Вы находитесь на странице: 1из 11


Upul Dompege1 , J.P.Karunadasa2, Kusum Shanthi3


Transmission Projects Branch, Ceylon Electricity Board, Sri Lanka Department of Electrical Engineering, University of Moratuwa, Sri Lanka 3 Transmission O & MS Branch, Ceylon Electricity Board, Sri Lanka 1 udompage@yahoo.com 2karu@elect.mrt.ac.lk 3kusumshanthi@ieee.org

Abstract: Ceylon Electricity Board (CEB) as many other utilities uses breaker switched capacitor (BSC) banks for voltage support and reactive power compensation in grid substations. At present it has a 320Mvar installed capacity in 33kV level and according to CEB transmission plan 70Mvar more to be added in next few years. The main intentions of the use of capacitor banks is to give voltage support at the substation level, reduction of losses in power transformers and transmission lines, and to release the capacity constraints in transformers and lines. CEB uses power factor regulation for switching these capacitor banks for above purposes but no studies have been done to evaluate its suitability. It is learnt that the switching based on this criteria does not fully match with the system requirement and therefore sometimes necessary to manually switch on them overriding the auto controllers or vice versa. Optimizing the use such an economical reactive power source to the specific intention for which they have been installed is a key issue to be addressed. The paper describes the works carried out to evaluate the present switching criteria of BSC banks in CEB and the proposal of a economical way of switching with considering all technical aspects related to capacitor bank operations in medium voltage level including simulation and real time monitoring.. I INTRODUCTION The 33kV capacitor banks in the CEB network are connected to the 33kV load bus at Grid sub stations. However at Pannipitiya the capacitors are connected to the 33kV tertiary winding of the 220 / 132 / 33 kV inter bus transformer. At all locations, the switching ON criterion is based on power factor at 33kV transformer incoming feeder. Switching off is based on leading reactive power limit or leading power factor. If voltage support is necessary, the banks should be switched considering the voltage at the point of connection. If the capacity constraints or loss minimization is concerned, then they shall be fully utilized to minimize drawing var from remote generation. Under these considerations, why CEB controls them in an indirect way like power factor is a question. It should be checked whether the requirements are best met with or the available resources are fully utilized with the present switching criteria [1]. As observed, there are situations where some of the 33kV capacitor banks at the grid substation are kept unused, while having an acute problem of heavy reactive power requirement in transmission system. This happens

mostly when power across the companys transmission system does not coincide with load conditions in locations where the capacitor banks are fixed. In some situations, the power factor may be within acceptable limits but the voltages are below the nominal or on load tap changer is forced on higher taps to take care of the voltage. The substation level capacitor bank can directly serve for voltage support or var support, without depending on power factor regulation which is an indirect measure of voltage or var requirement. The objective of this paper is to fill this void by presenting the work carried out in following areas. verify the applicability of present switching criteria check and ensure the possibility of connecting maximum capacitor banks installed without violating technical constraints review and optimizing the present switching parameters, if the present switching criteria is the optimal solution for the CEB. and to design and propose a suitable switching criteria for the capacitors by means of network simulation and practical implementation with continuous monitoring. II SITE SELECTION Precise data at the substations is beneficial for such an analysis but studying the total system is practically impossible in a live system. However, a case study is a sufficient and satisfactory solution for a research like this. The duration of data measurements shall cover a substantial duration to represent the actual system variations. The general practice of such a study is to have one week duration. Sub station at Panadura was selected as a pilot station and the research was based on the findings for this sub station. The load curves both real and reactive were compared with the system behaviour and found satisfactorily matching and representing the system as a whole. Details of the substation are as follows. Sub station capacity 2 x 31.5 transformers Incoming feeders T connection to Pannipitiya Matugama line / Double circuit No of feeders 6 No of capacitor banks 4 x 5 Mvar Maximum average night peak 46MW +27Mvar Minimum average load 19MW +12Mvar

LEM Qwave Premium power quality analyzer and Ellite 4 Pholyphase power meter was used for data measurement. MW and Mvar, 33kV bus voltage, power factor at 33kV incomer TF 1, tap position of on load tap changer TF 1, load side harmonics were measured at 33kV side. MW & Mvar, power Factor at 132kV bus bar and 132kV bus voltage were recorded at the 132kV level. III SWITCHING CRITERIA OF CEB There are two types of switching methods in the CEB system. In both types the criterion for switching on the banks is the lagging power factor. The controller evaluates the power factor of the 33kV transformer incomer feeder using voltage and current analogue signals and switches the first filter bank when the power factor is below a certain specified limit. Generally, this limit is 0.9800. The next banks are switched on as per the same condition considering the calculated power factor. In one type of controllers, switching off is based on leading power factor. In the second type, controller compares the reactive power calculated using measured power factor and measured the real power with the reactive power calculated using the set power factor and measured real power.

If the difference is greater than a multiple of minimum step of the banks, then the banks are switched off gradually. This multiple is calculated as (1+Hysterisis) where the hysteresis setting is generally about 10% [2]. In the CEB system, if more than one controller is used for set of banks on each bus section, these works as independent controllers when the bus section is open and in master slave mode if the bus section is in ON position. In independent operation, the controller switches the banks assigned to it, typically two. First is always the filter bank and compensator bank later. In the master slave mode, the master will control all the banks if the communication between the controllers is established IV PRESENT SWITCHING PATTERN The figure 1a below shows the behaviour of the capacitor banks over the full range of measurements (9 days) with the present switching criteria. It is more elaborated in figures 1b and 1c in a days window for two selected days. The figures 2a and 2b indicate HV side voltage and reactive power requirement at 33 bus, for same two days.

