Flexible Voltage Control to Support Distributed Generation in Distribution Networks Maciej Fila
Gareth A. Taylor
Jonathan Hiscock
Brunel University/EDF Energy
[email protected]
Brunel University - Brunel Institute of Power Systems
[email protected]
Fundamentals Ltd
[email protected]
Malcolm R. Irving
Peter Lang
Brunel University - BIPS
[email protected]
EDF Energy
[email protected]
Abstract - Increasing penetration of Distributed Generation (DG) in distribution networks significantly changes both the real and reactive power flows in the network and can create serious voltage control problems. Furthermore, traditional Automatic Voltage Control (AVC) schemes that can normally deal with the reverse power flows, are unable to cope with the voltage problems associated with the presence of DG under certain conditions. Several techniques have been deployed to improve distribution network voltage profiles in such cases, for instance network reinforcement or active voltage control with remote voltage sensing units. Another method that has been proposed and recently implemented is the SuperTAPP n+ relay scheme that is based on locally taken measurements at the substation level combined with a state estimation technique. Such an approach enables assessment of the voltage rise at the point of connection of DG and effective control of the voltage level at the substation. The first objective of this paper is to present the fundamental principles of innovative voltage control schemes for distribution networks with DG based on locally measured signals. Secondly the functionality of the most flexible scheme will be demonstrated via software simulation for a range of distribution network case studies based upon realistic EDF Energy network scenarios. Finally, the results from the modelling and analysis of the SuperTAPP n+ relay scheme and its feasible application will be discussed and detailed conclusions are presented. Keywords: advanced voltage control, distributed generation, TAPP scheme, distribution network, voltage control relay I. INTRODUCTION Modern distribution networks face numerous challenges. A key one is the cost-effective accommodation of distributed generation (DG). Distribution networks were originally designed to deliver electricity from grid supply points (GSP) to numerous small and medium sized customers in a reliable, efficient and statutory specified manner. In order to meet current and future government targets growing amounts of generation from renewable energy sources (RES) is being connected to the power network. The majority of the RES is connected at medium and low voltages and classified as DG. Increasing penetration of such DG is changing the original character of the distribution networks from uni-directional to
bi-directional power flow and forces distribution network operators (DNOs) to take new approaches to network management. One of the key issues while accommodating DG is voltage rise at the point of connection. Due to unacceptably high voltage levels new applications for DG are often suspended or the generation capacity limited. In the section II of this paper the statutory requirements for the voltage level in distribution networks are discussed and the theoretical background of voltage change in distribution networks is analysed. This section also presents and examines currently available and potential solutions to counteract the voltage rise effect caused by the presence of DG. Section III presents the principles of an innovative technique for flexible voltage control to support DG in distribution networks. The SuperTAPP n+ scheme functionality and performance are examined and case study results based on realistic EDF Energy networks are presented and discussed in Section IV. Finally conclusions are stated in section V. II. VOLTAGE RISE EFFECT ON DISTRIBUTED NETWORKS WITH DG A.
Regulatory requirements for voltage profile in distributed network
DNOs are obliged by the Electricity Safety, Quality and Continuity Regulations (ESQC) to maintain the voltage profile across the networks within statutory limits. The steady-state voltage magnitude for the HV networks (between 1 kV and 33 kV) should be kept with a tolerance of +/– 6% of the nominal voltage and for the LV networks within a tolerance of +10/–6 % [1]. Any change in the network, such as additional load or new connection of DG, changes the voltage profile of the network. The responsibility of DNO is to make sure that this system is still operated within permitted limits. Before new generation is connected, its effect on the network and the voltage profile must be investigated. The network operator usually considers the following worst case scenarios: - maximum output of the generator under minimum load demand
- maximum load demand and no generation - maximum load demand and maximum generation Under all of these conditions the voltage profile of the network must remain within statutory limits. If this criterion is not satisfied, one of the standard passive techniques or recently proposed active schemes can be used to deal with excessive voltage rise. B.
