1372 IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 29, NO.

3, JUNE 2014

Experimental Determination of the ZIP Coefficients

for Modern Residential, Commercial, and
Industrial Loads
Abdullah Bokhari, Student Member, IEEE, Ali Alkan, Rasim Dogan, Marc Diaz-Aguiló,
Francisco de León, Senior Member, IEEE, Dariusz Czarkowski, Member, IEEE, Zivan Zabar, Senior Member, IEEE,
Leo Birenbaum, Senior Member, IEEE, Anthony Noel, Member, IEEE, and Resk Ebrahem Uosef, Member, IEEE

Abstract—This paper presents the experimental determination gone through a significant development in ballast technology,
of the ZIP coefficients model to represent (static) modern loads and today electronic ballasts are more popular than magnetic
under varying voltage conditions. ZIP are the coefficients of a ballasts [1]–[5]. In addition to energy savings, electronic bal-
load model comprised of constant impedance , constant current
, and constant power loads. A ZIP coefficient load model is lasts have constant power consumption under voltage variation,
used to represent power consumed by a load as a function of better power regulation, and color consistency. On the other
voltage. A series of surveys was performed on typical residential, hand, magnetic ballasts behave as constant impedance loads.
commercial, and industrial customers in New York City. House- Similarly, outdoor lighting is driving innovation in high-inten-
hold appliances and industrial equipment found in the different sity lighting (HID). Within this segment, there has been a major
locations were tested in the laboratory by varying the voltage from
1.1-p.u. voltage to 0 and back to 1.1 pu in steps of 3 V to obtain the technology shift from mercury vapor to high-pressure sodium
individual – , – , and – characteristics. Customer load and, finally, to metal halide lights. New types of indoor and out-
tables were built using seasonal factors and duty cycles to form door lights are now being introduced to the market as, for ex-
weighted contributions for each device in every customer class. ample, induction lights and light-emitting diode (LED) lights,
The loads found in several residential classes were assembled and which are expected to dominate the lighting industry due to their
tested in the lab. It was found that modern appliances behave quite
differently than older appliances even from only 10 years back. better energy efficiency [3], [6]–[9].
Models of the different customer classes were validated against The objective of this paper is to present a ZIP coefficients
actual recordings of load variations under voltage reduction. model which accurately describes the steady-state behavior of
Index Terms—Commercial class, industrial class, load char- modern loads under varying voltage conditions. An investiga-
acteristic, load composition, load model, residential class, ZIP tion on the effect of varying load is introduced by analyzing
coefficients. individual and composite load structures by means of ZIP coef-
ficients. The work described here is part of a project intended to
estimate the impacts of conservation voltage reduction (CVR)
I. INTRODUCTION for customers and utilities. CVR is a technique commonly used
by power utilities to conserve energy by reducing the voltage
delivered to the loads. The main idea is that loads (devices, ap-
L OAD composition has changed substantially from a few
years back. In the last 10 years, the proliferation of power-
electronic supplies used in many household loads (for example:
pliances, etc.) consume less power when the applied voltage re-
duces. Validation of the composite load model and ZIP coeffi-
cients against field measurements are presented in [27].
flat screen TVs, fluorescent compact lights (CFLs), laptop and
The polynomial expression known as the ZIP coefficients
cell-phone chargers) has modified substantially the way loads
model represents the variation (with voltage) of a load as a com-
behave as the voltage varies [1]–[3]. The lighting industry has
position of the three types of constant loads , , and , ,
and stand for constant impedance, constant current, and con-
stant power loads, respectively. The expressions for active and
Manuscript received March 15, 2013; revised August 28, 2013; accepted
October 01, 2013. Date of publication October 28, 2013; date of current reactive powers of the ZIP coefficients model are
version May 20, 2014. Paper no. TPWRD-00307-2013.
A. Bokhari, A. Alkan, R. Dogan, M. Diaz-Aguiló, F. de León, D. Czarkowski,
Z. Zabar, and L. Birenbaum are with the Department of Electrical and Com- (1)
puter Engineering, Polytechnic Institute of New York University, Brooklyn,
NY 11201 USA (e-mail: [email protected]; rsmdgn marc.
[email protected]; [email protected]; [email protected]; (2)
[email protected]; [email protected]; [email protected]).
A. Noel is with Public Service Electric and Gas Company, Newark, NJ 07102
USA (e-mail: [email protected]). where and are the active and reactive powers at operating
R. E. Uosef is with Consolidated Edison Inc., New York, NY 10003 USA voltage ; and are the active and reactive powers at
(e-mail: [email protected]).
rated voltage ; , , and are the ZIP coefficients for
Color versions of one or more of the figures in this paper are available online
at http://ieeexplore.ieee.org. active power; and , , and are the ZIP coefficients for
Digital Object Identifier 10.1109/TPWRD.2013.2285096 reactive power.

