Should IEEE adopt IEC Flicker Standards
Voltage and Lamp Flicker Issues: Should the IEEE Adopt the IEC Approach?
Authors - Biographies at the end of the paper
S. Mark Halpin - Mississippi State University, Starkville, MS
Roger Bergeron - Hydro Quebec, Varennes (Quebec), Canada
Tom Blooming - Cooper Power Systems, Franksville, WI
Reuben F. Burch - Alabama Power Company, Birmingham, AL
Larry E. Conrad (Chair,TF on Light Flicker) - Cinergy Corp., Plainfield, IN
Thomas S. Key (Secretary, TF on Light Flicker) - EPRI Power Electronic Applications Center, Knoxville,
TN
Abstract
This provides an overview of the IEC flicker measurement and assessment procedures. The IEC procedures are
correlated with existing IEEE flicker standards to show the benefits of the IEC methodology. Application issues,
such as customer impact assessment, are also discussed. Three case studies are provided to show the correlation of
IEC flickermeter output values with documented customer complaints. The case studies also provide areas where
modifications to the IEC flicker standards may be necessary when developing a new IEEE flicker standard. The
Lamp Flicker Task Force is working to adopt and embrace IEC flicker standards as an IEEE recommended practice.
Introduction
Flicker is a difficult problem to quantify and to solve. The untimely combination of the following factors is required
for flicker to be a problem: 1) some deviation in voltage supplying lighting circuits and 2) a person being present to
view the possible change in light intensity due to the voltage deviation. The human factor significantly complicates
the issue and for this reason flicker has historically been deemed "a problem of perception." The voltage deviations
involved are often much less than the thresholds of susceptibility for electrical equipment, so major operating
problems are only experienced in rare cases. To office personnel, on the other hand, voltage deviations on the order
of a few tenths of one percent could produce extremely annoying fluctuations in the output of lights, especially if
the frequency of repetitive deviations is 5-15 Hz.
Due to the clear relationship between voltage deviation and light response, the term "flicker" often means different
things to different people with the interpretation primarily governed by the concerns of a particular discussion. In
each case, the deviation may or may not be strictly periodic and is usually expressed as a change (as indicated by the
change in rms value) relative to the steady-state level (expressed as an rms value averaged over some time period).
For voltage variations, the change is usually expressed as ∆V/V. A similar expression for light intensity variations
also exists.
From a utility application point of view, voltage fluctuations have usually been of interest, perhaps because voltage
changes are easily measured with existing instrumentation. Historically, these voltage changes have been used in
conjunction with "flicker curves" such as those shown in Figure 1 [1,2,3]. These curves, derived from controlled
experiments, offer thresholds of perception and/or irritability when periodic rectangular voltage fluctuations occur
continuously (only threshold of irritability curves are shown here). Even though the use of a simple curve leaves
much to be desired (including an accepted industry-wide definition of the essential ∆V/V term), it is comforting to
note that IEEE and UIE frequency weightings are very similar. The improvements that are now possible, based
primarily on existing IEC standards, are the subject of this paper and will be presented in later sections.
Página 1
Should IEEE adopt IEC Flicker Standards
Figure 1. Historical Flicker Curve
Standards-making bodies tend to focus on the changes in light intensity output in order to account for the human
observability factor. As standards have evolved, significant attempts have been made to include the years of
experience obtained using the "flicker curve" method described previously. There are, however, a number of degrees
of freedom that must be addressed in the development of a universally-accepted standard including lighting circuit
voltage, type of lamp involved, randomness of voltage fluctuation, and human factors which affect perception.
At this time, there are no widely-accepted flicker standards in the United States and Canada. In Europe and other
countries, however, the International Electrotechnical Commission (IEC) has developed a group of standards which
systematically account for many of the difficulties in the "flicker curve" methods. The IEEE Task Force on Light
Flicker is presently considering modifications to these IEC standards that are required for them to be considered for
adoption in the United States and Canada. The following sections describe the existing IEEE and IEC Standards.
IEEE Flicker Standards
The IEEE publishes voltage flicker limits in the form of recommended practice documents. The two most notable
are IEEE 519-1992 [4] and IEEE 141-1995 [3]. Although intended to be identical, they have very slight differences.
Both display the recommended practice on an xy graph as shown in Figure 1. The graph presents a borderline of
visibility and a borderline of irritation curve with each related to the continuity, the amplitude, and the frequency
aspects of the voltage fluctuations. Combinations of flicker frequency and magnitude below the borderline of
irritation are assumed to cause very few to no complaints.
The research behind the IEEE flicker curves is more than 50 years old. Researchers subjected people to a variety of
flicker magnitude and duration combinations from incandescent light bulbs. They used a variety of bulb wattages,
but 60 W bulbs dominate the research. The observers reported their feelings about each flicker dosage. They could
report that they did not see it, that they saw it but were not irritated, or that the flicker dosage was irritable to them.
The results have a statistical nature because observers do not always agree about visibility and irritation.
Researchers drew the visibility and irritation curves at "reasonable" levels. A flicker dosage just slightly below the
irritation limit might produce mild irritation from a very few observers. Increasing flicker dosage to slightly above
the irritation line will produce two results. First, a larger percentage of the population will be irritated. Second,
people irritated at the lower dosage will become more irritated. The percentage of irritable population and irritation
Página 2
Should IEEE adopt IEC Flicker Standards
level both increase with higher flicker dosage.
