Secondary Voltage Regulation Based on Average Voltage Control

TecnoLógicas Artículo de Investigación/Research Article ISSN-p 0123-7799 ISSN-e 2256-5337 Vol. 21, No. 42, pp. 63-78 Mayo-agosto de 2018

Secondary voltage regulation based on average voltage control

Regulación secundaria de voltaje basada en el control del voltaje promedio

Edwin H. Lopera-Mazo1, y Jairo Espinosa2

Recibido: 05 de febrero de 2018 Aceptado: 20 de abril de 2018

Cómo citar / How to cite

E. H. Lopera-Mazo, y J. Espinosa, Secondary voltage regulation based on average voltage control. TecnoLógicas, vol. 21, no. 42, pp. 63-78, 2018.

© Copyright 2015 por autores y Tecno Lógicas Este trabajo está licenciado bajo una 1 Licencia Internacional Creative MSc. in Engineering, Electrical Engineer, Facultad de Minas, Universidad Commons Atribución (CC BY) Nacional de Colombia, Facultad de Ingenierías, Instituto Tecnológico Metropolitano, Medellín-Colombia, [email protected] 2 PhD. in Applied Sciences, MSc. in Engineering, Electronic Engineer, Facultad de Minas, Universidad Nacional de Colombia, Medellín-Colombia, [email protected] Secondary voltage regulation based on average voltage control

Abstract This paper compares a conventional Secondary Voltage Regulation (SVR) scheme based on pilot nodes with a proposed SVR that takes into account average voltages of control zones. Voltage control significance for the operation of power systems has promoted several strategies in order to deal with this problem. However, the Hierarchical Voltage Control System (HVCS) is the only scheme effectively implemented with some relevant applications into real power systems. The HVCS divides the voltage control problem into three recognized stages. Among them, the SVR is responsible for managing reactive power resources to improve network voltage profile. Conventional SVR is based on dividing the system into some electrically distant zones and controlling the voltage levels of some specific nodes in the system named pilot nodes, whose voltage levels are accepted as appropriate indicators of network voltage profile. The SVR approach proposed in this work does not only consider the voltage on pilot nodes, but it also takes the average voltages of the defined zones to carry out their respective control actions. Additionally, this innovative approach allows to integrate more reactive power resources into each zone according to some previously defined participation factors. The comparison between these strategies shows that the proposed SVR achieves a better allocation of reactive power in the system than conventional SVR, and it is able to keep the desired voltage profile, which has been expressed in terms of network average voltage.

Keywords Secondary Voltage Regulation, Average voltage, Conventional SVR, Hierarchical Voltage Control System, Power System.

Resumen En este trabajo se realiza una comparación entre un esquema convencional de Regulación de Voltaje Secundario (RVS) que se basa en nodos piloto y un RVS propuesto, que toma en cuenta los voltajes promedio de las zonas de control. La importancia del control de voltaje para la operación de los sistemas de potencia ha promovido varias estrategias para enfrentar este problema. Sin embargo, el Sistema de Control de Voltaje Jerárquico (SCVJ) es el único esquema efectivamente implementado con algunas aplicaciones relevantes en sistemas de potencia reales. El SCVJ divide el problema de control de voltaje en tres etapas reconocidas. Entre ellas, la RVS es la encargada de gestionar los recursos de potencia reactiva para mejorar el perfil de tensión de la red. La RVS Convencional se basa en la división del sistema en algunas zonas eléctricamente distantes y en controlar los niveles de tensión de algunos nodos específicos del sistema denominados nodos piloto, cuyos niveles de tensión se aceptan como indicadores adecuados del perfil de tensión de la red. La RVS propuesta en este trabajo no solo considera el voltaje en los nodos piloto, sino que también toma los voltajes promedio de las zonas definidas para llevar a cabo sus respectivas acciones de control. Además, este nuevo enfoque permite integrar más recursos de potencia reactiva en cada zona de acuerdo con algunos factores de participación previamente definidos. La comparación entre estas dos estrategias muestra que la RVS propuesta logra una mejor asignación de la potencia reactiva en el sistema con respecto a la

[64] TecnoLógicas, ISSN-p 0123-7799 / ISSN-e 2256-5337, Vol. 21, No. 42, mayo-agosto de 2018, pp. 63-78 Secondary voltage regulation based on average voltage control

SVR convencional y que es capaz de mantener un perfil de voltaje deseado, el cual ha sido expresado en términos de la tensión media de la red.

