Power Factor Requirements for Electronic Loads in California

Brian Fortenbery, Electric Power Research Institute

ABSTRACT

Nonlinear loads, which are normally powered by ac-dc switching power supplies, are
typical in a wide range of electronic devices, such as computers, monitors, televisions, printers,
fax machines, copiers, audio equipment, and telecommunications equipment. Residential and
commercial miscellaneous loads, which include consumer electronics, are the fastest growing
segment of household energy use in the United States.
Power factor is a measure of the efficiency with which a load uses the current supplied to
it. Typical power factors are shown in Figure 6.The nonlinear loads described above have a poor
power factor, and thus draw more current than required for the DC load on their output. This is
because of harmonics that are inherent in the power supply current, due to their design. However,
this can be corrected using power factor correction (PFC) circuitry at the front end of the power
supply – an approach required in Europe and Japan for high wattage devices. In so doing, the
current flowing in the building wiring is reduced, and this in turn can reduce heating effects in
the wiring from that current. While efforts are underway to improve power supply efficiency
through standardized test methods, voluntary labeling programs, and mandatory efficiency
standards, parallel discussions are addressing whether or not to require PFC in the high
efficiency designs. Without any harmonic compensation, the highly distorted load currents of
computer workstations, such as those shown in Figure 1, as well as other electronic loads, can
lead to losses in the building wiring that are much higher than for undistorted load current.
This paper describes a project designed to definitively determine the energy savings
potential in building distribution wiring from power factor correction in various end use
electronic products, and recommend appropriate Title 20 requirements for securing those savings
cost effectively. In addition, there are other, non-energy benefits associated with PFC (reduction
of harmonics) that are of interest. The final conclusion shows that the opportunities for energy
savings are not as large as previously thought, but are still compelling.

Introduction
There are three major types of electric loads in commercial and residential buildings—
resistive, reactive, and nonlinear. Resistive loads are those from incandescent lamps and electric
resistance heaters. Reactive loads are typically inductive, such as electric motors that might be
found in pumps and compressors. These loads draw current that is out of phase with the source
voltage, which reduces the amount of active power transferred to the load. Nonlinear loads,
which are normally powered by ac-dc switching power supplies, are typical in a wide range of
electronic devices, such as computers, monitors, televisions, printers, fax machines, copiers,
audio equipment, and telecommunications equipment. As interest in cost effective opportunities
to reduce electrical consumption has grown, attention is increasingly turning to these nonlinear
loads, which include most electronic products previously classified as “miscellaneous load” or
“plug load.” Residential and commercial miscellaneous loads, which include consumer
electronics, are the fastest growing segment of household energy use in the United States. While
the relative energy intensity of applications such as heating and cooling is declining, the DOE’s

©2014 ACEEE Summer Study on Energy Efficiency in Buildings 9-160

 Annual Energy Outlook forecasts that the intensity of residential and commercial miscellaneous
end-uses, as defined in the 2007 Annual Outlook, will increase significantly by 2030.
The following is an overview of harmonics in building power systems, to establish the
causes and effects. After this, the energy savings potential will be explored.
Most harmonic currents in commmercial systems are caused by nonlinear loads, that is,
loads that draw currents whose frequencies differ from the frequency of the source. Many
electronic devices are nonlinear loads because they use solid state rectifiers at their inputs and
filter capacitors after the rectifiers. An example of this type of circuit is a personal computer
power supply, as shown in Figure 1.
1-phase
rectifier

Switch
Mode DC
Power Load
Supply

Figure 1. PC power supply circuit diagram.

Solid state rectifiers inherently draw current in pulses, when the AC line voltage is higher
than the voltage across the filter capacitor used with the rectifier. This pulsed current is very rich
in harmonics, as seen in Figure 2. The harmonic spectrum plot shows the presence of odd
harmonics, with relatively large magnitudes at the lower frequencies. As the frequency increases,
the magnitudes decrease.
200 4

150 3
Capacitor 120

100 Voltage 2
100

50 1
% of Fundamental

80
0 0
Current
60
-50 -1
40
-100 Line -2
Voltage 20
-150 -3
0
-200 -4 1 3 5 7 9 11 13 15 17 19 21 23 25
0 90 180 270 360
Harmonic Order
Degrees - 60 Hz

Figure 0. PC power supply current and spectrum.

