Beruflich Dokumente
Kultur Dokumente
Research Article
Abstract: The impact of distributed generation (DG) units on the voltage stability has become a challenging issue especially
when squirrel cage induction generator (SCIG)-based wind DGs are utilised. Optimisation methods are tools which can be used
to place and size the DG units in the distribution system, so as to utilise these units optimally within certain constraints. This
study aims to optimally size and allocate advanced wind energy based DG technology with innovative reactive power capability,
reduced capital cost, and improved energy capture capability to improve voltage stability. Therefore, a new combination of SCIG
and doubly-fed induction generator (DFIG) based DG configuration is proposed. In this configuration, the reactive power
absorbed by SCIG is supplied by DFIG, and therefore, the combined system operates at unity power factor, which makes it
feasible to comply with the IEEE 1547 standard. A methodology is proposed to optimally size and allocate the DG system with
an objective function to improve the voltage profile considering numerous technical and economic constraints. The performance
of the proposed DG configuration is compared with DGs that utilise SCIG with a parallel reactive power compensation. IEEE 30-
bus test system is used to demonstrate the effectiveness of the proposed methodology.
IET Renew. Power Gener., 2017, Vol. 11 Iss. 3, pp. 281-288 281
© The Institution of Engineering and Technology 2016
N
PSCIG = ∑ PSCIG i
(3)
i=1
M
PDFIG = ∑ PDFIG j
(4)
j=1
282 IET Renew. Power Gener., 2017, Vol. 11 Iss. 3, pp. 281-288
© The Institution of Engineering and Technology 2016
the largest and smallest values of ∂Pi /∂V i is the strongest and
weakest bus, respectively.
PDG
PF = (10)
PSystem
Nowadays, utility-scale onshore wind turbine sizes are 1.5 5 MW ∑nN= 1 V nprn
[16]. In this study, a 2 MW wind turbine size is adopted for both V Index = (12)
4
SCIG and DFIG. The total size of the integrated wind turbine-
based DG can be calculated by multiplying the number of selected
where prn is the probability density function, and it is given by
ratings. Fig. 2 shows flowchart of the algorithm for the proposed
DG sizing and placement. The detailed descriptions of each step
are given below. 1 −(((x − μ))/2σ2)2
prn = e (13)
σ 2π
4.1 Step 1: selection of candidate bus
μ and σ2 are the calculated mean and variance of the assumed
The candidate bus is selected using a voltage stability index that loads, respectively; N is the number of load scenarios; x is a
tests the voltage sensitivity to active power variation. The active random variable that is substituted by the load bus because a
power sensitivity index can be briefly derived based on the Newton probabilistic load flow that accounts for the load uncertainty is
Raphson power flow as follows [9]: adopted in this study. The highest voltage index indicates the
most improved voltage profile. The impact of DG installation
ΔP J 1J 2 Δδ can be evaluated based on the following three possible
= (9)
ΔQ J 3J 4 ΔV scenarios:
where J 1 and J 2 are submatrices in the Jacobian matrix that i. V Index < 1, DG units will worsen the voltage profile.
expresses the change in active power P with respect to power angle ii. V Index = 1, DG units will have no impact on the voltage
δ and bus voltage V. J 3 and J 4 define the change in reactive power profile.
with respect to the power angle and bus voltage, respectively. By iii. V Index > 1, DG units will improve the voltage profile.
assuming that the reactive power variation at a certain load bus is
zero, the voltage profile only becomes proportional to the active The voltage stability index measures the ability of DG units, for
power variation. Therefore, it is acceptable to set J 1 and J 4 of the given uncertain load conditions, to maintain the voltage profile
Jacobian matrix to zero. The diagonal elements of J 2 indicate the stable within the target range.
active power sensitivity of the ith bus. The bus corresponding to
IET Renew. Power Gener., 2017, Vol. 11 Iss. 3, pp. 281-288 283
© The Institution of Engineering and Technology 2016
4.4 Step.4: adding DFIG units to SCIG 1
Pout = C ρAv3 (18)
2 P
In this step, DFIG units are added to the existing SCIG units when
the voltage index is <1.0. The combined DG units will operate at
unity power factor and supply pure active power to the candidate where ρ is the air density (kg/m3), A is the rotor swept area (m2), v
bus to which it is connected. The number of installed DFIG units is the wind speed (m/s), P is the turbine power in kW, and CP is the
depends on the reactive power requirements of SCIG. The reactive power coefficient.