It is important to remember that most of the research assumed 60 W incandescent 120 V bulbs. Lower wattage
incandescent bulbs have shorter time constants. They tend to produce the first complaints especially for flicker
frequencies faster than one per second. Higher wattage bulbs have longer time constants and are less responsive.
New lighting technologies have yet different responses to voltage fluctuations. Recent test results using various
lamps are discussed in a later section.
It is also important to recognize that the laboratory experiments used sudden voltage changes. Rectangular
modulation of the 60 Hz sine wave was the most popular approach. Slower voltage changes, such as sinusoidal
modulation, are less likely to produce flicker complaints. It is also important to note that most of the experiments
used repetitive single point exposures. The experiments used steady voltages except for the single magnitude and
frequency combination. Most practical applications have combinations of several flicker dosages with multiple
amplitudes at various frequencies that possibly are produced by more than one source. Further, the voltage
modulation often is not rectangular and not periodic.
For example, consider the complexity of a typical residential circuit that also happens to serve a large seam welder.
The customer will experience flicker from motor starts and other load changes in his or her home. If they share a
common distribution transformer and secondary, they will also experience flicker from other customer loads on the
same transformer. They also see lamp flicker from the seam welder. The total flicker experience then is a
combination of these individual flicker dosages. The IEEE curves do not address multiple dosage issues.
The situation becomes even more complex if the utility adds an adaptive var compensator for flicker control. A seam
welder typically generates rectangular modulation. However, the single or half cycle var compensator controls add
significant complexity. The system sees the full voltage drop until the compensator responds. The voltage moves to
a new point which is unlikely to match the voltage prior to the weld. Voltage over-shoot is common at the end of the
weld due to compensator response time. This voltage modulation is far more complex than anything anticipated by
the IEEE flicker curves.
The IEEE flicker curves have served the industry well for many years. However, better techniques are available. A
flicker measurement protocol developed by the International Union for Electroheat (UIE) and embraced by the IEC
shows great promise. Cooperative efforts between the IEC, UIE, EPRI and IEEE allow the IEC standard to be
modified for a variety of lighting technologies and a variety of system voltages. This effort promotes one universal
standard for flicker.
It should be mentioned that there are many significant advances beyond the use of single curves such as those in
Figure 1. There are numerous major manufacturers in North America that offer flicker measurement products and
each is different. Furthermore, many of the utilities in North America have their own limits for flicker which are not
necessarily based on a formal measurement process. The large number of approaches presently in use is largely
responsible for this attempt by the Task Force on Light Flicker to standardize the measurement, evaluation, and
assessment procedures. The significant prior experience of the UIE and the IEC appears to offer the most promising
starting point for standardization in North America.
The IEC Flicker Measurement Standard
This section describes the IEC 1000-4-15 Standard (referred to in this paper as 4-15 for brevity) which gives the
specifics of a measurement approach for flicker that can be adapted to a wide variety of situations [5]. (Note that 415 has replaced the well-known IEC 868 flickermeter standard) The block diagram of the flickermeter specified in 415 is shown in Figure 2. The major portions of the flickermeter are 1) input processing, 2) "lamp-eye-brain"
response, and 3) output processing.
The input processing blocks are designed to established the "normal" voltage, V, and perform initial waveform
manipulations necessary to extract the frequency content of the voltage deviations from this "normal" value. The
primary steps here are 1) converting the rms value of the measured voltage to a reference level to insure that
percentage deviations are equal regardless of the input rms level and 2) squaring the input to ease the separation of
the low-frequency (0.5-25 Hz) variations from the power frequency components via filtering. This function is
referred to as a "squaring demodulator."
Página 3
Should IEEE adopt IEC Flicker Standards
referred to as a "squaring demodulator."
The "lamp-eye-brain" characteristic is obtained from a mathematical derivation of 1) the response of a lamp to a
supply voltage variation, 2) the perception ability of the human eye, and 3) the memory tendency of the human
Figure 2. IEC 1000-4-15 Flickermeter
brain. This section of the IEC flickermeter is where modifications can be made to fit particular needs. The transfer
function shown in (1) is provided as a reasonable model for the first two of these responses.
(1)
The coefficients in (1) are given by the IEC for 230 V, 60 W incandescent lamps. An amendment for 120 V, 60 W
incandescent lamps has been adopted by the UIE and is being circulated for consideration by the IEC. Recent
testing on other 120 V lamps (including magnetic and electronic ballast fluorescent and compact fluorescent lamps)
in the United States has resulted in appropriate lamp transfer functions for a very wide range of lamps. The response
characteristic in (1) can be modified to include different lamp characteristics as shown in (2) and used for virtually
any application.
(2)
The output processing of the flickermeter translates the output of block 4, called the instantaneous flicker sensation,
into the statistical indices Pst and Plt. The short-term flicker severity index, Pst, is a statistical quantification of the
instantaneous flicker sensation and is derived from a time-at-level analysis of the instantaneous flicker sensation. A
single Pst value is calculated every 10 minutes and Pst>1 corresponds to the level of irritability for 50% of the
persons subjected to the measured flicker. The long-term flicker severity index, Plt, is a combination of 12 Pst values.
Practical flicker limits are typically developed from 95 th and 99 th percentiles of a series of Pst and Plt values
collected over time periods perhaps as long as one week.
