Chapter 3.

Definitions Grady
June 2005
3. Definitions

3.1. RMS

The rms value of a periodic current (or voltage) waveform is defined as

t +T
1 2
I rms =
T ∫ i(t ) dt ,
t
so that
t +T
2 1 2
I rms =
T ∫ i(t ) dt . (3.1)
t
If the current is sinusoidal, then the rms value is simply the peak value divided by 2.
However, if the waveform has Fourier series

∞
i (t ) = ∑ I k sin(kω o t + θ k ) ,
k =1
then substituting into (3.1) yields
t +T ⎛ ∞ 2
1 ⎞
2 ⎜ ∑ I k sin( kω o t + θ k ) ⎟ dt
T ∫ ⎜⎝ k =1
I rms =
⎟
t ⎠
t +T ⎛ ∞ ∞ ∞ ⎞
1 ⎜ I 2 sin 2 (kω t + θ ) + 2 ⎟dt
2
I rms =
T ∫⎜ k∑ o k ∑ ∑ I m I n sin( m ω o t + θ m ) sin( n ω o t + θ n )
⎟
t ⎝ k =1 m =1, n =1,m ≠ n ⎠
t +T ⎛ ∞
1 ⎛ 1 − cos 2(kω o t + θ k ) ⎞
∫ ⎜⎜ ∑ I k ⎜⎝
2 2
I rms = ⎟
T 2 ⎠
t ⎝ k =1
∞ ∞
⎛ cos((m − n)ω o t + θ m − θ n ) cos((m + n)ω o t + θ m + θ n ) ⎞ ⎞⎟
+ ∑ ∑ ImIn⎜
⎝ 2
+
2
⎟ dt
⎠ ⎟⎠
(3.2)
m =1, n =1,m ≠ n

Equation (3.2) is complicated, but most of its terms contribute nothing to the rms value if one
thinks of (3.2) as being the average value for one fundamental period. The average value of each
cos 2(kω o t + θ k ) term is zero because the average value of a cosine is zero for one or more
integer periods. Likewise, the average value of each cos((m ± n)ω o t + θ m ± θ n ) term is also
zero. Thus, (3.2) reduces to
∞
t +T ⎛ ∞ ⎞
∑ I k2 ∞ 2
2 1 ⎜ ∑ I 2 • 1 ⎟dt = k =1 (t + T − t ) = ∑ ⎛⎜ I k ⎞⎟ ,
T ∫ ⎜⎝ k =1 k 2 ⎟⎠
I rms = ⎜ ⎟
2 k =1⎝ 2 ⎠
t

and yields the final result

Page 1 of 12
 Chapter 3. Definitions Grady
June 2005

∞ 2
⎛I ⎞ 1 ∞ 2,
I rms = ∑ ⎜⎜ k ⎟⎟ = ∑I (3.3)
k =1⎝ 2 ⎠ 2 k =1 k
or
2
I rms = I12, rms + I 22, rms + I 32, rms + L . (3.4)

Equations (3.3) and (3.4) ignore any DC that may be present.

From (3.3) and (3.4), it can be observed that cross products of unlike frequencies contribute
nothing to the rms value of the total waveform. The same statement can be made for average
power, as will be shown later. Furthermore, since the contribution of harmonics to rms add in
squares, and their magnitudes are often much smaller than the fundamental, the impact of
harmonics on rms is usually not great.

3.2. THD

The most commonly-used measure for harmonics is total harmonic distortion (THD), also known
as distortion factor. It is applied to both voltage and current. THD is defined as the rms value of
the harmonics above fundamental, divided by the rms values of the fundamental. DC is ignored.
Thus,
∞ 2 ∞
⎛I ⎞
∑ ⎜⎝ k2 ⎟⎠ ∑ I k2
k =2 k =2
THD I = = . (3.5)
I1 I1
2

THD and rms are directly linked. Note that since

1 ∞ 2
2
I rms = ∑I
2 k =1 k

and since
1 ∞ 2 ∞
∑I
2 k =2 k
∑ I k2 − I12
THD I2 = = k =1 ,
I12 I12
2

then combining the above two equations yields

∑ I k2 = I12 + I12 • THDI2 = I12 (1 + THDI2 ),

∞

k =1

Page 2 of 12
 Chapter 3. Definitions Grady
June 2005
so that
1 ∞ 2 I12
∑
2 k =1
Ik =
2
1 + THD I2 ,( )
or

2
I rms (
= I12, rms 1 + THD I2 . )
Thus the equation linking THD and rms is

I rms = I1, rms 1 + THD I2 . (3.6)

Because line losses are proportional to the square of rms current (and sometimes increase more
rapidly due to the resistive skin effect), then line losses always increase when harmonics are
present. For example, many PCs have a current distortion near 1.0 (i.e., 100%). Thus, the
wiring losses incurred while supplying a PC are twice what they would be in the sinusoidal case.

