Meters must be designed to represent the human factors involved in the perception and toleration of flicker so that the probability of customer complaints can be estimated
Flicker is a phenomenon that is very difficult to characterize due to the human factors involved. However, in order to study flicker effects, it was necessary to develop flicker units and magnitudes that would allow its measurement. When voltage fluctuations exist on a power system, the problem is usually not so much to measure the size of the fluctuations themselves as to determine the effect on various types of lamps and to estimate the probability that they will cause complaints of light flicker. For a given amplitude of flicker, the opinions of many observers differ widely, and the limit (or tolerable) level on a system must be one that only very few consumers find intolerable. Calibration of any particular meter in terms of consumer complaints is, therefore, likely to be difficult. The flicker meter should be designed to represent the physiological and psychological processes involved in the perception and toleration of flicker [1].
Flicker Meters
The measuring devices previously developed were designed to detect voltage fluctuations and their processing in order to indicate the impression of visual observer. In this way, the devices considered the limited visual repetition sensitivity and the effect of thermal time-constant of incandescent lamps. Also, the devices had built-in image fusion that was carried out by the eye/brain system when the illumination was rapidly changing (approximately at 30 Hz). Several devices have been developed (mainly in England, France, Germany, and Japan); all detect voltage fluctuations in the range 0.5-30 Hz and estimate a weighted average of such fluctuations, which provides an assessment of the dosage of flicker that would be annoying. The concept of flicker quantity was usually defined as the integrated flicker over a period of time, and it uses the percentile probability that a voltage fluctuation limit has been exceeded [2].
Maximum sensitivity is presented between frequencies of 6 and 10 Hz, as the range of interest is from 1 to 25 Hz. Isolated and frequent voltage changes are also annoying, as the brain remembers the previous perturbation and immediately relates it with the new one. In this case, the variation in magnitude and time interval between two successive perturbations should be considered. Due to the different weightings given to the voltage fluctuation component as well as the different interpretations used to define the flicker dose, the previously used flicker meters did not give consistent readings on common data.
The International Union for the Application of Electricity (UIE), which was formerly known as the International Union for Electroheat, adopted a standard assessment methodology that is based on the previously developed flicker meters proposed in 1979, taking into account the following facts:
•Specification of the design and function characteristics of the measurement device (flicker meter)
•Specification of the statistical assessment of the flicker phenomenon
•Estimation of the short- and long-term flicker severity.
Figure 1. Functional block diagram of the UIE flicker meter
To the authors' knowledge, there is no American standard related to flicker meter. The International Electrotechnical Commission (IEC) standardized a flicker meter that incorporates weighting curves that represent the response of the human eye to light variations produced in a 60 W, 230 V, 50 Hz, double-coiled filament incandescent lamp. The output of the meter is given as per-unit flicker voltage, where one per unit is the level that should cause noticeable and annoying light flicker, with the perception threshold for 50% of the human population. Flicker is defined in terms of incandescent lamps because of their common usage and sensitivity to voltage changes. Flicker is also observed with fluorescent lamps.
The European flicker measurement standards are IEC Standard 868 (initially presented in 1986), IEC Standard 868 Amendment 1 (1990), and IEC Standard 61000-4-15 (1997), which are under continuous revision in order to include the necessary changes for their application to 120 V, 60 Hz lamps, for instance [3], [4]. The instrument is a specialized amplitude-modulation (AM) analyzer in which the carrier frequency is the main frequency (50 or 60 Hz), having post-detection band-pass filtering to emulate the response characteristic ofthe lamp-eye-brain system [5].
EC Standard 868 was drafted for an analog flicker meter designed during the 1970s. For the last 15 years, analog flicker meters have been replaced gradually by digital versions that emulate each analog function. The major parts of the flicker meter are the input, the lamp-eye-brain response, and the output processing.
Flicker Meter Implementation
Figure 1 shows a block diagram for the complete flicker meter instrument described in the IEC standards. The main characteristics of each of the five blocks are described as follows.
