Short T.A. Electric Power Distribution Handbook

Подождите немного. Документ загружается.

568 Electric Power Distribution Handbook

vision. The perception process involves the thermal characteristics of the

ﬁlament and the human eye–brain response.

The European community has derived sophisticated methods of charac-

terizing and analyzing ﬂicker, developing a ﬂickermeter (IEC 868, 1986) and

a comprehensive set of standards. The IEEE is moving towards adopting the

European ﬂickermeter approach to quantify ﬂicker (Halpin et al.). The ﬂick-

ermeter models the complex lamp–eye–brain interaction to uniformly quan-

tify ﬂicker from a variety of sources, whether the offending load is an arc

furnace or repetitive motor starts. The ﬂickermeter measures voltage and

produces these outputs:

• Instantaneous ﬂicker sensation — This output is in “perception units,”

a unit-less quantity. One per unit is deﬁned as the threshold of

perception for 50% of the population.

• Short-term ﬂicker indicator (P

) — The short-term ﬂicker indicator is

a weighted factor based on probabilities of the instantaneous ﬂicker

sensation over a ten-minute monitoring period. A P

= 1 is the

threshold of irritation, the level that the majority of the population

ﬁnds annoying. P

values as low as 0.7 have been found to be visible

under some conditions.

• Long-term ﬂicker indicator (P

) — Long-term ﬂicker is based on 2 h

of ﬂicker measurement and combines 12 consecutive P

values

(using the cubed root of the sum of the cubes of the 12 P

values).

The ﬂickermeter produces comparable results to the GE ﬂicker curve when

the ﬂickermeter has a square-wave input as shown in Figure 11.21. Both the

GE curve and the IEC curve (calibrated for 120-V lamps) produce comparable

results. A ﬂicker curve analysis is easier for loads that ﬂuctuate periodically

in a square-wave fashion. For intermittent or chaotic loads like welders, a

ﬂicker curve is impossible to use. Converting a measured waveform to the

frequency domain with a Fourier transform and plotting the results on a

ﬂicker curve is imprecise. While the exercise reveals predominant frequen-

cies, the Fourier transform does not mimic the brain’s response to ﬂickering

light. The ﬂickermeter is much more appropriate for cases where loads

ﬂuctuate irregularly.

The P

is an rms quantity that we can treat like a per-unit voltage. If the

current draw from the ﬂuctuating load doubles (and the waveshape stays

the same), the voltage deviation (DV) will double, and P

will double. Just

as we can use voltage dividers, we can use impedance dividers to estimate

ﬂicker propagation and estimate ﬂicker severity at other locations on a circuit

(see Figure 11.22). Flicker decreases as one moves upstream of the ﬂuctuating

load on a radial circuit. Customers at or downstream of the ﬂuctuating load

experience the worst ﬂicker.

Multiple sources of harmonics can inﬂuence ﬂicker differently, depending

on the characteristics of the ﬂuctuating loads. Multiple sources combine as

1791_book.fm Page 568 Monday, August 4, 2003 3:20 PM

Other Power Quality Issues 569

UIE, the organization that developed many of the European ﬂicker stan-

dards (UIEPQ-9715, 1999), suggests the following values of m for combining

multiple ﬂuctuating loads (smaller m is more conservative):

m = 4: for arc furnaces operated to avoid simultaneous operations

m = 3: when the risk of simultaneous voltage variations is minimal (Most

unrelated loads are in this category.)

FIGURE 11.21

GE borderline of irritation curve compared to the voltage change necessary to produce P

= 1

from the IEC ﬂickermeter with a square-wave input.

FIGURE 11.22

Propagation of voltage ﬂicker.

GE IEC

0.1 1.0 10.0 100.0 1000.

