A

dpros©

K

1

ukakLr

rfcpLfnupe

0d

Electric Power Systems Research 77 (2007) 1314–1322

Voltage control with on-load tap changers in medium voltagefeeders in presence of distributed generation

Ferry A. Viawan a,∗, Ambra Sannino b, Jaap Daalder a

a Division of Electric Power Engineering, Department of Energy and Environment, Chalmers University of Technology, Gothenburg, Swedenb ABB, Corporate Research, Vasteras, Sweden

Received 15 July 2005; received in revised form 19 June 2006; accepted 28 September 2006Available online 13 November 2006

bstract

This paper discusses voltage regulation on medium-voltage feeders with distributed generation (DG) using on-load tap changer (LTC) and linerop compensation (LDC). The analysis shows that LTC is robust against DG, whereas DG can affect the effectiveness of the voltage regulationrovided by LDC. However, with proper coordination between DG and LDC, it is possible to ensure voltage regulation without unnecessarily

estricting the integration of DG. It is shown that, while lowering the LTC setting can increase the DG integration limit, even higher increase can bebtained by activating the LDC feature, which is present in most LTCs, but often not used. LDC regulation is also compared with other alternativesuch as using a DG unit with voltage control capability and installing a line voltage regulator.

2006 Elsevier B.V. All rights reserved.

on; Vo

artcw

wcHccitsait

eywords: Distributed generation; On-load tap changer; Line drop compensati

. Introduction

Voltage control of MV distribution feeders is achieved bysing on-load tap changers (LTCs) and capacitor banks. The LTCeeps the voltage at the substation busbar constant. Normally,LTC is also provided with line drop compensation (LDC) to

eep the voltage constant at a remote load center (LC) [1–3].TCs are widespread in distribution networks and are likely toemain in service for many years to come [3].

Distributed generation (DG) causes reduction or possiblyeversal of real power flows, which may impart significant poweractor (PF) changes detected by a HV/MV transformer. Thesehanges may affect the effectiveness of the voltage regulationrovided by LDC. Therefore, DG should be coordinated withDC to ensure that the distribution network will not loose the

unction of proper voltage regulation. Proper coordination is alsoecessary in order to ensure that the DG integration will not be

nnecessarily limited. DG integration limit is defined in thisaper as the maximum DG power that can be generated beforeither maximum allowed voltage variation, conductor thermal

∗ Corresponding author. Tel.: +46 31 7721661; fax: +46 31 7721633.E-mail address: ferry.viawan@chalmers.se (F.A. Viawan).

pL[aBod

378-7796/$ – see front matter © 2006 Elsevier B.V. All rights reserved.oi:10.1016/j.epsr.2006.09.021

ltage regulation; Voltage control

mpacity or rating of substation transformer are reached. Theeduction or reversal of active power flow due to DG will increasehe voltage along the feeder. DG coordination with switchedapacitor banks is also required to ensure that the capacitorsill not cause overvoltages [4].Maximum DG integration limit on multiple feeders with LDC

ithout changing LTC tap position is presented in [5]. Thehange in voltage profile can be minimized with this method.owever, the integration limit obtained by preventing LTC from

hanging its position will be very marginal. More DG poweran be introduced without violating permissible voltage lim-ts when the LTC tap position is allowed to change. In [6,7],he installation of a line voltage regulator (VR) is presented toolve unacceptable voltage variations. With this method, a largemount of DG can be connected to the feeder. However, VRnstallation means additional investment costs. The reduction ofhe LTC setting in a HV/MV substation to increase wind powerroduction is presented in [8]. Voltage regulation with multipleDCs on multiple feeders with different loading is proposed in

9]. This method is based on system wide coordination of volt-

ges using remote control, communication and optimization.esides the need of a communication link and the modificationf the existing LDC control system, the voltage control of theistribution network is far more complicated when using this

F.A. Viawan et al. / Electric Power System

mt

awmffltoc

2L

fbt

aa3lmiwa

icecc

2

ii

U

waa

ttavca

uasiab

iaips

aoeoa

L

i

2

Ltance RL and reactance XL, and performs voltage corrections toget the voltage at the LC ULC constant within the range.

ULB ≤ ULC ≤ UUB (2)

Fig. 1. Basic LTC arrangement.

ethod [8], which may be considered undesirable by many dis-ribution network operators nowadays.

