Impact of electric vehicle loads on electric power distribution systems


Material Information

Impact of electric vehicle loads on electric power distribution systems
Physical Description:
vii, 100 leaves : ill. ; 29 cm.
Buchh, Tariq Aslam
Publication Date:


Subjects / Keywords:
Electric vehicles   ( lcsh )
Electric power distribution   ( lcsh )
Electric power consumption -- Forecasting   ( lcsh )
Electrical and Computer Engineering thesis, Ph. D   ( lcsh )
Dissertations, Academic -- Electrical and Computer Engineering -- UF   ( lcsh )
bibliography   ( marcgt )
non-fiction   ( marcgt )


Thesis (Ph. D.)--University of Florida, 1998.
Includes bibliographical references (leaves 92-99).
Statement of Responsibility:
by Tariq Aslam Buchh.
General Note:
General Note:

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 029532204
oclc - 41284352
System ID:

This item is only available as the following downloads:

Full Text







In the Name of God, the Most Beneficent, the Most Merciful.

All the praises and thanks be to God, the Cherisher and Sustainer of the Worlds.
The Most Beneficent, the Most Merciful.
The Only Master of the Day of Recompense.
You (Alone) we worship and You (Alone) we ask for help.
Guide us to the Straight Way.
The way of those on whom You have bestowed Your Grace and not of those who
earned your anger, nor of those who went astray.

Translation of Opening Chapter of the Holy Ouran


I acknowledge the conviction of those who believe in the Truth and have the

courage to stand for it. Having said that, I would like to mention the names of those who

have been there when I needed them. A word of appreciation for my adviser Dr. Alex

Domijan Jr. for agreeing to chair my advisory committee and to Dr. Dennis Carroll, Dr.

Khai D Ngo, Dr. Herman Lam and Dr. Barney Capehart for being on the committee.

My mother, who has been a constant source of inspiration and drive, my father,

who is an embodiment of patience, my wife, Falaknaz, who sacrificed a part of her

dreams and aspirations and is taking excellent care of our kids, Shaima and Ziad, my

loving brothers, Mehmood and Khalid and my best friend, Ashfaq. There has also been a

huge treasure of friends and well-wishers who were there to support and pray for me

whenever the going got tough. A special word of thanks for Dr. Song for his constructive

criticism and suggestions.

Florida Power and Light Company, provided the funding and technical support

and I appreciate it.


ACKNOWLEDGEMENTS.. ...............................................................iii

ABSTRACT ............................ ..........................


1 INTRODUCTION ................................... ..................... ..... .......I
1.1. Background................................................. 1
1.2. Project Justification .................................................. 3
1.3. Objectives and Goals..................................... ....... ... .................. 4
1.4. M ajor Contributions ...................................................................... 4
1.4. Research Phases....................................................... 5
2 LITERATURE SURVEY............................... ...... ............................. 7

3 POWER QUALITY AND HARMONIC DISTORTION ..................................... 10
3.1. Harm onics........................................................... 11
3.2. Estimation of Harmonic Distortion............................. ...................... 12
3.2.1. C rest Factor ..................................................... ...................... 12
3.2.2. Percentage Total Harmonic Distortion (THD).................. 13
3.2.3. The K-Factor ............................................. 13
3.3. Transform er Derating .................................................................... 14
3.3.1. Transformer Capability Equivalent Calculation ...................... 15
3.3.2. Effect of Current Harmonics on the Transformer Neutral Current ....... 17
4.1. Introduction. .................................... ................ ......... .................... 19
4.2. General Considerations In Data Acquisition And Processing...................... 20
4.2.1. Data Collection...................................................... 21
4.2.2. Data Recording........................................................ 22
4.2.3. Data Preparation..................................................... 22
4.2.4. D ata Q ualification ............................................... .............. 23
4.2. Data Analysis..................................................... 24
4.3. Field Data M monitoring ............................................................. 26
4.3.1. Data Acquisition Methodology...................... ....... 26
4.3.2. Data M monitoring Instrument......................... .......... ... .................... 26
4.3.3. Data M management ........................................................... 27


5 DATA REPRESENTATION AND ANALYSIS..................... ....... 30
5.1. Introduction ............................................................ 30
5.2. Data Trends and Characteristics.......................... ........... ...................... 32
5.3. Data Observation ............................................... .................. 33
5.4. Load Characteristics ......................................................... 34
6 LABORATORY TESTING OF THE EV CHARGER........................................... 47
6.1. Testing Procedure and Results .............................................. ....... 47
6.2. Observation of Laboratory Results................................... 48
7 COMPUTER SIMULATION.................................................. 61
7.1. Component Models............................... .......................... 61
7.1.1. System M odel........................................................... ..................... 6 1
7.1.2. Electric Vehicle Charger And Battery Models .................................. 64
7.1.3. Transformer Model .......................................... 64
7.1.4. Domestic Load M odel.................................. .............. 66
7.2. Mathematical Model Of The EV Charger....................... 66
6.3. Simulation Results................................. .......................68
7.4. Simulation of Short Charging Time Chargers....................................... 69
7.4.1. Simulation Methodology For Fast Chargers..................................... 70
8.1. The Field D ata ............................................................ ........... 74
8.2. Laboratory Test....................................... ............74
8.3. Simulation Results...................... ........ ........ .......... 75
8.4. Comparison.................................................. 75
9 DERATING OF DISTRIBUTION TRANSFORMERS..................................... 83
9.1 M ethod of Calculation................................................... 83
9.1.1 ANSI/IEEE C 57.110 M ethod............................... .................... 83
9.1.2 Modified ANSI/IEEE C 57.110 Method.................................. ..... 85
9.2. Transformer Derating with Fast Time Chargers................................... 85
9.3. Conclusion.......................... .... ........................ 86
10 CONCLUSIONS.................................................. 89
10.1 .Summary and Conclusions................... ................... 89
10.2. Scope For Further Research ................... .................... 90

REFERENCES................................ .. ..........92

BIOGRAPHICAL SKETCH................... ............... 100

Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy



Tariq Aslam Buchh

August 1998

Chairman: Dr. Alexander Domijan, Jr.
Major Department: Electrical and Computer Engineering

The voltages and currents on four distribution sites (FPL_1 to FPL_4) have been

monitored. The load behavior with and without the EV load has been recorded. By

performing laboratory tests on the GMC EVI charger charging a fully discharged battery,

the characteristics of the charger have been determined. The harmonic spectrum of the

transformer current charging a fully discharged battery has been determined. The snap

shots of the current and voltage waveforms have been recorded and studied. Based on the

laboratory data a computer model of the charger and the distribution system has been

developed. Comparing the simulation results to the laboratory results and the field data

the simulation results have been validated. The simulated currents are roughly 4 to 10%

away from the field recorded currents, in their magnitude and distortion. Based on the

field data, laboratory results and the simulation results the derating of the distribution

transformer has been calculated. The two methods suggested for the calculation of

derating have been presented and compared. The maximum and minimum limits of

derating have been suggested. Using the first method the maximum derating is 20% and

the minimum value is 99%. Using the modified method the derating remains close to 99

% throughout the charging cycle. (It must be noted here, that a derating to 20% means

that the transformer can safely handle only 20% of its full rated load. Likewise a derating

to 99% means that the transformer can handle 99% of its full rated capacity. Thus, a

transformer derated to 20% is highly derated and a transformer derated to 99% is slightly

derated.) After establishing the reliability of the simulation, by comparing the simulation

results with the field data and laboratory results, a charging scenario with shorter

charging times (45 minutes and 1.2 hours) has been simulated. The distribution

transformer derating corresponding to these short-charging times has been calculated and

presented. Furthermore, recommended scope for further research has been suggested

which among other things, involves validating the derating formula for distribution




1.1. Background

From the state of being a concept vehicle, a possible means of future

transportation and a dream, the Electric Vehicle (EV) has become a reality. But even after

the first major commercial launch of the EVI by General Motor Corporation, the future

of EVs is still under a cloud of uncertainty and some unanswered questions.

The popularity of EVs as a future mode of transport can be attributed partly to

exhausting reserves of natural energy resources and environmental considerations. These

and many other factors have boosted the research and development of the EV technology

and a possibility of it being accepted as the popular mode of ground transport in the not

so distant future. There has been a great boost to EV research because of government and

industrial funding. This research has been mainly aimed at improving the battery

technology, charging techniques, weight, comfort and reliability. For the

environmentalist, the EVs' noiseless operation compliments its pollution free

performance, and, for the customer, the EV is a good performance machine with the

downside of short battery life, long recharging time and high initial cost.

The EVI manufactured by General Motors Corporation is powered by a Delco Valve-

Regulated Lead Acid battery pack, which contains 26 12-volt modules, capable of

carrying 16.2-kilowatt hours of energy. The batteries can be recharged from 0 to a full

state of charge in approximately 3 hours using a 240-volt inductive Magne charge

charger and about 15 hours from a 120-volt portable charger. Inductive charging is

achieved without metal-to-metal contact. The alternating current powers a device that

produces a fluctuating magnetic field in a paddle that is inserted into a port at the front of

the car. This induces an alternating emf in a winding inside the car. This alternating emf

is then rectified and fed into the batteries. The inductive charging system is inherently

safe even under hazardous climatic conditions, since the windings are safely encased in a

plastic cover. Typically this system can handle power levels from 1.5 to 25 kW with

overall efficiency of 90% while the power transfer frequency is 40-350 kHz [1]. The

specifications of the EV, as published by General Motors [PrEView Drive program

literature], are summarized as follows:

1. 137 H.P. three-phase ac induction motor.

2. 16.8 kWh lead-acid battery.

3. Inductive coupled standard charger (220 volts, 30 Amps)

4. IGBT power inverter module.

5. Range: 90 miles highway, 70 miles EPA city.

6. Acceleration from 0 to 60 mph in 8.5 seconds.

7. Electronically regulated speed of 80 mph.

8. Charging time from 85% depth of discharge is 2 to 3 hours.

9. Curb weight is 2970 pounds.

While the popularity and widespread use of EVs on one hand will take ground

transportation into a new era, on the other hand it can put additional and unprecedented strains

on power supply and distribution systems.

