General rights Copyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright owners and it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights.

Users may download and print one copy of any publication from the public portal for the purpose of private study or research.

You may not further distribute the material or use it for any profit-making activity or commercial gain

You may freely distribute the URL identifying the publication in the public portal If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately and investigate your claim.

Downloaded from orbit.dtu.dk on: Apr 27, 2020

Discharge characteristics in inhomogeneous fields under air flow

Vogel, Stephan; Holbøll, Joachim

Published in:Proceedings of the 25th Nordic Insulation Symposium on Materials, Components and Diagnostics

Publication date:2017

Document VersionPeer reviewed version

Link back to DTU Orbit

Citation (APA):Vogel, S., & Holbøll, J. (2017). Discharge characteristics in inhomogeneous fields under air flow. In Proceedingsof the 25th Nordic Insulation Symposium on Materials, Components and Diagnostics

Discharge characteristics in inhomogeneous fields under air flow

Stephan Vogel, Joachim Holbøll

Technical University of Denmark

AbstractThis research investigates the impact of high velocityair flow on Partial Discharge (PD) patterns generatedin strongly inhomogeneous fields. In the laboratory, aneedle plane electrode configuration was exposed to ahigh electrical DC-field and a laminar air flow up to 22m

s .The needle was connected to a variable DC potential ofup to 100kV over a grounded plate in order to triggerdifferent corona modes. The impact of the air flow on thespace charges created in the vicinity of the electrode isevaluated by means of PD measurements in time domain.The results indicate that the wind increases the frequencyand magnitude of partial discharges in the vicinity ofthe electrode due to an increased rate of space chargeremoval around the tip of the needle and in the gap.The positive polarity shows higher dependency on airflow compared to the negative polarity. It is shown thatpositive breakdown streamer corona can be extinguishedif wind speeds of more than 14.3m

s are applied to theelectrode.

1. IntroductionThe frequency of upward lightning attachment to tallstructures and wind turbines is biased by the height of thestructure, topography of the surrounding environment,meteorological conditions and nearby lightning activity[1]. Recent publications indicate that also the impact ofair flow may have a distinct impact on the distributionof space charges around the Lightning Protection System(LPS) of a structure which in turn may impact thefrequency of the formation of upward lightning leaders[2]. Stationary objects, such as Franklin rods installedon towers, exposed to negative charged thundercloudscreate positive charged space charges, which slowlypropagate towards the cloud and hereby reduce theelectric field around the extremities of the buildings.In the case of rotating wind turbines, generated spacecharges originating from the receptors are readily sweptaway from the electrode due to the dominant angularvelocity of the rotation. Therefore, the resulting electricfield around the leading edge of the wind turbine isexpected to be higher as compared to stationary objects,leading to a theoretical higher rate of initiation of upwardlightning leaders. This is of particular interest in winterlightning areas where frequent self-initiated upwardlightning discharges are observed from tall structures.

The source of space charges are PD in air, also known ascorona discharges, which are a well studied phenomenon

and several publications describe studies on the generaldischarge behaviour of positive and negative corona [3][4] [5]. Furthermore, the propagation of space chargein the atmosphere is well understood and studies of iondevelopment in space and time can be found for examplein [6] [7] [8]. Unlike electrons, ions are characterised bya relatively high mass and a comparably low propagationspeed, which enables the interaction with air flow such aswind[9].

Only few publications, however, deal with the impactof air flow to corona discharges. F.D. Alessandro [10]studied the impact of wind to negative and positivecorona currents at 3300m above sea level on earthed pinneedles created by the electric field of a thunderstormwhere natural wind speeds up to 14m

s were observed. In[11] the impact of air flow up to 16m

s was investigatedby means of a spark gap and a wind tunnel. The authorshowed that the onset corona voltage in his case wasindependent of the wind speed, however, there was anincrease in the measured current while wind was appliedto the gap. Similar findings are reported in [12]. Themagnitude of corona current varied in these publicationsfrom approx. 1uA to 300uA. All the above mentionedpublications investigated DC corona currents.

To the knowledge of the authors, so far no systematicevaluation of PD pattern exposed to strong windvelocities up to v = 22m

s has been performed.