0885-8977 © 2013 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
BOKHARI et al.: EXPERIMENTAL DETERMINATION OF THE ZIP COEFFICIENTS 1373

TABLE I TABLE II
SAMPLE SERVICE CLASSES AND THEIR STRATUM CLASSIFICATION TYPICAL RESIDENTIAL LOAD STRUCTURE

The tasks performed to obtain an accurate load model were:

1) field surveys were conducted for a number of residential,
commercial, and industrial customers in New York City. From
these surveys, information on the composition of their loads was
obtained. A list of the most common appliances and devices
used in each customer class was made; 2) voltage variation tests
(from 1.1 to 0, and then from 0 to 1.1 ) were performed A “load component” is here defined as the aggregate equiva-
in the laboratory on both individual devices, and on composite lent of all devices with similar behavior; for example: electrical
loads, found for the different customer classes; 3) from these ex- heating (ambient and water), motor loads (air conditioners,
periments, ZIP coefficient models were built by fitting quadratic pumps), and power-electronics loads (fluorescent lights, TVs,
functions to the data using a constrained optimization algorithm; chargers). Customers are generally grouped into three major
and 4) actual recordings of load variation under voltage reduc- classes: 1) residential; 2) commercial; and 3) industrial; see
tion were compared with results predicted from the models for [20]. In each class, the load consists of the sum of several
each customer class. different “load components,” each of which contributes some
fraction to the total load; see [21]. As a result, different types
II. LITERATURE REVIEW and quantities of loads may be connected or disconnected in a
Load modeling and load characterization studies have been power system during a day. It is known that the load composi-
performed for a long time. A study performed in 1923 obtained tion of each class exhibits different behavior with the change
a miniature model for an ac network with details of the network in weather conditions, economical situation, and culture [22],
components [10]. The study compared the actual system and the [23].
miniature model under varying load. A study performed in 1973 Three approaches exist for load modeling: measure-
addressed composite loads and load parameters and included the ment-based models, disturbance-based models, and com-
effects of feeders and distribution transformers [11]. The voltage ponent-based models [24], [25]. For the purpose of this study,
and frequency dependency of these composite loads were also the component-based approach is adopted, since it has been
investigated. widely used in load flow and stability studies; has the advan-
In 1992, an IEEE Task Force published a paper on “Load Rep- tage of not requiring field measurement; and can be applied to
resentation for Dynamic Performance Analysis,” summarizing different operating conditions [23].
the current status on power system load modeling [12]. Defini- To implement the component-based approach in this study,
tions of basic load modeling concepts were explained and the customers in New York City were identified based on their load
importance of further developments in load modeling was dis- shape for their hourly demand. Load shape depends on: 1) the
cussed. Numerous studies comparing the performance of load type of day (weekday or holiday); 2) service class (SC), which
modeling have been published (see [12]–[15]). A mathemat- represents a group of customer types with similar load charac-
ical expression relating load characteristics to depressed voltage teristics (i.e., residential, small commercial, large commercial,
levels by means of the sum of constant impedance (Z), constant and industrial); 3) stratum: each service class is decomposed
current (I), and constant power (P) components, was introduced into subgroups, or strata, within a service class based on cus-
by Kalinowsky and Forte in 1981 [16]. However, the use of such tomer size, as measured by a particular billing quantity (details
a linear combination was first introduced by Kent et al. in 1969 are given below); and 4) temperature. Table I shows samples of
[17]. The ZIP coefficients model has been widely used to rep- service classes with their stratum billing variable.
resent the relation between voltage and power characteristics of
loads [18], [19]. IV. LOAD SURVEYS
To investigate load profiles of each customer class in a tar-
III. CUSTOMER CLASSES geted network, surveys were conducted in several sites of the
Load composition of customers plays a significant role in de- network. This gave an accurate estimation of load composition.
termining the appropriate load model. Load composition varies Onsite visits were performed on different customer classes, and
according to many factors such as the type, size, behavior of the information on the details of each customer load was recorded.
customer, and recent advances and upgrades in equipment tech- The surveys were aimed to obtain the nature of loads in each
nology. In this study, load composition is organized based on customer class, the electrical data, the total number of equip-
the customer class with a recognized load profile. ment being used, and how often those pieces of equipment were
1374 IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 29, NO. 3, JUNE 2014