While much work remains to be done to properly model the flicker sensitivity of various lamps, the opportunities
suggested by (2) form the basis for the possible adoption of the IEC flickermeter specifications by the IEEE. The
Página 4
Should IEEE adopt IEC Flicker Standards
suggested by (2) form the basis for the possible adoption of the IEC flickermeter specifications by the IEEE. The
continuous-time methodology of the IEC flickermeter avoids the complications associated with existing IEEE
flicker curves. In addition to incorporating various lamps, the IEC flickermeter automatically incorporates the
effects of multiple flicker frequencies and non-standard (i.e. not square or sine wave) modulating waveforms. For
pre-evaluations of potential flicker-producing customers, the IEC provides shape factors that can be used to
translate many typical modulation waveforms into equivalent square or sine wave modulating waveforms so that
flicker estimates can be made based on pre-determined Pst=1 curves.
With all standards, updates are regularly requested. The 1986 IEC 868 standard was drafted for an analog
flickermeter designed during the 1970s. For the last 10 years, analog flickermeters have been gradually replaced by
digital versions which emulate each analog function. Very few North American digital flickermeters are available
and each new digital flickermeter that has been developed has been designed based on some interpretations of each
analog function described by the IEC. These interpretations are not obvious and justify a new digital specification
that could be addressed by IEEE.
The digital specification should not be considered trivial. Signal processing issues related to sampling rate,
quantization effects, and windowing (to list only three) are relevant factors which must be considered. At this time,
there are no uniform specifications covering these concerns.
Recent Lamp Test Results
In recent months, EPRI's Power Electronics Applications Center has performed flicker tests on several types of
modern lighting. It was observed during these tests that the lamp's amplifying characteristic, or gain factor, is the
important consideration for flicker due to voltage fluctuations. Additional tests show how interharmonics (noninteger harmonics) and phase-shifting harmonics on the power line can cause fluorescent lamps to flicker, despite
their having low gain factors when compared to incandescent lamps. Gain factor is defined and calculated by
measuring relative changes in light level while inducing controlled voltage fluctuations. By controlling the
magnitude and the frequency of voltage fluctuations, the lamp's flicker response can be determined using a
photometer to measure the lamp output. If the percentage of relative light fluctuation is greater than the percentage
of voltage fluctuation, the lamp is said to have an amplifying effect, or a gain factor greater than unity. Figure 3
shows an example of incandescent and fluorescent lamp gain factor.
Incandescent lamp gain drops off at higher frequencies because of the thermal inertia of the filament. Similarly,
when voltage fluctuations change gradually, as in a sine wave, rather than instantly, as in a rectangular wave, a
different flicker response is observed. For infrequent changes (1 Hz) a rectangular-shaped voltage change causes
about twice the level of flicker of a sine-wave shaped change of the same magnitude. As the frequency of
modulation increases, thermal inertia of the lamp filament begins to mask these shape differences. Furthermore, a
230-V lamp, typical in Europe, has a thinner filament than a 120-V lamp used in North America and therefore
allows illumination changes to occur more quickly, thus exhibiting more flicker.
Página 5
Should IEEE adopt IEC Flicker Standards
Figure 3. Gain Factor Variations for Different Lamps
In contrast, fluorescent lamps have very little thermal inertia and respond even faster. While the time constant for a
120 V incandescent lamp is about 28 ms and a 230V lamp about 19 ms, a typical fluorescent lamp has a time
constant of less than 5 ms. Consequently, fluorescent lamps are more sensitive to different voltage wave shapes,
harmonic-related phase shifts, and more rapid voltage fluctuations. Using gain factor to predict the flicker
performance of a variety of different type lamps has been effective in lab tests. More than 50 different lamps,
including incandescent, compact and 4-foot magnetic and electronic fluorescent, and high intensity discharge types
have been evaluated. A wide range of flicker performance is documented and these results are being used to
influence lamp manufacturers to design more flicker-free products. These test results are shown in Figure 4.
Lamp dimmers are also believed to play a role in the increased number of flicker-related complaints. The use of
incandescent dimmers in homes substantially increases lamp susceptibility to voltage changes. A typical electronic
dimmer will nearly double the change in light output for a typical voltage change compared to the same lamp with
no dimmer. Figure 5 shows test results for lamps with various percentages (0, 25, 50, and 75%) of dimming.
One relatively new source of flicker study is the unique voltage distorting characteristics of different types of arc
furnaces. Simple harmonic distortions of the voltage usually do not play a significant role in flicker
Página 6
Should IEEE adopt IEC Flicker Standards
Figure 4. Flicker Response Test Results
Figure 5. Dimmer Effects on Lamp Gain Factor
severity. Recently, however, special cases of harmonic distortion and harmonic phase shifting were found to be a
direct cause of lamp flicker in fluorescent lamps. To better understand how higher frequency harmonics can cause
low frequency flicker, tests were performed using a 5%, 185-Hz interharmonic component added to the
fundamental. This interharmonic alters the waveshape and effects the voltage peaks much more than the rms value,
which remains fairly constant on a cycle-by-cycle basis. The 185 Hz non-integer harmonic causes a cyclical "beat"
of 5 Hz and also cyclically changes the phase angle of the voltage peaks by a few degrees. It was found that while
incandescent lamps exhibited a slight amount of flicker, certain fluorescent lamps responded to the interharmonic
voltage by flickering at the "beat" frequency much more noticeably. Figure 6 shows test results of flicker caused by
interharmonic voltages.
