2012 International Conference on High Voltage Engineering and Application, Shanghai, China, September 17-20, 2012

Measurement of Ion Mobility in Transformer Oils for

HVDC Applications
Lijun Yang1, Mohsen Shafiee Zadeh2, Joachim Schiessling3, Olof Hjortstam3
Yuriy V. Serdyuk2 , Stanislaw M. Gubanski2
1

State Key Laboratory of Power Transmission Equipment & System Security and New Technology
Chongqing University, 400044, Chongqing, China
2
High Voltage Engineering, Chalmers University of Technology
SE-412 96, Gteborg, Sweden
3
Power Technologies, ABB Corporate Research
SE-721 78, Vsters, Sweden

Abstract-Conduction process in dielectric liquids usually

occurs due to ion migration and is therefore characterized by
concentration and mobility of the ionic charge carriers. For this
reason accurately determining these parameters and especially
ion mobility in insulating oils for HVDC applications needs to be
done precisely. In this report various methods for determining ion
mobility are described, including single polarity, reversal polarity
methods in time domain as well as frequency domain based
measurement of dielectric response. The advantages and
disadvantages of each of them are presented and illustrated with
the results of measurements on various oils using two different
test cells. The dependence of ion mobility on measuring voltage
level and temperature are analyzed and compared.

I.

INTRODUCTION

Rising energy demands and global efforts to reduce

greenhouse gases are strong pushes for harvesting renewable
energy sources in remote areas far away from consumption
centers. To transfer bulk power over long distances, High
Voltage Direct Current (HVDC) transmission technology is
utilized. One of the key components in HVDC transmission
systems is a converter transformer, insulations of which
consists basically of oil and paper/pressboard (impregnated
with oil) and is stressed by combined AC and DC voltages.
The latter initiates specific phenomena in the insulation, in
particular, generation and transfer of free charges in its
components. These are associated with charge accumulation on
liquid-solid interfaces and corresponding redistributions of
electric fields in the entire system. The processes are of
dynamic character and their intensities are defined mostly by
conductive properties of the liquid phase. The transient
behavior of HVDC insulation needs to be further understood in
order to properly design HVDC insulation systems for safe
operation during long service life time.
Whereas permittivity is an important parameter for AC
design, conductivity of insulation components determines the
electric field distribution under DC conditions. Conduction
process in dielectric liquids usually occurs due ions migration
and is characterized by concentrations and mobilities of ionic
charge carriers. Therefore, determining these parameters for
insulating oils used in HVDC applications, especially the ion
978-1-4673-4746-4/12/$31.00 2012 IEEE

mobilities, is crucial and needs to be done with high accuracy.

In this report, various methods for determining ion mobility are
described, including single polarity and reversal polarity
methods in time domain as well as that based on measurement
of dielectric response in frequency domain. The advantages
and disadvantages of each of them are presented and illustrated
with the results of measurements performed in various oils
using two different test cells. The dependencies of mobilities
on test voltage level and temperature are examined.
II. REVIEW OF THE METHODS FOR ION MOBILITY
MEASUREMENT

There are two basic techniques to measure ion mobilities in

insulating liquids: single polarity method and reversal polarity
method. In both the methods, so-called time of flight (TOF) is
determined from the measured current flowing through an oil
sample exposed to certain test voltage.
For the single polarity method, a voltage U of positive or
negative polarity is applied to the test cell. Thereafter TOF is
defined as the time interval between the moment of voltage
application and the instant when the resulting current reaches a
local maximum value, as shown schematically in Fig. 1a. The
voltage is switched off when the current is in a quasi-steady
state. Before next measurement can be started, the test cell

464

i (A) U (V)
TOF1

TOF2

U
trest

t (s)

trest

(a)
i (A)
TOF+

U (V)
I TOF
U

t (s)
-U
-I TOF

TOF-

(b)
Figure 1. Time variations of the current in the test cell during (a) single
polarity and (b) reverse polarity measurements.

2012 International Conference on High Voltage Engineering and Application, Shanghai, China, September 17-20, 2012

III.

-10

-10

x 10

x 10

2.5

i(t) /A

i (t) / A

1.5

-1

Effect of the polarity of previously applied test voltage

Test voltages were sequentially applied to the test cell in the
following order: -120V -120V +120V +120V 120V. During each voltage application, the transient current
was recorded until a quasi-steady state was reached and,
afterwards, the voltage was switched off. The resting time of
60 min was used before the next voltage application. The
current traces recorded during the last four voltage applications
are shown in Fig. 2. It is evident that the polarity of the
previously applied voltage affects the current profile. Thus for
the measurements with positive test voltage following the
negative one, the obtained TOF is equal to 31.8 s (Fig. 2b) and
yields the positive ions mobility of 1.510-9 m2/Vs. However if
the previously applied voltage is positive, the measured TOF is
12.0 s (Fig. 2c) and the corresponding mobility is about two
times higher 2.810-9 m2/Vs. For the negative polarity, no peak
in the current is observed if the preceding voltage is negative
(Fig. 2a) while the TOF of 30.8 s is obtained in case of
978-1-4673-4746-4/12/$31.00 2012 IEEE

