Static dynamic back-corona characteristics

Static and dynamic back-corona characteristics

Gianluca Bacchiega, Ivo Gallimberti,
IRS srl - via Vigonovese 81 - 35124 Padova - Italy (e-mail: bacchiega@irsweb.it)

Véronique Arrondel, Philippe Raizer, Jérôme Lecointre, Michel Hamlil
EDF R&D – 6, quai Watier – 78 401 Chatou - FRANCE (e-mail : veronique.arrondel@edf.fr)

Abstract

Voltage-current characteristics are a basic tool to analyse the operational conditions of an
electrostatic precipitator, and to detect the onset of Back-corona, that strongly affects the collection
performances. In this investigation, a laboratory set-up has been arranged, with a classical ESP wire-
plate configuration; both static and dynamic voltage-current characteristics have been measured and
analysed with different geometries, dust characteristics, and layer thickness. Similar results have been
also obtained in an industrial ESP, and compared with the laboratory results.

In the laboratory ESP, the static voltage-current characteristics and the collection efficiency
have been measured as functions of time; with high resistivity dust, back-corona takes place as layer
thickness increase, with a decrease of the efficiency and a sharp raise of the current. The effect of
electrode type and powder resistivity has been analysed.
With varying applied voltages, the dynamic voltage-current characteristics show an hysteresis cycle,
the recorded values depending on the slope of the applied waveform. The shape of the dynamic
characteristics changes significantly with the onset of back-corona.

A self-consistent numerical model has been applied to simulate static and dynamic voltage-
current characteristics in absence of back-corona. These simulations allowed to analyse the retarding
effects of the different physical process (capacitive charging, ion drift, space charge movements), and
to estimate the corresponding time constants. Finally, the dynamic effects of back-corona have been
analysed.

A second set of experiments have been realized in a 600MW power plant: ESP static and
dynamic characteristics have been measured with fly ash of low and high resistivity: the results are
very similar to those obtained in the laboratory ESP, and confirm that it is possible to define a specific
“signature”, on both static and dynamic V-I characteristics, of the different physical processes
associated to the particle captation, including back-corona.

1 Introduction

The collection efficiency of an Electrostatic Precipitator depends on a large number of
geometrical and physical parameters; the most important of them is the current emitted from the High
Voltage electrodes and collected by the plates [1]: for that reason the Voltage-Current characteristic,
and the actual working point on them, are used generally used to define the ESP operating conditions
[2]. It is well known that in general the higher is the current the better are the collection performances.
However, with high resistivity fly ash, backcorona discharges may form in the layer of deposited dust,
leading to a current increase, with a reduction of the collection performances.

Backcorona is a series of micro-discharges in the air spaces within the collected dust layer [3]:
it forms when an excessive field is induced in the layer by the charge flowing through its surface: it is
therefore related to the local values of the current density. Backcorona forms at spots on the surface
of the collecting plates, causing re-entrainment of previously collected dust. The current rises after
back corona onset, and tends to be tunnelled through existing back corona spots: this accelerates the
back corona growth.

Positive ions, injected by back corona into the volume of a negatively energised precipitator,
have the following effects [4]:

p.1
ICESP IX : Static and dynamic back-corona characteristics

- discharge previously charged dust particles;
- attach a positive charge to dust particles, causing their drift backwards to the emitting
electrodes;
- reduce the space charge, and therefore increase the electrical field on the emitting electrode
and the corona current;
- build-up nodule shaped dust deposits on the emitting electrode, that further increase local field
and corona current;
- form sparks at a much lower voltage level.

In the operation of an industrial ESP it is necessary to avoid all these negative effects of
backcorona: for this good diagnostic tools are needed, based on the modifications that backcorona
induces on easily measurable parameters.