The present work is structured as follows: At firstthe laboratory setup is described and some moredetails regarding the measurement setup is provided.Subsequently, negative and positive corona dischargepatterns are illustrated and the impact of the air flowis discussed. The measurements were performed in thelaboratory of the Technical University of Denmark.

2. Experimental laboratory setup of DC-corona under air flow

A high voltage DC source up to 100kV with 1mAmaximum current output is connected to a needle-planespark gap with distance d = 20cm and time domainPD measurements are performed with and without wind.The schematic of the setup is illustrated in Figure 1and an overview of layout is presented in Figure 2.The purpose of this test is to investigate if air flowcan change the patterns of discharges generated in theclose vicinity of the needle tip. The needle had adiameter of 4mm and an approximate tip diameter of

1mm. The PD measurements were performed by meansof a Power Diagnostix ICM Measurement System witha bandwidth of 40kHz to 800kHz, which detectedthe transient charge displacement across the spark gapvia a measurement capacitor CM = 400pC. Deadtime between measurements was set to the lowestpossible value of 2us. Lower noise threshold wasset to 2% of the respective PD measurement scale.Furthermore, acoustic measurements for PD localisationwere performed by means of an ultrasonic PD detector.A bleeding resistor RB = 330GΩ was added to ensuresave discharge of the measurement setup when powereddown. The DC voltage of 100kV applied to a needle-plane electrode configuration of 20cm represents nearbreakdown conditions, so all corona modes could bestudied. The wind source was a modified industrialfan for buildings which produced a laminar air flow onan rectangular air-outlet surface of 61cm times 30cm.The needle electrode was mounted 40cm from the outletof the fan and a maximum wind speed of 22m

s wasmeasured in the plane of the electrode.

+

PDMeas.

UDC CM RB

air ow

Fig. 1 – Circuit diagram for PD measurement on needle-planespark gap exposed to wind flow. The DC voltage UDC

is varied between 0 and 90kV. The air flow was 22ms

atn = 1500rpm. Partial discharges were measured with thelow impedance PD measurement impedance in series withelectrode gap and across the measurement capacitor CM .The bleeding resistor acts as safety equipment.

2.1. Methodology of measurements

The methodology for the PD measurements is thefollowing. Constant DC voltage was applied to theneedle tip and the PD patterns were measured during100 seconds. From t = 0s to t = 20s there was noair movement in the gap. At t = 20s the industrialfan was turned on and the wind speed increased in theelectrode configuration. The fan accelerated linearlyfrom n = 0rpm to n = 1500rpm in ten seconds. Fromt = 30s − 40s the maximum wind speed of v = 22m

swas applied. At t = 40s the fan was turned off andthe fan speed linearly decreased within ten seconds. Itneeds to be noticed that air movement continued slightly

Fig. 2 – Fan and needle-plane electrode configuration with PDmeasurement equipment.

Table 1 – Rotational speed vs. wind speed of industrial fan

Rotational Speed Wind Speed[rpm]

[ms

]250 3.4500 7.1750 10.51000 14.21250 17.51500 21.8

in the gap after the fan reached idle state due to theinertia of the air. At t = 60s, the fan was switched-onagain and the processed was repeated in similar fashion.Both positive and negative polarities at the needle wereevaluated and the results are presented the Figure 3 and4. Each illustration contains nine diagrams which showthe behaviour at discrete voltage levels. On the top ofeach illustration, the time dependent rotational speed ofthe fan is illustrated. The correlation between steady statewind speed and rotational speed is given in Table 1.

2.2. Negative corona

Figure 3 illustrates the impact of air flow to the PDpatterns under negative polarity at the needle. Thecorona onset was determined for the negative polarityat UDC = 3kV . Compared to the positive polarity, thePD started earlier since initiating electrons are readilyavailable from the negatively charged electrode. Whenincreasing the applied voltage, the negative coronaexperiences three different modes, which are Trichelstreamer (or Trichel pulses), pulseless negative glow,and negative streamer. Each mode has characteristicproperties as described by [13]. For further informationregarding corona modes please refer to the given source.