Palabras clave Regulación de Voltaje Secundario, Voltaje promedio, RVS convencional, Control de Voltaje Jerárquico, sistema eléctrico de potencia.

TecnoLógicas, ISSN-p 0123-7799 / ISSN-e 2256-5337, Vol. 21, No. 42, mayo-agosto de 2018, pp. 63-78 [65] Secondary voltage regulation based on average voltage control

1. INTRODUCTION HVCS solves the voltage control problem by geographically and temporarily Keeping all node voltage levels within dividing the electric power system consid- an appropriate operating range and con- ering three hierarchical levels. This divi- trolling reactive power flow constitute sion aims to limit interference between some of the most important tasks in the involved control actions. These three levels operation of an electric power system [1]. are known as Primary (PVR), Secondary Some relevant electric service aspects such (SVR), and Tertiary Voltage Regulation as efficiency, stability, safety, reliability (TVR) [5], as shown Fig. 1. and energy quality directly depend on the PVR is based on the operation of Auto- regulation of these variables [2]. matic Voltage Regulator (AVR), which Many proposals have been studied and changes the magnetic field applied to the developed in order to improve voltage con- network’s synchronous generators and trol in transmission systems. Among these other kinds of reactive compensation con- alternatives, the Hierarchical Voltage Con- trollers, such as capacitor and reactor trol System (HVCS) has been widely rec- banks, and on-load tap changers, among ognized as a viable solution, because it has other. This level only operates with local been satisfactorily adopted in some coun- voltage measurements and presents re- tries [3]. HVCS has shown superior per- sponse times of some fractions of a second formance compared with early schemes for node voltage correction where this con- where secondary voltage regulation partic- trol is applied [6]. ularly has been done in a manual way [4]. According to the theoretical foundation of conventional SVR, voltage control is

Fig. 1. Hierarchical structure of the coordinated voltage control. Source: Authors’ own work.

[66] TecnoLógicas, ISSN-p 0123-7799 / ISSN-e 2256-5337, Vol. 21, No. 42, mayo-agosto de 2018, pp. 63-78 Secondary voltage regulation based on average voltage control essentially a local problem where electri- but the principles behind system partition cally close nodes exhibit similar voltage are basically the same [12]: 1) a strong variation in the presence of network dis- coupling between nodes in the same zone; turbances. Therefore, the previous step 2) reactive controllable resources in each before applying conventional SVR consists zone; and 3) appropriate number of zones in dividing the system into voltage control to implement regional control. The most zones with low electrical coupling. Howev- commonly-used method considers short er, many factors could affect this coupling circuit level of nodes and it uses a voltage between the defined areas; as a result, sensibility matrix derived from the sys- conventional SVR’s performance would be tem’s Jacobian matrix. The analytical pro- significantly degraded after a particular cedure is based on the sensibility matrix of disturbance [7]. node voltage changes regarding injected After the definition of these zones, con- reactive power changes in each node when ventional SVR executes a coordinated con- only PVR operates [10]. trol of the reactive power resources related In conventional SVR, each voltage con- to the PVR in each of these areas. If the trol zone is essentially characterized by its PVR is based on AVR operation, conven- own pilot node voltage; therefore, the con- tional SVR dynamically adjusts the opera- trollers cannot perceive significant variation points of this equipment. The genera- tions in the voltage levels of electrically tors chosen for participating in voltage remote areas from this node, which can regulation in each zone are known as con- adversely affect power electric perfor- trol generators. The main objective of these mance. This problem could be more rele- generators is to adjust voltage levels of vant in nodes near the boundaries of two or some specific nodes from each zone, named more control zones, because these nodes pilot nodes [8], [9]. Finding a specific should be electrically more distant from weighted partition of the reactive power the pilot nodes of each zone. provided from each zone control generator This paper presents a method that is the most-commonly used criterion in this characterizes network voltage profile traditional scheme. The aim is to keep a through a performance indicator based on convenient reactive power margin in each some (or perhaps all) node voltages that control zone to face eventual voltage con- belong to each area, instead of using only tingencies. SVR takes more time to re- pilot nodes. The proposed SVR includes spond than PVR and it exhibits time con- more node voltages, even those located stants close to three minutes. near bordering areas, and it provides In a superior level of this hierarchical greater flexibility because changing the structure is the TVR. This stage defines participant nodes is easier. This method reference values for pilot node voltages also conveniently controls reactive power normally based on an optimal power flow injection in each zone in comparison with a analysis for this estimation. Typically, this conventional SVR scheme. Moreover, the stage tries to minimize system energy proposed SVR offers great possibilities for losses or to maximize loading according to progressively adapting the control scheme security and economic constraints. TVR to topology changes that could be present presents time constants near to ten in the network. minutes [10]. The results presented in this paper In order to define the voltage control could have a relevant impact on currently zones and their corresponding pilot nodes implemented SVR schemes based on the several methods have been proposed [11]. pilot nodes, because this new proposal This task constitutes one of the most im- maintains voltage control zone partition- portant issues for HVCS implementation, ing. Additionally, although the proposed

