Distributed Generators Placement for Loadability Enhancement based on Reactive Power Margin

Tareq Aziz T. K. Saha N. Mithulananthan

School of Information Technology and Electrical Engineering, The University of Queensland, Qld 4072, Australia e-mail address: { taziz, saha, mithulan}@itee.uq.edu.au
Abstract This paper proposes a simple methodology for placing two principal types of DG units - synchronous machine and induction machine with an objective of enhancing loadability of distribution system. The proposed methodology is based on the concept of reactive power margin. Buses have been ranked based on the reactive power margin and grouped as strong and weak buses for finding suitable location of DG units. The effect of size of DG unit on loadability is also examined along with grid loss measure which leads to suitable size selection. The proposed methodology is successfully applied to a modified 16 bus primary distribution system using a commercially available analytical tool and the results are verified using a research analytical software tool. Keywords- Grid integration, distributed generators, loadability, suitable size, suitable location, reactive power margin.

location, keeping the size of DG constant using analytical methods. For example ref. [6] uses tangent vectors for ranking buses and selects the best location for only one type of DG i.e. synchronous generator keeping its size constant. Ref. [8] optimizes both location and size but it considers DG as a real power source without considering its reactive power capability. This paper proposes techniques to identify proper location and size of two principal types of DG units (i.e. induction and synchronous machine) for enhancing loadability. Depending on chosen voltage stability index the buses of a primary distribution system have been ranked first, which limits their number for placing DG units resulting in reduced number of load flow computations. Then based on those results along with loadability study a decision is made for placement of SG and IG in the system followed by determination of suitable size of each DG unit on the selected buses. However, to achieve maximum system benefits in a deregulated environment generator companies can work in collaboration with the transmission and distribution companies to select the location of generators based on the proposed methodology. This paper is organized as follows. Section II gives a brief introduction to the distributed generators used in this paper. Then section III describes the static voltage stability indices employed to determine the location of placing DG units. Section IV describes the methodology for ranking the suitable location of buses for placing SG and IG along with fixing the maximum size of each type. Considering the practical constraints section V presents and describes the results for location and sizing of SG and IG in the test system along with a small introduction to the test system. Section VI summarizes the major contributions and conclusions. II. DISTRIBUTED GENERATORS IN STUDY

I.

INTRODUCTION

In recent years, there has been a considerable increase of penetration of renewable and non-renewable based distributed generation (DG) resources in distribution grids all over the world. With the increased share of DG in power system, allocation and sizing of DG has become the most important concerns for power system stability. Inappropriate selection of location and size of DG may lead to increased system loss and unacceptable voltage profile as have been found in several studies [1],[2]. Due to lack of new generation and increase in demand the existing infrastructure is already facing high grid loss and poor voltage profile. Moreover, it has been found that radial distribution systems with a high resistance to reactance ratio causes a greater amount of loss and are more prone to voltage instability [3]. However, in spite of all these constraints, maximizing loadability has been a good choice for the distribution system operator who wants to optimize their resources and maximize their profit in the current deregulated market scenario. Study on selecting proper location of DG unit is a comparatively new area, unlike selecting location for reactive power compensators. The existing methodologies can be classified into two main categories. The first category is based on artificial intelligence techniques (e.g. genetic algorithm) [4],[5]. These techniques demand a large number of computations resulting in slow convergence. On the other hand, analytical methods with repetitive load flow stands as the second category to decide the locations for DG in power system [6]-[8]. Ref. [6] and [7] however, optimizes only
This work was supported by the CSIRO Intelligent Grid Flagship Collaboration Research Fund.

Synchronous generators connected to a distribution network as DG units are mostly operated with constant active power as their resource is not very volatile (bagasse based CHP plant, gas turbines, solar thermal plants and internal combustion engines). For SG, power factor control mode is usually adopted by the independent power producers as their target is to maximize active power production [9]. So power factor control mode has been employed in this paper. The active and reactive power generation of a synchronous generator can be expressed as given in (1) and (2) [10].

