476 IEEE TRANSACTIONS ON POWER SYSTEMS, VOL. 22, NO.

1, FEBRUARY 2007

Centralized and Distributed Voltage Control: Impact

on Distributed Generation Penetration
Panagis N. Vovos, Aristides E. Kiprakis, A. Robin Wallace, and Gareth P. Harrison, Member, IEEE

Abstract—With the rapid increase in distributed generation sess voltage profile and dispatch DG and other network elements
(DG), the issue of voltage regulation in the distribution network accordingly [4]. The approach also requires investment in sen-
becomes more significant, and centralized voltage control (or sors and communication assets.
active network management) is one of the proposed methods.
Alternative work on intelligent distributed voltage and reactive An alternative approach of ensuring maximum capacity
power control of DG has also demonstrated benefits in terms of with minimal voltage impacts is through use of distributed,
the minimization of voltage variation and violations as well as the “intelligent,” power factor, and voltage control of DG and
ability to connect larger generators to the distribution network. other network components such as transformers with on-load
This paper uses optimal power flow to compare the two methods tap changers (OLTCs). Previously published work [5] demon-
and shows that intelligent distributed voltage and reactive power
control of the DG gives similar results to those obtained by cen- strates the benefits in terms of the minimization of steady-state
tralized management in terms of the potential for connecting voltage variations (and violations) as well as the ability to
increased capacities within existing networks. connect larger generators to the distribution network. While the
Index Terms—Dispersed storage and generation, optimal power capacity benefit could be easily quantified for individual DGs,
flow, power distribution planning, power generation control, the advantage of widespread usage of intelligent distributed
voltage control. control of DG has not been explored fully.
The aim of this paper is to compare the centralized and dis-
tributed approaches for controlling distribution network volt-
I. INTRODUCTION
ages in terms of the capacity of DG that could be accommo-
ONNECTION of distributed generation (DG) fundamen-
C tally alters distribution network operation and creates a va-
riety of well-documented impacts with voltage rise being the
dated within existing networks as well as contrasting them with
the current power factor control approach. To achieve this, it
was necessary to draw on earlier work [6]–[8] that used optimal
dominant effect, particularly in rural networks [1]. A range of power flow (OPF) techniques to evaluate the network capacity
options have traditionally been used to mitigate adverse impacts, available for connecting distributed generators. The technique
but these generally revolve around network upgrades, the costs has been extended to incorporate the intelligent generator con-
of which may be considerable. This occurs partly as a result trol algorithms and in doing so could find the maximum level of
of current distribution network operator (DNO) practice of as- DG penetration possible in each case.
sessing DG connections on a “first come-first served basis” as This paper is set out as follows: Section II explores the theo-
well as the “fit and forget” approach to DG operations that re- retical basis of the voltage variation and the potential methods
quire DG to operate at fixed power factors with capacity limited of control. Section III sets out the OPF method and the neces-
to minimize adverse impacts. These effects conspire to limit the sary augmentation required to incorporate the two voltage con-
capacity available within distribution networks to connect DG. trol techniques. Sections IV and V deal with a case study, while
However, a range of alternative operational approaches have Section VI discusses the implications of the results.
been proposed to raise the level of DG capacity that may be ac-
commodated within networks while limiting the need for capital II. THEORY
investment in new network assets.
A. Voltage Variation in Weak Networks
Active management, a form of centralized control of distri-
bution networks, has been proposed as a means of enhancing Traditionally, DNOs required that all distributed generators
connectable capacity [2]. Taking a similar approach to that used connected to the distribution network should operate in power
in transmission systems, a distribution management system con- factor control (PFC) mode. However, PFC has an adverse effect
troller would be used for wide area voltage control and reactive on generator terminal bus voltage, especially in the case of weak
power management [3]. It would employ state estimation to as- distribution networks. Fig. 1 illustrates the simplest form of an
electrical network that consists of two buses connected through
Manuscript received April 6, 2006; revised August 7, 2006. This work was
a single line.
supported by the Scottish Funding Council for the Joint Research Institute with The voltage drop across the line is approximated by the
the Heriot-Watt University, which is a part of the Edinburgh Research Partner- following equation:
ship. Paper no. TPWRS-00207-2006.
The authors are with the Institute for Energy Systems, Joint Research Institute
for Energy, School of Engineering and Electronics, University of Edinburgh, (1)
Edinburgh EH9 3JL, U.K. (e-mail: [email protected]; Aristides.Kiprakis@
ed.ac.uk; [email protected]; [email protected]).
Color version of Fig. 6 is available online at http://ieeexplore.ieee.org. where and are the line resistance and reactance, respec-
Digital Object Identifier 10.1109/TPWRS.2006.888982 tively, and and the active and reactive power exported from
0885-8950/$25.00 © 2007 IEEE
Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY BOMBAY. Downloaded on June 04,2025 at 10:20:57 UTC from IEEE Xplore. Restrictions apply.
 VOVOS et al.: CENTRALIZED AND DISTRIBUTED VOLTAGE CONTROL 477

