683656

8/10/2019 683656

1/6

Loading Capacity of Standard Oil Transformers onPhotovoltaic Load Profiles

Georg Kerber1 & Prof. Dr.-Ing Rolf Witzmann2

1 Dipl.-Ing. Georg Kerber; Associated Institute of Power Transmission Systems; Technical Universityof Munich Arcisstrae 21; D-80290 Munich;[email protected]; www.een.ei.tum.de 2 Prof. Dr.-Ing Rolf Witzmann; Associated Institute of Power Transmission Systems; TechnicalUniversity of Munich Arcisstrae 21; D-80290 Munich;[email protected]; www.een.ei.tum.de

Abstract

Oil immersed distribution transformers arewidely used in public networks. The rating isoptimized for the known load profile. Theincreasing infeed of distributed generation inthe networks, esp. by photovoltaic(PV)-generators, leads to modified loading profiles.For the investigation of the capacity ofdistribution networks to cope with fluctuatingdistributed generation, the loading capacity ofthe transformers under these conditions is apoint of interest. This paper describes asimulation model to calculate temperature andageing according to the Standard DIN-EN40076-7 [1]. Parameter studies with measuredPV-infeed and ambient temperature profileswere carried out. Indoor and outdoor

disposition is considered. The influence ofuser-load together with PV-generation isanalyzed. The results indicate that a PV-powerof 1.5 times the rated power of the transformerneither leads to overheating nor tounacceptable ageing of the transformer. Allvalues are covered by the recommendations forcyclic loading given in the Standard.

1 Introduction

Due to many reasons, such as CO2-issue,governmental promotion and substanciallyrising prices of oil and other fossil energysources, an increase in dispersed electric powergeneration can be noticed. Therefore theimpact of PV-generators in the public lowvoltage grid emerges as a matter of interest forthe local network operators concerning systemoverloading. One question is the behaviorand/or limitation of standard oil immerseddistribution net transformers to a typical PV-generated fluctuating load profile.At present the rated values of thesetransformers refer to a continuous load at anambient temperature of 40C.

2 State of the Art

Distribution transformers are widelystandardized. The key data is described in DIN 42 500 [2]. The typical units used in Germandistribution grids are 100, 120, 250, 400 and630 kVA transformers [3]. In residential areasthe transformer type is selected assuming amaximum load estimation depending on thetype (grade of electrification) and the numberof households which are supplied [4]. Thosenetworks were only designed for energyconsumption. The increasing infeed by PV-power plants leads to a partial inversion of thetransformer-loadflow in the distribution grids.Due to undetermined behavior of distributiontransformers in case of fluctuating power flow

the rated power of the transformer was set tolimit the PV-infeed. Studies to determine themaximum capacity of low voltage grids forPV-infeed [5] show that for a voltagelimitation of 10 % UN (network nominalvoltage UN = 400 V, 50 Hz) the ratedtransformer power is the main bottleneck.

3 Simulation Model

To test the impact of PV-load profiles on atransformer a thermal model according toDIN-EN 40076-7 [1] has been set up. Thedifferential equations solution was used (seeFig. 1). Input parameters are the load factor K(load current / rated current) and the ambienttemperaturea. The model calculates the oil-and hotspot temperature (o, h). The relativeageing-rate V is calculated according to Eq. 1for non-thermal upgraded paper.

World Renewable Energy Congress (WRECX) Editor A. Sayigh 2008 WREC. All rights reserved. 1198

8/10/2019 683656

2/6

Fig. 1 Block diagram representation of thethermal transformer model [1]

( 98) /62 hV = Eq. 1

The investigated transformer was a 20/0.4 kV,400 kVA, uK = 4% ONAN type. The relevantmodel parameters R, k11, k12, k22 and the timeconstants W and O were calculated accordingto the Standard based on a manufacturerstemperature rise run and the data sheet. Themodel was implemented as a script in acommercial network analysis program. Themodeling was verified with the dynamic loadprofiles given by the Standard, with load stepresponses and a measured temperature rise run.The maximum divergence between the

simulated temperature rise run and themeasured data was below 3 K (Fig. 2). Theverified model is based on an outdoordisposition of the transformer. For simulationsof transformers with enclosure (Kiosk) atemperature correction of 15 K was used asdefined in the Standard.

