You are on page 1of 8
letical Power and Energy Systems 27 (2005) 335-342 Evecruc, POWER. ENERGY SYSTEMS A simulation model for AGC studies of hydro—hydro systems KC. Divya, P.S. Nagendra Rao* Deparment of Electrical Engineering, Indian nstite of Science, Bangalore S60012, India Received 19 March 2003; revised 25 November 2004; accepted 29 December 2004 Abstract AGC studies peta to hyclro-hydro systems have received Fite attention. In this paper a simulation model for AGC studies of such systems has been proposed. The difficulty in extending the traditional approach [Elgerd OL. Electric energy systems theory, New York: Me- Graw Hill; 1983.| for such systems is overcon ‘by assuming that all areas in a system operate at the sa simulation model is obtained by ignoring the difference in frequeney between control arcas, unlike the traditional approach, whe e frequency. The proposed cach area is assumed to operate ata different frequency. The features of the proposed model have been demonstrated through simulation studies. © 2005 Elsevier Ltd. All rights reserved, ‘Keywonls: Automatic generation coat; Simulation models: Hydro systems 1. Introduetion ‘The normal operation of an interconnected multi-area power system requires that each area maintain the load and, generation balance. This is normally achieved by means of fan automatic generation controller (AGC). AGC tries to achieve this balance by maintaining the system frequency and the tie line flows at their scheduled values. The AGC action is guided by the area control error (ACE), which is a Function of system frequency and tie line flows. The ACE, represents a mismatch between area load and generation taking into account any interchange agreement with the neighboring areas. The ACE for the ith area is defined as ACE = APic + BAF w where Pic Pic staal ~ Pic wasted aNd Py isthe net tie Tine flow; Af=frciustfashedutca and f is the system frequency: B, is referred to asthe frequency bias factor ‘This control philosophy is widely used and is generally refered to asthe tie line bias contol ‘AGC studies are generally carried out using simulation ‘models. The origin of the simulation model that has been * Comresponing author. Tel: +91 8022932365; fax: +91 802360648 Exmail addresses: divyaseniise-emotin (K.C. Divya), nagen exteciiscemetin PS. Nagendea Rao} (0142-06158 ~sce front mater © 2005 Elsevier Lu. All ihts reserved 10-1016 .jepes. 2004.12.04 Widely used [1] for AGC studies can be traced {© [2} Sulpsequently, the state space and diserete versions of this model have also been used [3.5] In [1,3] this model has been used for AGC studies of a two-area non-reheat thermal system. Later, this model has been used to study the AGC for two area reheat thermal system [4-6] and hydro thermal system [7]. Further, it has been extended for multi-area systems, which comprises of three [8] as well as four area systems [9,10]. Itappears thatthe studies carried out sofar are limited to only thermal or hydrothermal systems. Even though the interconnected hydro systems are quite common, AGC of such systems does not seem to have been studied so far. ‘This paper presents a study of the AGC for a two area hydro system. As an attempt to extend the conventional ‘model [1] tured out to be unsuccessful, an alternate simulation model for AGC studies of such system has been proposed here. The AGC performance of a 1wo area test system has een studied using the proposed model with an integral controller. 2. The proposed model ‘The primary requirement of the power system model in the context of AGC studies is that it should enable the computation of deviations in frequency and area tic line flows. In the traditional models, each area with in a system is assumed to be operating at a different frequency. ‘The frequency deviation in cach area is obtained by 26 KC. Divya, PS. Nagendra Rao / Electrical Power and Energy Systems 27 (2005) 335-242 considering the area dynamics (aggregate dynamics of the governors, turbines, generators and loads present in the area), The tie line low deviations between the areas is ‘computed as the product of the tie line constant and the ‘angular difference between the two areas. “The proposed model adopts a different approach. Here, the entire system is considered to be operating al a single frequency (all area frequencies are equal). This approach though uncommon, is not unjustified. In fact, the IEEE task force on AGC [11] explicitly recommends this approach. In [11] ithas been stated that “in AGC studies, the momentary difference between the frequency of different areas can be ignored. For all AGC purposes the frequency used for one area to compute ACE should be the same as used in the other ‘areas so long as they remain interconnected’. A conceptual frame work of such a simulation model (common system frequency approach) had also been outlined earlier by Athay [12] in his comprehensive review of AGC. In the proposed model, the frequency (system frequency ‘common to all areas) is determined by integrating the net system accelerating/decelerating power (i.e. difference of total system generation and load). Since the difference between the area frequencies is neglected, the traditional approach cannot be used to compute the tie line flow deviations. In order to obtain the tic line lows the area power balance equations has been used. The power balance ‘equation for the ih area is written as ys a Po + Pa; — PLD) = Hi, 2 where Pacis the te Hine power Now: Pg; isthe actual