There is an increasing concern that the qual-ity of power delivered by Egyptian electric distribution companies may be adversely affected by nonlinear loads.

Nonlinear loads, such as arc furnaces and static power converters, generate harmonic currents that may produce excessive harmonic distortion on the electric distribution system. Harmonic currents and voltages can cause several harmful effects on an electrical system, including:

Increased losses. Overheating of conductors and equipment. Over-stressing of insulation. Telephone interference. Disruption of electronic equipment. Improper operation of protective relaying. Metering errors.

Capacitors have increasingly been used to improve the system power factor. Capacitors, while not generating harmonics themselves, can cause resonance conditions that can magnify harmonic levels. To address this issue, the Electrical Power Transmission & Distribution Branch of the Electricity & Power Research Council in Egypt has begun a program of harmonic measurements at different voltage levels throughout Egypt.

As a part of this effort, the team of Black & Veatch International (BVI), Sabbour Associates and the Alexandria Electricity Distribution Co. (AEDC) is surveying harmonic distortion levels on the AEDC medium-voltage system. The system is mainly 11-kV underground cable serving 300 kVA to 1500 kVA distribution-service transformers in open loop configuration. The looped feeders are fed either directly from distribution substations or from distribution points (switching stations). The team is also making detailed computer harmonic analyses of selected portions of the system. The results of one analysis, at Nehas 33/11 kV Substation, are included in this article.

Distortion Levels on the Medium-Voltage System Voltage distortion levels at some points exceed the limits recommended by the IEEE Recommended Practices and Requirements for Harmonic Control in Electrical Power Systems, IEEE Std 519-1992. A limit of 5% for total harmonic distortion (THD) is recommended at medium-voltage service points. THD is the root sum squared of individual harmonic frequency components divided by the fundamental component. Individual harmonic voltages are limited to 3% of the fundamental voltage.

Power and harmonic distortion measurements were taken at 39 locations, most extending over a 24-hour period. Figure 1 shows the distribution of THD measurements with the time of day. A marked association is evident, with the highest distortion occurring between 6:00 and 7:00 a.m. The total data distribution of voltage THD measurements is shown in Fig. 2. Overall, 5.6% of the measurements exceeded the recommended limit. The measurements were not made on a scientific random sampling basis, so the results cannot be inferred to the overall AEDC system. Most of the measurements were at distribution substation buses, but large isolated nonlinear loads will distort the voltage more at buses closer to the load. On the other hand, we were looking for situations that might be subject to harmonic distortion and made more measurements at locations with known nonlinear loads.

While measuring the harmonic bus voltage distortion, the team also measured the current distortion in selected feeders. The feeder current total harmonic distortion was also generally highest in the morning. The percentage distortion was expected to be high because this is the lightest load period. The magnitude of the individual harmonic current in amperes, however, was also highest at the time when the fundamental load current was lowest. Figure 3 shows a typical distribution of individual harmonic currents with time along with the fundamental load current. The three highest harmonics are detailed. High harmonic current in the early morning explains why the voltage distortion is highest at this time. The harmonic voltage occurs as a result of the current flowing through various parallel impedance paths, with the least impedance generally being to the source. Some harmonic current also flows into linear loads and some into shunt capacitors. Probable explanations of this phenomenon are:

During the early morning hours, many of the linear loads on a feeder that could absorb harmonic currents are turned off. Much of the nonlinear load may be street lighting that is off during the day. When the load is light, capacitors that normally absorb harmonic currents are switched off.

Nehas 33/11-kV Substation Nehas 33/11-kV Substation is located in an industrial area of Alexandria, Egypt and primarily serves a single factory complex, the Egyptian Copper Works (Nehas). This complex includes steel-making and aluminum works as well as copper works. The system was selected for detailed measurement and analysis because the load includes two large electric arc furnaces equipped with harmonic filters and several dc motor drives, which are sources of harmonic distortion. The electrical one line diagram of the substation and the medium-voltage system is shown in Fig. 4.