This paper examines LTC/LDC regulation of a MV feeder,nd how it is affected by DG. The voltage regulation constraintsith LTC and LDC are reformulated for the case of one orore DG units connected along a feeder in a single- or multi-

eeder system. Simulation results are presented for differenteeder models, which compare the maximum DG integrationimit obtained in presence of DG with LTC and LDC regula-ion. These regulation methods are also compared with the usef a VR and the installation of a DG unit with voltage controlapability.

. Voltage regulation in conventional MV feeder withTC/LDC

The objective function of voltage control in conventionaleeders, i.e., feeders with only loads and no DG connected, cane either to minimize feeder losses or to operate the feeder closero the nominal voltage.

One means of voltage control is the LTC regulator, which is anutotransformer with automatically adjusted taps. Commonly,LTC provides a regulation range from −10% to 10% using

2 steps [1,2]. The LTC will keep the voltage constant at theocal busbar, or at a LC, if provided with LDC. In practice,

any LTCs are operated with the LDC feature disabled, whichs much simpler. In this paper, voltage regulation using both LTCith LDC disabled (called LTC regulation) and LTC with LDC

ctivated (called LDC regulation) are analyzed.Another means of voltage control are capacitor banks, which

nject reactive power into the feeder, thus decreasing the lineurrent and losses and increasing the voltage. Capacitors canither be fixed or switched by capacitor control. Many capacitorontrols can be used in conventional MV feeders, such as timeontrol, voltage control, current control and VAR control [1].

.1. Voltage control with LTC regulation

The basic arrangement of voltage control with LTC regulation

s shown in Fig. 1. The LTC will keep the local busbar voltage,.e., the sending-end voltage U0, constant within the range:

LB ≤ U0 ≤ UUB (1)

s Research 77 (2007) 1314–1322 1315

here ULB = Uset − 0.5 bandwidth is the lower boundary volt-ge, UUB = Uset + 0.5 bandwidth is the upper boundary voltagend Uset is the setpoint voltage.

The voltage on a conventional MV feeder decreases towardshe end. The LTC shall then be set to ensure that the voltage athe feeder end is higher than the minimum allowed voltage Umin,nd the sending-end voltage is lower than the maximum allowedoltage Umax. Since the LTC keeps the sending-end voltage U0onstant, it is possible to operate the feeder with minimum lossest any load condition.

No load or feeder parameters appear in Eq. (1), so LTC reg-lation will not be affected by changes in the PF or reversal ofctive/reactive power. With the sending-end voltage kept con-tant, multiple-feeders controlled by the same LTC can be treatedndividually, as the load on one feeder will not affect the volt-ge profile on adjacent feeders, except in a small range of LTCandwidth.

When the feeder is too long, sometimes it is necessary tonstall a VR. This is an autotransformer with automaticallydjusted taps, which is also provided with a LDC function. Here,t should be ensured that the voltage is higher than Umin at therimary side of the VR and lower than Umax at the secondaryide.

Capacitor banks, when present, may overcompensate the linend increase the losses when feeder load is low, or even causevervoltage. The capacitor then needs to be switched off throughither voltage or current or VAR control. When the load cyclen the feeder is predictable throughout a day, time control is alsoppropriate.

Finally, the voltage constraints for a conventional feeder withTC regulation can be simply formulated as:

i. U0,i ≤ Umax, ∀ secondary side of LTC transformer or VRs.i. Uend,i

∣∣max load ≥ Umin, ∀ nodes at feeder-ends or primary

side of VRs.

.2. Voltage control with LDC regulation

Voltage control with LDC regulation is shown in Fig. 2. TheDC estimates the line voltage drop based on line current I, resis-

Fig. 2. LTC with line drop compensation.

1 Systems Research 77 (2007) 1314–1322

Pf

R

X

wpr

i

wei

Epwfts

U

wo

Fa

Fa

t

U

It

U

Ti

L

U

w

U

w

316 F.A. Viawan et al. / Electric Power

roperly adjusting RL and XL to the turns ratios of current trans-ormer (CT) and potential transformer (PT) yields.

set = NCT

NPTRL (3)

set = NCT

NPTXL (4)

here Rset and Xset are LDC settings for resistive/reactive com-ensation, NCT the turns ratio of the CT and NPT is the turnsatio of the PT.