Utilities have stated that when investigating a flicker complaint, it is very helpful to understand the flicker
Página 7
Should IEEE adopt IEC Flicker Standards
Utilities have stated that when investigating a flicker complaint, it is very helpful to understand the flicker
characteristics of different lamps. In some cases, it is possible to reduce the flicker level by simply re-lamping. This
can be demonstrated to utility customers complaining of flicker by providing sources of flicker-free lamps (i.e. gain
factors less than 1.0). The other solutions to flicker include installing mitigation equipment or properly addressing
service requirements to flicker-producing loads. A critical step in utilizing the IEC methodology is the acceptance
of the correlation between Pst and Plt values (being greater than 1.0) and the
Figure 6. Lamp response to a 5%, 185-Hz inter-harmonic voltage and a 5%, 3 rd harmonic voltage with ±90° phase
shift.
occurrence of customer complaints and the ability to use these concepts to develop useable limits. The application
methodology needed to base flicker limits on the IEC flicker measurement procedure is discussed in the following
section.
Pst and Plt: Application Methodology for the IEC Flickermeter
Values of Pst and Plt are directly available from the IEC flickermeter. As such, it is directly possible to define flicker
limits based on these values for equipment that is already in service. In many instances, however, it is necessary to
evaluate the flicker emissions of a potential customer before service is provided. Due to the wide variety of
equipment, operating voltages, and service designs (e.g. radial or mesh), the IEC has established three different
categories of limits for 1) low-voltage equipment with rated current less than 16 A, 2) low-voltage equipment with
rated current greater than 16 A, and 3) medium and high voltage equipment. Limits are given for both the statistical
parameters Pst and Plt as well as maximum rms voltage deviations. The following overviews of the relevant IEC
standards are intended to show how the use of the Pst and Plt concepts can help to overcome the problems that have
plagued North American utilities for years when trying to apply simple "flicker curves." The reader is strongly
cautioned to consult the actual standards before conducting any flicker assessment.
IEC Standard 1000-3-3
IEC Standard 1000-3-3 [6] gives limits and evaluation procedures for low-voltage equipment with current ratings
less than 16 A. Table 1 provides the different possible methods for evaluating Pst for limit compliance evaluations.
Página 8
Should IEEE adopt IEC Flicker Standards
less than 16 A. Table 1 provides the different possible methods for evaluating Pst for limit compliance evaluations.
As shown in Table 1, any voltage fluctuation can be assessed using the flickermeter measurement procedure. Direct
measurement is obviously most appropriate for
Table 1. Methods of Evaluating Pst (IEC 1000-3-3)
Types of voltage fluctuations
Methods of evaluating Pst
All fluctuations
Direct measurement
Voltage fluctuations with
known U(t)
Simulation; direct
measurement
Voltage fluctuation waveforms Analytical method;
corresponding to existing "shape simulation; direct
factor" curves
measurement
Rectangular voltage changes at
known frequency
Use Pst=1 "flicker curve"
loads already connected to a supply. If the rms voltage variation waveform, U(t), is known, computer simulation
(including a flickermeter simulation algorithm) can be used. If the waveform U(t) is not known, but the potential
load is known to produce rms voltage variations of a certain type (e.g. motor starting), then existing "shape factors"
defined in 1000-3-3 can be used to estimate Pst analytically. Only when the rms voltage variations are known to
resemble square waves can the traditional "flicker curve" approach be used to estimate Pst values. Using the curve
methodology, if a given voltage variation at a given frequency locate a point above the curve, the resulting Pst will
be greater than 1.0. Regardless of the evaluation approach used, each Pst value is to be determined over a ten minute
observation window. As defined by the IEC, one Plt value can be calculated based on N=12 successive Pst values
using (3).
(3)
Care should be taken in any evaluation other than direct measurement (under actual operating conditions) to match
the duty cycle of the equipment. IEC 1000-3-3 specifies limits of Pst£1 and Plt£0.65 for low-voltage equipment with a current
rating less than (or equal to) 16 A. Furthermore, this equipment shall not produce a maximum relative rms voltage fluctuation of more than 4%.
IEC Standard 1000-3-5
IEC Standard 1000-3-5 [7] gives limits and evaluation procedures for low-voltage equipment with current ratings
greater than 16 A. The limits in 1000-3-5 are those given in 1000-3-3. It is recognized, however, that a lower supply
impedance will be required to meet these requirements for larger equipment. In addition, 1000-3-5 recognizes
equipment that may produce voltage fluctuations at a rate of less than one per hour. In these cases, the limits of Pst
and Plt are not applicable. The maximum rms voltage deviation is limited to 1.33 times the 4% limit of 1000-3-3.
IEC 1000-3-5 specially recognizes that low-voltage equipment with a current rating greater than 75 A should be
evaluated based on the actual supply impedance at the connection point. Pst can then be estimated based on the
relative size of the load VA and the supply transformer VA rating. The Plt limit is set equal to 0.65Pst for equipment
with current ratings >75A.