Utilization of Cap Banks under present sytem


4 N of ca b k o p an s

20.01.2009 21:30:00 21.01.2009 17:30:00 22.01.2009 13:30:00 23.01.2009 09:30:00 24.01.2009 05:30:00 25.01.2009 01:30:00 25.01.2009 21:30:00 26.01.2009 17:30:00 27.01.2009 13:30:00 28.01.2009 09:30:00

Time of Day No of Cap Banks at Panadura GSS at present scheme No of Max Cap Banks If only pF control is used

Fig.1a Switching pattern of capacitor banks over full measurement period

Utilization of Cap Banks (22.02.09)
45.00 40.00 35.00 30.00 25.00 20.00 15.00

Utilization of Cap Banks (24.02.09)

6 45 40 35 5 30 25 20 15 4 5 6

No of Cap Banks

Phase angle

10.00 5.00 0.00 -5.00 -10.00 -15.00 -20.00 -25.00 -30.00 -35.00 -40.00 -45.00

Phase angle

10 5 0 -5 -10 -15 -20 -25 2 3

-30 -35 -40

22.01.2009 00:00:00

22.01.2009 03:20:00

22.01.2009 06:40:00

22.01.2009 10:00:00

22.01.2009 13:20:00

22.01.2009 16:40:00

22.01.2009 20:00:00

0 22.01.2009 23:20:00

-45 24.01.2009 00:00:00

24.01.2009 03:20:00

24.01.2009 06:40:00

24.01.2009 10:00:00

24.01.2009 13:20:00

24.01.2009 16:40:00

24.01.2009 20:00:00

0 24.01.2009 23:20:00

Time of Day Phase Angle at 33 bus (no Caps) Ph. angle (Nominal) No of Max Cap Banks Phase Angle with Cap Banks (calculated) No of Cap Banks
Phase Angle at 33 bus (no Caps) Ph. angle (Nominal) No of Max Cap Banks

Time of Day
Phase Angle with Cap Banks (calculated) No of Cap Banks

Fig.1b Switching pattern of capacitor banks 22nd Jan 2009

Fig.1c Switching pattern of capacitor banks 24th Jan 2009

No of Cap Banks

Reactive power and HV side voltage (22.01.09)

20.00 80.00 78.00 15.00 76.00 74.00

Reactive power and HV side voltage (24.01.09)

20.00 78.00 76.00 15.00 74.00 10.00 72.00 70.00 5.00 68.00 0.00 66.00 64.00 -5.00 62.00 -10.00 24.01.2009 00:00:00 24.01.2009 03:20:00 24.01.2009 06:40:00 24.01.2009 10:00:00 24.01.2009 13:20:00 24.01.2009 16:40:00 24.01.2009 20:00:00 60.00 24.01.2009 23:20:00





70.00 68.00


66.00 64.00 62.00


-10.00 22.01.2009 00:00:00 22.01.2009 03:20:00 22.01.2009 06:40:00 22.01.2009 10:00:00 22.01.2009 13:20:00 22.01.2009 16:40:00 22.01.2009 20:00:00

60.00 22.01.2009 23:20:00


Time of Day
Reactive power 132kV side voltage Nominal voltage

Time of Day
Reactive power 132kV side voltage Nominal voltage

Fig.2a HV side voltage and reactive power on 22.01.09

Fig.2b HV side voltage and reactive power on 24.01.09

The figures show that the present switching pattern does not fully utilize the installed capacitor banks with the present power factor regulation switching criteria. Comparing figures 1b with 2a and 1c with 2b, we can see that at the start of green arrow, the load phase angle becomes close to the setting value but next step of the banks is not switched on, since the phase angle is marginally above the set point. However, during this period, the voltage goes down and the reactive power consumption is high. Therefore, for both days, the 4th bank could be switched on at around 10.00 hrs. As for those two days, the situation is generally common through out and therefore the present switching criteria is not a suitable solution. V OBSERVATIONS FROM MEASUREMENTS Behaviour of the substation load is cyclic and has two distinct peak points. There is a load peak in the early morning hours and highest peak is in the night.
MW / Mvar Curve - Panaduara GSS (20th to 28th Jan 2009) Measured at 33kV side

Power factor - Panaduara GSS (20th to 28th Jan 2009) Measured at 33kV bus and 132 kV bus


Power Factor




20.01.2009 21:30:00 21.01.2009 17:30:00 22.01.2009 13:30:00 23.01.2009 09:30:00 24.01.2009 05:30:00 25.01.2009 01:30:00 25.01.2009 21:30:00 26.01.2009 17:30:00 27.01.2009 13:30:00 28.01.2009 09:30:00

Time of the day

PF at 33 bus PF at 132 bus

Fig.4a Power factor over the full measurement period

Power factor - Panaduara GSS (21st Jan 2009) Measured at 33kV bus and 132 kV bus



Power Factor




MW / Mvar




0.00 1.40 3.20 5.00 6.40 8.20 10.00 11.40 13.20 15.00 16.40 18.20 20.00 21.40 23.20

10 PF at 33 bus PF at 132 bus

20.01.2009 21:30:00 21.01.2009 17:30:00 22.01.2009 13:30:00 23.01.2009 09:30:00 24.01.2009 05:30:00 25.01.2009 01:30:00 25.01.2009 21:30:00 26.01.2009 17:30:00 27.01.2009 13:30:00 28.01.2009 09:30:00

Fig.4b Power factor in a days window

Time of the day MW Mvar

Fig.3 Daily load pattern

Day time load is considerably flat and has a drop at lunch time and at the close of office hours. The morning and night peaks are generally due to lighting loads and day time industrial and commercial load is naturally inductive. This is shown in the figure 3. Behaviour of the system power factor at the medium voltage bus and high voltage bus describes the composition of the load. During the said morning peak and night peak the PF is comparatively high.