TecnoLógicas, ISSN-p 0123-7799 / ISSN-e 2256-5337, Vol. 21, No. 42, mayo-agosto de 2018, pp. 63-78 [67] Secondary voltage regulation based on average voltage control

SVR is not based on optimization algo- voltages is a first relevant feature of the rithms as in some recent approaches, this conventional SVR. The other important scheme allows to easily integrate new con- aspect of this stage corresponds to the trol strategies into the HVCS. application of weighted participation factors for all devices injecting reactive power in each zone, in order to have an appropri- 2. METHOD ate reactive power reserve in the network to support voltage contingencies. These 2.1 Conventional SVR two features encourage establishing voltage control zones with low electric coupling Fig. 2 shows the conventional SVR as coefficients [13]. commonly used in a specific voltage control zone within the HVCS structure. In this 2.2 Definition of voltage control zones and scheme, AVR sets the PVR and its main pilot nodes task is manipulating excitation system of control generators in each area in order to Several methods have been proposed reach a specific stator voltage reference. for dividing an electric network into volt- Meanwhile, SVR leads the voltage refer- age control zones and defining their re- ence value of each control generator to- spective pilot nodes. The most widely-used wards a new specific reference. This sec- and implemented method for this division ondary stage has normally adopted two is based on electrical distances and short approaches to change AVR voltage refer- circuit levels of nodes, which is derived ences with longer time constants. Handling from a stationary analysis of the system’s the injected reactive power in each control variables [11]. zone according to the changes of pilot node

Measured voltage at Pilot Node

Voltage PILOT reference NODE Proportional – Integral Controller Measured reactive power

Reactive power Reactive control loop Power reference Measured stator voltage Control generator Other Stator units voltage Automatic Voltage reference Regulator (AVR) variation

VOLTAGE Stator voltage Excitation CONTROL reference Excitation system ZONE

voltage

Fig. 2. Conventional SVR approach. Source: Adapted from [6].

[68] TecnoLógicas, ISSN-p 0123-7799 / ISSN-e 2256-5337, Vol. 21, No. 42, mayo-agosto de 2018, pp. 63-78 Secondary voltage regulation based on average voltage control

The restrictions of a power system can of them. If a high number of pilot nodes is be represented in the linearized form [10] selected, some voltage control zones with shown in (1): high electric coupling among them could be created. In that situation, a control action Δ푃 퐽푃θ 퐽푃푉 Δ휃 [ ] = [ ] [ ] (1) in a region could strongly affect the voltage Δ푄 퐽푄휃 퐽푄푉 Δ푉 behavior of the neighboring nodes, which could have adverse effects on the applica- where Δ푃, Δ푄, Δ휃, Δ푉 represent incre- tion of the secondary regulation strategy. mental changes in real power, reactive On the other hand, an exceptionally low power, voltage angle, and voltage mag- interaction between control zones, i.e. a nitude of system nodes, respectively; small number of control zones, could pro- and 퐽푃θ, 퐽푃푉, 퐽푄휃, 퐽푄푉 are the Jacobian duce a serious deterioration of the network matrix elements. voltage profile, because some node voltage If active power is kept constant, i.e. disturbances will probably be neglected by Δ푃 = 0, the expression (2) is obtained from controllers. Therefore, setting a sensibility (1), threshold that defines the interaction level of control zones and their pilot nodes is a fundamental aspect for the SVR scheme. Δ푄 = 퐽푅 Δ푉 (2)