TABLE III • elevators: hydraulic, pneumatic, and traction;

CLASSIFICATION OF RESIDENTIAL STRATA BASED ON ANNUAL CONSUMPTION • air handling: HVAC, chillers, fans, and heat pumps;
• pumps: hot and cold water circulation, chiller pumps, fire
pumps, and centrifugal pumps;
• compressors: air compressors and industrial freezers;
• household appliances: refrigerators, freezers, warmers,
ovens, and microwaves;
• power-electronics devices: power supplies, chargers, TVs,
game consoles, computers, and peripherals.
For the purpose of creating an accurate customer load model,
each customer load was classified based on running load versus
annual load. Since running loads change depending on the
temperature, seasons were chosen according to the climate
of New York City. Load tables for every surveyed customer
were formed into seasonal results. Since climate and energy
consumption are closely coupled, each customer class was
formed with three seasons: spring/fall, summer, and winter, as
loads and consumption vary depending on the season. Spring
and fall seasons were combined together, since there is no
significant difference between them.
Using the survey information, load tables were built with load
component information for every season. Each load component
was correlated with a duty cycle, and with a seasonal use factor,
since loads do not run all of the time, and their characteristics
Fig. 1. Reconstructed active and reactive power curves for small commercial vary. The total seasonal energy consumption was computed by
customers. These power curves use the ZIPs coefficients obtained with actual
cutoff voltage reported in the second half of Table VII.
summing the correlated load component power consumptions.
This was done in order to build an accurate detailed load model
for each class, taking into account climate change, energy con-
TABLE IV sumption, and the running loads during a season. The load tables
EQUIPMENT CONTRIBUTION FOR HIGH SCHOOL AND HOTEL
so obtained were validated by comparing the calculated hourly
power consumption with utility records of hourly energy con-
sumption for a whole year for every surveyed customer. Con-
sumption reports are divided based on defined seasons, and the
average hourly peak load for every season was formed. Seasons
are defined as follows: spring/fall: March, April, May, June, Oc-
tober, and November; summer: July, August, and September;
winter: December, January, and February. The energy consump-
tion for every season matched very well with the actual sea-
sonal average peak kilowatt-hours for the surveyed customers.
The constructed load table for a typical residential customer is
shown in Table II.
In Table II, the duty cycle indicates the relative time that
the equipment is connected. Season factor indicates the rela-
tive time that the equipment is in use. Season factor varies as
load configuration changes with season (spring/fall, summer,
and winter). Both season factor and duty cycle are fractions of
a 24-h period, and have a value between 1 and 0.
The actual power of a single item of equipment, and the total
connected or disconnected during the day. In addition, the be- component power in a class, is calculated as
havior of each load class varies with the change of weather, type
of activity, and economic situation. Survey data provided guid-
ance for the selection of the equipment to test in the laboratory. Actual equipment power
The survey data, for a service class, were extracted and cate- equipment rated power
gorized based on equipment characteristics. The types of loads duty cycle season factor (3)
observed during the surveys fall into the following categories:
Total component power
• lighting equipment: fluorescent, incandescent, halogen,
compact fluorescent lights (CFL), and high-intensity dis- actual equipment power
charge (HID) lights; Total number of equipment (4)
BOKHARI et al.: EXPERIMENTAL DETERMINATION OF THE ZIP COEFFICIENTS 1375

TABLE V
EQUIPMENT CONTRIBUTION WEIGHT IN% PER CLASS (SUMMER)

TABLE VI A. Residential Customers

ACTIVE AND REACTIVE ZIP MODEL PER CUSTOMER CLASS (SUMMER)
Surveys were performed on different size dwellings (ranging
from studio apartments to a large house) to determine, in detail,
which household appliances were used in each dwelling. This
information was used to construct a set of residential load tables
depending on the size and power consumption of the residential
unit.
Consolidated Edison Inc (the utility company of New York
City) classifies the residential class into six subclasses (or strata)
A, B, C, D, E, and F based on annual peak power consumption
records; see Table III.