It should be noted that IEC 1000-3-3 and 1000-3-5 are considered to be "equipment standards" by which
manufacturers of low-voltage equipment can design their products. However, the limits and evaluation procedures
apply equally well to both the "design" phase and the "operational" phase of low-voltage equipment. The limits set
forth serve to limit fluctuations below thresholds of irritability for users connected to the same supply circuit
regardless of whether or not equipment was designed according to these standards. For this reason, the limits of
1000-3-3 and 1000-3-5 can be applied in North America where limits are presently not placed on equipment
Página 9
Should IEEE adopt IEC Flicker Standards
1000-3-3 and 1000-3-5 can be applied in North America where limits are presently not placed on equipment
manufacturers. For uses other than equipment design, however, existing levels of background fluctuations should be
taken into account.
IEC Standard 1000-3-7
IEC Standard 1000-3-7 [8] gives limits and evaluation procedures for equipment connected to medium and high
voltage supply systems. Note that the IEC defines medium voltage (MV) as 1kV<MV£35kV and high voltage as
35kV<MV£230kV; Extra-high voltage (EHV) is considered to be >230 kV. Specific limits are not given that must be
followed; 1000-3-7 recognizes that the limit values for Pst and Plt will vary between utilities depending on the
specifics of the loads served and the supply network. Indicative planning levels, which are the quality targets of a
supplying utility, are given in Table 2.
Table 2. Planning Levels for MV, HV, and EHV Systems
Planning Levels
MV
HV-EHV
Pst
0.9
0.8
Plt
0.7
0.6
These levels are evaluated on a statistical basis. As a general guideline, Pst and Plt should not exceed the planning
levels more than 1% of the time, with a minimum assessment period of 1 week. IEC Standard 1000-3-7
distinguishes between Pst and Plt values measured throughout a supply system and those associated with a particular
fluctuating load. Planning levels (usually denoted as LPst and LPlt) apply throughout a supply system; the aggregate
effects of all fluctuating loads must be taken into account. Emission limits for individual loads (denoted as EPst and
EPlt) must be set so that the combined effects do not exceed planning levels.
IEC Standard 1000-3-7 presents a three-step procedure for evaluating fluctuating loads. The first step is an
"automatic acceptance" procedure that can be applied to assess the impact of a potential customer without detailed
analysis. Table 3 shows the criteria for MV connections which specify the maximum allowable ratio of load power
variation, ∆S, to the available short circuit power, SSC, as a function of the fluctuation rate. Fluctuating loads
connected directly to a HV supply can be accepted without further study provided the ratio Smax/SSC<0.1% where
Smax is the maximum load power.
Table 3. Maximum permissible load fluctuations ∆S/SSC for automatic acceptance of MV loads
r (# of variations/minute)
∆S/SSC (%)
r>200
0.1
10£r£200
0.2
r<10
0.4
Assuming that a particular fluctuating load does not meet the criteria of Table 3, Pst and Plt limits are allocated to
each individual load in a MV system based on each individual load's portion of the total load on the MV system as
shown in (4) for Pst with the quantity GPstMV defined in (5). The factor FMV is used to account for the nonsimultaneous nature of all flicker contributions and will always be less than or equal to 1.0 (more typical values for
non-simultaneous fluctuations are 0.2-0.3). Similar relations apply for Plt.
Página 10
Should IEEE adopt IEC Flicker Standards
(4)
(5)
The following definitions apply to (5):
GPstMV=Maximum global contribution of local loads to the flicker level in the MV system,
LPstMV=Planning level of the flicker level in the MV system,
LPstHV=Planning level of the flicker level in the supplying (upstream) HV system, and
TPstHM=flicker transfer coefficient from the upstream HV system to the MV system.
The use of a single limit value for the entire MV system under consideration and then allocating contributions to
this total limit among various disturbing loads is a method which allows larger loads (which generate greater
revenue) a greater share of the system's ability to absorb the effects of fluctuating loads. To avoid overly restricting
smaller loads, basic levels of EPst,i=0.35 and EPlt,i=0.25 may be allowed for all loads.
For MV systems with loads that do not individually meet the limits of (4) for EPst and EPlt, a more thorough analysis
is required. In this stage 3 analysis, all available information about present and expected future system and load
conditions should be considered. Any concessions made for an individual load to exceed emission levels
determined as in (4) should not result in the violation of GPstMV and GPltMV limits.
Stage 2 evaluations for loads connected to HV transmission nodes is similar to (4). Equation (6) illustrates the stage
2 evaluation for Pst limits for an individual (the ith) load served at high voltage. StHV represents the total load
supplied from the HV point serving the ith load. In many cases, the total load served at a given point can be difficult
to determine. IEC Standard 1000-3-7 considers various facets of this problem and proposes alternative methods for
determining StHV. The important point is that emission limits for individual loads are provided in terms of the
system's total ability to absorb fluctuations; multiple fluctuating loads are automatically taken into account.
(6)
Stage 3 evaluations for HV loads follow the same logic as that proposed for MV loads. In general, detailed studies
incorporating large amounts of system and load data are needed to insure that HV planning levels for LPst and LPlt
are not exceeded.
Case Studies Using the IEC Flickermeter
Three cases studies covering different flicker-producing loads are presented in this section. Where appropriate,
flicker criteria considerations that are appropriate for power contracts are discussed.
Resistance Welding Machines, Case 1
Resistance welding sometimes causes flicker problems for the steel fabrication industry. This is especially true when
one transformer feeds many welders that have random and independent operation. Occasionally too many welders
fire at the same time. The voltage drop at that moment causes numerous cold welds. Large (single) automated
welding machines also sometimes cause flicker problems for electric utilities. This is especially true for welding
machines that repeat the welding one time per second or faster. The frequent repetitive voltage drop from the welder
Página 11
Should IEEE adopt IEC Flicker Standards
machines that repeat the welding one time per second or faster. The frequent repetitive voltage drop from the welder
can cause a very noticeable flicker in lighting.