For power systems, the EV loads pose twofold problems. These are:

1. Increase in power demand

2. Decrease in power quality.

The power distribution system looks at the EV as an additional load of about 7kw/EV

(based on the GMC EV ), which was almost certainly not foreseen when most of the present

distribution systems were designed more than a few decades ago. The fact that the EV charger

load is a non-linear load with power electronic devices makes matters worse, for it is a source

of harmonics and distortion in the distribution system. These harmonics which get injected

into the system have a significant impact on the distribution system in terms of losses and

overall performance of the power system. The non-linearities and harmonics caused by EV

chargers can adversely affect other sensitive loads that may be connected to the distribution

system. The power quality can become a major concern if the battery and charger technology

is developed such as to put an extra strain on the distribution system. For example, research is

now being aimed at developing batteries and their chargers which would make it possible to

charge a completely exhausted battery in a very short period of time (15 minutes).

1.2. Project Justification

Existing distribution systems may not have the capability to cater to the new

needs of customers if the use of EVs becomes widespread. The electric power utility

companies need to be aware of and prepared to face such load scenarios. The utility

companies need to first have a reliable assessment of the possible future load scenarios

with EV loads. After this, they need to have a ready design of the new distribution system

equipment in terms of their sizes and ratings and also in terms of the configuration or the

layout of the entire distribution system. Thus, in case of an increase in the use of EVs by

the customers, the utilities will find themselves well prepared and thus continue to

reliably supply quality power to their customers.

1.3. Objectives and Goals

The aim of this project is to provide Florida Power and Light Company with an

assessment of the possible load effects with EV loads. The outline of the objectives of the

project is given below.

* To study the normal load behavior for a typical distribution system feeding two to ten

* To study the load behavior when one of the houses in the above case owns an Electric

* To perform laboratory tests to determine the behavior of the General Motor EVI
charger, in terms of its current requirements and harmonic effect on the input line.

* Based on the above data to prepare a computer model of the distribution system
feeding loads consisting of EV loads.

* Prepare the mathematical model for various components and the whole system.

* Perform basic statistical analysis of the recorded data.

* Use computer simulations to study various possible load scenarios.

1.4. Maior Contributions

Major contributions of this dissertation work are itemized below.

Actual field data was recorded from distribution transformer sites. This data
contains information about the load behavior and characteristics at these
particular sites. It has been for the first time, that such data has been made
available, for load scenario comparison between loads with and without the
EV loads.

Harmonic analysis has been performed on the field-recorded data, to identify
the signature that the EV load leaves on the system. This has provided with an
understanding of the EV load behavior when put on the system along with
other domestic loads.

Actual laboratory testing of the EV charger, charging into a fully discharged
battery has been carried out. For the first time, GMEVI, charger has been
tested when operated in isolation from any other external disturbance or
interference. The testing procedure has been developed and applied to find the
current, power, voltage, power factor and harmonic characteristics of the EV

Computer simulation of the distribution system feeding EV loads has been
carried out. The simulation results have been compared with the actual field
and laboratory results. Thus, establishing the accuracy of the simulation

Fast charging time EV chargers have been simulated and their expected
behavior has been studied. Although these chargers are not commercially
available, however, these simulation results have provided a basis for
understanding the expected behavior of such chargers once such a technology
becomes available.

Generalized load profile on a distribution transformer site has been developed.

An IEEE/ANSI method for the calculation of the derating of transformer
feeding non-sinusoidal loads has been modified. The proposed method yields
more accurate and realistic results.

1.4. Research Phases

The objectives of the thesis outlined in 1.3 have been achieved by dividing the

whole research work into various phases. These phases are outlined as follows and later

on in the section they are explained in more detail.

These phases are:


This phase consists of:

The recording and analysis of the field load data with and without the EV
being a part of the load.

The laboratory testing of the characteristics of the EV charger.


This phase consists of:

The physical model of the various components of the distribution system.


This phase consists of:

Computer simulation of the power distribution system and the analysis of
simulation results.
Create and study of different load scenarios with multiple EV loads.
Create and study load scenarios with EV loads of variable charging times.


This phase consists of:

Calculation of expected derating of a distribution system transformer
based on the simulation results for:
Normal charging time chargers.
Short charging time chargers. (Quick Chargers)


The issue of power quality is as old as the issue of electric power itself But due to

changing load conditions and customer needs, the importance and significance of supplying

high quality power has changed with time. In the last decade, due to the increase in the use of

power electronic equipment, power quialr, has attracted the attention of the utilities and

customers alike. The utilities look to supply quality power to its customers at a price

dependent on the amount of distortion that the customer is causing. The customers on the

other hand want a reliable, sag free power to be delivered to them at a highly competitive


The study and research in the field of electric power quality can be traced to 60's [1]

and 70's [2,3]. The fact though is that such studies were sparse. Power quality during that era

was just a limit applied to fluctuations of frequency and voltage, to the voltage unbalance,

voltage transients and flickers and power cuts [1,2]. Owing to the absence of vast amount of

non-linear loads, power quality and harmonics was a factor more of an academic importance

rather than a serious practical concern. The harmonics in the power systems were mostly

caused due to load switching and telephone interference. During the 80's with the first power

electronic revolution, the use of power electronic equipment was on an increase. Most of the

power electronic loads are switching loads and are highly non-linear; these loads thus cause a

lot of harmonics to be injected into the system. This prompted the academia to study the

power system harmonics in a greater detail, this included devising methods for the actual

measurement of system harmonics, making case studies and estimating their effect on

transformers, feeders and lines [4,5,6]. In the mid 80s [7] the increase in the presence of non-

linear power electronic loads was obvious and a further increase was anticipated. This increase

was owing to first the change in the technology e.g. faster and more efficient power electronic

devices, better converter topologies and converter control methodologies, second the advent of

fast microprocessors and their application to power electronic converter drives. The

microprocessor controlled drives were more efficient and reliable, had the flexibility of speed

control over a wide range, were easy to control and were compact in size. The need for

technical understanding of the nature of electric disturbances caused by these non-linear loads

was felt. Thus, in the face of changing load scenarios, there was a need for power quality

education that included definitions, grounding practices, power line dynamics, case experience

and trouble shooting [8]. In and after the late 80s a sharp increase in the published literature on

power quality and harmonics can be found. The number of these publications is so large that it

is impossible to list all of them, thus a selection of these papers has been listed here. Decrease

in power quality that not too long ago was a mere speculation, was now viewed as a serious

threat to the normal operation of power systems [10]. Research was thus directed towards

development of harmonic mitigation techniques [11] as well as development of power

electronic converters which would cause minimum harmonic distortion. Papers

recommending different methods for the estimation and monitoring of power system

harmonics were published [12,13,14]. References 15 25 include papers on the study and

characterization of specific non-linear loads and their effect on power systems. The studied

loads consist of fluorescent lighting loads, electric drive loads, railroad loads and some other

types of non-linear loads. In addition to these papers, scores of publications dealing with the

study of the effect of harmonics on various power system equipment can be found in the

literature [25 30].

The last few years have seen a phenomenal increase in the use of computers by the

power customers. The computer loads, which are highly non-linear [31], although, exist

significantly as commercial loads but the number of home computers is also increasing every

day. [32]. Power quality is no longer an issue that is linked with just reliability, efficiency,

voltage sags and flickers; today power quality is understood literally as the quality of the

actual current and voltage waveforms. The power customer today has become very sensitive

to the quality of power in terms of waveforms that the utility is providing. This sensitivity is

owing to the changing needs of the customers, for instance in some cases the equipment

powered from the line is highly sensitive to harmonics. Also, both commercial and domestic

customers are more aware of harmonics and the impact they have on the performance of

equipment and utility bill.

One of the very sources of harmonics in the power system is used to assess and

remedy the situation. Computer modeling and simulations are being used for the study of

various non-linear loads and the impact they have on power systems.[33-37]. In the midst of

the talk and concern about increasing disturbance and harmonics, yet another type of non-

linear load has appeared on the horizon. This is the electric vehicle load. The electric vehicle

load has attracted the attention of the researchers and the utilities alike.


Power quality, in simple words can be defined as the amount of distortion in the

system current and/or voltage waveforms. The power systems and equipment are designed for

pure sinusoidal current and voltage waveforms. Nowadays, widespread use of computers,

florescent lights, Adjustable Speed Drives and many other power electronic equipment by

customers cause a considerable amount of distortion of the voltage and current waveforms.

The distortion is caused due to the inherently non-linear characteristics of these types of loads.

In the presence of this increasing level of non-linear loads, the utility companies need to

continue catering to the customer needs, which is supplying quality power with a high degree

of reliability. This could necessitate new and stricter distortion guidelines for power

customers, as well as a variable rate structure being developed. Also, utility engineers need to

deal with the issue by analyzing the distortion and plan to control it by employing methods to

either minimize the distortion by using some mitigation techniques or derate the existing

system equipment to better cope with the steadily increasing non-linear load scenario.

Electrical energy consumption in the United States in the year 1992 was of the order of 600 *

109 kWh/year [38]. With the advances in power electronic converters, computers and many

other non-linear loads, more and more of the proportion of this power would be used by such

non-linear, distortion causing loads.

3.1. Harmonics

The non-linear characteristics of the loads cause the distortion of the current

and/or voltage waveforms in a power system. These current and voltage waveforms are

periodic in nature but they are not pure sinusoids. The distorted waveforms can be

expressed in the form of a Fourier series. Using the Fourier series expansion, a distorted

periodic wave can be expressed as the sum of an infinite number of sine waves, whose

frequencies are an integral multiple of the fundamental frequency.

The distorted periodic waveform f (t) can be expressed as:

f(t) = ao + ,[a,, cos(ncot) + b ni ...... i

ao is the dc offset of the waveform
a,, is the amplitude of the cos term
b,, is the amplitude of the sin term
n is the order of the harmonic

To study the effect of the distorted voltage and current waveforms on the power

system, the principle of superposition is used. This means that the effect of each

constituent (harmonic) of the distorted wave is studied separately and then the overall

effect is found by adding the effect of individual harmonics. The fundamental constituent

of the distorted waveform has a frequency of 60 Hz, the second harmonic has a frequency

of 120 Hz, the third 180 Hz, and so forth. The equipment used in the power system is

designed for a supply frequency of 60 Hz i.e. the fundamental frequency. The other

constituent waveforms (harmonics) certainly have an undesirable impact on the

equipment and the overall power system and its performance.