While current distortion in loads varies from a few percent to more than 100%, voltage distortion
is generally less than 5%. Voltage THDs below 0.05, i.e. 5%, are considered acceptable, and
those greater than 10% are definitely unacceptable and will cause problems for sensitive
equipment and loads.

3.3. Average Power

Harmonic powers (including the fundamental) add and subtract independently to produce total
average power. Average power is defined as

t +T t +T
1 1
Pavg =
T ∫ p(t )dt = T ∫ v(t )i(t )dt . (3.7)
t t

Substituting in the Fourier series of voltage and current yields

t +T ⎛ ∞ ∞ ⎞
1 ⎜ ∑ Vk sin( kω o t + δ k ) • ∑ I k sin( kω o t + θ k ) ⎟dt ,
Pavg =
T ∫ ⎜ ⎟
t ⎝ k =1 k =1 ⎠

and expanding yields

t +T ⎛ ∞
1
Pavg =
T ∫ ⎜⎜ ∑ Vk I k sin(kω o t + φ k ) • sin(kω o t + θ k )
t ⎝ k =1

Page 3 of 12
 Chapter 3. Definitions Grady
June 2005
∞ ∞ ⎞
+ ∑ ∑ Vm I n sin(mω o t + φ m ) sin(nω o t + θ n ) ⎟⎟dt ,
m =1, n =1,m ≠ n ⎠

t +T ⎛ ∞
1 ⎛ cos(φ k − θ k ) − cos(2kω o t + φ k + θ k ) ⎞
Pavg =
T ∫ ⎜⎜ ∑ Vk I k ⎜⎝ 2
⎟
⎠
t ⎝ k =1
∞ ∞
⎛ cos((m − n)ω o t + θ m − θ n ) cos((m + n)ω o t + θ m + θ n ) ⎞ ⎞⎟
+ ∑ ∑ ImIn⎜
⎝ 2
+
2
⎟ dt
⎠ ⎟⎠
m =1, n =1,m ≠ n

As observed in the rms case, the average value of all the sinusoidal terms is zero, leaving only
the time invariant terms in the summation, or

∞ V I ∞
Pavg = ∑ k k
2
cos(φ k − θ k ) = ∑ Vk ,rms • I k ,rms • dpf k = P1,avg + P2,avg + P3,avg + L , (3.8)
k =1 k =1

where dpf k is the displacement power factor for harmonic k .

The harmonic power terms P2, avg , P3, avg , L are mostly losses and are usually small in relation
to total power. However, harmonic losses may be a substantial part of total losses.

Equation (3.8) is important in explaining who is responsible for harmonic power. Electric utility
generating plants produce sinusoidal terminal voltages. According to (3.8), if there is no
harmonic voltage at the terminals of a generator, then the generator produces no harmonic
power. However, due to nonlinear loads, harmonic power does indeed exist in power systems
and causes additional losses. Thus, it is accurate to say that

• Harmonic power is parasitic and is due to nonlinear equipment and loads.

• The source of most harmonic power is power electronic loads.

• By chopping the 60 Hz current waveform and producing harmonic voltages and currents,
power electronic loads convert some of the “60 Hz” power into harmonic power, which
in turn propagates back into the power system, increasing system losses and impacting
sensitive loads.

For a thought provoking question related to harmonic power, consider the case shown in Figure
3.1 where a perfect 120Vac(rms) power system with 1Ω internal resistance supplies a triac-
controlled 1000W incandescent lamp. Let the firing angle is 90°, so the lamp is operating at
half-power.

Page 4 of 12
 Chapter 3. Definitions Grady
June 2005

Current i

1Ω
i
+ +

vs vm Customer 14.4Ω

– –
Figure 3.1. Single-Phase Circuit with Triac and Lamp

120 2
Assuming that the resistance of the lamp is = 14.4Ω, and that the voltage source is
1000
v s (t ) = 120 2 sin(ω o t ) , the Fourier series of current in the circuit, truncated at the 5th harmonic,
is
i (t ) = 6.99 sin(ω o t − 32.5 o ) + 3.75 sin(3ω o t − 90.0 o ) + 1.25 sin(5ω o t − 90.0 o ) .