A good example of the continuous welding machine flicker problem is a steel fabrication plant in Indiana. The
plant has enjoyed success and growth into new product lines. Calculations for the newest product line showed
uncompensated flicker levels would exceed the borderline of irritation on IEEE Standard 141-1993. Cost estimates
clearly favored adaptive var compensation compared to lowering the system impedance. Unfortunately, the one
cycle response of the compensator has a bit of a doubling effect on the frequency of the voltage fluctuations. It was
impossible to use IEEE 141 to account for the complex voltage fluctuations produced by the welder/compensator
combination.
The welders and compensator were served from a distribution circuit with residential customers. There are
approximately 1,600 customers who experience the point of common coupling voltage fluctuation magnitudes.
About 25 of those customers have complained with some making repeated complaints. The flicker caused some
customers to worry about the wiring in their homes. One customer worried their perception of flickering lights was
an early warning sign for a seizure or heart attack. Some customers seemed satisfied to know the utility was aware of
and working on the situation. Some took comfort in knowing their house wiring was not the problem.
The first test of the proposed IEC flickermeter showed excellent results. As discussed previously, a principal output
of the flickermeter is short term flicker severity level Pst. The recommended maximum flicker level before complaints
is about Pst =1.0. The flickermeter output ranged from Pst =1.0 to 1.2 during the situation that generated customer
complaints. The number of customer complaints confirmed the "borderline of irritation" interpretation of Pst=1.0.
Motor Starting and Load Torque Variations
Figure 7 shows a plot of rms voltage data collected on a rural distribution feeder in the southeastern U.S. The feeder
supplied several hundred residential customers and a process industry. The system supplying the distribution
substation was relatively weak (35 MVA short circuit capacity at the 12kV distribution bus) and the industrial load
with four large (2-350 and 2-500 hp) induction motors was located at the end of the feeder. Frequent motor starts
and variations in load torque (up to breakdown torque) characterize the plant load behavior.
These measurements were collected in response to numerous complaints to the local utility offices and the Public
Service Commission. While an IEC flickermeter was not available for use at this site, the entire feeder and motor
loads were simulated in EMTP with the IEC flickermeter modeled in TACS. The UIE 120 V, 60 W incandescent
bulb characteristics were used in the TACS flickermeter implementation.
Figure 7. RMS Voltages (120 V base) Collected on 12 kV Feeder Near Residential Service Points
Figure 8(a) shows the simulated feeder voltage (on 120 V base) for motor start and no-load acceleration followed by
a torque (step) increase from no load to slightly less than breakdown torque. It is clear that the simulated feeder
voltage deviations are within the ranges in Figure 7. Figure 8(b) shows the corresponding IEC flickermeter Block 4
output (instantaneous flicker sensation). (The long simulation times required eliminate the practicality of obtaining
Pst and Plt values.) These results clearly indicate values of instantaneous flicker sensation significantly greater than
Página 12
Should IEEE adopt IEC Flicker Standards
1.0. If statistically analyzed over several repetitions, these high levels of instantaneous flicker sensation would most
likely indicate Pst>1. Of course, Pst>1 supports the large number of customer complaints.
(a) 12 kV Feeder Voltage (120 V base)
(b) IEC Flickermeter Block 4 Output Value
Figure 8. EMTP Simulation of Distribution Feeder with Fluctuating Motor Load and IEC Flickermeter
Several features of the flickermeter output of Figure 8(b) are worth noting. The different "peaks" in the output
correspond to different voltage variations where the "sharper" and "smoother" variations correspond to higher and
lower meter outputs, respectively. It would be difficult to apply conventional flicker curves such as those in IEEE
519 and 141 due to the aperiodic and non-standard nature (i.e. not rectangular fluctuations) of the variations; the
IEC shape factors could be applied easily to predict these results before the customer was initially connected.
Resistance Welding Machines, Case 2
In Evansville, Indiana, a manufacturer of condenser tubes for residential refrigerators purchased a three-phase, 1500kVA resistive spot welder to fasten the condensers to steel wire used for heat dissipation and structural support. To
fill a backlog of condenser orders from a major refrigerator manufacturer, the condenser manufacturer installed the
welder and operated it at full capacity, 24 hours a day.
During the first 13 days of operating the new welder, 107 different residential customers of the local electric utility
complained about flickering lamps, with as many as 25 calls in one day. The complaints compelled the utility and
the user to determine ways to remedy the problem without shutting down the welding operation.
Within a few days investigators began looking for a solution. The approach included taking measurements with a
UIE flickermeter configured for 120-V lamps and recording the power system configuration. At the same time
factory personnel experimented with welder set up.
The flickermeter, shown in Figure 9, used during the investigation was a portable ensemble of interface module,
laptop computer, and cables, all small enough to fit in a briefcase. The computer software for the flickermeter
translates measurements of the voltage fluctuations into a number, in units of Pst, indicating level of light flickering.