power requirements of SCIG depend on its designated operating
power and penetration factors. The reactive power capability of 5 Results and discussion
DFIG is limited by both its operating power factor and penetration
level. In this study, operational power factor of 0.9 is adopted for A modified IEEE 30-bus system is used as a test system [20]. Bus
the DFIG units. While various lagging power factors over a range 1 is a slack bus with constant voltage of 1.05 pu. Buses 2, 5, 8, 11,
of 0.9 lagging and 0.98 lagging are tested for sizing the proposed and 13 are PV buses, generating total active power of 161 MW and
DG. The optimal size of the proposed DG is the smallest size that reactive power of 70.7 MVar. The rest of the system buses are load
gives a voltage index and candidate bus voltage higher than 1 and buses (PQ buses). The system has total active load of 366 MW and,
0.95 pu, respectively. The flowchart in Fig. 2 shows how the sizing total reactive load of 176.8 MVar, respectively. Fig. 3 shows the
algorithm tests a variety of SCIG and DFIG combinations to single-line diagram of the IEEE 30-bus test system.
determine the optimal DG size based on both technical and
economic considerations. 5.1 Selection of candidate bus for DG units installation
Voltage stability analysis using the ∂Pi /∂V i index is performed in a
4.5 Step.5: conduct cost analysis
MATLAB environment. Table 1 shows the voltage stability
The total cost of the SCIG-based DG is evaluated by adding the analysis results. Bus 26 is the weakest bus and has the smallest
cost of the SCIG units and the parallel compensation cost. The index value. The corresponding voltage magnitude at bus 4 is 0.93
developed algorithm compares the capital installation cost of the pu, which is below the voltage stability limit of 0.95 pu. Therefore,
proposed DG to the cost of SCIG with compensation. Thus, the DG units are installed at bus 26 to improve the voltage profile to
objective of minimising the cost of the DG can be formulated as the desired limit. The most weak buses in the system are ranked in
a manner where the bus with the smallest ∂Pi /∂V i is ranked first.
CostSCIG + DFIG = (PtotalksPF + PtotalkDPF ) (14)
S D
5.2 Installation of SCIG-based DG
CostSCIG = (PtotalksPF + MQCap) (15)
S The SCIG-based DG is installed at candidate bus 26. The active
load demand at bus 26 is 8.5 MW, which is the highest MW
where Ptotal is the total load of the system, ks is the capital cost of a demand in the system. The corresponding bus voltage is 0.93 pu.
single SCIG unit in $/kW, kD is the capital cost of a single DFIG When the SCIG unit generates the maximum 8 MW (4 SCIG units
unit in $/kW, PF is the penetration factor of SCIG units, PF is the × 2 MW) at a lagging power factor of 0.953 (2.2% penetration
S D
factor), it requires 2.7 MVar from the grid. The bus 26 voltage
penetration factor of DFIG units, QCap is the reactive power moves to 0.936 pu which is not within the permissible limits (0.95
injected by the parallel compensation in kVar, and M is the cost of pu < V < 1.05 pu). The installed SCIGs at bus 26 absorb reactive
reactive power compensation in $/kVar. Minimising the capital cost power and increase the reactive demand on bus 26 from 5 MVar to
of the proposed DG is another objective function that must satisfy 8.2 MVar. The increase in the reactive demand deteriorates the
the following condition: voltage stability at bus 26. Table 2 shows the impact of the SCIG-
based DG on bus 26 voltage.
CostSCIG + DFIG ≅ CostSCIG (16) It is clear that, relatively low lagging power factor of SCIG
results in an increased value of the absorbed reactive power and
The cost of the proposed DG can be minimised by sizing DG with total reactive demand at the candidate bus. Consequently, the
a minimum number of SCIG and DFIG units. The average capital voltage magnitude at the candidate bus decreases.
installation cost of a 2 MW scale wind turbines is roughly $3–$4
million. The National Renewable Energy Laboratory estimated the 5.3 Installation of SCIG-based DG with parallel reactive
capital installation cost of 1.5 MW unit in $/kW from $1400 to power compensation
$2900 [17]. In addition, the cost of DFIG is 10% higher than that
of SCIG [18]. In this study, the installation cost of a single SCIG- A capacitor bank, SVC, or static synchronous compensator
based DG and DFIG is estimated using installation cost of 1500 (STATCOM) can be used to supply the reactive power
and 1650 $/kW, respectively, as (see equation below) requirements of the SCIG and improve the DG power factor to
unity. Table 3 summarises the results for different DG sizes at
different lagging power factors and corresponding bus 26 voltages.