There is an input transformer before block 1; its function is the insulation and adaptation of the instrument input circuit to the level of the measured signal, allowing nominal input voltages from 55 to 415 V at line frequency.
Input Voltage Adapter
The primary function of the input voltage adapter (block 1) is to provide a normalized rms voltage to the input of the demodulator (block 2). An automatic gain control circuit with a 10 to 90% step-response characteristic of 1 minute provides the necessary functionality. Besides, it possesses two filters to eliminate dc components and double-frequency ripple. This circuit emulates a well-known characteristic of the human perception for which moderate-level, constant stimuli to the senses gradually become imperceptible. Block 1 includes a calibration generator.Demodulator
Block 2 specifies the use of a squaring multiplier as a demodulator. The purpose of this block is to recover modulating signals and simultaneously suppress the main-frequency carrier signal via filtering, as these signals are the only desired output.
Weighting Filters
Block 3 includes three filters connected in series and a ranging circuit. The first filter is a first-order high-pass filter with the cut-off frequency set to 0.05 Hz. The second is a sixth-order Butterworth low-pass filter with a corner frequency of 35 Hz. These filters remove the dc component and the 100 Hz doubled carrier, with its associated sidebands, from the signal output by block 2. The third filter gives a band-pass response centered at 8.8 Hz, providing a very specific weighting function within the frequency band of interest (0.05 to 35 Hz), simulating the response of the lamp-eye-brain system for the average observer.
The lamp-eye-brain characteristic is obtained from a mathematical derivation of:
• Response of a lamp to a supply voltage variation
• Perception ability of the human eye
• Memory tendency of the human brain.
The filters are precisely specified by means of the required transfer function in the complex frequency domain. The ranging function that is required for instruments using certain types of statistical classifiers resides inside block 3. Full-scale ranges corresponding to voltage change from levels of 0.5 to 20% are defined with the requirement of minimum resolution.
Squaring Multiplier and First-Order Sliding Mean Filter
Block 4 implements the remainder of the lamp-eye-brain model. The squaring operator simulates nonlinear eye/brain response characteristics, while the first-order filter emulates perceptual storage effects in the brain with the time constant of 300 ms. When the instrument gain is properly set, modulation levels corresponding to the mean human threshold for flicker sensation will generate values of 1 at the output of this block. The output of block 4 is called instantaneous flicker sensation, denoted by Pf5.
Statistical Classifier
Block 5 emulates human irritability due to flicker stimulation; it is a sampling A/D converter followed by a statistical classifier. This classifier translates the output of the previous block into short-term flicker severity index (Pst) and long-term flicker severity index (Plt). Pst is a statistical quantification of the instantaneous flicker sensation and is derived from a time-at-level analysis of the Pf5. It consists of a weighted sum of percentiles of the cumulative probability distribution of the flicker sensation, with the purpose of providing objective information on the flicker severity level independently of the type of voltage fluctuation, time variation law, and evolution.
While much work remains to be done in order to properly model the flicker sensitivity of various lamps, it is very probable that the IEEE will adopt the IEC 868 flickermeter specifications.
In February 1998, the IEEE P1453 Flicker Task Force voted unanimously to embrace and enhance the IEC flicker meter mea¬surement protocol for the IEEE Recommended Practice [6], [7].
Regulations or Flicker Limits
IEEE Approach
Flicker curves were unified and included in IEEE Standard 141-1993 and IEEE Standard 519-1992. The subject is addressed briefly in the IEEE standards. In IEEE Standard 141, both the borderline of irritation and borderline of visibility of flicker are given, which were extracted from a handbook published in 1960. On the other hand, IEEE Standard 519 presented similar information taken from different sources, offered just as a guide for planning. The curves from the two standards are not precisely equivalent; the general tendency is similar, but IEEE Standard 519 borderlines are more demanding than IEEE Standard 141 in voltage fluctuation for maximum sensitivity frequency (at approximately 8.8 Hz).