The voltage at the LC during minimum and maximum loads approximated as:

ULC = U0,max − Imax(RL cos φ + XL sin φ)

ULC = U0,min − Imin(RL cos φ + XL sin φ) (5)

here U0,max and U0,min are maximum and minimum sending-nd voltage, respectively, Imax and Imin are maximum and min-mum line current cos φ is PF at LTC location.

As LDC regulation employs load and feeder parameters, seeq. (5), any changes in the PF or direction of active/reactiveower will affect the performance of the regulator. One case ishen the X/R ratio of the setting is poorly adjusted.Consider a

eeder with a load at the LC regulated by LDC in Fig. 3(a). LDCries to keep the voltage at the load constant by adjusting theending-end voltage as:

0 = Uset + I(Rset,HV cos φ + Xset,HV sin φ) (6)

here Rset,HV and Xset,HV are Rset and Xset read on primary sidef the CT and PT.

ig. 3. (a) Feeder with one load at LC regulated by LDC. (b) Voltage at LC asfunction of X/R setting at different load power and PF; Uset = 0.98 pu.

dd

U

wtcoa

csarwa

nasaf

c

ig. 4. Feeder with one load at LC and a capacitor located between the LDCnd LC.

With this sending-end voltage, the actual voltage at LC ishen:

LC = U0 − I(RL cos φ + XL sin φ) (7)

f Rset is properly adjusted, i.e., Rset = RL, the voltage error athe LC is:

LC − Uset = IRL sin φ

(Xset

Rset− XL

RL

)(8)

he error increases with decreasing PF and increasing load, fornstance as shown in Fig. 3(b).

If the same feeder has a capacitor located between LDC andC, see Fig. 4, LDC will adjust the sending-end voltage as:

0 = Uset + I2(RL cos φ2 + XL sin φ2) (9)

hereas the voltage at the LC with the capacitor on is:

LC = U0 − I2y(RL cos φ2 + XL sin φ2)

−I1(1 − y) (RL cos φ1 + XL sin φ1) (10)

here I1 and cos φ1 are current and PF at the LC.I2 and cos φ2 are current and PF sensed by the LDC y is the

istance between the LDC and the capacitor as a fraction of theistance between the LDC and the LC.

The error of the voltage at the LC is then:

LC − Uset = (1 − y)I2(RL cos φ2 + XL sin φ2)

− (1 − y)I1(RL cos φ1 + XL sin φ1) (11)

hich, for the same load, increases the bigger the capacitor andhe closer to the sending-end, as shown in Fig. 5. When theapacitor is connected after the LC, it will affect the performancef the LDC less, as the current and PF sensed by the LDC arepproximately the same as at the LC.

LDC regulation is more complicated when multiple feedersontrolled by one LTC/LDC are loaded differently. Fig. 6(b)hows that, when the feeders are loaded uniformly, both LTCnd LDC regulation yield voltage variations within allowableanges. But when the feeders are loaded differently, the feederith highest load will suffer undervoltage with LDC regulation,

s shown in Fig. 6(c).Thus it should be noted that, on a feeder with LDC regulation,

ot only the maximum load defines the regulation constraint, butlso the load factor difference (LFD) among the feeders. Off-lineimulations and proper commissioning should be performed to

djust the setting properly so that voltage variation on all feedersor all possible loadings and LFD will be within allowed limits.

Finally, activating LDC requires readjustment of capacitorontrol setting when the capacitor control is of voltage control

F.A. Viawan et al. / Electric Power System

Fig. 5. LDC regulation on a feeder with loads uniformly distributed at 10 nodes.LDC regulated point is node-4. QL is the total load reactive power and QC is thereactive power injected by capacitor.

Fig. 6. (a) Two-feeder system regulated by LTC/LDC; (b) voltage profile withLTC and LDC when both feeders have the same loading; (c) voltage profilewith LTC and LDC when loading is 50% for feeder-1 and 100% for feeder-2. Uset = 1.0425 pu for LTC, and Uset = 1.0 pu with regulated point 1 km fromsubstation for LDC.

tropwodal

r

i

3

3

uacafic

t

i

faw

f

U

iehte

pcka

s Research 77 (2007) 1314–1322 1317

ype. When the feeder is regulated by the LTC, the voltage iselatively high during low load and the capacitor will be turnedff by its voltage control. But when LDC is activated, voltagerofile during low load decreases, as shown in Fig. 6(b), andith the same voltage setting of the capacitor control, the turn-ff voltage might never be reached and capacitor will stay onuring low load. Though this condition does not lead to overvolt-ge, it causes overcompensation and therefore increases feederosses.