2.3 SVR based on the average voltage of the where is known as the reduced Jacobian 퐽푅 control zones matrix of the system, which is given by (3),

−1 The conventional SVR scheme assumes 퐽푅 = 퐽푄푉 − 퐽푄휃 퐽푃휃 퐽푃푉 (3) that pilot nodes characterize the voltage behavior of each control zone. Consequent- Besides, from (2), the relationship be- ly, in this second scheme, the voltage between Δ푉 and Δ푄, will be (4), havior of remaining nodes is practically neglected. This situation could disregard −1 Δ푉 = 퐽푅 Δ푄 (4) some important voltage variations in specific areas, which can affect the general −1 where 퐽푅 is the reduced 푉푄 Jacobian, system voltage profile. 푡ℎ where its 푖 diagonal element is known as An innovative approach to improve 푡ℎ the 푉푄 sensibility at 푖 system node. SVR performance has been proposed in The matrix of electrical distances plays this work in order to face these specific an important role in the study of inde- aspects. The proposed SVR calculates the pendent voltage control zones [14] and it is average voltage of each control zone and defined as (5): defines an appropriate injection of reactive 퐷푖푗 = 퐷푗푖 = − log(훼푖푗 훼푗푖) (5) power for each control generator by changing the voltage reference of the AVR. In 휕푉푖 휕푉푗 the new scheme, a previously established where 훼푖푗 = ( ) / ( ) , and 훼푗푖 = 휕푄푗 휕푄푗 weighted balance between injected reactive 휕푉 휕푉 ( 푗) / ( 푖), whose terms are obtained powers in all the control generators is 휕푄 휕푄 푖 푖 maintained as in the conventional SVR. from the Jacobian matrix. The proposed SVR defines the voltage Matrix 퐷 defines the electric coupling 푖푗 control zones by using the sensibility ma- level between any pair of nodes in the sys- trix in a way that is comparable to the tem. As a result, it is used for establishing conventional approach. However, this in- voltage control zones and selecting pilot novative approach characterizes these nodes and control generators for each one zones based on their average voltages, thus

TecnoLógicas, ISSN-p 0123-7799 / ISSN-e 2256-5337, Vol. 21, No. 42, mayo-agosto de 2018, pp. 63-78 [69] Secondary voltage regulation based on average voltage control possibly including all load and generation 푉 nodes in each zone. The highest level of 퐴푘 HVCS could define the participation percentages of all control generators. + _ PI Controller 푘 Δ푉퐴푘 In the proposed approach, the average voltage 푉 of the 푘푡ℎ control zone is given 퐴푘 푉 by (6): 퐴 푟푒푓 푘

1 Fig. 3. Average voltage variation in the proposed 푉 = ∑ 푉 SVR approach. Source: Authors’ own work. 퐴푘 푚 푖 (6) 푖휖Ω control zone), this approach assumes that where there are n control generators per zone,

and the use of the term average voltage nodes set of the 푡ℎ control zone includ- Ω: 푘 finally allows to involve all nodes in the ed in the SVR approach control strategy. number of nodes of 푚: Ω The proposed SVR takes the total reac- : 푡ℎ node voltage of 푉푖 푖 Ω tive power generated at each control zone

Q , and carries out a specific distribution To integrate this new variable into the T of its injection into the zone according to SVR scheme, an average voltage reference the participation factors (α , α , … , α ), value ( ) should be defined for each 1 2 n 푉푟푒푓 푘 which define the percentage of reactive voltage control zone, whose magnitude power to be applied by each generator to could correspond to the solution of an op- the area. Even if there is a single control timization problem provided at tertiary generator in a specific zone (as it usually stage of HVCS. As shown in Fig. 3, to limit happens in the conventional SVR), the the interference of these new control ac- voltage profile of each area is improved tions, a PI controller has been incorpo- because the average voltage value would rated. Its parameters allow to obtain re- be more representative of voltage behavior sponses with greater time constants than than the voltage of only one node, as occurs those provided by the PVR. with the use of pilot nodes. The proposed SVR scheme with average In this new approach, the PVR only ac- voltage in a specific control zone is schema- counts AVRs voltage controlling; however, tized in Fig. 4. In contrast to conventional other kinds of continuous reactive power SVR (where there is only one control gen- resources could easily be incorporated into erator and pilot node per voltage this scheme.