2 Voltage-current characteristics in a laboratory ESP model

2.1 Laboratory ESP description

The experimental tests have been performed in the Test Laboratory of SAMES in Grenoble
(France) on a small scale vertical precipitator (Figure 1), made of a rectangular transparent PVC tube
(0.30 x 0.30 m), with three fields, each realized with a couple of plates (Pl.1,Pl.2 and Pl.3), and two
emitting electrodes. In order to test the effect of the geometry, cylindrical wires five millimetres in
diameter and ribbon bands with points have been used as emitting electrodes. The applied voltage
was generated by three DC switching power supplies (60kV, 500 µA), with low ripple (less than 1%
peak to peak), controlled by a remote PC: this enabled to program different voltage waveshapes on
each field (ramp, square, etc.).

During test operation a constant air flow of 270 m3/h, at room temperature, was supplied to the
ESP. Different types of dust were fed at the top of the precipitator duct : synthetic powders (époxy
polyester, with different metal additives to vary the resistivity) and coal fly ash collected in the hoppers
of an industrial ESP. The size distribution of the synthetic particles was concentrated below 10 um,
while that of the coal fly ash was between 1 and 100 um. The particle resistivity could vary from
medium to very high values (1010 to 1012 ohm cm). An almost constant particle concentration (10
gr/m3) was maintained at the ESP inlet during all the test period.

Voltage and current measurements were performed at each generator, with a fast recording
oscilloscope and a datalogger; the air and powder flow rates were systematically checked, to establish
the inlet concentrations; the collecting plates were periodically removed and weighted (at specified
time intervals 15, 30, 60 min, 1, 2, 4, 6 hr), to establish the deposition rate and the collection efficiency
of each field.

Figure 1 : Laboratory precipitator layout
p.2

2.2 Current characteristics

The corona current, under negative DC voltage, has specific patterns [5,6]]: at onset level low
amplitude pulses appear, with a regular repetition rate (Trichel Pulses); as the voltage is increased,
the frequency of the Trichel Pulses increases, while a continuous component is developed (Glow
Corona); at even higher voltage a second impulse component appear (Pre-breakdown Streamers),
with large amplitude and random pattern (Figure 2 b). If the statistical distribution of pulse amplitude is
analysed, a typical double population is observed; in the present case (Figure 3), with dust load but no
back-corona, the Trichel Pulses present a narrow distribution centred around 100 µA, while the pre-
breakdown streamers present a wider distribution with a maximum around 250 µA.

When back-corona is established, the current pattern and the statistical distribution of pulse
amplitude change completely ( Figure 2 a): a new population appears of very small pulses centred
around 50 µA, with random repetition rate, while the two previous ones disappear (Figure 3). These
pulses have to be associated to the back-corona microdischarges in the dust layer deposited on the
plates; due to their sharp form and random repetition rate they produce a wide H.F. electromagnetic
noise, that makes difficult to obtain clean measurements.
400

a
Current ( µA )

200

0
400

b
Current ( µA )

200

0
0 10 20 30 40 50 60
T im e (µs)
Figure 2 : current signal versus time; b: with dust load, V= 40 Kv, below backcorona thresold;
a : with dust load, V= 50 Kv, above backcorona thresold
400

a
300
Peak number

200

100

b

0
0 100 200 300 400 500
M a x im u m p e a k c u rr e n t ( µ A )

Figure 3 : Statistical distribution of current pulse amplitude; b: with dust load, V= 40 Kv, below
backcorona thresold; a : with dust load, V= 50 Kv, above backcorona thresold

p.3

2.3 Static Voltage-Current characteristics

The static Voltage-Current characteristics of each field have been measured with a step by step
slowly increasing voltage waveshape: with no dust load, above the corona inception threshold, all the
characteristics present a typical quadratic dependence (see curve a in Figure 4). When a high
resistivity powder is injected at top of the duct, it is possible to see that the voltage-current
characteristics change in time: this variation is faster for the first plate and slower for the third one.
The characteristics of Pl.3 are presented in Figure 4, after deposition times of 2, 4, 6 minutes): as the
dust layer deposited on the plates increases in thickness, a back-corona discharge forms within the
layer and raises the discharge current. Consequently the measured curves move to higher currents,
and change shape from quadratic to linear, above the back-corona threshold.
In normal operation, the injection of low resistivity particles lowers the current values, moving
the voltage-current characteristics to the right of the reference curve, with no dust load: this may be
attributed to the attachment of part of the ionic charge to the particles, that have a much lower
mobility; in addition, their space charge reduces the electric field on the emitting electrodes, thus
reducing the corona current. On the contrary, in the present case, even in absence of back-corona,
with short deposition times the characteristics is moved to the left of reference curve. This has to be
attributed to an increase of the wire emissivity, associated to dust deposition on the wire surface, that
increases the surface roughness and therefore the local field distortion: in this case, the field reduction
due to the particle space charge is compensated by the field increase due to local distortion.

