SEMINARYA "Light for Architecture and Urban Space: Design, quality and energy efficiency “ Energy efficiency in lighting with LED technologies: Deepening in the case of public lighting.

1. The source of this presentation is the following journal paper: Electric Power Systems Research Volume 96, 2013, Pages 16-
22 “A novel technique for energy savings by dimming high pressure sodium lamps mounted with magnetic ballasts using a
centralized system” (Review). Authors: Alessandro Burgio, Daniele Menniti, Department of Electronic, Computer and System
Science, University of Calabria, Via Pietro Bucci, cubo 42C, 87036 Arcavacata of Rende - (CS), Italy.
Abstract The return on investment from adopting centralized energy saving systems for street lighting can be compromised
by voltage drop in long distribution lines. Large, centralized systems can reduce the output voltage sufficiently to dim distant
high intensity discharge lamps. In this paper, the authors explore alternative dimming techniques and study the behavior of a
400 W HPS lamp when supplied by trapezoidal or rectangular voltages rather than the conventional sinusoidal supply. The
efficiency and effectiveness of these alternatives are assessed by means of laboratory experiments.
High-Intensity Discharge Lamps in Centralized Systems: Exploring New
Techniques for Operating, Controlling and Driving
Dr. Alessandro Burgio
Department of Electronic, Computer and System Science, University of Calabria
SEMINARY
A "Light for Architecture and Urban Space: Design, quality and energy efficiency “
Energy efficiency in lighting with LED technologies: Deepening in the case of public lighting.

2. When the instantaneous value of the supply voltage is greater
than a given value (+Vk in the positive half-period, −Vk in the
negative half-period) the process of glow discharge begins. Such
a phenomenon is characterized by a significant and sudden
increase in current flow. In the presence of an electric arc, the
voltage between electrodes falls to values typically around 10–
40V, while at the center of the current column, the current
density can reach ten thousand amperes per square centimeter
at a temperature of several thousand degrees Celsius. If voltage
and current are not adequately controlled by the supply circuit,
the transition from the glow discharge to the arc discharge is
discontinuous (causing the well-known flickering effect seen in
aging street lamps) as well as the conduction mode (causing an
unstable current flow). To avoid flickering, an inductor is placed
in series with the lamp. Termed as magnetic ballast, the inductor
ensures a high quality of lighting and a continuous-conduction
mode.

3. A magnetic ballast is an inductor that is wound on a
ferromagnetic core with an air gap. Typically placed in series
with the lamp, the magnetic ballast significantly increases the
phase angle between voltage and current, thereby
decreasing the power factor to values around 0.6–0.4. In
order to reduce the phase angle, a capacitor is connected in
parallel to the magnetic ballast and lamp. To adjust the light
output of an HPS lamp using a magnetic ballast and a
capacitor, one of two techniques is typically used. The first
technique consists of reducing the amplitude of the supply
voltage Vac to decrease the energy provided to the lamp.
However, the voltage amplitude cannot be reduced below
the value Vk = 120 V because establishing the arc between
the two electrodes requires a minimum supply voltage to
avoid extinguishing the arc. The second technique
adjusts the light output of the HPS lamps controlling
the current Iac. This is accomplished by increasing the
reactance of the magnetic ballast while maintaining the
supply voltage. This reduces the current Iac and therefore
the energy provided to the lamp. To increase the
reactance, an additional inductor L1 is connected in
series to the inductor L. By opening the switch S, the
reactance changes from the value ωL to ω(L + L1), and
the lamp is operated in a reduced power mode.
Therefore, two operating levels are defined by the values
of L and L1.

4. Modern centralized energy saving systems ensure the
functioning of street lighting even in the presence of voltage
sags that occur on the utility side. However, because they
achieve energy savings by reducing the voltage amplitude
as in the case of the adjustable-ratio transformers, they
otherwise suffer from the same limitations. To overcome this
limit, a supply voltage having a trapezoidal rather than
sinusoidal waveform can be adopted.

5. A comparative study of the response of a test system when supplied by a rectangular or trapezoidal voltage rather
than a conventional sinusoidal voltage is presented. The test system consists of a 400 W high pressure sodium lamp
in series with a magnetic ballast having an inductance equal to L = 110 mH and a parallel capacitor having a
capacitance equal to C = 50 F. The goal of this comparative study is to identify new techniques for dimming HPS
lamps using centralized energy saving systems for street lights. The effectiveness and the efficiency of these new
techniques are assessed by means of laboratory experiments conducted according to the following steps: (a) ensuring
all conditions to safely exercise the lamp when an unconventional voltage, i.e., rectangular or trapezoidal, is
considered; (b) determining the values of parameters that define the unconventional voltage so that the lamp emits its
own nominal light output; and (c) modifying the values of parameters to operate the lamp in a reduced power mode.
Laboratory results are divided into three sets. In the first set, the conventional sinusoidal supply voltage is
considered, and in the second and third sets, a rectangular and a trapezoidal supply voltage are studied,
respectively. Voltage and current are accurately measured capturing 1028 samples a period of the fundamental
(the sampling rate is 51.400 kHz) and using an oscilloscope type HP mod. 54602B 150 MHz mounted with a
GBIP interface type Agilent mod. 54657A. Collected data are so used to determine the harmonic voltage and
current distortion via Matlab. According to the type of laboratory tests and studies, all powers are calculated in
accordance with the definitions provided by the IEEE Working Group on nonsinusoidal situations, practical
definitions for powers in system with nonsinusoidal waveforms and unbalanced loads.