total ‘area generation; PL) is the total area load, which is a function of frequency: H(dfayld) is accelerating or decelerating power of each area. H, is the total generator inertia of the area and f,, is the system frequency, may be noted that the concept of common system frequency and power balance equation based te line low computation have been incorporated in a power system simulator developed to verify the long term dynamic behaviour as well as AGC performance of the Pacific Northwest system [13]. Is claimed in [13] that with some fe I fine tuning of the prime mover parameters (based on system ‘measurements) the simulator response matched closely with, the measured responses. However, this approach does not, seem fo have been adopted by others for AGC studies subsequently. It can be seen that the proposed model has 1wo distinet advantages as compared with the traditional approach. The first is that, this model does not require the computation of the te line constant, This calculation is not straight forward, {or systems wherein two areas are connected by more than. fone tie line. The other advantage is that this model does not, require the use of a composite prime mover model representing the entire area. Thus, this model can preserve the identity of each unit and non-linearities like dead band, and generation rate constraint (GRC) of different units can be separately incorporated. 3. Simulation results Results of simulation studies of two systems. are presented here in order to illustrate the performance of the proposed model. The first test system (Test system-1) corresponds to a two area hydro-hydro system with each, area being represented by a composite model. The other test system (Test system-2), having six hydro units, is derived from the IEEE 30 bus test system. 3. Test system-1 ‘The choice of this test system (with composite hydro unit model) is prompted by the following considerations: (@) Representing area dynamics through a composite turbine model is an extensively used practice in AGC literature (2-7] over the last 50 years. ii) This choice facilitates the comparison of the primary response obtained using the proposed model with that obtained using the conventional model ii) The parameters for the composite hydro unit model of ‘an area are readily available from easlier work [7] dy, [=m AO ea [Liv Fig. 1. Conventional model—block diagram. KC. Divya, PS. Nagendra Rao / Elecrcal Power and Energy Systems 27 (2005) 35-342 a7 Table 1 Pacumctr values: two aca hydo sytem Paramctrs Proposcd model 5 = % : 4 87 0519 he 5 i 1 Ripa) 24 24 Ta (00867 . Ko - ca Toe : 20 1 (puMtwira) . 04 Bs (poM WH) . 0 First the responses of both the models (conventional and proposed) have been oblained and compared in the absence. of AGC (primary response). Later, the response of the ‘proposed model has been studied with an integral controller. 3.1.1. AGC models for test system-1 Fig. 1 shows the block diagram representation of the conventional model for the two area hydro system considered here. The parameters of each of the areas are identical and are given in Table 1. They correspond, to the hydro area model used in [7]. The value of the tie line constant is also identical to that used in [7], which is, the same as that used for two area non-reheat thermal system in [3] and the two area reheat thermal system, in [4-6] ‘The block diagram representation of the proposed ‘model for this system is shown in Fig. 2. Comparing Fig. 2 with Fig. I it can be seen that the tie Tine flow computation block is absent in the proposed model. Further, a single system frequency is obtained by integrating the net system accelerating power (APiy). However, in the conventional model shown in Fig, 1, two different area frequency deviations (Afi, Af) are obtained, by integrating the individual area accelerating powers (AP a1. BP). 3.1.2. Primary response ‘The responses of the proposed and conventional models, in the absence of AGC, have been studied for a step load disturbance in one of the area, This response essentially, shows how a new load is shared by the two interconnected ‘areas in the absence of AGC. The variation in frequency and tie line power flow deviations obtained using the conventional and the proposed mode! (with out the controller) for a 0.01 pu step load increase in area-2 have been shown in Figs. 3 and 4, respectively. From these plots, itis evident that the response ‘obtained with the proposed model is stable and does not contain any high frequency oscillations and that obtained with the conventional model contains high frequency oscillations whose magnitude increases with time, I is obvious that the conventional model response is unstable. This inference is further reinforced by the location of the poles of the conventional model given in Table 2. From, ‘Table 2 it is seen that a pair of complex poles are situated in the right half ofthe s-plane. However, itis well known that a umber of hydro units can operate in parallel in a very stable ‘manner, even when there is no AGC. This implies that the traditional model is incapable of capturing the right Ben [igo | Pag a cea yg AP) AP Ty | Ay > 1 bP + Migs ¢ BPiys = MPa + OF 1 propose AGC model: tet sytem-2. Pn amste) Tat /2) Pilot valve and Scromolortime constant Servomotor gain Main servamotor time Reset time Permanant droop Temporary droop Ky te te 6 Fi 9. Hydro unit mode ‘that the system frequency falls considerably immediately after the disturbance, rises as the