4.6 Step.6: DG output power estimation
The reactive power requirement of the SCIG-based DG with sizes
The performance of the wind turbine can be characterised by a of 4, 6, and 8 MW are listed in Table 3 for various operating power
power curve. The power curve plays a vital role in predicting the factors.
output power of a wind turbine for a given wind speed [19]. A DG The reactive power absorbed by the DG varies greatly with the
typically consists of multiple identical wind turbines that can be DG size and operating lagging power factor. A DG with lower
modelled with an equivalent power curve. The developed power factor and larger MW capacity draws more reactive power.
algorithm compares the output power of the proposed DG to that of All the proposed sizes in Table 3 have enhanced bus 26 voltage to
the SCIG-based DG at a given wind speed as follows: the desired limit (> 0.95 pu). However, the capital cost plays a vital
role in selecting the optimal size. The capital cost increases with
PSCIG + DFIG > PSCIG (17) the DG size and the required parallel compensation. However, the
limited ability to efficiently extract power at variable wind speed is
The DG output power is calculated as a main drawback of the SCIG-based DG.
6 Conclusion
This paper presented a method to optimally size and allocate
advanced wind energy based DG technology with innovative
reactive power capability to improve voltage stability. A new DG
configuration based wind energy is proposed. The proposed DG
configuration combines both SCIG and DFIG units. The reactive
power absorbed by SCIG is supplied by DFIG (self VAR control),
Fig. 7 Aggregated power curves of SCIG-based DG with compensation and therefore, the combined system operates at unity power factor,
and the proposed DG configuration which makes it feasible to comply with the IEEE 1547 standard.
The sizing and allocation method has been extensively simulated
Table 6 Capital installation cost of SCIG-based DG with for an IEEE 30 bus test system. Simulation results indicate that the
compensation proposed optimisation method can provide a framework for the
DG, Power SCIG SCIG SCIG sizing and allocation of the introduced DG in the electrical
MW factor of +capacitor, $ +SVC, $ +STATCOM, $ distribution system to achieve the target of improving the voltage
SCIG millions millions millions stability.
Based on a probabilistic load flow that accounts for load
4 0.900 6.030 6.074 6.104
variations, a voltage index is used to analyse the impact of the DG
4 0.951 6.025 6.065 6.092 on the voltage profile. The impact of integrating SCIG, SCIG with
6 0.900 9.057 9.120 9.195 compensation, and the proposed DG on the voltage stability has
6 0.971 9.051 9.100 9.183 been investigated using the developed voltage index. The voltage
8 0.900 12.056 12.143 12.186 index is calculated for low and high wind speeds of 6 and 12 m/s,
8 0.987 12.025 12.063 12.088 respectively. Simulation results show that the highest voltage index
value is obtained when the proposed DG is installed which gives
the best voltage improvement case. This is due to a substantial
Table 7 Capital installation cost of proposed DG saving in reactive power and the capability of the proposed DG to
extract more active power lower wind speeds. It was also
DG, No. of SCIG No. of DFIG SCIG + SCIG concluded that at higher DG sizes, the capital cost of the proposed
MW units units DFIG, $ units pf DG becomes more competitive with SCIG due to the increase in
millions the compensation cost of SCIG. Furthermore, this study indicated
4 1 unit (2 MW) 1 unit (2 MW) 6.3 0.900 that the proposed DG configuration incorporates the attractive
6 2 units (4 MW) 1 unit (2 MW) 9.3 0.971 advantages of both variable and fixed speed wind systems. This
8 3 units (6 MW) 1 unit (2 MW) 12.3 0.987 work will be specifically useful to electric power utilities to strive
towards increased penetration of wind power generation.
7 References
[1] Raja, P., Selvan, M.P., Kumaresan, N.: ‘Enhancement of voltage stability
margin in radial distribution system with squirrel cage induction generator
based distributed generators’, IET Gener. Transm. Distrib., 2013, 7, (8), pp.
898–906
[2] Hrishikesan, V.M., Venkatraman, K., Selvan, M.P.: ‘Performance of custom
power devices in SCIG based wind farms during abnormal grid conditions’.
2014 Annual IEEE India Conf. (INDICON), Yashada, India, 11–13 December
2014, pp. 1–5
[3] Dahraie, M.V., Najafi, H.R., Ebadian, M.: ‘Analytical investigation of the
effect of wind farm equipped with SCIG on voltage stability’. Second Iranian
Conf. on Renewable Energy and Distributed Generation (ICREDG), Tehran,
Iran, 6–8 March 2012, pp. 121–126
[4] IEEE standard for distributed resources interconnected with electric power
systems, IEEE P1547 Std, 2002
[5] Hedayati, H., Nabaviniaki, S.A., Akbarimajd, A.: ‘A method for placement of
Fig. 8 Cost variation for SCIG-based DG with compensation and DG units in distribution networks’, IEEE Trans. Power Deliv., 2008, 23, pp.
proposed DG configuration 1620–1628
[6] Alonso, M., Amaris, H.: ‘Voltage stability in distribution networks with DG’.
