Benefits of Active Management of Distribution Networks with Distributed Generation J. Mutale, Member, IEEE
Abstract--Electric power distribution systems are in a state of transition from traditional passive systems with unidirectional power flows from high to low voltages to active networks with multidirectional power flows. This change is driven primarily by the growing penetration of distributed generation of different technologies including new and renewable types. The process of change is likely to be painful for network utilities as it requires abandoning some well established, tried and tested methods of planning and operating their systems. In order for the change to be embraced by all stakeholders including regulators, utilities and customers there is need for the costs and benefits of active management to be quantified. This paper discusses the work that has been undertaken at The University of Manchester in this area focusing on the benefits of active management in the UK. Index Terms— Active management, Cost benefit analysis, Integration of Distributed generation, Distribution network
basic principles of active management are then reviewed. A case study to illustrate the benefits of different active management strategies is then presented. The paper concludes with a discussion on incentive mechanisms devised to encourage distribution network operators in the United Kingdom to connect more DG and to seek innovative solutions to the challenge of cost effective integration of DG. II. DRIVERS OF ACTIVE MANAGEMENT Fig. 1 is a pictorial depiction of two possible trajectories of total system costs as DG penetration grows with time. The primary driver of active management is to reduce the overall system costs so as to avoid trajectory B in Fig. 1 and to follow instead trajectory A. System costs following Trajectory B would arise if the present passive network management philosophy continues.
I. INTRODUCTION
A
CTIVE management of distribution networks is central to the cost effective integration of DG into power systems. The current approach of connecting DG is many cases based on a so-called “fit and forget” policy, which is consistent with passive network management. A fundamental underlying hypothesis of this policy is that control problems should be solved at the planning stage by providing adequate network capacity. For example, if, as is often the case, DG output leads to voltage rise during periods of low demand, either the capacity of DG that can be connected is constrained or alternatively, a larger cable or overhead conductor size is used. It is not difficult to see how such an approach would lead to higher total system cost in the long term. It also unduly penalizes DG developers by limiting the capacity that could be connected. Where the DG is based on renewable sources, this approach would undermine the objective of maximizing use of renewable energy sources to meet climate change targets of reducing CO2 emissions from power generation. This paper reviews the work that has been carried out at the University of Manchester in the area of DG integration focusing on the quantification of the benefits of active management of distribution networks. It starts by briefly revisiting the drivers behind the increasing penetration of DG and restating the case for active management as a means of effective integration of DG to reduce total system costs. The J. Mutale is a Senior Lecturer in the School of Electrical and Electronic Engineering, The University of Manchester, PO Box 88, Manchester, M60 1QD, UK (e-mail:
[email protected])
142440178X/06/$20.00 ©2006 IEEE
Fig. 1. Trajectories of future system costs
It should, however, be noted that for system costs to follow trajectory A it is necessary that as DG penetration increases displacing the energy produced by centrally dispatched generation, at the same time novel control systems and strategies must be developed that will enable DG either as individual units or as aggregates to have the flexibility and controllability similar to central plant. The latter attributes underpin system security. DG being equipped with such attributes will guarantee that it not only displaces energy (MWh) produced by central plant but also displaces the associated capacity (MW). In the short term, change in control and operating philosophy of distribution networks is likely to increase costs to cover research, development and deployment of new technologies and required information and communication infrastructure as suggested in Fig. 1.
601
PSCE 2006
However, full integration of DG and demand side via decentralised control philosophy will deliver benefits. This paper’s principal focus is on active voltage control in distributions system with DG. The basic principles of this control strategy are explained in the next section. III. PRINCIPLES OF ACTIVE MANAGEMENT The voltage rise effect is illustrated using a simple circuit shown in Fig. 2. This figure represents the basic features of a distribution system into which a distributed generator is connected (assumed at 11kV). This generator (DG) together with a local load (PL, QL) and a reactive compensator (QC) is connected to the distribution system (DS) via a weak rural distribution overhead line with impedance Z = (R+jX) and a 33/11 kV transformer with an On Load Tap Changer (OLTC).
