Dielectric Frequency Response of Oil-paper

510 J. Liu et al.: Dielectric Frequency Response of Oil-paper Composite Insulation Modified by Nanoparticles

1070-9878/12/$25.00 © 2012 IEEE

Dielectric Frequency Response of Oil-paper Composite Insulation Modified by Nanoparticles

Jun Liu, Lijun Zhou, Guangning Wu, Yingfeng Zhao

College of Electrical Engineering, Southwest Jiaotong University No.111. Sec.1, North Erhuan Road Chengdu, Sichuan, 610031, China

Ping Liu and Qian Peng Electrical Test and Research Institute of Sichuan Province

No.24, Qinghua Road Chengdu, Sichuan, 610072, China

ABSTRACT

The research on dielectric frequency response of oil-paper composite insulation modified by nanoparticles has been carried out in order to seek possible way to improve insulation properties thereof. Before and after nanoparticles being added, the dielectric frequency responses of oil-paper composite insulation with different moisture contents at different temperatures have been measured within the frequency from 0.1 mHz to 1 MHz. The Cole-Cole relaxation model has been used to analyzing the measured data. The results show that the variation rules of insulation condition can be indicated by the parameters of relaxation model. By adding nanoparticles, a new low frequency relaxation appeared.

Index Terms — Dielectric frequency response, oil-paper composite insulation,

Cole-Cole model, nanoparticles modification, complex dielectric coefficient.

1 INTRODUCTIONOIL filled transformers are widely used in extra high voltage

and ultra high voltage power transmission systems. Thus the oil-paper composite insulation condition will seriously affect the electrical life of transformer insulation [1-2]. With increasing transmission voltage level, it is more difficult for the traditional oil-paper composite insulation to meet the demands of small size, high tolerance to field strength, and high reliability. The emergence of nanotechnology leads to the development of composite materials into a new era. Much research has been carried out recently incorporating various nanoparticles into existing solid dielectric systems [3-4]. It showed that their breakdown characteristics, interfacial properties, space charge, partial discharge, etc. have been improved [5-6]. The early research of nanofluids focused on their excellent cooling features, for example, nanoparticles were added into the transformer oil to form colloid for enhancing the cooling effect of the transformer windings [7]. The research on breakdown characteristics of transformer oil modified by nanoparticles under ac, dc and lighting voltages has shown that the addition of nanoparticles plays an important role in improving its aging and damage performance [8-9]. Nanoparticles modifying technology provides a new effective way to improve the performance of oil-paper insulation system. However, the dielectric response of nanoparticles modified oil-paper composite insulation has not been reported yet.

Dielectric response method as a new technique of insulation diagnosis was presented at the commencement of the 1980’s. Time-domain dielectric response methods such as return voltage measurements (RVM), polarization and depolarization current (PDC) measurements emerged successively [10-11]. Frequency domain spectroscopy (FDS) measurements were introduced for diagnosis of transformer insulations in the late 1990’s [12-16]. The relationship between the transformer insulation condition and dielectric response was researched by Saha using return voltage measurement [13]. Gubanski applied the frequency response method to diagnose moisture in transformer insulation [17-20]. In this paper, the dielectric frequency responses of oil-paper composite insulation with different moisture contents at different temperatures have been measured before and after nanoparticles being added within the frequency range from 0.1 mHz to 1 MHz. The Cole-Cole relaxation model has been used to analyze the measured results. Parameters of the Cole-Cole relaxation model were obtained by a least squares fitting. The influence of nanoparticles addition on the relaxation parameters is discussed.

2 EXPERIMENTAL SETUP The TiO2-oil nanofluids were prepared by solvothermal

[21]. The oil used in experimental was new pure DB-#25 transformer oil produced in Kelamayi (moisture content is 8 ppm), which is applied in the ultra high voltage Manuscript received on 28 July 2011, in final form 12 December 2011.

IEEE Transactions on Dielectrics and Electrical Insulation Vol. 19, No. 2; April 2012 511

transformer about 1000 kV. The steps of the nanoparticles synthesis are as follows:

1) The analytical reagent oleic acid (12.5 ml), triethylamine (10 ml) and cyclohexane (50 ml) were filtered and dried. They were mixed and stirred 10 min by constant temperature heating magnetic stirrer (DF-101S, made by the Shanxi Taikang Biological Technology Co., Ltd. in China) at 25 °C.

2) 2.5 ml tetrabutyl titanate was added dropwise into the mixture by burette. The rate was about 0.5 ml/s. After being stirred for 30 min at 25 °C, they were moved into the pressure vessel. The reaction was kept at 150 °C for 48 h.

3) When the solution cooled to room temperature, the transparent TiO2 cyclohexane sol was formed, which precipitated after adding excess ethanol. The precipitate was centrifuge washed three times by 300 ml ethanol to neutral (PH≈7). The semi-dried cake was obtained after filtration.

4) The precipitates obtained from the above procedure were centrifuged four times (at 5000 rpm for 3-min intervals). Then it was ground to pale yellow powder by mortar manually.

5) The powder was calcined in muffle furnace for 2 h at 700 K. The white TiO2 nanoparticles were formatted. Average particle size is about 25 nm. The crystal structure and type is the rutile. The SEM micrographs of the TiO2 powders were shown in Figure 1.