The flickermeter automatically takes measurements in ten-minute intervals. If the Pst measured is above 1.0, then
Página 13
Should IEEE adopt IEC Flicker Standards
The flickermeter automatically takes measurements in ten-minute intervals. If the Pst measured is above 1.0, then
utility customers connected to the same feeder will likely notice flicker in incandescent lamps (the threshold is
higher for fluorescent lighting as indicated in a previous section). Inside the facility, flicker was observed in
fluorescent lamps, indicating serious voltage fluctuations.
Figure 9. UIE/IEC Flickermeter Installed at the Substation Oil Circuit Breaker Serving the Factory
A single-line diagram of the utility and user systems is shown in Figure 10. With the flickermeter at the factory
service entrance (a few hundred electrical feet from the welder transformer, point A in Figure 10) voltage
fluctuations were measured at 2 to 2.4 Pst. The plant power consumption (kVA), voltage distortion (THDV), and
current distortion (THDI) were also monitored and recorded at regular intervals using a three-phase power monitor.
On the same day, the flickermeter was moved to the utility 69-kV substation about two miles away. There, the meter
was installed at a 12.47-kV oil circuit breaker on the feeder serving the condenser factory (point B in Figure 10).
While monitoring the power system voltage, different welder electrical and mechanical configurations were tried.
To determine whether the older welder significantly contributed to the voltage fluctuations, its operating schedule
was also recorded and compared to the flickermeter records. In this way, sufficient data was obtained to guide
electrical system and welder changes that mitigated the fluctuations.
Power system tests included varying the 1200-kVAR reactive compensator at the substation (on or off), the 900kVAR reactive compensator at the facility (on or off), and the bus tie at the substation (open or closed). Changes in
reactive compensation and the bus tie did not significantly affect the voltage fluctuations at the substation or
service entrance. Closing the bus tie shown in Figure 10 slightly lowered the Pst at the factory service entrance from
2.4 to 2.0. It also led to more flicker complaints from newly affected customers now tied to the distribution feeder
that served the welder.
Página 14
Should IEEE adopt IEC Flicker Standards
Figure 10. Single-Line Diagram Showing the Utility System, Factory Service, and Monitoring Points.
Meanwhile at the welder, operation was maintained at a high throughput with hope that production levels could be
maintained without producing flicker. Investigators suspected that simultaneous firing of electrode pairs was the
main cause of the light flicker, and changed the firing sequence from two at a time to one at a time and decreased
the weld time from two to one cycle per weld. When the welder was reconfigured to fire electrode pairs individually
for a duration of one cycle, the recorded Pst at the service entrance indeed dropped significantly, from >2.0 to about
0.8. See the recorded flickermeter output in Figure 11.
Next, the cam speed and weld heat were varied to determine the maximum throughput of high-quality welds. After
each change in welder configuration, investigators tested the weld strength of random condenser samples by pulling
on structural wires using a weld test jig. In this way, the investigators determined that a maximum cam speed of 107
RPM and a weld heat of 83 percent achieved the greatest throughput of high-quality welds.
Figure 11. Recorded Flickermeter Output (Pst)
The condenser manufacturer and the utility agreed that changing the firing sequence of the electrode pairs from two
at a time to one at a time and the cam speed from 60 RPM to 107 RPM was the most cost-effective option for
Página 15
Should IEEE adopt IEC Flicker Standards
at a time to one at a time and the cam speed from 60 RPM to 107 RPM was the most cost-effective option for
reducing flicker and maintaining a high production level. By doing so, the condenser manufacturer reduced voltage
fluctuations at the substation bus to a level that did not create observable flicker. Reconfiguring the welder also
benefited the condenser manufacturer by decreasing the energy consumption per weld and increasing the welder
throughput from a speed of 60 RPM to 107 RPM. After the welder was reconfigured, complaints about flicker in the
area ceased, as predicted by the IEC flickermeter Pst values falling below 1.0. Except for the cost of the
investigation, reconfiguring the welder was a cost-free option that enabled the condenser factory to maintain a high
production level and meet the backlog of orders for refrigerator condensers.
Conclusions
There are many practical situations where existing "flicker curve" methodology can not be applied in a consistent
manner. The IEC has moved toward a standardized measurement technique and has developed flicker and voltage
fluctuation limits based on this technique. Significant experience with the measurement technique has validated
the approach for European power systems and the necessary modifications required to adapt the measurement
procedure and the limits to North American power systems are now available. The case studies presented in this
paper demonstrate the correlation between customer complaints and flickermeter output which indicate that flicker
and voltage fluctuation limits based on the IEC methods can be effectively used in North America.
References
[1] R.C. Seebald, J.F. Buch, and D.J. Ward, "Flicker Limitations of Electric Utilities," IEEE Transactions on Power
Apparatus and Systems, Vol. PAS-104, No. 9, pp. 2627-2631, September 1985.
[2] M. Sakulin and T.S. Key, UIE/IEC Flicker Standard for Use in North America: Measuring Techniques and
Practical Applications," Proceedings of PQA'97, March 1997, Columbus OH.
[3] IEEE Standard 141-1993: Recommended Practice for Power Distribution in Industrial Plants, IEEE, 1993.
[4] IEEE Standard 519-1992: Recommended Practices and Requirements for Harmonic Control in Electrical
Power Systems, IEEE, 1993.
[5] IEC Publication 868, "Flickermeter: Functional and Design Specifications," CEI, 1986.
[6] IEC Standard 1000-3-3, "Limitation of Voltage Fluctuations and Flicker in Low-Voltage Supply Systems for
Equipment with Rated Current £ 16 A," CEI, 1994.