0.5
0

-2

100

200

300

400

500

100

200

(a)

400

500

(b)

-10

x 10
2

300
t / s

t / s

x 10

-11

TOF=12.0 s

i (t) / A

1.5
1

-5

-10

0.5

-15

TOF=30.75 s

100

200

300
t / s

400

500

100

200

300

400

500

t /s

(c)

(d)

Figure 2. Influence of previously applied polarity on the current profile with

the rest time of 60 min. T=25oC
TABLE I.
TOF AND ION MOBILITIES OBTAINED BY SINGLE POLARITY METHOD
trest (s)

Previous
polarity

TOF (s)

60

+
-

+
12.0
31.8

120

9.5

43.0

COMPARISONS BETWEEN VARIOUS METHODS

A. Single polarity method

Although single polarity method has been used to measure
ion mobilities in dielectric liquids, polymers and liquid crystals
[4-6], our experiences demonstrated that it suffers from poor
reproducibility of the results. The reason for this is the fact that
the peak in the transient current is sometimes hardly obtainable
and, hence, TOF and mobility value cannot be determined. In
addition, the spread of TOFs is large when repeating
measurements under identical voltages. To clarify this
phenomenon, the following tests were performed.

TOF=31.75 s

i(t) /A

must rest for relatively long time trest. Knowing TOF, the
mobility can be found as = d 2/(UTOF), where d is the gap
size [1, 2].
In the reversal polarity method, a step voltage of one polarity
is first applied to the test cell to sweep out the ions from the oil
bulk and to accumulate them in the vicinity of electrode
surface. Thereafter, the polarity is suddenly reversed (the
switching time of the voltage polarity change must be much
faster than the transit time of ions). By doing so, the ions
accumulated at the electrodes are forced to move towards the
opposite electrodes and their mobility is found using TOF from
the current trace recorded after voltage reversal (see Fig. 1b).
When performing the mobility measurements, special
attention must be paid to the selection test voltage levels and
the gap size. A low field stress (below ~1 kV/cm) is preferable
to minimize charge-carrier generation either in the bulk (by
field-enhanced dissociation) and/or at the electrodes (by
injection). Furthermore, the amplitude of the voltage must be
relatively low to prevent electro-hydrodynamic motion of the
fluid leading to enhanced apparent conduction and, at the same
time, higher than possible electrochemical polarization voltage
of the electrodes [3]. In addition, the inter-electrode gap should
be much larger than the Debye length to avoid the conductivity
increase caused by the interpenetration of the electrical ionic
double layers present on the electrode surface.

30.8
Not
measurable
Not
measurable
Not
measurable

Ion mobility
(10-9 m2/Vs)
+
2.8
1.1
1.5
none
3.5

none

7.8

none

positive-negative voltage sequence providing mobility value of

1.110-9 m2/Vs (Fig. 2d). Changing the rest time from 60 min
to 120 min and repeating the above described procedure
provided a new set of TOFs and mobility values (Table I). The
large discrepancies in the mobility values make the results
obtained with the single polarity method hardly acceptable.
Effect of the rest time
In this test, a voltage of -10 V was applied to the cell for
1000 s. It was switched off and the sample was left for resting
during time trest. Thereafter, a voltage +10 V was applied and
the transient current was recorded. By changing the duration of
trest (0 min, 10 min, 20 min, 30 min, 60 min, 120 min and 12
hours), the effect of rest time on the calculated mobility could
be observed. Note that at trest = 0 s, the method becomes
equivalent to the reversal polarity method.
The recorded currents traces are shown in Fig.3. One can
observe that the most pronounced peak corresponds to the rest
time equal to 0 s (marked as No.1). For trest equal to 10 min
and 20 min, the current peaks are still observed, but their
amplitudes become lower and corresponding TOFs are shorter.
Further increase of the rest time leads to a weakening of the
peaks and for the rest times longer than 120 min they become
undetectable and, hence, the TOFs cannot be obtained anymore.
These results clearly show that the single polarity method is

465

2012 International Conference on High Voltage Engineering and Application, Shanghai, China, September 17-20, 2012
x 10

3.2

-11

No.1, 0min
No.2,10min

No.3, 20min

2.8

No.4, 30min

I (A)

No.5

No.5, 60min

No.6

2.6

No.6, 120min
No.7, 12hours

No.1

2.4

No.7

2.2

No.4
No.3
No.2

2
1.8

50

100

t(s)