The PC power supply is only one of many possible harmonic producers in the 120-volt
range. Other loads that inject harmonic currents include office equipment for communications,

©2014 ACEEE Summer Study on Energy Efficiency in Buildings 9-161

 printing and copying and lighting with high efficiency electronic ballasts. Nonlinear loads in the
480 V range include adjustable-speed drives (ASDs) for HVAC, larger computers,
uninterruptible power supplies, and 277-V lighting.

Problems Related to Harmonics

Harmonics result in an increase in the rms value of current, which increases line losses
(heating in wiring, given by I2R, where I is the rms current, and R is the resistance in the wire).
The rms value of the current is given by


I
2 2 2 2
I h  I1  I 2  I 3  
h 1

Where I is the total rms current,

h is the harmonic number,
I1, I2, etc, are the current components at the first and second harmonic, etc
Ih is the current component at the hth harmonic

Since the rms value of the current is higher when THD is higher, the heating effect
associated with I2R losses is higher than it would otherwise be. (I2R is the value of the power
consumed in the wiring, where I is the current and R is the resistance in the wire). These losses
manifest themselves in a number of specific ways:

 Heating effects in phase conductors as well as neutrals

 Heating in 3-phase, dry-type transformers
 Harmonic voltage related heating in other equipment, such as motors
 Harmonic voltage stress on system capacitors and equipment capacitors
 Resonance with power-factor-correction capacitors
 Voltage distortion at the point of common coupling

Wiring I2R losses will vary, depending on the length of the line segments, the amount of
load traveling through them, and the nature of the load. The distorted current with low power
factor leads to relatively higher losses-per-watt of connected load. Without any harmonic
compensation, the highly distorted load currents of computer workstations, such as those shown
in Figure 1, as well as other electronic loads, can lead to losses in the building wiring that are
much higher than for undistorted load current. Also, the effectiveness of harmonic elimination
methods will be highly dependent on their locations in the building wiring. Usually, the best
location for harmonic filters is at the offending load. Power factor correction (PFC) circuits can
be added to the front end of nonlinear devices to eliminate their harmonic contribution to the
power system.
Transformer losses are well defined, with derating for harmonics covered in ANSI
C57.110. Losses are divided between load and no-load losses. The load losses include I2R losses
and stray losses. It is the stray losses that are most affected by the harmonic content of the
current waveform. There are eddy-current losses that cause heating in many transformer parts,

©2014 ACEEE Summer Study on Energy Efficiency in Buildings 9-162

 but it is the heating in the windings that is of most interest. This heating is proportional to the
square of the load current and the square of the frequency, which leads to the notion of a K factor
promoted by some transformer manufacturers. The K-factor serves as an indication of the
additional eddy current heating in the winding. This K factor is given by

K
I h 2
h
2

I 2
h

Where h is the harmonic number,

Ih is the hth harmonic component of the current.

The derated transformer current is given by

1  PEC  R
I RMS 
1  K * PEC  R

Where PEC  R represents the rated eddy-current losses for that transformer design.

This K factor takes into account the additional eddy current heating in the windings due
to the harmonic components of the nonlinear current.

Power Factor Correction

One of the impacts of harmonic content in load current is an increase in the rms value of
the current, as shown before.
This results in additional heating of the building wiring and transformers. Since the harmonic
content also manifests itself in the power factor, power factor can be used as a measure of the
level of harmonics as well as a measure of the losses they cause in the power system. These
problems can be avoided by the manufacturer through the use of a power factor correction
circuit, which effectively removes the harmonics from the power supply current.
Previous work for the CEC showed that the use of PFC power supplies inherently reduces
the current in the building wiring and thus reduces the heating or I2R losses associated with the
harmonic-rich currents. It was of great interest to study the topic and develop an understanding
of what typical energy savings might be achieved in the building distribution wiring through
such an approach. In order to do this, a model was developed and lab tests performed to establish
some typical values of possible energy savings. In addition, there are other, non-energy benefits
associated with PFC (reduction of harmonics) that are of interest. Both the results of the tests and
calculations and the non-energy benefits of interest are outlined in the following sections.