If a wattmeter is placed immediately to the left of the triac, the metered voltage is

v m (t ) = v s (t ) − iR = 120 2 sin(ω o t )
(
− 1 • 6.99 sin(ω o t − 32.5 o ) + 3.75 sin(3ω o t − 90.0 o ) + 1.25 sin(5ω o t − 90.0 o ) )
= 163.8 sin(ω o t + 1.3o ) + 3.75 sin(3ω o t + 90.0 o ) + 1.25 sin(5ω o t + 90.0 o )

and the average power flowing into the triac-lamp customer is

Pavg =
163.8 • 6.99
2
(
cos 1.3o − (−32.5 o ) + )
3.75 • 3.75
2
(
cos 90.0 o − (−90.0 o ) )
+
1.25 • 1.25
2
(
cos 90.0 o − (−90.0 o ) )
= 475.7 – 7.03 – 0.78 = 467.9W

Page 5 of 12
 Chapter 3. Definitions Grady
June 2005
The first term, 475.7W, is due to the fundamental component of voltage and current. The 7.03W
and 0.78W terms are due to the 3rd and 5th harmonics, respectively, and flow back into the power
system.

The question now is: should the wattmeter register only the fundamental power, i.e., 475.7W, or
should the wattmeter credit the harmonic power flowing back into the power system and register
only 475.7 – 7.81 = 467.9W? Remember that the harmonic power produced by the load is
consumed by the power system resistance.

3.4. True Power Factor

To examine the impact of harmonics on power factor, it is important to consider the true power
factor, which is defined as

Pavg
pf true = . (3.9)
Vrms I rms

In sinusoidal situations, (3.9) reduces to the familiar displacement power factor

V1 I1
P1, avg cos(δ 1 − θ1 )
dpf1 = = 2 = cos(δ 1 − θ1 ) .
V1, rms I1, rms V1 I1
2

When harmonics are present, (3.9) can be expanded as

P1, avg + P2, avg + P3, avg + L

pf true = .
V1, rms 1 + THDV2 • I1, rms 1 + THD I2

In most instances, the harmonic powers are small compared to the fundamental power, and the
voltage distortion is less than 10%. Thus, the following important simplification is usually valid:

P1, avg dpf1

pf true ≈ = . (3.10)
V1, rms I1, rms 1 + THD I2 1 + THD I2

It is obvious in (3.10) that the true power factor of a nonlinear load is limited by its THD I . For
example, the true power factor of a PC with THD I = 100% can never exceed 0.707, no matter
how good its displacement power is. Some other examples of “maximum” true power factor
(i.e., maximum implies that the displacement power factor is unity) are given below in Table 3.1.

Page 6 of 12
 Chapter 3. Definitions Grady
June 2005
Table 3.1. Maximum True Power Factor of a Nonlinear Load.

Current Maximum
THD pf true
20% 0.98
50% 0.89
100% 0.71

3.5. K Factor

Losses in transformers increase when harmonics are present because

1. harmonic currents increase the rms current beyond what is needed to provide load power,

2. harmonic currents do not flow uniformly throughout the cross sectional area of a
conductor and thereby increase its equivalent resistance.

Dry-type transformers are especially sensitive to harmonics. The K factor was developed to
provide a convenient measure for rating the capability of transformers, especially dry types, to
serve distorting loads without overheating. The K factor formula is

∞
∑ k 2 I k2
K = k =1 . (3.11)
∞
∑ I k2
k =1

In most situations, K ≤ 10 .

3.6. Phase Shift

There are two types of phase shifts pertinent to harmonics. The first is a shift in time, e.g. the
2T
± among the phases of balanced a-b-c currents. If the PC waveform in Figure 3.2 is delayed
3
by ΔT seconds, the modified current is

∞ ∞
i (t − ΔT ) = ∑ I k sin(kω o (t − ΔT ) + θ k ) = ∑ I k sin(kω o t − kω o ΔT + θ k )
k =1 k =1
∞ ∞
= ∑ I k sin (kω o t + (θ k − kω o ΔT )) = ∑ I k sin(kω o t + θ k − kθ o ) , (3.12)
k =1 k =1

where θ o is the phase lag of the fundamental current corresponding to ΔT . The last term in
(3.12) shows that individual harmonics are delayed by kθ o .