Fig. 1. Simple two-bus system.

the DG bus toward the bulk supply point (BSP). Any fluctuation
in real power brings about a proportional fluctuation of voltage.
Fig. 2. Vector diagram of the generator voltage and current showing the re-
This is particularly evident in cases of weak networks where the sponse of the “intelligent” distributed controller.
line resistance is high relative to its reactance. In power factor
control mode, the ratio is maintained constant, so fol-
lows any variation of , tending to augment the voltage varia-
for DGs, which assumes a more flexible attitude from DNOs
tion. It is for these reasons that voltage rise is regarded as one of
concerning the voltage control by DGs. The authors’ target
the major impacts of, and constraints on, the connection of new
was to develop a voltage control method capable of keeping
DGs on the network, particularly in rural areas [1].
the DGs online during light or heavy demand conditions by
In (1), if was allowed to compensate for the feeder voltage
combining the advantages of AVR and PFC. The method was
rise or drop created by by adjusting in the opposite direction
termed automatic voltage/power factor control (AVPFC), and
(with ), then voltage could be maintained within limits al-
its steady-state response has the effect of relaxing power factor
lowing greater active power export. For voltage rise, this would
when voltage approaches the statutory limits. Its operation is
be achieved by defining a more leading power factor at which
depicted in the vector diagram of Fig. 2.
the generator is to be controlled. Power factor settings could
The operating point of the generator (tip of vector ) always
be specified so that DG operates at lagging power factor to
moves along the thick, dashed line. When voltage approaches
export reactive power during high demand periods while im-
the statutory limits, or , the PFC is deactivated and the
porting during low demand. While this appears to be a rela-
DG adjusts the production of reactive power to support voltage.
tively simple approach, it would require analysis to ensure that
The generator decreases the ratio when voltage drops to
voltage is maintained appropriately under all normal operation
the lower threshold, , while it increases the ratio
cases and may require a degree of central coordination.
when voltage reaches the upper threshold . Obviously, in
B. Voltage Regulation or Power Factor Control altering the ratio, the power factor is also being changed:
it must be restricted between the minimum and max-
DNOs have hitherto been reluctant to allow any operation by imum operating power factors.
independent generators, which could potentially disrupt the pas- The major advantage of this approach is that it does not re-
sive role of the distribution network to supply demand. Specif- quire knowledge of the distribution network as a whole, relying
ically, distributed generators are not permitted to perform au- purely on the voltage signals at its own terminals. As such, the
tomatic voltage regulation (AVR), an inherent feature of syn- approach would not require extensive deployment of sensors
chronous generators to regulate the terminal bus voltage by ad- and communications equipment and the “intelligence” can be
justing their reactive power output, as it may destabilize the embedded within the DG control system. However, the con-
automatic OLTCs of some distribution transformers. A further trollers would need to be set up such that they did not conflict
reason for the avoidance of AVR in most distributed generation with others, e.g., by causing hunting.
applications is that should a small generator with AVR con- 2) Centralized Voltage Control: Theoretically, voltage con-
trol attempt to correct for voltage drops, it has to inject great trol of the distribution network can be achieved centrally, in a
amounts of reactive power in order to raise the bus voltage. This similar manner to the transmission system through dispatch of
may result in high field currents and overheating for the gen- active and reactive power from distributed generators and other
erator, triggering the excitation limit or over-current protection network elements.
and disconnecting the generator from the network. Such an active management scheme would consist of a distri-
For the above reasons, most distributed generators have been bution management system controller accepting voltage, power
required by the DNOs to operate in power factor control mode. flow, and equipment status measurements at selected locations
PFC is less disruptive for network voltage control devices such in the distribution network. The controller would then use state
as OLTCs and results in much lower field currents brought about estimation to estimate network power flow and voltage profiles
by voltage drops under AVR control, thereby reducing thermal before dispatching plant according to economic dispatch from
stresses on the generator [9]. OPF [4] or a rule-based method.
The downsides to the approach include the necessary invest-
C. Alternative Voltage Control ment in sensors, communications, and dedicated controllers as
1) “Intelligent” Distributed Voltage/Power Factor Control: well as the balance required to ensure adequate accuracy of the
Kiprakis and Wallace [5] proposed a voltage control method state estimator.
Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY BOMBAY. Downloaded on June 04,2025 at 10:20:57 UTC from IEEE Xplore. Restrictions apply.
 478 IEEE TRANSACTIONS ON POWER SYSTEMS, VOL. 22, NO. 1, FEBRUARY 2007