Trichel streamer corona was identified betweenUDC = 3kV − 40kV , pulseless negative glowbetween UDC = 50kV − 60kV and negative streamerfrom UDC = 70kV . The main discharge band ofnegative PD are located between a few picocoulombs

0.0 50.0 100.0[s] 0.0 50.0 100.0[s] 0.0 50.0 100.0[s]

0.0 50.0 100.0[s] 0.0 50.0 100.0[s] 0.0 50.0 100.0[s]

0.0 50.0 100.0[s] 0.0 50.0 100.0[s] 0.0 50.0 100.0[s]

137

68.4

0

[pC]

137

68.4

0

[pC]

137

68.4

0

[pC]

137

68.4

0

[pC]

137

68.4

0

[pC]

137

68.4

0

[pC]

137

68.4

0

[pC]

137

68.4

0

[pC]

137

68.4

0

[pC]

n

[rpm]

0

1500n

[rpm]

0

1500n

[rpm]

0

1500

fan

on

fan

on

fan

o

fan

o

fan

on

fan

on

fan

o

fan

o

fan

on

fan

on

fan

o

fan

o

10 kV 20 kV 30 kV

40 kV 50 kV 60 kV

80 kV 90 kV70 kV

Fig. 3 – Negative polarity: 100 seconds of PD measurements with air flow from t = 20s − 50s and t = 60s − 90s. Trichel streamercorona mode from UDC = 3kV − 40kV , pulseless negative glow corona from UDC = 50kV − 60kV , streamer corona fromUDC = 70kV − 90kV .

and 137 picoculomb. At the higher DC voltages aboveUDC = 70kV , there were also repetitive dischargesobserved which were characterised by a low frequencyand a charge magnitude of up to several nanocoulombs(Frequency here is understood as pulse repetition rate).These results are related to negative streamer modeswhich reached further into the gap.

As can be seen in Figure 3, at the negative polarity, the airflow increases mainly the magnitude of the discharges.In the Trichel streamer mode between UDC = 3kV −40kV , the maximum amplitude increases between 5%and 100% depending on the DC voltage. Since thedischarges are spread within a wider band, also thefrequency of discharges increases. When air flow isapplied in the negative glow corona state between UDC =50kV − 60kV , the characteristic high frequency PDpattern increase slightly in magnitude and various largerdischarges are superimposed on the measurement. Theair flow changes the PD pattern from pulseless negativeglow to negative streamer corona. From UDC = 70kV −

90kV , the air flow increases slightly the frequency andmagnitude of the PD pattern.

2.3. Positive corona

Four different positive corona modes can be distinguishedwhich are burst corona, onset streamer, glow corona,and breakdown streamer [13]. The measurement resultsare illustrated in Figure 4. Corona onset for the positivepolarity was determined to Uonset = 12kV . Burst coronaand onset streamer corona can appear within a very shortvoltage range [13] and were determined simultaneouslybetween a voltage range of UDC = 12kV − 14kV .Compared to the negative polarity, the PD of the initialpositive corona mode was characterised by a highercharge magnitude (up to Q = 500pC) and a lowerfrequency. This is in good agreement with the gasdischarge theory since continious availability of startingelectrons at the negative polarity enable more frequentdischarges with a lower amplitude, whereas PD atpositive biased electrodes trigger at random electronsin the ambient air [14]. The glow corona mode started

0.0 50.0 100.0[s] 0.0 50.0 100.0[s] 0.0 50.0 100.0[s]

0.0 50.0 100.0[s] 0.0 50.0 100.0[s] 0.0 50.0 100.0[s]

0.0 50.0 100.0[s] 0.0 50.0 100.0[s] 0.0 50.0 100.0[s]

547

273

0

[pC]

547

273

0

[pC]

547

273

0

[pC]

547

273

0

[pC]

2.73

1.73

0

[nC]

2.73

1.73

0

[nC]

2.73

1.73

0

[nC]

2.73

1.73

0

[nC]

2.73

1.73

0

[nC]

n

[rpm]

0

1500n

[rpm]

0

1500n

[rpm]