Figure 1. SEM micrographs of TiO2 for 2×105 times zoom in. The scale is 50 nm.

6) Appropriate amount of TiO2 powder was added to the transformer oil. The concentration of nanoparticles in the oil was controlled about 0.3 g/l. The bulk density of TiO2 is 0.48 g/cm3, the estimate of the volume fraction would be about 0.06%. In order to avoid agglomerate, the ultrasonic dispersion instrument (FS-450, made by Shanghai Sonxi ultrasonic equipment company in China) was used to disperse the nanoparticles evenly throughout the oil. The TiO2-oil nanofluids were transparent. The dispersed phase particles cannot be observed with the unaided eye. For observed the dispersion of nanoparticles in the oil, the test paper was immersed horizontally in the nanofluids and then taken out for TEM analysis. The

result was shown in Figure 2. Although the nanoparticles had a little agglomeration, they were also in the nanometer level.

Figure 2. TEM photo of oil with TiO2 nanoparticles for 2×105 times zoom in. The scale is 50 nm.

The pressboard samples used for the measurements were discs with thickness of 1 mm and diameter of 55 mm. The weight of the dried paper was 2.36 g. The diameter of the electrodes was 50 mm. The oil used for impregnation was 1970 ml. The preparation of oil impregnated papers included the following steps: paper drying, paper moisturizing, impregnating with transformer oil. First, the papers were dried at 80 °C in a thermal-vacuum chamber for 2-3 days. During this process, their weights were monitored using the precision scale EP114C, made by OHAUS in America, for identifying whether they became dried out. After drying, the papers were kept in an open atmosphere for absorbing moisture from the air. Weights of the papers were again monitored to obtain predefined levels of moisture intakes. The papers were put into containers filled with virgin and nanoparticles modified transformer oil for impregnation. Five groups of samples were prepared with moisture concentrations of 0.5, 1, 2, 3 and 4%.

The first group of samples was used to measure frequency response of oil-paper composite insulation with different moisture contents. The papers with 0.5, 1, 2, 3 and 4% moisture contents were immersed in pure virgin transformer oil and TiO2 nanoparticles modified transformer oil. They were sealed and placed in the oven at 50 °C for two weeks, so that the moisture diffusion in oil-paper composite insulation achieved steady state. Moisture contents in oil and paper before measurement are shown in Table 1. Take the paper with 3% moisture contents as an example. Before equilibrium, the moisture from that in the paper is about 0.07299g (0.07299/(2.36+0.07299)=3%). The moisture from that in the oil is about 8×10-6×1970 ml×0.87 g/ml=0.01371 g. The all moisture is about 0.0867 g. After equilibrium, the moisture from that in the paper is about 0.050623 g (0.050623/(2.36+0.050623)=2.1%). The moisture from that in the oil is about 21.5×10-6 × 1970 ml×0.87 g/ml= 0.036848 g. The all moisture is about 0.0874 g. Considering the test error, the moisture gets balances before and after equilibrium.


Table 1. Moisture contents of oil-paper composite insulation before dielectric frequency response measurement at 50 °C.

Moisture contents

before equilibrium

Moisture contents

after equilibrium

In paper (%) In oil (ppm) Sum (g) In paper (%) In oil (ppm) Sum (g)

0.5 8 0.026 0.7 5.3 0.026

1 8 0.037 1.0 8.0 0.037

2 8 0.062 1.5 14.6 0.061

3 8 0.087 2.1 21.5 0.087

4 8 0.112 2.3 32.1 0.110

The second group of samples was used for measuring

frequency response of oil-paper composite insulation at different temperatures. The 2% moisture intakes samples were immersed in pure new transformer oil and TiO2 nanoparticles modified transformer oil. Then they were placed in sealed oven set the temperature about 10, 30, 50 and 70 °C for two weeks. After the moisture diffusion reached steady state, the moisture contents in oil and paper were shown in Table 2.

Table 2. Moisture contents of oil-paper composite insulation before dielectric frequency response measurement at different temperature Moisture contents

before equilibrium

In paper (%) In oil (ppm) Sum (g)

2 8 0.062

Moisture contents

after equilibrium In paper (%) In oil (ppm) Sum (g)

10 °C 2.4 2.3 0.062

30 °C 2.2 5.0 0.062

50 °C 1.5 14.6 0.061

70 °C 1.0 22.3 0.062

(a) Principle of test system

(b) Setup used for experiment

Figure 3. Setup of the test system.

The following text used the moisture contents of about 0.5, 1, 2, 3 and 4% which refer to the samples based on their nominal moisture concentration before equilibration. The measurement setup was designed as Figure 3. The tested papers were placed between the voltage electrode and measuring electrode. The oven was used to control the temperature. In order to eliminate edge effects, a guard electrode was employed. The dielectric frequency response of oil-paper from 10-4 Hz to 104 Hz were tested by the frequency domain insulation diagnostic analyzer IDAX300 manufactured by Megger Group Limited. The applied effective voltage was 140 V. The dielectric responses in higher frequency range about 104 to 106 Hz was performed by the LCR testers HIOKI 3532-50. The applied effective voltage was 5 V.

