IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. 47, NO. 4, JULY/AUGUST 2011 1983
Comparison Between Different Discharge La
Models Based on Lamp Dynamic Conductan
Cecilio Blanco Viejo, Juan Carlos Alvarez Antón, Arturo Robles, Francisco Ferrero Martín, Memb
Juan Carlos Viera, Member, IEEE, Sounil Bhosle, Member, IEEE, and Georges Zissis, Senior Mem
Abstract—Discharge lamp models based on dynamic lamp con-
ductance are derived from physical equations that describe lamp
behavior. Lamp constructive data are not necessary to build the
models, just the lamp current and voltage data are necessary. In
addition, these models have a relatively low complexity, and not
much calculation time is necessary to obtain them. They can be
used to simulate low- and high-pressure lamps at low and high
(20 kHz < f < 1 MHz) frequencies. Two of these models, de-
rived from different physical equations, are tested and compared
using fluorescent and high-pressure sodium lamps.
Index Terms—Discharge lamp, dynamic conductance, model.
I. INTRODUCTION
D
ISCHARGE lamps are probably the most used lamps
in artificial lighting because of their high luminous ef-
ficiency. They can be operated at different frequencies with
electromagnetic or electronic ballasts. Therefore, to have mod-
els for discharge lamps at different frequencies is important
because these models can be used in ballast design or electrical
power distribution studies to reduce development costs.
A good option is to use models based on the lamp dynamic
conductance G(t) by means of a simple differential equation.
Knowledge of mathematical models for G(t) allows deriving
the current and voltage waveforms of discharge lamps. Physical
assumptions regarding plasma are used to find a mathematical
expression of discharge conductance.
Thesemodelsare usefulat low and high frequencies
(> 20 kHz) because dynamic conductance of discharge lamps
is always ohmic when the supply frequency stays lower than the
Manuscript receivedMay 31, 2010;revisedSeptember28, 2010,
November 9, 2010, and February 23, 2011; accepted February 24, 2011. Date
of publication May 19, 2011; date of current version July 20, 2011. Paper 2010-
ILDC-226.R3, presented at the 2008 Industry Applications Society Annual
Meeting, Edmonton, AB, Canada, October 5–9, and approved for publication
in the IEEE T RANSACTIONS ONINDUSTRY APPLICATIONSby the Industrial
Lighting and Display Committee of the IEEE Industry Applications Society.
C. Blanco Viejo, J. C. A. Antón,F.Ferrero Martín, and J.C. Viera are
with the Electrical Engineering Department, University of Oviedo,33204
Gijon, Spain (e-mail: cecilio@uniovi.es; anton@uniovi.es; ferrero@uniovi.es;
viera@uniovi.es).
A. Robles is with the Mathematics Department, University of Oviedo, 33204
Gijon, Spain (e-mail: aroblespeso@uniovi.es).
S. Bhosle is with OLISCIE, 31600 Lherm, France (e-mail: sounil.bhosle@
oliscie.fr).
G. Zissis is with the University of Toulouse-Laboratoire Plasma et Conver-
sion d’Energie, 31062 Toulouse, France (e-mail: georges.zissis@Laplace.univ-
tlse.fr).
Digital Object Identifier 10.1109/TIA.2011.2155014
plasma frequency [1]. In our case, supply frequency is alw
lower than plasma frequency.
In discharge lamps (low and high pressure), the positive
umn occupies the largest part of the discharge volume. Th
itive column is locally and globally an electrically neutral e
(electron density equal to ion density). Under this conditio
plasma behaves macroscopically as a nonlinear ohmic resi
This hypothesis is valid until the “plasma frequency” is hig
than several hundreds of gigahertz (above this limit value, an
imaginary term appears in the local conductivity expressio
In our case, the operating conditions for lamps are limited to
less than 100 kHz. We should note that a drastic change in
the resistance behavior of the positive column occurs when
the frequency exceeds the “ambipolar diffusion characteris
frequency.” This frequency is situated, for standard lamps, in
the region from 500 Hz to 1.5 kHz (depending on the plasm
composition, dimensions, and pressure). The disappearanc
reignition peaks in the lamp voltage waveform is the expre
of this change.
In this paper, two different models based on lamp dynamic
conductance are compared. Both models are deduced from
plasma physics, while other models currently used are purely
empirical. It should be underlined that model parameters h
a physical meaning and they could be determined from the
theory. However, this task is complex. In addition, fundam
lamp data are missing or uncertain in several cases (partic
for cross sections, radiative coefficients, and so on).
The objective of this paper is to compare two conductanc
models in order to recognize their ability to reproduce lam
behavior at low and high frequencies. The mathematical eff
to calculate parameters for both models are also compared
II. THEORETICAL APPROACHES
Models for G(t) are based on different postulates which t
into account macroscopic variables inside the discharge tu
We are dealing with both fluorescent lampsand high-
pressure sodium (HPS) lamps. The operating pressure is low
(few millibars) in fluorescent lamps,and for this reason, the
plasma is far from local thermodynamic equilibrium condit
In the case of HPS lamps, the pressure is higher (at the lev
atmospheric pressure), and the electron density exceeds a
two orders of magnitude of Griem’s criterion (limit value g
by Griem: 5 × 10
15
cm
−3
) for local thermal equilibrium (LTE)
achievement. Under these conditions, it is not possible to use
the same physical models to describe the local plasma elec
0093-9994/$26.00 © 2011 IEEE