Vous êtes sur la page 1sur 4

2010 International Conference on Solid Dielectrics, Potsdam, Germany, July 4-9, 2010

Measurement and Modelling of Partial Discharge Behaviour in a Spherical Cavity within a Solid Dielectric Material as a Function of Cavity Diameter
Hazlee A. Illias, George Chen, Paul L. Lewin
School of Electronics and Computer Science University of Southampton Southampton, United Kingdom hai07r@ecs.soton.ac.uk

AbstractPartial discharge (PD) measurement is an important diagnostic tool for insulation system assessment, which can be used to determine degradation of the insulation. Thus, PD monitoring has become an essential tool in assessing the performance of insulation materials and the modelling of PD events has been used widely to attain a better understanding of this phenomena. In this paper, a Finite Element Analysis (FEA) model for simulation of PD activity in spherical cavities of different diameters within a homogeneous dielectric material has been developed. The model has been used to study the influence of spherical cavity size on PD activity within a solid dielectric material by comparison of the simulation and measurement results. It was found that certain parameter values in the model vary for different cavity sizes. Critical parameters affecting PD activity for different spherical cavity sizes have been identified. Keywords-component; partial discharge, finite element analysis method

developed to study PD behaviour for different cavity sizes. Through modelling PD behaviour and comparison of simulation and measurement results, critical parameters that affect PD activity have been identified. These include inception voltage, detrappable electron effective lifetime, cavity surface conductivity, electron generation rate, extinction voltage and temperature decay time constant. II. TEST OBJECT

I.

INTRODUCTION

The experiment uses a commercial PD detection system [46]. Fig. 1 shows a schematic diagram of the test object, consisting of an artificial spherical void of a diameter, d located within an epoxy resin of 3 mm thick and 38 mm diameter. The values for d are 1.1 and 2.35 mm while the thickness of smaller epoxy block, hsmall for each d is 1.7 and 2.7 mm respectively. The test object is immersed in mineral oil to eliminate discharges around the edge of the electrodes. A 50 Hz, 18 kV ac sinusoidal voltage is applied to the test object. Fig. 2 shows details of the two-dimensional (2D) model geometry that has been developed using FEA software, together with its mesh elements. The model consists of a hemispherical cavity of diameter 1.1 mm, a cavity surface of thickness 0.05 mm and a homogenous dielectric material ( r=4.4) of 3 mm thickness and 10 mm diameter. The horizontal line in the cavity center represents the cross sectional area used to integrate current density to calculate the current in the cavity during a discharge event. A 50 Hz, 18 kV ac sinusoidal voltage is applied at the upper electrode while the lower electrode is always grounded. III. PD MODELLING

PD behaviour within a cavity in a solid dielectric material depends on many factors. One of the factors that influences PD activity is the cavity size. This is because the electric field and temperature distributions in the cavity are influenced by its geometry, which determines the pattern of PD occurrence. Previous work has reported that there were reduced numbers of PDs per cycle for smaller cylindrical cavity diameters within a dielectric material [1]. Higher total apparent charge per cycle for larger cylindrical cavity diameters has also been reported [2]. Other work has shown that the maximum discharge magnitude increases but the inception field decreases when the cavity size increases [3]. However, published research on the influence of different spherical cavity sizes on PD within solid dielectric materials has not been widely reported. In this work, measurement of PD activity for different diameters of spherical cavity in an epoxy resin has been undertaken. It has been found that PD activity in the cavity is strongly dependent on the cavity size. The measured inception voltage, discharge phase and amplitude distributions, PD repetition rate, total apparent charge per cycle, mean charge magnitude and maximum discharge magnitude were found to vary as a function of cavity size. A PD model has been

The electric field and temperature distributions in the model are solved using FEA software. The electric field distribution is solved by a using field model, where the governing equation is

(V) (V / t ) = 0

(1)

where V is the electric potential, is the conductivity and is the permittivity. If the temperature in the cavity is assumed