III. SIMULATION TECHNIQUE where , and are the coefficients of the quadratic benefit
function and are negative. DG capacity is always positive, and
A. Optimal Power Flow as a Tool for Capacity Allocation is the benefit for the DNO from a new generator per unit
While OPF is traditionally used as an operating tool in power capacity connected at bus . The use of this formulation is nec-
systems, Harrison and Wallace [6] presented a means of using it essary as by minimizing negative costs DG capacity and benefit
to assess the availability of network capacity to accept DG. The is maximized.
approach arose as a means of allowing DNOs to guide developers 3) Transmission Network: The energy transfers to and from
to appropriate locations and avoid “sterilization” of the network the higher voltage transmission network or with adjoining distri-
or expensive upgrades. With an increasing number of DG con- bution systems are simulated as generators with quadratic cost
nections, their impacts become increasingly interdependent and functions. The coefficients of the cost functions are negative for
OPF presents a reliable means of dealing with this aspect. exports and positive for imports. The outputs of the generators
The approach set out in [6] represented DG as negative load are negative when they represent exports and positive when they
and employed the load-shedding algorithm in a proprietary OPF represent imports.
to maximize DG capacity such that network voltage and thermal 4) Existing Capacity and Loads: Existing generation ca-
constraints were respected. It was successfully able to demon- pacity is simulated as generators with constant active power
strate the consequences for available capacity when a small gen- output, equal to their maximum capacity, and given reactive
erator was located inappropriately. With the network constraints power injection capabilities. Loads are simulated as sinks of
limited to voltage and thermal limits, the approach was suit- constant active and reactive power.
able for rural networks, but it had limited applicability in urban 5) Network Constraints: The network must be operated
networks within which fault level constraints may predominate. within a narrow range of voltage to ensure safe operation of
This shortcoming was addressed in [7], where Vovos et al. pre- power system equipment and quality of supply
sented a bespoke OPF that explicitly represented DG capacity as
generators as well as developing a method for incorporating the (3)
constraints imposed by the fault level capability of switchgear.
where and are the lower and upper bounds of the
This aspect was developed further in [8] with fault level con-
voltage of bus around the rated value.
straints incorporated directly into the OPF formulation as simple
The thermal capacity of a line or transformer, , also sets a
nonlinear inequality constraints resulting in much greater accu-
limit to the maximum apparent power (MVA) transfer
racy and speed. Both [7] and [8] showed that the additional con-
straints had a significant impact on the ability of the network to (4)
accept new DG. The following outline of the method is based
broadly on the formulation in [7] and [8]. where is the apparent power, and is the thermal limit.
1) New Distributed Generation Capacity: At each location The connection of DG raises network fault currents at all net-
where DG may be connected (termed “capacity expansion lo- work locations with the impact being dependent on generator
cations” in [7] and [8]), the available DG capacity is simulated impedance and bus voltages. In [8], these effects are consid-
as a generator with quadratic cost functions. This formulation ered by relating the additional impedance introduced into the
is subtly different to traditional OPF as here the functions have impedance matrix to the MVA capacity of the DG. The fault
negative coefficients to indicate that development of DG results level constraints, , given by the fault capacity, , of each
in a negative cost or, in other words, a benefit. As such, DG with set of switchgear at bus
larger coefficients (i.e., more negative) will be favored, which al-
lows locational preferences to be expressed, although it is gen- (5)
erally assumed [6]–[8] that these are the same for all new DG in
order to avoid biasing the analysis. B. Enhanced OPF for Alternative Voltage Control
A key requirement for the approach is that the operating ca- The formulation in the preceding section is directly appli-
pability of the DG must “grow” with the capacity rather than cable for determining network capacities where the generator
being fixed like traditional generator models in OPF. In the case is operated in power factor control mode. This section shows
of power factor-controlled DG, new capacity has constant power the necessary enhancements to the OPF formulation in order to
factor (normally between 0.9 and unity) including those inter- allow analysis of the effect of the distributed and centralized
faced via inverters [10]. voltage control schemes.
2) Capacity Benefit Model: The target of the optimization With the benefit function (2) associated purely with DG ca-
procedure is to maximize the benefit from new generation ca- pacity, the voltage control scheme employed will only influence
pacity for the DNO. This is consistent with electricity regulators the objective function through the amount of DG connectable.
acknowledging the positive effect of DG with, for example, the As such, the approach is capable of encapsulating the impact of
U.K. regulator, Ofgem, providing localized monetary incentives different voltage control schemes on the benefit from new gen-
to DNOs to encourage them to connect DG to their networks eration capacity. Furthermore, the mathematical description of
[11]. The total benefit from new capacity across all candi- the contribution of DGs to fault levels is common for all voltage
date buses forms the following quadratic objective function control schemes.
(2) The main difference in the OPF formulation between DGs
operated under power factor control (PFC) and the distributed
Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY BOMBAY. Downloaded on June 04,2025 at 10:20:57 UTC from IEEE Xplore. Restrictions apply.
 VOVOS et al.: CENTRALIZED AND DISTRIBUTED VOLTAGE CONTROL 479