3.2. Estimation of Harmonic Distortion

Before assessing the qualitative and quantitative effect of harmonics on the power

system, some kind of estimate or quantitative expression of harmonic distortion needs to

be defined. Three methods of estimating harmonic load content are practiced [41]. The

Crest Factor and The Total Harmonic Distortion (THD) are the most common methods.

The third method is the K-Factor method.

3.2.1. Crest Factor

The Crest Factor gives a very simple estimate of the harmonic content of the

waveform. It is defined as the ratio of the peak of the wave to its Root Mean Square (RMS)

value. A perfect sine wave by definition will thus have a crest factor of 1.414.

Crest Factor = Peak Value of the Waveform
RMS Value of the Waveform

If the Crest Factor of the waveform (voltage or current) is other than 1.414, it

indicates some kind of distortion in the wave. This technique has a limitation, it does not

provide information about the constituent harmonic frequencies. Knowledge of the

harmonic frequencies and their respective amplitudes is important to study the effect of

wave distortion on the system.

3.2.2. Percentage Total Harmonic Distortion (THD)

Total Harmonic Distortion (THD) is defined as the ratio of the sum of the RMS

values of the harmonics to the RMS value of the fundamental.

THD = 2

The above expression shows that THD provides with a measure of the distortion

of a waveform. Like the Crest Factor, the THD is also limited in that it does not provide

any information about the frequencies and amplitudes of individual harmonics. While

computing the THD for a wave, the relative phases of the various harmonics are also not

considered. Nevertheless both Crest Factor and the THD are very useful to establish the

presence or absence of waveform distortion.

3.2.3. The K-Factor

The K-Factor while providing an estimate of distortion in a waveform takes into

account the harmonic frequencies and amplitudes. This fact makes it the most accurate

available method of estimation of non-linear load harmonics in a system. This is the method

used for the calculation of derating of dry-type power distribution transformers [42].

Reference 4 proposes an expansion of the K-Factor method. The author [41] proposes to use

the K-Factor as a weighted sum. Each type of load is considered separately and then all the K-

Factors are combined to create a composite harmonic fingerprint.

The K-Factor is defined as:

K = I/,(pu)2h'

3.3. Transformer Delating

Power transformers and distribution transformers are designed for sinusoidal

currents and voltages. Name plate ratings of transformers hold for purely sinusoidal

supply and loads. If the current gets non-sinusoidal, for instance because of the use of

switching power supplies, solid state inverters, converters, Adjustable Speed Drives etc.,

the indicated ratings of transformers no longer hold. A high harmonic content in the

output current waveform effects the transformer ratings. Load harmonic currents add a

disproportionate amount of additional transformer heating, particularly in the form of

winding eddy current losses and stray losses which are supply frequency dependent.

Several methods are currently used in the industry to assess the derating of transformers

for a harmonic application. The most widely used method is IEEE/ANSI C57.110. The

IEEE/ANSI method was developed for power transformers and does not specifically

address other applications such as distribution and dry type transformers. On the other

hand, the K factor method can be applied to only dry type transformers. In general, the

application of the IEEE/ANSI method provides conservative results when applied to a

distribution transformer. This standard outlines the following method for calculating the

transformer harmonic capability.

3.3.1. Transformer Capability Equivalent Calculation.

In this method a per unit value of non-sinusoidal current is calculated which will

result in load losses in the transformer just equal to the highest loss region for 60 Hz rated

operation. The winding eddy current loss under rated conditions at the point of maximum

loss density is 15% of the local 12R losses. Taking the fundamental component of current

as 1 per unit and taking the measured per unit values of the current harmonics up to the

SI 1h harmonic the derating of a transformer is calculated.

The maximum loss density of the local 12R losses is assumed as 15%. Solving

Equation 8 in ANSI/IEEE C 57.110 for the permissible current.

Following the procedure in the standard:


I + f x 0.15

Using the values shown in Table. I. and obtaining the amplitude of each harmonic

from Figure 4, the maximum permissible non-sinusoidal load current is given by:

113.089 0.501

Hence the ratio of transformer rating with and without the harmonic loads = .501

Table. I Harmonic Content of the Sample Waveform


(H) FH2 FH2H2

1 1 1 1

2 .16 .0256 .102

3 1.3 1.69 15.21

4 .267 .0712 1.139

5 1 1 25

6 .101 .010 0.36

7 .68 .462 22.638

9 .54 .291 23.57

11 .447 .199 24.07

Total 4.75 113.089

3.3.2. Effect of Current Harmonics on the Neutral Current.

The previous sections of this chapter discuss the effect of harmonics on the

distribution transformer. However there is some more concern for harmonic loads on multi-

grounded wye systems [43,44]. The current harmonics in a Y-Y distribution transformer

cause a high neutral current and an elevated emf of the neutral.

In an ideal case when the loads on the transformer are balanced, the neutral current

which is the vector sum of the three phase currents is zero.


I'A = Asincot
1B = Asin(t +2400)
Ic = Asin(cot+1200)

where A is the amplitude 'A 1, and Ic are the currents in phase A, B and C respectively

The neutral current is the vector sum of the three phase currents and is given by:


'N = Asin at + A sin(w t + 2400)+ A sin(w t + 1200)
.'. IN =0

Now if the load current is rich in the third harmonic, the third harmonic component

with amplitude T can be written as:

I'A = T sin(3a)
1B =T sin(3 t+ 2400)
Ic = Tsin(3ot +1200)

And as before


NI = T sin(3o t) + T sin(3co t + 2400)+ T sin(3o t +1200)
.'. I, =3Tsin(3wt)

The net neutral current because of all the other tripplen harmonics will yield similar

results. As reported by Mansoor et al. [44], in a power system, even with load diversity, there

is very little cancellation of the third harmonic. These high neutral currents cause the

following problem [45]:

1. The heavy neutral currents cause the over loading of feeder transformers. They
can cause further distortion of the voltage because they push the transformer into
saturation. The saturated transformer then becomes a source of harmonics.

2. When the neutral conductor carries harmonic currents, additional heat is generated
and the current carrying capacity of the feeder is reduced.

3. The neutral conductor for transformers is generally sized same as the phase
conductor, but due to the tripplen harmonic currents the neutral conductor current
can exceed its rated value thus overloading the neutral that can lead to equipment

4. The balanced tripplen currents flow as circulating currents in the transformer delta
primary winding. As a result, more current flows in the winding of the transformer
than is detected by the transformer primary circuit over current protection device.
This can result in the overloading of the transformer.

These high neutral harmonic currents cause a voltage differential between the neutral

and the ground since the ground wire at harmonic frequencies can cause significant neutral

conductor voltage drop. This is known as Common Mode Noise Voltage.


4.1. Introduction.

The effect of the EV loads on the power distribution system would depend on a

number of factors. Once the EVs become popular one of the factors that is going to be of

significance, is the usage pattern of the EV by its users. The usage pattern would

obviously vary from individual to individual, it would depend on the lifestyles and

specific needs of the customer, e.g. nature of his job, weather conditions, age of the user,

so on and so forth. To generalize the usage pattern and come up with an accurate

generalized model is surely a nontrivial problem. Based on field-recorded data and

knowledge of the EV load behavior, a typical load profile of a distribution node feeding

EV loads can be constructed. Even though this load model won't be accurate but it will

contain enough information to enable the utilities to design the distribution transformers,

so that they are able to supply power reliably and economically. The usage pattern of

users on the same distribution transformer shall determine the load diversity factor, the

expected peak load on the transformer and the overall derating of the transformer.

The objective of the field data monitoring thus is threefold. First to get the load

profile during two separate periods viz. when none of the customers are using an EV

and when one of the customer is using an EV. Second to try and identify the signature or

effect of the EV load when it is connected to the system. This effect is in terms of

increase in power demand and possible decrease in power quality. Lastly, this field

recorded data can form the basis for configuring system computer model used for

carrying out computer simulation study.

The objectives of field data monitoring have been achieved by monitoring the

field data (system voltages, currents, power factor etc.) first for a period of time (two

weeks) when the EV load is not a part of the loads on the distribution transformer. This is

followed by a two-week period then when one of the customers is using the EV. The data

obtained during the above two load scenarios has been analyzed. The comparison of these

load scenarios has been used to establish a generalized load scenario when one of the

customer on a distribution node is using an EV.

4.2. General Considerations In Data Acquisition And Processing

Appropriate techniques for the acquisition and processing of random data are

heavily dependent upon the physical phenomenon represented by the data and the desired

engineering goals of the processing. In broad terms, however, the required operations

may be divided into five primary categories as follows:

a. Data collection

b. Data recording (including transmission)

c. Data preparation

d. Data qualification

e. Data analysis

Each of these categories involves a number of sequential steps. The purpose of

this section is to summarize basic considerations associated with each of these key steps.

4.2.1. Data Collection

The primary element in data collection is the instrumentation transducer. In general

terms, a transducer is any device which translates power from one form to another. In an

engineering context, this usually means the translation of some measure of a physical

phenomenon of interest into an analog signal with a calibrated relationship between the

input and output quantities. This translation may involve up to three basic operations; (a)

a mechanical conversion of the physical quantity of interest into an intermediate

mechanical quantity, (b) a pick off step which converts the intermediate mechanical

quantity into an intermediate electrical quantity, and (c) an electrical conversion into a

final electrical quantity, usually voltage. Some transducers may combine any two or all

three of the operations, depending upon the physical quantity being measured and the

specific nature of the transducer.

For example, a thermocouple, which is widely used as a temperature transducer,

converts a difference in temperature directly to a difference in voltage without

intermediate steps. On the other hand, a resistance thermometer, another commonly used

temperature transducer, first converts temperature to a change in resistance, and then

converts the resistance change to an electrical voltage change.

Ideally, the above operations would be accomplished without distortion or

modification of the time history of the physical quantity being measured. In other words,

if the input time history is x(t) and the output time history is y(t), the perfect transducer

would provide an analog output, y(t) = c x(t), where c is a simple calibration constant.

Unfortunately, this ideal situation is difficult to achieve in practice. Gain and phase

modifications as well as distortion producing non-linearities are often inherent in the

transducer operations. This fact makes the transducer a potential source of error in any

data acquisition and processing program.