3 MEASUREMENT RESULTS 3.1 DIELECTRIC FREQUENCY RESPONSES OF OIL-PAPER COMPOSITE INSULATION WITH DIFFERENT MOISTURE CONCENTRATIONS

The dielectric frequency response of the oil-paper composite insulation with the moisture contents about 0.5, 1, 2, 3 and 4% have been measured at 50 °C. The results are shown in Figure 4.

10-5 10-4 10-3 10-2 10-1 100 101 102 103 104 105 106100

101

102

103

Rea

l Pa

rt o

f P

erm

itti

vity

Frequency (Hz)

'

0.7% moisture content 1% moisture content 1.5% moisture content 2.1% moisture content 2.3% moisture content

(a) Real part of permittivity

10-5 10-4 10-3 10-2 10-1 100 101 102 103 104 105 10610-3

10-2

10-1

100

101

102

103

104

Imag

inar

y P

art

of

Per

mit

tivi

ty

Frequency (Hz)

"


(b) Imaginary part of permittivity

Figure 4. Dielectric frequency response of oil-paper with different moisture contents.


At 50 °C, with the moisture content increasing, complex permittivity of oil-paper composite insulation increases significantly at low frequency. However, they tend to be constant at high frequency. Above 10 kHz, the imaginary parts are rising with increasing frequency, which indicates a presence of another loss peak at high frequencies. That is different from the typical dielectric relaxation model equation (e.g., Debye equation, Cole-Cole arc equation, Davidsion-Cole asymmetric arc equation). Consequently, the Cole-Cole expression with two relaxation times was used to fit them. The corresponding real part of permittivity could be written as [20, 22-23]:

1 2

'

1 2(1 ) (1 )

1 2

Re1 ( ) 1 ( )

nA

j j

(1)

The imaginary part of complex permittivity is equal to the Kramers-Kronig(K-K) transformation of formula (1) plus the dc loss [19, 20].

1 2

" dc

0

1 2(1 ) (1 )

1 2

cot (1 )2

Im1 ( ) 1 ( )

nA n

j j

(2)

10-5 10-4 10-3 10-2 10-1 100 101 102 103 104 105 106100

101

102

Measured Fitted

Frequency (Hz)

Re

al P

art

of

Pe

rmit

tiv

ity

'

(a) Measured and fitted real part of permittivity

10-5 10-4 10-3 10-2 10-1 100 101 102 103 104 105 10610-3

10-2

10-1

100

101

102

103

Frequency (Hz)

Measured Fitted

Imag

inar

y P

art

of

Per

mit

tivi

ty

"

(b) Measured and fitted imaginary part of permittivity Figure 5. Measured and fitted complex permittivity of oil-paper with 2% moisture content at 50 °C

The parameters A, n, α1, α2, τ1, τ2, △ε1, △ε2, σdc were used to describe the different contributions to dielectric response. Figure 5 is the measured and fitted complex permittivity of oil-paper composite insulation with 2% moisture content at 50 °C. There are two relaxations at high frequency and low frequency, respectively. The σdc give much contribution at the low frequency range. The decrease part of permittivity with the increasing frequency at low frequency range was modeled by an inverse power dependence on frequency, Aω-n, whereas the remaining part was modeled by Cole-Cole expression [20] with two relaxation times. The 1/2πτ1, △ε1, 1-α1 represent the relaxation center frequency, the relaxation amplitude and the slope of the first Cole-Cole relaxation in the low frequency range, respectively. Its center frequency is about 10-2-102 Hz. So do the α2, τ2, △ε2 to the second Cole-Cole relaxation in the high frequency range, which is mainly contribute to the increase part of permittivity after the minimum. Its center frequency is upper than 109 Hz. The parameters consequently estimated by means of the least square technique to obtain the best fit of the measured permittivity. A good agreement between the measured data and fitted response based on Cole-Cole expression with two relaxation times can be observed. Fitting the dielectric responses of oil-paper with different moisture contents at 50 °C, the results are shown in Figure 6. The estimated parameters by the least squares for the dielectric responses of the oil-paper samples are listed in Table 3.

10-5 10-4 10-3 10-2 10-1 100 101 102 103 104 105 106100

101

102

103

0.7% measured 0.7% fitted 1% measured 1% fitted 1.5% measured 1.5% fitted 2.1% measured 2.1% fitted 2.3% measured 2.3% fitted

Rea

l Par

t o

f P

erm

itti

vity

'


10-5 10-4 10-3 10-2 10-1 100 101 102 103 104 105 10610-3

10-2

10-1

100

101

102

103

104

105


Ima

gin

ary

Par

t o

f P

erm

itti

vity

Frequency (Hz)

"

(b) Measured and fitted imaginary part of permittivity

Figure 6. Measured and fitted complex permittivity of oil-paper with different moisture contents at 50 °C.

Frequency (Hz)


Table 3. Estimated Cole-Cole model parameters of oil-paper with different moisture content at 50 °C.