The IEEE flicker curves have served the industry well for many years, but the status of electric power systems is becoming much more complex than anticipated due to the presence of new types of lamps and also to new phenomena, such as multiple frequency dosage, flicker modulation, inter-harmonics, and subharmonics. This means that IEEE flicker curve methodology urgently needs to be updated, taking into account the fact that new powerful techniques are now available. Cooperative efforts between the IEC, UIE, Electric Power Research Institute (EPRI), and IEEE allow the IEC standard to be modified for a variety of lighting technologies and system voltages. This effort promoted one universal standard for voltage flicker [8].
It should be indicated that IEC and EN standards are of widespread application in Europe, being adopted by most of the European countries, and also by many countries outside of Europe. Values of Pst and Plt are directly available from the IEC flicker meter, and it is then possible to define flicker limits based on these values for equipment that are already in service. In many instances, however, it is necessary to evaluate the flicker emissions of a potential customer before service is provided.
IEC Approach
Due to the wide variety of equipment, operating voltages, and service designs, the IEC has established three different categories of limits for:
• Low-voltage equipment with rated current less than 16 A (IEC Standard 61000-3-3)
• Low-voltage equipment with rated current greater than 16 A (IEC Standard 61000-3-5)
• Medium and high voltage equipment (IEC Standard 61000-3-7).
Limits are given for both of the statistical parameters (Pst and Plt) as well as maximum rms voltage deviations. The following explanation of the relevant IEC standards are intended to show how to use the Pst and Plt concepts in comparison with the flicker curves methodology.
IEC Standard 61000-3-3. This standard provides limits and evaluation procedures for low-voltage equipment with current ratings less than 16 A. The individual emission limits (Pst =1, Plt = 0.65) should be measured on the supply point through reference impedance values. The defined impedance values correspond to the 90% system impedance on single-phase European low-voltage systems. Furthermore, this equipment should not produce a maximum voltage fluctuation of more than 4% [5].
The different possible methods for evaluating Pst for limit compliance evaluations are:
• Direct measurement using the flicker meter measurement procedure, which is most appropriate for loads already connected to the supply
• Computer simulation, if the rms voltage variation waveform U(t) is known
• Defined "shape factors" to estimate Pst analytically, if the waveform U(t) is not known but the potential load is known to produce rms voltage variations of a certain type.
IEC Standard 61000-3-5. This standard provides limits and evaluation procedures for low-voltage equipment with current ratings greater than 16 A. The limits are those given in IEC Standard 61000-3-3. It is recognized, however, that lower supply impedance will be needed to meet these requirements for larger equipment. 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 be estimated then based on the relative size of the load and the supply transformer rating. It should be noted that IEC Standard 61000-3-3 and IEC Standard 61000-3-5 are equipment standards by which manufacturers of low-voltage equipment can design their products [8].
IEC Standard 61000-3-7. This standard gives limits and evaluation procedures for equipment connected to medium voltage (1 to 35 kV) and high voltage (35 to 230 kV) supply systems. Specific limits that must be followed are not given, recognizing that the limit values for Pst and Plt will vary among utilities depending on the specifics of the loads served and the supply network characteristics.
Indicative planning levels, which are the quality targets of a supplying utility, are given in table format. Planning levels apply throughout a supply system; the aggregate effects of all fluctuating loads must be taken into account. Emission limits for individual loads must be set so that the combined effects do not exceed the planning levels.
Affected Devices
The main effect of voltage fluctuations is on illumination (precisely on electric lamps) resulting in light flicker, which has been a cause of engineering concern since the start of the electric power industry. The light flicker problem is now a very hot issue due to the current increasing voltage distortion in medium and low voltage systems. Near rated voltage, the percentage variation of light output from incandescent filament lamps is on average 3.8 times the percentage voltage change causing it, which varies only between 4.1 and 3.4, for lamps rated between 15 and 1,500 W. Illumination relative fluctuations (RI), measured by a photometer adapted to the human eye response, have been introduced. The relation between the percentage of light variation and the percentage of voltage change is usually called gain factor (or lamp amplifying characteristic).