0

1500

fan

on

fan

on

fan

o

fan

o

fan

on

fan

on

fan

o

fan

o

fan

on

fan

on

fan

o

fan

o

12 kV 20 kV 30 kV

40 kV 50 kV 60 kV

80 kV 90 kV70 kV

Fig. 4 – Positive polarity: 100 seconds of PD measurements with air flow from t = 20s − 50s and t = 60s − 90s. Burst corona andonset streamer corona mode from UDC = 12kV − 14kV , positive glow corona from UDC = 14kV − 50kV , breakdown streamercorona from UDC = 60kV − 90kV .

at approximately UDC = 14kV and continued untilUDC = 50kV . Few positive glow corona dischargeswere detected with the PD-measurement equipment.The reason for this appearance is most likely the highfrequency of glow corona pulses in DC fields of severalMegaherz [5]. With an upper cut-off frequency of800kHz from the PD measurement equipment, manypulses may be not able to be detected. Breakdownstreamer corona starts at UDC = 60kV and continuesuntil UDC = 90kV . Notice, the y-axis scale in Figure4 changes from Q = 547pC to Q = 2.73nC. Withincreasing voltage, the frequency of the pulses as well asthe magnitude increases.

The impact of the air flow is more pronounced at positivepolarity compared to negative polarity. The strongestimpact can be seen at burst corona/onset streamer coronaand breakdown streamer corona as can be seen in Figure4. The reason for the observation is that streamerspropagate further into the gap and are not only containedaround the vicinity of the gap. When partial discharge

channel propagate further into the gap, also the air flowhas higher impact since it can influence the movement ofcharged ions.

In the case of UDC = 12kV illustrated in Figure 4,the frequency of pulses increase significantly, whereasthe magnitude of pulses decreases slightly. Glow coronafrom UDC = 20kV - UDC = 30kV could not bemeasured with the PD-measurement equipment withoutair flow, however, when air flow was applied, pulsesclearly could be detected. A possible explanation is thatthe air flow removes the shielding ions from the electrode,leading to discharges which are higher in magnitude andlower in frequency. From UDC = 60kV - UDC = 90kVbreakdown streamer corona appears. It can be observedthat the magnitude of PD increase while the fan is startingup and the air flow is increasing in the gap. If the airflow reaches a critical wind speed, the occurrence of PDstops. The breakdown streamer corona is extinguishedand possibly transformed into a reduced glow coronadischarge pattern which takes place only in the vicinityof the electrode.

Positive breakdown streamer corona extinguished byhigh air flow

0.0 50.0 140.0[s]

2.73

1.73

0

[nC]

v

[m/s]

0

22 21.817.5

14.210.5

7.13.4

0

70 kV

0.0 50.0 140.0[s]

1.3E4

6.5E3

0

[N]

Fig. 5 – Breakdown streamer corona exposed to variableamount of air flow. The magnitude of PD increases withincreasing air flow and the frequency (number of discharges)of PD increases first at v = 3.4m

sand decreases from

v = 3.4ms

until breakdown streamer corona is extinguished.

In order to see the impact of air flow to the breakdownstreamer corona, a separate experiment was performedwhere the wind magnitude was varied at a steady DCvoltage of UDC = 70kV . As can be seen in Figure 5,already at slight wind speeds of v = 3.4m

s the magnitudeof PD discharges increases as well as the frequency of thedischarges. At v = 7.1m

s , the magnitude does not varydistinctly; however, the frequency (Number of dischargesN ) of the PD decreases. At v = 10.5m

s , the magnitude ofPD is further increased but the amount of discharges aresignificantly reduced. From v = 14.2m

s , the breakdownstreamer corona is extinguished. There remains a highfrequency PD pattern which is of similar nature of theglow corona mode.

2.4. Comparison between the impact of air flowfrom positive and negative polarity

Following conclusions can be drawn from the laboratoryexperiment:

1. Corona caused by negative polarity is characterised

by lower charge magnitudes compared to positivepolarity, however:

2. The frequency of the discharges at negative polarityis much higher.

3. The corona discharge pattern of the positive polarityis more influenced by the effect of wind comparedto negative polarity, especially for DC voltages overUDC = 60kV .

4. Positive breakdown streamer corona can be extin-guished or reduced to glow corona mode whenhigher wind speeds than v = 14.2m

s are exposedto the electrode.