Parameters Oil-paper with different moisture content at 50 °C

0.5% 1% 2% 3% 4%

A 0.24 0.068 0.8 1.88 0.628

n 0.55 0.68 0.48 0.62 0.785

△ε1 0.6 1.8 0.11 0.1 14.9

τ1 24 28 18 0.00318 0.8

α1 0.96 0.42 0.735 0.8 0.5

△ε2 3.6 3.5 4.2 4.4 4.6

τ2 2.5×10-12 2.6×10-12 3.3×10-12 7.1×10-12 1.9×10-11

α2 0.68 0.75 0.757 0.77 0.73

σdc 8.5×10-14 3.2×10-13 3.3×10-12 7.1×10-12 1.2×10-11

Similarly, dielectric frequency response of oil-paper composite insulation modified by nanoparticles with different moisture contents about 0.5, 1, 2%, 3 and 4% have been measured at 50 °C. The results are shown in Figure 7.

10-5 10-4 10-3 10-2 10-1 100 101 102 103 104 105 106100

101

102

103

104

Rea

l P

art

of

Per

mit

tivi

ty


Frequency (Hz)

'

(a) Real part of permittivity

10-5 10-4 10-3 10-2 10-1 100 101 102 103 104 105 10610-3

10-2

10-1

100

101

102

103

104

Imag

inar

y P

art

of

Per

mit

tivi

ty


Frequency (Hz)

"

(b) Imaginary part of permittivity

Figure 7. Measured complex permittivity of oil-paper modified by nanoparticles with different moisture contents at 50 °C.

Comparing Figures 7 and 4, the dielectric frequency responses of oil-paper composite insulation shows that some changes after adding nanoparticles, especially in the low frequency. The measured data was fitted by Cole-Cole expression with two relaxation times. The result is shown in Figure 8.

10-5 10-4 10-3 10-2 10-1 100 101 102 103 104 105 106100

101

102

103

Measured Fitted

Frequency (Hz)

Rea

l Par

t o

f P

erm

itti

vity

'


10-5 10-4 10-3 10-2 10-1 100 101 102 103 104 105 10610-2

10-1

100

101

102

103

104

Measured Fitted

¡±

Frequency (Hz)

Imag

inar

y P

art

of

Per

mit

tivi

ty


Figure 8. Measured and fitted complex permittivity of oil-paper modified by nanoparticles with 2% moisture content at 50 °C based on Cole-Cole expression with two relaxation times.

There is a clear deviation when the Cole-Cole expression with two relaxation times was used to fit the dielectric frequency response of oil-paper composite insulation modified by nanoparticles, especially at the low frequency. Considering a new relaxation peak appeared below 0.01 Hz, a low frequency relaxation was added to fit the result. The real part of complex permittivity based on Cole-Cole expression with three relaxation times can be obtained:

1 2

3

'

1 2(1 ) (1 )

1 2

3(1 )

3

1 ( ) 1 ( )Re

1 ( )

nA

j j

j

(3)


The corresponding imaginary part of complex permittivity is equal to the Kramers-Kronig (K-K) transformation of formula (3) plus the dc loss.

1 2

3

" dc

0

1 2(1 ) (1 )

1 2

3(1 )

3

cot (1 )2

1 ( ) 1 ( )Im

1 ( )

nA n

j j

j

(4)

The parameters A, n, α1, α2, α3, τ1, τ2, τ3, △ε1, △ε2, △ε3, σdc were consequently estimated by means of the least square technique to obtain the best fit of the measured permittivity.

10-5 10-4 10-3 10-2 10-1 100 101 102 103 104 105 106100

101

102

103

104

Measured Fitted

Frequency (Hz)

Rea

l Pa

rt o

f P

erm

itti

vit

y

'


10-5 10-4 10-3 10-2 10-1 100 101 102 103 104 105 10610-3

10-2

10-1

100

101

102

103

104

¡± Measured Fitted

Frequency (Hz)

Im

agin

ary

Par

t o

f P

erm

itti

vity


Figure 9. Measured and fitted complex permittivity of oil-paper modified by nanoparticles with 2% moisture content at 50 °C based on Cole-Cole expression with three relaxation times.

Figure 9 is the measured and fitted complex permittivity of oil-paper modified by nanoparticles with 2% moisture content at 50 °C based on Cole-Cole expression with three relaxation times. After nanoparticles adding, a new relaxations modeled by the Cole-Cole expression was contributing to the dielectric response. Its center frequency is lower than 10-4 Hz. Compared to Figure 8, the established new relaxation model can reflect the relaxation peak produced by nanoparticles at the low frequency band. Fitting the dielectric responses of nanoparticles modified oil-paper

with different moisture contents at 50 °C, the results are shown in Figure 10. The estimated model parameters by the least squares for the dielectric responses are listed in Table 4.

10-5 10-4 10-3 10-2 10-1 100 101 102 103 104 105 106100

101

102

103

104


Frequency (Hz)

Rea

l P

art

of

Per

mit

tivi

ty

'


10-5 10-4 10-3 10-2 10-1 100 101 102 103 104 105 106

10-3

10-2

10-1

100

101

102

103

104

105


"

Frequency (Hz)

Imag

inar

y P

art

of

Per

mit

tiv

ity


Figure 10. Measured and fitted complex permittivity of oil-paper modified by nanoparticles with different moisture contents at 50 °C. Table 4. Estimated model parameters of oil-paper modified by nanoparticles with different moisture contents at 50 °C.

parameters

Oil-paper modified by nanoparticles with different moisture contents at 50 °C.