500

d
400

c
Current (mA)

300
b

200 a

100

0
0 10 20 30 40 50 60 70 80
A p p lie d v o lta g e (k V )

Figure 4 : Voltage-Current characteristics for different deposition times (Pl.3 - cylindrical wires;
a=0 min; b=2 min; c=4 min; d=6 min)
The establishment of back-corona also affects the particle collection efficiency (see Figure
5). At the beginning of the particle injection, each plate has a slightly different efficiency (with an
average around 65%), depending essentially on particle size distribution and concentration at the inlet
of each plate. As the dust layer thickness increases and the back-corona forms on the plates, the
particle collection efficiency decreases sharply and falls down to about 20%; the overall efficiency falls
from 96% to 54%.

100%

Tot
80%
P l. 1
60%
Efficiency

P l. 2

40%
P l. 3

20%

0%
0 1 2 3 4 5 6
T im e (m in )

Figure 5 : Collection efficiency of the different plates as a function of deposition time (cylindrical
wires, V=50 kV)
p.4

Static voltage-current characteristics with back-corona are influenced by the particle resistivity.
Figure 6 show how curves change as powders with different resistivity and humidity are used. Curve a
is given as reference without particles; curve b and c show the results for medium and high resistivity;
curve d show the results for very high resistivity. As the resistivity increases, the field across the
deposited dust layer increases, together with the back-corona intensity and the discharge current.

500

d
400
c
b
Current (µA)

300

a
200

100

0
0 10 20 30 40 50 60 70 80 90
A p p lie d vo lta ge (kV )

Figure 6 : Effect of particle resistivity on Voltage-current characteristics (Pl. 1 - cylindrical wires;
a = no particle load; b = medium; c = high; d = very high)

The Voltage-current characteristics are also influenced by the geometry of the ESP
electrodes. In Figure 7 the results are shown for two types of emissive electrodes: bands with points
(a) and cylindrical wires of 5 millimetres in diameter (b). With no dust load, the corona voltage
threshold is lower for the sharper geometry, and higher current values are obtained for lower applied
voltages. A similar trend is shown by the curves with high resistivity particles (a1 and b1): with the
sharper geometry also the back-corona threshold becomes lower.

500

400
a1 a2
Current (µA)

300

b1 b2
200

100

0
0 10 20 30 40 50 60 70 80 90

Figure 7 : Effect of the geometry of emissive electrodes on Voltage-Current characteristics (PL: 1;
a: band with points; b: cylindrical wires; 1 : with dust, 2 : without dust)

It is well known from the literature [7] that the Voltage – Current characteristics has quadratic
form. In the present case the form k*U*(U – U0)has been adopted, where U is the actual voltage, U0
the corona inception voltage, and k a field distortion coefficient, that represents the effect of
protrusions or points on a smooth surface: it has be assumed 1.0 for the wire, and 1.43 for the band
with points. With this normalization, in absence of backcorona, the characteristics are linear, with an
average slope between 0.08 and 0.10 uA/kV2; after backcorona inception, the slope suddenly
changes, and identifies with no ambiguities the inception voltage (see Figure 8). This procedure may
be useful in practical applications.

p.5

500

a1
400
b1
b2

Current (µA)
300

a2
200

100

0
0 1000 2000 3000 4000 5000 6000
2
k U(U-U 0 ) (kV )

Figure 8: Curve normalization (PL: 1; a: band with points; b: cylindrical wires; 1 : with dust, 2 : without
dust; a2, b2 : points = experimental, lines = simulation results)

In order to verify the static characteristics on the base of the physical processes, the
ORCHIDEE simulation code [8,9,10,11] has been applied to the specific conditions of the laboratory
ESP. The results are reported as solid lines in Figure 8, while the points represent the experimental
results: the agreement is quite satisfactory.