The voltage constraints for a conventional feeder with LDCegulation can then be formulated as:

i. U0,i

∣∣max load ≤ Umax, ∀ secondary side of LTC transformers

or VRs.ii. Uend,i

∣∣max load ≥ Umin, ∀ nodes at feeder-ends or primary

side of VRs.ii. Uend,i

∣∣max LFD ≥ Umin, ∀ nodes at ends of the feeder with

highest loading.

. Impact of DG on voltage regulation

.1. DG in feeder with LTC regulation

As explained in Section 2.1, the performance of LTC reg-lation is not affected by changes in PF or direction ofctive/reactive power, and, for multi-feeder systems, each feederan be treated individually. This means that only the voltage riset the DG connection point and its impact on the voltage pro-le of the particular feeder where DG is connected needs to beonsidered.

Voltage constraints for LTC regulated feeder with DG canhen be formulated as:

i. U0,i ≤ Umax, ∀ secondary side of LTC transformers or VRs.ii. Uend,i

∣∣max load,no DG ≥ Umin, ∀ nodes at feeder-ends or pri-

mary side of VRs.ii. UDG|min load, max DG ≤ Umax.

Other constraints are conductor ampacity Imax and trans-ormer rating STX,max. To obtain safe results, one should takes reference UUB when the limiting factor is Umax, and ULBhen the limiting factor is Umin, Imax or STX,max.The linear voltage drop/rise approximation for a simplified

eeder with one load and one DG at LC.

DG = U0 − RL(PL − PG) + XLQL

UDG(12)

ndicates that the DG integration limit can be increased by low-ring the sending-end voltage at LTC terminal U0. On the otherand, the current will increase when U0 decreases, which meanshat the ampacity constraint could be violated. However, theffect on the voltage profile is more significant.

One drawback of LTC regulation is that the maximum DG

ower that can be connected to the feeder is relatively low, espe-ially for DG connection far away from the substation. LTCeeps the sending-end voltage constant, which causes the volt-ge along the feeder to be high when the feeder is lightly loaded,

1 Systems Research 77 (2007) 1314–1322

sc

3

Pc

ttaddPhcwio

namtpb

wlro

tat

c

i

ii

ivv

bibsit

rlpn

4

4

eafilT[oarmfopv

318 F.A. Viawan et al. / Electric Power

o there is very little margin before the power produced by DGauses overvoltage.

.2. DG connection to feeder with LDC regulation

As explained in Section 2.2, LDC is affected by changes inF and by power direction reversal, which can occur due to DGonnection.

Eq. (8) is still valid, with I being the net current flow due tohe combined effect of load and DG. Rewriting Eq. (8) usinghe net active power (PL − PDG), it can easily be concluded thatlow value of PDG, less than about twice the load power, willecrease the net active power flow and thereby reduce the errorue to poorly adjusted X/R of the setting. On the other hand, theF also decreases, which counteracts the error reduction. Forigher values of PDG, the error will increase as compared to theonventional feeder and with increasing DG power. Moreover,hile DG operating at unity PF will have the beneficial effect of

ncreasing the overall PF, the situation is aggravated if the DGperates at leading PF, which should be avoided.

Similarly to the case of the capacitor shown above, DG con-ected between LTC and LC will alter the current and PF seent the LTC and thereby introduce an error. This situation can beore serious than in the case of the capacitor, since the capaci-

or is normally smaller than the total load, while the DG can inrinciple be much larger. The regulated point should in this casee moved to before the DG connection point.

DG will also worsen the voltage variations among the feedersith different loading when it is connected to the feeder with the

owest load. In this particular case, a solution can be to move theegulated point closer to the source (with appropriate adjustmentf Uset).

Thus, on feeders with LDC regulation, DG will not only affecthe voltage profile on the feeder where it is connected, but willlso cause more voltage drop on adjacent feeders, and decreasehe sending-end voltage U0.