978-1-4244-7944-3/10/$26.00 2010 IEEE 1

Applied voltage Cylindrical electrode Epoxy resin

cav =

max{1

exp[(|Ucav/Uinc| + |Icav/Icrit|)]}

(6)

Cavity

hsmall

3 mm

where max is the maximum cavity conductivity during PD, Icav is the PD current in the cavity and Icrit is the critical current to start electron avalanche [6, 7]. The physical and apparent charges are calculated by time integration of current through the cavity and lower electrode respectively. The surface charge after a PD event can decay through conduction along the cavity wall. In this work, surface charge decay is modeled using cavity surface conductivity, s which is dependent on the polarity of Ucav and Us [6]. If their polarities are opposite to each other, s is set to an initial value, s0 but if their polarities are the same, s is increased and is calculated using
s

6 mm 25 mm 38 mm

Figure 1. Schematic diagram of the test object

uniform, the temperature distribution can be solved using

s0exp|Ucav/Udecay+Ts/T0|

(7)

C p T / t (kT) = Q

(2)

where is the mass density, Cp is the specific heat capacity, k is the thermal conductivity and Q is the heat source density. Equations (1) and (2) are coupled via the Q term, where Q can be determined by multiplying the current density, J, with electric field, E, in the cavity during the PD event. When the voltage across the cavity, Ucav exceeds the inception voltage, Uinc and if an initial free electron is available, PD may occur in the cavity. The total electron generation rate, Net is calculated by Net = (Ned+Nes+Nei)exp|Ucav/Uinc| (3)

where Udecay is the voltage across the cavity causing the surface charge to decay, Ts is the temperature on the cavity surface and T0 is the initial temperature. s is limited by a maximum to avoid numerical surface conductivity value, smax convergence problems. Since the surface charge decay can affect the electron generation rate, Nes can be written as Nes = Nes0|UPD/Uinc||Us/UsPD| (8)

where Nes0 is the initial Nes and UsPD is the voltage due to free surface charge immediately after a PD occurrence. After a PD event, the temperature in the cavity, Tcav increases. This is because heat energy is released due to electron ionization during PD interval. In this model, the heat density is increased from zero during the PD interval, resulting in an increase in the temperature of the cavity. Assuming that the cavity volume does not change after each PD event, the increase in temperature causes the pressure in the cavity to increase. Thus, using the ideal gas law, P1/T1 = P2/T2 (9)

where Ned, Nes and Nei are the electron generation rate due to charge detrapping from the shallow traps near the cavity surface, free charges on the cavity surface and charge injection from the electrode respectively. Ned can be calculated by Ned = Ned0|UPD/Uinc|exp(t/
trap)

(4)

where Ned0 is the initial Ned, UPD is the voltage across the cavity immediately after previous PD event, t is the time elapsed since previous PD event and trap is the detrappable electron effective lifetime. Ned0 is subdivided into two values, Ned0H which is used after the polarity of Ucav changes following a PD and Ne0L when there is no polarity change of Ucav between PD events [6]. In order to consider the statistical aspect of a PD event, a probability of PD occurrence is calculated using P = 1exp(Net dt) (5)

where P1 and P2 are the previous and new pressure and T1 and T2 are the previous and new temperature in the cavity. The new
Spherical cavity High voltage electrode Dielectric material

where dt is the time stepping interval [6, 7]. Every time Ucav exceeds Uinc, P is calculated and compared with a random number R, which lies within 0 to 1. If P is greater than R, a discharge will occur. During a PD event the cavity conductivity is calculated using

Symmetry axis

Cavity surface

Ground electrode

Figure 2. 2D axial-symmetric model geometry with mesh elements

pressure, P2 will modify the inception field, Einc in the cavity, which can be calculated using Einc = 24.2P2(1+8.6/(2P2r)0.5) (10)

where r is the cavity radius [8, 9]. Uinc is the voltage across the cavity when the electric field in the cavity is equal to Einc in the absence of surface charge from the FEA model. Table I details the definition of cavity size independent parameters used in the simulation. IV. RESULTS AND DISCUSSION