Voltage profile on distributed networks with DG
The voltage variation along the feeder might be described by the following equation (1) ΔV = (R + jX )I = (R + jX )
P − jQ V
(1)
By modifying equation (1) and neglecting the imaginary term, it can be seen that the voltage at a remote point on the feeder, V2, is governed by source voltage V1, impedance of the line R+jX and active and reactive power flows P and Q [2]. V2 ≈ V1 +
R⋅ P + X ⋅Q V1
(2)
Considering DG connected on the feeder, and the associated active and reactive power output, PG and QG respectively, equation 2 can take the following form: V2 ≈ V1 +
R ⋅ (PG − PL ) + X ⋅ (± QG − QL ) V1
(3)
Theoretically, equation 3 describes how the voltage rise effect at the point of connection of the DG can be overcome. The voltage V2 can be reduced by the following actions: - reducing source voltage V1 - decreasing R and/or X - changing one of the terms R.P or X.Q or both C.
Methods to overcome voltage rise effect caused by DG
There are several proposed solutions to improve the voltage profile in distribution networks with DG. Solutions such as network reinforcements; line re-conducting, building a dedicated line are currently available and used in distribution networks. Active voltage control with remote voltage sensing units (such as GenAVC [3]), line voltage regulation, scheduling of DG and several other active techniques have became recently become available or are still in the development and trial stages. All of these methods can also be applied in combination to increase the capability of the network to accommodate DG. One of the passive solutions to reduce voltage rise is network reinforcement. By upgrading the conductor size of the feeder to accommodate DG, the resistance and reactance are reduced. That means that both terms in equation 3, R.(PGPL) and X.(±QG-QL) are smaller under the same load and generation conditions and therefore the voltage rise is effectively lessened. This method provides satisfactory
increase in the availability of the connection and also improves power quality, reduces losses and customer interruptions. However implementation of this solution is associated with significantly high costs. A common technique that is employed to manage voltage profiles in transmission networks is generator reactive power control. In this technique, the voltage rise caused by generator real power output P, and consequently the increase of term R.(PG-PL) in Eq. 3, is compensated for by absorbing reactive power -Q and hence reducing term X.(±QG-QL). This technique maintains desirable voltage levels at the point of connection of the generator. Due to the fact that the X/R ratio in distribution networks is typically much smaller than in transmission networks this technique is not efficient for 33 and 11 kV cable feeders. It might provide reasonable effectiveness of voltage control when a generator is connected to overhead lines, due to the fact that its reactance is 4 times higher than a cable. However it can place heavy demands on the generator to absorb reactive power. This may cause additional operational costs for the generator, lead to increase in losses in the network and adversely affect AVC performance [4]. Managing the voltage rise in distribution networks by reactive compensation can be justified in some cases, nevertheless its application is limited and it is common practise for DNOs to instruct distributed generator to operate at or close to unity power factor. Under minimum load conditions a generator output might significantly surpass demand on the feeder (PG >> PL), causing unacceptable voltage rise. Despite the fact that this extreme situation, minimum load and maximum generation, needs to be considered before DG is connected to the network, it occurs rather occasionally and in specific operational time periods. Therefore it may be beneficial to connect a larger generator and restrict its output when voltage level at the point of connection rises above acceptable limits. However, it is important to note that this technique has significantly negative economic and technical implications, particularly for the generators and may be challenging in reality for a number of generators. Voltage V2 at the point of connection of the DG can be decreased by reducing voltage V1 at the primary substation. This coordinated voltage control scheme actively manages AVC voltage set point, taking into account voltage level at the DG and overall voltage profile on the network. A commercially available system that uses this technique is GenAVC. The GenAVC system operates by taking voltages and power flow measurements at the substation level as well as additional strategic locations across the network via communications links with remote terminal units (RTUs). State estimation is used to evaluate voltages throughout the network and desirable voltage targets are determined for the primary transformer AVC [3]. GenAVC is a very effective scheme to maximise generator export while maintaining voltage profile within statutory and operational limits. However, considerable costs are also associated with the implementation and operation of this scheme, but even more importantly costly reinforcement of the network can be avoided.