6. The test system was supplied with a sinusoidal voltage having an amplitude and
frequency equal to 325.270 V and 50.000 Hz respectively; the laboratory test
measurements are shown in Table 1 and are assumed as a reference in subsequent
tests. The amplitude and frequency of the sinusoidal supply voltage were adjusted, one
at a time, to dim the lamp and reduce the light output and the energy consumption.
First, the amplitude was adjusted and the authors name this subset of experiments
SinusV. The results obtained in the laboratory by implementing this technique are
presented in Table 2. The reduction of the energy consumption that results from
dimming the HPS lamp can also be achieved by increasing the frequency instead of
reducing the amplitude of the sinusoidal supply voltage; the authors name this subset
of experiments SinusF. Varying frequency is not a technique commonly used to dim
HPS lamps mounted with magnetic ballasts and capacitors, but it is also an effective
technique as demonstrated by the results of the laboratory experiments reported in
Table 3.

7. When the normalized frequency of the sinusoidal supply voltage
increases from 1 (exactly 50.000 Hz) up to 1.240 (approximately
62.000 Hz), the lamp continues to operate properly but at a reduced
power level. Setting f = 1.240, the measured illuminance is
approximately 0.600 (a reduction of about 40% with respect to the
nominal value), and the apparent power S is 0.850. This technique
has the great advantage of ensuring the perfect gas discharge even
in a reduced power mode because the voltage amplitude remains
unchanged and equal to 1. Unfortunately, the experimental results
indicate that this technique does not scale well; above 1.240, the
apparent power grows rapidly. Indeed, when the frequency is set to
greater values, the illuminance and the active power continue to
decrease, but the apparent power increases. This trend is because,
for frequencies greater than 1.240, the capacitive reactance
decreases, leading to increased reactive power. Effectively, increasing
frequency detunes the RLC circuit formed by the lamp and ballast,
leading to a significant increase in the apparent power S. In the
system response, the current THD remains almost constant, while the
current I1 tracks the apparent power S.

8. In the second set of experiments, the test system is supplied by
a rectangular voltage. Such a voltage waveform is defined by only
three parameters, its amplitude, frequency and pulse duration (Ton),
all of which can be appropriately adjusted to operate the lamp at a
nominal or reduced power mode. To operate the test system safely,
the parallel capacitor must be removed; otherwise, the current drawn
by the capacitor, being proportional to the derivate of voltage with
respect to time, would have dangerously high peak values. Once the
capacitor was removed, the authors first set the frequency of the
rectangular waveform equal to 1 and adjusted the amplitude and Ton
so to measure the nominal illuminance. Several couples of values of
amplitude and Ton were determined. As expected, in these couples
the voltage amplitude increases with the decreasing of Ton. In
particular, the amplitude increases to 1.070 when Ton is set to equal
5.200 ms.

10. The test system response is studied under the trapezoidal supply
voltage. Such a voltage waveform is compatible with the presence of
the capacitor provided that the phase angle ˛ is lower than 78◦. In fact,
for values greater then 78◦, the current peak became dangerously
high, and safe operation of the experiments was no longer possible.
Taking into account the voltage drop along a long distribution line,
typically around 3–4% for a well designed feeder line, the amplitude
of the trapezoidal voltage was set to 1.030, i.e., 3% greater than the
nominal value. Then, to determine the value of angle, the authors
considered a trapezoidal voltage having the same rms value as the
reference sinusoidal voltage so that the angle ˛ would consequently
be 39◦. The frequency of the trapezoidal voltage was initially set to
equal 1.

11. To dim the HPS lamp and save energy, the test system response was first studied when reducing the
amplitude of the trapezoidal Voltage. The test system can be operated in reduced power mode by
increasing the frequency of the trapezoidal supply voltage rather than reducing the amplitude; in this
subset of experiments, named TrapF, the amplitude is set to 1.030, and the frequency is increased
from 1.000 to 1.240.

12. The comparison among these waveforms takes into account the real and the apparent power (relevant for tax purposes) and
the rms current (relevant for a proper sizing of power lines). The real power delivered to the test system decreases with the
illuminance for all waveforms. The lower the value of real power measured when RectangA is adopted, the higher the value of
apparent power measured. Indeed, the apparent power delivered to the test system under RectangA is about 3–4 times the
apparent power measured in the case of SinusV, SinusF and TrapV; the same consideration is also valid for the rms current.
Therefore, although RectangA could be used to operate HPS lamps at full or reduced power, such a supply voltage waveform
and the related technique of adjusting the amplitude would be very inefficient. Regarding the remaining waveforms SinusV,
SinusF and TrapV, the test system responses are quite similar to each other even in the reduced power mode. No appreciable
differences exist in terms of real power. On the contrary, higher values of apparent power and rms current are measured when
SinusF is used to achieve a reduced power mode with respect to the SinusV and TrapV. Only TrapV offers similar performance
to SinusV.

13. Thank you for your attention
Dr. Alessandro Burgio
alessandro.burgio@unical.it