B. Managing the voltage rise effect by generation curtailment It is important to observe that the probability of such an extreme situation (coincidence of minimum load with maximum generation) actually occurring is generally low, and hence it may be beneficial to accommodate a larger generator at busbar 2 and curtail it when voltage at busbar 2 rises to the limit. The effect of generation curtailment on the capacity that can be connected is given by (3) below. max g
P
cur G
|P
V2max V1 R
(3)
The likelihood of the coincidence of minimum load with maximum generation will determine the total annual energy curtailed. As the price of electricity is primarily driven by load demand, and generation curtailment occurs typically during periods of low load, the value of this energy curtailed is likely to be relatively low.
DS
Fig. 2 Simple system for modeling voltage rise
The voltage at busbar 2 ( V2 ) can be approximately calculated as follows:
V2 | V1 R Pg PL r Qg QL r QC X
(1)
This simple equation can be used to analyze qualitatively the relationship between the voltage at busbar 2 and the amount of generation that can be connected to the distribution network, as well as the impact of alternative control actions. A. Worst Case Scenario (Minimum Load Maximum Generation) Approach Under this scenario the capacity of generation that can be connected to a distribution circuit is determined by analysing the extreme conditions of the coincidence of minimum load and maximum generation. This policy enables Distribution Network Operators (DNOs) to continue to operate their systems as if generators were not connected at all. The effect of such a connection policy on the amount of generation that can be connected to an existing system can be analysed by the following expression (for the of sake simplicity unity power factor operation is assumed, i.e. rQGrQC=0):
V2 | V1 RPGmax
(1)
The capacity of the generator that can be accommodated in the existing system is clearly limited by the maximum voltage at busbar 2:
PGmax d
impedance, R, is critical for the amount of generation that can be connected (the value of reactance, X, is not relevant as the generator is assumed to operate with a unity power factor). This resistance is determined by conductor size and is assumed constant for a given system.
V2max V1 R
C. Managing the voltage rise effect by reactive compensation Managing of reactive power injections can make a considerable impact on the capacity of generation that can be connected to weak overhead distribution networks. If reactive power, Qimport ( i.e. r QG QL r QC ), is absorbed from the network, the amount of generation that can be connected under no load conditions can be increased:
PGmax |
V
max 2
V1 Qimport X R R
(4)
Observe that the effectiveness of reactive power import is greatly influenced by the value of line reactance X. In this context, reactive compensation is considerably more effective on overhead networks (with typical reactance of X OH | 0.4[: / km] ), than on cable networks (with typical reactance of X C | 0.1[: / km] ). It is also important to bear in mind that absorbing reactive power would lead to an increase in losses, and the evaluation of this control option should therefore include loss assessment. D. Managing the voltage rise effect using coordinated voltage control Control of voltage at busbar 2 by regulating voltage V1, at busbar 1, using the OLTC, can considerably increase the capacity of distributed generation. In this control option, the min
OLTC is used to lower voltage to the minimum value V1 enabling larger injection of active power at busbar 2:
(2)
It is important to observe that the real part of the network
602
PGmax d
V2max V1min R
,
(5)
However, in a more complex network, the value of this
voltage, and the corresponding tap position of the OLTC, would have to be optimised. All these three methods of regulating voltage can be applied in combination. It should also be noted that reinforcing the system could also enhance the amount of generation that can be accommodated.
was a basis for constructing wind generator output for various installed capacities. Reactive demand of the generator is approximately modelled and together with two sizes of power factor correction applied, average power factors of 0.95 and 0.98 are achieved (absorbing VARs). The benefit of following active management options were explored using optimal power flow simulations:
IV. CASE STUDY A. System description The case studies presented below are based on the work reported in [1]. The 33kV distribution network on which the case studies were carried out is shown in Fig. 3. The network is fed from a 132 kV network (busbar 1) through an OLTCtransformer. Loads are connected to busses 2, 3, 4 and 5. The load at busbar 2 represents the aggregated loads of the remaining part of the system. Distributed wind generation is connected at bus 6, where power factor correction is also connected. Branch parameters of this network are given in Table 1.
