Capacitors
December 2000 — Greenwood SC
Partial Discharge Considerations
in Capacitor Design
KILOVAR BRIEFS
7
Research and development provides the means of
investigating the parameters and processes influencing
capacitor performance and reliability in a sound scientific manner. It provides the expertise to develop the
links between the phenomenon observed and the physical processes responsible. It also leads to improvements in performance and reliability through the development of new materials, new designs and new manufacturing procedures.
An important example of the contribution of research
and development to capacitor reliability has been in the
area of partial discharge phenomenon and its role in
capacitor design. When the voltage across the plates of
a capacitor dielectric system is slowly raised, a level is
reached where a multitude of partial discharges begins
to occur at a consistent voltage level. This is referred to
as the Partial Discharge Inception Voltage (DIV) of the
dielectric system.
Partial discharges are very short-duration, minute
current pulses that have been observed to occur in
dielectric systems under high electrical stresses. Partial
discharges are normally detected using current-sensing
instrumentation connected to the dielectric system
which responds with a resonant output to the shortduration current pulses within the dielectric device.
Present scientific data associate partial discharges with
the electrical breakdown of gas bubbles in regions of
high electrical stress. The gas bubbles could be present
naturally in the dielectric medium or could actually be
evolved from liquids as a result of the electric field. The
bubbles permit gaseous phase discharges which can be
measured with the test equipment. When the voltage on
the dielectric system is slowly raised, partial discharges
first occur in the region where the electrical stress is the
highest.
Figure 1 shows a field plot of a parallel plate dielectric
system representative of a power factor correction
capacitor. The equipotentials are shown as parabolas.
The distance between the equipotentials is a measure
of the electrical stress in that region. As the equipotentials come closer together, the stress increases. The
equipotentials are equally spaced well to the right of the
foil edge. The stress in this region is considered the
nominal dielectric stress.
December 2000 • Supersedes 6/86
Printed in USA
Figure 1.
Plot of Thick Parallel-Plate Dielectric System.
Figure 2.
Field Plot of Thin Parallel-Plate Dielectric System.
As can be seen in Figure 1, the equipotential lines
group together near the foil edge. This means that the
electrical stress is highest in this region. In a wellimpregnated capacitor dielectric system, the inception of
partial discharge activity will take place at the foil edge
as this is where the electrical stress is the highest.
Figure 2 shows a field plot of a thinner capacitor dielectric system. The potential difference between the foils
has to be correspondingly reduced such that the
nominal dielectric stress (well to the right of the foil
edge) is the same as that in Figure 1. Figure 2 again
shows that equipotentials group together near the foil
edge. Unlike the thicker dielectric system shown in
Figure 1, the equipotentials in Figure 2 do not group
together as closely near the foil edge in the thinner
dielectric system, resulting in a lower voltage gradient.
Since the edge stress magnification is less for a thin
dielectric, the thin dielectric system has a higher DIV
per unit thickness than a thick dielectric layer.
1
DIV
Partial Discharge Considerations in Capacitor Design
THIN
THICK
DIELECTRIC THICKNESS
Figure 4.
DIV vs Thickness.
Figure 3.
Miniature Capacitor.
Cooper Power Systems utilizes miniature capacitors,
Figure 3, to obtain partial discharge data of a basic
dielectric system as described in Figure 4. Miniature
capacitors rather than full-size units are used because
of the problems associated with partial discharge
measurements of large capacitors. Full-size capacitors
usually make use of more than one series group and
have large active areas. As a result, the magnitude of
partial discharges becomes infinitesimal with respect to
the total charge flow. Because of this and other interfering factors, the measuring sensitivity is reduced and
the possibility of errors is greatly increased. It has been
found that through the use of miniature capacitors,
many hundreds of samples can be tested with relative
ease and great assurance of proper data. The miniature
systems make use of dielectric configurations, materials
and processing which are identical to full-size capacitors. Thus, with respect to obtaining characteristic
partial discharge data, the miniature system is a very
effective representation of the actual full-size capacitor.
Once the DIV of an electrical system is described as a
function of the dielectric thickness, it can be used as a
design tool. Figure 5 is a representation of the DIV as it
is considered for the design of capacitors. Experience
2
DESIGN
REGION
SAFETY
FACTOR
DESIGN STRESS
THIN
THICK
DIELECTRIC THICKNESS
Figure 5.
Capacitor Design Criteria.
and testing must establish a design stress which
assures adequate life of the materials. Additionally, a
safety factor should be applied to the design stress to
prevent the capacitor from operating in corona during
nominal/rated applications. Cooper Power Systems’
capacitors utilize a minimum safety factor of 180% of
design stress at room temperature. This results in a
design region for the dielectric thickness which is limited
on one end by the discharge inception voltage and by
other factors on the thinner dielectric end. As the dielectric thickness decreases, the material consistency
becomes less precise.
KILOVAR BRIEFS
Thus, the design region is limited on both ends by
several factors. At one end, the limiting factors are
discharge inception voltage and the safety factor, while
at the other end, the limitation is material consistency.
Although characteristic corona data is obtained using
miniature capacitors, the basic design data is verified
using full-size production units. Since corona tests on
full-size units are poor at best, an energization test is
utilized. For example, tests have been conducted by
operating units starting at 130% of rated voltage and
increasing the voltage in 10% increments. The result of
such a test is shown in Figure 5. The unit that was used
in this test failed in a matter of minutes when the
voltage was raised above the DIV level established from
miniature samples.
200
% of Rated Voltage
Operation of the dielectric system in corona can rapidly
cause permanent damage and eventually lead to failure.
However, proper design parameters utilizing an
adequate safety factor will allow long operational life of
the dielectric system.
7
180
160
140
120
100
0
60
120
180
Time (Minutes)
Figure 6.
Step Voltage Stress Test.
The characteristic partial discharge inception level can
be influenced by other factors besides the maximum foil
edge stress. Experiments conducted by Cooper Power
Systems’ Research and Development program were the
first to document the importance of dielectric fluid pressure on corona discharge characteristics. New dielectric
materials, films and fluid, and designs, such as foil edge
treatments and increased stacking factor designs,
continue to be evaluated for improved partial discharge
characteristics which will ultimately result in further
improvements in capacitor performance.
The adequacy of partial discharge considerations in
capacitor design is verified by the product’s performance. Cooper Power Systems continues to perform
basic research on dielectric systems applicable to
power capacitors. Discoveries as a result of this
research continue to improve the performance of dielectric systems implemented within the power capacitor
products.
3
P.O. Box 1640
Waukesha, WI 53187
http://www.cooperpower.com
©2000 Cooper Industries, Inc.
Printed on Recycled Paper