B. Commercial Customers
Commercial sites are designated as either “large” or “small”
by Consolidated Edison. Each of these two groups is divided
into different subclasses, according to their annual peak power,
as for the residential subclasses. Surveys were conducted of
four small commercial businesses: a supermarket; a restaurant;
a laundromat; and an optical store. Also, two large commercial
establishments (a school and a hotel) were surveyed.
1376 IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 29, NO. 3, JUNE 2014

TABLE VII
ACTIVE AND REACTIVE ZIP MODEL. FIRST HALF OF THE ZIPS WITH 100-V CUTOFF VOLTAGE.
SECOND HALF REPORTS THE ZIPS WITH ACTUAL CUTOFF VOLTAGE

Detailed information was obtained. Load tables for each tion was computed. Utility records of annual consumption with
commercial customer were built and the total energy consump- hourly recordings for each surveyed location exist. Comparison
BOKHARI et al.: EXPERIMENTAL DETERMINATION OF THE ZIP COEFFICIENTS 1377

Fig. 2. Residential strata constructed active (left) and reactive (right) power
curves. These power curves are using the ZIPs coefficients obtained with actual
cutoff voltage reported in the second half of Table VII.

Fig. 4. Comparison of active and reactive powers between old and new
appliances.

a higher variation of active and reactive power under voltage

reduction. The supermarket load is dominated by the HVAC
system with a constant power load behavior. The laundromat
almost has a constant current load model. Load topology of
a small commercial class varies depending on their type of
business activity and their location in New York City. These
different load combinations were used to define multiple small
commercial classes with a weighted percentage given to each
customer load table in order to create a general small commer-
cial load model. A formula for computing small commercial
class ZIP coefficients is provided below; see (7)
2) Large Commercial Customers: Load tables for those sur-
veys were constructed and checked against utility records of the
Fig. 3. Active and reactive test results with constrained curve fitting. The ZIP average consumption for the three seasons. Table IV shows the
curve with the 100-V cutoff is shown in the solid line and ZIP with the actual
cutoff voltage in the dashed line. The two sets of ZIPs are shown in Table VII. percent contribution of equipment for the two customers under
study.
Customers in this class have high energy demand and dif-
of these records with computed power consumption for each ferent load behavior depending on the type of business. In gen-
season showed very good agreement. eral, to fully represent loads in this category, public as well
1) Small Commercial Customers: Commercial customers as private customers should be included in the survey. In ad-
vary depending on the type of load used by them to conduct dition, heating, cooling, and street lighting loads largely con-
their business activities. For example, laundromat loads are tribute in this class, but may differ because of building struc-
mostly motor driven, while the load of a restaurant is pri- ture and construction year. To take into account the different
marily resistive. The selection of surveyed customers included customer load compositions in this class, an electrically heated
businesses with: an HVAC system with resistive heating; a customer (hotel) and a gas-heated customer (high school) were
centralized ac system with gas heating, and a normal air con- surveyed.
ditioning system. This was done in order to ensure a complete
investigation of the small commercial class. C. Industrial Customers
Fig. 1 shows how active and reactive powers vary for Two large industrial customers were surveyed. One is a
each surveyed customer. Customers having mainly constant pipeline transportation and energy storage customer and the
impedance equipment (i.e., restaurant and optical store) have other is a water treatment plant. Most of the loads found in the
1378 IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 29, NO. 3, JUNE 2014

plants are motor-driven pumps and compressors. The collected and a controllable dynamometer was controlled to offer con-
load information for the power plants was used to formulate stant torque.
the load table and to compute equipment weight. The total During the voltage reduction procedure, loss of load func-
calculated energy consumption matched utility records of the tionality determined the cutoff voltage . We noted,
average seasonal peak in the annual metered demand. however, that some pieces of equipment continued to consume
In general, equipment weight percentages and class ZIP co- reactive power even when no work was delivered; examples
efficients were calculated as follows: are induction motors and air compressors. Other loads, such as
ballasts and LCD TVs, did not fully stop working, but mani-
fested fluctuations. These experiments provided the correlation
between voltage and power (active and reactive) of those loads.
Equipment weight The recorded data that fell into the range between 100 to 130 V
load component power were used for the curve-fitting process. This is so because the
(5)
class power final objective of the project is to assess the load behavior for
conservation of voltage reduction studies. Table VII shows the
equipment tested in the lab and the generated ZIP coefficients.
(6)
B. Assembled Customer Classes