2.4 Dynamic voltage-current characteristics

The static Voltage-Current characteristics have been measured by applying a step by step
slowly increasing voltage; if a fast varying voltage waveshape is applied, the dynamic characteristics
show an hysteresis cycle, in both cases with and without dust load: an example of this behaviour is
presented in Figure 9, without dust load, for bands with points as emissive electrodes, and intermittent
energisation (trapezoidal wave shape with rise and fall times of the order of 30 ms). The measured
current values are different depending on whether the applied voltage is increasing or decreasing: with
a rising slope the current is higher than the static value, the opposite being true with a decreasing
slope. This type of dynamic hysteresis cycle is typical of processes with a delayed response, which
may characterized by a Time Constant: the amplitude of the hysteresis cycle depends on the
steepness of the applied variation, specifically on the ratio between the Time Constant and the rise or
fall times.

500

400
Current (µA)

300

200
ris e
s ta tic
100
fa ll

0
0 10 20 30 40 50 60 70 80

Figure 9 : Voltage-Current characteristics: static and dynamic (PL. 1; bands with points; no dust load)

In the present case, without dust load, the current is carried by the negative ions generated
in the corona discharge: with a transient voltage, due to the finite mobility, their distribution between
wires and plates may respond to the voltage variation only with the delay of their transport processes.
Therefore, all the physical parameters associated to the current emission (space charge, electric field
on the wire surface, intensity of the corona discharge) respond to voltage transient with the Time
Constant of the ion transport.
p.6

500

400
3 ms

Current (µA)
300

200 30 m s
3 ms
30 m s
100
static

0
10 20 30 40 50 60 70
A pplied V oltage (kV )

Figure 10 : Experimental (points) and simulated (lines) static and dynamic Voltage-Current
characteristics (rise-fall time for simulations: 3 and 30 ms)
In order to describe in detail the dynamics of the physical processes, the ORCHIDEE
simulation code [11,12] has been applied to the specific conditions of the laboratory ESP: a
trapezoidal voltage waveshape has been applied, with different rise and fall times, with no dust load.
The simulation results are presented in Figure 10; the lines represent the computed data, the points
the experimental results. It can be clearly seen that the amplitude of the hysteresis cycle depends very
much on the voltage variation rate; for a rise and fall time of 30 ms, the experimental and computed
results are in quite good agreement.

The computed results show that, on the rising side of the dynamic characteristics, the
voltage is in advance with respect to the space charge distribution: in fact, the charges generated
around the wire need time to cross the gap and fill the volume; as a consequence of the reduced
“chocking effect” of the space charge, the electric field on the wire surface is higher, leading to higher
current emission. The opposite is true in the falling side: the delay in the space charge dissipation
sustains the “chocking effect”, and therefore reduces the field and the current.

The same behaviour is observed with the dust load, in absence of back-corona (clean
plates): the amplitude of the hysteresis cycle is slightly higher, indicating a longer Time Constant for
space charge transport; this is clearly due to the attachment of ions to the dust particles, and the
consequent reduction of the charge average mobility. The reduced Time Constant may be assumed
to result from a weighted average of the mobility Time Constants for negative ions and charged
particles.

A very different behaviour is observed after the inception of back-corona (Figure 11): while
the static curve steeps up rapidly, the rising side of the dynamic curve increases less rapidly than the
corresponding stationary values; the dynamic values pass from the left to the right side of the static
characteristics. This indicates a delay in the establishment of the back-corona discharge within the
dust layer: it may be represented with a new Time Constant, which is switched on above the back-
corona threshold. It is worth noting that this delay has nothing to do with that observed in par. 2.3,

500

400
Current (µA)

300

200
ris e

100 s ta tic

fa ll
0
0 10 20 30 40 50 60 70 80

Figure 11 : Voltage-Current characteristics: static and dynamic (PL. 1; bands with points; dust load)

p.7

which was essentially associated to the rate of layer deposition on the plates. An opposite behaviour is
observed in the falling part of the voltage wave (Figure 11): however, the delay for back-corona
extinction appears to be quite short than that for its inception.

