A Quality Information System Based on Distributed Data Acquisition

Transcription

1 Electrical Power Quality and Utilisation, Magazine Vol. I, No. 2, 2005 A Quality Information System Based on Distributed Data Acquisition Peter G.V AXELBERG Unipower AB, Sweden Summary: Traditionally the control and supervision of a power network has mainly been focused on the operation of the network or for protecting system components during fault conditions etc. Relatively little attention has been focused on the product itself the electrical energy. This old approach is now changing since many processes in the modern society are depending on having access to electrical energy with high reliability and quality. The trend among energy suppliers and power quality sensitive industries is to put increased focus on the supervision of the power quality in order to detect power quality disturbances before they cause unplanned interruptions etc. This paper discusses a system Quality Information System for permanent monitoring of power quality. The paper also discusses the new standard IEC This standard defines calculation schemes for the most important power quality parameters. The paper also discusses a monitor that is producing power quality information in accordance with this new standard. Key words: power quality, permanent power monitoring system, IEC , distributed data processing 1. INTRODUCTION Today s society depends on having access to electrical energy with high quality in order to function satisfactorily. The blackouts that appeared in the US and Europe in 2003 have put reliability and power quality in focus even more. Another reason why power quality is of growing interest is the deregulation process within the electrical industry. The deregulation opens for new and interesting strategies regarding the cost structure since the electrical energy is now treated as a product. In a near future energy suppliers will be able to offer differentiation in the energy price depending on the level of power quality delivered. Another obvious task for both energy suppliers and their customers is to optimise investments and reduce the maintenance costs. To meet this new situation both suppliers and customers need more information regarding the status of the power network compared to what is normally available today. Therefore, a permanent monitoring system is needed for this purpose. This article discusses the structure of such a system. This article will also highlight the new power quality measurement standard IEC that has to be implemented in a permanent power quality system. conditions. Relatively little attention has been given to the supervision of the quality of the product itself (i.e. electrical energy). This (old) approach of operating a power network is changing rapidly and there is an increasing awareness among energy suppliers that supervision of parameters related to the product energy itself is of increasing importance and cannot be ignored any longer. In order to understand network operation and monitoring five different layers are defined (see igure 1). The first layer represents the power network itself. The second layer represents basic operation administrated by the SCADA system. The third layer represents the protection system of the network which means everything concerning the control of the system during fault conditions. The fourth layer is the new one and can be thought of as a layer for supervision of the product itself the electrical ig. 1. The structure of a modern power network 2. A QUALITY IN ORMATION SYSTEM Traditionally the control and supervision of a power network has mainly been focused to operate the power network and preventing damages of system components during fault Peter G.V. Axelberg: A Quality Information System Based on Distributed Data Acquisition 47

2 ig. 2. Block diagram of a QIS energy. inally, layer five represents the loads. The new type of system used in layer four is named Quality Information System (QIS) and is used by energy suppliers and consumers (i.e. industrial plants) for proactive control of the reliability and quality of delivered (or consumed) energy. The QIS monitors are permanently installed in strategic points throughout the power network and examples of locations are substations, incoming feeders to industrial plants and power quality sensitive loads. The supervision is also performed on equipments polluting the network with bad power quality. Installing a QIS is justifiable from an economical standpoint since great savings can be achieved if a power quality disturbance is detected by the QIS before it causes a costly interruption. The structure of a QIS is shown in igure 2. The key components in the system are the monitors and the database. The information in the database is evaluated using specially designed analysis software and report generators. Some of the information can also be presented in an ordinary web browser. The web browser doesn t have all functionalities compared to the dedicated analysis software, but it is good as a complement. The information produced by the QIS is based on the new standard IEC which defines calculation schemes for the most important power quality parameters (see paragraph 3). The information is thereby quality-assured and fully comparable with information produced by other systems based on the same standard. The QIS monitors are also monitoring digital status signals from circuit-breakers, relays and other system components. A digital status signal from an external device will start a recording on the analogue input channels so that the origin of a disturbance can more easily be traced. All monitors at