978-1-4244-7397-7/10/$26.00 2010 IEEE

740

IPEC 2010

E q P = E I cos = E sin G t t X t s
E E2 q E cos t Q = E I sin = t G t t X X s s

(1)

A. Voltage Sensitivity Factor Based on the general concept, SF (sensitivity factor) index for a system represented by F ( z , ) can be defined as
SF = dz d

(2)

[15]. When SF becomes large, the system turns

Where, E is the terminal voltage of the generator with per t unit system and E = X i represents the excitation voltage
q ad fd

insecure and ultimately collapses. Here the system voltages are checked with respect to the change in loading which results in a Voltage sensitivity factor (VSF) calculated as VSF = dV
dP

. High sensitivity means even small changes

fd s represents synchronous reactance and internal rotor angle. But for a given real power output of a synchronous generator the reactive power generation is bounded by both armature and field heating limits.

due to field current i

. Here X

and , respectively

in loading causes large changes in voltage magnitude, which indicates weakness of the bus. B. Reactive Power Margin Reactive power margin is measured as a distance between the lowest MVAr point of Q-V curve and voltage axis as shown in Fig. 1 [16],[17]. The negative values of reactive supply indicate the increasing reactive load. Thus reactive power margin indicates how much further the loading on that particular bus can be increased before its loading limit is expired and voltage collapse takes place. Reactive power margins are used in [18] to evaluate voltage instability problems for coherent bus groups. These margins are based on the reactive reserves on generators, SVCs and synchronous condensers that exhaust reserves in the process of computing a Q-V curve at any bus in a coherent group or voltage control area. In this study, this index is used to measure the strength of the buses of primary distribution system with a single feeder. The validity of this index in our study has been justified by another index i.e. voltage sensitivity factor which has been used earlier for the same purpose [6]. With the combined results from measurement of these two indices, the methodology described in section IV has been used to identify suitable location and appropriate size of DG unit in the test system.

Because of the subtle nature of wind, induction generators are largely used in wind power plants [11]. Applications of induction generator are also found in micro turbines of small hydro plants and internal combustion engines. In this paper, the squirrel cage rotor induction generator has been employed which consumes reactive power from the system. With a terminal voltage E and rotor current I , the real and reactive t 2 power generated by the induction generator is given by (3) and (4), respectively [12].
* P = Re( E I ) G t 2 * Q = Im( E I ) G t 2

(3) (4) (5) (6)

Where
I = Et 2

/( jX m (R2 / s + jX2 ) /( jX m + R2 / s + jX2 ) + Z1 )

And

Z1 = R1 + jX 1

Here, R1 =Stator resistance, X 1 = Stator leakage reactance,

X m = Magnetizing reactance, R2 = Rotor resistance (referred

to stator),

X 2 = Rotor leakage reactance (referred to stator).

When an induction generator is placed on a bus, a portion of reactive power is usually locally supplied [13]. In this paper, the capacitor size will be decided later depending on the grid losses in presence of the induction generator. STATIC VOLTAGE STABILITY AND INDICES One of the principal factors of voltage collapse has been identified as the increased load demand which is generally accompanied by an increase in reactive power demand. Distance to collapse can be measured in terms of different physical quantities such as loadability, reactive power reserve etc. Also, a number of performance indices have been developed by the researchers to determine proximity to voltage collapse [14].
Figure 1. Example Q-V curve and reactive power margin.

III.

741

IV.