The team measured harmonic voltage and current distortion on the 11-kV substation bus 1 and on all of the outgoing substation feeders. During two of the measurements on the 25 ton furnace feeder, the harmonic filter was out of service, which provided the opportunity to measure the unfiltered harmonic currents. While the voltage harmonic distortion on the substation bus did not exceed the recommended limits, the harmonic current in the feeder serving the arc furnaces was excessive.

Detailed Analysis A detailed harmonic analysis of the Nehas system was made using a commercial software program. The program provides three basic types of analysis: load flow, harmonic distortion and system frequency response. Filter parameter calculations are also included to assist in the specification of harmonic filters. A fundamental frequency load flow analysis is made before analyzing harmonic distortion to determine base values of voltage and current. The harmonic distortion analysis is made by calculating the system admittance matrix for each frequency. For each harmonic order, a current source is applied at the location of every harmonic load; bus voltages and branch currents are then calculated. The system frequency response is calculated by injecting one ampere of current at each harmonic source location and calculating the voltage at the selected buses over a range of frequencies.

The detailed analysis at Nehas was limited to 11-kV bus 1, where the electric arc furnaces are served. An alternate source to some of the loads is provided by the 11-kV bus 2. During the harmonic measurements, breaker 51 had no load, indicating that transformer point 587 A&B and presses distribution point (DP) were served from 11-kV bus 2. The 25-ton furnace filter is a damped single frequency tuned filter with damping resistors and capacitors loosely coupled magnetically to the main filter reactor. This filter is switched with the furnace load. The ladle furnace filter is a single frequency tuned filter but is not normally in operation because it is not needed for reactive power compensation.

Background Harmonic Distortion One difficulty in analyzing an existing system is that analysis software assumes the existence of a pure sinusoidal voltage source and requires that all sources of distortion be expressly defined. Real systems are complex. Sources of distortion are so widely dispersed that they cannot be separately defined. Neither can the system impedance be completely defined. Analysts are inevitably faced with measured distortions that do not fully match the distortion calculated by the theoretical model.

At Nehas, the distortion measurements included times when the arc furnaces were not in operation and no other known harmonic sources were connected to the bus. Nevertheless, there was distortion in the bus voltage. To account for this distortion, we applied a fictitious "background" harmonic source to the high side of the substation transformer. The harmonic spectrum of the background source was designed to produce the average measured 11-kV bus voltage when the arc furnace was off. This solution is not ideal because this fictitious source will not react to alternative circuit configurations, loads and elements in precisely the same manner as the actual system.

Base Case Analysis The system identified as the base case represents the situation where measurements were made on the 25 ton arc furnace circuit when the furnace and its harmonic filter were in operation. A project specific arc furnace harmonic current source was modeled to match the average of the highest 50% of current distortion measurements made on the furnace. Several alternative circuit configurations are compared to this base configuration. The calculated harmonic voltage distortion for 11-kV bus 1 for this base case was reasonably close to the measured values. The total harmonic voltage distortion was measured as 2.2%, but calculated to be 1.7%.

Alternative Configuration Analyses An analysis without the arc furnace filter showed that the filter has a limited effect on the harmonic distortion. Figures 5 and 6 show the voltage and current harmonic spectra, respectively, without the filter compared to the base case. The filter reduces third order and higher harmonic distortion but increases the second order distortion. The increase in the second order distortion is a result of resonance between the filter and the system.

