1 of 13 June 2016, IEEMA Journal, Page No. 94 to 98

Parallel Operation of Transformers with Large Non-identical Taps for Reactive Power Compensation

K Rajamani Abhijit Mandal

Reliance Infrastructure Ltd 1.0 Introduction

A PV solar plant of 40MW capacity is established in Dhursar, Rajasthan, India. The

DC output from individual solar panels is converted to AC through inverters. Inverter

outputs are summed up and the consolidated output is stepped to 33kV with 380V /

33kV transformers. The transformers feed 33kV bus of switchyard through over

head lines. In the switchyard, 33kV voltage is further stepped up using two numbers

220/33 kV Step UP Transformers (SUTs). A 32KM overhead line, owned by Power

Plant Operator, connects Dhursar 220kV substation to Deechu substation of State

utility (Grid). The power evacuation scheme is shown in Fig 1. Circuit breaker

positions are omitted to simplify the network details. Details of CSP (Concentrated

Solar Plant), installed at the same location, are omitted as this is not directly

relevant to present analysis and discussions.

2 of 13 June 2016, IEEMA Journal, Page No. 94 to 98

During night time, when the PV plant is down, small auxiliary power to the extent of

500KW is drawn over the Deechu – Dhursar EHV line. In the vicinity of plant, MV or

LV lines are not present that could have supplied the auxiliary power. Tariff meter at

Deechu substation is used for billing purpose towards import of power from grid to

plant. The lightly loaded EHV line generates not so insignificant capacitive charging

MVAR. In the present case, assuming 0.14MVAR/KM for line charging, the 32KM

long line will generate about 4.5MVAR. Depending on the actual voltage at which

EHV line operates, charging VAR will vary (proportional to V2). Though the active

power drawn on the line is (maximum) 0.5MW, because of charging VAR of line, the

tariff meter at grid station registers maximum demand of about 5MVA. Assuming

demand charges of Rs 160/KVA/month, fixed charges work out Rs 8 lacs per

month.

It is desirable to reduce the contract demand to a minimum so that high fixed charges are not paid for drawing just 500KW during night time. One straight forward

and well known solution is to install a shunt reactor of 4 to 5 MVAR at 33kV to nullify

the capacitive charging current from line. The reactor can be switched in during

night time and switched off during day time. This will directly reduce the demand

within 1 MVA.

Another (unconventional) alternative is to operate the two 220/33kV transformers in parallel and deliberately keep the taps of transformers very different (say one at+5%

and the other at -5%). This results in circulating current between the two

transformers. The circulating current produces reactive loss and thus acts like a

shunt reactor. The reactive loss in transformer compensates the capacitive

generation from line. This results in reduced demand from grid station.

This article presents the results obtained from tests done at site operating the

transformer in parallel with non-identical taps.

3 of 13 June 2016, IEEMA Journal, Page No. 94 to 98

2.0 Analysis prior to site testing Before attempting this novel exercise at site, extensive analytical and simulation

studies were done for parallel operation with different taps to estimate the differential

voltage to be kept to reduce the demand at grid substation to less than 1 MVA.

Parameters of Step Up Transformer (SUT) are given below:

Rating: 50 / 60 / 75 MVA (ONAN / ONAF / OFAF)

Voltage: 220 / 33 kV

Tap Range: 10% in steps of 1.25%

Tap 1 242 / 33 kV

Tap 9(N) 220 / 33 kV

Tap 17 198 / 33 kV

Rated Impedance on 75 MVA: 11. 6% on Tap 9 (Nominal)

: 12.08% on Tap 1

: 11.70% on Tap 17

For simulation purposes, transformer impedance is considered as 11.8%.

Base current B = 75 / (1.732 x 33) = 1.312 kA

2.1 Permissible tap range to avoid overfluxing During the testing, transformers should not be subjected to over fluxing condition [1].