푉퐺1 푉푟푒푓1 푄퐺1 Δ푉퐴푘 푉푟1 _ _ 푉 푄퐺1 + + 푟1 + AVR 1 Generator 1 + PI Controller 1 + + + 훼1 푉 퐺2 푉푟푒푓2 푄퐺2 Δ푉퐴푘 푉푟2 _ _ 푉 푄퐺2 푄푇 + + 푟2 + AVR 2 Generator 2 + 훼 + PI Controller 2 + + 2

M 푉퐺푛 푉푟푒푓푛 - 푄퐺푛 Δ푉퐴푘 푉푟푛 _ - - -_ - 푉 + 푄- 퐺푛 - - - + + 푟푛 AVR n Generator n + 훼 + PI Controller n + + 푛

Primary Voltage Regulation Secondary Voltage Regulation

Fig. 4. Proposed SVR scheme with average voltage control. Source: Authors’ own work.

[70] TecnoLógicas, ISSN-p 0123-7799 / ISSN-e 2256-5337, Vol. 21, No. 42, mayo-agosto de 2018, pp. 63-78 Secondary voltage regulation based on average voltage control

3. RESULTS AND DISCUSSION tween these two partitions: generators 1, 2, and 3 belong to the same voltage control In order to evaluate the performance of zone, and nodes 6, 11, 12, and 13 are the proposed SVR scheme, the well-known grouped in another similar zone. The dif- 14-node IEEE system has been imple- ferences between these partitions are mented. Its parameters and dynamical mainly due to the electrical distance of the model are described in [15]. A conventional thresholds chosen in each case. As can be SVR was initially applied to this electrical noted in Fig. 5, nodes 8 and 14 are initially system, followed by the proposed SVR excluded from any voltage control zone approach based on average voltages. The because they exceed the defined electrical voltage sensibility matrix of the system distance threshold. The pilot nodes for was defined to apply both schemes. Pilot control zones 1 and 2 correspond to nodes 5 nodes, voltage control zones, a specific and 13, respectively. sensibility threshold, and the short circuit The three generators located in voltage level of all network nodes were selected control zone 1 were included in imple- from this matrix. As a result, Fig. 5 shows mented SVR approaches, i.e. all generators the two defined voltage control zones. in this area were taken as control genera- The resulting system division is similar tors, while in voltage control zone 2 there to that published in [16], where a different is only one control generator, which corre- system partitioning method was used. sponds to generator 6. The reactive power There are some comparable aspects be- participation percentages injected by each

Voltage Control Zone 2

Voltage Control Zone 1

Fig. 5. Defined voltage control zones in 14-node IEEE system. Source: Adapted from [15].