This value corresponds to necessary voltage reduction at the substation in order to bring the voltage level at the point of connection of DG within statutory limits. An additional advantage of this method is the improved performance of load drop compensation (LDC). In order to eliminate the error in the LDC performance caused by DG, the generator current is removed from the summation of transformer currents ITL. I LOAD = (ITL − I G ) = (ITL − I FG ) ⋅ (1 + EST )
III. VOLTAGE CONTROL ON THE NETWORK WITH DG BASED ON LOCAL MEASUREMENTS It appears that a remote terminal unit (RTU) with a communications link to the AVC is the only way to provide information regarding the generator output and voltage level. Hence, enabling AVC schemes to reduce the associated voltage target according to network requirements. However, it is important to note that by using only measurements at the substation level and resemblance of the load pattern on each feeder, in certain cases similar results can also be achieved. Figure 1 shows a system arrangement where two parallel transformers are controlled by SuperTAPP n+ relays. The innovative technique employed in the SuperTAPP n+ relay is the ability to estimate output of the generator which is connected at a remote point on the feeder. This is accomplished by the additional current measurement IFG on the feeder with DG and ratio EST which represents the load share between feeders with embedded generation to those without generators [7].
The LDC voltage boost calculation is based on the true load current ILOAD, not on the total transformer current as is done in a standard AVC relay [6]. Due to the fact that both generator compensation bias and the LDC voltage bias calculation are based on the EST ratio, any deviation of this factor introduces an error in the AVC target voltage. Figure 2 shows an example of the load profile on an 11 kV substation over a week. The bottom line on the plot represents load current on the feeder prior to the connection of the DG. The line in the middle represents the total load on the substation. Under ideal circumstances the load on each feeder, and consequently the load on the substation, follow the same pattern and the ratio EST is constant. In real systems there are some discrepancies in the load current ratio on the feeders which are reflected in fluctuation of the factor EST. The magnitude of variation in EST is represented by the top line on figure 1 with the values on the right-hand axis.
load on the feeder with generator I I FG (3) = 1 = load on the feeders without generators I 2 ITL − I FG
The factor EST is calculated prior to the connection of DG or when its output is zero. During operation of the DG this ratio is used to estimate generator output as follows: I G = (E ST ⋅ (I TL − I FG )) − I FG
(4)
Total Substation Load
1600
Load on Feeder with DG
Est
0.45 0.40
1400
Load Current (Amps)
EST =
(6)
0.35
1200
0.30
1000
0.25 800
0.20 600
0.15
400
0.10
200 0 15-Feb
Est ratio
Another innovative active scheme, based on coordinated voltage control, is SuperTAPP n+. Firstly, it offers very effective AVC performance based on the Enhanced TAPP algorithm [5,6] and secondly an innovative technique for voltage control in the distribution network with DG. The key benefit of this scheme is that all measurements are taken or estimated locally and there is no need for remote communication with the generators.
0.05 0.00 16-Feb
17-Feb
18-Feb
19-Feb
Time
20-Feb
21-Feb
22-Feb
Fig. 2. Load profile on the substation and EST ratio fluctuation.
Fig. 1. SuperTAPP n+ relay arrangement.