Methodology: Laboratory Tests for Desktop Computers

Lab testing was performed to ensure that the theoretical effects of PFC on system wiring
losses were reasonable and achievable. In order to determine the energy savings due to PFC
loads, the following procedure was adopted. The test was conducted in an environment where the

©2014 ACEEE Summer Study on Energy Efficiency in Buildings 9-163

 source voltage could be controlled. A total of 24 computer power supplies were chosen for this
test—8 units without PFC, (referred to as NPFC), 8 PFC units and 8 80 PLUS units. The 80+
PLUS units are power supplies that achieve an efficiency over 80% at 20%, 50% and 100% load
and include power factor correction. Non-80 PLUS units have efficiencies between 60 and 80%.
Each of the 24 power supplies had identical 200-W output power ratings. All the power supplies
were loaded identically; that is, each power supply was loaded with constant resistors on the dc
side having total dc power output of 97 W. The 12-V bus was loaded to 72 W (2-ohm resistor)
and the 5-V bus was loaded to 25 W (1-ohm resistor). The tests were set up as shown in the
diagram below (Figure 4).

Figure 2. Wiring diagram for cable loss tests.

Measurements were taken first using 8 power supplies without PFC and then again using
8 PFC-equipped power supplies, then finally with 8 80 PLUS power supplies.

Test Results

Test results indicated that while the change in wiring losses is small, there are reductions
in the power consumed by the power supply itself.
The loads were carefully controlled to ensure that the output power was the same in all
cases. As the line current was reduced, the power factor increased, automatically reducing the
power consumed.

In a linear system,

P  V RMS I RMS cos 

Where  is the angle between the voltage and the current.

©2014 ACEEE Summer Study on Energy Efficiency in Buildings 9-164

 In systems with inductive loads, there is a displacement between the voltage and current
given by this angle  . The cos of this angle is known as the displacement power factor since it
depends only on the angle between voltage and current. There is a term known as true power
factor, which is equal to the ratio of real to apparent power.
If there is no distortion, true power factor is equal to displacement power factor. In the
presence of distortion, however, these two are no longer equal and must be studied further.

The above equation still holds true, although it is more correct to write it as follows:

P  V RMS I RMS PFTRUE

Where

PFDISP
PFTRUE 
2
1  THDI

However the single-phase power supplies of interest here are a special case in which the
angle between the voltage and current is 0, making the cosine equal to unity. Therefore, any
reduction in power factor is due to the presence of harmonic distortion, and so power factor is
typically directly related to harmonic distortion in power supplies.
One result is that the benefits, or savings, from using 80 PLUS power supplies can be
attributed to three effects:

1. Savings by virtue of more efficient power supply.

This is the original intent of using 80 PLUS power supplies, and is the most significant
effect.
2. Indirect savings in wiring due to lower current requirement of the more efficient
power supply.
This is a side effect of high efficiency loads, and while it is relatively small, it is still real
and measurable (and has usually been ignored in past analyses).
3. Indirect savings in wiring due to power factor correction.
This is a known and measurable effect of PFC loads, larger than the indirect savings from
higher efficiency and a significant added savings of energy.

The results of the testing and modeling are shown in Figure 4, where the savings from the
efficient power supply are normalized to 100%. This allows the magnitude of the other effects to
be compared. A major result from these findings was that EPA’s Energy Star program added a
power factor correction requirement to the desktop computer power supply efficiency
specification in 2007. This specification was Energy Star Computer Power Supply Specification
Version 4.