150

200

240

Figure 3. The effect of rest time on the result of single polarity method.

highly sensitive to the time interval between two consecutive

tests therefore not very useful in practice.
B. Reversal polarity method
In order to examine the reproducibility of the result
obtained with reversal polarity method, the effect of sweeping
out time, tsweep on TOFs was investigated. Fig. 4 presents the
current traces and the measured TOFs recorded for the
amplitude of the test voltage U = 30 V and temperature 40 C.
One can notice that more than 400 s was required for the
current to reach a quasi-steady state. When the sweep out time
was shorter (curves for 129 s and 220 s in Fig. 4a), i.e. the
polarity reversal took place before the transient was completed,
the TOFs were strongly dependent on tsweep (Fig. 4b). While
reversing polarity after the quasi-steady state is being reached,
the TOFs become practically independent of tsweep. The ion
mobilities obtained from the current traces recorded after
reversing the voltage polarity during transient process are
therefore not the true mobilities and therefore a long enough
sweeping out time is necessary for determining them precisely.

measure the mobilities of ions in three types of transformer oils

marked below as A, B and C. The first one is a highly refined
mineral oil with enhanced gas-absorbing properties. The
second one is inhibited naphthenic oil with extremely good
electrical properties and excellent ageing stability. Oil C is
similar to A and, in addition, provides exceptional oxidation
stability and ultra-low sulphur content. Oils A and C are from
the same manufacturer. In the tests, oil sample was placed in a
stainless steel HAEFELY 2903a cell, which is designed as a
cylindrical capacitor with a shielded measuring electrode
(guard ring capacitor) with the gap size of 2 mm, providing the
geometric capacitance of ~60 pF. Keithley 6517A electrometer
was used as both the voltage source and the ampere meter
during the measurements.
The ion mobility values obtained for the three oils are
presented in Fig. 5 as functions of the test voltage and its
polarity. As it is seen, the mobilities remain practically
constant at low test voltages. Once the latter rises beyond a
certain threshold, the mobility starts to increase. The threshold
voltage is different for different oils. For oils A and C, the
threshold values were respectively ~60 V and ~40 V, while for
oil B the threshold was not observed in the investigated voltage
range. This phenomenon might be caused by field-enhanced
dissociation and electrohydrodynamic (EHD) motion of the
fluid in the test cell. The former can take place if the electric
field strength exceeds ~100 V/mm and enhances of the
conduction process in the liquid with further increasing voltage
[7]. The EHD motion is related to macroscopic movement of
oil molecules and can contribute to the drift of ions. Thus, the
-8

10

Measurements of ion mobilities in three types of oils

The reversal polarity method was further utilized to

(m /Vs)

-9

10

oil A, T=25oC

-12

10.5

x 10

-10

10

10

20

10
9.5

60

100

-9

2290

x 10

+
Mobility (m2/Vs)

I (A)

8.5
220

8
7.5

129

7
6.5
6

150 200

(a)

712

U (V)

40

2
Oil B, T=25oC

1
50

100

150

200

250
t (s)

300

350

400

450

10

(a)

20

40

60

120

U (V)

(b)
-8

10

160

(m /Vs)

TOF (s)

140
120

-9

10

100
80

120

250

500

1 000 1 500

10

tsw eep (s)

(b)
Figure 4. Effect of sweeping time on (a) transient current and (b)
O
TOF, U=30V, T=40 C. Numbers in (a) indicate tsweep in seconds.

978-1-4673-4746-4/12/$31.00 2012 IEEE

Oil C,T=25oC

-10

2 500

10

20

30
U (V)

60

80

100

(c)
Figure 5. Effect of the test voltage and its polarity on ions mobility
in (a) oil A, (b) oil B and (c) oil C, T = 25 C

466

2012 International Conference on High Voltage Engineering and Application, Shanghai, China, September 17-20, 2012
10

TABLE II. ION MOBILITY FOR THREE TYPES OF OIL

2

(10 m /VS)
+
7.09
7.60
24.5
20.8
3.31
3.41

Oil type
A
B
C

-10

(10 m /VS)
+
1.63
3.99
0.18
1.6
0.68
0.84

10

The single polarity method yields results highly dependent

on the previous sample history, i.e. earlier applied voltage
polarity and the rest time between measurements. Therefore,
the single polarity method cannot be recommended for
determining the ion mobility in insulating oils. In contrary, the
use of the reversal polarity method allows to obtain reliable
results. For measurements performed at low test voltages, the
obtained mobility values are independent of the voltage level.
When the voltage is increased to a higher level, the apparent
mobility rises, mostly due to the effect of EHD movements or
electric field enhanced dissociation. An attempt to estimate the
ion mobility from results of FDS measurements shows good
agreement with the mobility levels obtained through the
reversal polarity method.