During day time the power factor is low due to highly inductive industrial load. However it takes a somewhat flat profile showing that both real and reactive loads increase in same proportion. Figures 4a and 4b illustrate this pattern. The intention of CEB in using the capacitor banks is a key factor in the analysis. It is quite clear that CEBs intention is to give a voltage support at the medium voltage bus which drops due to increasing load. Dropping the system voltage at the load centres is a critical problem especially in locations where there is no



close by generation for reactive power compensation. Release of sub station capacities and reduction of losses are secondary expectations in CEBs point of view although they too are very important. As said earlier, voltage decreases due to large resistive loads at night and morning peaks and due to heavy industrial and commercial loads at day time. Voltage improves in mid night till early morning with decreasing loads but considerable base reactive load exists through out. Comparison of voltage at high voltage bus and power factor measured at medium voltage bus is shown in figure 5a.
Comparison of 132kV Voltage & Phase angle measured at 33 bus
20.00 15.00 10.00 5.00 0.00 77.00 79.00

that if the first come banks correct the power factor then the others will not come even if there is a possibility of compensating more reactive power or increasing the bus voltage. As explained earlier, this is clearly observed in figures 1b, 1c, 2a and 2b. The possibility of stepping to the 4th bank is still there at around 9.00 hrs on both days despite the phase angle is just above the setting value. Analysis of measured data shows a considerable uncompensated reactive power with the present switching criteria. It is further explained in figures 6a and 6b for two other days with in measurement period and indicates unsuitability of the present switching criteria.
Uncompensated Var (26.01.09)
15.00 10 9 10.00 8 7


Phase angle

Voltage (kV)

-5.00 -10.00 -15.00 -20.00 -25.00 69.00 -30.00 -35.00 -40.00 -45.00 21.01.2009 21.01.2009 22.01.2009 23.01.2009 24.01.2009 25.01.2009 26.01.2009 26.01.2009 27.01.2009 00:00:00 20:00:00 16:00:00 12:00:00 08:00:00 04:00:00 00:00:00 20:00:00 16:00:00
Time of Day PF at 33 bus (no Caps) 132kV Voltage without caps Nominal132 Voltage




6 5 0.00 4 -5.00 3 2 -10.00 26.01.2009 00:00:00 1 26.01.2009 23:20:00




26.01.2009 03:20:00

26.01.2009 06:40:00

26.01.2009 10:00:00

26.01.2009 13:20:00

26.01.2009 16:40:00

26.01.2009 20:00:00

Time of Day
Unserved Var No of Cap Banks No of Max Cap Banks

Fig.5a Comparison of HV bus voltage and PF

Comparison of 132kV Voltage & Phase angle measured at 33 bus on 21st Jan 2009
20.00 15.00 10.00 5.00 0.00 Phase angle -5.00 -10.00 -15.00 -20.00 -25.00 69.00 -30.00 -35.00 -40.00 -45.00 0.00 2.00 4.00 6.00 8.00 10.00 12.00
Time of Day

Fig.6a Uncompensated reactive power under present switching criteria 24.01.09

Uncompensated Var (27.01.09)

10 9


8 7

75.00 Voltage (kV)




6 5 0.00 4 -5.00 3 2 -10.00 27.01.2009 00:00:00 1 27.01.2009 23:20:00



65.00 14.00 16.00 18.00 20.00 22.00

27.01.2009 03:20:00

27.01.2009 06:40:00

27.01.2009 10:00:00

27.01.2009 13:20:00

27.01.2009 16:40:00

27.01.2009 20:00:00

Time of Day
Nominal132 Voltage

PF at 33 bus (no Caps)

132kV Voltage without caps

Unserved Var

No of Cap Banks

No of Max Cap Banks

Fig.5b Comparison of HV bus voltage and PF in a days window

Fig.6b Uncompensated reactive power under present switching criteria 27.01.09

Power factor goes high during night and morning peaks causing tendency to switch off the capacitor banks but bus voltage goes down. Due to this, we may deliberately ignore a possibility of improving bus voltage due to gradual disconnecting of capacitor banks during night peak or delay in picking up the banks in the morning. In other words, either it is possible to keep some capacitor banks for extended time or some banks can be connected bit earlier. Therefore, voltage and power factor behaves contradictorily. In the other case, again the voltage decreases during day time with high inductive loads and power factor becomes low. It comes to an approximate flat profile later. The figures 5a and 5b show this clearly. This means that the increase of real and reactive power is in same proportion. The point that has to be considered is

On Load Tap Changer (OLTC) and the AVR in such scale utility substations is also available to adjust the LV bus voltage. Use of capacitor banks in voltage support is the most economical since it reduces apparent power drawn from the system hence reducing losses. Tap changer improves the voltage by changing the tap position and reduces only small amount of reactive power and overall effect on reactive power due to increased voltage is an increase. Therefore, full utilization of already installed capacitor banks is better in adjusting the bus voltage than adjusting the taps. Figure 7 which contain the pattern of the tap position with no capacitor banks shows that the system operates at higher tap positions during day time with decreasing voltages.

