BETTER MODELING TOOLS

The new way of looking at flicker also has suffered from the lack of commonly available tools for accurately modeling the PST of prospective load. While this subject can become complex, there are some better and simpler tools emerging now to meet this challenge, such as the Electrical Pollution Screening (EPS) tool (Fig. 3a).

The EPS is designed to be a 5-minute modeling and screening tool with software that runs on any handheld device using the WindowsMobile operating system. Developed by Dr. Mark Halpin and colleagues from Auburn University, with funding by PacifiCorp, its function is to model most loads connected to their corresponding distribution systems, and to test these loads against limits for voltage sags (motors), harmonic distortion and flicker. If any of these limits is violated, the output screen displays a frowning face instead of a smiley face (Fig. 3b). Additionally, the degree of compliance with the limits is displayed in industry-standard terms, something slightly more quantifiable than a smiley face.

The EPS is being released at nominal cost as a service to the power industry, so that utilities and their customers can accurately and comprehensively screen for potentially “polluting” devices from their power systems. The EPS can even perform a basic frequency scan to check for resonance involving capacitors. It will soon be offered by the Power Quality Service Center at www.pqsc.org.

WORKING WITH CUSTOMERS

The two case studies summarized on page 29 illustrate a few steps in working with customers on power-quality issues:

1. Have the flicker standards in place, if possible, to handle issues before loads appear on the scene; this goes beyond ANSI C84.1 (to IEEE 1453 for flicker)

2. Be sure that load sheets for larger new-customer loads include the right questions for power-quality issues

3. Learn how to calculate P_ST flicker, distortion and sagging with correct modeling tools

4. Obtain the proper measurement equipment for before versus after measurements. If flicker were identified as the issue, the proper equipment would include an accurate P_ST flickermeter (that uses CIGRÉ protocol)

5. Be friendly yet hold firm to power-quality standards in contracts and tariffs.

These steps apply to all power-quality issues, including flicker. If a utility does not yet have the proper programs, policies, equipment and training in place to handle flicker as outlined in the five steps, taking just one step at a time is a good approach. Dedicating some resources to the process is a good place to start, and competent consultants can help fill in the gaps while a utility is in transition.

Regarding flicker specifically, the old GE curve can get a utility by with the simpler loads until the utility has a handle on the new method. But remember, for oddball flickering loads such as an arc furnace (Fig. 4), the new way is really the only game in town.

FLICKER-CONTROL BENEFITS

For many years, handling fluctuating loads of inconsistent periods, magnitudes or shapes required a lot of hand waving and storytelling. Thankfully, those days are now behind RMP. When a flicker issue comes up that pits a sensitive customer against a customer with a fluctuating load, RMP now has a technically defensible middle ground using IEEE 1453.

In fact, one of the benefits for the industrial customer with a fluctuating load is that the new way of looking at flicker is often less strict than the old way, because it gives allowance for softer transitions during the voltage fluctuations. In this way, it allows some loads to come onto the system that would not have been allowed before, and this is good business.

---

Dennis Hansen is a principal power-quality engineer at PacifiCorp. In his first 18 years as an engineer, he worked in protective relaying, research and development, and generation engineering. For the last 13 years, he has been working primarily in power quality and reliability. He serves as chair of the IEEE PES Voltage Quality working group. In 1998, he received PacifiCorp's Spirit of Excellence award for his role in mitigating the harmonics in the Salt Lake area ski resorts. Dennis.Hansen@PacifiCorp.com

CASE STUDIES

Case 1: Too little, too late

A hypothetical sawmill had been in production for many years in a rural area. As the nearby town grew, the rural area became only semi-rural and residents started to complain about light flicker. Because the sawmill had been in production for several years, its management said it didn't see a problem that needed fixing. Flicker measurements were not made with an accurate method at the outset and the load grew gradually; therefore, the problem was never caught at the design stage. More recent measurements by the utility clearly showed a flicker problem, but the Public Utility Commission wouldn't back the utility because of how long it took to discover the problem.

Case 2: A Success Story

This is how to work with a customer to resolve a flicker problem. RMP customer Energy Solutions asks if it can install a 6000-hp motor to shred low-level nuclear waste at a desert waste site. This is similar to a car-shredding operation, only bigger. RMP fills in the appropriate load sheets for large customers. The utility engineers determine that the power system is too weak, and the motor can only be installed with a static var compensator (SVC) to handle the flicker. RMP suggests some qualified SVC bidders to Energy Solutions. The waste-shredding unit is built with the SVC as an integral component. Testing shows that the PST flicker is no worse than before, though the area with only industrial customers is marginal in meeting flicker limits. This is considered acceptable by the utility. Everyone is pleased.