The voltage constraints for a LDC regulated feeder with DGan then be formulated as:

i. U0,i

∣∣max load, no DG ≤ Umax, ∀ secondary side of LTC trans-

formers or VRs.i. Uend,i

∣∣max load, no DG ≥ Umin, ∀ nodes at feeder-ends or pri-

mary side of VRs.i. Uend,i

∣∣max LFD, max DG ≥ Umin, ∀ nodes at feeder-ends with

highest loading, with DG located on the feeder with lowestloading.

. UDG|min load, max DG ≤ Umax.

. U0,i

∣∣min load, max DG ≥ Umin, ∀ secondary side of LTC trans-

formers or VRs.

Similarly as with LTC regulation, Imax and STX,max will alsoe additional constraints. When the limiting factor is constraintii. or vi., the maximum allowed DG power can be increased

y increasing Uset or moving the regulated point closer to theource. When the limiting factor is constraint v., it can bencreased by decreasing Uset or moving the regulated point closero DG (farther from the source). Note however that moving the

at

f

Fig. 7. System studied models.

egulated point needs readjustment of Uset anyway. When theimiting factor is Imax or STX, max, the maximum allowed DGower can be increased by increasing Uset, but the increase willot be significant.

. Case study

.1. System model

The method presented in Section 3 is tested on three differ-nt models, as shown in Fig. 7. The simulations presented herere mainly performed with DIgSILENT PowerFactory [10]. Therst model is a simple radial feeder with uniformly distributed

oads along the feeder. The second model is the IEEE 34 Nodeest Feeder [11]. All feeder and load parameters are as given in11], with the addition of a 330 k VA 3-phase capacitor installedn node-890 to improve the voltage at this node (which, beforend after the installation of the capacitor is 0.92 and 0.97 pu,espectively, with all other conditions as in [11]). The thirdodel is a MV network consisting of three feeders with uni-

ormly distributed load of power PL at each node. Parametersf system model-1 and model–3 are presented in Table 1. Theower transformer for both model-1 and model-3 has nominaloltage 33/6.6 kV, short circuit impedance X = 10%, X/R = 10,

nd rating 7.5 and 12 MVA for model-1 and model-3, respec-ively.

Minimum feeder loading is 20% nominal and maximumeeder loading is 100% nominal. Maximum LFD among the

F.A. Viawan et al. / Electric Power Systems Research 77 (2007) 1314–1322 1319

Table 1Parameters of the system model-1 and model-3

Model/feeder Feeder length Conductor Load Capacitor

r (m�/km) x (m�/km) Rating (A) PL (MW) Power factor QC (MVar) Location (node)

1 5 120 350 610 0.5 0.85 1.2 83/Fd-1 6 270 350 360 0.2 0.85 1 113/Fd-2 5 270 350 360 0.34 0.85 1 73/Fd-3 5 270 350 360 0.4 0.85 1 6

Fm

fv

4

proptwfwvtUrw

vUTi

Fig. 9. Voltage profile along system model-1 with both LTC (Uset = 1.04 pu) andLm

tistla

tdapfiL

TT

D

2371

ig. 8. DG integration limit PDG at different DG connection points for systemodel-1 with different regulation methods.

eeders for model-3 is 20%. Maximum and minimum allowedoltage are Umax = 1.05 pu and Umin = 0.94 pu, respectively.

.2. Results and discussion

The DG integration limit with DG at different connectionoints is presented in Fig. 8 for system model-1 and differentegulation methods. Besides LTC and LDC, the other meth-ds considered are LTC with lowered setting, DG with reactiveower control capability (operating at leading PF) and installa-ion of VR. The original LTC setting is assumed to be 1.04 puith 0.015 pu bandwidth. Voltage profile with this LTC setting

or the case of maximum load with no DG and minimum loadith maximum DG connected at node-10 is shown in Fig. 9. Theoltage profile with minimum load and maximum DG ensureshat, when U0 = UUB = 1.0475 pu, see Eq. (1), UDG is lower than

max. If PDG is increased further by 100 kW, either voltage, cur-ent or transformer rating constraint, as explained in Section 3,ill be violated, which is shown in Table 2.Note in Fig. 9 that Uset can be lowered provided that the

oltage at the end of the feeder (node-10) is still higher thanmin when U0 = ULB with maximum load and no DG. Moreover,able 2 shows that with the original LTC setting, DG integration

s limited by the overvoltage constraint. Hence, lowering the set-

icwa

able 2echnical constraints that will be violated when DG power is increased 0.1 MW abov