4.2.2. Data Recording

For some applications, it is possible to perform all desired data processing directly

on the transducer signals in real time. For most applications, however, this is not

practical and some form of storage (and perhaps remote transmission) of the transducer

signals will be required. The most desirable and convenient type of data storage system

is the magnetic recorder or a computer hard drive memory. Although other types of

recorders could be used, the magnetic recorder has the advantage of being able to store

large quantities of data, and to reproduce them in electrical form. The most desirable

way to transmit data signals is through electrical lines or telephone lines. There are

obvious situations where this is not feasible; for example, retrieving data from a

spacecraft in earth orbit. For such cases, radio transmission (telemetry) of the transducer

signals is usually required.

4.2.3. Data Preparation

The next key phase in data acquisition and processing is the preparation of raw data

for detailed analysis. The raw data are usually supplied from the recorder in the form of

voltage time histories (or as direct analog signals from the transducers, if the data are being

analyzed on-line). A number of operations are needed at this point to make the voltage time

histories suitable for detailed analysis. The first of these operations is generally classified as

data editing.

Data editing refers to those pre-analysis procedures which are designed to detect

and eliminate spurious and/or degraded data signals which might have resulted from

acquisition and recording problems such as excessive noise, signal dropout, loss of signal

due to transducer malfunctions. Editing can often be accomplished through visual

inspection of the data time history signals by a talented analyst. In more elaborate data

acquisition and processing systems, a specific instrument for quick-look evaluation might

be employed. Real-time spectrum analyzers are popular for this application. It should be

noted that the data editing step is more critical for the case of digital processing than for

analog processing. This is true because once the data have been converted to a digital

format, it is often difficult to detect even the most obvious errors in the original signal.

For the case of analog data processing, data preparation beyond editing usually

includes only conversion into engineering units (calibration).

4.2.4. Data Qualification

The correct procedures for analyzing random data, as well as interpreting the

analyzed results, are strongly influenced by certain basic characteristics which may or

may not be exhibited by the data. The three most important of these basic characteristics

are the stationarity of the data, the presence of periodicities in the data, and the normality

of the data. Stationarity is of concern because the analysis procedures required for non-

stationary data are generally more complicated than those which are appropriate for

stationary data. Periodicities in the data should at least be identified to avoid erroneous

interpretations of later results. The validity of an assumption that the data (excluding

periodicities) have a Gaussian probability density function should be investigated since

the normality assumption is vital to many analytical applications for random data.

Qualification of sampled data in terms of these basic characteristics is indicated as a

separate operation to be performed prior to detailed data analysis. In practice, however, it

is often accomplished as an integral part of the data analysis phase. Practical

considerations and procedures for such qualification will now be discussed.

4.2. Data Analysis

The procedures for analyzing the properties of random data may be divided logically

into two categories: the procedure for analyzing individual sample records, and the procedure

for analyzing a collection of sample records given the properties of the individual records.

Applicable data analysis procedures for these two categories are now outlined.


The first step is a mean and mean square value (or variance) measurement. This step

is almost universally performed for one or more of three sound reasons. First, since the mean

and mean square values are the basic measures of central tendency and dispersion, their

calculation is generally required for even the most rudimentary applications. Second, the

calculation of short time averaged mean and mean square value estimates provides a basis for

evaluating the stationarity of the data. Third, mean and mean square value estimates can be

extracted from other descriptive properties (probability density plots, correlograms, and/or

power spectra) which might be measured later. The comparison of directly measured mean

and mean square values estimates to the corresponding estimates extracted from other

analyses provides an excellent method for checking the data analysis equipment or computer

programs for correct operation.


Perhaps the most important single descriptive characteristic of stationary random

data is the power spectral density function, which defines the frequency composition of

the data. For constant parameter linear physical systems, the output power spectrum is

equal to the input power spectrum multiplied by the square of the gain factor of the

system. Thus power spectra measurements can yield information concerning the

dynamic characteristics of the system. To be more general, the mean square value of the

data in any frequency range of concern is determined by the area under the power

spectrum bounded by the limits of that frequency range. Obviously, the measurement of

power spectra data, will be valuable for many analysis objectives. Secondary

applications include its use for the detection of periodicities and as an intermediate step in

the calculation of autocorrelation functions.


The last fundamental analysis included in the procedure is probability density

analysis. Probability density analysis is often omitted from a data analysis procedure

because of the tendency to assume that all random phenomena are normally distributed.

In some cases, however, random data may deviate substantially from the Gaussian form.

If such deviations are detected by a test for normality, then the probability density

function of the data must be measured to establish the actual probabilistic characteristics

of the data. Furthermore, a probability density function estimate is sometimes used as a

basis for a normality test.

4.3. Field Data Monitoring

4.3.1. Data Acquisition Methodology.

The data was recorded in the form of one cycle snap shots of voltage, current and

power factor. These waveforms were acquired once every one minute interval. The

voltage and current waveforms at the distribution transformer output terminals were

recorded and stored temporarily in the local hard drive of the data acquisition system.

The stored data was then down loaded to a personal computer over a telephone line using

modems. The distribution nodes under study were feeding typical domestic loads at 110

volts and EV loads at 220 volts. At each of these locations the distribution transformer

has typical domestic loads of 2 to 10 homes. At each location one EV was provided to

one of the customers for a period of two weeks. The customer was asked to put the EV to

normal use. The data at each location was recorded during time intervals that can be

classified into three states. The first state is the period (about two weeks) when there was

no EV load. The second state is the period (about 2 weeks) when one of the houses on the

distribution transformer was using an EV. The third state was again when there was no

EV load. Thus the field data collected represents the load conditions before; during and

after the EV load was connected to the system. Figure 4.1 shows the data acquisition

system layout.

4.3.2. Data Monitoring Instrument.

The field data was recorded using several BMI models 8020 Plus PQNodeTM data

acquisition instruments. The BMI model 8020 Plus PQNodeTM monitors power quality

phenomena in electric power systems. It combines the capabilities of instruments that

monitor continuous quantities, (such as voltage level, load level, power factor, and

harmonic distortion), with the capabilities of instruments that record disturbances

(impulses, wave shape faults, swells and sags, outages, cold load pickups). Monitoring

continuous quantities requires that the monitored signals be sampled at a regular basis,

while monitoring disturbances requires that the signals be continuously sampled, and

recorded only if the signals exceed specified values. The BMI 8020 Plus PQNodeTM

continuously samples up to four voltages and four current three phases and neutral. The

instrument periodically records waveforms, RMS levels, frequency and temperature so

those trends can be calculated. The PQNodeTM is designed to interface with a personal

computer via telephone line, and is equipped with an internal modem for this purpose.

4.3.3. Data Management

The data acquired by the BMI data acquisition system was stored by the BMI unit

in the hard drive of the device in the form of ASCII files. Each set of data was stored as a

4 K file. The actual information is in only 1 kilobyte of the 4 K space. After downloading

the data files to the personal computer in the Power Quality and Power Electronics Lab,

at The University of Florida, the data needed to be stored safely with adequate backups

and at the same time consuming optimum space. The data also needs to be easily

restorable whenever needed for analysis purposes. Using data acquisition and data storing

software, the data was archived in the computer hard drive. The archived data occupies

only one fourth of the space since the files are crunched together. The data was also

backed up on tape drives connected to the computer. In this way it was made certain that

the precious data is not lost in the event of a hard drive failure. For the purpose of

analysis, only the required amount of data is restored and after it is processed it is deleted

from the hard drive and a fresh batch of data to be analyzed and processed is restored

from the archives.

For each batch of data the THD trend is plotted. Also plotted are the voltage and

current trends for all the harmonics from the second through to the fiftieth harmonic

component. In addition to the harmonic trends, steady state trends are also plotted; these

are the real power, apparent power, RMS current and power factor trends for the two

phases between which the EV charger is connected. Each of the trend waveforms is

displayed one by one.

After examining the harmonic and steady state trends for the selected days,

typical real time steady state voltage and current waveforms at certain instances of time

were also plotted. It may be mentioned here that the BMI instrument acquires the

distribution node data in the form of real time waveforms of voltages and currents. Using

these waveforms which are acquired during every fixed period of time, the computer

software uses them to create various harmonic and steady state trends.

Figure 4.1. Data Acquisition System Layout


5.1. Introduction

This section deals with the graphical representation of the recorded data. A visual

inspection of the data trends is enough to show that the data represents stationary random

process. It is assumed that all random phenomena are normally distributed. [data book] In

some cases, however, random data may deviate substantially from Gausian form. Looking at

the data trends included in this chapter, the properties of the recorded data can be described at

any instant of time by computing the average values over a collection of sample functions that

define the random process. The stationary random data represented here falls in the category

of ergodic random data. The properties of ergodic random process can be determined by

performing time averages over a single sample function. Fortunately, in practice, random data

representing stationary physical phenomena are generally ergodic. It is for this reason that the

properties of stationary random phenomena can be measured properly, in most cases, from a

single observed time history record.

The recorded data is presented in the form of various trends and waveforms. Each data

trend for a particular site with and without the EV load are presented together to facilitate

comparison between the two load scenarios namely when the EV load is the part of the load

and when the EV is not a part of the load. The data recorded using the BMI data acquisition

units was synthesized using the software PASSTM. Broadly speaking two types of data trends

were created and displayed graphically, viz. the steady state trends and the harmonic trends.

For the purpose of studying the impact of Electric Vehicle loads on the distribution system,

only the following trends have been selected:

1. The Steady State real power trend.

2. The Load Current Total Harmonic Distortion (THD) trend.

3. The various harmonic trends of the load current.

Although there are many other steady state and harmonic trends that were

displayed using the PASSTM software, only the above mentioned three trends have been

used for data analysis for the following reasons:

The EV charger load is a real power load, which consumes a peak power of 7

kilowatts (based on GMI charger). The 7-kilowatt power is consumed when a fully

discharged EVI battery is connected to the charger. The power decreases as the charge .

on the battery increases. So one of the impressions that the EV charger load causes in the

system is an increase in real power. Thus the real power trends for the distribution node

with and without the EV load were created. This made it possible to compare the load

scenario with and with out the EV load. From the laboratory testing of the EV charger, it

was found that the EV load causes a distortion of the input line current. The General

Motors EVI is known to cause very little input voltage distortion [2].