No of Cap Banks

No of Cap Banks

132kV voltage pattern, tap position & no of cap banks (25th to 27th)
85 83 16 15 14


No of Cap Banks / Tap position

132kV bus voltage (ph_E) kV

81 79 77 75 73 71 69 67

13 12 11 10 9 8 7 6 5 4 3 2 1

65 25.01.2009 25.01.2009 25.01.2009 26.01.2009 26.01.2009 26.01.2009 27.01.2009 27.01.2009 27.01.2009 00:00:00 08:20:00 16:40:00 01:00:00 09:20:00 17:40:00 02:00:00 10:20:00 18:40:00

Time of Day
132kV bus voltage Max. Continuous 132 voltage No of Max Cap Banks Nominal 132kV voltage No of Cap Banks Tap position

Fig.7 Behaviour of tap position with no capacitor banks

The voltage rise obtained by raising one tap position up, is 1.5 % of the voltage at the point of measuring. This is as per the specifications of the OLTC. At 33kV voltage this rise is about 0.495 kV. The approximated percentage voltage rise given by switching one 5Mvar capacitor bank is given as (kvar / kva) * Xt Where kvar = addition of reactive load, kva = transformer rating and Xt = transformer reactance in % [3]. When two transformers are in parallel, this value becomes 0.79% and the voltage rise is 0.260kV at 33kV. As these figures suggests, the effect of rise in one tap step is same as adding two 5Mvar capacitor banks when two transformers are paralleled or one 5Mvar banks when one transformer is connected. Considering the above, the system could be operated at least with two taps below if the capacitor banks are connected.

Following effects due to the switching of capacitor banks to the system was studied by modelling the network with PSCAD which is widely used simulation software for network simulations [4]. Maximum voltage rise due addition of capacitor banks at the bus bar The capability of transformer OLTC and AVR to handle those voltage variations by changing tap position, when necessary. The capability of OLTC to handle the current through it without exceeding its current switching capacity during back feeding reactive power into the system The effect of resonance when adding more capacitor banks under various load conditions and system harmonic levels Effects on voltage distortion caused by load harmonics at 33kV bus, when adding more capacitor banks Cost analysis considering the reduction of losses due to power factor improvement, release of system component capacities etc. and many others. VII VOLTAGE RISE The substation model was adjusted to have same measurement condition as measured without capacitor banks. The changes in parameters when simulating the switching of the capacitor banks as per present criteria, for maximum var compensation and for maximum capacitor banks were recorded next. The maximum voltage rise which may occur at maximum source voltage and minimum load with all capacitors was also simulated and recorded.
Multiple Run Output File All Caps Load -17.2 MW 9.6Mvar

Run #

Tap HV Position Voltage 13 12 11 10 9 74.78 74.79 74.80 74.82 74.83

LV Voltage 33.50 32.98 32.47 31.98 31.50

Ph Ang_LV -23.51 -23.51 -23.51 -23.51 -23.51

LV_MW LV_MVar 47.12 45.66 44.27 42.95 41.68 20.50 19.87 19.26 18.68 18.13

TF_HV_ Current (pk) .17 .16-------(B) .16 .15 ------(A) .15

1 2 3 4 5

Multiple Run Output File 3 cap banks

Run # Tap HV Position Voltage 13 12 11 10 9 75.17 75.17 75.17 75.18 75.18 LV Voltage 34.45 33.91 33.38 32.87 32.38 Ph Ang_LV -6.17 -6.17 -6.17 -6.17 -6.17 LV_MW LV_MVar 49.83 48.28 46.80 45.38 44.03 5.38 5.22 5.06 4.90 4.76 TF_HV_ Current (pk) .16 .15 .15 .14-----(C) .14

Run # Tap HV Volt.(kV-phE) LV Volt(kV) HV_Current (pk-A) 1 8 83.08 36.86 .105 2 7 83.07 36.31 .102 3 6 83.06 35.78 .099 4 5 83.05 35.26 .096 5 4 83.04 34.76 .094 6 3 83.03 34.28 .091 7 2 83.03 33.80 .089 8 1 83.02 33.34 .086 No Caps Multiple Run Load -17.2 MW 9.6Mvar Output File Run # Tap HV Volt.(kV-phE) LV Volt(kV) HV_Current (pk-A) 1 8 82.55 35.49 .102 2 7 82.56 34.97 .099 3 6 82.56 34.46 .097 4 5 82.57 33.97 .094 5 4 82.56 33.49 .091 6 3 82.58 33.03 .089 7 2 82.59 32.58 .086 8 1 82.59 32.14 .084

1 2 3 4 5

Table 1 An abstract of the results from network

The table1, an abstract of the results from network simulation shows that if start from point (A) with LV bus voltage 31.98kV, the tap position changes from 10 to 12 until it adjust the bus voltage to 32.98kV (Point B). Switching of 3 capacitor banks of 5Mvar can keep the bus voltage 32.87kV while retaining at the same tap (point C) but reduce the HV side current in one transformer by about 14A.