The corresponding value for fluorescent lamps is of the order of unity. Consequently, filament lamps are inherently more sensitive to voltage fluctuations than fluorescent lamps. The percentage variation for high-pressure mercury vapor lamps and sodium vapor lamps is from 2.8 to 3.3, and it is 0.5 for low-pressure sodium lamps [1].
Incandescent Lamps
Incandescent Lamp with Sinusoidal Voltage. The heat transfer behavior of an incandescent lamp can be studied using the analog circuit shown in Figure 2.
Figure 2. Incandescent lamp thermal equivalent circuit
The filament is represented by the thermal resistance Rt in parallel with the thermal capacity Ct. The filament temperature rise T above the ambient is represented by the voltage across the resistance Rt. The current source (P) supplying the resistive and capacitive loads is equal to the power dissipated by the filament [9].
The luminous flux produced by the lamp is a nonlinear function of the filament temperature, physical characteristics, and geometry. Typical 120 V incandescent lamps, with power between 45 and 200 W, have thermal time constants between 10 and 200 ms [1], [9]. The time constant of a 230 V lamp is equivalent to that of a 120 V lamp having nearly half power due to the reduction in the filament thickness in order to have the same rated power.
Incandescent Lamp with Nonsinusoidal Voltage. The instantaneous power has quite a complicated expression, which yields a cumbersome equation for the filament temperature. Only interharmonics with order smaller than 1.5 for 60 Hz and 1.67 for 50 Hz are visible. A recurrent flicker is produced only when the voltage waveform contains noninteger harmonics with frequencies in the range of 25 to 90 Hz [9].
Voltage with Square-Wave Modulation. This is the most common form of voltage flicker. It is characteristic in situations when large loads are recurrently switched on and off. This type is most likely to produce light flicker complaints.
Following the analysis in [9], the component of filament temperature that is causing visible light fluctuation can be found. Analytical and graphical study of the relative change of illumination for square-wave modulation (as a function of voltage fluctuation), lamp thermal time-constant, and fluctuation frequency have shown the following.
• Higher wattage lamps that have larger thermal time-constant cause less annoyance.
• As the frequency of voltage fluctuation increases, for the same fluctuation magnitude, the irritation decreases.
• For the same lamp thermal time-constant, the luminous flux variation (and then annoyance) increases almost lin¬early with the voltage fluctuation magnitude.
Voltage with Sine-Wave Modulation. Sinusoidal modulation causes lesser relative illumination fluctuation and smoother changes than the square-wave modulation. The lamp thermal time-constant effect is less noticeable and the illumination fluctuation decreases faster, when the frequency increases, than for square-wave modulation.
Use of Incandescent Lamp Dimmers. The wide-spread use of lamp dimmers are 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, due to the conduction-angle change ofthe electronic control principle. A typical electronic dimmer would nearly double the change in light output for a typical voltage change compared to the same lamp with no dimmer. The flicker gain factor depends on the attenuation setting and perturbation frequency, ranging from nearly 1 for 25 Hz (invisible perturbation) to a maximum of 8 at very low frequency. The gain factor reaches the value of 6 for 75% attenuation and maximum sensitivity frequency (8.8 Hz) [8]. Dimmer behavior is highly af¬fected by interharmonics existence.
Gaseous-Discharge Lamps
ФFluorescent and other forms of discharge lighting are much less sensitive to voltage magnitude fluctuations than the incandescent type, with gain factors of about 1.2 versus 2.7, which is practically nonvariable with the fluctuation frequency [10]. Discharge lamps having virtually no energy storage respond instantly to changes in voltage (time-constant is less than 5 ms) [11]. Recent tests and reported problems show how interharmonics and phase-shifting on the power line can cause fluorescent lamps to flicker at locations far away from the disruption source [12].
Most of the existing voltage fluctuation (or light flicker) standards are based on observations of annoyance caused by incandescent lamps. The physical mechanism of electric energy conversion between the incandescent and fluorescent lamps differs drastically.