By using the survey information, several customer class loads

(7) were reconstructed in the laboratory. Residential strata A, B,
C, and D were assembled and voltage reduction tests were per-
formed following the same procedure used for isolated appli-
(8) ances. To ensure the accuracy of customer models, the assem-
bled load composition test results were compared with the re-
Equipment contribution weight in percent per class is shown constructed load shape of the strata. The reconstruction process
in Table V for the summer season and the ZIP model per cus- started with the calculation of active power and reactive
tomer class in Table VI. power demand for each appliance used in the particular res-
idential load under study. Next, active and reactive powers were
calculated for each load. A simple calculation of the active and
reactive powers as a function of voltage for an LCD television
V. EXPERIMENTS
is shown in the following equation:
The setup for the laboratory experiments consisted of a power
source, a power analyzer, and computer software. These three
elements were used in automated testing to perform tests under (9)
identical conditions so that reliable results would be obtained.
Voltage reduction tests were performed on each device, and
the behavior of these loads was recorded. Prior to each test,
loads were operated under nominal voltage in steady state for (10)
30 min to avoid the transient behavior of the cold start. Then,
the voltage was decreased from 130 to 0 V in steps of 3 V and where is the input voltage and varies from 130 to 100 V in
kept for 20 s at each step. A settling down period was allowed steps of 3 V.
at each voltage step to prevent any undesired transient. For each Then, the sum of active and reactive powers for each appli-
step, voltage , current , active power , and re- ance in the load table was calculated with
active power were recorded.

A. Isolated Appliances and Equipment

(11)
To create a load profile for each surveyed customer, the ef-
fect of varying load demand on equipment was investigated.
Laboratory experiments were performed on single-phase and on
three-phase loads. For several of these tests, loads were built (12)
in order to simulate real operating conditions. In most cases,
the actual device was used for the test. However, some of the Then, active and reactive power curves were plotted and com-
motor loads were emulated. For example, a water pump system pared with measurements of the assembled load composition
was constructed using a centrifugal pump connected to a storage tests. Fig. 2 shows the reconstructed active and reactive power
tank through pipes; then pressure was applied using valves. The curves using the computed ZIP coefficients for different resi-
operation of elevators was emulated with an induction motor dential strata.
BOKHARI et al.: EXPERIMENTAL DETERMINATION OF THE ZIP COEFFICIENTS 1379