decay such as smax are smaller for the smaller cavity than for the larger cavity. This is because there an increased number of PDs per cycle for the smaller cavity. For the smaller cavity, surface charge decay through conduction along the cavity wall might be less significant, a lower value of smax but higher value of Udecay are assigned to the smaller cavity than the larger cavity. However, for the larger cavity, charge conduction along the cavity wall is more significant because there is more free charge left on the cavity surface after each PD event. These charges decay through surface conduction, resulting in charge loss through recombination [5]. In the simulation, the initial inception voltage, Uinc0, is set according the measured applied inception voltage, which is then used to determine the inception electric field, Einc obtained using the FEA model. The value of Einc is used to calculate the initial pressure in the cavity, P0 by using equation (10). It was found that P0 is around 70 kPa. The initial extinction voltage, Uext0 is set according to the minimum discharge magnitude. The critical current for avalanche, Icrit is lower in the smaller cavity because the current through the smaller cavity is lower than the larger cavity. Fig. 4 shows the simulation of PD charge magnitude and temperature in the cavity against phase of the applied voltage for two different spherical cavity diameters. The average temperature decay time constant, Tdecay for the larger cavity is 9 ms while for the smaller cavity is 1.8 ms. In the smaller cavity size, Tdecay is lower because heat can dissipate faster through the surrounding material. An average 400 pC event causes 1C temperature rise in the smaller cavity while in the larger cavity, 1C temperature rise is caused by an average 1130 pC event. This is because in the larger cavity, the temperature rises slower because of its larger volume. Because of the increased number of PD events within the smaller cavity, the average temperature rise per PD is 0.8C while for the larger cavity it is 2.7C.

Fig. 3 shows -q-n plots from the measurement and simulation results of PD activity for spherical cavities of diameter 1.1 mm and 2.35 mm within a dielectric material of thickness 3.0 mm for a 50 Hz of 18 kV ac applied voltage. The rabbit-ear curve on -q axes for the smaller cavity is well defined but the rabbit-ear curve of the larger cavity is more widely dispersed. Discharges that occur near the minimum charge magnitude for the smaller cavity are more concentrated than for the larger cavity [4]. Table II details the comparison between the measurement and simulation results for different cavity sizes while Table III shows the definition of cavity size dependent parameters used in the simulations. From Table II, the simulation data are in good agreement compared with the measurement results for both cavity sizes. From Table III, parameters that relate to electron generation rate such as Ned0H, Ned0L, Nes0, Nei and trap are higher while the parameters that relate to surface charge
TABLE I. DEFINITION OF PARAMETERS USED IN THE SIMULATION Definition Time step during no PD Time step during PD Material relative permittivity Cavity surface relative permittivity Cavity relative permittivity Material conductivity Initial cavity surface conductivity during no PD Cavity conductivity during no PD Maximum cavity conductivity during PD Cavity initial temperature Specific heat capacity of the material Specific heat capacity of air Thermal conductivity of the material Thermal conductivity of air Density of the material Density of air Symbol dt dt
rmat rs rair mat

Value 1/720f 1 4.4 4.4 1 1x10-18 1x10 0 1x10-2 293 1179 1005 0.35 0.0257 561 1.205
-18

Unit s ns

S/m S/m S/m S/m K J/(kgK) J/(kgK) W/(mK) W/(mK) kg/m3 kg/m3 Figure 3. -q-n plots of the measurement (a-b) and simulation (c-d)

s0

0 max

T0 Cpmat Cpair kmat kair


mat air

TABLE II.

MEASUREMENT (M) AND SIMULATION (S) DATA 1.1 M 6.5 651 101 373 80 S 6.5 687 106 352 83 M 2.5 5420 2165 4763 913 2.35 S 2.5 5615 2248 4748 914

TABLE III.