site are connected to the database via a communication channel. In order to reduce installation costs, already existing communication channels like a public telephone line, Ethernet or twisted pair cable are used. Using existing communication channels means that the capacity of the channels differs from one site to another. That is the first reason why the size of information to be transferred must be kept as small as possible. The second reason is the risk for a communication interruption. The risk increases with increased size of data. The third reason is the cost for transfer of data. This cost is approximately proportional to the amount of data when using a public telephone line. Using distributed calculation with a minimum of data to be transferred is therefore both more reliable and more cost-effective. In a (far) future when high speed alternatives are available on all monitoring sites, it will be possible to change from a distributed data processing approach to a central processing approach. 3. THE NEW IEC STANDARD The usefulness of a QIS is dependent on the accuracy of the information produced by the system. The information must be quality assured and measurement methods and algorithms shall be used according to accepted standards. The increasing need for power quality measurements has driven the requirement for standards describing measurement methods and how the different power quality parameters are calculated and interpreted. IEC has defined a series of standards, called Electromagnetic Compatibility Standards that, among others, deal with power quality issues. The most important standard document concerning power quality measurement methods is IEC (Testing and measurement techniques Power quality measurement methods) valid since August This standard is the first overall standard available covering the measurement techniques and calculations for the most important power quality parameters. The aim of the standard is to make it possible to obtain reliable, repeatable and comparable results regardless of the instrument being used. The standard prescribes methods to calculate parameters for the following power quality disturbances: requency variations Variations in the magnitude of the supply voltage 48 Electric Power Quality and Utilization, Magazine Vol. I, No 2, 2005

3 Voltage fluctuations (flicker) Voltage dips, swells, transients and interruptions Unbalance Harmonics and interharmonics in voltage and current Mains signalling voltages Implementing IEC in the QIS monitors will guarantee sufficient quality assurance of the produced information. A. Class A and Class B instruments There are different applications of power quality measurements like load analysis, trouble-shooting and contractual measurements. Different types of applications require different instruments. An instrument used for load analysis or troubleshooting does not need the same accuracy as an instrument used for resolving contractual disputes. IEC therefore has defined two different classes of accuracy (Class A and B). Class A is the higher accuracy class used for measurements like verifying compliance with standards, contractual measurements etc. Another intention with a class A instrument is that any measurements of a parameter carried out with two different class A instruments measuring in the same point should give identical results. Compliances with the standard as a class B instrument does not immediately mean that the instrument is less reliable or less accurate. It is just that the way of calculating various parameters may be different from the well-defined Class A methods. As will be shown below, the class A methods are rather complex in several cases. The customer may decide to settle for a cheaper instrument that gives somewhat different parameter estimates. Even for class B instruments the method of calculation shall be fully documented so that any customer is able to determine for themselves what the impact of the different way of calculation is likely to be. urthermore, the standard document covers both 50 Hz and 60 Hz systems. In some cases different methods are prescribed for different power system frequencies. Unless otherwise noted, the text below refers to Class A instruments in a 50 Hz system. B. Variations and events When studying power quality disturbances it is important to distinguish already at an early stage between variations and events [1][2]. Variations are continuous or steady-state disturbances that typically involve small variations around a normal or nominal level. Examples are frequency variations and harmonic distortion. Events are large disturbances like voltage dips and transient overvoltages. The difference between variations and events becomes very clear when doing measurements: variations are those disturbances that can be measured at predefined instants; events require a trigger. Integration times or power quality variations a measurement window has to be determined before any measurement can be done. The parameter that actually varies (frequency, voltage magnitude, harmonic voltage) is determined first over a basic measurement window. or frequency variations the basic measurement window is 10 seconds, for all other variations the basic measurement window is 10 cycles in a 50 Hz system and 12 cycles in a 60 Hz system. As the window length is approximately 200 ms in each case, the basic measurement window is often referred to as the 200 ms window. With this time integration window as a base, three measuring intervals are defined. These are: 150 cycles (180 cycles in a 60 Hz system), 10 minutes and 2 hours. The 150 cycles rms value is calculated as the root mean square