Fig. 3. Case Study - 6 Bus Distribution Network
TABLE I BRANCH PARAMETERS
A mixture of residential, industrial and commercial loads is allocated to each of the busbars. Hourly averages of active and reactive power are used to form annual load profiles. These take into account daily and seasonal variations of load and may be considered to be typical. Reactive demand profiles are also modelled following generally accepted rules characteristic for the majority of load (variations in active load are considerably greater than in reactive load). Peak values of the loads are given in Fig. 3, together with line flows and losses. Modelling of the hourly output of wind generation is based on the Stochastic Markov Model presented in [2]. This model is used to create a normalised annual generation profile, which
-
Generation curtailment Reactive compensation and voltage control Area based voltage control by OLTC Area based voltage control by OLTC and voltage regulator The impact of voltage controls on losses was also assessed.
-
The results are presented and discussed below. B. Base case scenarios Base case scenarios represent the present, passive approach to voltage control. Two characteristic conditions are considered: (i) minimum load – maximum generation and (ii) maximum load – maximum generation. In all studies it was assumed that the voltage could vary within +/- 3% from the nominal value. Applying the first criterion (i) it was found that a 10MW wind generator could be connected, while the criterion (ii) reduces this to only 6 MW. In [1], the concept of optimal power flow (OPF) is applied to schedule the available controls optimally and to quantify the benefits of alternative voltage control strategies. This benefit is expressed in terms of the increased amount of DG that can be connected. The main purpose of an OPF applied in the context of active management of distribution networks is to determine the optimal schedule of available controls (generation curtailment, VAR absorption, turns ratio of the OLTC transformer, load shedding, etc) that minimise the total cost of taking these actions while satisfying voltage and thermal constraints. In this particular case, the general optimisation task is reduced to a problem of minimising the amount of generation that has to be curtailed in order to satisfy voltage constraints. Clearly, if the voltage rises above the maximum value due to high wind generation output the OPF would enforce the satisfaction of the violated constraints through curtailing the generation, while maintaining power factor of the wind farm unchanged. This problem, for each hourly period t, may be stated mathematically as follows: Objective function, Minimise
I ( PGicur )
(7)
Subject to:
603
PGi PLi PGicur QG1 QC QLi S ij d S ijmax
Vi
min
d Vi d Vi
Pi inj (V ,T , T ) inj i
Q (V ,T , T )
(8) (9) (10)
max
(11)
PGimin d PGicur d PGimax min C
Q
max C
d QC d Q
min k
d Tk d T
cur Gi
cur Gi
T
max k
Q
f (P
)
(12) (13) (14) (15)
Where,
PLi , QLi
Active and reactive load at node i, at time t
PGi , QGi
Active and reactive generation at node i, at time t
cur Gi
P
cur Gi
,Q
Active and reactive generation curtailment or increase at node i, at time t (if possible)
QC
Fig. 4 Energy produced and curtailed at 0.98 power factor
Reactive power generated/absorbed by a
Fig. 4 shows the resultant annual energy produced and curtailed for various installed capacities ranging from 4 MW to 20 MW, in steps of 2 MW. Based on the current practice, the capacity of DG allowed for connection is generally limited by the extreme condition of minimum or maximum loading and maximum generation output. This condition only allows 6 MW of generation to be connected. For this level of output no generation curtailment occurs and any increase in the level of penetration will lead to violation of voltage limits at the connection point. The light bars represent the net energy generated in the course of one year, while the dark bars represent the curtailed energy. Net energy generated increases until penetration reaches 12 MW. The value of energy generated and curtailed can be calculated from energy prices prevailing in the market.