0.5% 1% 2% 3% 4%

A 0.038 0.09 0.122 0.1 0.19

n 0.73 0.71 0.68 0.65 0.59

△ε1 3.45 4.52 7.13 9.12 10.9

τ1 935 420 200 110 92

α1 0.42 0.56 0.5 0.45 0.43

△ε2 1.8 5.9 6.2 8.1 8.8

τ2 2.2×10-12 6.5×10-12 8.7×10-12 9.8×10-12 1.1×10-11

α2 0.76 0.72 0.68 0.66 0.65

△ε3 56 98 2023 2902 5256

τ3 1953 1617 978 406 218

α3 0.14 0.16 0.2 0.25 0.28

σdc 7.2×10-14 1.2×10-13 4.5×10-12 8.9×10-12 1.6×10-11


3.2 DIELECTRIC FREQUENCY RESPONSES OF OIL-PAPER COMPOSITE INSULATION AT

DIFFERENT TEMPERATURE In order to study the influences of temperature on the

dielectric frequency response, the complex permittivity of oil-paper composite insulation with 2% moisture content has been measured at 10, 30, 50 and 70 °C. According to the Cole-Cole expression with two relaxation times, the results were fitted. The measured data and fitting curves are shown in Figure 11. The parameters are listed in Table 5.

10-5 10-4 10-3 10-2 10-1 100 101 102 103 104 105 106100

101

102

10°C measured 10°C fitted 30°C measured 30°C fitted 50°C measured 50°C fitted 70°C measured 70°C fitted

Rea

l P

art

of

Per

mit

tivi

ty

Frequency (Hz)

'


10-5 10-4 10-3 10-2 10-1 100 101 102 103 104 105 10610-3

10-2

10-1

100

101

102

103

104

Imag

inar

y P

art

of

Per

mit

tivi

ty


Frequency (Hz)

"


Figure 11. Measured and fitted complex permittivity of oil-paper with 2% moisture content at different temperature Table 5. Estimated model parameters of oil-paper with 2% moisture content at different temperature

Parameters Oil-paper with 2% moisture content at different temperature

10 °C 30 °C 50 °C 70 °C

A 0.073 0.11 0.8 2.5

n 0.44 0.55 0.4 0.45

△ε1 3.9 8 0.11 1.28

τ1 4600 200 18 0.24

α1 0.37 0.12 0.735 0.32

△ε2 3.8 4.4 4.2 3.6

τ2 3×10-10 4.38×10-11 3.33×10-12 1.38×10-12

α2 0.75 0.74 0.757 0.7

σdc 7.8×10-14 4.9×10-13 3.25×10-12 2.8×10-11

Also, the complex permittivity of oil-paper modified by nanoparticles with 2% moisture content has been measured and fitted at different temperature levels of 10, 30, 50 and 70 °C. The Cole-Cole expressions with three relaxation times were established to analyzing the results. The measured data and fitting curves are shown in Figure 12. The parameters are listed in Table 6.

10-5 10-4 10-3 10-2 10-1 100 101 102 103 104 105 106100

101

102

103

104

Rea

l Par

t o

f P

erm

itti

vity


Frequency (Hz)

'


10-5 10-4 10-3 10-2 10-1 100 101 102 103 104 105 10610-3

10-2

10-1

100

101

102

103

104


Imag

inar

y P

art

of

Per

mit

tivi

ty

Frequency (Hz)

¡±


Figure 12. Measured and fitted complex permittivity of oil-paper modified by nanoparticles with 2% moisture content at different temperature. Table 6. Estimated model parameters of oil-paper modified by nanoparticles with 2% moisture content at different temperature.

parameters

Oil-paper modified by nanoparticles with 2% moisture content at different temperature

10 °C 30 °C 50 °C 70 °C

A 0.143 0.192 0.122 0.173 n 0.78 0.74 0.68 0.62

△ε1 5.41 7.62 7.13 5.42 τ1 162 178 200 268 α1 0.16 0.32 0.5 0.362 △ε2 7.25 7.6 6.2 6.8

τ2 3×10-10 1.1×10-11 8.74×10-12 7.2×10-12

α2 0.723 0.72 0.68 0.652 △ε3 23 243 2023 2902 τ3 345 425 978 4675 α3 0.11 0.13 0.2 0.18 σdc 1.25×10-13 8.5×10-13 4.5×10-12 9.7×10-12


4 ANALYSIS AND DISCUSSION

4.1 THE INFLUENCE OF MOISTURE AND

TEMPERATURE ON THE DIELECTRIC FREQUENCY RESPONSE OF OIL-PAPER

COMPOSITE INSULATION

From Figures 6 and 11, an agreement between the measured and fitted responses could be reached. Therefore, the response of oil-paper could be represented by the proposed fitting method. From the Tables 3 and 5, the center frequencies of the two relaxations are 10-2-102 Hz and 1011 -1012 Hz, respectively.

With the increase of moisture content, the real part of complex permittivity of oil-paper composite insulation increases below 10 Hz. The value of the imaginary part of complex permittivity becomes larger. Their shapes remain unchanged. △ε1 tends to increase. △ε2 has a slightly increment also. The time constant τ1 is highly sensitive to the moisture content. τ1 significantly decreases with the increase of moisture content. The polarization at low frequency moves to the high frequency direction. It can be clearly distinguished in Figure 6. On the other hand, increasing moisture content causes a light increment of τ2. The polarization at high frequency slightly shifts to lower frequency.