TecnoLógicas, ISSN-p 0123-7799 / ISSN-e 2256-5337, Vol. 21, No. 42, mayo-agosto de 2018, pp. 63-78 [71] Secondary voltage regulation based on average voltage control control generator in zone 1 were main- When network disturbances occur, the tained as indicated by the power flow solu- reactive power flowing between lines 10 tion in the initial operating point. On the and 11 totally disappears and, consequent- other hand, after analyzing the electrical ly, the amount of reactive power delivered distances between nodes, it was decided to the network by generators has to that generator 8 would not participate in change. Therefore, as expected, this situa- any SVR scheme; hence its voltage will tion affects all network voltage levels. Fig- only be controlled by its own AVR. ures 6 and 7 particularly show the node- In order to compare the performance of voltage dynamical responses directly asso- the conventional SVR with the proposed ciated with this failure resulting from the alternative, a topological change was made two strategies under analysis. to the network taking into account electri- Network average voltages during the cally remote nodes with respect to the first stationary condition were taken as a established pilot nodes. For this purpose, reference value, i.e. it was assumed that system voltage profile, i.e. the average this initial power flow reflected an appro- voltage of control zones, was analyzed priated operation point and satisfied some before and after the outage of the trans- constraints defined by the TVR stage. In mission line between nodes 10 and 11, that sense, it is desirable for any SVR which is one of the lines connecting the two strategy applied to power systems to lead defined voltage control zones. This abrupt the voltage profile to a similar condition topological change was made exactly 50 after any contingency, at least until a new seconds into the simulation; hence, before optimal power flow is established to give this time, the system exhibits a quasi- other reference values to the system. How- stationary condition. At this point, it ever, PVR only compensates specific node should be clarified that, in the proposed voltages in the network. Additionally, con- SVR, the voltage of node 14 was integrated ventional SVR acts on a few special load into the calculation of the average voltage nodes, which are identified as the pilot of zone 1 in order to explore the method’s nodes. flexibility. Nevertheless, as it can be veri- Unfortunately, as Fig. 6 shows, it is fied, this small variation did not produce possible that some nodes barely benefit significant effects on results. from conventional SVR implementation. In this work, after the considered

1.054

1.053

1.052 Conventional SVR 1.051 Proposed SVR

1.05

Node 10 voltage (pu) 1.049

1.048 0 50 100 150 200 250 300 Time (s) Fig. 6. Node 10’s voltage response. Source: Authors’ own work.

[72] TecnoLógicas, ISSN-p 0123-7799 / ISSN-e 2256-5337, Vol. 21, No. 42, mayo-agosto de 2018, pp. 63-78 Secondary voltage regulation based on average voltage control

1.068

1.067

1.066

1.065

1.064

1.063

1.062

1.061 Conventional SVR 1.06 Node 11 voltage (pu) Proposed SVR 1.059

1.058 0 50 100 150 200 250 300 Time (s) Fig. 7. Node 11’s voltage response. Source: Authors’ own work. disturbance, the conventional SVR produc- pilot nodes and leads them near the volt- es practically no important effect on node age levels prior to the failure. These previ- 10’s voltage, because only the PVR acts on ous magnitudes were taken as reference this variable. On the other hand, the pro- values for these specific voltage nodes ac- posed SVR at least tries to recover this cording to this traditional approach. Con- voltage to the previous value, because this versely, as can be noted in the same plots, strategy includes its variation in the aver- the proposed SVR leads these voltages to age voltage terms in the algorithm. other levels because this strategy has a In a comparable way, the proposed SVR different purpose, which requires changing scheme identifies a voltage control problem these magnitudes in accordance with the on node 11 due to the line outage and tries desired voltage profile. to improve the average voltage of control The proposed SVR applies an appropri- zone 2 by increasing the reactive power ate reactive power injection in each control flow, which allows to slightly recover this zone through the participant control gen- voltage, as can be seen in Fig. 7. In other erators in order to reach their correspond- words, Figures 6 and 7 allow to visualize ing average voltage reference values. Fig- some relevant effects of the proposed SVR ures 10 and 11 show the average voltages on communicating nodes between different of control zones 1 and 2, respectively, be- voltage control zones. These figures reveal fore and after the disconnection of the the important efforts by this scheme to transmission line. In the first stage of recover the network voltage profile to a transitory response, just after the disturb- level near where it was before the disturbance occurred at second 50, PVR’s fast ance. action is observed. After this event, the Figures 8 and 9 show the voltage re- proposed SVR recovers the average voltage sponses of pilot nodes 5 and 13, respective- in each control zone in about 3 minutes. If ly, due to a transmission line outage locat- that average voltage is accepted as a satis- ed between nodes 10 and 11. Some seconds factory general indicator of network volt- after this disturbance, conventional SVR age profile, it is quite important to realize recovers the voltage magnitudes of the how the proposed SVR can recover this

TecnoLógicas, ISSN-p 0123-7799 / ISSN-e 2256-5337, Vol. 21, No. 42, mayo-agosto de 2018, pp. 63-78 [73] Secondary voltage regulation based on average voltage control

1.0232

1.023

1.0228

1.0226 Conventional SVR 1.0224 Proposed SVR

1.0222 Reference voltage

1.022

1.0218

Node 5 voltage (pu)

1.0216

1.0214 0 50 100 150 200 250 300 Time (s) Fig. 8. Node 5’s voltage response. Source: Authors’ own work.