Fig. 3. Distributed voltage control strategy.

Fig. 4. Smoothing of control strategy transition.

or centralized voltage control schemes is that power factor is
allowed to vary, within the DG operating limits, when voltage
drops or rises beyond a critical threshold value. Since the focus
is on capacity planning, it is logical to expect that new capacity
will only raise voltage levels. Thus, in order to simplify the anal-
ysis, it is assumed that the power factor constraint is relaxed only
when the generator voltage rises to a critical value .
In addition, in order to consider both leading and lagging power
factors, the angle is constrained
rather than the power factor is positive for lagging
and negative for leading power factors).
Finally, the minimum and maximum operating
power factors are roughly the same for various sizes of DGs.
Therefore, it is assumed that , and are
common for all new DGs. Furthermore, is usually the
rated power factor (generation at rated output), so it is usually
the target of PFC. Both assumptions can be described in
Fig. 5. Central voltage control region of DGs.
the OPF formulation. First, the constraints for an allowed range
of power factor
where the functions , and are defined in the
(6) Appendix. This approximation creates a smooth transition
around for with respect to .
and for power factor controlled DG 2) Centralized Voltage Control: When DGs are centrally dis-
patched, their voltage control region is restricted only by their
(7) PF operating limits and the statutory voltage regulations

1) Distributed Voltage Control: The voltage control strategy (10)

of the distributed voltage controller is described by the curve in (11)
the voltage-reactive power graph in Fig. 3.
Equation (8) describes this control strategy mathematically Constraints (10) and (11) are shown graphically in Fig. 5.