generators pick up the load and finally settles down at 59,9920 Hz, As the load disturbance is in area-1, the tic line flow deviation of area-1 is negative. Further, from this plot, it ean be seen that the lie Tine flow deviations of the two areas oscillate around 0401 and then settle down at this value (in the absence of supplementary control). However, from the tie line flow riation obtained (using the proposed model) shown in Fig. 4, it is seen that the tie line flow immediately settles down 10 its final value without any oscillations. ‘This is because the test system-1 both the areas are chosen, to be identical, while in test system-2, each unit is different ‘and the corresponding dynamies of the wo areas are different, 3.2.3. Response with controller ‘The AGC response of test system-2 is obtained using the proposed mode! with an integral controller. The design procedure followed here is identical to that used earlier for {est system-1. The performance index J has been obtained for various values of gain and K,=0.01 is seen to minimize J. ‘The response of the controller, for this Value of gain is shown in Fig, 11. From this it can be seen that the governor set point inereases almost monotonically and finally settles down at a value equal to the load disturbance in area-1; no futile control actions are seen. Hence, a value of K;=0.01 has been chosen. as the integral controller gain for this system. ‘The response of test system-2 with the integral controller (K,=0.01) has been obtained. Fig. 12 shows the variation in ACE for a 0.02 pu step load disturbance in atea-1. From this, ‘Tehles Hydro generator unit data Gen Gen. bueno, Area nO fp K A ° a he aeaner x T T bos S0 50 vor 10 ase 8 1 om 30 55 oot a a 25s c 1 os 485 50 oot oa Lo as sna D 2 om SH 50 oor Lo ce E 2 oo a9 50 oor ot Lo 23 F 2 ous sz 45 oot oa og 20 KC. Divya, PS. Nagendra Rao / Elecrcal Power and Energy Systems 27 (2005) 335-342 Va 2 tf oa 00 “Tie line flow variation : Area-t ——_———_——————— 010 Bt time (sec) Ti tine Now variation = Area-2 ————————————— 10 Panay pw 2 Boon Soon a ol ee a ne ee) f = 5.06) “aod oh i tna) Fig 12 Variation in ACE with intgral conwoller(comtller gain 0.01): test system2. it ean be seen that the magnitude of the ACE of area-I (area 4. Conclusions ‘in which load increases) reaches a value which is about four times the magnitude of load disturbance immediately after In this paper a simulation model has been proposed for the load disturbance. While this rise in ACE appears to be large, it is 10 be noted that in the absence of the supplementary control the overshoot seen in ACE signal is, 0.16 pu. Consequently, the integral control action has not only reduced the steady-state error to zero, but also reduced, the overshoot in ACE signal. In this ease also, similar to test system-1, the ACE signal remains in a band around zero (5% of the disturbance magnitude) after about 100 s, AGC studies of @ hydro-hydro system. It has been shown, that this model, unlike the conventional model, provides a stable representation for an interconnected multi-area hydro system. An additional feature of this model is that it filters ‘out the fast varying components—a desirable feature according to the IEEE task force on AGC [11] ‘The AGC response of two different test systems obtained. using the proposed model have been studied, In this study, we KC. Divya, PS. Nagendra Rao / Electrical Power and Energy Stems 27 (2005) 335-242 ‘an integral control strategy has been used and a suitable controller gain has been obtained, References [1] Blgerd OL. Electric energy systems theory: New York: Me-Graw Hill, 1983, [2] Concoatia C, Kirchmayer LK. Tieine power fequeney contol of clectical power systems, Part 1, Trans AIEE 1953:562-72 Elgerd Ol, Foss CE. Optimum megawat-requency contol of multiarea electric energy’ system. IEEE Trans Power Apparatus Syst TOTD:PAS-89(4):856-62. ‘Tripathy SC, Hope GS, Malik OP. Optimization of load frequency contol parameters for power systems with reheat steam trbines ind [govemor dead band no-lineaity, IEE Proo-—Put © 19821291016 Koshart ML, Satsang PS, Nanda J Sample data automatic generation control_of interconnected cheat thermal systems considering "snerstion rte contin IEEE Trans Power Apparates Syst 19815 PAS. 104a):2334-42. [6 Koti ML Nanda , Kothari DP, Das D. Discrete mode age of 10 ara reheat thermal sytem with new area controler. IEEE Trans Power Syst 1989:42)730-8 [7 Nanda J, Khai ML, Satsngi PS. Automatic generation control of imerconnected hydro thermal systems in continuous and discrete ‘mode considering generation rate consta IEProc—PartD 1988: 130(1):17-27 [8] Lim KY, Wang Y, Zhou R, Robust decentralized lad frequency sonal of muliaea power systems. TEE Proc—Part C 1996:14% 377-46, [9] Malik OP, Kumar A, Hope GS. A toad frequency conto algorithm based on genenized approach. IEEE Trans Power Syst 198A: 375-82 [10] Chang CS, Fa W. Area lead fraqueney contol using faery fin scheduling of Pl controllers. Elect Power Syst Res 199742: 145-92, [11} Jabot N,_ Vansyek LS, Evart DN, Fink LH, Hottmann AG. Understanding sttomtic generation control, IEEE Trans Power Syst 1992-713; 1106-21 {U2} Athay TM. Generation scheduling and contol Proc TEEE. 1987.75: 1392-606, [13] Taylor CW, Cresap RL. Real-time power systom simulation fo sutomatic generation contol. IEEE Trans Power Syst 1976:DAS- 951) 375-94, [14] NERC. Policy 1 generation contol and performance. Oper Man 196:Decs1-14, 1} hapwww pserecomeleda

You might also like