The data represented in this chapter can be broadly classified into two categories


I. When the distribution transformer is feeding normal domestic loads.

II. When the distribution transformer is feeding normal domestic loads that
also include Electric Vehicle Load.

Representing the data in this manner has facilitated the comparison of the load

scenarios with and without the EV load.

5.2. Data Trends and Characteristics

Figure 5.1 through 5.7 are sample data trends created from the actual monitored field

data. Each of these figures consist of two parts, viz. part a and part b. Parts a are the data trends

without the EV loads, parts b are the data trends when the EV load is a part of the loads on the

transformers. Following is the summary of highlights of each of the figures 5.1 to 5.7.

Figure 5.1: This figure represents the Total Harmonic Distortion trends of the line

voltage. The voltage THDs for both cases are very close. Maximum voltage THD variation

between "a" and "b" is 2%. The presence of EV loads does not have a significant effect on the

input voltage THD. This fact is verified by the laboratory testing of the EV charger.

Figure 5.2: Unlike the voltage, the current THD trends shown in figure 5.2, are

sensitive to EV loads. Comparing 5.2 a and 5.2 b, it can be seen that the current THD trend for

two cases is not the same. When EV was a part of the domestic load, we see in 5.2 b, the

current THD between the hours of 2:15 and 4:40 is very high. The current THD during this

time varies between 80% to 40%. For the remaining part of the day, the current THD for both

cases is close to each other.

Figure 5.3: These figures show the real power trends. The real power trend, when EV

is a part of the load is not much different from the case, when EV is not a part of the load. This

could be due to the fact that the customers do not fully discharge the batteries before

recharging them. Thus each time the battery charger load is put on the system, it draws only a

fraction of its maximum rated power of 7 kilowatts.

Figure 5.4: These show the power factor trends. The EV load does not effect the

power factor. Thus we see, in both parts of this figure the power factor trend stays close to .8.

Figure 5.5: These figures represent the third harmonic current trends. As in the case of

the overall current THD trends, the third harmonic current trend between the hours of 2:15 and

4:00, show a high, for the case when EV is part of the load. The third harmonic component

touches a high of 1.5 amperes during this period. This again indicates a possible presence of

EV loads during these hours.

Figure 5.6 and 5.7: These are the fifth and the seventh harmonic current trends. The

conclusions drawn from these trends are similar to those drawn from figure 5.5.

5.3. Data Observation

Observing the data trends represented in figures 5.1 through 5.7, the following

conclusions can be drawn:

I. The EV load does not have a distorting effect on the input line voltage. Thus
the input line voltage THD is not effected by the presence of EV loads.

II. The EV loads cause a considerable distortion of the line current. The line
current THD is particularly high when the amplitude of the fundamental of the
line current is small. Thus the current distortion is maximum close to the end
of the charging cycle, when the charger draws a small current.

III. The EV users generally would not be expected to fully discharge the batteries
before recharging them. Thus the EV loads may not have a significant effect
on the expected peak load on the distribution transformers.

IV. The EV battery charger operates at a high power factor. Thus the reactive
power drawn by the charger is low.

V. The EV charger injects a significant amount of lower odd current harmonics
into the power line.

5.4. Load Characteristics

This section contains the computed maximum and the average daily power output of

the distribution transformer. The daily average power is the average of the load over a period

of two weeks. The load average has been found, first for the period weeksk) when EV was

not a part of the loads and next for the period of 2 weeks, when EV was a part of the load.

Also, for these two two-week periods the maximum load profile has been computed. The

figure X.8 a, has been derived by computing the load average for a 2-week period when EV

was not a part of the loads. The figure 5.9b is the 2-week load average when EV was a part of

the loads. Comparing, 5.8a and 5.8b, we see that there is an increase in the average power

consumed, when the loads contain the EV load. This increase is particularly visible, during the

early morning hours, this reinforces the previous conclusion about the times at which the EV

load was connected to the system.

Figure 5.9 a and b, show the maximum power recorded during the two two-week

periods. Plot in figure 5.9a is the maxima of the daily load curves of the two weeks period

when EV was not a part of the loads. Plot 5.9b is the maxima of the daily load curves of the

two-week period when EV was a part of the loads. It must be noted, the EV loads although do

not effect the peak loads, nevertheless being an additional load it does affect the average daily

power consumed.

Figures 5.10 a and 5.10 b, show the snap shots of the recorded line current waveforms.

The line current snap shot of figure 5.10a was recorded when EV was not expected to be a

part of the loads and 5.1 la was recorded, when EV was expected to be a part of the loads.

From the Fast Fourier Transforms (FFTs) of waveforms in figure 5.10a, shown in figure

5.1 Ob, we see that the line current does not have a high content of the lower odd harmonics

(viz. 3rd, 5th and the 7"). This is not the case with figure 5.1 lb, where these harmonics are


Based on the preceding discussion, the load profile of Figure 5.9, qualifies for a

generalized load profile of loads on a distribution transformer, for that particular site.

Voltage THD Trend: No EV Loads

Time of Day


Voltage THD with EV Loads

0 0 0 (N (0 S mo 0 V 2

Time of Day


Figure 5.1 Voltage Total Harmonic Distortion Trends

Current THD Trend: No EV Load

o 20

I 15


U. I


'I U, vi 10 CO

Time of Day


Current THD Trend: EV Load


80 -

O. Z 0

Time of Day


Figure 5.2. Line Current Total Harmonic Distortion Trends

Real Power Trend: No EV Load

D 4
3 3
co 2

77o 7 -4 N N o o in 9 o Sn n p

Time of Day


Real Power Trend: EV Load


6 ---

I g
0 0
t 2

'N Ip .O. ., ; z b d,

Time of Day


Figure 5.3. Real Power Trends

Power Factor Trend: No EV


a0 0

0-~ s _b 0 l .- -. <. ,:. ---<- IV N.- N' 4-N
&u ^ n' 41> 1' 41. 41 <1 41 41 ,* 4'-1 441 41. (? >^ n i

Time of Day


Power Factor Trend: EV Load

*5 1
_026 -------------- -- ---------- I
M 04

1 aS az o~o -i**''oOO.. 41 .; 41 41 41 \0 i 41

Time of Day


Figure 5.4. Power Factor Trends

Third Harmonic Current Trend: No EV Load

E 5
< 4

-- 3? N 'N- N N N
2- j--- ---- -

C ) C- N V V N '


N C CN T t C C CO N rN 0

Time of Day


Third Harmonic Current Trend: EV Load

3 -----------------


Time of Day

-n +,- A

'2 2 2

Figure 5.5. Third Harmonic Current Trends


Fifth Harmonic Current Trend :No EV Load

0- 25
E 2
C 15


25 --------------------------.--.-----.------.-------,--

Time of Day


Fifth Harmonic Current Trend :EV Load

m 2
< 1.5

0 0

Time of Day


Figure 5.6. Fifth Harmonic Current Trends

Seventh Harmonic Trend : No EV Load

S 12
E- 1 2 ----------------------------
E 08

Time of Day


Seventh Harmonic Current Trend : EV Load


S 0.6 -- --
q 04
` <, 8; 1,( I' 0*^.^*.^**{*' *^ *^ ^y^^^^ '.

Time of Day

.E 0 .6 ,-------- -----


Figure 5.7. Seventh Harmonic Current Trend

Average Power: No EV Load

6 1 1

C 6 I... .i .$ A Ji._ ..i-M ., ...A ,---
(a 0

'S. ^ y *y O{* (6 6 N6 C8. 3. 1 *w

Time of Day


Average Power: EV Load

8 &:
o> 1 2r---------------------------------

. 0 .

11P '?1* o"1 o11P 11 N 11% 14" NIP le 11 6N ^'10
Time of Day


Figure 5.8. Average Power Computed on the Basis of Two Week Daily Load

Max Power Demand: No EV Loads

12 ---- --

4 -PP 0'-- P *s -.P P- -- -
2 -------------------------------------

Time of Day


Max Power Demand : EV Load

20 -
c 15

10 -


4' 4' "''0 S,> ,0 0 "* e 0 4 ,0 e* ",?-l .

Time of Day


Figure 5.9. Maximum Power Demand Computed over Two Week Period

radwClirventrwawemi wi NEV Lo[ad

5 10 15
noe ( 5


FFrOftie wad curritWaeirm wj WNo EV ad

S: 3.310773
RMSi : 0

06 --- --- ----- RMS 32.6343
AsJM : 395032
TF 130.916
04- ---r :427332

00 1 1- 1 I I II 1. I I ll II l i. II ... II I I I
120 720 1320 1920 2520 3120


Figure 5.10. a) Line Current Waveform Snap Shot with No EV Load
b) Fast Fourier Transform of the Current Waveform of "a"


lo~---- -
-10 _

L -<



CF : 1.61574
-FFr: 112339



Fe i Recorded b Jrnentwil EV Loads

Hr, I.,'1.-
Ag,: 533581
Abs: 10.7422
RMS: 6/3585
CF : 175933
FF : 114432



Hrmon m ii Coentofte Fe-iRecrGedLad Current
r- -: 6
Rrd 153EI
T-H :12278
TD :0
TD :12S78
RMSh 197152
-- RMSi: 0
RMSft: 15A95
RMS :15A95

T : 356.55

--t -.. -. _- ___

720 1320 1920 2520 3120


Figure 5.11.
a.) Field Recorded Current Waveform.
b.) FFT of the Field Recorded Waveform in Figure a.



| 100





i : iL, ru tff-l-c



6.1. Testing Procedure and Results

The EVI charger is a 240 volt split single-phase device. In the laboratory

testing of the charger, a nearly fully discharged (5% of full charge) battery was connected

to the charger. The voltages and currents on both input lines were measured; the

equipment set up for the laboratory testing is shown in figure 6.1. Electrical

measurements were taken using the power monitor, automatically once every one-minute

interval for the entire charging cycle. The fully discharged battery was connected to the

charger at 10:17 AM. The battery reached its almost full charge at 2 PM, a charging time

of a little less than 3 hours. The EV1 charger when charging a fully discharged battery

draws a current of 30 amps at 240 volts and is also a source of significant harmonics as

shown in the following laboratory results. These results have been presented in the form

of waveforms and graphs that depict the electrical characteristics of the EVI charger.