of fifteen 10-cycles rms values. The windows shall be continuous and non-overlapping so that it is easy to prove that the calculated 150 cycle value is the correct rms value obtained. The aggregation from 150 cycles to 10 minutes is more complicated since the actual frequency will vary. When the system frequency is exactly 50 Hz, there are exactly 200 intervals. or a frequency of 49.5 Hz the 150 cycles become 3.03 seconds long and there will be only 198 of them in a 10-minute interval. In most cases the number of intervals is not an integer number and the last interval is discarded in the calculation. Despite the discarded data, it remains safe to calculate the U rms/(10-min) as the rms voltage. Each 10-minute interval must begin on an absolute 10 minutes clock time, +/ - 20 ms. These intervals are used when calculating the voltage magnitude, harmonics and interharmonics and the voltage unbalance. C. lagging A sudden event like a voltage dip, swell or an interruption will influence other power quality parameters like flicker [3], unbalance etc. In order not to count an event twice, IEC defines a flagging concept. When an event such as a voltage dip, swell or a short interruption occurs the instrument shall record that specific event and indicate that the other Peter G.V. Axelberg: A Quality Information System Based on Distributed Data Acquisition 49

4 ig. 3. Definition of a voltage sag according to IEC Voltage swell and interruption have similar definition parameters may be affected by this event. Therefore the interval is flagged and during the interpretation phase of the measurements it should be decided if the value obtained during this interval is to be included in the final result. The other power quality parameters shall not be recorded. Instead, the interval shall be flagged, meaning that it is marked to show the specific event and that other measured data during the interval should be ignored. The flagging concept should be handled with care however. When a contract or regulatory requirement does not require the reporting of voltage dips, it may be considered inappropriate to leave out the flagged value due to voltage dips. As voltage dips are not counted, there is no case of double counting. lagging should be treated especially careful when reporting voltage-magnitude variations. Voltage-dip and voltage-swell thresholds are typically set at 90% and 110% of nominal voltage, respectively. When dips and swells lead to removal of voltage magnitude values, the voltage magnitude will never be outside of the range from 90 to 110%. When a 200 ms interval is flagged the 3 second, 10 minute and 2 hour intervals shall also be flagged. This could lead to the removal of a large part of the data. One minor voltage dip does not significantly affect the 2 hour values. Removal of flagged data remains an issue of discussion. inally, the standard document does not state anything about flagging due to voltage transients. A site with a large number of transients may show a significant number of intervals with anomalous data that is not flagged. D. requency The frequency shall be calculated every 10 seconds for Class A instruments. To calculate the frequency, zero-crossings during 10 seconds are counted. The accuracy for Class A shall be better or equal to ± 10 mhz and less than ±100 mhz for a Class B instrument. An accuracy of 10 mhz requires a measurement window of 10 second ± 2 ms. E. Voltage rms value The voltage rms value is calculated every 10 cycles for Class A instruments. Based on this rms value, the 150 cycles and 10 minute rms values are calculated. The accuracy for Class A shall be better than or equal to ± 0.10 % of nominal voltage and for Class B ± 1.0 %. licker The flicker calculations for Class A instruments shall follow the restrictions in the standard IEC ( lickermeter functional and design specifications) [4]. urthermore, measurement methods how to trace a flicker source is described in [5] and [6]. Sags, swells and interruptions As mentioned before, power quality events like sags, swells and interruptions, require a triggering mechanism to be detected. The triggering mechanism prescribed in IEC is based on the 1 cycle rms values of the voltage updated every ½ cycle for a Class A instrument. When an rms value exceeds or falls below a stated trigger level (in one of the phases), the instrument shall start recording and continue until the rms value has returned to normal (on all phases). The first instant is referred to as the start of the event, the second as the end of the event. The time between the start and the end of the event is called the duration of the event. The lowest rms value (in any of the phases) for a voltage dip is called the retained voltage. The accuracy for Class A instruments shall be within ± 0.2 % of the stated nominal voltage and ± 2.0 % for class B. Note that there are three different thresholds involved here: a voltage-dip threshold to detect beginning and end of voltage dips; a voltageswell threshold to detect beginning and end of voltage swells; and a voltage-interruption threshold to detect beginning and end of voltage interruptions. The standard document does not give any values for these thresholds, but typical values used are 90% for the voltage-dip threshold, 110% for the voltage-swell threshold and 10% for the voltage-interruption threshold. Some discussion on threshold setting can be found in IEC technical report [7]. rom the above discussion a number of interesting conclusions can be drawn: Every voltage interruption is also counted as a voltage dip. Some other documents, notably IEEE Std. 1159, use a lower limit for the retained voltage of a voltage dip. IEC does not. 50 Electric Power Quality and Utilization, Magazine Vol. I, No 2, 2005