PROPOSED TECHNIQUES

Overall procedure of the proposed approach can be summarized in the following steps. Step 1: Q-V curves are drawn for all buses to determine reactive power margin of each bus. For a very large distribution system Q-V curves can be drawn by considering the availability of primary resources. Step 2: The buses are ranked in descending order of the values of reactive power margin to form a priority list in terms of strength. Bus having the lowest reactive power margin is the weakest bus. i.e. in an N bus system the weakest bus (WB) can be defined as follows: m arg in m arg in m arg in m arg in WB Min Q ,..., Q ,Q ,Q N 3 2 1 m arg in where Q is the reactive power margin of each i bus in MVAr. Step 3: P-V curves are drawn to get VSF and loadability. All the loads in the base case are presented as constant PQ load and these are increased according to the following relation, keeping the power factor constant: P = P (1 + )

compensator size is determined, which makes the grid loss with IG equal to base case grid loss. V. SIMULATION RESULTS AND DISCUSSION

A. Test distribution System and Analytical Tool In this study, the 16-bus distribution system as shown in Fig. 2 is used. This system is a 23 kV balanced distribution system with a total load of 28.7MW and 17.3MVAr, respectively. This system is a modified form of the one used in [20]. All the results presented in this paper were simulated with the DIgSILENT PowerFactory 14.0 [21], a commercial tool and also have been verified using a research analytical tool PSAT [22]. B. Ranking of Buses Based on the proposed steps a ranking of buses from this test system is made from the calculation of reactive power margin and VSF of system buses. Figures 3 and 4 show the reactive power margin and VSF, respectively for all the load buses. As can be seen from Fig. 3, bus 8 is the bus with the highest margin of 150.35MVAr whereas bus 7 is the bus with the lowest margin of 22.94MVAr. If we expand the concept of this index, it can be noted that when an amount of reactive power equal to reactive power margin is drawn from that bus by loads then it can experience collapse. For example if 150.35MVAr of reactive power is drawn from only bus 8, with loads on all other buses remained unchanged, the system will experience collapse though the total load at base condition counts to be only 17.3MVAr. So based on this comparison index, our study clearly defines bus 8 as the strongest bus and bus 7 as the weakest bus. Fig. 4 plots VSF of all load buses near to the point of collapse when maximum value of loadability or loading point has been reached (here this value is 2.615 times the base load). Bus 7 comes out as the weakest bus while bus 8 stands out as the strongest bus, even in terms of VSF. After closely examining these two plots together we can make a ranking of buses based on their strengths.

= Q (1 + ) 0 L

Where P and Q represents the base case loading and 0 0 is the loading factor. However, the use of realistic load direction can be used to get a more practical solution in this static voltage stability study [19]. The weakest bus in an N bus system has dV dV1 dV 2 dV 3 N WB Max , , ,......, dP dP dP dP Now from this result a ranking of buses based on strength is decided and this list is used to verify results found in step 2. Based on these index values two separate lists of buses weak bus and strong bus are made. The above procedure reduces the solution space to these few buses in the list. Step 4: Two different types of DG with increasing machine sizes are placed on ranked buses. Locations for SG and IG are chosen individually with an objective of increased reactive power margin and loadability. Step 5: Once the location for SG and IG has been fixed for each bus in a priority list SG is placed and the size is varied from minimum (0MW) to a higher value in small steps until minimum grid loss is found by running repetitive load flow. Step 6: Step 5 is now repeated with IG on preferred locations (derived from the first 4 steps) with the same targets. Step 7: With an induction generator connected to the system, grid loss becomes greater than the base case value when real power injection by the generator becomes zero due to lack of primary resources. So the minimum

Figure 2. Single line diagram of test system.

742

Figure 3. Reactive power margin of load buses of the test system. Figure 5. Change in Reactive Power Margin in weak area with inclusion of IG.

Figure 4. Voltage Sensitivity Factor of load buses of the test system.