Page 7 of 12
 Chapter 3. Definitions Grady
June 2005

5
delayed

Amperes
0

-5

Figure 3.2. PC Current Waveform Delayed in Time

The second type of phase shift is in harmonic angle, which occurs in wye-delta transformers.
Wye-delta transformers shift voltages and currents by ± 30 o . ANSI standards require that,
regardless of which side is delta or wye, the a-b-c phases must be marked so that the high-
voltage side voltages and currents lead those on the low-voltage side by 30 o for positive-
sequence, and lag by 30 o for negative sequence. Zero sequences are blocked by the three-wire
connection so that their phase shift is not meaningful.

3.7. Phase Sequence

In a balanced three-phase power system, the currents in phases a-b-c are shifted in time by
± 120 o of fundamental. Therefore, since

∞
ia (t ) = ∑ I k sin(kω o t + θ k ) ,
k =1

2π
then the currents in phases b and c lag and lead by radians, respectively. Thus
3
∞
2π
ib (t ) = ∑ I k sin(kω o t + θ k − k 3
),
k =1
∞
2π
ic (t ) = ∑ I k sin(kω o t + θ k + k 3
).
k =1

Picking out the first three harmonics shows an important pattern. Expanding the above series,

ia (t ) = I1 sin(1ω o t + θ1 ) + I 2 sin(2ω o t + θ 2 ) + I 3 sin(3ω o t + θ 3 ) ,

Page 8 of 12
 Chapter 3. Definitions Grady
June 2005
2π 4π 6π
ib (t ) = I1 sin(1ω o t + θ1 − ) + I 2 sin(2ω o t + θ 2 − ) + I 3 sin(3ω o t + θ 3 − ) , or
3 3 3

2π 2π
= I1 sin(1ω o t + θ1 − ) + I 2 sin(2ω o t + θ 2 + ) + I 3 sin(3ω o t + θ 3 − 0) .
3 3

Note that the first harmonic of ib (t ) is positive sequence because the currents are balanced. The
second harmonic of ib (t ) leads the first harmonic of ia (t ) by 120 o and is therefore negative
sequence. The third harmonic of ib (t ) is in phase with the first harmonic of ia (t ) and is
therefore zero sequence. Now, for phase c,

2π 4π 6π
ic (t ) = I1 sin(1ω o t + θ1 + ) + I 2 sin(2ω o t + θ 2 + ) + I 3 sin(3ω o t + θ 3 + ) , or
3 3 3

2π 2π
= I1 sin(1ω o t + θ1 + ) + I 2 sin(2ω o t + θ 2 − ) + I 3 sin(3ω o t + θ 3 − 0) .
3 3

Note that the first harmonic of ic (t ) is negative sequence because the currents are balanced. The
second harmonic of ic (t ) lags the first harmonic of ia (t ) by 120 o and is therefore positive
sequence. The third harmonic of ic (t ) is in phase with the first harmonic of ia (t ) and is
therefore zero sequence. The phase sequence pattern developed is shown in Table 3.2.

Table 3.2. Phase Sequence of Harmonics in a Balanced Three-Phase System

Phase
Harmonic Sequence
1 +
2 –
3 0
4 +
5 –
6 0
… …

If a system is not balanced, then each harmonic can have positive, negative, and zero sequence
components. However, in most cases, the pattern in Table 3.2 can be assumed to be valid.

Since “triplen” harmonics (i.e., multiples of three) in a balanced system are zero sequence, they
cannot flow in three-wire systems or loads. Thus, a delta - grounded wye transformer at the
entrance of an industrial customer effectively blocks the flow of triple harmonic load currents
into the power system, provided that the load is reasonably balanced. Unfortunately, the
transformer does nothing to block the flow of non-triplen harmonics.

Page 9 of 12
 Chapter 3. Definitions Grady
June 2005
Only the triplen harmonic currents are affected by grounding paths. The other harmonics are
either positive or negative sequence, which sum to zero at neutral points and do not flow in
neutral wires or ground connections.

Another interesting observation can be made concerning Table 3.2. Line-to-line voltages never
have zero sequence components because, according to Kirchhoff’s voltage law, they always sum
to zero. For that reason, line-to-line voltages in commercial buildings are missing the 3rd
harmonic that dominates line-to-neutral voltage waveforms.