0

20

40

60

80

0 200 400Time [min]

T e m p e r a

t u r e

[ C ]

Simulation

Measurement

Fig. 2: Simulated and measured top-oiltemperature of the temperature rise run

A long term measurement of the AC-power-output of the inverter of a 1 MW photovoltaicpower plant located in Munich [6] togetherwith the ambient temperature were used tosimulate the PVinfeed. The time resolution ofthe data was 1 min (mean values). Taking intoaccount the hot spot time constant of thetransformer of about 4 min a simulation timestep of 1 min is sufficient [1]. The measuredand simulated time period was the completeyear 2005. The energy-yield with950 kWh/kWp of the PV-plant was slightlybelow average in relation to other years.Therefore the transformers temperatures andageing might be slightly higher in other years.To model PV-generators with a different ratedpower the measuered AC-power-output was

linearly scaled. Symmetric loading wasassumed. The power factor was set to cos() =1.0. As the effect of harmonics on transformertemperatures is not covered by the Standardthis influence was neglected.

To investigate the impact of the user load the15 min. average VDEW standard load profilefor households was taken into account. The1 min values were linearly interpolated to thesimulation time step of 1 min. An annuallyenergy consumption of 4500 kWh/a per

household (4 persons.) was assumed [7].The VDEW standard load profile describes theaverage user load of many households.Because the load fluctuation increases withfewer households the use of the load profile isrestricted to networks with more than 150households [8] But even with less than 150households the simulations are a goodindication for the temperatures to be expectedbecause the effects of a more dynamic loadprofile are dampened by the thermal timeconstants of the transformer.


8/10/2019 683656

3/6

1 2 3 4 5 6 7 8 9 10 11 120

100

200

300

400

500

600

Month of year

P V - A

C P o w e r

[ k W ]

A B

Fig. 3 PV-infeed / transformer load in 2005; (PR = 1.5; LR = 0)

Mark A: max power peak; Mark B: max hotspot temperature

1 2 3 4 5 6 7 8 9 10 11 12-20

0

20

40

60

80

100

120

Month of year

T e m p e r a

t u r e

[ C ]

Hotspot Temp.Oil Temp.Ambient Temp.

BA

Fig. 4 Measured and simulated temperatures; (PR = 1.5; LR = 0)

Mark A: max power peak; Mark B: max hotspot temperature

4 Simulation Results

In order to investigate the impact of differentloading of the transformer on the hot-spot andoil temperature as well as on the ageing thepower ratio PR and the load ratio LR weredefined:

rated power of PV plant PR

rated power of transformer

= Eq. 2

LR number of households loadprofile= Eq. 3

Parameter studies for PR = 0.5 2.25, LR = 0 -330 and indoor and outdoor disposition wereconducted. For every parameter combination asimulation for the whole year 2005 wasperformed.Due to the simple consideration of the housingdefined by the Standard the results withhousing mainly show a temperature offsetcompared to the results without housing. The

temperature behavior versus time and theresulting ageing differ only marginally. As the

main point of interest is the maximum loadingcapacity the results are discussed only forindoor disposition.

Loading Capacity with PV-Load only

Fig. 3 gives the the PV-generator outputrespectively the transformer current for apower ratio PR = 1.5. The measured ambienttemperature together with the calculated oiland hotspot temperature for the investigatedtransformer is shown in Fig. 4.

Fig. 3 indicates that the maximum power isfed in only on few days of the year. Theefficiency of the PV-generator depends on theambient temperature. This effect leads tosignificantly reduced power generation onsunny days because the temperature of the PV-cell is much higher than the nominaltemperature of 20C, which is used todetermine the nominal power output of the cell[9]. Therefore, the maximum power output of


8/10/2019 683656

4/6

the PV-generator occurs on rather cool days inspring and autumn and especially whenmoving clouds are causing high gradients ofsolar radiation on the cells. This behavior isadvantageous for the transformer loadingcapacity because the peak power is generatedonly for a short period of time.

Fig. 4 shows, that the hotspot-temperaturealmost always stays below 100 C, althoughthe ambient temperature changes significantlywith the seasons. This is caused by the fact,that especially on days with a high ambienttemperature, when the transformer loadingcapacity is the limited, the PV-generatorsefficiency is reduced too.