Voltage dips per minute

DV/V

Fluctuating

load

st TOTAL st i

()

1791_book.fm Page 569 Monday, August 4, 2003 3:20 PM

570 Electric Power Distribution Handbook

m = 2: simultaneous operation of devices producing random-noise type

ﬂuctuations, such as arc furnaces operating at the same time

m = 1: when there is a high probability of simultaneous operation

The ﬂickermeter can effectively quantify ﬂicker for existing loads; it does

not help much for predicting the ﬂicker at new installations. It is possible to

model the time variations of ﬂuctuating loads and run a software “ﬂicker-

meter” on the results. If a similar load exists on the system, another approach

is to measure P

at the existing location and scale it by the relative differences

between impedances at the existing location and the new location:

where

st,A

= the short-term ﬂicker indicator measured at a location with an

impedance back to the source of Z

(the three-phase short-circuit

impedance)

st, B

= the short-term ﬂicker indicator for the same ﬂuctuating load moved

to a point B which has a short-circuit impedance of Z

A basic screening criteria based on the stiffness of the distribution circuit

and the amount of ﬂuctuation also helps determine if ﬂicker could be a

problem at new installations. The Europeans have adopted a set of criteria

depending on the rate of change of the load and the ﬂuctuation relative to

the stiffness of the supply (Table 11.5). The change in load (DI) is referenced

to the short-circuit current at the nearest customer (I

). At less than ten

changes per minute, with a short-circuit level on the primary of 1000 A, a

load could ﬂuctuate by 4 A (as measured on the primary) without causing

ﬂicker if the ﬂuctuations are fewer than 10/min. North American limits could

be less conservative than this because the lower-voltage North American

light bulb is less sensitive to ﬂicker than ﬁlaments at higher European volt-

ages (120-V vs. 230-V bulbs).

Motor starts are a special case of voltage ﬂicker. Most motors start only a

few times per day, which is possibly off the ﬂicker curve. While motors may

TABLE 11.5

Thresholds Where Fluctuating Loads

Will Not Cause Flicker Problems per

(IEC 61000-3-7, 1995; UIEPQ-9715, 1999)

Number of Voltage changes

per minute, r DD

I / I

, %

r > 200 0.1

10 £ r £ 200 0.2

r < 10 0.4

st B

st A,,

1791_book.fm Page 570 Monday, August 4, 2003 3:20 PM

Other Power Quality Issues 571

start infrequently, when they do start, the voltage change is sharp, and light

change can be deep and easily visible. Motors normally draw ﬁve to six times

full-load current during starting. Normally, the voltage drops suddenly and

then gradually recovers over several seconds as the motor comes up to speed.

Utilities normally have a criteria for motor starts based on the change of

voltage and how often the customer starts the motor. Willis (1997) reported

that many utilities use a criteria of 3% during motor starting (Table 11.6),

but some utilities vary on how they deﬁne the percentage (either relative to

the nominal voltage, the minimum voltage, or the voltage at the time of the

motor start). Also, utilities often limit the size of motors or the starting

current allowed by end users, depending on the voltage (see Table 11.7 for

one utility’s limits).

The source impedance at the customer with the motor (or other ﬂuctuating

load) is an important component. As with harmonics, we are most concerned

TABLE 11.6

Voltage Flicker Limits at Several Utilities

during Motor Starting

Utility Voltage Criteria

Dense urban area 3%

Urban/suburban 3%

Suburban and rural 3%

Urban and rural 3%

Rural, mountainous 4 V (on a 120-V base)

Rural, mountainous none

Source: Willis, H. L., Power Distribution Planning

Reference Book, Marcel Dekker, New York, 1997.

TABLE 11.7

One Utility’s Allowable Motor Starting

Currents

System

Maximum Allowable

Starting Current, A

Single Phase

120 V 100

208 V 160

240 V 200

Three Phase

208 V 1554

240 V 1346

480 V 673

2400 V 135

Note: Automatically controlled motors are

limited to half of the allowable start-

ing currents in the table.