3.8. Transformers

Consider the example shown in Figure 3.3 where twin, idealized six-pulse current source ASDs
are served by parallel transformers. The top transformer is wye-wye or delta-delta, thus having
no phase shift. The bottom transformer is wye-delta or delta-wye, thus having 30 o phase shift.
To begin the analysis, assume that the per-unit load-side current of the top transformer is the
standard six-pulse wave

THD I = 30.0%

ΔΔ
Six-Pulse
or ASD
YY
THD I = 14.2%

Net Twelve-Pulse
Σ line-side load-side

Operation ΔY
Six-Pulse
or ASD
YΔ

THD I = 30.0% THD I = 30.0%

Figure 3.3. Current Waveforms of Identical Parallel Six-Pulse Converters

Yield a Net Twelve-Pulse Converter

Page 10 of 12
 Chapter 3. Definitions Grady
June 2005

itop, loadside (t ) = I1 sin(1ω o t )

I I
+ 1 sin(5ω o t + 180 o ) + 1 sin(7ω o t + 180 o )
5 7
I1 I1
+ sin(11ω o t ) + sin(13ω o t )
11 13
I1 I
+ sin(17ω o t + 180 o ) + 1 sin(19ω o t + 180 o ) + L
17 19

Note that the even-ordered harmonics are missing because of half-wave symmetry, and that the
triple harmonics are missing because a six-pulse ASD is a three-wire balanced load, having
characteristic harmonics k = 6n ± 1, n = 1,2,3,... . Because the transformer has no phase shift,
then the line-side current waveform (in per-unit) is the same as the load-side current, or

itop , lineside (t ) = itop , loadside (t ) .

Now, because the fundamental voltage on the load-side of the bottom transformer is delayed by
30 o , then each harmonic of the load-side current of the bottom transformer will be delayed by
k • 30 o , so that

ibottom, loadside (t ) = I1 sin(1ω o t − 30 o )

I I
+ 1 sin(5ω o t + 180 o − 150 o ) + 1 sin(7ω o t + 180 o − 210 o )
5 7
I I
+ 1 sin(11ω o t − 330 o ) + 1 sin(13ω o t − 390 o )
11 13
I I
+ 1 sin(17ω o t + 180 o − 510 o ) + 1 sin(19ω o t + 180 o − 570 o ) + L
17 19

The current waveform through the top transformer is not shifted when going from load-side to
line-side, except for its magnitude. However, the various phase sequence components of the
current through the bottom transformer are shifted when going to the line-side, so that

ibottom, lineside (t ) = I1 sin(1ω o t − 30 o + 30 o )

I I
+ 1 sin(5ω o t + 180 o − 150 o − 30 o ) + 1 sin(7ω o t + 180 o − 210 o + 30 o )
5 7
I1 I
+ sin(11ω o t − 330 o − 30 o ) + 1 sin(13ω o t − 390 o + 30 o )
11 13
I1 I
+ sin(17ω o t + 180 o − 510 o − 30 o ) + 1 sin(19ω o t + 180 o − 570 o + 30 o ) + L .
17 19

Page 11 of 12
 Chapter 3. Definitions Grady
June 2005
Combining angles yields

ibottom, lineside (t ) = I1 sin(1ω o t )

I I
+ 1 sin(5ω o t ) + 1 sin(7ω o t )
5 7
I I
+ 1 sin(11ω o t ) + 1 sin(13ω o t )
11 13
I I
+ 1 sin(17ω o t + 180 o ) + 1 sin(19ω o t ) + L .
17 19

Adding the top and bottom line-side currents yields

2 I1 2I
inet (t ) = itop , lineside (t ) + ibottom, lineside (t ) = 2 I1 sin(1ω o t ) + sin(11ω o t ) + 1 sin(13ω o t ) + L .
11 13

The important observation here is that harmonics 5,7,17,19 combine to zero at the summing
point on the line-side. Recognizing the pattern shows that the remaining harmonics are

k = 12n ± 1, n = 1,2,3,... ,

which leads to the classification “twelve-pulse converter.”

In an actual twelve-pulse ASD, a three winding transformer is used, having one winding on the
line-side, and two parallel wye-delta and delta-delta windings on the load-side. The power
electronics are in effect divided into two halves so that each half carries one-half of the load
power.

Summarizing, since harmonics in a balanced system fall into the predictable phase sequences
shown in Table 3.2, it is clear that a wye-delta transformer will advance some harmonics by 30 o
and delay other harmonics by 30 o . This property makes it possible to cancel half of the
harmonics produced by ASDs (most importantly the 5th and 7th) through a principle known as
phase cancellation. The result is illustrated in Figure 3.3, where two parallel six-pulse converters
combine to yield a net twelve-pulse converter with much less current distortion. Corresponding
spectra are given in the Appendix.

Transformer phase shifting may be used to create net 18-pulse, 24-pulse, and higher-pulse
converters.

Page 12 of 12