3. ConclusionIn this paper, the impact of air flow on dischargecharacteristics in inhomogeneous fields was investigatedand clearly confirmed. PD measurements wereperformed and the impact of wind on the different coronamodes was demonstrated. The time domain PD patternof the positive needle polarity shows higher dependenceon air flow compared to the negative polarity. Airflow increases the magnitude and frequency of coronadischarges in the needle-plane electrode configuration,proofing the importance of the space charges in thegap and their displacement by means of the air flow.An exception is the positive breakdown streamer coronaUDC > 60kV which is extinguished by air flow fromv = 14m

s .

4. AcknowledgementsThe research has been partly funded by the DanishEnergy Agency through the Energy Technology Devel-opment and Demonstration Program (EUDP), as part ofthe ELITE [15] project.

[1] Stephan Vogel, Joachim Holbøll, Javier Lopez,Anna Candela’ Garolera, and Søren Find Madsen.Lightning Attachment Estimation to Wind Turbinesby Utilizing Lightning Location Systems. In Inter-national Lightning Detection Conference, 2016.

[2] Joan Montonaya, Oscar van der Velde, and EarlWilliams. Lightning discharges produced bywind turbines. Journal of Geophysical Research,119:1455–1462, 2014.

[3] Leonard B. Loeb and Arthur F. Kip. Electricaldischarges in air at atmospheric pressure: Thenature of the positive and negative point-to-planecoronas and the mechanism of spark propagation.Journal of Applied Physics, 10(3):142–160, 1939.

[4] R. Morrow. Theory of negative corona in oxygen.Physical Review A, 32(3):1799–1809, 1985.

[5] R Morrow. The theory of positive glowcorona. Journal of Physics D: Applied Physics,30(22):3099–3114, 1999.

[6] Serge Soula and Serge Chauzy. Multilevelmeasurement of the electric field underneath athundercloud: 2. Dynamical evolution of a groundspace charge layer. Journal of GeophysicalResearch, 96(D12):22327, 1991.

[7] Marley Becerra, V. Cooray, S. Soula, and S. Chauzy.Effect of the space charge layer created by coronaat ground level on the inception of upward lightningleaders from tall towers. Journal of GeophysicalResearch Atmospheres, 112(12):1–11, 2007.

[8] Marley Becerra. Glow corona discharges and theireffect on lightning attachment: Revisited. 2012 31stInternational Conference on Lightning Protection,ICLP 2012, pages 2–6, 2012.

[9] Stephan Vogel, Javier Lopez, Joachim Holbøll,Søren Find Madsen, Anna Candela Garolera, andKim Bertelsen. Numerical Simulation of the effectof wind removing the corona space charge overgrounded structures under thunderstorm conditions.In International Conference on Lightning & StaticElectricity, 2015.

[10] F. D’Alessandro. Experimental study of the effect ofwind on positive and negative corona from a sharppoint in a thunderstorm. Journal of Electrostatics,67(2-3):482–487, 2009.

[11] David F Ariza, Oscar R Montero, Francisco JRoman, and Oscar F Escobar. Influence of theW ind for P ositive C orona in a S ingle- needle-electrode on T hunderstorm-like C onditions. InInternational Symposium on Lightning Protection(SIPDA), pages 101–106, 2011.

[12] F. Avella, J. Dıaz, D. Ariza, O. Escobar, andF. Roman. Avella, F., Dıaz, J., Ariza, D.,Escobar, O., & Roman, F. (2012). Influenceof thunderstorm-like wind velocities on pointdischarge corona currents. 2012 31st InternationalConference on Lightning Protection, ICLP 2012,0–3. https://doi.org/10.1109/ICLP.2. In 31stInternational Conference on Lightning Protection,ICLP, pages 0–3, 2012.

[13] N. Giao Trinh. Partial Discharge XIX: Discharge inAir Part I: Physical Mechanisms. IEEE ElectricalInsulation Magazine, 11(2):23–29, 1995.

[14] Loeb. Electrical coronas. University of CaliforniaPress, Cambridge University Press, 1965.

[15] S F Madsen, S Vogel, J Lopez, A C Garolera,and K Bertelsen. ELITE – ENHANCED LIGHT-NING EFFECTS TESTING CAPABILITIES. InInternational Conference on Lightning and StaticElectricity (ICOLSE), pages 1–8, 2015.