G node LTC, Uset = 1.04 LTC, Uset = 1.03

UDG,max UDG,max

UDG,max UDG,max

UDG,max UDG,max

0 UDG,max UDG,max

DC regulation with maximum load and no DG and with minimum load andaximum DG at node-15.

ing to Uset = 1.03 pu, with bandwidth unchanged, will obviouslyncrease the DG integration limit, as shown in Fig. 8. Moreover,ince voltage rise due to DG is a function of PDG multiplied withhe line resistance, see Eq. (12), the increase of DG integrationimit gets lower when the DG connection point moves fartherway from the source.

The use of LDC in this single feeder system proves effectiveo increase the DG integration limit, as shown in Fig. 8. Therawback is power loss increase that is, however, not significant,s shown in Table 3. In this particular example, the regulatedoint for LDC is chosen such that the voltage profile along theeeder with maximum load and no DG with LDC regulations approximately the same as the voltage profile obtained withTC regulation with the original setting. The regulated points node-2 and the LDC setting is as shown in Table 4. The

orresponding voltage profile for the case of maximum loadith no DG and minimum load with maximum DG connected

t node-10 is shown in Fig. 9.

e corresponding PDG in Fig. 8

LDC DG at leading pf Install a VR

I2,max I2,max I2,max

I3,max I3,max I3,max

I7,max I7,max I7,max

UDG,max I10,max I10,max

1320 F.A. Viawan et al. / Electric Power Systems Research 77 (2007) 1314–1322

Table 3Distribution line losses of system model-1 (kW) with different regulation methods

Without DG With DG and 20% load

Load (%) LTC, Uset = 1.04 LTC, Uset = 1.03 LDC DG LTC, Uset = 1.04 LTC, Uset = 1.03 LDC

Node PDG (MW)

20 7.0 7.2 7.5 2 3.7 24.1 24.5 26.450 33.9 34.5 35.8 3 3.2 24.4 24.8 26.875 79.1 81.1 82.1 7 2.4 27.8 28.3 30.2

100 157.5 159.2 157.5 10 1.9 25.9 26.4 28.2

Table 4LDC regulator setting

Model LTC location CT ratio PT ratio Rset (�) Xset (�) Uset (V) Band-width (V)

1 Substation 600 32 2.27 6.61 120.6 1.82 Substation 100 120 6.46 4.8 121.2 2223

cpT1par

Vttbsfid

F(

etamtDcfsta

814-850 100 120852-832 100 120Substation 1000 32

The DG integration limit when using DG with reactive powerontrol capability shown in Fig. 8 is obtained by operating DG atower factor 0.985–0.99 leading, with the original LTC setting.he limit increases significantly, at the expense of an additional.0–1.4 MVAr of reactive power, depending on the connectionoint, flowing from the substation to the DG. This will requiresource somewhere else in the system to provide the required

eactive power.The DG integration limit is also increased significantly by the

R installation. The example in Fig. 8 is obtained by installinghe VR at node-6 with the setting Uset = 1.01 pu, and changinghe LTC setting to Uset = 1.02 pu, with 0.015 pu bandwidth foroth LTC and VR. The voltage profile with the VR installed is

hown in Fig. 10. There is a margin to decrease the VR settingurther, but, as the DG integration at either node-7 or node-10s already limited by current constraints, there is no benefit toecrease it more.

ig. 10. Voltage profile along the system model-1 with VR installation at node-6Uset = 1.02 pu for the LTC and Uset = 1.01 pu for the VR).

iimToD

ds

Fm

2.7 1.6 122 22.5 1.5 124 23.4 4.41 121.2 1.8

The DG integration limits for system model-2 with differ-nt regulation methods are presented in Fig. 11. The constraintshat will be violated when further increasing PDG by 100 kWre presented in Table 5. Compared with system model-1, thisodel has lower voltage drop during maximum load. Therefore,

he LTC setting can be decreased until 1.02 pu to allow higherG integration. For the case of DG with reactive power control

apability, the limit is obtained by limiting the minimum poweractor of the DG to 0.90. The VR is installed at node-808 withetting 1.01 pu. After installation of this VR, the regulator set-ings are 1.00 pu for the LTC and 1.02 for both VRs at 850-814nd at 852-832.