The maximum hotspot temperature is reachedat a quite low PV-generator output of about75 % of the nominal power (Fig. 5 and Fig.6). The temperature peak of oil and hotspot iscaused by the high ambient temperature ofabout 35C and the continuous transformerloading.

0 6 12 18 240

100

200

300

400

500

600

Hour of day

15.07.2005

P V A

C p o w e r

[ k W ]

Fig. 5 Transformer load on the day with themaximum hotspot temperature (PR=1.5; LR=0).

0 6 12 18 240

20

40

60

80

100

120

Hour of day

15.07.2005 ThotToil

Tamb

ThotToil

Tamb

T e m p e r a

t u r e

[ C ]

Fig. 6: Temperatures on the day with themaximum hotspot temperature (PR = 1.5;LR = 0).

The day of the maximum power output of thePV-generator is shown in Fig. 7 and Fig. 8.The low ambient temperature and highlyfluctuating radiation leads to high powerpeaks. On the other hand, the low ambienttemperature and the dampening of thefluctuations by the transformers thermal timeconstants are limiting the transformer hotspotand especially oil temperature on this day.

0 6 12 18 240

100

200

300

400

500

600

Hour of the day

P V A

C p o w e r

[ k W ]

14.05.2005

0 6 12 18 240

100

200

300

400

500

600

Hour of the day

P V A

C p o w e r

[ k W ]

14.05.2005

Fig. 7: Transformer load on the day of themaximum PV-Power peak (PR = 1.5; LR = 0)

0 6 12 18 240

20

40

60

80

100

120

Hour of the day

T e m p e r a

t u r e [ C ]

ThotToilTamb

14.05.2005

Fig. 8: Temperatures on the day with themaximum PV-Power peak (PR = 1.5; LR = 0).

To evaluate the maximum loading capacity fora cyclic loading of the transformer theStandard mentions different criteria to meet.

The hotspot temperature of 120Cshould not be exceeded.

The oil temperature should not exceed105C.

To limit overheating on bushings andcable-end connections and otherdevices the current should be restrictedto 1.5 times the rated current.

Apart from this the cumulated ageing shouldnot exceed 360 days per year to ensure theexpected lifetime.


8/10/2019 683656

5/6

For that reason for every parameter set themaximum oil and hotspot temperature and thecumulated ageing in the simulation periodwere determined.The results for in- and outdoor disposition areshown in Fig. 9 to Fig. 11:

hotspot temperature limit of 120C isreached at PR ~ 1.75.

oil temperature of 105C is reached athigher values of PR (>2) because ofthe high thermal oil time constant(~180 min) compared to the short loadperiods, even on sunny days

ageing is of lower relevance becauseof low average load and ambienttemperature compared to thetransformer design (recommendation

is exceeded at PR > 2)The difference for indoor and outdoordisposition leads to slightly higher tempera-tures and increased ageing for PR < 1.75.

The simulation results indicate that thedecisive criterion for the loading capacity of anoil immersed distribution transformer in caseof PV-power infeed is the limitation of thecurrent to 1.5 times the rated current.Therefore the power ratio PR is limited to 1.5.

1 1.5 2 2.50

50

100

150

T e m p e r a

t u r

i n C

Power ratio P PV,r /ST,r

TH,MAXTO,MAXTH,MEANTO,MEAN

Fig. 9: Transformer temperatures on PV-infeedfor outdoor disposition without user load.

1 1.5 2 2.50

50

100

150

Power ratio PPV,r /ST,r

T e m p e r a

t u r

i n C

TH,MAXTO,MAXTH,MEANTO,MEAN

Fig. 10: Transformer temperatures on PV-infeedfor indoor disposition without user load.

1 1.5 2 2.50

60

120180

240

300

360

A n n u a

l a g e

i n g

i n d a y s

indoor dispositionoutdoor disposition

Power ratio P PV,r /ST,r Fig. 11: Cumulated annual transformer ageingfor indoor and outdoor disposition on PV-infeedwithout user load.