The shape of a 8, with the crossing of the two sides of the dynamic characteristics around
the back-corona inception, seems to be a typical “signature” of the presence of back-corona
discharges within the deposited dust layer: the amplitude and form the dynamic cycle may be
associated to the different time constants for charge transport and discharge formation.

3 Voltage-current characteristics in an industrial ESP

3.1 Industrial ESP description

The tests on the industrial precipitator have been realized in a 600 MWe coal burning unit of
the EDF Power Plant in Le Havre (France). The ESP installed at the boiler outlet is realized with three
parallel casing: each of them is composed by 5 fields; the first and the last of these fields are equipped
with traditional power supplies, while the other ones are equipped with new DC switching power
supplies able to feed low ripple voltage (+/- 1% peak to peak). Each field has 60 gas ducts with a
spacing of 0.28 m, and 7 collecting plates (0.46 m long and 12.4 m high). Emissive electrodes are 2
barbed wires per plate, placed inside rigid frames at a distance of 0.12 m each other. Rapping
equipment is realized by falling hammers.

During the tests, the electrofilter flue gas flow varied from 1.25 106 Nm3/h to 1.6 106 Nm3/h
for electrical loads varying from 410 to 530 Mwe; the fly ash concentration was roughly constant
around 9 g/Nm3 (6% dry O2). Four test campaigns have been performed between March and April
2002; four different coals have been used, to test different fly ash resistivities and gas flow rates:
South African coal ( 7 1010 Ohm cm – Pe = 517 Mwe), South African coal (5 1010 -2 1011 Ohm cm –
Pe = 400 MW), Australian coal (3 109 Ohm cm – Pe = 520 MW), Polish coal (very low resistivity – Pe =
500 Mwe).

3.2 ESP static voltage-current characteristics

The industrial test have been realized in order to verify that the typical “signatures” of the
backcorona formation and development, observed in the laboratory ESP model, can be effectively
observed also in a full scale industrial plant.

The static Voltage-Current characteristics of each field have been measured with a step by
step slowly increasing voltage. Preliminarily, a test has been realized with the plant burning high
resistivity South African coal, just after the arrest of the unit. With cold precipitator (60 °C) and without
dust load, two static characteristics have been recorded, before and after the evacuation of the dust
layer from the plates by rapping (Figure 12). It can bee clearly seen, above 28 kV, the typical pattern
of backcorona inception in the deposited dust layer.

500

400

a
Current (mA)

b
300

200

100

0
0 10 20 30 40 50
Applied voltage (kV)

Figure 12 : Static Voltage-Current characteristics with (a) and without (b) dust layer on the plates

p.8

Similar results have been obtained under normal operating conditions: Figure 13 shows the
static characteristics obtained in three different campaigns, respectively with high (a), medium (b) and
low (c) coal resistivity.

500

400
c b
Current (mA)

300
a
200

100

0
0 10 20 30 40 50

Figure 13 : Static Voltage-Current characteristics in three different campaigns, respectively with
high (a), medium (b) and low (c) coal resistivity.

3.3 ESP dynamic voltage-current characteristics

The dynamic Voltage-Current characteristics have been recorded by applying an fast
intermittent voltage waveshape: also in the industrial plant, they show an hysteresis cycle, in both
cases with and without dust load: an example of this behaviour is presented in Figure 14, with dust
load of low resistivity of the dust particles. As in the laboratory case, the measured current values are
different depending on whether the applied voltage is increasing or decreasing: with a rising slope the
current is higher than the static value, the opposite being true with a decreasing slope.

500

400
Current (mA)

300

200
rise
static
100
fall

0
0 10 20 30 40 50

Figure 14 : Voltage-Current characteristics: static and dynamic (Polish coal, with low resistivity).