VI. CALCULATION OF ZIP COEFFICIENTS and 0.8 input voltage. The behavior of reactive power
for magnetic and electronic ballasts has also changed as old bal-
A. Method lasts resembled constant current loads while the reactive power
To obtain this model, for each appliance, active power and of new ballasts tends to increase or decrease depending on the
reactive power for different voltage levels were measured (see supplied voltage level.
Fig. 3). With the fast growth of switching power supplies used in
LCD TVs and laptop chargers, the consumption and behavior
B. Calculation of Parameters differ from old appliances (CRT TVs). Some appliances today
If the equipment is a three-phase load, the nominal voltage are equipped with power factor correction; for example, the
for the fit is set to 208 V. On the other hand, if the load is a LCD TVs have a slightly leading power factor. By comparing
single-phase load, the nominal voltage is set to 120 V. To obtain tests of the CRTs from the old study with the new test for the
the values of and corresponding to 120 V or 208 V, we LCD TV, one can see that the reactive power behaves quite dif-
performed a preliminary least-square fit in the nominal voltage ferently. Most of the compared appliances indicate that major
region . To increase the accuracy of the modeling changes have occurred in their active and reactive power be-
procedure, a constrained least square was introduced with an havior. The study shows that it is essential to update the model
optimization routine implemented to deliver the best fit with the for appliances to obtain the correct analysis and design of the
constraint that the sum of the three ZIP coefficients has to add network.
to 1. The optimization problem is formulated as follows:
VII. CONCLUSION
The paper has presented experimentally verified ZIP coeffi-
cient models for the most commonly used appliances and for
(13) different customer classes. Surveys were performed to deter-
mine which appliances and pieces of equipment were available
Subject to (14)
in the different dwellings and businesses in New York City. The
and loads found in several residential classes were assembled and
tested in the lab. It was found that modern appliances behave
quite differently than older appliances even from 10 years ago.
The model of the different customer classes has been validated
against actual recordings of load variations under voltage reduc-
(15) tion in several networks served by Con Edison.
Subject to (16)
REFERENCES
Fig. 3 shows active and reactive power curves of the voltage [1] K. P. Schneider, J. C. Fuller, and D. P. Chassin, “Multi-state load
reduction tests for selected pieces of equipment. The figure also models for distribution system analysis,” IEEE Trans. Power Syst.,
shows the corresponding fitted curves obtained between 100 and vol. 26, no. 4, pp. 2425–2433, Nov. 2011.
[2] F. L. Quilumba, W. Lee, H. Huang, D. Y. Wang, and R. L. Szabados,
130 V (to cover the range for CVR studies) and between the “Load model development for next generation appliances,” in Proc.
actual cutoff voltage and 130 V to be used for other studies; for Ind. Appl. Soc. Annu. Meeting, 2011, pp. 1–7.
example, cold-load pickup studies. [3] B. Zhao, Y. Tang, W. Zhang, and Q. Wang, “Modeling of common
load components in power system based on dynamic simulation exper-
iments,” in Proc. Int. Conf. Power Syst. Technol., 2010, pp. 1–7.
C. Comparison Between Old and New Loads [4] Calif. Energy Comm., “2008 building energy efficiency standards for
A study similar to the one described here was performed in residential and nonresidential buildings,” Sacramento, CA, USA, Rep.
no. CEC-400-2008-001-CMF, Dec. 2008.
1999 for Consolidated Edison for different appliances and to de- [5] U.S. Dept. Energy, “CFL market profile: Data trends and market in-
termine residential and commercial models as static loads [26]. sights,” Silver Spring, MD, USA, Sep. 2010, D&R International Ltd.
Due to the rapid change in technology, load models needed to be [6] Philips Advance, “Pocket guide to high intensity discharge lamp bal-
lasts, Philips lighting electronics,” Rosemont, IL, USA, RT-8100-R02
updated to account for the variation of power system load char- 12/09, 2009.
acteristics. A comparison is now presented between the new re- [7] C. Dilouie, What’s New in HID Lamps and Ballasts, Jul. 17, 2012.
sults and those of the previous study to see how the behavior of [Online]. Available: http://lightingcontrolsassociation.org/whats-
new-in-hid-lighting/
modern loads differs from that of older loads. [8] W. Yan, S. Y. R. Hui, and H. Chung, “Energy saving of large-scale
Fig. 4 shows a comparison of selected appliances; one can high-intensity-discharge lamp lighting networks using a central reac-
note important differences. Within the past decade, electronic tive power control system,” IEEE Trans. Ind. Electron., vol. 56, no. 8,
pp. 3069–3078, Aug. 2009.
ballasts with their constant power behavior replaced magnetic [9] “Standard load models for power flow and dynamic performance sim-
ballasts that follow a constant current curve. New electronic bal- ulation,” IEEE Trans. Power Syst., vol. 10, no. 3, pp. 1302–1313, Aug.
last technology has significantly improved from 10 years ago 1995.
[10] O. R. Schurig, “A miniature A-C. transmission system for the practical
with a better active power regulation over a variety of input solution of networks and transmission-system problems,” Trans. Amer.
voltages. One of the most important characteristics of new elec- Inst. Elect. Eng., vol. 42, pp. 831–840, Jun. 1923.
tronic ballast circuits is the degree to which they control the [11] G. J. Berg, “Power-system load representation,” Proc. Inst. Elect. Eng.,
vol. 120, no. 3, pp. 344–348, 1973.
lamp power with changes in input voltage. Tests show that new [12] “Load representation for dynamic performance analysis of power sys-
electronic ballasts consume constant active power between 1.1 tems,” IEEE Trans. Power Syst., vol. 8, no. 2, pp. 472–482, May 1993.
1380 IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 29, NO. 3, JUNE 2014