1791_book.fm Page 571 Monday, August 4, 2003 3:20 PM

572 Electric Power Distribution Handbook

about the voltage drop at the point of common coupling, the point where

other customers can be tapped off. For customers fed from the same trans-

former, the point of common coupling is at the transformer. The impedance

at the transformer is dominated by the transformer’s impedance; the pri-

mary-side impedance is normally small. If customers share secondary cir-

cuits, also add the impedance of the shared secondary. For larger customers

with their own dedicated transformer, the point of common coupling is on

the primary. To evaluate the ﬂicker to the customer that has the ﬂuctuating

load, consider the secondary and service drop to the meter. Hydro Quebec

estimated reference impedances for Canadian residential and industrial facil-

ities (Ba et al., 1997). At the panel, they estimated that 95% of customers

have lower impedances than the values in Table 11.8, which we could use

as a ﬁrst approximation for motor starting and other ﬂuctuation limits.

Interharmonics — harmonic distortions that are not integer multiples of

the fundamental — can cause voltage ﬂicker (Gunther, 2001). Noninteger

harmonics are less common than integer harmonics, so this problem is not

particularly widespread. Cycloconverters, arc furnaces, arc welders, and

induction furnaces inject interharmonics. Interharmonics also come from

loads like ovens or furnaces with integral cycle control, where the load

controls the average voltage by either giving the heater full voltage or no

voltage on a cycle-by-cycle basis (60% average voltage could come from six

cycles on, then four cycles off). Standard six-pulse or twelve-pulse power

converters do not normally create noninteger harmonics, but a converter can

create noninteger harmonics if its electronic switches ﬁre at the wrong time.

Misﬁring can come from incorrect control settings or a variety of hardware

problems. The Wisconsin Electric Power Company had such a problem with

a dc arc furnace that caused ﬂicker (Tang et al., 1994).

Two superimposed frequencies beat against each other at this frequency:

where

= frequency of the interharmonic

= integer multiple of the fundamental frequency closest to f

TABLE 11.8

Reference Impedances as Seen from the Electrical Panel

Board

Location Impedance, WW

120-V residential, phase to neutral 0.19 + j0.062

240-V residential, phase to return phase 0.20 + j0.080

600-V industrial, phase to neutral 0.58 + j0.107

600-V industrial, phase to phase 0.57 + j0.135

Source: Ba, A. O., Bergeron, R., and Laperriere, A., “Source im-

pedances of the Canadian distribution systems,” CIRED 97, 1997.

ff f

1791_book.fm Page 572 Monday, August 4, 2003 3:20 PM

Other Power Quality Issues 573

So, for n = 3.1, the beat frequency is |3.1–3| = 0.1 per unit, or |186 Hz–180

Hz| = 6 Hz (see Figure 11.23), right at the most sensitive ﬂicker frequency.

Frequencies of n = 1.9, 2.1, 2.9, 3.1, 3.9, and 4.1 per unit each beat at the same

frequency: 0.1 per unit. Lower-frequency interharmonics — those near the

ﬁrst through third harmonic — are most likely to cause ﬂicker. The IEC

ﬂickermeter detects ﬂicker due to rms changes. Low-frequency interharmon-

ics below the second harmonic change the rms, especially near the funda-

mental. But interharmonics above the second harmonic only modulate the

peak; so at higher-frequency interharmonics, where the rms stays constant,

the IEC ﬂickermeter does not detect ﬂicker (even though the ﬂuctuations in

the peak can result in noticeable ﬂicker on ﬂuorescent lights).

Incandescent lamps respond to rms voltage. Interharmonics below the

second harmonic can cause ﬂicker; higher frequencies do not. Fluorescent

lights with electronic ballasts are particularly sensitive to the waveform peak;

as the peak ﬂuctuates, light output ﬂuctuates. Electronic ballasts rectify the

ac, which tracks the waveform peaks.

Just as capacitors can amplify integer harmonics, resonances can amplify

noninteger harmonics. Solutions to ﬂicker from interharmonics can include

some of the solutions to regular ﬂicker discussed in the next section. Also,

harmonic solutions are appropriate in some situations, especially removing

resonances by resizing or moving capacitors.

11.4.1 Flicker Solutions

11.4.1.1 Load Changes

Often, ﬂicker is most economically solved at the source of the problem, the

ﬂuctuating load.

FIGURE 11.23

A 20% interharmonic frequency of n = 3.1 (186 Hz) causing a beat frequency of 6 Hz.

0.0 0.2 0.4 0.6 0.8 1.