Table 2 Results from simulations for worst case analysis

Under such a worst case, LV bus voltage with no capacitor banks is 33.494 kV at tap position to 4. When all banks are connected at this stage, AVR & tap changer is capable to maintain the bus voltage at

33.344kV at tap changes from 4 to 1. Practically this is not a desired condition but such a worst case will not be allowed by the system operator. Table 2 shows the results from the simulations. With present configuration, the maximum effective reactive power injection through a transformer when either transformers in parallel, or transformers are independent, is 10Mvar (since each transformer is connected with two banks). Addition of 20 Mvar gives a rise of about 1.04kV at 33kV bus voltage. Rise of 1 tap position changes the voltage by 0.015 pu and this is about 0.495kV at 33kV and therefore the effect of rise in voltage over the nominal value due to addition of maximum capacitor banks can be handled with two tap positions. The simulation results were studied on voltage at the HV bus with the no banks, maximum banks, when the banks are switched to give optimum var compensation and when banks are switched with the present scheme. As per the above results, switching the maximum capacitor banks under any real time condition is obviously possible as far as the voltage rise at bus bar is concerned. Real time data measurement of 132 kV voltages with all 4 capacitor banks in ON condition was done and compared with the simulation results.
Variation of tap position - 21st & 22nd January 2009
18 16 14

Tap - real measurements 18th Feb to 21st Feb 2009

16 15 14 13 12

Tap position

11 10 9 8 7 6 5

4 18.02.2009 18.02.2009 19.02.2009 19.02.2009 19.02.2009 20.02.2009 20.02.2009 20.02.2009 21.02.2009 21.02.2009 12:00:00 19:30:00 03:00:00 10:30:00 18:00:00 01:30:00 09:00:00 16:30:00 00:00:00 07:30:00

Time of day
Tap - real measurements

Fig.9b Tap position variation with all cap banks (actual measurements)

model with slight modifications to add harmonic current source, distortion level measurement and frequency scan module blocks. The simulations were done for different load combinations and for different substation configurations as well. The simulation results are shown in figures 10a to 10e. Followings are the observations

Number of taps

12 10 8 6 4 2 0 21.01.2009 00:00:00 21.01.2009 09:00:00 21.01.2009 18:00:00 22.01.2009 03:00:00 22.01.2009 12:00:00 22.01.2009 21:00:00 24.01.2009 06:00:00 24.01.2009 15:00:00

Time of day
Tap - actual measurement with no cap banks Tap - Simulated results with maximum cap banks Tap - Simulated results - under present scheme

Fig.9a Tap position variation to give constant LV bus voltage (simulated results)

VIII VOLTAGE CONTROL BY OLTC & AVR Behaviour of tap changer position in response to rise of voltages beyond the nominal values due to capacitor banks were simulated and Figure 9a shows the comparison. The figure 9b shows the tap position recorded in real time monitoring with all capacitor banks connected. Tap position behaves within the acceptable range.The current through the OLTC with maximum capacitor banks does not exceed rated current under any condition and therefore, it is not a decisive factor. IX RESONNANCE AND VOLTAGE DISTORTION The effects of resonance due to switching on the capacitor banks were studied using the same sub station

With only filter banks, two resonance points are observed, one with highest impedance and one with lowest. When filter banks are mixed with normal banks, one additional high impedance point is observed where the value is higher than the first one. Minimum resonance point (close to 250Hz /tuned to 5th harmonic) and the first high impedance point(190Hz-inter harmonic or close to 4th harmonic) is same under all conditions. Therefore impacts are negligible. The second high impedance point varies with the configuration. Higher the load lower the magnitude of the impedances For only normal banks, the minimum resonance point not seen (detuned banks) The effect of voltage harmonic distortion mainly due to second highest impedance point is the fact to be considered

Resonance Characteristics (Single TF/1 filter bank - for different loads)

130 120 110 100 90

Impedenace (ohm)

80 70 60 50 40 30 20 10 0 50 150 250 350 450 550 650 750 850 950 1050 1150 1250

Frequency (Hz)
Minimum tf load Max day load share for 1 tf Max night peak share for 1 tf Max tf load

Fig.10a Frequency scan single t/f & one filter bank

Resonance Characteristics (Single TF/2 CAP banks - for different loads)

130 120 110 100 90 130 120 110 100 90

Resonance Characteristics (2 TF/1 filter bank - for different loads)

Impedenace (ohm)

70 60 50 40 30 20 10 0 50 150 250 350 450 550 650 750 850 950 1050 1150 1250

Impedenace (ohm)


80 70 60 50 40 30 20 10 0 50 150 250 350 450 550 650 750 850 950 1050 1150 1250

Frequency (Hz)
Minimum tf load Max day load share for 1 tf Max night peak share for 1 tf Max tf load Minimum tf load

Frequency (Hz)
Max day loadf Max night peak Max tf load

Fig.10b Frequency scan single t/f & 1filter & 1 normal bank
Resonance Characteristics (2 TF/2 filter bank - for different loads)
130 120 110 100 90

Fig.10c Frequency scan 2 t/f & 1filter bank

Resonance Characteristics (2 TF/3 filter bank - for different loads)

130 120 110 100 90

Impedenace (ohm)

Impedenace (ohm)

80 70 60 50 40 30 20 10 0 50 150 250 350 450 550 650 750 850 950 1050 1150 1250

80 70 60 50 40 30 20 10 0 50 150 250 350 450 550 650 750 850 950 1050 1150 1250

Frequency (Hz)
Minimum tf load Max day loadf Max night peak Max tf load Minimum tf load

Frequency (Hz)
Max day loadf Max night peak Max tf load

Fig.10d Frequency scan 2 t/f & 2filter banks

Fig.10e Frequency scan 2 t/f & 2filter 1normal banks

Resonance Characteristics (2 TF/4 filter bank - for different loads)

130 120 110 100 90

switched capacitor rating at time slot tn and tn is taken as 10 min interval.