Knowing the current output of the generator IG, the voltage rise at the point of connection can be evaluated and appropriate generator compensation bias (buck) can be applied to the AVC. The generator compensation bias is calculated in reference to the voltage rise at the maximum generator current IGMAX as shown in equation 5. VG = VGMAX % ⋅
IG I GMAX
(5)
In the example above the average EST ratio is 0.35 with the deviation approximately +/–20%. This range of variation in the EST ratio is consistent in the networks but it is influenced by the character of the load. For the feeders with the majority of the residential customers it tends to be less significant, whereas for the feeders with a considerable amount of commercial load, variation in EST has a tendency to increase. The research showed that under the latter condition it might rise to 35% off the average value. Another factor affecting generator voltage bias calculations is the amount of generation on the feeder. When the amount of DG is a small proportion of the feeder load, the change in EST, caused by DG, is also low. Any deviation in the load
ratio from the set value introduces a significant error. As the DG share in the feeder load increases, the error in the voltage bias calculation reduces. The curves in figure 3 correspond to various DG to load ratios and the generator voltage bias errors for a range of EST deviation. 90 10%
80
20%
Voltage bias Error %
70
30%
60
40%
50
50% 60%
40
80% 100%
30
130% 170% 200%
20 10 0 0
5
10
15
20
25
30
35
40
45
50
Est deviation %
Fig. 3. Voltage Bias Error in respect to generator to load ratio on the feeder and EST deviation
The voltage rise effect becomes a problem in the network when the term RP+XQ in equation 2 becomes positive. For DG operating at unity power factor, this situation takes place when the generator output exceeds demand on the feeder which causes the generator output to load ratio to become greater than 100%. Taking into consideration the EST variation in the range of 20%, the maximum error introduced in the SuperTAPP n+ performance does not exceed 18%. Further improvement in the performance can be achieved by attentive setting of the EST ratio in the relay. When SuperTAPP n+ is going to be used to increase the ability of the network to accommodate DG, detailed investigation of the network characteristics must be carried out. Several factors such as load characteristics of the feeders, variation of EST, generator output and required generator and LDC voltage bias need to be considered to guarantee effective and secure performance of the SuperTAPP n+ relay. IV. CASE STUDY - SOFTWARE SIMULATION AND RESULTS The following case study is used to demonstrate simulation results of the performance of the SuperTAPP n+ scheme in a realistic EDF Energy network. Figure 4 shows a one-line diagram of the 132/11 kV radial network. The load of the substation is supplied by two parallel 30 MVA transformers. On one of the 15 feeders DG with a total capacity of 5 MW is connected. Due to the unacceptable voltage increase for the period of low demand the total export of the DG has been limited to 3.8 MW.
Fig. 4. Distribution network diagram with the 5 MW of DG.
Functionality of the SuperTAPP n+ scheme has been implemented into the Operation and Control of Electrical Power Systems (OCEPS) software [8]. Together with the model of the network described above, this was used to analyse the opportunity to increase the output of the DG and improve the voltage profile in the system by using SuperTAPP n+ to control the voltage at the 11 kV bus-bar. From the analysis of the historical load data on the feeder and substation it was observed that EST varies in the range of 0.18 to 0.25. Through the software simulation it was identified that the best performance of the SuperTAPP is achieved with the setting EST=0.22. Furthermore, to obtain an effective voltage target, the voltage generation bias was set to provide 2% attenuation while the generator is exporting maximum power and 2% LDC boost under maximum load conditions. Figures 5 and 6 represent the load profile of the feeder with DG and the feeder with the highest voltage drop for the two worst case scenarios; minimum load-maximum generation and maximum load-maximum generation respectively. For the minimum load and maximum generation condition, the voltage level at the point of connection rises above the statutory limit to 11.74 kV with the current AVC scheme (master-follower). It can be observed that there is voltage headroom in the network to reduce the voltage at the primary substation and bring the voltage on the feeder with DG back within the statutory limit. When the SuperTAPP n+ scheme is in use the AVC voltage target is reduced by 2% and the voltage at the point of connection of DG is decreased to 11.6 kV and within the limit.
Standard AVC
CONCLUSIONS
SuperTAPP
11.80 11.70 Upper Voltage Limit
11.60 11.50 11.40
Voltage (kV)
11.30 11.20 11.10 11.00 10.90 10.80 10.70 10.60 10.50 10.40
Lower Voltage Limit
10.30 1
6
11
16
21
26
31
36
41
46
51
56
61
Substation
Fig. 5. Load profile for minimum load and maximum generation.