Time, seconds

1791_book.fm Page 573 Monday, August 4, 2003 3:20 PM

574 Electric Power Distribution Handbook

• Welders — For single large welders, changing the electrode ﬁring

sequence to draw smaller but more frequent current pulses can

reduce ﬂicker (EPRI PEAC Case Study No. 1, 1997). For multiple

welders, control equipment can add a short delay to some welders

to prevent simultaneous ﬁring of several units. Also, on some weld-

ers, users can lower the weld “heat” (and current draw) without

sacriﬁcing the quality of the weld.

• Motors — For motor-starting ﬂicker, one of several reduced-voltage

starters will reduce the current draw and voltage deviation during

starting. Common ways to reduce the voltage include an autotrans-

former start, a reactor start, and electronically switching the voltage

(like a light dimmer) during starting. Reducing voltage to the motor

during starting reduces the current inrush from 5 to 6 times normal

current to under 2 per unit in some cases. Electronic motor starters

can ramp the current to a preset maximum level. However, reduced-

voltage starts also reduce the starting torque, so mechanical load

issues can limit how much we can reduce the voltage. For motors

susceptible to stalling (such as wood grinders), more careful opera-

tions on the mechanical side can reduce ﬂicker (such as controlling

the stock fed to the grinder).

• Resistive heaters — Heaters with on–off type controls ﬂicker from

repeated on and off cycling. Often, more precise control with less

ﬂuctuation is possible by upgrading the heater controls — regulating

the voltage to the heating elements or splitting one heater into sep-

arately controlled elements.

• Operational limits — A simple option requiring no physical changes

is to limit operation of offending loads to times when neighboring

customers are unlikely to be bothered by ﬂickering lights. Midnight

to 6 a.m. is normally safe. Daytime operation might be tolerated.

Operation from 6 p.m. to 11 p.m. is most likely to cause complaints.

For facilities with multiple ﬂuctuating loads, facilities can stagger

operations to limit overlapping operation of multiple units.

Another option is to provide solutions where the customers are complain-

ing. Review the conditions when ﬂicker occurs and what lighting ﬁxtures

are ﬂickering. A review of the lighting may reveal especially sensitive light

ﬁxtures such as some types of dimmers. These light ﬁxtures can be replaced

by ﬁxtures that ﬂicker less. One option is to convert from incandescent to

ﬂuorescent lights, which do not ﬂicker as much.

If the facility causing the ﬂuctuations is the only one experiencing the

ﬂicker, one solution is to isolate the lighting circuits from the circuits with

ﬂuctuating loads. Run separate circuits to the lighting loads. For single-

phase ﬂuctuating loads, put them on one phase and the lighting on the other

two phases.

1791_book.fm Page 574 Monday, August 4, 2003 3:20 PM

Other Power Quality Issues 575

11.4.1.2 Series Capacitor

On distribution lines, much of the voltage drop is due to a circuit’s induc-

tance. If we add a capacitor in series with the inductance, the capacitor

cancels out a portion of the inductance (see Figure 11.24). Now, the circuit

has less total inductance, so ﬂuctuations in load cause less voltage drop. A

series capacitor is a passive circuit element. It responds instantaneously and

automatically. Series capacitors help with voltage regulation as well as

improving voltage ﬂicker.

The best place for a series capacitor depends on where the complaining

customers are relative to the offending load. Customers upstream of the

series capacitor see no difference in ﬂicker. Voltage improves downstream

of the series capacitor. Sizing and placement involves tradeoffs. Ideally, we

want to size the capacitor such that its impedance equals the circuit imped-

ance to the ﬂuctuating load (smaller series capacitor impedances are more

expensive: more kvars, more units in parallel). For placement, a good rule

of thumb is electrically half way between the source and the ﬂuctuating load

— the point where the voltage drop is half of the total voltage drop due to

the ﬂuctuating load. Locations closer to the source reduce ﬂicker for more

customers but increase fault duties. An IEEE working group suggests a rule-

of-thumb location where the voltage drop is one-third to one-half of the total

drop (Miske, 2001).

FIGURE 11.24

Series capacitor.