The design flux density is 1.7T at all taps. Testing was planned after 7PM when the

PV plant shuts down. Based on recent records of 220kV grid voltage profile after

7PM, the maximum grid voltage expected during testing was 230kV. For applied

voltage of 230kV, the operating flux density at different taps is shown in Table 1.

Table 1

For example:

Operating flux density at Tap 1 = (230/242) x 1.7 = 1.62T

Operating flux density at Tap 9(N) = (230/220) x 1.7 = 1.78T

Operating flux density at Tap 17 = (230/198) x 1.7 = 1.97T

Tap No 1 5 9 10 11 12 14 17

HV Vol kV 242 231 220 217.25 214.5 211.75 206.25 198

HV Vol % +10 +5 N -1.25 -2.5 -3.75 -6.25 -10

BOPE T 1.62 1.69 1.78 1.80 1.82 1.85 1.90 1.97

LV Vol kV 33

4 of 13 June 2016, IEEMA Journal, Page No. 94 to 98

The above gives a clue that initially keep the tap of one transformer at 9 and

progressively change the tap of other transformer towards 1 (positive maximum). In

this way, there is no danger of over fluxing. If the demand from Deechu to Dhursar

does not fall below 1MVA, even after keeping the tap at 1 on one transformer,

change the tap of other transformer towards 17 (negative maximum). But in this case

we must ensure that operating flux density does not exceed saturation flux density of

1.9T. As a measure of abundant caution, it was decided to restrict operating flux

density to below 1.85T. This corresponds to a tap 12. Hence the tap range available

is 1(+10%) for one transformer and 12(-3.75%) for other transformer. Studies were

done varying the taps within this permissible range.

2.2 Approximate differential voltage estimation

Assume reactive compensation requirement = Q = 5MVAR

Q = 5 / 75 = 0.0667pu

XT = 11.8% = 0.118pu

Let differential voltage when taps of the two transformers are non-identical = V

Refer Fig 2. When switch S is closed, circulating current flows. Refer Cl 6.2 [2].

Circulating current C = V / 2XT

Calculated Reactive Loss = C2 x 2XT = V2 / 2XT = Q

V = Sqrt(Q x 2XT) = 0.1255pu

5 of 13 June 2016, IEEMA Journal, Page No. 94 to 98

The approximate voltage difference required is 12.55%. This will create circulating

current that will produce reactive loss of 5 MVAR.

This is verified by detailed load flow simulation described in next section.

2.3 Load flow studies

Refer Fig 3 for base case when both transformers are at nominal tap. The circulating current is zero. The demand from grid is 5.041 MVA.

Next, tap of SUT1 is kept at 8 (+1.25%) while that of SUT2 is unchanged at 9. Refer

Table 2 and Fig 4. The circulating current between the transformers is 69A which

produces reactive loss. The demand from grid reduces to 4.993MVA.

6 of 13 June 2016, IEEMA Journal, Page No. 94 to 98

Table – 2

Tap No (%) Differential

Voltage

(%)

CIR Amps Reactive Compensation

MVAR

Demand from Grid

MVA

SUT1 SUT2 Calculated Measured Calculated Measured Calculated Measured

9 (0) 9 (0) 0 0 0 0 0 5.041 4.855

8

(+1.25) 9 (0) 1.25 69.0 69.92 0.048 0.040 4.993 4.815

7

(+2.50) 9 (0) 2.50 135.5 140.17 0.189 0.125 4.853 4.730

6 (+3.75)

9 (0) 3.75 201.0 205.07 0.415 0.265 4.628 4.590

5

(+5.00) 9 (0) 5.00 265.0 269.60 0.721 0.570 4.324 4.285

4

(+6.25) 9 (0) 6.25 327.5 332.37 1.101 0.945 3.947 3.910

3

(+7.50) 9 (0) 7.50 388.0 392.33 1.548 1.353 3.504 3.505

2

(+8.75) 9 (0) 8.75 447.5 453.52 2.059 1.833 2.999 3.025

1

(+10.0) 9 (0) 10.00 505.5 509.83 2.628 2.376 2.440 2.485

1

(+10.0)