The Egyptian Copper Works installed a harmonic filter on the 25 ton furnace primarily to avoid power factor penalties. The filter helped avoid problems that might occur if unfiltered capacitors were applied in the presence of high harmonic distortion. To see the extent of these problems, the filter in the model was replaced with a 3500 kVAR capacitor bank, which is large enough to compensate for the furnace reactive power requirements. The unfiltered capacitor bank resulted in a resonance with the system at the tenth harmonic with high distortions in both the bus voltage and the feeder current, as shown in Figs. 7 & 8. Power factor correction with capacitors alone is unacceptable in the presence of the large nonlinear arc furnace load. Although the harmonic filter itself does not greatly reduce harmonic distortion, the filter is necessary to prevent the resonance that would occur if capacitors alone were used. To determine if a better solution than the existing harmonic filter existed, the analysis program's harmonic filter calculator was used to model several alternative filters. Table 1 shows the characteristics of the filters investigated and the resulting distortion in the 11-kV bus 1 voltage and the 25 ton DP feeder current. The first alternative, a single tuned filter where the Q factor is Q=32, was determined by calculating the optimum Q for voltage distortion reduction. The capacitor kilovolt shown in Table 1 is the phase to neutral rating; the capacitor kilovar is the three phase capacity at rated voltage. The effective kilovar is the net filter capacity at the fundamental frequency. The target harmonic order is indicated as h0.

All of the alternative filters have considerably higher losses than the existing, loosely coupled, damped filter. None of the alternatives suppress the bus voltage distortion as well as the existing filter, although the single tuned filter with Q=32 and the dual single tuned filters suppress the branch current distortion somewhat better.

Conclusion Preliminary harmonic distortion measurements made on the AEDC medium-voltage system show that there are isolated areas where harmonic distortion exceeds the recommended limits; however, in most locations studied, the utility bus voltage distortion limits are exceeded only for a short time in the early morning.

No problems, such as blown capacitor fuses, have been reported in utility operations. In the few instances where customer problems have been identified, circuit configuration changes solved the problems. The lack of reported problems, however, does not necessarily mean that there are no harmful effects from the harmonic distortion present on the system. Increased losses may be occurring without being recognized. Overheating of conductors and equipment and over-stressing of insulation may cause elevated failure rates in the future. Metering errors also may go undetected.

This study has shown that the measurement and analysis of harmonic distortion is feasible and useful. Obtaining the necessary data to model the system is a problem, however. The measured levels of harmonic distortion are difficult to match by applying theoretical harmonic sources because of uncertainty in both the characteristics of the sources and the configuration of the system at the time of the measurements. For reliability reasons, the medium voltage networks generally have multiple paths to serve each load. The circuit configurations at any particular time are often known only to local operating personnel, requiring close coordination between study and operating personnel.

In spite of the difficulties, the team made a useful analysis of the Nehas 33/11-kV Substation medium-voltage network. The analysis confirms that the existing arc furnace filter is appropriate and necessary. Although the harmonic current distortion in the arc furnace feeder exceeds recommended limits, it does not cause high-voltage distortion on the AEDC bus. Alternative filter designs will not reduce the current distortion significantly without high filter losses. TDW

James T. Ghrist, distribution engineer, received the BSEE from West Virginia University in 1968 and the MBA from the University of Pittsburgh in 1973. He is responsible for studies and designs associated with the USAID-financed Alexandria Electrical Network Modernization project, including harmonic distortion measurement and analysis, load management, outage analysis and power factor correction. Ghrist is a registered professional engineer in five U.S. states and has been with Black & Veatch, Kansas City, Missouri, U.S., since June 1973.

Nazineh G. Eassa, received the B. Sc and M. Sc. degrees from Alexandria University in 1975 and 1982, respectively. She is the manager of the Capacitors Department in Alexandria Electricity Distribution Co. Eassa has been involved in the USAID-financed Alexandria Electrical Network Expansion and Modernization projects since 1983.

Ibrahim Y. Megahed received the B. Sc degree from the University of Alexandria, Egypt in 1960 and the Ph. D degree from the University of London, England in 1965. He is chairman of the Electrical Engineering Department of Alexandria University and consultant to the Alexandria Electricity Distribution Co. Megahed is a fellow of IEE, UK.