The increase in DG integration limit by operating DG at lead-ng power factor in this model is shown to be less effective than its in system model-1. The main reason is that the line in system

odel-2 has much higher resistance than in system model-1.he analysis of the effect of line resistance on the effectivenessf reactive power control with DG to limit voltage rise due to

G is given in [12].Finally, the DG integration limits for system model-3 with

ifferent regulation methods are presented in Fig. 12. The con-traints that will be violated when DG power is increased further

ig. 11. DG integration limit PDG at different DG connection points for systemodel-2 with different regulation methods.

F.A. Viawan et al. / Electric Power Systems Research 77 (2007) 1314–1322 1321

Table 5Technical constraints that will be violated when DG power is increased 0.1 MW above corresponding PDG in Fig. 11

DG node LTC, Uset = 1.04 LTC, Uset = 1.02 LDC DG at leading PF

808 UDG,max UDG,max STX,max UDG,max

814 UDG,max UDG,max UDG,max UDG,max

854 UDG,max U814,max U814,max UDG,max

858 U852,max U852,max U852,max U852,max

Table 6Technical constraints that will be violated when DG power is increased 0.1 MW above corresponding PDG in Fig. 11

DG node LTC, Uset = 1.04 LTC, Uset = 1.03 LDC DG at leading PF Install a VR

2 UDG,max UDG,max I3,max I3,max U2,max

3 UDG,max UDG,max U U U7 UDG,max UDG,max

10 UDG,max UDG,max

Fm

acfwd

siiisbtLsfts

5

wolo

sdDhp

(igipbt

epfaattt

fiaovhw

A

f

ig. 12. DG integration limit PDG at different DG connection points for systemodel-3 with different regulation methods.

re presented in Table 6. For the case of DG with reactive powerontrol capability, the limit is obtained with minimum poweractor of the DG equal to 0.90. The VR is installed at node-2ith setting 1.01 pu. The LTC setting could only be decreasedown to 1.03 pu.

Comparing this multi-feeder system with the single-feederystem in model-1 and model-2, one can conclude that thencrease in DG integration limit by activating LDC or bynstalling a VR is less effective in a multi-feeder system thann a single-feeder system. One reason is that, in a multi-feederystem, the voltage profile with LDC regulation will be definedy the average of all feeder voltages at the regulated points. Fur-hermore, when a VR is installed in a multi-feeder system, theTC setting cannot be decreased as much as in a single-feederystem, as the decrease will affect the voltage profile on othereeders. After the VR installation, the LTC setting was changedo 1.03 pu in system model-3, compared to 1.02 and 1.00 pu inystem model-1 and model-2, respectively.

. Conclusions

In this paper, voltage regulation in medium-voltage feeders

ith distributed generation (DG) has been treated. The principlef operation of on-load tap changers (LTCs) with and withoutine drop compensation (LDC) has been reviewed and the effectf DG on LTC and LDC regulation has been explained. Based on

R

DG,max DG,max DG,max

UDG,max UDG,max UDG,max

UDG,max UDG,max UDG,max

imulations on three different feeder models, the effectiveness ofifferent regulation methods (LTC with reduced setting, LDC,G with reactive power control capability and VR installation)as been analyzed and shown to depend on feeder structure,arameters and DG connection point.

It has been demonstrated that the use of on-load tap changersLTCs) with line drop compensation (LDC) can significantlyncrease the maximum size of DG that can be connected to aiven feeder without disrupting voltage regulation. By revis-ng the LTC settings and activating the LDC feature, which isresent in most LTCs but often not used, connection of DG cane allowed without the need for additional equipment to coun-eract problems such as voltage rise in low load conditions.

The use of LDC to increase the DG integration limit should bexplored as an alternative before, e.g., operating DG at leadingower factor, which implies additional reactive power flowingrom the substation to the DG, and VR installation, which meansdditional investment cost. If one of the latter measures must bedopted due to unwillingness of the distribution network opera-or (DNO) to change the LTC/LDC settings, it can be argued thathe DNO should then also be responsible for the cost of puttinghem in place.