Impedenace (ohm)

80 70 60 50 40 30 20 10 0 50 150 250 350 450 550 650 750 850 950 1050 1150 1250

Frequency (Hz)
Minimum tf load Max day loadf Max night peak Max tf load

Fig.10f Frequency scan 2 t/f & 2filter 2normal banks

Under all configurations, the voltage distortion at 33kV bus level is below 7.2% level (6.5% is the planning value as per IEC 61000-3-6 and 8% tolerance value as per EN 50160) [5] . High distortion is resulted when all capacitor banks are connected. Therefore the impact to allowable voltage distortion levels by maximum use of capacitor banks is under acceptable levels. X NEW SWITCHING CRITERIA Utilization of the capacitor banks under present criteria, calculated on daily average and with reference to maximum utilization is about 75 % for power factor / var control switching criteria and 70.03 % for pure power factor control. (Utilization = (Mvar1*t1 + Mvar2*t2 + --------+

Although these values are comparatively high, it does not indicate the optimality of the use. The present scheme contains unnecessary utilization at certain time periods and periods of partial utilization of capacitor banks even the opportunity is there to fully use them. In real situation, sometimes the network operators manually switch off the banks to avoid high leading power factor and bus voltage rises or switch on the banks which are already in off position due to improved power factor. Therefore, the high utilization factor is not the mere deciding factor for the optimal usage. Loss minimization, voltage support, releasing capacity constraints etc., are the factors to be considered. As discussed in the previous chapters, the possibility of connecting the maximum number of capacitor banks into the LV bus under any system conditions is obvious. The analysis shows that the harmful effects can be maintained with marginally affecting the regulations and not violating the technical limitations. Therefore, following conclusions can be made. For the selected substation, it is possible to connect all four capacitor banks under any system condition. Therefore, any other combinational arrangement, to suit the local requirements is also possible.

Mvarn*tn) / Maximum Mvar * 24 where Mvarn =

The first point can be considered in the system point of view when capacitor banks in substations are kept unused while transmission system needs reactive power for other locations. This happens mostly when power across the companys transmission system does not coincide with load conditions in locations where the capacitor banks are fixed. This situation can be mostly experienced in substations which are heavily interconnected. Under those conditions, keeping a definite economical reactive energy source underutilized or unutilized depending on local requirements, while generation or some other means producing and transmitting them in the system, is not justifiable. CEB has to take advantages of ON Demand Control to use the already installed capacitor banks in this manner. If the transmission system needs var as explained and if a centralized network control center monitor the load flow in its transmission system, then switching of unused capacitor banks at such a time can be used to inject reactive power. This needs a comprehensive load flow study, fully pledged SCADA system and sometimes remote station control facility etc., to implement the above schemes. Interestingly, those are already in touch with the CEB transmission network. Therefore, if necessary CEB can use its maximum installed capacitor banks without any difficulty. Second option is to meet local requirements in each substation. As explained voltage or var control or a combination of both seems to be better compared to power factor. It is always an indirect measure of reactive power. And also it has no concerns over the effects beyond the substation, such as voltage rises due to predominant line capacitance during very light loaded conditions. In such cases, considerable lagging reactive load at load centres is beneficial. If the substation reactive power requirement is fully compensated during these periods, the voltage rise at receiving ends will be a problem. In such cases capacitor bank switching based on voltage control may have more benefits. XI SWITCHING CRITERIA BASED ON REACTIVE POWER Switching based on reactive power requirements is a more flexible and natural means of capacitor control concepts. It adds a fixed amount of lagging reactive power into the system regardless of most other conditions. Losses and the capacity release are directly proportional to the apparent power. Injecting the reactive power reduces apparent power and consequently losses. In var control based switching, due consideration has to be given to avoid hunting or PUMPING of the banks. Hysteresis or restraint control is suggested to avoid such a problem. In general, switching ON based on about 2/3 of a step and switching off based on slightly beyond 1/3 of the step in leading direction. To avoid responding to sudden reactive power changes, restraint control or integration of inputs over certain time period can be

used. These are available in most of the capacitor bank controllers. Considering the above basis, parameters for reactive power control switching for master slave control mode was suggested as follows. The minimum step setting depends on controller, based on master slave mode or independent mode. Considering the results obtained by simulations, following points can be considered in a reactive power control based switching criteria for CEB. When transformers are paralleled, one controller feels only a half of the capacity of a switched bank. Step size of a bank is 5Mvar. Switching ON when lagging reactive power exceeds 2.5 *2/3 =1.6Mvar (lag) Switching OFF when leading reactive power exceeds (2.5 *1/3) *1.4 1.2Mvar(lead)

Switching points were selected from simulation results with approximated AVR control and shown in figure 11 The switching points based on lowest reactive power drawn from system and power factor close to unity (optimum compared to losses) was also show in the diagram.
Comparison of var control Vs present scheme 21st &22nd January 2009