The first graph, Figure 6.2 shows the power consumed by the EVI charger over

the entire charging cycle. The load varied from the maximum of 7kw at the beginning of

the cycle to about Ikw when the battery is nearly completely charged. During the entire

charging cycle the input voltage remained unaffected as far as either the magnitude or the

harmonic distortion is concerned (Figure 6.3).

The input current however varied significantly in percentage harmonic content.

Figures 6.4, 6.5, 6.6 and 6.7 show the current waveforms recorded at specified instants of

time. It can be observed that as the amplitude of the current decreases the waveform distortion

becomes more and more visible. Figure 6.9 shows the RMS values and figure 5.10 the Total

Harmonic Distortion (THD) of the input currents for phases a and b, over the entire charging

cycle. The figure 6.11 shows the harmonic content of the input current. It was seen that the

amplitude of various harmonics throughout the charging cycle does not vary significantly.

Since the amplitude of the fundamental component decreases with time and the THD, which

is a percent measure of harmonics to the fundamental shows a significant increase with time.

6.2. Observation of Laboratory Results

The EV charger charging into a fully discharged battery draws a line current of 30

Amps at 240 Volts. This corresponds to a power consumption of approximately 7 kW at time

10:17 as shown in Figure 6.2. The voltage waveform as shown in Figure 6.3., stays distortion

free throughout the charging cycle. The input line current distortion at the beginning of the

charging cycle is less visible (Figure 6.4). This is because the amplitude of the line current is

high and its THD which is expressed as a percent of the fundamental RMS current, is low.

Note that in Figure 6.10 the THD at time 10:18 is just about 2 % for both line currents laand

Ib. As the charge on the battery increases and the amplitude of the line current decreases, the

current distortion becomes more and more visible (Figure 6.5 to 6.8). This is verified by

Figure 6.10 where in at the end of the charging cycle, the THD of the line current climbs to

almost 30 %. Although the THD of the line current increases as the charge on the battery

increases, the actual harmonic content of the line current stays fairly constant. Figure 6.11


shows the line current harmonic content up to the 50h harmnic. The most significant

harmonics are the 3rd, 5h and the 7t.The even harmonics are almost not present.

Line 1
Line 2
Neutral I


Power Monitor
Digital Oscilloscope
Signal Analyzer
Power Harmonics


Single Phase


Figure 6.1. Experimental Setup for EV Charger Testing


Charger Power Output

m 6000
1 4000

-1 i-1 r-i I--I --i 1--I -I --

Figure 6.2. Power Curve for the EVI Charger.



-too i- / : ; : /


S- \50

U3 -100-

0 5 10 15 20 25
Time (ms)

Figure 6.3. Typical Charger Output Voltage Waveform



20 \


-40 ,

0 5 10
Time (ms)

Figure 6.4. Charger Input Current Waveform Sampled at 11:19





40 -



10 -


S-10 -


0 5 10 15 20 25
Time (ms)

Figure 6.5. Charger Input Current Waveform Sampled at 11:40

I 1



Ti me


Figure 6.6. Charger Input Current Waveform Sampled at 12:07

Tj me

\ /

5 20 25

Figure 6.7. Charger Input Current Waveform Sampled at 12:36

1 .0 1 1 111

0.5 --





10 15
Time (ms)

Figure 6.8. Charger Input Current Waveform Sampled at 13:53




Current RMS


30 -- 11 RMS
25 --


15 -

10 -

10:18 10:42 11:02 11:22 11:42 12:02 12:22 12:42 13:02 13:22 13:42

Figure 6.9. The Charger Current RMS Variation over the Charging

Current THD



2la THD
25 lb THD

10 20



10:18 10:42 11:02 11:22 11:42 12:02 12:22 12:42 13:02 13:22 13:42

Figure 6.10. The Charger Current THD Variation Over the Charging Cycle.

Ia Max Harmonics

0 0.6



0\ a 'r aO M 0 CM,1 'D 00 0 N *' D 0 0 0 CM x, c) 0
o 'N M M M M ( 't 't I

Figure 6.11. The Charger Input Current FFT.


This section contains the results of computer simulations of the power distribution

system with domestic and Electric Vehicle (EV) loads. The domestic load included in

these simulations is arbitrary. The EV model used is based on the laboratory data

obtained by testing the actual charging characteristics of an EV charger used by the GMC

EVI. The aim of this simulation is to see the effect of EVs connected to the distribution

system. The transformer current with EVs present in the system is simulated. The FFT of

the transformer current with and without EVs are shown and different harmonic

components are found to study the effect of these harmonics on the distribution

transformer in terms of the derating of transformers.

The power distribution system with an EV charger load was simulated using

SABERTM. This section includes the simulation results and discussion about the

simulation model.

7.1. Component Models

7.1.1. System Model

The system model consists of the distribution transformer, domestic load and the

EV charger and battery load. Figure 7.1 shows the layout of the simulated power

distribution system. The distribution transformer has typical domestic loads connected at

Figure 7.1. Distribution System Model.

Figure 7.2. Simplified Model of the Simulated EV Charger

120 volts between the center tap and the outer terminals. The outer terminals, which are

at 240 volts, are connected to a full bridge rectifier. The bridge rectifier is used to model

the inductive charger used by the EV to charge its batteries.

7.1.2. Electric Vehicle Charger And Battery Models

The EV charger is modeled on the basis of laboratory data obtained during the

charging of a fully discharged EV battery connected to charger. The full bridge rectifier

represents the charger. The input to the rectifier is 240 V from the distribution

transformer. The EV charger and battery model is shown in figure 7.2. The battery has

been modeled as a series combination of a resistor and a capacitor. The resistance

represents the loss component of the battery. It is presumed that the charging pattern of

the capacitor shall closely represent the charging pattern of the EV battery. The values of

the resistor and the capacitor can thus be chosen to achieve the desired charging current

and charging time. In the simulation while the full wave bridge rectifier represents the

EV load, a composite current source has been used to represent the harmonic effect of the

EV charger. The composite current source is injecting harmonics of the same amplitudes

and respective orders as shown in figure 3.2. The actual waveform of the harmonic

current injected by the composite current source is shown in Figure 7.3.

7.1.3. Transformer Model

The transformer used for the simulation is a gap less three winding transformer

with one primary and two secondary windings. The transformer in the simulation has the

following specifications.

3 i m ------ -. _________ --- _

-3 Hroict ha re

Figure 7.3. The Harmonic Current Injected by the Charger

These specifications were provided by FPL.

7.1.4. Domestic Load Model

The domestic load has been modeled as a combination of some lighting and heating

load represented by resistive load. The load has been modeled such as to exhibit the

characteristics similar to the actual load characteristics as determined from the field data.

Figure 7.4 shows a typical domestic load model used in simulation.

7.2. Mathematical Model Of The EV Charger.

The EV charger can be represented by a single-phase bridge rectifier with an R-C

load, as shown in figure 7.1 can represent the EV charger. A current source model is used for

the injected harmonics when there is very little voltage distortion [9]. Since the EV charger is


Turns Ratio 7260, 120/240

Material of the Core Grain oriented silicon steel

Thickness of Lamination .009 inches

Net cross sectional area of the core 28.9 sq. inches

Path length 25.74 inches

Magnetization Saturation 20.0 Kilogauss

Figure 7.4. Domestic Loads Model

known to cause very little voltage distortion, the harmonics injected by the charger into the

line can be modeled by a composite current source. The behavior of the charger can thus be

determined by a second order system.

LC +RC-dvc + v =v (1)
dt2 dt

where v, the transformer secondary voltage

vc the capacitor voltage

L the equivalent battery inductance

C the equivalent battery capacitance

R the equivalent battery internal resistance

The net harmonic current from the composite current source can be expressed as:

iH Imk sin kcol (2)

where it, is the net harmonic current

k is the order of the harmonic

Ik is the amplitude of the k th harmonic current

The SABERTM software package was used to simulate the system, both transient

and steady state operating conditions, including RMS values, THD etc can be obtained

using SABERTM. In the battery model the L is small and hence can be neglected.

Therefore the charging time constant (T) for the battery can be given by:

T = RC (3)

7.3. Simulation Results

In the computer simulation two types of loads have been considered, viz. the

lighting loads and the EV load. The simulation results with two houses and a single EV

load are presented. The lighting load current is taken as a pure sinusoidal current. The

current harmonics injected into the mains by the EV charger are shown in figure 7.3. This

is based on the experimental results obtained by testing the EVI charger in the

laboratory. The harmonic components of the charger input currents (figure 7.5) are

supplied by the composite current source. The charger input current in the simulation is

controlled by varying R. We know from the lab test data that the maximum current that

the charger draws (for fully discharged battery) is approximately 30 Amps.

By modeling the GM EVI charger as a combination of a full wave bridge rectifier

with a resistive and inductive load and a composite current source, simulation results very

close to the lab results were obtained. A detailed discussion on the comparison of the

laboratory results, the simulation results and the field-recorded data has been included in

Chapter 7.

The simulation runs of the distribution system made it possible to check the various

load scenarios in the distribution system. The system design can now be checked for an

arbitrary number of EVs connected to it. This data obtained from simulations is important and

makes it possible to optimally design/redesign future distribution systems. This includes

replacing and/or adding transformer units in the existing systems in order to make them able to

withstand the new load conditions caused by the EV load and the harmonic distortion

associated with it.

Based on the field and laboratory data the EV charger has been successfully modeled.

The simulation results are close to the laboratory results obtained during the actual charging of

the EV battery. The waveforms on the input side of the transformer have a low order current


7.4. Simulation of Short Charging Time Chargers

The computer simulations were used to check scenarios when the charging time

for charging the battery is much shorter. The GM EV1 charger takes about three and a

half hours to charge a fully discharged battery (from about 15% of full charge). The long

charging time is a drawback with the EV and research is being done to reduce this

charging time substantially. The fast charging time chargers are not commercially

available and their behavior in terms of harmonic effect is not known. The computer

simulations have been used to provide an estimate of the effect of the fast charging EV on

the distribution system. The results of these simulations are based on the researcher

intuition and experience of working with highly non-linear loads. The fast charger study

is aimed at providing an idea about how such a technology, when developed can effect

the distribution system. These results need to be verified once the fast charging

technology becomes available.