5 A three-phase event with a voltage drop in one phase and a rise in another, is counted both as a voltage dip and as a voltage swell. The standard document does not resolve problems like this, as it would involve interpretation of the results. A document dedicated to voltage-dip measurements is currently under development within IEEE [8]. This document will address some of the interpretation issues.. Unbalance To fulfil the requirements of Class A, unbalance shall be calculated using the method of symmetrical components. rom the measured phases, the three symmetrical components are calculated (positive, negative and the zero sequence component) over a 10-cycle window. The unbalance factor is calculated as the absolute value of the ratio between the negative and the positive sequence component expressed as a percentage. Unbalance shall be calculated over 10-cycle, 150-cycle and 10- minute intervals. G. Harmonics and interharmonics To fulfil the requirements for Class A, the calculations shall be made in accordance with IEC [9]. This document describes normative measurement methods for harmonics and interharmonics up to 2 khz. The standard gives also guidelines on how to design an instrument fulfilling the requirements in IEC The basic interval for harmonic measurements is again the 10-cycle interval. A discrete ourier transform (D T) over a 10-cycle window gives a spectrum with a frequency resolution of 5 Hz. This implies that in between the harmonic frequencies (integer multiples of 50 Hz), nine additional values are available. The lowest and the highest of these are added to the (integer) harmonic. The remaining seven together form the interharmonic. Thus for the interval from 245 Hz to 305 Hz: 255 Hz is added to 250 Hz and 245 Hz to form the 5th harmonic; 295 Hz, 300 Hz and 305 Hz form the 6th harmonic. The remaining values from 260 Hz to 290 Hz will form the interharmonic 6.5. This method of calculation does not find the frequency of discrete interharmonics but it does result in an index for the level of distortion present between two harmonic frequencies. H. Miscellaneous IEC also contains testing methods to verify an instrument according to the standard. The standard further contains a section describing installation, measurement techniques, transducers etc. Even before IEC was published a discussion had started on further development of the document. The document is an enormous step forward in defining methods for power quality monitoring. or the first time it is possible to talk about the characteristics of a voltage dip or about the level of interharmonic distortion without having to define the terminology in tedious detail. The advantages of this are obvious, for bilateral contracts, for regulatory requirements, but also for academic publications, measurement campaigns, textbooks etc. There are however a few places where the documents needs further development in the future. Some of this is already addressed in a number of informative annexes with the documents. Other material still needs to be developed. Examples of disturbances that are not addressed at all in the standard are voltage transients and higherorder harmonics. This is certainly in part due to the difficulties in accurately measuring them. Other parts where further development is needed are in the processing of three-phase measurements and in the power-related parameters. 4. THE QIS MONITOR The monitors are the cornerstones in a QIS since they are responsible for sampling and calculation of most of the data. A badly designed monitor with too low accuracy or lack of norm compliance makes the produced information less valuable. The information must be quality ig. 4. Block diagram of a QIS monitor Peter G.V. Axelberg: A Quality Information System Based on Distributed Data Acquisition 51

6 assured and a condition for that is that the monitor calculates the power quality parameters according to the IEC Class A. Beside norm compliance; another important thing is to use a monitor that reduces the sampled data as much as possible. Thereby a minimum of data has to be transferred over the communication channel. On the other hand it is a demanding task for the monitor to fulfill the calculations needed for compliance with IEC Class A because it requires both high absolute accuracy and high calculation speed. This is difficult to combine since a higher calculation speed requires a higher processor speed leading to an increased internal noise and thereby a decrease in the absolute accuracy etc. igure 4 shows a block diagram that includes the most important parts of a monitor that is used in a QIS. A modern monitor is always designed for three-phase measurements. The monitor allows for at least four voltage channels and four current channels. Three of the voltage channels are dedicated to measure the phaseto-neutral voltages or the phase-to-phase voltages. The fourth channel can be used to measure any voltage like the neutral-toprotective ground voltage etc. Three of the current channels are dedicated to measure the phase currents. The fourth channel is often used to measure the current flowing in the neutral. A. Analogue inputs and A/D conversion The input signal conditioning and the A/D conversion forms the basis of a modern monitoring system. The digitalization simplifies the design of analogue circuitry and provides greater flexibility for altering the algorithms to be used for manipulating (processing) the (sampled) data. The overall accuracy of the system depends on the input dynamic range, the sampling rate, and the