[13] Y. Li, D. Chiang, B. K. Choi, Y. T. Chen, D. H. Huang, and M. G. Marc Diaz-Aguiló was born in Barcelona, Spain. He received the M.Sc. degree
Lauby, “Representative static load models for transient stability anal- in telecommunications engineering from the Technical University of Catalonia
ysis: Development and examination,” Proc. Inst. Elect. Eng., Gen., (UPC), Barcelona, Spain, in 2006 and the M.Sc. degree in aerospace controls
Transm. Distrib., vol. 1, no. 3, pp. 422–431, May 2007. engineering from a joint program between Supaero, Toulouse France, and the
[14] A. P. Alves da Silva, C. Ferreira, A. C. Zambroni de Souza, and G. Massachusetts Institute of Technology, Cambridge, MA USA, in 2008, and the
L. Torres, “A new constructive ANN and its application to electric Ph.D. degree in aerospace simulation and controls from the Technical University
load representation,” IEEE Trans. Power Syst., vol. 12, no. 4, pp. of Catalonia, Barcelona, Spain, in 2011.
1569–1575, Nov. 1997. Currently, he is a Postdoctoral Researcher at the Polytechnic Institute of New
[15] M. Sadeghi and G. A. Sarvi, “Determination of ZIP parameters with York University, Brooklyn, NY, USA. His research interests are in power sys-
least squares optimization method,” in Elect. Power Energy Conf., tems, controls, smart-grid implementations, and large systems modeling and
2009, pp. 1–6. simulation.
[16] S. A. Kalinowsky and M. N. Forte, “Steady state load-voltage charac-
teristic field tests at area substations and fluorescent lighting compo-
nent characteristics,” IEEE Trans. Power App. Syst., vol. PAS-100, no.
6, pp. 3087–3094, Jun. 1981.
[17] M. H. Kent, W. R. Schmus, F. A. Mccrackin, and L. M. Wheeler, “Dy-
namic modeling of loads in stability studies,” IEEE Trans. Power App. Francisco de León (S’86–M’92–SM’02) received the B.Sc. and the M.Sc.
Syst., vol. PAS-88, no. 5, pp. 756–763, May 1969. (Hons.) degrees in electrical engineering from the National Polytechnic
[18] L. M. Hajagos and B. Danai, “Laboratory measurements and models of Institute, Mexico City, Mexico, in 1983 and 1986, respectively, and the Ph.D.
modem loads and their effect on voltage stability studies,” IEEE Trans. degree in electrical engineering from the University of Toronto, Toronto, ON,
Power Syst., vol. 13, no. 2, pp. 584–592, May 1998. Canada, in 1992.
[19] S. J. Ranade, A. Ellis, and J. Mechenbier, “The development of power He has held several academic positions in Mexico and has worked for the
system load models from measurements,” in Proc. Transm. Distrib. Canadian electric industry. Currently, he is an Associate Professor at the Poly-
Conf. Expo., 2001, vol. 1, pp. 201–205. technic Institute of New York University, Brooklyn, NY, USA. His research
[20] E. Vaahedi, M. A. El-Kady, J. A. Libaque-Esaine, and V. F. Carvalho, interests include the analysis of power phenomena under nonsinusoidal con-
“Load models for large-scale stability studies from end-user consump- ditions, the transient and steady-state analyses of power systems, the thermal
tion,” IEEE Trans. Power Syst., vol. 2, no. PWRS-4, pp. 864–870, Nov. rating of cables, and the calculation of electromagnetic fields applied to ma-
1987. chine design and modeling.
[21] W. W. Price, K. A. Wirgau, A. Murdoch, J. V. Mitsche, E. Vaahedi,
and M. El-Kady, “Load modeling for power flow and transient stability
computer studies,” IEEE Trans. Power Syst., vol. 3, no. 1, pp. 180–187,
Feb. 1988.
[22] C. Concordia and S. Ihara, “Load representation in power system sta- Dariusz Czarkowski (M’97) received the M.Sc. degree in electronics from
bility studies,” IEEE Trans. Power App. Syst., vol. PAS-101, no. 4, pp. the AGH University of Science and Technology, Cracow, Poland, in 1989, the
969–977, Apr. 1982. M.Sc. degree in electrical engineering from Wright State University, Dayton,
[23] P. Kundur, Power System Stability and Control. New York: Mc- OH, USA, in 1993, and the Ph.D. degree in electrical engineering from the Uni-
Graw-Hill, 1994, pp. 17–279. versity of Florida, Gainesville, FL, USA, in 1996.
[24] J. R. Ribeiro and F. F. Lange, “A new aggregation method for deter- In 1996, he joined the Polytechnic Institute of New York University,
mining composite load characteristics,” IEEE Trans. Power App. Syst., Brooklyn, NY, where he is currently an Associate Professor of Electrical and
vol. PAS-101, no. 8, pp. 2869–2875, Aug. 1982. Computer Engineering. He is a coauthor of Resonant Power Converters (Wiley,
[25] “EPRI report load modeling for power flow and transient stability com- 2011). His research interests are in the areas of power electronics, electric
puter studies,” New York, EPRI Final Rep. EL-5003, Jan. 1987, vol. 2. drives, and power quality.
[26] D. Shmilovitz, J. Duan, D. Czarkowski, Z. Zabar, and S. Lee, “Charac-
teristics of modern nonlinear loads and their influence on systems with
distributed generation,” Int. J. Energy Technol. Policy, vol. 5, no. 2, pp.
219–240, 2007.
[27] M. Diaz-Aguilo, J. Sandraz, R. Macwan, F. d. León, D. Czarkowski, C.
Comack, and D. Wang, “Field validated load model for the analysis of Zivan Zabar (M’76–SM’81) was born in Hadera, Israel, in 1939. He received
cvr in distribution secondary networks: Energy conservation,” in IEEE the B.Sc., M.Sc., and D.Sc. degrees in electric power and power electronics
Trans. Power Del., accepted for publication. from Technion-Israel Institute of Technology, Haifa, Israel, in 1965, 1968, and
1972, respectively.
Abdullah Bokhari (S’12) received the B.Sc. degree in electrical engineering He is Professor of Electrical Engineering at the Polytechnic Institute of New
from King Saud University, Riyadh, Saudi Arabia, in 2004 and the M.Sc. degree York University, Brooklyn, NY, USA. He served as the Head of the Electrical
in electrical engineering from the Polytechnic Institute of New York University, and Computer Engineering Department, Polytechnic Institute of New York Uni-
Brooklyn, NY, USA, in 2009, where he is currently pursuing the Ph.D. degree versity, for three years (from 1995 to 1998). He has six patents and more than 50
in electrical engineering. papers published in technical journals. His areas of interest are electric power
His research interests include power system modeling and analysis, power systems, electric drives, and power electronics.
theory, and electrical machines. Dr. Zabar is a member of Sigma Xi.