©2014 ACEEE Summer Study on Energy Efficiency in Buildings 9-165

 Savings Expressed as % of Direct Power Supply
160% Benefit from PFC
Cable Savings from Efficiency Change
140%
Direct PS Load Savings

120%
Savings from 80+

100%

80%

60%

40%

20%

0%
40 60 80 100 200
Cable Length (Feet)

Figure 4. Comparison of savings from PFC and efficiency improvement in percent.

These are conservative estimates, because they neglect other loading that might be on the
same cable, such as laser printers, copiers, and monitors. Because losses are proportional to the
square of the rms current, the incremental change in losses is twice the present loading level.

dPC dI 2 RCABLE
  2 IRCABLE
dI dI

For example, if the load were 10 A, and were reduced by 1 A, the difference in losses
(savings) would be

10 2 R  9 2 R  19R watts

But if the load were 25 A, and were reduced by 1 A, the savings would be

252 R  24 2 R  49R watts.

This finding is particularly noteworthy as the share of nonlinear loads continues to grow
in commercial buildings.

©2014 ACEEE Summer Study on Energy Efficiency in Buildings 9-166

 Current Project

Building on the findings of the previous CEC work, the current project was designed to
do a similar analysis for ALL electronic loads. The research question to be answered is whether
PFC requirements might garner enough energy savings to warrant Title 20 regulation.
Because the harmonic wiring losses depend on building wiring topology, it was necessary
to determine typical topologies for residential and commercial buildings. To narrow down the
variety in residential and commercial buildings, buildings with high annual energy use1 and well-
known loads were selected.
Floor space was used as the criteria to narrow down residential to three representative
sizes:

 Small 800-1000 square feet (typically 2 bedroom, 1 bath with dining space in kitchen)
 Medium 1500-2000 square feet (typically 3 bedroom, 2 bath with separated dining space)
 Large 2500-3000 square feet (typically 4 bedroom, 3 bath with formal dining space)

These categories were chosen because the National Electric Code (NEC) is driven by floor
space.
A sample of homes in California were chosen to physically measure the loads on
individual branch circuits. The primary reasons for these measurements were to validate the load
testing in the laboratory, and to determine some idea of typical coincidence of load, since it can
affect the savings achieved, as described above.
Lab tests revealed power factor levels for various electronic loads, which will be used in
the analysis. A sample of these values are shown in Figure 6.

1
Based on CA stock estimates and energy intensity per square foot

©2014 ACEEE Summer Study on Energy Efficiency in Buildings 9-167

 POWER FACTOR OF COMMON HOUSEHOLD ELECTRONICS

Element 41" Plasma TV

Playstation 3
PS3 Move
Sharp Aquos 3D TV
Microwave
Xbox 360
Xbox Kinect
Digital Picture Frame
Wii
Philips 52" Projection TV
Digital Picture Frame
Digital Picture Frame
Digital Picture Frame
Magnavox Projection TV
Dell Monitor
ViewSonic Monitor
Digital Picture Frame
Samsung 70" 3D Bluray
Magnavox Projection TV - standby

0.0 0.2 0.4 0.6 0.8 1.0

POWER FACTOR
Figure 6. Measured power factor of selected electronic loads.

In this scenario, the analysis is simplified by the fact that harmonic mitigation is the only
change. With the power supply work, the devices were more efficient which led to three energy
savings mechanisms. Here, the only means of achieving energy savings is through the
reduction/elimination of harmonic distortion, which reduces the RMS value of the current.
Therefore, to model the system, it is only necessary to use the current spectrum measured in the
lab, calculate the losses using the RMS value measured, applying the coincidence of other loads.
To compare, the same scenario is adjusted by removing the harmonic components of the
current. This is accomplished by simply using the fundamental component of the current as the
rms value. The energy savings is then calculated as the difference between the two scenarios.
Previous projections were that energy savings from power factor correction could be just
under 2% of all plug loads. This equates to approximately 1.4 Billion kWh per year for the state
of California. Stakeholder outreach efforts have shown strong support for inclusion of PFC
requirements for electronic loads in California. This project, however, through more rigorous
analysis, demonstrated that those projections were somewhat optimistic, and some adjustments
have been made to the projected energy savings.
Conclusions
Through the process of determining the savings potential of implementing PFC in
electronic devices across California, there were several interesting discoveries that are of note.
1. The amount of savings in the residential sector is much smaller than previously
estimated. As a result, power factor correction in residential appliances is not likely to be