number of bits in the ADC. Two different signal paths exist on the input channels. The signal from the first path is used for calculation of power quality parameters referred as variations (voltage rms, harmonics, flicker etc.) while the signal from the second path is used to detect events like transients etc. The signal from the first path passes an anti-alias filter in order to fulfill alias attenuation required by IEC Class A. All analogue signals are connected to a multiplexer which is a channel selector for the ADC. The multiplexer must switch between the input channels at a sufficient high speed rate; otherwise the time delay between the conversions will introduce an error when certain parameters like the power quantities are calculated. B. Digital input and output channels As mentioned before, a QIS monitor must be able to monitor other types of phenomena than pure power quality parameters. Therefore, also digital input and output channels are implemented. One reason for implementation is the increasing demand of a monitor which is a combination of a traditional power quality analyzer and a fault recorder. The digital inputs monitor the (actual) status of relays, circuit breakers etc. When such a device trips, the information from the digital channels is stored along with the traditional power quality parameters. A fault can thereby be evaluated more accurately. C. Generation of the sampling frequency (f s ) Traditionally, a sampling frequency of 6400 Hz has been used as a de facto standard among manufacturers of power quality monitors. A sampling frequency of 6400 Hz means 128 sampling points per period at 50 Hz. The T algorithm can be used since 128 is a power of 2. To follow the IEC Class A, the situation will be slightly different. The requirement in IEC Class A is a frequency resolution of 5 Hz. With the required base window of 10 periods, and a wish to use the T algorithm, 1024-point T calculation must be performed on each base window. The new sampling frequency will be 5120 Hz (or a multiple). Harmonics up to the 51st can be evaluated with a resolution of 5 Hz. The sampling frequency is locked to a multiple of the power frequency and thereby also a multiple of the 10-period time window by using a Phase Locked Loop (PLL). igure 5 shows a block diagram of a PLL. The PLL consists of a comparator, a phase detector, a low-pass filter (LP-filter), a Voltage Controlled Oscillator (VCO) and a divider (counter). The purpose of the PLL is to take a signal from the VCO, divide the frequency by an integer N and compare (in the phase detector) that result with the frequency of the reference signal U ref =U L1 (i.e. power network signal from phase L 1 ). The phase detector output is fed back to the VCO through a lowpass filter. The loop works to adjust the output of the phase detector to zero automatically. That means the output of the divider is precisely on the same frequency as the reference signal. If the frequency of the reference signal is changing, the output of the phase detector will give a non-zero output. The frequency of the VCO will thereby change until the PLL is locked to this new frequency. The output from 52 Electric Power Quality and Utilization, Magazine Vol. I, No 2, 2005

7 the VCO is used to trigger A/D conversion at a sampling rate which is exactly N times higher than the actual power frequency. D. Time synchronization In a power quality survey (especially troubleshooting) is it important to know when a certain phenomenon occurs. Therefore, all power quality monitors have a built-in real-time clock. Moreover, if it is necessary to carry out simultaneous measurements at different geographical locations, time synchronization is a must. Depending on the requirements in such a measurement, the accuracy of the internal realtime clock is often enough. But in some application more accurate time synchronization is needed. In those cases, a GPS receiver connected to the monitor will manage the time synchronization. The cost for such a device has decreased significantly over the past years, making it an attractive solution in many applications where extreme accuracy in real-time is needed. One possible disadvantage of a GPS is that the signals from the satellites sometimes are too weak. This happens especially at indoor locations. E. Memory The memory size and structure used in a monitor depends on the amount of data to be processed and transferred. or example, an 8-channel monitor using a 14 bit ADC (requires 2 byte per conversion) at a sampling frequency of 5120 Hz generates a data flow per second of 2x8x5120 bytes = 80 kb/s. To be able to administrate and calculate that amount of data, special care must be taken when designing the monitor. or example circular memory buffers must be used for quick access to the data when calculations are performed. Normally a monitor has memories of two different types. One type is a fast memory (e.g. SRAM) in which the software is stored and running. The other one is a non-volatile memory (e.g. flash memory) in which the measurement data is stored. In some monitors, a hard drive is used to store measurement data. The advantage of using a hard-drive is the memory capacity. On the other hand, it is a disadvantage to use a hard-drive in portable instruments since it is sensitive to mechanical stress.. Error The total error in a measurement is a sum of errors that can (mainly) be divided into three different categories: instrument errors (quantization, offset- and linearity errors etc.) transducer errors errors due to the measurement signal (low signal level etc.) The (theoretical) maximum error that can occur is the sum of the absolute