IV. TEST CASE

when
when (8) A. Network Topology
The generic 12-bus 14-line distribution network presented in
In order to avoid the optimization burden that the discrete Fig. 6 has three potential DG connection points at buses 1, 10,
transition between power factor and voltage control creates, (8) and 11 [7], [8]. Connected to the 132-kV system at Bus 12, the
is approximated by the equality constraint below which repre- network has a common rated bus voltage level at 33 kV, ex-
sents the Sigmoid function in Fig. 4 cept for the load buses that are at 11 kV. The connection points
are connected to the 33-kV network through 30-MVA 33:11-kV
transformers, with taps operated fixed for simplicity. The au-
(9) tomatic tap changer on the 90-MVA 132:33-kV transformer is
Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY BOMBAY. Downloaded on June 04,2025 at 10:20:57 UTC from IEEE Xplore. Restrictions apply.
 480 IEEE TRANSACTIONS ON POWER SYSTEMS, VOL. 22, NO. 1, FEBRUARY 2007

TABLE II
OPTIMAL CAPACITIES WITH FAULT LEVEL CONSTRAINTS IGNORED

Fig. 6. The 12-bus 14-line test network [7], [8].

TABLE I
BRANCH CHARACTERISTICS FOR THE 12-BUS NETWORK [7]
D. Economics
The U.K. energy regulator Ofgem set an incentive for DNOs
for connecting DG of 1.50/year for each new kW of DG. This
represents the capital expenditure component of the incentive
[11] and is used as the marginal benefit, coefficient , in (2). In
this specific case, the other coefficients, and , are not required
and are therefore set to zero.

V. SIMULATION RESULTS
Two sets of analyses were performed for each of the three
voltage control modes. The first set includes the network voltage
and thermal constraints as a proxy for a rural network, while the
second set aims to represent a more urban network by including
the fault level constraints.
The initial capacity allocation for the “rural” version of the
network (i.e., with fault level constraints ignored) is presented in
enabled to operate in voltage control mode and regulate within Table II. It is clear that power factor control results in the lowest
% of the rated voltage at the low voltage side within a % total new capacity and export to the 132-kV network, while
tap range. The characteristics of transformers and lines are pre- the more relaxed operating modes offer greater connectable ca-
sented in Table I. The loads on buses 1, 3, 5, 6, 8, 10, and 11 are pacity. It is intuitive that the broader the operating voltage region
assumed to be drawing constant complex power. A pre-existing of the generators (see Figs. 3 and 5), the broader the solution
15-MW generator is installed on Bus 5, capable of providing up space for the OPF. Consequently, the benefit the DNO receives
to 10 MVAr of reactive power. for connecting DG (i.e., the objective function) increases as the
control over power factor is relaxed. Distributed and central-
B. Constraints ized voltage control, respectively, offer 72% to 86% gains in
Line 2–5 is constrained by a thermal limit of 14 MVA, 4–9 by connectable capacity and corresponding financial gain for the
a thermal limit of 40 MVA, while all other lines are considered DNO. However, the impressive total capacity achieved from the
to be unconstrained. Up to 100 MW of active and 60 MVAr methods with relaxed power factor comes at a cost: that of high
of reactive power can be exchanged with the 132-kV external losses that are between 4.8 and 5.8 times greater than with PFC.
network without affecting secure operation. Finally, bus voltage High losses have previously been suggested as an outcome of
fluctuations were limited to % around the nominal values. active (centralized) voltage control [13].
Switchgear is rated at 250 MVA at 11 kV, 1000 MVA at 33 kV, The high losses are due to a combination of increased active
and 3500 MVA at 132 kV, which are typical U.K. ratings [12]. power export from the generators and their operation at more
leading power factors (increased reactive power imports) to
C. Voltage Control Properties regulate local voltage levels. The action of the two voltage
The rated power factor of each DG is assumed to be 0.9 lag- control techniques is clearly shown by inspecting the power
ging. The distributed and centralized voltage control strategies factors of the DG at each location: in all cases, these have
allow relaxation from this value within the range of 0.9 lag- become more leading to hold the voltage down by absorbing
ging and 0.9 leading. With distributed voltage control, the power reactive power.
factor is allowed to vary once the generator voltage reaches In the second set of analyses for the “urban” version of the
. network, it became clear that connecting DG in the capacities
Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY BOMBAY. Downloaded on June 04,2025 at 10:20:57 UTC from IEEE Xplore. Restrictions apply.
 VOVOS et al.: CENTRALIZED AND DISTRIBUTED VOLTAGE CONTROL 481