No of banks




0 21.01.2009 00:00:00

5 21.01.2009 06:00:00 21.01.2009 12:00:00 21.01.2009 18:00:00 22.01.2009 00:00:00 22.01.2009 06:00:00 22.01.2009 12:00:00 22.01.2009 18:00:00

Time of day
Present scheme Optimum based minimum loss proposed Var control Mvar with no capacitors

Fig.11 Switching pattern with var control for 21st & 22nd Jan 2009

The figure shows that the proposed switching policy based on reactive power control goes neck to neck with the loss optimized switching pattern than the present switching criteria (Blue and red curves). No of switching operations per day is on the safe side. A typical capacitor bank switch can operate 6times per day considering 50,000 no of operations and 20 years life time. The table 4 shows the results. Number of switching Bank 2 Bank 3 Bank 4 0 2 2 0 2 2 0 1 2

Date 21.01.09 22.01.09 24.01.09

Bank 1 0 0 0

Table 4. No of switching operations on selected days

The utilization factor is 80% and better than the present system and also closer to the theoretical loss optimized


pattern. Increase or decrease of energy loss compared to present switching criteria was calculated based on the point that the losses are directly proportional to I2. Three days were considered and a decrease of 1.8%, 4.9% and 5.04% was observed with an average reduction of 3.94% (Considering only the transformer losses). The capacity release of the substation was calculated as below. The same capacity will be released from the generation as well [3].

The difficulty in voltage control based switching is that it has to coordinate with the voltage regulator of the power transformers since the latter also tries to control the voltage. Coordination is required in such a case and following factors have to be considered. During switching on for decreasing bus bar voltages, the capacitors shall come first if the reactive power load is more than a specified percentage of the minimum step of a bank otherwise the tap changer can increase the voltage. The purpose of this is to minimize the losses by maintaining the power factor close as much as possible to unity. During switching off for increasing terminal voltages, reactive power at the time of decision must be considered and the algorithm shall decide the most economical step, whether to reduce the tap or to switch off a capacitor bank. The purpose of above two conditions is to maintain the bus voltage while reducing the losses to minimum. If the only requirement is to control the voltage, then proper dead band selection for two controllers also can serve the purpose. Differentiate the integration time, the time period over which the measurement is averaged, also can be used with hysteresis control to make the control philosophy more simple.

Where KVAs - release of substation KVAs KVAR - Capacity of substation

- Capacity

of next step of the banks

Cos and Sin - Cos and sine of power factor before adding next step

For the selected substation, addition of 5Mvar for 2 * 31.5MVA transformers at the conditions as at 8.30hrs on 24th January 2009, the capacity release KVAs was calculated as,
MVAs = [ {1-(5 *Cos7.13/63)2}+Sin7.13 * (5 / 63) - 1]63 = 0.425 MVA

With the simulation results as in table 5 for same time slot on the selected day, it can be calculated as; MVAs = (33.012 +4.072) - (32.62742+.896852) = 0.620 MVA But this is with a tap position change as well. Therefore the simulation results can be justified. Considering the simulation results, total average energy released by switching from present scheme to proposed var control scheme is 15.64 MWh per day (calculated based on 30min sample time). The scheme maintains the tap close to nominal tap while keeping the 33 kV voltages also within the range.

A voltage selection scheme based on a hysteresis control as in the figure12 is evaluated for comparison with the present and proposed var control schemes.
AVR tap raise Cap bank ON Nominal Voltage Cap bank OFF AVR tap lower

32.50 kV





Fig.12 Switching points for cap bank controller and AVR

Date & Time MW 24.01 . 09 08:30 :00 Mvar 33 Volt Under Present criteria

132 Volt

No of Bank s

Ph angl e

Utiliz ation



The approximated switching points of capacitor banks based on above voltage control scheme, was selected using the simulation results for three selected days and figure 13 shows the comparison.









Comparioson present scheme Vs voltage control 21st & 22nd Jan 2009
5 80 79

Proposed var control scheme No of Ban ks 4


No of banks


Mva r

33 Volt

132 Volt

Ph angl e

Utili zatio n 10.0 0



76 75

24.01. 09 08:30:00



32.7 8

75.0 9


73.7 6

74 73

1 72 71 0 21.01.2009 00:00:00 70 21.01.2009 06:00:00 21.01.2009 12:00:00 21.01.2009 22.01.2009 22.01.2009 22.01.2009 22.01.2009 18:00:00 00:00:00 06:00:00 12:00:00 18:00:00 Proposed voltage control 132kV Voltage with no caps

Table 5. Simulation results 8.30hrs 24th Jan 2009

Time Present scheme Optimum switching based on minimum loss


Fig.13 voltage control switching compared with present & loss optimized criteria 21st & 22nd January