reactive compensator, at time t
Pi inj , Qiinj
Active and reactive power injection at node i, at time t
Tk
Tap setting of the tap-changer k, at time t
S ij
Load flows of the branch ij, at time t
S ijmax
Maximum control load flow in branch ij, at time t
Ti
Voltage angle at node i, at time t
Vi
Voltage at node i, at time t
The objective function (7) minimises the total cost of generation curtailment. In general, however, exercising each of the available control actions could be associated with some cost. Equations (8) and (9) represent nodal power balance. The optimisation is also subject to the branch thermal constraint (10) and network voltage limits (11). The maximum amount of active generation curtailed will be limited by the capacity of DG connected (12). Reactive power curtailment may be correlated with the active power curtailment, which is modelled through (9). Reactive power support is limited by the capacity of reactive compensation installed (13). The tapchanger setting Tk will be optimised and can vary within the bounds given by (14). C. Generation curtailment In this exercise, generation curtailment is used to manage voltage rise effect. The OLTC transformer maintains a constant voltage at its terminals. The effect of applying generation curtailment to larger wind farm schemes was investigated. Two case studies are performed (i) operation with average power factor of 0.98 and (ii) operation with average power factor of 0.95. The results are summarised in Fig. 4 and Fig.5.
Fig. 5. Energy produced and curtailed at 0.95 power factor
In the case of power factor of 0.95 (Fig. 5), the net energy generated increased from 11,152 MWh (for a 6 MW installed capacity) to about 14,631 MWh (for an 8 MW installed capacity), where 239 MWh (or 1.61 %) is curtailed. For the 8 MW wind farm, the reduction in average power factor from 0.98 to 0.95 reduces the amount of energy curtailed from 898 MWh to 239 MWh. This clearly shows that requiring
604
wind farms to operate with unity power factor (used in some utilities) will severely limit the amount of generation that can be connected. D. Area based control by OLTCs In previous cases the OLTC is used to maintain voltage magnitude at busbar 2 to a constant value of 1.0 p.u. In this case the tap position of the OLTC is optimised in order to minimise generation curtailment. The results are shown in Fig. 6.
a similar way as an OLTC with the ability to change the turns ratio from 0.9 to 1.1 p.u. continuously. The results of this case are shown in Fig. 7. It is clear that this voltage control policy increases the amount of generation that can be connected (up to 20MW) with almost no curtailed energy.
Fig. 7. Energy produced and curtailed with Application of OLTCs and Voltage Regulators
Fig. 6. Energy produced and curtailed with application of OLTC
Year round analyses using the OPF tool show that penetration levels up to 14 MW can be achieved with virtually no energy being curtailed (for a 14 MW wind farm, only 87 MWh is expected to be curtailed in one year). For a wind farm of 16 MW, the amount of energy curtailed is 1578 MWh or 5.3 %. Clearly area-based voltage control by OLTC transformers allows a considerable increase in penetration of DG. It is however important to remember that this technique of voltage regulation will need implementation of a Distribution Management System (with appropriate communication systems). E. Application of OLTCs and Voltage Regulators As indicated above, minimum load - maximum generation conditions are usually critical for the amount of generation that can be connected, due to the voltage rise effect. However, it may also be necessary to consider maximum load – maximum generation conditions. This is because, the use of OLTC transformer to reduce the voltage on the feeder where the generator is connected, may produce unacceptable voltage drop on adjacent feeders that supply loads. In this case it may be beneficial to separate the control of voltage on feeders that supply load from the control of voltage on the feeder to which the generator is connected. This can be achieved by the application of voltage regulators. In order to examine the benefits of this option, a voltage regulator is inserted at the beginning of the feeder, which accommodates the wind farm. This allows an independent voltage regulation on feeders with loads by the OLTC, while the voltage regulator controls the voltage on the feeder with the wind farm. In the OPF the voltage regulator is modelled in
F. Impact of Voltage Controls on Losses Connecting DG alters the loss performance of distribution networks. Small DG penetrations tend to reduce network power flows, and thus network losses. However, when penetration is large then DG will export power to the grid and may cause losses to increase. In this particular system, with no generation being connected network losses are 2860 MWh/year. Fig. 8 shows the level of network losses for various levels of penetration and various voltage control strategies applied.