©2014 ACEEE Summer Study on Energy Efficiency in Buildings 9-168

 a cost-effective option for legislative efforts, except for large, relatively power hungry
devices like televisions. However, commercial savings are more easily and cost
effectively realized, and should still be considered as a viable option for codes and
standards bodies such as Title 20.
2. Homeowners that participated in the field study were turning their devices (computers,
laptops, computer monitors) off when not in use rather than leaving them on to idle for
significant periods of time. Previously, estimates to determine energy consumption in
residential applications of office equipment assumed a device would be turned on, used
for a period of time, and then left on to sit idle while not it use. Data collected in the field
shows that in the twenty-two homes surveyed, the computers and laptops were actually
turned off when they were not in use, and the duty cycle was between 20% and 40%.
3. Since the savings potential of implementing PFC increases with higher currents, and
kitchens typically house the appliances with some of the highest currents (coffee makers,
toasters, microwaves, etc.), it was expected that the kitchen could provide a reduction in
energy losses from PFC. However, kitchens do not present much of a savings opportunity
in most cases. This is due to almost every appliance in a kitchen being either power factor
corrected, a resistive load, or a motor load (such as a refrigerator). Resistive loads
(toaster, coffee maker, etc.) already have a PF of 1 and do not draw a current with
harmonic content so there is no room for improvement in these devices. Motor loads,
such as refrigerators or dishwashers, may have a PF less than 0.9, but it is due to
displacement and not distortion. Refrigerators are primarily motor loads with the fan
motors and a compressor motor, and motors do not distort the current, but rather shift it
out of phase with the voltage causing displacement. This type of low power factor cannot
be corrected with the removal of distortion, which is the subject of this report.
Significance

The research contained in this report can be used to inform the Title 20 rulemaking
process, and increase the overall energy efficiency of California’s electronic devices. Currently,
improving power factor in electronic devices is a largely untapped avenue for significant energy
savings. In the United States, California is the only state to regulate PF at all, and that presently
only affects televisions of 100 W or higher. This work shows that there is a much greater
opportunity than televisions alone.
Title 20 Impacts

EPRI measured the power factor of a number of home electronics and appliances to
determine the potential impact of statewide minimum power factor standards.
With the exception of state level regulations on television power factor, no other state or
federal power factor standards apply to plug loads. In California, televisions must be power
factor corrected to 0.9 when they exceed 100 W during operation. Some voluntary specifications,
such as the 80 Plus program and Energy Star for desktop power supplies, require specific power
factor performance levels for plug loads. Many products that would benefit from improved
power factor are also already subject to regulation related to other aspects of device performance.
The relative level of effort required to implement power factor correction policy was
rated as high due to the diversity of products covered under potential regulations. Manufacturers
from dissimilar product classes would have to be engaged and the technical implementation of

©2014 ACEEE Summer Study on Energy Efficiency in Buildings 9-169

 power factor correction equipment evaluated. Energy savings were seen as low, but cost
effective due to low per product upgrade costs. Existing state and federal device regulations may
cause preemption issues and should be further studied.
Considering the above factors, the EPRI research team developed a recommended
specification that could inform Title 20 policy that would require all plug-loads larger than 50
watts to have power factors of at least 0.9 at 50 and 100 percent load. With this policy, an
estimated energy savings of 241 GWh per year after full stock turnover could be achieved.
There is a research need to investigate the potential savings from PFC further, so that
energy savings projections can be even more exact. In order to further refine these findings, it
will be necessary to perform even more detailed load monitoring on a larger sample of both
residential and commercial spaces, increasing the granularity of the data.

©2014 ACEEE Summer Study on Energy Efficiency in Buildings 9-170