1.059

1.0585

1.058 Conventional SVR

1.0575 Proposed SVR Reference voltage 1.057

1.0565

1.056

1.0555

1.055

Node 13 voltage (pu) 1.0545

1.054 0 50 100 150 200 250 300 Time (s) Fig. 9. Node 13’s voltage response. Source: Authors’ own work. value in each control zone, while the con- On the other hand, Fig. 12 highlights ventional SVR does not value its signifi- the reactive power responses of all control cance. In that sense, Fig. 10 allows to visu- generators in zone 1 to face the disturb- alize how conventional SVR leads the av- ance when the proposed SVR is used. In erage voltage below the value before the contrast with the traditional SVR ap- failure, leaving it close to the value given proach, the proposed scheme uses several by the PVR stage. This means that on generators in the control zone. In this par- average all node voltages decreased in this ticular case, it uses all of them to manipu- zone. Alternatively, Fig. 11 shows how late the required reactive power and to conventional SVR unnecessarily increases compensate system voltage levels after the the average voltages of all nodes in zone 2. failure.

[74] TecnoLógicas, ISSN-p 0123-7799 / ISSN-e 2256-5337, Vol. 21, No. 42, mayo-agosto de 2018, pp. 63-78 Secondary voltage regulation based on average voltage control

1.0434

1.0432

1.043

1.0428 Conventional SVR Proposed SVR 1.0426 Reference voltage 1.0424

1.0422

1.042

Average zone voltage of 1 (pu) 1.0418 0 50 100 150 200 250 300 Time (s) Fig. 10. Response of average voltage in control zone 1. Source: Authors’ own work.

1.066

1.065 Conventional SVR Proposed SVR 1.064 Reference voltage

1.063

1.062

1.061

Average zone voltage of 2 (pu) 1.06 0 50 100 150 200 250 300 Time (s) Fig. 11. Response of the average voltage of control zone 2. Source: Authors’ own work.

Fig. 12 also reveals how participation where in fact this generator provides most factors are applied to distribute reactive reactive power. Fig. 13 is congruent with power resources of each area in a weighted zone 1’s voltage behavior shown in Fig. 11. way, in contrast to what happens in pres- The proposed SVR requires lower reactive ence of pilot nodes, where a single genera- power injection from the control generator tor is assigned per area to compensate the in control zone 2 than the conventional voltage level of each one of these specific option. This low injection improves re- nodes. For instance, as can be seen in this serves of reactive power to face other pos- figure, after the failure, node 2’s control sible contingencies in this zone but guar- generator contributes to more reactive antees an appropriate voltage profile since power than the other generators. This is it takes into account all node voltages by because participation factors were adjusted average voltage control. according to an initial stationary condition,

TecnoLógicas, ISSN-p 0123-7799 / ISSN-e 2256-5337, Vol. 21, No. 42, mayo-agosto de 2018, pp. 63-78 [75] Secondary voltage regulation based on average voltage control

0.7

0.6

0.5

0.4

0.3 Q1 Q2 0.2 Q3 0.1

Reactivepower inzone 1(pu) 0

-0.1 0 50 100 150 200 250 300 Time (s) Fig. 12. Reactive power resources of the control zone 1. Source: Authors’ own work.

Conventional SVR Proposed SVR 0.4

0.35

Reactivepower ofGenerator 6(pu) 0.3 0 50 100 150 200 250 300 Time (s) Fig. 13. Reactive power resource of the control zone 2. Source: Authors’ own work.