Effects of additional user load

In distribution network transformer stations thePV-infeed is usually combined with the userload. This leads to lower power peaks due tothe compensation of generation andconsumption in the low voltage grid.Therefore, lower temperatures are to beexpected. On the other hand the overnightconsumption increases the utilization of thetransformer and interferes with the nightly cooldown of the transformer. To investigate theseeffects, simulations with additional user loadwere performed. All simulations show that, -compared to exclusive load or PV-infeed - thecombination of both reduces the temperaturesand ageing. This effect can be visualized bestwith a high user load eg. LR = 330 (Fig. 12and Fig. 13).

It can be concluded, that the worst case for atransformer is loading with PV-generationonly. Therefore a limitation of PR = 1.5 evenwith user load is a conservative estimation.


8/10/2019 683656

6/6

TH,MAX

TO,MAX

TH,Mittel

TO,Mitte l

TH,MAX

TO,MAX

TH,Mittel

TO,Mitte l

TH,MAX

TO,MAX

TH,Mittel

TO,Mitte l

0 0.5 1 1.5 20

40

80

120

160

Power ratio P PVr / S Tr

T e m p e r a

t u r e

i n C

Fig. 12: Transformer temperatures on PV-infeed and a user load of 330 households (indoordisposition)

0 0.5 1 1.5 20

0.5

1

1.5

Power ratio P PVr / S Tr

A n n u a

l a g e

i n g

i n d a y s

Fig. 13: Cumulated annual ageing on PV-infeedand a user load of 330 households (indoordisposition)

5 Conclusion

The maximum loading capacity for an oil-immersed distribution transformer wasdetermined by means of simulation based onmeasurements of PV-infeed and ambienttemperature during one whole year (2005). Theeffect of additional user load and transformerhousing was considered.It is shown that a 400 kVA oil immersedtransformer in indoor disposition does notoverheat with a PV power ratio (rated PV-inverter power by rated transformer power)below 1.75. Additional consumption increasesthe maximum loading capacity of thetransformer.According to DIN-EN 40076-7 [1] a current

greater than 1.5 times the nominal currentshould not be used for cyclic loading becauseother components of the transformer station

such as bushings or cable-end terminationscould be overloaded.With a specification of a maximum PV-inverter power of 1.5 times the transformerrated power (PR = 1.5), safe operation of atransformer is given.The marginal overloading capability of PV-inverters ensures a current within theacceptable range for cyclic loading. Byneglecting the influence of the user load, anadditional security margin is available. Theinfluence of annual fluctuations on ambienttemperature and energy-yield is covered by theuncritical transformer temperature and ageingat PR = 1.5.Smaller transformers should be even moretolerant to PV-load profiles, because they are

less vulnerable to loading beyond thenameplate rating.

References

[1] DIN EN 60076-7; Power Transformers-Part 7: Loading guide for oil-immersedpower transformers; Dec. 2004 VDEVerlag

[2] DIN 42500 Drehstrom-l-Verteilungs-transformatoren 50 Hz, 50 bis 2500 kVA;Dec, 1993

[3] Kerber, G. ; Witzmann, R.; StatisticalDistribution Grid Analysis and ReferenceNetwork Generation; ew - Jg 107 (2008),Heft 6; VWEW-Energieverlag

[4] Kaufmann W.; Planung ffentlicherElektrizittsverteilungs-Systeme;VWEW-Energieverlag 1995; ISBN 3-8007-2141-4

[5] Kerber, G.; Witzmann, R.; Capacity ofDistribution Networks due to Power fromPhotovoltaic; ew Jg. 106 (2007), Heft 4,;VWEW-Energieverlag

[6] Solarenergiefrderverein Bayern e.V.;http://www.sev-bayern.de/ ;2008-04-21

[7] Energieverbrauch der privaten Haushalteund der Sektoren Gewerbe, Handel,Dienstleistung (GHE), Final reportBMWI; Project ID 17/02.;www.isi.fhg.de/publ/downloads/isi04b15/ghd-hauptbericht.pdf; 2008-04-21

[8] Dr.-Ing. Klaus Engels; ProbabilistischeBewertung der Spannungsqualitt inVerteilungsnetzen; RTWH Aachen; 2000

[9] Crystalline silicon terrestrial photovoltaic(PV) modules Design qualification andtype approval (IEC 82/269/CD:2001)


Documents

683656