7.4.1. Simulation Methodology For Fast Chargers

The charger model for the fast charging charger is essentially the same as that for

the normal charging time charger. The difference however is only in the values of the

resistor and the capacitor that represents the battery. Also the amplitude of the harmonic

currents has been increased proportional to the decrease in the charging time. Thus the

fast charging time charger has also been simulated as a combination of a full-wave bridge

rectifier and a harmonic current injection source. As mentioned earlier, these simulation

results are based on intuition and experience of working with non-linear loads. These

results can only be verified once fast charging technology becomes available. Figures 7.6

through 7.8 show the current waveforms and their respective FFTs for simulated

scenarios where the charging time is reduced as indicated in the figures. The distribution

transformer derating corresponding to this charging scenario is shown in Section 8.2.

Figure 7.5 Transformer Current for the Fast Charging Charger with Charging
Time of 45 minutes.

Trmuuformar Cun Lr


o1ask 1~k e.75k 1.0K I ItK 1.Ek

Figure 7.6 FFT of the Transformer Current for the Fast Charging Charger with
Charging Time of 45 minutes.

(A): I(S
Trmfanner Curuwmn


Figure 7.7. Transformer Current for the Fast Charging Charger with Charging
Time of 1.2 Hours.



0.6k 1.0k

Mlg(A) : f(Hz)
FFT of the Trunsformw Curnw

1.5k 2.0k

Figure 7.8. FFT of the Transformer Current for the Fast Charging Charger with Charging
Time of 1.2 hours.

(A) : t()
Transonner Cumnrt


In this section, the field data and laboratory test results are compared with those

done by computer simulation to present the effect of EV loads on the power distribution


The aim of this project is to study the impact of the EV load on the power distribution

system. In summary, this goal has been accomplished by recording the field data with and

without the EV load, recording the behavior of the EV charger by the laboratory testing of the

GM EVl charger charging a fully exhausted battery. The data acquired from the field provides

the information about the characteristics of the distribution loads when EV load is a part of the

load and when there is no EV load. The laboratory testing of the charger provides the input

current characteristics of the charger and the line current harmonic content of the charger.

Using the data from the laboratory test, the computer simulation model of the EV

charger has been built. The details of the computer model have been discussed in Section

8.1. Thus the laboratory testing of the charger forms the basis on which the computer

model is built. The charger computer model has been incorporated into the overall model

of the distribution system. The distribution system consists of the transformers, feeders,

different types of normal domestic loads and the EV loads.

First of all, in this section, the accuracy of the computer model of the EV charger

will be established. This has been achieved by comparing the laboratory test results of the

EV charger with the computer simulation results. Second, the similarity of the field-

recorded data with the simulation results will be discussed.

8.1. The Field Data

The field data was recorded in the form of one-cycle snapshots of current and

voltage as well as power factor at the distribution transformers at four different locations

in South Florida and then transmitted to a computer over the telephone line. The detailed

description about the data monitoring and acquiring methodology are mentioned in

Section 4.2.

8.2. Laboratory Test

For the laboratory test the EV charger is selected as a 240 volt, split single-phase

device as described in the previous section. The voltage and current input into the charger

were monitored and recorded over the full charging cycle for the EV, which was nearly

completely discharged to a level of 5% of full charge. Electrical measurements were

taken using a power monitor automatically every minute over the entire charging period

of about three hours.

From the measurement it was found that the current supplied to the EV charger

varies throughout the charging cycle. The charger current variation is shown in Figure

5.9. The actual amplitude of each of the harmonics in the charger current does not vary

significantly during the charging cycle.

8.3. Simulation Results

The power distribution system consisting of the feeder, distribution transformer,

the EV charger, together with domestic loads (mainly the lighting lamps) was simulated

using SABERIM. The models of various components used for simulation are based on

both field data recorded and the laboratory data. To simulate variation of input current

and distortion at different instants over its charging period, an adjustable resistor, as well

as a composite current source which is based on the FFT analysis of the laboratory test

data, is introduced in the loop. By varying the resistance, the identical situations to those

in the laboratory test can be established. Therefore, current waveform snapshots over a

whole cycle of 60 Hz are obtained and then compared with those recorded in the

laboratory test.

8.4. Comparison

The field data, laboratory test data and simulation results are compared in terms of

the waveform snapshots with the same magnitude of charger input current for all three

cases viz. the field data, the laboratory test data and the simulation results. The aim is to

establish the reliability of the laboratory test data; the simulation model and simulation

results by comparing them to each other and the field recorded data. Some of the

waveforms that have been included in previous chapters have been put in this section for

convenience of comparison.

FeI Recorded Currentwiti EV Load

Mi: -302734
Ag: 1527
Ats: 302734
RMS: 181012
CF : 167246
IF:1 85

5 10 15 20


Ha ont ATf flkide oftie F bRAecordeaurrent

Fi :179963

TD :0
To :10.79
RMh : 194302
.-- ----- ---- ------- RMSi :0
RMSih: 1:0
RMS : 18009
P511M 27.771
T :289335

'. I..'.' 3120


Figure 8.1.
a.) Field Recorded Current

b.) Waveform FFT of the Field Recorded Waveform.

^^ L^TbnitBTiE

20 *- -


^ o-----




M34 'fZl
Mr. 1l".C
Ag: 533581

CF : 1.7533
FF :114432



Hanmormt rc ntoftie iefiRecordi Qarent --

TM :12B78
S TD :0
T :123278
RMs: 197152

RMS :1SA95

-- 5

r 5518.64

25 -

120 720 1320 1920 2520 3120


Figure 8.2.
a.) Field Recorded Current Waveform.
b.) FFT of the Field Recorded Waveform in Figure 8.3.

-," _____

C 0
ia r^ l /a~





* 075-




0 10 1I 20 0i
Time (mS)

Figure 8.3. Laboratory Test Transformer Current at 11:40 AM

Time (ms)

Figure 8.4. Laboratory Test Transformer Current at 12:36 PM

Figure 8.5. FFT of the Current Waveform of Figure 8.6.

muld a Chrgr RCur wth corr-pong wFFT
(correpondnig to Lab Te CuMrent 12:0)
Mwg(A): ;f(H


0.5k 1.0k Aik 2.0k 2.k 0.0k

SO IKhorNt-xjn2)

0Oj2 0028 0.06 45 004

Figure 8.6. The Simulated Current Waveform and the Corresponding FFT

The SkAuld Current Wvetornm nd Nb FT
(Conoepondnngto 11:40 urent Wefnm of the Lab TeM)
1. ___~_________ MCgA) t:b)