Table I represents the first four weak buses whereas Table II shows the first four strong buses. For example, in Table I bus no. 7 stands out as the weakest bus with the lowest reactive power margin (22.94MVAr) as well as the highest sensitivity factor (0.038161 Volts p.u. /MW).
TABLE I. WEAK BUSES Bus No. 7 6 1 4 Reactive power Margin (MVAr) 22.94 24.41 27.78 32.54 VSF (Voltage p.u./MW) 0.038161 0.036145 0.028193 0.026833

Figure 6. Change in Reactive Power Margin in weak area with inclusion of SG.

have been assumed with SG too. SG used in this study is operated in power factor control mode with a power factor of 0.8 lagging. These two plots clearly show that the inclusion of IG on the weak buses does not improve the reactive power margin or strength of these buses. For example, the reactive power margin of weak buses like bus 7 remains unchanged around 23 MVAr with the change in real power injection. But with the inclusion of SG this margin and strength has greatly improved which is more prominent with higher real power injection. Here it is observed that the reactive power margin of bus 7 increases from 22.94MVAr to 29.74MVAr with an increase in real power injection by the SG from 0MW to 6MW. So the increase in reactive power margin indicates the growing strength of the weak buses in the presence of the SG which is not achievable with an IG on weak buses. System loadability is clearly affected with the inclusion of DG unit into the system. The results have been shown for DG with increasing sizes: SG in Fig. 7 and IG in Fig. 8. With the increase of size of SG, loadability improves in every case (greater than the base case loading margin = 2.615 p.u.). But, in this study, it has been found that the rate of increase of loadability with respect to machine size is higher for weak buses than strong buses. In a real scenario, this loadability improvement with increasing machine size is limited by the thermal limit of the network components. But as the size of the induction machine is increased the loadability decreases for almost every bus except the buses in the strong area (as mentioned in Table II). The rate of increase of loadability with IG is much lower than SG and after some point it tends to decrease.

TABLE II. STRONG BUSES Bus No. 8 10 9 14 Reactive power Margin (MVAr) 150.35 78.22 74.34 65.97 VSF (Voltage p.u./ MW) 0.002927 0.005134 0.006029 0.008107

C. Determining Locations for SG and IG To find the effect of DG units on reactive power margin, IG and SG are placed separately on weak areas and the resulting reactive power margins are plotted in Figs. 5 and 6, respectively. The total system load was 28.7MW (i.e. around 30MW) hence two machine sizes - 3MW (around 10% of total demand) and 6MW (around 20% of total demand) have been chosen along with the base case with no DG (0MW representing 0% penetration). To keep consistency for comparison of results the same values of real power injection

743

Figure 9. Change in Grid losses and Reactive power intake with SG size Figure 7. Variation in Loadability with the change of Synchronous machine size. TABLE III. SUITABLE SIZE OF SYNCHRONOUS GENERATOR Bus No. Machine Size(MW) with p.f.=0.8 9.81 9.76 9.86 9.68 Grid loss (MW) Without SG 1.78 1.78 1.78 1.78 With SG 0.80 0.76 0.81 0.70 Grid loss (MVAr) Without With SG SG 1.95 0.93 1.95 0.89 1.95 0.91 1.95 0.77 (p.u.)

7 6 1 4

3.39 3.38 3.22 3.23

Figure 8. Variation of Loadability with the change of Induction machine size.

Hence, in order to enhance the overall loadability and reactive power reserve of the distribution system the SG units need to be placed on weak buses (bus no. 7, 6, 1 and 4 as in Table I) whereas the IG units need to be placed on strong buses (bus 8,10,9 and 14 as in Table II). D. Determining Size of Synchronous Generator As the synchronous machine loadability keeps on improving with increasing machine size the suitable size would be that size which corresponds to minimum grid loss. Fig. 9 shows the variation of loss with increasing SG size on bus 7. However, in the real scenario another important factor needing to be addressed is that DG has not been designed to supply reactive power to power the grid/feeder. Usually the grid/feeder always supplies the amount of reactive power required by system demand. So when SG units are placed in the system and the total reactive power demand of the system is delivered from that SG then there will be zero reactive power flow from grid into the DG connected system. Based on a number of load flows with increasing size of SG the suitable sizes are determined and shown in Table III for all the weak buses in the test system. Any SG with a rated value lower/greater than this suitable value will result in a greater amount of real and reactive power losses in the system. With this suitable size the reactive power intake of the feeder becomes zero as shown in Fig. 9. According to the findings shown in Table III placement of a SG of around 10MW on any of the weak buses will reduce the grid loss by more than 50% of the original losses.