[7] IEC Standard 1000-3-5, " Limitation of Voltage Fluctuation and Flicker in Low-Voltage Power Supply Systems
with Rated Current > 16 A," Cei, 1994.
[8] IEC Standard 1000-3-7, "Limitation of Voltage Fluctuation and Flicker for Equipments Connected to Medium
and High Voltage Power Supply Systems," CEI, 1995.
Authors' Biographies
S. Mark Halpin (M 93) received his BEE, MS, and PhD degrees from Auburn University in 1988, 1989, and 1993,
respectively. He is currently an Assistant Professor in the Department of Electrical and Computer Engineering at
Mississippi State University. His teaching interests include power systems, control systems, and network analysis.
His research interests are in the areas of modeling and simulation techniques for large-scale power systems, power
system transients and harmonics, and computer algorithms. He is active in the IEEE Power Engineering Society and
Industry Applications Society, where he serves as chairman of the IEEE-IAS Working Group on Harmonics.
Roger Bergeron (Eng M'74) After graduating with a B.SC. Degree in 1974 in Electrical Engineering from Sherbrooke
University, he spent the next five years building up his knowledge and acquiring experience in private corporations
such as Québec Iron and Titanium. In 1980, he joined Hydro-Québec where he contributed to many research
programs involving worker safety and service quality. He is the taskforce convener of IEEE P1159.1 on "Recorder
Página 16
Should IEEE adopt IEC Flicker Standards
programs involving worker safety and service quality. He is the taskforce convener of IEEE P1159.1 on "Recorder
Qualification and Data Acquisition Requirements for Characterization of PQ Events", convener for CSA 311.4
committee on the Canadian standard regarding Electromagnetic Compatibility of Low Voltage and Low Frequency,
IEC77A WG1 TF2 and IEC 77A WG9 member on drafting the IEC 1000-4-7 and IEC 1000-4-30, and UIE WG2
secretary which is the UIE committee on power quality.
T. M. Blooming (S '89, M '94) received a B.S. degree in electrical engineering from Marquette University in 1992, an
M.E. degree in electric power engineering from Rensselaer Polytechnic Institute in 1994, and an M.B.A. degree from
Keller Graduate School of Management in 1998. He is a Power Systems Engineer with the Systems Engineering
Group of Cooper Power Systems, Franksville, WI. His primary responsibility is to perform analytical studies on
industrial power systems. This involves on-site harmonic, flicker, and transient measurements. It also includes
system modeling by applying computer programs to perform harmonic, power flow, stability, short circuit, and
protective device coordination analysis. He is an instructor in the Cooper Power Systems Power Quality and
Harmonics, Overcurrent Protection, and Transformer Application and Protection Workshops. He is active in the
IEEE voltage flicker task force.
Reuben F. Burch, IV (M 70, SM 97) was born in Eastman, GA, on August 5, 1948. He received his BEE degree from
Auburn University in 1970. He is a Principal Engineer in Enhanced Power Quality at Alabama Power Co. in
Birmingham, AL where he mainly performs voltage flicker and harmonic analyses. He is a member of IEEE, PES, and
IAS and is a registered professional engineer in Alabama and Georgia.
Larry Conrad (M'74-SM'91) earned his B.S.E.E. in 1974 and his M.S.E.E degree in 1993 from Rose-Hulman
Institute of Technology. He has been employed at Cinergy since 1974 as Engineer, Project Engineer, Senior
Engineer, and Technical Services and Power Quality Manager. He now is Manager of Operations Engineering and
responsible for power quality, the POWER CLINIC(R) power quality services, reliability, and other duties. He has
worked with voltage sag issues since 1985. He is generally credited for inventing voltage sag predictive techniques
and equipment coordination methods for IEEE Std. 493 and P1346. Mr. Conrad is a registered Professional
Engineer in the states of Indiana and Ohio. He is a member of the IEEE Industry Applications Society and the Power
Engineering Society. Mr. Conrad is an IEEE representative to ANSI C84 and an advisor to the US National
Committee for IEC SC77A and a member of WG2. Mr. Conrad is Chairman of IEEE voltage flicker project P1453
and Chairman of the IEEE Std 493, Gold Book, Voltage Sag Working group. He is also a member of the IEEE P1346,
ASD and PLC compatibility, Working Group. He also is responsible for the voltage considerations chapter in the
next revision to IEEE Std. 141, Red Book.
Thomas S. Key was born in Indianapolis, Indiana, in 1947. He received a BSEE from the University of New Mexico
in 1970 and an ME in power engineering from Rensselaer Ploytechnic Institute in 1974. He is currently the
Technical Director at EPRI's Power Electronics Applications Center in Knoxville, Tennessee, and is responsible for
power quality related research, development, and testing. During the ten years before joining PEAC in 1989, he
managed electrical power system design and power conditioning system applications for renewable sources of
energy at Sandia National Laboratory in Albuquerque, New Mexico. In 1978, while in the Navy, he collaborated
with the Computer Business Equipment Manufacturers Association to create the well-known "CBEMA curve,"
which is the industry standard voltage tolerance envelope for information technology equipment. He is a senior
member of the IEEE and is active in the development of standards in the areas of power quality, photovoltaics, and
power system design, and 1996 recipient of the IEEE Outstanding Engineer in Region 3.
Página 17