The performance of LDC is affected by changes in poweractor and direction of power flow, which can occur with thenstallation of DG. However, with a proper commissioning andset of off-line simulations, it can be ensured that a given sizef DG can be connected at a given location without violating theoltage regulation constraints, for all load conditions. Finally, itas been shown that power loss increase due to the use of LDC,ith and without DG, is not significant.

cknowledgement

The authors gratefully acknowledge the financial supportrom Goteborg Energi Research Foundation, Sweden.

eferences

[1] T.A. Short, Electric Power Distribution Handbook, CRC Press LLC,2004.

1 System

[[

[

FoSCphoS

AftESfiRt

JhaopT

322 F.A. Viawan et al. / Electric Power

[2] T. Gonen, Electric Power Distribution System, McGraw-Hill Book Com-pany, 1986.

[3] M. Thomson, Automatic voltage control relays and embedded generationI, Power Eng. J. 14 (2000) 71–76.

[4] P. Brady, C. Dai, Y. Baghzouz, Need to revise switched capacitor controlson feeders with distributed generation, in: Proceedings of 2003 IEEE PESTransmission and Distribution Conference and Exposition, vol. 2, 2003,pp. 590–594.

[5] T.E. Kim, J.E. Kim, A method for determining the introduction limit ofdistributed generation system in distribution system, in: Proceedings of2001 IEEE Power Engineering Society Summer Meeting, vol. 1, 2001, pp.456–461.

[6] L.A. Kojovic, The impact of dispersed generation and voltage regulatoroperations on power quality, in: Proceedings of CIGRE 2005 Athens Sym-posium, 2005.

[7] C.L. Masters, Voltage rise the big issue when connecting embedded gen-eration to long 11 kV overhead lines, Power Eng. J. 16 (2002) 5–12.

[8] S. Repo, H. Laaksonen, et al., A case study of voltage rise problem due to alarge amount of distributed generation on a weak distribution network, in:Proceedings of. 2003 IEEE Bologna PowerTech Conference, vol. 4, 2003.

[9] J.H. Choi, J.C. Kim, Advanced voltage regulation method of power distribu-tion systems interconnected with dispersed storage and generation systems,IEEE Trans. Power Deliver. 16 (2) (2001) 329–334.

10] DIgSILENT PowerFactory, http://www.digsilent.de/.11] Distribution System Analysis Subcommittee Report, Radial distribution

test feeders, 2001 IEEE Power Engineering Society Winter Meeting, vol.2, pp. 908–912.

12] F. Viawan, A. Sannino, Analysis of voltage profile on LV distribution feed-ers with DG and maximization of DG integration limit, in: Proceedings ofCIGRE 2005 Athens Symposium, 2005.

tDNae

s Research 77 (2007) 1314–1322

erry A. Viawan received the B.Sc. and M.Sc. degrees from Bandung Institutef Technology, Indonesia in 1996, and Chalmers University of Technology,weden in 2003, respectively. He worked as a Power System Engineer at PTaltex Pacific Indonesia from 1996 to 2003, where he worked on operation,lanning and protection of a transmission and distribution system. Since 2004,e is a Ph.D. student at the Division of Electric Power Engineering, Departmentf Energy and Environment, Chalmers University of Technology, Gothenburg,weden.

mbra Sannino received the M.Sc. and Ph.D. degrees in Electrical Engineeringrom the University of Palermo, Italy in April 1997 and February 2001, respec-ively. From January 2001–December 2004 she was with the Department oflectric Power Engineering of Chalmers University of Technology, Gothenburg,weden, first as Assistant Professor and since January 2004 as Associate Pro-essor. From Chalmers University she also received the D.Sc. degree (Docent)n Power Systems in 2003. Since October 2004 she is with ABB, Corporateesearch, Vasteras, Sweden. Her interests include applications of power elec-

ronics in power systems, distributed generation, wind power and power quality.

aap Daalder received his D.Sc. degree in Power Engineering from The Eind-oven University of Technology, The Netherlands. He was Associate Professort the same university until 1984, when he left for Norway to become a Directorf Technology and a Member of the Board of a subsidiary of the ABB Com-any in Skien. In 1993 he was appointed Full Professor at Chalmers University ofechnology, Gothenburg, Sweden. He has been Head of the Department of Elec-

ric Power Engineering and currently leads the Group of Power Systems in theepartment of Energy and Environment. He is a board member of the Swedishational Committees of CIGRE and CIRED and of the ELEKTRA program. His

reas of interest are power systems and environmental issues related to powerngineering.