E. Determining Size of Induction Generator With the placement of an IG the loadability of the system starts to decrease after a certain machine size as shown in Fig 8. So the suitable size would be that size of IG, which corresponds to maximum loadability. The inclusion of an IG in the distribution system usually results in a greater amount of grid losses as mentioned earlier and this amount of loss is always more than with an SG in the system. Also the reactive power consumption through the grid tends to increase with increasing machine size. Fig. 10 shows the change in MW losses and reactive power intake in the system with the increase of IG size (rated mechanical power) placed on bus 14. However, suitable size needs to ensure less grid loss than base case loss. As a result these two issues of loadability and grid loss are combined to decide the suitable size of IG. With a primary objective of maximizing loadability the grid loss is calculated with increasing machine sizes. The suitable/maximum size as shown in Table IV represents the machine for specified buses which offers maximum loadability with grid loss lower than the base case. It is interesting to note that the suitable size of IG at bus 8, which is the strongest bus in the system, turns out to be 21MW though the system demand is around 30MW. However, in other locations, the suitable sizes are in the range of 2.2MW up to 9MW which is around 30% of the total demand. The minimum values of compensation which will bring the loss (at zero real power injection) at most equal to the base loss has been worked out and tabulated in Table V. For example, placing a capacitor with a minimum value of 1.6MVAr at bus no. 9 (with the specified size of 7.5MW induction machine) results in a grid loss of 1.78MW at zero real power injection.

744

REFERENCES
[1] C. L. T. Borges and D. M. Falcao, "Impact of distributed generation allocation and sizing on reliability, losses and voltage profile," in Power Tech Conference Proceedings, 2003 IEEE Bologna, 2003, p. 5 pp. Vol.2. S. K. Salman, "The impact of embedded generation on voltage regulation and losses of distribution networks," in Embedded Generation on Distribution Networks (Digest No. 1996/194), IEE Colloquium on the Impact of, 1996, pp. 2/1-2/5. M. Moghavvemi and M. O. Faruque, "Technique for assessment of voltage stability in ill-conditioned radial distribution network," Power Engineering Review, IEEE, vol. 21, pp. 58-60, 2001. G. Celli and F. Pilo, "Optimal distributed generation allocation in MV distribution networks," in Power Industry Computer Applications, 2001. PICA 2001. Innovative Computing for Power - Electric Energy Meets the Market. 22nd IEEE Power Engineering Society International Conference on, 2001, pp. 81-86. M. Gandomkar, M. Vakilian, and M. Ehsan, "Optimal distributed generation allocation in distribution network using Hereford Ranch algorithm," in Electrical Machines and Systems, 2005. ICEMS 2005. Proceedings of the Eighth International Conference on, 2005, pp. 916918 Vol. 2. N. Mithulananthan and T. Oo, "Distributed Generator Placement to Maximize the Loadability of Distribution System " IJEEE vol. 43, pp. 107-118, April 2006. W. Caisheng and M. H. Nehrir, "Analytical approaches for optimal placement of distributed generation sources in power systems," in Power Engineering Society General Meeting, 2005. IEEE, 2005, p. 2393 Vol. 3. N. Acharya, P. Mahat, and N. Mithulananthan, "An analytical approach for DG allocation in primary distribution network," International Journal of Electrical Power & Energy Systems, vol. 28, pp. 669-678, 2006. W. Freitas, J. C. M. Vieira, A. Morelato, L. C. P. da Silva, V. F. da Costa, and F. A. B. Lemos, "Comparative analysis between synchronous and induction machines for distributed generation applications," IEEE Transactions on Power Systems, vol. 21, pp. 301-311, 2006. P. Kundur, N. J. Balu, and M. G. Lauby, Power system stability and control. New York: McGraw-Hill, 1994. T. Ackermann, Wind power in power systems, Chichester, West Sussex, England, John Wiley, 2005. V. Akhmatov, Induction generators for wind power. [Brentwood]: MultiScience Pub., 2005. N. Jenkins and Institution of Electrical Engineers., Embedded generation. London: Institution of Electrical Engineers, 2000. "Voltage stability assessment, procedures and guides " IEEE/PES Power System Stability Subcommittee Technical ReportJanuary 2001. Y. Mansour, "Suggested techinques for voltage stability analysis," IEEE/PES, Techincal Report1993. C. W. Taylor, N. J. Balu, and D. Maratukulam, Power system voltage stability, New York, McGraw Hill, 1994. T. Van Cutsem and C. Vournas, Voltage Stability of Electric Power Systems, Kluwer, 1998. R. A. Schlueter, "A voltage stability security assessment method," IEEE Transactions on Power Systems , vol. 13, pp. 1423-1438, 1998. A. Sode-Yome, N. Mithulananthan, and K. Y. Lee, "Effect of Realistic Load Direction in Static Voltage Stability Study," in IEEE/PES Transmission and Distribution Conference and Exhibition: Asia and Pacific, 2005, pp. 1-6. S. Civanlar, J. J. Grainger, H. Yin, and S. S. H. Lee, "Distribution feeder reconfiguration for loss reduction," IEEE Transactions on Power Delivery, vol. 3, pp. 1217-1223, 1988. DIgSILENTGmbH, "DIgSILENT PowerFactory V14.0 -User Manual," DIgSILENT GmbH, 2008. F. Milano, "PSAT, Matlab-based Power System Analysis Toolbox," 2002.