Figure 11 shows the dielectric response of oil-paper with 2% moisture content at different temperature. The changes of △ ε1 are not obvious, which represent that the temperature has little effect on the low frequency polarization. △ε2 are almost unchanged. But the time constants τ1 and τ2 are very sensitive to temperature. They all decrease with increasing temperature. The overall responses move to the high frequency direction. Figure 11 reflects this tendency.

4.2 THE INFLUENCE OF ADDING

NANOPARTICLES ON THE DIELECTRIC FREQUENCY OF OIL-PAPER

From Figures 10 and 12, there is a relaxation at low frequency range. The relaxation center frequency is corresponding to about 10-4 Hz. Comparing with normal oil-paper composite insulation system, the ε'' also increases at the frequency range. At the same time， ε' greatly increases after adding nanoparticles. The τ1, τ2, △ε1, △ε2

of oil-paper composite insulation modified by nanoparticles present the same trends with normal oil-paper. The △ε3

increases with the increase of temperature and moisture content. The τ3 increases with increase of temperature, the center frequency of the low frequency relaxation shifted to lower frequency. The τ3 decreases with the increase of moisture content, the center frequency of the low frequency relaxation shifts to high.

The addition of metal oxide nanoparticles produces a new polarization mechanism. It increases the real and imaginary part of the complex permittivity of oil-paper composite insulation. Figure 13 shows that the ac (50 Hz) breakdown strength of the oil fluids with and without

nanoparticles, measured in accordance with IEC-156 standard, using the MJD-3 oil breakdown voltage tester manufactured by the Zhuhai Sanchang electric appliance company in China. Figure 14 shows that the partial discharge inception voltage (PDIV) of oil-paper composite insulation with and without nanoparticles. The needle and sphere electrodes were used to test the partial discharge. The gap is about 1 mm, which is equal to the thickness of the oil impregnated paper. The samples were the same with the ones be measured their dielectric responses in the preceding text.

5 10 15 20 25 30 35

30

35

40

45

50

Brea

kdow

n Vo

ltag

e (V

)

mo i s t ur e c ont ent ( ppm )

nor mal o i l nanopar t i c l s modi f i ed o i l

Figure 13. The ac (50 Hz) breakdown strength of the oil fluids with and without nanoparticles

0. 0 0. 5 1. 0 1. 5 2. 0 2. 5 3. 0 3. 5 4. 0 4. 56

7

8

9

nor mal o i l - paper nanopar t i c l s mod i f i ed o i l - paper

part

ial

disc

harg

e in

cept

ion

volt

age

(V)

mo i s t u r e c ont ent i n i mpr egna t ed paper ( % ) Figure 14. The partial discharge inception voltage (PDIV) of oil-paper composite insulation with and without nanoparticles

As Figure 13 shows that the ac breakdown voltage decreases for both the oil with and without nanoparticles when moisture content increases. The nanoparticles improved withstand field strength of the oil. The nanoparticles modified oil shows much less dependence of electric breakdown on the moisture. Adding nanoparticles could be responsible for alleviating the detrimental effect of moisture on the electric breakdown. As Figure 13 shows, the partial discharge inception voltage (PDIV) of oil-paper


with and without nanoparticles confirmed that the nanoparticles improved withstand field strength of the oil-paper composite insulation.

The dielectric loss tanδ was calculated by the formula (3.5) in [20]， the comparison of the oil-paper composite insulation system before and after adding nanoparticles was shown in Figure 15.

10-4 10-3 10-2 10-1 100 101 102 103 104 105 10610-3

10-2

10-1

100

101

0.7% Normal0.7% Modified1% Normal1% Modified1.5% Normal1.5% Modified2.1% Normal2.1% Modified2.3% Normal2.3% Modifiedta

n

Frequency (Hz)

(a) Dielectric loss of oil-paper with different moisture contents at 50 °C before and after modified by nanoparticles.

10-4 10-3 10-2 10-1 100 101 102 103 104 105 10610-3

10-2

10-1

100

101

102

tan

Frequency (Hz)

10C Normal10C Modified30C Normal30C Modified50C Normal50C Modified70C Normal70C Modified

(b) Dielectric loss of oil-paper with 2% moisture content at different temperature before and after modified by nanoparticles Figure 15. Comparison of oil-paper dielectric loss before and after modified by nanoparticles.

Figure 15a shows the dielectric loss of oil-paper with different moisture contents at 50 °C before and after modified by nanoparticles. Compare to normal oil-paper insulation, the dielectric loss of oil-paper modified by nanoparticles decrease below 10-2 Hz or above 10 Hz, while increase within the frequency range from 10-2 Hz to 10 Hz. The moisture content do not changes this rule. As Figure 16a, the addition of nanoparticles is propitious to reduce the dielectric loss of the transformer which works in

the power frequency, although the moisture content increases.

Figure 15b shows the dielectric loss of oil-paper with 2% moisture content at different temperature before and after modified by nanoparticles. The same rule could be concluded comparing to Figure 15a. However, when the temperature gets 70 °C, the dielectric loss of nanoparticles modified oil-paper insulation is more than the normal one from 10-2 Hz to 102 Hz. The high temperature makes the adverse impact to the oil-paper insulation modified by nanoparticles work in the power frequency, as shown in Figure 16b.