10

(-1.25) 11.25 575.5 581.70 3.410 3.266 1.682 1.605

1

(+10.0)

11

(-2.50) 12.50 647.5 647.10 4.318 4.152 0.859 0.715

7 of 13 June 2016, IEEMA Journal, Page No. 94 to 98

In Fig 5, tap of SUT1 is at 1 (+10%) while that of SUT2 is at 9. The circulating current

is 505.5A. The demand from grid reduces to 2.44MVA.

In Fig 6, tap of SUT1 is at 1 (+10%) while that of SUT2 is at 11(-2.5%). The

circulating current is 647.5A. The demand from grid reduces to 0.859MVA. Thus with

a differential voltage of 12.5%, the demand from grid reduces below 1MVA which is

the desired objective.

8 of 13 June 2016, IEEMA Journal, Page No. 94 to 98

3.0 Testing at site

The above theoretical analysis gave us confidence to go ahead with testing at site.

Before starting the test, all the existing switchyard protections and schemes were

checked and corrective actions where ever required were ensured to prevent

inadvertent tripping during testing. A template was made to note down the following

for each set of taps:

Tap numbers of SUT1 and SUT2

Grid Voltage

MVA and pf from grid as registered in tariff meter at Deechu

Currents on 33kV side of transformers

OTI and WTI readings

Operating current and restraining current as registered by differential

protection for each transformer. Automatic control of OLTC was disabled. Tap changing was done locally. Since this

type of testing is one of a kind and rarely attempted before, engineers were

stationed locally near the transformers to notice any abnormal increase in vibration

or noise during testing.

9 of 13 June 2016, IEEMA Journal, Page No. 94 to 98

3.1 Measurement of circulating current On 33kV side, phase currents (magnitude) for both transformers are measured. The

circulating current is derived as follows:

Three phase currents from SUT1: R1, Y

1, B1

Three phase currents from SUT2: R2, Y

2, B2

Measured Circulating current C = (R1 + Y

1 + B1 + R

2 + Y2 + B

2) / 6

3.2 Measurement of reactive compensation achieved

Reactive power on tariff meter at Deechu end is measured.

(i) With both transformers on nominal tap (Tap 9), Measured reactive power = Q0

(ii) With non-identical taps, measured reactive power = QK

Measured reactive compensation achieved = Q0 - QK

3.3 Measurement of MVA

The MVA demand is a direct measurement read from tariff meter at Deechu end.

3.4 Comparison between measured and calculated values

Refer Table 2.

(i) Testing started with taps of SUT1 and SUT2 kept at nominal values (Tap 9). The

measured values are:

Circulating current 0

Demand = 4.855MVA

Power factor = 0.043

MVAR = 4.8505

(ii) Tap of SUT1 is changed to 8(+1.25%) while tap of SUT2 tap is unchanged.at 9.

The measured values are:

Circulating current = 69.92A

Demand = 4.815MVA

Power factor = 0.044

MVAR = 4.8103

Reactive compensation realized = 4.8505 – 4.8103 = 0.0402MVAR

10 of 13 June 2016, IEEMA Journal, Page No. 94 to 98

(iii) Tap of SUT1 is changed to 7(+2.5%) while tap of SUT2 is unchanged.at 9. The

measured values are:

Circulating current = 140.17A

Demand = 4.73MVA

Power factor = 0.043

MVAR = 4.7256

Reactive compensation realized = 4.8505 – 4.7256 = 0.1249MVAR

(iv) Similar measurements were taken till SUT1 tap is at 1(+10%) with tap of SUT2 is

unchanged at 9. The demand has come down to 2.485MVA (Refer Table 2). Next,

the tap of SUT2 was raised to 10 and then to 11 (-2.5%) with tap of SUT1 at 1. The

measured values are:

Circulating current = 647.1A (49% of RAT)

Demand = 0.715MVA

Power factor = 0.214

MVAR = 0.6984

Reactive compensation realized = 4.8505 – 0.6984 = 4.1521MVAR

(v) Measured reactive compensation for differential voltage of 1.25% is 0.0402

MVAR {Refer Cl(ii) above}. When differential voltage is increased ten times (12.5%),

{Refer Cl(iv) above} the measured reactive compensation increases by almost 100

times to 4.1521 MVAR. This exponential increase in reactive compensation

(proportional to C2) with increase in differential voltage can be seen from Fig 7.

(vi) Further increase in tap of SUT2 to Tap 12 will make the drawl from grid reactive

but the demand will be almost the same with tap of SUT2 at Tap 11. Hence the

testing was terminated with taps of SUT1 and SUT2 at Tap 1 and Tap 11

respectively.

(vii) With tap of SUT1 at 1 and tap of SUT2 at 11, the goal to get the demand at

Deechu below 1MVA is achieved. This corresponds to a differential voltage of 12.5%

and is in line with analytical predictions.

11 of 13 June 2016, IEEMA Journal, Page No. 94 to 98

(viii) Comparisons between calculated values (from load flow studies) and values

obtained from test at site are shown in Fig 7, Fig 8 and Fig 9. The calculated and test

values are in close agreement.

12 of 13 June 2016, IEEMA Journal, Page No. 94 to 98

Minor errors could be attributed to following:

Calculated values assume constant voltage on EHV side. During

measurement at site, grid voltage is not steady and varies when readings are

taken at different instances of time. Grid voltage varied between 225.6kV and

227.4 kV during the testing period.

Calculated values assume constant impedance at all taps. In practice, there

is a small variation in impedance at different taps.

Since the quantity measured is low (less than 5 MVA at 220kV), inherent

meter error can’t be avoided.

(vii) During the entire testing duration transformers were operated under ONAN

conditions. WTI and OTI readings of both the transformers were monitored. The

maximum recorded values were 45C and 42C for WTI and OTI. These are much

below the alarm and trip settings which are in the range of 90C to 100C.

4.0 Acknowledgement The authors acknowledge valuable inputs received from Sanjiv Srivastava on over

fluxing aspects.

The authors thank Sonu Karekar for conducting the load flow studies.

The authors acknowledge the contribution of the site team (K Sheshadri, Anil Jain,

Alpesh Prajapati, Amit Jain, Deepak Paswan and Shivdan) at Dhursar for helping to

conduct this unique test successfully and safely.

The authors are grateful to Tanvi Shrivastava. The technical support extended by her

greatly facilitated in getting the experiment through at site.

13 of 13 June 2016, IEEMA Journal, Page No. 94 to 98

5.0 Conclusion The conventional wisdom during parallel operation of transformers is to keep the

taps of both transformers identical. Specific master – follower control schemes have

been developed for OLTC operation to achieve this ‘golden rule’. The main reason is

to avoid circulating current between transformers which only adds to heating of

transformer. In the present case, the ‘golden rule’ has been deliberately broken. The

taps of both transformers are kept widely different to circulate substantial current

between the transformers. The circulating current produces reactive power loss and

the effect of shunt reactor is achieved without a physical reactor being present. The

reactive loss in transformer compensates capacitive VARs produced in EHV system. This has been successfully demonstrated at site at 220kV level. In India, this may be

one of the few instances where parallel operation with such large deviation in taps at

EHV level has been attempted. The same idea could be extended by system control

operators for mitigating over voltage problems even at grid levels. Another

interesting application could be for testing Differential / REF schemes passing

substantially large primary currents. 6.0 References [1] ‘Transformer engineering – Design and practice’, S V Kulkarni and S A Khaparde,

Marcel Dekker, 2004.

[2] ‘Power transformers - Application guide’, IEC 60076-8, 1997