[2]

[3]

[4] Figure 10. Change in Grid losses and reactive power intake with Induction machine size. [5] (p.u.) [6] 2.66 2.61 2.34 2.58

TABLE IV. SUITABLE SIZE OF INDUCTION GENERATOR Bus No. 8 10 9 14 Machine Size (MW) 21 9 7.5 2.2 Grid Loss (MW) With out IG 1.78 1.78 1.78 1.78 With IG 1.14 1.36 1.23 1.62 Grid Loss (MVAr) With With out IG IG 1.95 1.30 1.95 1.53 1.95 1.34 1.95 1.79

[7]

TABLE V. MINIMUM SIZE OF COMPENSATOR Bus No. Induction Machine Size (MW) 21 9 7.5 2.2 Grid Loss (MW) With out IG 1.78 1.78 1.78 1.78 With IG (Real power injection =0) 1.78 1.78 1.78 1.78 Compensat or size (MVAr) 3 1.9 1.6 0.5

[8]

[9]

8 10 9 14

[10] [11] [12] [13] [14] [15] [16] [17] [18] [19]

VI.

CONCLUSIONS

A methodology has been proposed based on Q-V curve to determine the location of two major classes of DG synchronous generator (SG) and induction generator (IG) considering the reactive power issues of the system and these machines. Based on the result, it can be said that SG need to be placed on weak bus whereas IG need to be placed on strong bus to enhance system loadability. It is interesting to note that the rate of improvement of loadability with SG on weak bus is greater than the case where SG is placed at a strong bus. With IG on strong bus loadability improves up to certain machine size and then starts to decrease. Once the locations have been fixed, lookup tables have been formed using the proposed methodology, which can be used to restrict the size of each type of DG in the system. At suitable size of SG, reactive power intake of the system has been found zero with minimum grid loss. However, in reality an independent power producer would go for the closest match to the sizes mentioned in the market and deviations in grid loss and loadability due to this choice are within the tolerable range.

[20]

[21] [22]

745