As in the laboratory case, a very different behaviour is observed with high resistivity dust (5):
when the static curve steeps up rapidly, the rising side of the dynamic curve increases less rapidly
passing from the left to the right side of the static characteristics; the opposite is true for the falling side
of the characteristics. Also in the industrial ESP, the shape of a 8, with the crossing of the two sides of
the dynamic characteristics around the back-corona inception, seems to be the typical “signature” of
the presence of back-corona discharges within the deposited dust layer.

p.9

500

400

Current (mA) 300

200

100 rise
static
fall
0
0 10 20 30 40 50

Figure 15 : Voltage-Current characteristics: static and dynamic (South African coal, with high
resistivity).

Conclusions
In this paper the typical “signatures” , that identify the onset of backcorona, have been
presented and discussed:

- a change in the impulse component of the discharge current, which moves the statistical
distribution of the peak amplitudes towards lower values;
- a steep change of the static Voltage-Current characteristics at onset threshold, particularly
evident on normalized curves;
- the appearance of a 8 shaped form in the dynamic Voltage-Current hysteresis cycle, with a
crossing point around the onset conditions;

The analysis of these “signatures” may be recommended as a technique to determine the
onset of back corona, and to analyse the susceptibility of a precipitator to back corona, under specific
operating conditions.

Acknowledgements

The authors wish to thank the Company SAMES in Grenoble, its technical director Caryl
Thomé, and its personnel, for the use of the laboratory equipment, and for the help in running the
tests. The authors are also indebt with the personnel of the Le Havre power plant for the participation
to the industrial ESP tests.

References

[1] S. OGLESBY, G. NICHOLS: “Electrostatic precipitation”, Dekker, New York, 1978

[2] R.J. TRUCE, W. REIBELT: "New technologies improves electrostatic precipitator performances”,
VII ICESP*, Kyongjiu (Korea), 1998

[3] S. MASUDA, A. MIZUNO: “Initiation condition and mode of back-discharge”, Journal of
Electrostatics 4, 1977-78

[4] R.J. TRUCE, “Back Corona and its Effects on the Optimisation of electrostatic Precipitator
Energisation control”, Seventh International Clean Air Conference, August 24-28, 1981

[5] L. B. LOEB: “Electrical Coronas”, University of California Press: Berkeley, CA ,1965

p.10

[6] A. VALAGUSSA: “Prove di laboratorio per lo studio della captazione elettrostatica”, Rapp. CESI
Lab-93/3138, 1995

[7] I. GALLIMBERTI : “The mechanism of the long spark formation”, J. de physique, colloque C7,
supplement au n.7, Tome 40, Juillet 1979

[8] I. GALLIMBERTI, A. GAZZANI, U. TROMBONI, E. LAMI, F. MATTACHINI, G. TREBBI:"Physical
simulation of the particle migration in ESP, Part I - Model description", VI ICESP*, Budapest, 1996.

[9] I. GALLIMBERTI, A. GAZZANI, U. TROMBONI, E. LAMI, F. MATTACHINI, G. TREBBI:"Physical
simulation of the particle migration in ESP, Part II - Application results", VI ICESP*, Budapest, 1996.

[10] E. LAMI, F. MATTACHINI, I. GALLIMBERTI, R. TURRI, U. TROMBONI: "A Numerical Procedure
for Computing the Voltage-Current Characteristics in ESP Configuration", Journal of Electrostatics 34
p.385-399, 1995.

[11] I. GALLIMBERTI: "Recent advancements in the physical modelling of electrostatic precipitators",
Journal of Electrostatics 43 (1998) 219-247, 1998
.
[12] V. ARRONDEL, G. BACCHIEGA, I.GALLIMBERTI: "ESP modelling: from University to Industrial
Application", VIII ICESP, Birmingham (USA), 2001

p.11

Static dynamic back-corona characteristics

More Related Content

Viewers also liked

Similar to Static dynamic back-corona characteristics

More from IRS srl

Recently uploaded

Static dynamic back-corona characteristics