TABLE III A. Distributed or Centralized?

OPTIMAL CAPACITIES WITH FAULT LEVEL CONSTRAINTS INCLUDED
It is clear from the results that both voltage control schemes
have a major impact on the potential penetration of DG. It is
also interesting to note that while the centralized control scheme
delivers greater DG penetration, the difference is restricted to
4% to 8% across the two cases. There is also a more significant
impact on “rural” networks that is intuitively correct given the
predominance of voltage rise in such networks and the fact that
they would benefit most from voltage control. It is clearly in the
interest of DNOs to consider means of raising the penetration of
DG, particularly when they benefit financially.
The focus in this paper has been on voltage control and the
benefits in terms of improving DG penetration. A significant
issue for comparing the two competing approaches is the costs
and risks associated with them. In particular, this includes the
upfront costs associated with the centralized, active manage-
suggested in Table II would violate the breaking capacity of the ment approach. Further work is required to provide a thorough
switchgear connected to buses 1 and 10. When fault levels are cost benefit comparison of the two approaches to distribution
included as a network constraint, the overall level of connectable network voltage management. Other areas requiring attention
capacity alters (see Table III) with reductions in overall capacity would be to explore the network performance with additional
of 3.5%, 8.5%, and 12%, for PFC, distributed, and centralized active network elements, OLTCs, SVCs under distributed or
voltage control, respectively. Due to the proportionately larger centralized control of active and reactive power and voltage.
impact on the voltage control schemes, the relative capacity im-
provement over PFC reduces to 63% for distributed control and B. Losses
70% for centralized. The financial impact for the DNO is that With DNOs in the U.K. now incentivized to manage losses as
benefits would be reduced by between 2500 and 15 500. a means of improving economic efficiency and environmental
The balance of capacity between the locations also alters. impact [11], there would be concern over the increase in losses
For power factor control, the effect of relieving fault level con- suggested here. The incentive scheme sees DNOs rewarded for
straints is to reallocate all capacity from bus 1 to the other lo- losses below and penalized for higher than a historic loss bench-
cations. With the other two control schemes, it is Bus 10 that mark value. With benchmark losses in this example calculated
suffers reductions in capacity with bus 1 accepting much of the to be 2 MW, a loss incentive of 3 p/kWh suggests that DNOs
additional capacity. would face an hourly penalty of 60, 636, and 753 for power
The power factors of each DG highlight the active con- factor, distributed, and centralized control, respectively.
straints on capacity at each location. For Bus 10 the power Given that loss penalty per hour is far in excess of the
factor remains at 0.9 lagging under both distributed and central- yearly benefit the DNO receives for connecting DG capacity
ized voltage control. This implies that there is no voltage rise ( 1.50/year), it was deemed useful to consider how the DNO
problem requiring control action at the capacities indicated, might locate DG when the loss incentive was taken into account
i.e., that fault levels are the binding constraint. and what effect this would have on the relative performance of
The impact on network losses is more complex in this case the voltage control schemes. To do this, the OPF formulation
with all but distributed control seeing modest reductions, consis- was enhanced to include losses. The original capacity benefit
tent with the decrease in total capacity. With distributed voltage function (2) was restated to account for the incentive, , (3
control, the losses actually increase as a result of greater ca- p/kWh), provided by the difference between actual, ,
pacity being located at Bus 1, which is electrically remote from and benchmark losses,
the 132-kV network. This raises questions regarding the extent
of the net benefit (in energy terms) that arises from increasing
the installed capacity within a weak distribution network. It is (12)
possible that by increasing DG capacity in such a system, the
losses would be dramatically increased, causing the net MW de- The OPF was run with this new objective function for each
livered to decrease below the levels experienced when pure PFC of the earlier control methods and with and without fault level
is utilized. This would imply that the increased losses would constraints. With the results being identical whether fault levels
have to be partially covered by central generation. Hence, in- were included or not, Table IV contains the results from the
creased DG would have a negative impact to the energy balance OPF.
of the system. What is immediately clear from Table IV is that apart
from minor differences in power factor and the corresponding
VI. DISCUSSION impact on imports and benefits values, all three cases allow the
There are two areas of interest here: the relative merits of the same combination of DG capacities. The capacity connectable
distributed and centralized control schemes and the losses. is around 60% of the original amount under power factor
Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY BOMBAY. Downloaded on June 04,2025 at 10:20:57 UTC from IEEE Xplore. Restrictions apply.
 482 IEEE TRANSACTIONS ON POWER SYSTEMS, VOL. 22, NO. 1, FEBRUARY 2007