DEFINITION OF PARAMETERS USED IN THE SIMULATION FOR DIFFERENT CAVITY SIZE Cavity diameter Definition Symbol 1.1 2.35 5.5x10-9 6410 810 50 1x10 1.5 1500 200 3 1.6
-5

Cavity diameter (mm) Total PDs per cycle Total charge per cycle (pC) Mean charge (pC) Maximum PD magnitude (pC) Minimum PD magnitude (pC)

Unit 5x10-13 3690 340 100 1x10 2.5 35000 2000 12 2.5
-6

Maximum surface conductivity for charge decay Initial cavity inception voltage Initial extinction voltage Cavity voltage for charge decay Critical current for avalanche Detrappable electron effective lifetime Initial electron generation rate due to charge detrapping Initial electron generation rate due to free surface charge Electron generation rate due to electrode charge injection

smax

S/m V V V A ms 1/s 1/s 1/s 1/s

Uinc0 Uext0 Udecay Icrit


trap

0.8 PD charge(nC) 0.6 0.4 0.2 0 0 297 Temperature(K) 296 295 294 293 292 0 360 720 1080 1440 1800 2160 2520 2880 3240 3600 Phase(degree) 360 720 1080 1440 1800 2160 2520 2880 3240 3600 Phase(degree)
Tcav T0

Ned0H Ned0L Nes0 Nei

behaviour under different cavity sizes have been identified. The initial electron generation rate is higher for the smaller cavity and the effect of surface charge decay through conduction along the cavity wall and the temperature change due to discharges are less significant for the smaller cavity than the larger cavity. REFERENCES
[1] C. Forssen, "Partial discharges in cylindrical cavities at variable frequency of the applied voltage," Licentiate Thesis, Royal Institute of Technology (KTH), Stockholm, Sweden, 2005. [2] C. Forssen and H. Edin, "Influence of cavity size and cavity location on partial discharge frequency dependence," Conference on Electrical Insulation and Dielectric Phenomena, pp. 438-441, 2004.

(a)
8 PD charge(nC) 6 4 2 0 0 305 Temperature(K) 300 295 290 0 360 720 1080 1440 1800 2160 2520 2880 3240 3600 Phase(degree)

[3] M. Budde, V. Ermel, and M. Kurrat, "Partial discharge diagnostics of micro cavities and its modelling," International Symposium on High Voltage Engineering, 2009. [4] H. A. Illias, G. Chen, and P. L. Lewin, "Partial Discharge Measurements for Spherical Cavities within Solid Dielectric Materials under Different Stress and Cavity Conditions," Conference on Electrical Insulation and Dielectric Phenomena, 2009. [5] H. A. Illias, G. Chen, and P. L. Lewin, "Modelling of partial discharge activity in different spherical cavity sizes and locations within a dielectric insulation material," International Conference on Properties and Applications of Dielectric Materials, pp. 485-488, 2009. [6] H. A. Illias, G. Chen, and P. L. Lewin, "Modelling of surface charge decay in a spherical cavity within a solid dielectric material using Finite Element Analysis," International Symposium on High Voltage Engineering, 2009. [7] H. A. Illias, G. Chen, and P. L. Lewin, "Partial discharge modelling in a spherical cavity within a dielectric insulation material as a function of frequency," IEEE Electrical Insulation Conference, pp. 55-59, 2009. [8] L. Niemeyer, "A generalized approach to partial discharge modeling," IEEE Transactions on Dielectrics and Electrical Insulation, vol. 2, pp. 510-528, 1995. [9] F. Gutfleisch and L. Niemeyer, "Measurement and simulation of PD in epoxy voids," IEEE Transactions on Dielectrics and Electrical Insulation, vol. 2, pp. 729-743, 1995.

Tcav T0

360

720

1080 1440 1800 2160 2520 2880 3240 3600 Phase(degree)

(b) Figure 4. Simulated temperature in the cavity against phase for (a) 1.1 mm and (b) 2.35 mm spherical cavity diameter

V.

CONCLUSIONS

A model of partial discharge in a spherical cavity within a dielectric material has been developed and used to simulate PD activity for different spherical cavity sizes. It was found that PD behaviour is strongly influenced by cavity size because the electric field and temperature distributions in the cavity are dependent on this parameter. Through comparison between measured and simulated data, critical parameters affecting PD

Vous aimerez peut-être aussi