Under heavy load conditions it is important to keep the voltage at the primary substation reasonably high to make sure that customers’ voltage at the end of the feeders is above the minimum voltage limit. As can be seen in figure 6 the voltage magnitude at the 11 kV busbar is kept the same for both AVC schemes. This is due to the fact that generation voltage bias is overset by LDC voltage bias and the effective voltage target for SuperTAPP n+ is not changed. Under such conditions the voltage profile on the feeder with generation remains below the permitted +6% limit and on the heaviest loaded feeder the voltage remains well above the -6% limit. Standard AVC
SuperTAPP
11.80 11.70
Upper Voltage Limit
11.60 11.50 11.40
Voltage (kV)
11.30 11.20 11.10 11.00 10.90
This paper presents the fundamental theory behind voltage rise effect in the distribution network and considers a number of possible solutions to counteract this issue. Furthermore review of up-to-date techniques to increase ability of the network to accommodate DG is discussed, considering commercial and technical challenges. Functionality of the novel SuperTAPP n+ scheme is investigated. Due to the fact that this method is based on estimation techniques and assumptions on the load pattern it is susceptible to certain error. Examination of performance and factors affecting the scheme showed that in common network circumstances the maximum error does not exceed 20% which is an acceptable value. Therefore this scheme is feasible and an effective solution to control the voltage level in a distribution network with significant penetration of DG, despite requiring a study of the network configuration, load patterns, and voltage profile. The main advantages of SuperTAPP n+ are simple installation and configuration and no need for communications links which imply very low implementation and operational cost of the scheme. An additional benefit of the scheme is the simplicity of accommodating additional DG on the network. By installing an extra current measurement on the feeder with new DG and updating relay settings, SuperTAPP n+ can control the voltage profile on the network with a number of DGs with at least one reference feeder without DG. This paper provides evidence that the OCEPS software with an advanced AVC and DG functionality is an excellent tool for the analysis of the potential application of SuperTAPP n+ in a variety of the distribution networks. On the basis of the numerical results for the specific system it is demonstrated that the software can be used to determine network specifications and the accurate settings for the relay to accommodate DG in the constrained system.
10.80
REFERENCES
10.70 10.60 10.50 10.40
Lower Voltage Limit
10.30
1
6
11
16
21
26
31
36
41
46
51
56
61
Substation
Fig. 6. Load profile for maximum load and maximum generation.
The simulation was performed for a range of load conditions and a maximum generation output of 5 MW, taking into consideration the highest EST deviation. Under all circumstances the voltage profile in the network was maintained within statutory limits. It is demonstrated that with the provision of voltage generation bias and LDC voltage bias in the SuperTAPP n+ relay, the maximum generation might be increased from the restricted 3.8 MW to the maximum capacity of the DG of 5 MW.
[1] The Electricity Safety, Quality and Continuity Regulations 2002, http://www.berr.gov.uk [2] C.L. Masters, “Voltage rise the big issue when connecting embedded generation to a long 11 kV line”, Power Engineering Journal, February 2002 [3] “GenAVC technical specification”, www.econnect.com [4] G. Strbac, N. Jenkins, “Integration of Operation of Embedded Generation And distribution Networks”, DTI report Number K/EL/00263/REP, URN Number 02/1145, 2002 [5] M. Fila, G.A Taylor, J. Hiscock, “Systematic modelling and analysis of TAPP voltage control schemes”, 42nd International Universities Power Engineering Conference, UPEC 2007, pp. 327-334, Brighton, UK, 4-6 September 2007 [6] M. Fila, G.A Taylor, J. Hiscock, “Modelling and Analysis of Enhanced TAPP scheme for distribution networks”, 16th Power Systems Computation Conference, PSCC 2008 (accepted) [7] "Technical Specification for Advanced Voltage Control Relay SuperTAPP n+”, Fundamentals Ltd, 2006. [8] M.R.Irving, M.J.H.Sterling, “Efficient Newton-Raphson algorithm for load flow solution in transmission and distribution networks”, Proc IEE,C,134,5,1987,pp325-328.