HV voltage (Ph-E) kV

Date & Time


The criterion does not maximize the utilization. Gradual switching off of banks around 17.00 to 18.00 hrs also observed due to reduction of loads after office hours. There is a voltage rise during this period and the load rises after that due to lighting. Voltage control scheme does not coincide with the curve with minimum loss, like in reactive power control. As the data shows, following conclusions can be made. Maximum switching operations per 3rd and 4 th banks is about 4 so that the switching does not cause any unnecessary impact. Utilization when voltage control is used seems to be low compared to loss optimized switching pattern. It is 55%, 58% for the two selected days. Due to reduced utilization, energy losses and substation capacity release also not be economical. However it matches with the voltage properly. Therefore, if the need is to give voltage support, then this kind of switching policy is very satisfactory. XIII A PRACTICAL APPROACH In serving both voltage support and loss reduction purposes, combining the two concepts, voltage control and var control will give more benefits. However, the AVR is to be inoperative while the capacitor controller is in action but redundancy to be maintained by AVR when the capacitor controller is faulty, after banks are fully utilized, during a tripping of capacitor banks or the voltages are beyond certain extreme ends. It is learnt from the analysis that if reactive power control is used aiming to manually or automatically switch off the banks at specified time intervals (from around 22.30hrs to 7.00hrs) to avoid high bus voltages during low load conditions, the above combinational effect is achievable. . The comparison is shown in figures 14.
Voltage control Vs var control with Manual OFF at night - 21st &22nd Jan

disadvantage is the functionality of such a manual auto mixed control. However, if both voltage and reactive power combined controller having multiple variable or Boolean switching controllers can be used to switch the banks considering voltage and var, it could be a better solution. XIV ANALYSIS AND RESULTS i. Using capacitor banks at 33kV sub distribution level to compensate reactive power requirement and therein, to maintain voltage stability at same level is economical and effective in the CEB system. Occasions where the capacitor banks are switched ON and OFF manually by over-riding the auto controller was frequently observed. This says that the switching criteria are not fully fit to the requirements in CEB system. The observations also show that present switching criteria at the selected substation neither maximize nor optimize the utilization. The study proves the technical feasibility of maximum capacitor bank connections to the point at which they are fixed without violating voltage rise due to reactive power injection, effects to voltage distortion and resonance due to harmonics with additional capacitor banks, switching capabilities of the on-load tap changer and the capabilities of AVR to handle voltage variations due to reactive power injection. The maximum voltage rise under different capacitor bank combinations (with effective Tap control) for 21st, 22nd & 24th are 77.57kV, 77.8kV & 77.17kV respectively. The maximum percentage rise for high voltage side is .33% and that for low voltage side is 0.95%. Effects due to resonance for the selected substation are negligible. Voltage distortion levels remains marginally below 8% hence acceptable Local voltage variation due to added reactive power can be handled by the AVR and tap changer controls so that any combination of banks is feasible to connect. The current through the tap changer does not exceed its switching capacity. Reactive power controlled based switching is a very much economical method of capacitor bank controlling as far as the utilization, loss reduction and capacity release is concerned. Only problem a utility may face is that, some times especially in light load conditions with long transmission lines, there may be a necessity to have some reactive power to reduce the Ferranti effects. In such cases, minimizing reactive power consumption is not desired. In practice, for a utility like CEB where most of the generation is concentrated to certain areas, maintaining bus voltages may be difficult and be important than reducing losses using capacitor banks. In such a, voltage control based capacitor switching will be a good solution.




v. vi. vii.

viii. ix.

No of banks

0 21.01.2009 00:00:00 21.01.2009 06:00:00 21.01.2009 12:00:00 21.01.2009 18:00:00 22.01.2009 00:00:00 Time of day proposed Var control from 7.00 to 22.30hrs Prposed Voltage control scheme 22.01.2009 06:00:00 22.01.2009 12:00:00 22.01.2009 18:00:00

Fig.14 Voltage control Vs var control with manual off at night. (21 and 22 Jan 09)


As we see from these figures, if reactive power controlled switching can be used as above, it is similar to the voltage control scheme but less complex. The

XIV CONCLUSION Considering all these factors discussed so far, followings are the conclusions from this research study. i. Present capacitor bank switching philosophy based on power factor regulation does not give maximum benefits to the CEB transmission network. This scheme neither maximizes nor optimises the utilization.

XVI ACKNOWLADGEMENT Authors wish to thank to Dr H.M. Wijekoon (CEplanning/Distribution Region 3, CEB for his valuable comments and Mr. L.A.S.Fernando and his staff of Operations and Maintenance branch of CEB for facilitation data collection and measurements. XVII REFERENCES [1] Kusum Shanthi K.P Benchmark the Sri Lankan power system by power quality monitoring & analysis Master Thesis, University of Moratuwa , 2005. Chapter 7.2 pp 51. [2] User Manual for POCOS control reactive power controller and harmonic analyzer [3] Technical paper in web nciurro@pssamerica.com Economics when applying shunt capacitors pp6-7 [4] User Manual for PSCAD [5] Kusum Shanthi K.P, Rangith Pererra, Sarath Pererra Benchmarking the Sri Lankan power system by a power quality monitoring program 2005. Section III D pp 4.

ii. Considering the installed capacities and step sizes in each substation, it is technically possible to utilize the full installed capacities in all substations without violating the technical standards. iii. Therefore, it is technically feasible to back feed the excess capacitor bank capacity for reactive power compensation in the transmission network. iv. Use of a switching policy based on reactive power control or voltage control is more useful as far as the CEB system is considered. Reactive power based switching which is simple, is useful for loss minimization and voltage based control is useful when voltage stability is concerned. v. Considering the factors discussed in 7.1 viii and ix, for network like CEB, it is useful to consider the controllers with multi-parameter or Boolean switching options. Reactive power and voltage can be the parameters to be considered in the switching decisions.