Ali Alkan was born in Konya, Turkey. He received the B.Sc. degree in elec-
trical and electronics engineering from TOBB University of Economics and Leo Birenbaum (S’45–A’48–M’55–SM’70) was born in New York City in
Technology, Ankara, Turkey, in 2008 and the M.Sc. degree in electrical en- 1927. He received the B.E.E. degree from the Cooper Union, New York, USA,
gineering from Polytechnic Institute of New York University, Brooklyn, NY, in 1946, and the M.E.E. and M.S. (Phys.) degrees from Polytechnic Institute,
USA, in 2012. Brooklyn, NY, USA, in 1958 and 1974, respectively.
Currently, he is Professor Emeritus at the Polytechnic Institute, where, for
many years, he taught courses in electric circuits, electromechanical power con-
version, electromagnetic fields, and rotating machinery. He has conducted re-
Rasim Dogan received the B.Sc. degree in electrical and electronics engi- search in a number of areas: microwave components and transmission, biolog-
neering from Gazi University, Ankara, Turkey, the M.Sc. degree in electrical ical effects of microwave and low-frequency electromagnetic fields, electro-
engineering from the Polytechnic Institute of New York University, Brooklyn, magnetic launchers, and electric power distribution. He is a coauthor of ap-
NY, USA, and is currently pursuing the Ph.D. degree in electrical engineering proximately 40 peer-reviewed papers, and he holds three patents on microwave
from Polytechnic Institute of New York University. devices.
His research interests are power system load modeling and the calculation of Prof. Birenbaum is a member of Sigma Xi, Tau Beta Pi, the BioElectroMag-
electromagnetic fields applied to machine design. netics Society, and the New York Academy of Sciences.
BOKHARI et al.: EXPERIMENTAL DETERMINATION OF THE ZIP COEFFICIENTS 1381

Anthony Noel (M’81) was born in Brooklyn, NY, USA. He received the B.Sc. Resk Ebrahem Uosef (M’01) received the B.Sc. and M.Sc. degrees in electrical
degree in electrical engineering from Columbia University, New York, USA, in engineering from Alexandria University Faculty of Engineering, Alexandria,
1981. Egypt, in 1979 and 1981, respectively, and a second M.Sc. degree in electrical
In 1993, he completed the General Electric Power Systems Engineering engineering, and the Ph.D. degree in electrical engineering from Polytechnic
Course, General Electric, Schenectady, NY, USA. From 1982 to 2013, he University, Brooklyn, NY, USA, in 2007 and 2011, respectively.
was an Electrical Engineer with Consolidated Edison, New York, involved in He was an Engineer in a hydropower generating station in Egypt, and then
distribution system planning and analysis. Presently he is an Asset Management he was the owner of a consulting firm for an electric construction company
Expert with Public Service Electric & Gas, Newark, NJ, USA. in Egypt. He joined Con Edison’s Distribution Engineering Department, New
York, USA, in 2003 and is currently responsible for Con Edison’s distribution
system design and analysis.
Dr. Uosef is a Registered Professional Engineer in the State of New York.