10

10

10

85oC

-2

fc =0.00613Hz

25oC
fc =0.02512Hz

-4

fc =0.01259Hz

Oil A

-6
-3

10

-2

10

-1

10

10
f (Hz)

10

10

10

Figure 6. Frequency dependence of tg for oil A.

TABLE III. ESTIMATED ION MOBILITY BY FDS
T (oC)

fc(Hz)

25
55
85

0.00613
0.01259
0.02512

estimate

measured[9]

(10-10 m2/Vs)
12.3~4.09
25~8.39
50.2, 16.7

(10-10 m2/Vs)
7.09
20.1
27.0

ACKNOWLEDGMENT
The authors would like to thank ABB Transformers in
Ludvika, Sweden, and especially Mats Berglund for financially
supporting the project. The support provided by the National
111 Project (B08036) and Fundamental Research Found for
the Central University of China (CDJRC 10150004) for Dr.
Lijun Yang during her research in Sweden (July 2011 - January
2012) is gratefully acknowledged.
REFERENCES
[1]

[2]

[3]
[4]

CONCLUSION

978-1-4673-4746-4/12/$31.00 2012 IEEE

55oC

measured mobility becomes an apparent mobility, which is

normally larger by an order of magnitude than the true values
for ions [8].
The average values of ion mobilities in the studied oils
measured under the voltages of both polarities below the
thresholds are shown in Table II. As it is seen, the highest
values were obtained for oil B and they are in agreement with
the result of [5], where a value of 25.510-10 m 2/Vs was
registered. The mobilities in oil A are approximately three
times less than those for oil B and the lowest values were
found for oil C. Note also that the standard deviation of the
results are strongest for oil A.
C. Frequency domain spectroscopy
To investigate applicability of FDS method for estimations
of ion mobility, dielectric response measurements were carried
out in the frequency range from 1 mHz to 1 kHz utilizing IDA
200 measuring system. The frequency dependencies of the
dissipation factor for oil A obtained under different measuring
voltages and temperatures are presented in Fig. 6. One may
identify in the spectra a frequency point fc at which tg begins
to behave nonlinearly. If attributing this point to the arrival of
ionic carriers to the electrodes, ion mobility can be evaluated
as = d 2fc/U. The frequencies fc deduced from the results of
FDS measurements in Fig. 6 and the mobilities estimate
calculated with the corresponding test voltages (20-60 V) are
given in Table III. It is notable that the obtained values are
within the same range of magnitude or even close to the earlier
measured ones measured. However, this approach needs further
investigations as it is very difficult to distinguish the inflection
frequency at various voltages only through reading from the
real and imaginary permittivity graphs. One way for
identifying the frequency fc is through analyzing the harmonic
content of the current in the dielectric response measurements.
IV.

1V
60V
120V

tg

-10

[5]
[6]
[7]
[8]
[9]

467

Dikarev B., Karasev G., Romanets R., and Shimon N. "Modelling and
experimental research of the non-stationary processes of conduction and
space charge accumulation in dielectric liquids", Proc. IEEE Int. Conf.
Diel. Liquids, 1999, pp. 33-36.
Watanabe M., Sanui K., Ogata N., Kobayashi T., and Ohtaki Z. "Ionic
conductivity and mobility in network polymers from poly(propylene
oxide) containing lithium perchlorate", J. Appl. Phys., Vol. 57, No.1, pp.
123-128, 1985.
Tobazeon R., Filippini J.C., and Marteau C. "On the measurement of the
conductivity of highly insulating liquids", IEEE Trans. Diel. Elec. Insul.,
Vol. 1, No.6, pp. 1000-1004, 1994
Romanets R., and Yoshino K. "Transient currents in oganic dielectric
liquids, polymers, and liquid crystals", J. Phys., Vol. 49, No. 12A, pp. 5560, 2004.
Zadeh M. S. Measurement of ion mobility in dielectric liquids, MSc
thesis, Chalmers University of Technology, Gothenburg, Sweden, 2011.
Winokur P. S., Roush M. L., and Silverman J. "Ion mobility
measurements in dielectric liquids", J. Chem. Phys., Vol. 63, No. 8, pp.
3478-3489, 1975
Hilaire M., Marteau C., and Tobazeon R. "Apparatus developed for
measurement of the resistivity of highly insulating liquids", IEEE Trans.
Elect. Insul., Vol. 23, No. 4, pp. 779-787, 1988.
Atten P., "Electrohydrodynamic instability and motion induced by
injected space charge in insulating liquids", IEEE Trans. Diel. Elec.
Insul., Vol. 3 No. 1, pp. 1- 17, 1996
Yang L., Gubanski S.M., Serdyuk Y.V. and Schiessling J. Dielectric
properties of transformer oils for HVDC applications, IEEE Trans. Diel.
Elec. Insul., 2012 (submitted).