Proc. of IEEE PowerTech, Bucharest, Romania, 2009, pp. 1–6
It is clear that as the DG size increases (higher penetration [7] Al Abri, R.S., El-Saadany, E.F., Atwa, Y.M.: ‘Optimal placement and sizing
factor), the capital cost increases accordingly. In addition, the cost method to improve the voltage stability margin in a distribution system using
distributed generation’, IEEE Trans. Power Syst., 2013, 28, (1), pp. 326–334
is higher when SCIG units operate at lower lagging power factor [8] Kroposki, B., Sen, P.K., Malmedal, K.: ‘Optimum sizing and placement of
due to the increase in their MVar requirements. This is because of distributed and renewable energy sources in electric power distribution
the fact that the reactive power compensation cost mainly depends systems’, IEEE Trans. Ind. Appl., 2013, 49, (6), pp. 2741–2752
on the amount of reactive power supplied at a connected bus ($/ [9] Hung, D.Q., Mithulananthan, N.: ‘Multiple distributed generator placement in
primary distribution networks for loss reduction’, IEEE Trans. Ind. Electron.,
kVar) [21]. At lower power factor, the reactive power amount is 2013, 60, (4), pp. 1700–1708
higher resulting in an increase in the total VAR compensation cost. [10] Hong, Y.-Y., Luo, Y.-F.: ‘Optimal VAR control considering wind farms using
Table 7 presents the capital cost results of the proposed DG. The probabilistic load flow and gray-based genetic algorithm’, IEEE Trans. Power
results show the cost for various DG sizes and lagging power Deliv., 2009, 24, (3), pp. 1441–1449
[11] Khushalani, S., Solanki, J.M., Schulz, N.N.: ‘Development of three-phase
factors of SCIG units, and the power factor of DFIG units is kept unbalanced power flow using PV and PQ models for distributed generation
constant at 0.9 leading. As the of DG size increases, the capital cost and study of the impact of DG models’, IEEE Trans. Power Syst., 2007, 22,
increases accordingly. Furthermore, the cost is proportional to the (3), pp. 1019–1025
number of DFIG units which depends on the VAR demand of [12] Li, H., Chen, Z.: ‘Overview of different wind generator systems and their
comparisons’, IET Renew. Power Gener., 2008, 2, (2), pp. 123–138
SCIG units. [13] Muljadi, E., Singh, M., Gevorgian, V.: ‘Doubly fed induction generator in an
Fig. 8 shows the variation of the capital installation cost with offshore wind power plant operated at rated V/Hz’. Energy Conversion
changes in the DG size. It can be concluded that at higher DG Congress and Exposition (ECCE), Raleigh, NC, USA, 15–20 September
2012, pp. 779–786
IET Renew. Power Gener., 2017, Vol. 11 Iss. 3, pp. 281-288 287
© The Institution of Engineering and Technology 2016
[14] Guerrero, R.: ‘Grid code interrelation, wind generation evaluation and Electrical Engineering, Eindhoven University of Technology, Barcelona,
reactive compensation. Special topics inside a grid code’. Integration of Spain, 2013
Renewables into the Distribution Grid, (CIRED Workshop), Lisbon, Portugal, [19] Wan, Y.H., Ela, E., Orwig, K.: ‘Development of an equivalent wind plant
29–30 May 2012, pp. 1–4 power curve’. NREL Technical Report, NREL/CP-550-48146, 2010
[15] Glover, J.D., Sarma, M.S., Overbye, T.J.: ‘Power system analysis and design’ [20] Abuella, M.A.: ‘Study of particle swarm for optimal power flow in IEEE
(Cengage Learning, Stamford, CT, USA, 2011, 5th edn.) Benchmark systems including wind power generators’. MS thesis,
[16] Muljadi, E., Pasupulati, S., Ellis, A., et al.: ‘Method of equivalencing for a Department of Electrical and Computer Engineering, Southern Illinois
large wind power plant with multiple turbine representation’. IEEE PES University Carbondale, Carbondale, IL, USA, 2012
General Meeting, Pittsburgh, PA, USA, 20–24 July 2008, pp. 1–9 [21] Li, F.F., Kueck, J., Rizy, T., et al.: ‘A preliminary analysis of the economics of
[17] Tegen, S., Lantz, E., Hand, M., et al.: ‘2011 cost of wind energy review’. using distributed energy as a source of reactive power supply’. First Quarterly
NREL Technical Report, NREL/TP-5000-56266, 2013 Report, The U.S. Department of Energy, 2006
[18] Aydin, E.: ‘Determination of best drive train technology for future onshore
wind turbines as a function of the output power’. MSc thesis, Department of
288 IET Renew. Power Gener., 2017, Vol. 11 Iss. 3, pp. 281-288
© The Institution of Engineering and Technology 2016