values of each individual maximum error. This error is probably greater than the actual one since a summation of the absolute values will give the worst case error. A more realistic estimation of the total possible error is achieved by the concept of uncertainty where each error contribution can be considered statistically distributed. If the uncertainties can be considered independent and if they follow the normal (Gaussian) distribution, the total uncertainty is calculated as the root square sum of the individual errors. An example: Consider an instrument contributing with a maximum error of 0.2%, the transducer is contributing with a maximum error of 0.5% and the signal itself is contributing with a maximum error of 0.5%. The maximum error according to the summation of the absolute values of each individual error gives a total maximum error of 1.2%. Using the concept of uncertainty, the root square sum gives a total uncertainty of only 0.73%. G. Time window As mentioned before, a time integration window is defined as a number of sampling values used as a base for different kinds of calculations. rom this base window (e.g. 10 periods for 50 Hz according to IEC ) other time windows can be defined (150 periods, 10 seconds, 10 minutes and 2 hours according to IEC ). The principle of aggregation of a base window to another time window is shown in igure 6. ig. 5. unctional blocks of a Phase- Locked Loop (PLL) ig. 6. Example of time aggregation from a number of base windows Peter G.V. Axelberg: A Quality Information System Based on Distributed Data Acquisition 53

8 5. CONCLUSIONS There is an increasing demand for accurate supervision of the power quality in the electrical network. Previously most of the measurements were performed using portable analysers installed at site for a few days or a week. The new trend is to use a combination of portable analyzers and a permanently installed quality information system (QIS) for supervision of the power quality. The portables are used both for supervision of the power quality and for troubleshooting, while the QIS mainly is used for supervision of the power quality. The output produced by such systems is very much dependent on the performance of the analyzers and the QIS. It is most important that the information is based on an accepted measurement method. The standard IEC is valid since August Using the measurement methods described in this standard makes it possible to obtain reliable, repeatable and comparable results regardless of the instrument being used. This is always important but particularly important for power companies and their customers entering into contractual relationships for the delivery of an assured quality of supply. RE ERENCES 1. Bollen M.H.J.: Understanding power quality voltage sags and interruptions. 2000, New York: IEEE Press. 2. Bollen M.H.J.: What is power quality? Electric Power Systems Research, Vol. 66, no.1 (July 2003), pp Koponen P., Mäkinen A., Seesvuori R.: Voltage dips caused problems with digital flickermeters. IEEE Porto Power Tech Conference, September 2001, Porto, Portugal. 4. IEC , lickermeter functional and design specifications. 5. Axelberg P.: Measurement Methods for Calculation of the Direction to a licker Source. Thesis for the degree of Licentiate of Engineering, Chalmers University of Technology, Dept. of Electric Power Engineering, Gothenburg, Sweden, Axelberg P., Bollen M.H.J., Gu J.: A measurement method for determining the direction of propagation of flicker and for tracing a flicker source. CIRED 2005, Turin, Italy. 7. IEC , Voltage dips and short interruptions on power electric power supply systems with statistical measurement results. 8. IEEE P1564, Voltage sag indices, draft workinggroup document. This document is still under development. Check for the status of the document. 9. IEC , Ed. 2.0, Testing and measurement techniques General guide on harmonics and interharmonics measurements and instrumentation for power supply systems and equipment connected thereto. ACKNOWLEDGEMENTS Thanks are due to Math Bollen (currently at STRI AB) for his help with the section describing IEC Peter G.V. Axelberg received the M.Sc and Tech. Lic degrees from Chalmers University of Technology, Gothenburg, Sweden in 1984 and 2003, respectively. rom 1984 to 1992 he was at ABB Kabeldon in Alingsås, Sweden. In 1992 he co-founded Unipower where he is currently active as manager of business relations and research. Since 1992 he is also a senior lecturer at University College of Borås, Sweden. His research activities are focused on power quality measurement techniques. Address: Unipower AB, Sweden, pax@unipower.se. Math H.J. Bollen is responsible for the product area power quality and EMC at STRI AB, Ludvika, Sweden. He received the M.Sc. and Ph.D. degrees from Eindhoven University of Technology, Eindhoven, The Netherlands, in 1985 and 1989, respectively. Before joining STRI in 2003, he was a research associate at Eindhoven University of Technology, a lecturer at University of Manchester Institute of Science and Technology, Manchester, UK and professor in electric power systems at Chalmers University of Technology, Gothenburg, Sweden. His research interest covers a wide range of power system issues with special emphasis on power quality, reliability and related subjects. He has published a number of fundamental papers on voltage dip analysis and a textbook on power quality. Math Bollen serves as a co-chair of the voltage sag indices task force in IEEE PES and was an active member of CIGRE working group C4-07 (power quality indices and objectives). 54 Electric Power Quality and Utilization, Magazine Vol. I, No 2, 2005