TABLE IV Here
OPTIMAL CAPACITIES CONSIDERING LOSS INCENTIVE

and is the intersection of the smoothed curve with the

voltage threshold . is a real number margin-
ally greater than 1, which defines the steepness of . The
higher the value, the smoother the transition from to .
A value of 1.01 for produces a quite smooth function
without significant loss in precision.
control. The strength of the loss reduction incentive appears
REFERENCES
to be such that the benefits of voltage control would not be
[1] C. L. Masters, “Voltage rise: The big issue when connecting embedded
recognized. generation to long 11 kV overhead lines,” Power Eng. J., vol. 16, no.
Clearly as the analyses have been carried out under condi- 1, pp. 5–12, Feb. 2002.
[2] S. N. Liew and G. Strbac, “Maximising penetration of wind genera-
tions of maximum generation and minimum load, the maximum tion in existing distribution networks,” Proc. Inst. Elect. Eng., Gen.,
possible reverse power flows are captured under these circum- Transm., Distrib., vol. 149, no. 3, pp. 256–262, May 2002.
stances. As such, the power losses and correspondingly the [3] T. Bopp, A. Shafiu, I. Cobelo, I. Chilvers, N. Jenkins, G. Strbac, H. Li,
and P. Crossley, “Commercial and technical integration of distributed
hourly financial penalties are overstated as load will be larger at generation into distribution networks,” in Proc. CIRED, 17th Int. Conf.
all other times. This clearly represents the worst-case scenario Electricity Distribution, 2003.
[4] A. Shafiu, T. Bopp, I. Chilvers, and G. Strbac, “Active management
in terms of DG penetration. Further work is required into the and protection of distribution networks with distributed generation,” in
balance between DNO incentives to reduce losses and those for Proc. IEEE Power Eng. Soc. General Meeting, Jun. 6–10, 2004, vol. 1,
pp. 1098–1103.
connecting distributed generation. [5] A. E. Kiprakis and A. R. Wallace, “Maximising energy capture from
distributed generators in weak networks,” Proc. Inst. Elect. Eng., Gen.,
Transm., Distrib., vol. 151, no. 5, pp. 611–618, Sep. 2004.
VII. CONCLUSION [6] G. P. Harrison and A. R. Wallace, “OPF evaluation of distribution net-
work capacity for the connection of distributed generation,” Proc. Inst.
If DNOs move away from strict power factor control of Elect. Eng., Gen., Transm., Distrib., vol. 152, no. 1, pp. 115–122, Jan.
distributed generation, the remaining capacity in the existing 2005.
[7] P. N. Vovos, G. P. Harrison, A. R. Wallace, and J. W. Bialek, “Optimal
network can be better exploited. Intelligent distributed and power flow as a tool for fault level constrained network capacity anal-
centralized voltage control methods offer significant gains in ysis,” IEEE Trans. Power Syst., vol. 20, no. 2, pp. 734–741, May 2005.
connectable capacity, particularly in rural networks. Ironically, [8] P. N. Vovos and J. W. Bialek, “Direct incorporation of fault level con-
straints in optimal power flow as a tool for network capacity analysis,”
it appears that consequent losses appear to increase substan- IEEE Trans. Power Syst., vol. 20, no. 4, pp. 2125–2134, Nov. 2005.
tially. This was not an issue until now, but with the newly [9] T. W. Eberly and R. C. Schaefer, “Voltage versus VAr/power-factor
regulation on synchronous generators,” IEEE Trans. Ind. Appl., vol.
imposed loss penalties, the financial implications of losses 38, no. 6, pp. 1682–1687, Nov.–Dec. 2002.
increase have to be carefully assessed by the DNOs. Further [10] S. R. Wall, “Performance of inverter interfaced distributed generation,”
work is required on the relative costs and benefits of both in Proc. IEEE/Power Eng. Soc. Transmission Distribution Conf. Expo.,
2001.
voltage control approaches given the upfront costs associated [11] Ofgem, Electricity Distribution Price Control Review, Final Proposals,
with centralized, active network management. Further research Office of Gas and Electricity Markets. London, U.K., Nov. 2004.
into the incentives given to DNOs to manage losses and connect [12] B. M. Weedy and B. J. Cory, Electric Power Systems, 4th ed. Chich-
ester, U.K.: Wiley, 1998.
distributed generation is also warranted. [13] G. Strbac, N. Jenkins, P. Djapic, M. Hurd, and G. Nicholoson, “Benefits
of Active Management,” 2003, Dept. of Trade and Industry. London,
U.K.
APPENDIX
EXPANSION OF (9) Panagis N. Vovos was born in Athens, Greece, on
October 23, 1978. He received the M.S. degree from
Equation (9) provides a continuous approximation of the the Electrical Engineering Department, University of
discrete transition between power factor and voltage control Patras, Patras, Greece, in 2002 and the Ph.D. degree
modes from the University of Edinburgh, Edinburgh, U.K.,
in 2005.
His special fields of interest include cross-border
congestion management, efficient capacity alloca-
tion, and enhanced OPF.

Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY BOMBAY. Downloaded on June 04,2025 at 10:20:57 UTC from IEEE Xplore. Restrictions apply.
 VOVOS et al.: CENTRALIZED AND DISTRIBUTED VOLTAGE CONTROL 483

Aristides E. Kiprakis was born in Heraklion, Crete, Gareth P. Harrison (M’02) is a Lecturer in Energy
Greece. He received the B.Eng. degree in electronics Systems in the School of Engineering and Elec-
from the Technological Education Institute of Crete tronics, University of Edinburgh, Edinburgh, U.K.
and the Ph.D. degree in electrical power engineering In addition to his work on integrating distributed
from the University of Edinburgh, Edinburgh, U.K. generation into electricity networks, he is involved
He is currently a Research Fellow with the Uni- in analyzing the impact of climate change on the
versity of Edinburgh, working for the Supergen Ma- electricity industry with emphasis on hydropower,
rine Energy Consortium. His research interests in- marine energy, and electricity demand.
clude modeling and control of distributed generation Dr. Harrison is a Member of the Institution of
and renewable energy resources. Engineering and Technology, U.K., and a Chartered
Dr. Kiprakis is a member of the Institution of En- Engineer.
gineering and Technology, U.K.

A. Robin Wallace is Head of the Institute for Energy

Systems at the University of Edinburgh, Edinburgh,
U.K. In addition to his research in network integra-
tion of distributed generation, he is Co-Director of the
U.K. Energy Research Centre, leads the Engineering
and Physical Sciences Research Council’s Supergen
Marine Energy Consortium, and is a Consultant on
small hydropower and distributed generation issues.
Prof. Wallace is a Fellow of the Institution of En-
gineering and Technology and a Chartered Engineer.

Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY BOMBAY. Downloaded on June 04,2025 at 10:20:57 UTC from IEEE Xplore. Restrictions apply.