0 . 0 0 . 5 1 . 0 1 . 5 2 . 0 2 . 5 3 . 0 3 . 5 4 . 0 4 . 5

0 . 0 0

0 . 0 2

0 . 0 4

0 . 0 6

0 . 0 8

0 . 1 0

0 . 1 2

0 . 1 4

mo i s t u r e c o n t e n t ( p p m )

n o r ma l o i l - p ap e r n a no p a r t i c l s mo d i f i e d o i l - p a p e r

tan

(a) Dielectric loss of oil-paper with different moisture contents at 50 °C before and after modified by nanoparticles at 50 Hz.

10 20 30 40 50 60 700. 000

0. 005

0. 010

0. 015

0. 020

0. 025

0. 030

0. 035

0. 040

0. 045

0. 050

0. 055

0. 060

0. 065 nor mal o i l - paper nanopar t i c l s mod i f i ed o i l - paper

tan

t emper at u r e ( C)

(b) Dielectric loss of oil-paper with 2% moisture content at different temperature before and after modified by nanoparticles at 50 Hz

Figure 16. Comparison of oil-paper dielectric loss before and after modified by nanoparticles at 50 Hz.

5 CONCLUSIONS 1. The relaxation function built based on Cole-Cole

expression with two relaxation times can better reflect the relaxation process of oil-paper composite insulation and micro-polarization mechanism of relaxation. The relaxation parameters are related to the moisture content and temperature.


2. By adding nanoparticles, a new low frequency relaxation phenomena appeared. There are changes of the dielectric frequency response of the oil-paper. The Cole-Cole expression with three relaxation times could be fitted.

3. The nanoparticles improved withstand field strength of the oil. The nanoparticles modified oil shows much less dependence of electric breakdown on the moisture. Adding nanoparticles could be responsible for alleviating the detrimental effect of moisture on the electric breakdown. The partial discharge inception voltage (PDIV) of oil-paper composite insulation with and without nanoparticles confirmed that the nanoparticles improved withstand field strength of the oil-paper composite insulation.

4. Comparing with the normal oil-paper, the real and imaginary part of the complex permittivity of oil-paper composite insulation modified by nanoparticles are increased. The addition of nanoparticles is propitious to reduce the dielectric loss of the transformer which works in the power frequency, although the moisture content increases. However, the high temperature (above 70 °C) makes the adverse impact to the oil-paper insulation modified by nanoparticles work in the power frequency.

ACKNOWLEDGMENT This project is supported by National Nature Science

Foundations of China (51177136, 51107105/E0705), Sichuan youth Science and Technology Foundation (2011JQ0009) and the Sustaining Fund of China Southern Power Grid. The authors would like to thank Electrical Test and Research Institute of Sichuan Province, China for their technical support. The authors would also like to thank Xuancheng Jingrui New Material Co., Ltd for the nanoparticles test.

REFERENCES [1] L. E. Lundgaard, W. Hansen, D. Linhjell, T.J. Painter, “Aging of

oil-impregnated paper in power transformers”. IEEE Trans. Power. Del., Vol.19, pp. 230-239, 2004.

[2] L. J. Zhou, G.N. Wu and J. Liu, “Modeling of Transient Moisture Equilibrium in Oil-paper Insulation”, IEEE Trans. Dielectr. Electr. Insul., Vol.15, pp. 872-878, 2008.

[3] C. Yang, P. C. Irwin and K. Younsi, “The future of nano dielectrics in the electrical power industry”, IEEE Trans. Dielectr. Electr. Insul., Vol.11, pp. 797-807, 2004.

[4] T. Tanaka, G. C. Montanari and R. Mülhaupt, “Polymer nanocomposites as dielectrics and electrical insulation: perspectives for processing technologie, material characterization and future applications”, IEEE Trans. Dielectr. Electr. Insul., Vol.11, pp. 763-784, 2004.

[5] T. J. Lewis, “Interfaces are the Dominant Feature of Dielectrics at the Nanometric Level”, IEEE Trans. Dielectr. Electr. Insul., Vol.11, pp. 739-753, 2004.

[6] M. Roy, J.K. Nelson, R. K. MaeCrone, L. S. Schadler, C. W. Reed and R. Keefe, “Polymer nanocomposite dielectrics: the role of the interface”, IEEE Trans. Dielectr. Electr. Insul., Vol. 12, pp. 739-753, 2005.

[7] B. Berkovski and V. Bashtovoy, Magnetic Fluids and Applications Handbook, New York, Begall House Inc., pp. 657-752, 1996.

[8] Y. X. Zhou, Y. S. Wang, J. H. Tian and Y. C. Sha, “Breakdown characteristics in transformer oil modified by nanoparticles”, High Voltage Engineering, Vol. 36, No. 5, pp. 1155-1159 (in Chinese).

[9] S. Vladimir, H. Arne, R. Arnold, N. David and K. Raj, “AC (60 Hz) and impulse breakdown strength of a colloidal fluid based on transformer oil and magnetite nanoparticles”, IEEE Int’l. Sympos. Electr. Insul., Arlington, Virginia, USA, pp. 619-622, 1998.