Fig. 8. Impact of Voltage Controls on Losses
It is clear from Fig. 8 that for a low level of DG penetration losses are reduced regardless of the voltage control. For the higher level of penetration, enabled by the OLTC and the voltage regulator based controls, losses are significant due to the increased amount of DG output. G. Benefits of active management: UK rural 11kV networks A study was carried to quantify the benefits of active management across the UK. The results are summarised in Table 2. It is clear from this table that the benefits of active
605
management depend on the level of penetration of DG as well the characteristics of the networks involved. For low lowdensity rural networks DG penetration of up to 5GW (UK has a system peak demand of the order of 60GW), active management has no benefits. TABLE 2 BENEFITS OF ACTIVE MANAGEMENT: UK RURAL 11KV NETWORKS [M$]
In contrast to the IFI, the second scheme known as Registered Power Zones (RPZs) [4] is focused specifically on the connection of generation to distribution systems. RPZs are intended to encourage DNOs to develop and demonstrate new, more cost effective ways of connecting and operating generation that will deliver specific benefits to new distributed generators and broader benefits to consumers generally. VI. ACKNOWLEDGMENT The author is grateful to Professor G. Strbac, Imperial College London, Professor N. Jenkins, University of Manchester and P. Djapic, Imperial College London for undertaking the work and the case studies reported in this paper.
For high-density networks the benefits of active management begin to accrue at a penetration level of 5GW. As penetration grows up to 10GW, the benefits of active management are very significant exceeding $1Billion for the high-density systems.
VII. REFERENCES [1]
[2]
V. CONCLUSION Studies have shown that application of active management of distribution networks in the UK would enable connection of significant amounts of DG without reinforcing existing networks. However the effective integration of DG into distribution networks requires the cooperation of distribution network operators [3]. In recognition of this, two incentive schemes aimed at encouraging DNOs to connect more DG have been introduced in the UK. The first one is called Innovation Funding Incentive (IFI) [4]. This scheme is intended to provide funding for projects focused on the technical development of distribution networks to deliver value (i.e. financial, supply quality, environmental, safety) to end consumers. IFI projects can embrace any aspect of distribution system asset management from design through to construction, commissioning, operation, maintenance and decommissioning. A DNO is allowed to spend up to 0.5% of its Combined Distribution Network Revenue on eligible IFI projects.
[3]
[4]
G Strbac, N. Jenkins, M. Hird, P. Djapic, G. Nicholson, "Integration of operation of embedded generation and distribution networks”, A Project report, 2002, Available: http://www.sedg.ac.uk/ Clare Louise Masters, Joseph Mutale, Goran Strbac, Srdjan Curcic, Nicholas Jenkins, “Statistical Evaluation of Voltages in Distribution Systems with Embedded Wind Generation”, IEE proceedings on Generation, Transmission and Distribution, 2000 Martin Scheepers, Michiel van Werven, Joseph Mutale, Goran Strbac, and David Porter, “Distributed Generation in Electricity Markets, its impact on Distribution System Operators, and the role of Regulatory and Commercial Arrangements”, International Journal of Distributed Energy Resources, January-March 2006, Issue Number 1, Volume 2 Gareth Evans, “Innovation Funding Incentive (IFI) and Registered Power Zones (RPZ)”, 2005, Available: http://www.ofgem.gov.uk/ofgem/shared/template2.jsp?id=10498
VIII. BIOGRAPHY Joseph Mutale (M’96) is a Senior Lecturer at the University of Manchester in the Electrical Energy and Power Systems Group. He holds a B.Eng. in Electrical Machines and Power Systems obtained in 1981 from the University of Zambia, an MSc in Electric Power Transmission and Distribution and a Ph.D in Power System Economics from the University of Manchester obtained in 1987 and 2000 respectively. His research interests are in pricing of transmission and distribution networks, technical and economic integration of distributed energy resources into power systems as well as rural electrification
606