Fluctuating

load

Overvoltage

protection

Source

Voltage drop along

the circuit due to the

fluctuating load

With

capacitor

Without

capacitor

1791_book.fm Page 575 Monday, August 4, 2003 3:20 PM

576 Electric Power Distribution Handbook

Series capacitors can superbly compensate ﬂicker from ﬂuctuating induc-

tive loads. For resistive loads, series capacitors provide much less beneﬁt.

The voltage drop through the distribution system is approximately I

R +

X. Reducing X with a series capacitor helps reduce the I

X term for ﬂuc-

tuating I

; but if just I

ﬂuctuates, series capacitors provide little beneﬁt.

Electrically, series capacitors elegantly solve voltage ﬂicker; in practice,

they are not widely used, mainly because of

• Reliability of short-circuit protection — Historically, spark gaps were

used to protect the capacitors during downstream faults. Utilities

have had many problems with failures.

• Cost — Series capacitors are nonstandard and must be custom engi-

neered.

• Unusual — Line crews and ﬁeld engineers ﬁnd series capacitors

strange. For example, if crews switch in a shunt capacitor down-

stream of a series capacitor, the voltage may go down instead of up.

Increased fault currents downstream of the capacitor can make coor-

dination of protective devices more difﬁcult.

• Ferroresonance — Series capacitors may also ferroresonate with

downstream transformers under the right conditions.

A key design issue is the voltage across the capacitor during faults down-

stream of the unit. During a fault downstream of the series capacitor, the

voltage across the capacitor units is the short-circuit current times the capac-

itor impedance. Depending on the impedances upstream of the capacitor

relative to the capacitor size, the voltage across the series capacitor can

exceed the system’s nominal voltage. When calculating the available fault

current downstream of the capacitor, include the impedance reduction

caused by the capacitor. The fault current normally can rise past the capacitor

(depending on the amount of capacitance relative to the system impedance

and the line X/R ratio). The highest fault current can be some distance from

the capacitor. Figure 11.25 shows an example on a 12.5-kV system for a series

capacitor placed 3 mi from the substation to compensate for a ﬂuctuating

load at 8 mi from the substation. The series capacitor signiﬁcantly alters the

fault-current proﬁle, and the fault current impresses signiﬁcant voltage

across the capacitor. Some sort of protection is needed to prevent this.

The increased fault current raises the primary voltage and the customers’

voltage during the fault. Figure 11.26 shows a proﬁle of the line voltages for

a fault at 4.75 mi for the example shown in Figure 11.25. Under this scenario,

the fault current is leading (capacitive), so voltages rise along the line until

the series capacitor. These overvoltages are unacceptable, so some overvolt-

age protection across the capacitor is vital to prevent them. This example

has a rather poor choice of location and impedances, but it illustrates how

important it is to consider these applications carefully. Reducing the imped-

ance of the capacitor reduces the overvoltages and reduces the fault currents.

1791_book.fm Page 576 Monday, August 4, 2003 3:20 PM

Other Power Quality Issues 577

Lowering the capacitor’s impedance to maintain a total line impedance that

is always reactive (never capacitive) eliminates the problem of overvoltages.

But we still have higher fault currents downstream of the capacitor, and

lowering the capacitor’s impedance reduces the ﬂicker-prevention capability.

Several protection arrangements have been used to protect capacitors from

overvoltages. A spark gap is the simplest device but has been problematic.

Duke Power reported that many of the gaps on their units kept arcing and

started ﬁres (Morgan et al., 1993). Gap erosion also reduced protection and

FIGURE 11.25

Fault-current proﬁle on a 12.5-kV circuit with a series capacitor located 3 miles from the

substation (the same line parameters as Figure 7.11, 500-kcmil AAC conductors). The series

capacitor has no short-circuit protection.

6810

-2

6810

Distance from the substation, miles

= 4W

= 2.6W X

= 3.1W

Vo ltage across the

capacitor for a

fault at the given

location, kV

Impedances to the

fault, W

Three-phase fault

current, kA

1791_book.fm Page 577 Monday, August 4, 2003 3:20 PM