[10] T. K. Saha and Z. T. Yao, “Experience with return voltage measurements for assessing insulation conditions in service-aged transformers”, IEEE Trans. Power. Del., Vol. 18, pp. 128-135, 2003.

[11] T. K. Saha and P. Purkait, “Investigation of polarization and depolarization current measurements for the assessment of oil-paper insulation of aged transformers”, IEEE Trans. Dielectr. Electr. Insul., Vol. 11, pp. 144-154, 2004.

[12] D. Linhjell and L. Lundgaard, “Dielectric response of mineral oil impregnated cellulose and the impact of aging”, IEEE Trans. Dielectr. Electr. Insul., Vol. 14, pp. 156-169, 2007.

[13] T. K. Saha and P. Purkait, “Understanding the impacts of moisture and thermal ageing on transformer’s insulation by dielectric response and molecular weight measurements”, IEEE Trans. Dielectr. Electr. Insul., Vol. 15, pp. 568-581, 2008.

[14] T. K. Saha, ‘‘Observation of Giant Dielectric Constant in an Assembly of Ultrafine Ag Particles’’, Phys. Rev. B, Vol. 69, pp. 1254161-125464, 2004.

[15] W. S. Zaengl, “Applications of Dielectric Spectroscopy in Time and Frequency Domain for HV Power Equipment”, IEEE Electr. Insul. Mag., Vol. 19, pp. 9-22, 2003.

[16] J. Blennow and S. M. Gubanski, “Field experiences with measurements of dielectric response in frequency domain for power transformer diagnostics”, IEEE Trans. Dielectr. Electr. Insul., Vol. 21, pp. 681-688, 2006.

[17] C. Ekanayake, S. M. Gubanski, A. Graczkowski and K. Walczak, “Frequency response of oil impregnated pressboard and paper samples for estimating moisture in transformer insulation”. IEEE Trans. Power. Del., Vol. 21, pp. 1309-1317, 2006.

[18] C. Ekanayake, Diagnosis of Moisture in Transformer Insulation- Application of Frequency Domain Spectroscopy”, Ph.D. thesis, Goteborg, Sweden, Chalmers University of Technology, pp. 5-8, 2006.

[19] S. M. Gubanski, P. Boss, G. Csépes and V. Der, “Dielectric response methods for diagnostics of power transformers”, IEEE Electr. Insul. Mag., Vol. 19, No. 3, pp. 12-18, 2003.

[20] A. K. Jonscher, Dielectric relaxation in solids, Chelsea Dielectrics Press, London, pp. 66-110, 1983.

[21] Q. Lu, L. Q. Xiang, J. X. Huang and X. P. Zhao, “Synthesis and rheological properties of TiO2-oil nanofluids”, Chinese J. Material Research. Vol. 22, No.5, pp. 500-504, 2008 (in Chinese).

[22] W.S. Zaengl, "Dielectric Spectroscopy in Time and Frequency Domain for HV Power Equipment, Part I: Theoretical Considerations", IEEE Electr. Insul. Mag., Vol. 19, No. 5, pp. 5-19, 2003.

[23] A. K Jonscher, “Dielectric relaxation in solids”, J. Phys. D: Appl. Phys., Vol. 32, pp. R57–R70, 1999.

Lijun Zhou was born in Hangzhou, China, on 8 May, 1978. He received the B.Sc., M.Sc. and Ph.D. degrees in electrical engineering, respectively in 2001, 2004 and 2007 from Southwest Jiaotong University, Chengdu, where he is a associate professor of electrical engineering. His fields of interest are condition monitoring, fault diagnosis and insulation life-span evaluation for electric power equipment.

Jun Liu was born in Chengdu, China, on 17 September 1984. He received the B.Sc. degree from the Southwest Jiaotong University, Chengdu, China in 2006. He is currently pursuing the Ph.D. degree in the school of Electrical Engineering at Southwest Jiaotong University. He is engaged in research on fault diagnosis for electric power equipment.


Guangning Wu （M’97-SM’07） was born in

Nanjing, China, on 26 July, 1969. He received the B.Sc., M.Sc. and Ph.D. degrees in electrical engineering, from Xi’an Jiaotong University, respectively in 1991, 1994 and 1997. Currently, he is a Professor in the School of Electrical Engineering, Southwest Jiaotong University. His research interests include condition monitoring, fault diagnosis and insulation life-span evaluation for electric power equipment.

Yingfeng Zhao was born in Zhejiang，China，on 18 August 1988. He received the B.Sc. degree from the Southwest Jiaotong University, Chengdu, China in 2010. He is now a postgraduate student of the College of Electrical Engineering of the Southwest Jiaotong University. He is engaged in research on the on-line monitoring and fault diagnosing for electric power equipment.

Ping Liu was born in Guangan，Sichuan, China, in January 1971. He received the Master's degree from the Chongqing University, Chongqing, China, in 2009. He is the superintendent of the high-voltage technology department, Electrical Test and Research Institute of Sichuan Province, mainly engage in high voltage insulation equipment testing and technical management.

Qian Peng was born in Jingzhou, Hubei, China, on 4th February 1983. He received the Master's degree from the Southwest Jiaotong University, Chengdu, China in May 2009. He joined Electrical Test and Research Institute of Sichuan Province in 2009. He is engaged in research on high voltage and insulation technology.