S | I I
O.k Ik 1.6k 8k 2.k Mk

~~~____~~_____(A) ,t(*)
210 |l(horwttn)

0.0 0m01 0. 0.4 0.04

Figure 8.7. The Simulated Current Waveform and the Corresponding FFT

The waveform of figure 8.lshows the field recorded current for the distribution

transformer feeding two homes (FPL_1I). This is the distribution transformer current when the

load on the transformer contains EV load. Figure 8.2 is the FFT of the waveform in Figure

8.1. The observation after comparing the current waveform of Figure 8.1 and its FFT in figure

8.2 with the laboratory recorded waveform in Figure 8.5 and its corresponding FFT in Figure

8.6 are presented in a tabular form in Table II. Both the currents (of Figure 8.1 and 8.5) are

then compared to the waveform of Figure 8.8. Figure 8.8 shows the simulated transformer

current. Also compared are the FFTs of the three waveforms. (Figure 8.2, 8.4 and 8.8).

Table. II

Laboratory Test Simulation Results Field Data

(Figure 8.5 and 8.7) (Figure 8.8) (Figure 8.1 and


Peak Value of 30 Amps 29 Amps 27.82 Amps
RMS Value of 19.8 Amps 20.2 Amps 18.1 Amps
Crest Factor 1.515 1.435 1.537

THD 14.1 11.2 12.8

Holding the field-recorded data as reference the relative percent values of the

laboratory and simulation results is shown below in Table III.

Table. III

Looking at the waveforms figure 8.1 through figure 8.8 and Tables II & III, we

note that the field recorded and the simulated current contain the domestic load current.

The laboratory test waveform however is only the charger current. In the simulation the

domestic load is taken as linear load (Figure 6.4) while in reality (Field recorded) the

domestic loads may contain some non-linear loads. Thus the THD of the field current is

higher than the THD of the simulated current. The laboratory test current consists of only

the charger current.

Laboratory Test Simulation Results

(Figure 8.5 and 8.7) (Figure 8.8)

Peak Value of Current 7.8% 4.24%

RMS Value of Current 9.39% 11.6%

Crest Factor -1.43% -6.636%

THD 10.15% -12.5%


9.1 Method of Calculation

9.1.1 ANSI/IEEE C 57.110 Method

The method used for calculating the derating of the distribution transformer is the

ANSI/IEEE C 57.110 (see Section2.3.1). Using the method described in Section 2.3.1, the

derating has been calculated as a function of the current drawn by the charger. The derating

curve is shown in Figure 9.1. In the calculations the derating is calculated taking the charger

current as the base. So fh in equation 2.1 is calculated dividing each harmonic component by

the amplitude of the fundamental charger current. Thus, in this method the value of fh for the

fundamental component is always 1. This is the procedure employed by ANSI/IEEE C57.110

method. From the curve of Figure 9.1. we get the minimum and maximum limits of derating.

Table IV. Maximum and Minimum Derating Using ANSI/IEEE C57.110

Derating of Distribution Transformer

.-6 .4p SP3 .<^ .<^ .61 SP .(, 0(Z OZ .Q .Q .(

Time of Charging

Figure 9.1. The Transformer Derating Curve for the Complete Charging Cycle

9.1.2 Modified ANSI/IEEE C 57.110 Method

The ANSI/IEEE C57.110 method calculates the derating of the transformer

based on the charger fundamental current. In reality the transformer in a distribution

system is supplying other loads. Also from figure 9.1 it can be seen that the

transformer derating at 13:53 hours is as low as 20%. This calculation is done taking

the charger fundamental as base, and from Figure 5.9 we see that the charger current at

13:53 hours is just about 5 Amperes. This would form only a small percentage of the

total load on a distribution transformer of 37 KVA rating. Thus, although the derating

based on C57.110 method is very high the effect of the small current on the

transformer would be much less significant. In the modified ANSI/IEEE C57.110

method the calculation of transformer derating is done based on the maximum current

carrying capacity of the transformer. Since the transformer derating should be a

concern when the load on the transformer is full or close to full load this method

calculates the derating based on full load current of the transformer. The derating of a

37 KVA transformer using the modified method is shown in Figure 9.2.

92. Transformer Derating with Fast Time Chargers

Based on the simulation results of section 6.4, the modified ANSI/IEEE

C57.110 method has been used to calculate the derating of distribution transformers.

The derating has been calculated based on the 37 KVA distribution transformer whose

specifications are given in Section 6.1.3. Figures 9.3 and 9.4 show the transformer

derating for the two simulated load scenarios.

9.3. Conclusion

The derating of the distribution transformer based on ANSI/IEEE C57.110 method is

significant but it does not consider the current carrying capacity of the transformer when

calculating the derating. The modified approach calculates the derating of the transformer

based on the full load current of the transformer. It is obvious that based on this method we see

that the GM EVI charger does not significantly derate the transformer. One thing that may be

noted here is that the ANSI/IEEE C57.110 method is generally applied to find the derating of

large power transformer and gives highly conservative results for distribution transformers. It

is strongly recommended that the equation for derating of distribution transformers be

experimentally determined with KVA ratings of concern to FPL. (See Chapter 10). The

derating for the fast charging time chargers as seen from figures 9.3 and 9.4 is much higher. It

is expected that the fast chargers will be more prevalent than slow chargers if consumer

preference is not modified by the utility industry.

Derating of Distribution transformer

S-T- of Charging
Time of Charging

Figure 9.2. Transformer Derating using the Modified ANSI/IEEE Method


)) (( i

Derating of Transformer



*Derated Value


Derating of Transformer


gDerated Value


Figure 9.4. Derating Corresponding Short Charging Times a.) 1.2 Hours b.) 45 minutes


10.1. Summary and Conclusions.

The objectives of the project outlined in section 1.3 have been achieved. The

behavior of four distribution nodes (FPL_1I to FPL_4) has been monitored. The load

behavior with and without the EV load has been recorded. Volume II of the final report

contains samples of the recorded data with an executive summary. By performing the

laboratory test on the GMC EV 1 charger charging a fully discharged battery, the charging

characteristics of the charger has been determined. The charger takes about 3 hours to

charge a nearly fully discharged (15% charge) battery to full (100%) charge. The

harmonic spectrum of the transformer current charging the discharged battery has been

determined. The snap shots of the current waveforms have been recorded and studied.

Some of these waveforms have been included in this report. Based on the laboratory data

a computer model of the charger and the distribution system has been developed.

Comparing the simulation results to the laboratory results and the field data has validated

the simulation results. The simulated currents are roughly 4 to 10% away from the field

recorded currents, in their magnitude and distortion. Based on the field data, laboratory

results and the simulation results the derating of the distribution transformer has been

calculated. Based on the field data, laboratory results and the simulation results the

derating of the distribution transformer has been calculated. The two methods suggested

for the calculation of derating have been presented and compared. The maximum and

minimum limits of derating have been suggested. Using the first method the maximum

derated value is 80% and the minimum value is 1%. Using the modified method the

derating remains close to 99 % throughout the charging cycle. The transformer derating

results corresponding to fast charging (45 minutes and 1.2 hours) chargers has been


A Microsoft Excel based program has been provided to FPL. This program shall

enable the user to find the derating of the transformer using the harmonic content

information of the load on the transformer.

10.2. Scope For Further Research

The transformer derating has been calculated using the IEEE /ANSI

method. The formula is an empirical formula. To get a highly accurate measure of

the derating of transformers, laboratory tests should be performed. These tests can

be performed in the Power Laboratory in the Electrical and Computer

Engineering Department at the University of Florida. Using the real time program

R4TM, the computer can generate low voltage controlled harmonic signals. These

signals will be then amplified using the power amplifier in the laboratory and fed

to the transformer under test. Temperature sensors placed selectively on and

inside the transformers will sense the rise in the core and winding temperatures.

The change in the temperature rise with a change in harmonic content from pure

sinusoidal to very high harmonic content can be gauged. The transformers can be


tested for different harmonic scenarios and loads. Thus an exact formula for the

derating of transformers can be evolved. As the harmonic generation can easily be

controlled using the computer, affect of individual harmonics can also be studied.

This information can be particularly useful to FPL for the design of harmonic

filters that may need to be installed in places where the load is highly sensitive to

harmonics. Also, at sites where the load is known to generate a significant amount

of a particular harmonic, these tests can provide a strong basis for calculating the

derating of equipment and thus the design of appropriate transformers.


1. J. Pages, "On the quality of public electricity supply" Revue Francaise de I'Electricite,
vol.41, no.223, p. 14-21, 1968.

2. C. H. Hilger, "What is the quality of electric power?" Elektroteknikeren, vol.68,
no.19, p. 418-22, October 7, 1972.

3. H. Heikkila, "Harmonics in a power system" Saehkoe, vol.49, no.1, p. 27-30, January

4. G. L. Johnson, A. N. Mortensen, "A power system digital harmonic analyzer" IEEE
Trans. Instrum. Meas. (USA), IEEE Transactions on Instrumentation and
Measurement, vol.37, no.4, p. 537-40, 1988.

5. C. K. Duffey, R. P. Stratford, "Update of harmonic standard IEEE-519: IEEE
Recommended Practices and Requirements for Harmonic Control in Electric Power
Systems. Industrial Applications Society 35th Annual Petroleum and Chemical
Industry Conference pp. 249-55, 1988.

6. P. Meynaud, G. Hensman, J. L. Javerzac, L. Lagostena, M. J. Lahtinen, F. Lionetto,
W. Monbauer, P. Herager, A. Robert, J. Smid, G. Allen, C. Britten, A. Emanuel,
K. Murotani, P. Ribeiro, P. Wright, Yu Zhelezko, "Equipment producing harmonics
and conditions governing their connection to the mains power supply." Electra,
no.123, p. 20-37, March 1989.

7. T. S. Key, R. J. Ferraro, R. W. Taylor, "Electric power for the 21st Century: changing
loads-changing needs." Power Quality '89. Official Proceedings of the First
International Conference, p. 613, 58-73, 1989.

8. R. J. Gilleskie, "Power quality and emerging electrotechnologies their relationship,
impact and suggested precautions." Power Quality '89. Official Proceedings of the
First International Conference, p. 613, 108-23, 1989.

9. W.E. Kazibwe, R. J. Ringlee, G. W. Woodzell, H. M. Sendaula, "Power quality: a
review." IEEE Computer Applications in Power, vol.3, no.1, p. 39, January 1990.

10. M. S. Padbidri, D. G. Pantulu, "Studies on harmonic pollution in power systems."
Power Systems for the Year 2000 and Beyond. Proceedings of the Sixth National
Power Systems Conference, p. xvi+704, 31-5, 1990.

11. A. Domijan, Jr., E. Embriz-Santander, "Harmonic mitigation techniques for the
improvement of power quality of adjustable speed drives (ASDs)." APEC '90. Fifth
Annual Applied Power Electronics Conference and Exposition. Conference
Proceedings. p. 839, 96-105, 1990.

12. D. Ollivier, "Simulation of harmonic disturbances (in power systems)." Bulletin de la
Direction des Etudes et Recherches, Serie B, no.1, p. 1-31, 1991.

13. J. K. Phipps, J. P. Nelson, P. K. Sen, "Power quality and harmonic distortion on
distribution systems." 1991 Rural Electric Power Conference. Papers Presented at the
35th Annual Conference. p. xi+164, Dearborn, MI, USA; 28-30 April 1991.

14. D. A. Van Den Eijkel, D. A. Marshall, "The simulation and measurement of system
harmonic impedance." CIGRE. Proceedings of the 33rd Session. International
Conference on Large High Voltage Electric Systems, p. 2 vol. 2772, 36-302/1-4
vol.2. Paris, France; 26 Aug.-I Sept. 1990.

15. H. M. Beides, G. T. Heydt, "Power system harmonics estimation and monitoring."
Electric Machines and Power Systems, vol.20, no.2, p. 93-102. March-April 1992.

16. B. M. Hughes, J. S. Chan, D. 0. Koval, "Distribution customer power quality
experience." Conference Record of the 1991 IEEE Industry Applications Society
Annual Meeting. p. 2 vol. 1910, 1556-63 vol.2.Dearbom, MI, USA; 28 Sept.-4 Oct.

17. R. R. Verderber, 0. C. Morse, W. R. A.bir, "Harmonics from compact fluorescent
lamps." Conference Record of the 1991 IEEE Industry Applications Society Annual
Meeting. p. 2 vol. 1910, 1853-8 vol.2. Dearborn, MI, USA; 28 Sept.-4 Oct. 1991.

18. R. Christiansen, "Effects of high levels of harmonics from lighting equipment and
systems." Conference Record of the 1991 IEEE Industry Applications Society Annual
Meeting (Cat. No.91CH3077-5), p. 2 vol. 1910, 1859-62 vol.2. Dearborn, MI, USA;
28 Sept.-4 Oct. 1991.

19. K. Johnson, R. Zavadil, "Assessing the impacts of nonlinear loads on power quality
in commercial buildings-An overview." Conference Record of the 1991 IEEE
Industry Applications Society Annual Meeting (Cat. No.91CH3077-5), p. 2 vol. 1910,
1863-9 vol.2.Dearbom, MI, USA; 28 Sept.-4 Oct. 1991.

20. R. H. Kitchin, S. Etminan, "Computer simulation of harmonic waveform distortion,
and comparison with measured results." IEE Colloquium on 'Simulation of Power
Systems', p. 62, 4/1-7 London, UK; 2 Dec. 1992.

21. R. S. Wallace, "The harmonic impact of variable speed air conditioners on residential
power distribution." APEC '92. Seventh Annual Applied Power Electronics

Full Text
xml version 1.0 encoding UTF-8
REPORT xmlns http:www.fcla.edudlsmddaitss xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.fcla.edudlsmddaitssdaitssReport.xsd
INGEST IEID EDKGIPLV4_XZ3UL5 INGEST_TIME 2013-02-07T16:44:33Z PACKAGE AA00012982_00001