• Incandescent lamps use the Joule heating process, their volt¬age/current characteristic is linear, and the filament average temperature is proportional to the squared rms voltage.
• Fluorescent lamps belong to the electric discharge lamp family; they convert the electric energy into light by transforming electric energy into kinetic energy of moving electrons and ions. The voltage/current characteristic is nonlinear, and the arc voltage remains nearly constant during each half-cycle.
The average power produced in the discharge lamp is a func¬tion of the ignition angle: the larger the angle, the smaller the current and power. Besides, the angle depends on the waveform and magnitude of the supply voltage. For the same rms voltage, a flat waveform will produce a small angle and a bell-shaped waveform will increase the angle. Integer harmonics will not cause light flicker, since the voltage waveform remains the same every half-cycle. Noninteger harmonics will cause continuous change of voltage distortion from one half-cycle to the next. Such variations do not have an effect on the incandescent lamp, but they may significantly affect the fluorescent lamp performance [11].
Arc-discharge lamps will gradually replace incandescent lamps, and voltage distortion is going to remain as an acute power quality problem for many years, which means that proper research and quantification of this type of flicker is necessary. The annoyance curves obtained for incandescent lamps may not be useful for fluorescent lamps; more research work is needed in this area.
Conclusions
From the present analysis, the following can be concluded.
• Voltage fluctuation limits cannot be simply related to flicker-caused annoyance.
• Flicker meter capability of being easily emulated or PC implemented allows the quick study of problems in its early level.
• Application of flicker curves does not give a clear and complete idea for flicker problem assessment.
• Flicker meter specifications have not reached the final stage yet due to new types of lamps and lighting requirements.
• Possible flicker-caused problems should be completely analyzed from the system design point of view.
• Determination of flicker limits requires a thorough analysis in order to allow the proper system exploitation..
References
[1] P.G. Kendall, "Light flicker in relation to power-system voltage fluctuation," Proc. Inst. Elect. Eng., vol. 113, no. 3, pp. 471-479, 1966.
[2] J.D. Lavers and P.P. Biringer, "Real-time measurement of electric arc-furnace disturbances and parameter variations," IEEE Trans. Ind. Applicat., vol. 22, no. 4, pp. 568-577, 1986.
[3] IECFlickermeter Functional and Design Specifications, IEC Stan¬dard 868, 1986.
[4] IEC Electromagnetic Compatibility—Part 4: Testing and Measure¬ment Techniques, sec. 15, "Flickermeter: Functional and design specifications," IEC Standard 61000-4-15, 1997.
[5] J. McKim, "The UIE flickermeter demystified," Compliance Engi¬neering, vol. 16, no. 3, pp. 60-71, 1999.
[6] IEEE Flicker Task Force, P1453 draft. Available: http://grouper.ieee.org/groups/1453.
[7] S. Caldara, S. Nuccio, and C. Spataro, "Digital techniques for flicker measurement: Algorithms and implementations analysis," in Proc. IEEE Instrumentation and Measurement Conf., 1999, pp. 656-661.
[8] M. Halpin, L. Conrad, and R. Burch, Tutorial on Voltage Fluctua¬tions and Lamp Flicker in Electric Power Systems, IEEE Power En¬gineering Society publication 01TP151, 2001.
[9] L. Peretto and A.E. Emanuel, "A theoretical study of the incandes¬cent filament lamp performance under voltage flicker," IEEE Trans. Power Delivery, vol. 12, no. 1, pp. 279-288, 1997.
[10] M.K. Walker, "Electric utility flicker limitations," IEEE Trans. Ind. Applic., vol. 15, no. 6, pp. 644-655, 1979.
[11] A.E. Emanuel and L. Peretto, "The response of fluorescent lamp with magnetic ballast to voltage distortion," IEEE Trans. Power De¬livery, vol. 12, no. 1, pp. 289-295, 1997.
[12] B. Bhargava, "Arc furnace flicker measurements and control," IEEE Trans. Power Delivery, vol. 8