Although many authors have proposed Among these features are its ability to several algorithms for the SVR [10], cur- reach different operation points from set- rently the most-commonly implemented tings provided by TVR due to the flexibility SVR schemes use an approach quite simi- given by its participation factors, and the lar to the one presented in this work: con- possibility of using the voltage measure- ventional SVR with one control generator ments of all the nodes in a zone instead of per voltage control zone. The SVR proposed only one, which could reduce measurement in this work represents a step ahead for errors. this technique because it could offer more However, as a negative aspect, this ap- advantages regarding voltage control. proach will require more communication,

[76] TecnoLógicas, ISSN-p 0123-7799 / ISSN-e 2256-5337, Vol. 21, No. 42, mayo-agosto de 2018, pp. 63-78 Secondary voltage regulation based on average voltage control measurement and control equipment, 5. ACKNOWLEDGMENTS which will increase its implementation costs. As a numerical example comparing The authors appreciate the support by additional costs, the conventional SVR Instituto Tecnológico Metropolitano and applied to the test system only requires Universidad Nacional de Colombia in Me- two devices to measure and communicate dellín to carry out this work. two pilot node voltages to the corresponding regional controllers, but the proposed SVR requires at least 13 of them, because 6. REFERENCES almost all the nodes in the system have to participate in this strategy. [1] V. Venkatasubramanian et al., “Hierarchical Moreover, this approach should incor- Two-Level Voltage Controller for Large Power Systems,” IEEE Trans. Power Syst., porate a specific control algorithm able to vol. 31, no. 1, pp. 397–411, Jan. 2016. adjust the system configuration online and, [2] V. Alimisis and P. C. Taylor, “Zoning consequently, reorganize measured and Evaluation for Improved Coordinated control signals for each established region- Automatic Voltage Control,” IEEE Trans. al regulator. Therefore, it will be necessary Power Syst., vol. 30, no. 5, pp. 2736–2746, Sep. 2015. to hire more qualified and trained engi- [3] A. Morattab, O. Akhrif, and M. Saad, neers to maintain the system’s operation, “Decentralised coordinated secondary voltage which in general terms will demand a deep control of multi-area power grids using cost/benefit analysis to evaluate the con- model predictive control,” IET Gener. Transm. Distrib., vol. 11, no. 18, pp. 4546– venience of this strategy in each case. 4555, Dec. 2017. [4] Q.-Y. Liu, C.-C. Liu, and Q.-F. Liu, “Coordinated Voltage Control With Online 4. CONCLUSIONS Energy Margin Constraints,” IEEE Trans. Power Syst., vol. 31, no. 3, pp. 2064–2075, May 2016. The 14-node IEEE system was divided [5] A. Morattab, A. Dalal, O. Akhrif, M. Saad, into two voltage control zones using the and S. Lefebvre, “Model Predictive electrical distance technique. As a result, a Coordinated secondary voltage control of partition like those presented by other power grids,” in 2012 International Conference on Renewable Energies for researchers (who applied a different algo- Developing Countries (REDEC), 2012, pp. 1– rithm for this task) was obtained. 6. The proposed SVR led the average volt- [6] H. Lefebvre, D. Fragnier, J. Y. Boussion, P. age of each control zone to its previous Mallet, and M. Bulot, “Secondary values after the disturbance under analy- coordinated voltage control system: feedback of EDF,” in 2000 Power Engineering Society sis. This strategy maintained the partici- Summer Meeting (Cat. No.00CH37134), pation percentages of all control generators 2000, vol. 1, pp. 290–295. according to the reactive power injected by [7] J. Rios, A. Zamora, M. R. A. Paternina, A. each of them before the failure. This al- Lopez, and E. Vazquez, “Secondary voltage control areas through energy levels,” in 2016 lowed to conveniently distribute the control IEEE PES Transmission & Distribution efforts between these devices. Conference and Exposition-Latin America This approach considers all node volt- (PES T&D-LA), 2016, pp. 1–6. ages in each zone, which improves the [8] H. Vu, P. Pruvot, C. Launay, and Y. recognition of different voltage problems in Harmand, “An improved voltage control on large-scale power system,” IEEE Trans. the network. Furthermore, even disturb- Power Syst., vol. 11, no. 3, pp. 1295–1303, ances in nodes distant from pilot nodes are 1996. detected and control actions are taken to [9] G. Grigoras, B.-C. Neagu, F. Scarlatache, address this phenomenon. and R. C. Ciobanu, “Identification of pilot nodes for secondary voltage control using K-

TecnoLógicas, ISSN-p 0123-7799 / ISSN-e 2256-5337, Vol. 21, No. 42, mayo-agosto de 2018, pp. 63-78 [77] Secondary voltage regulation based on average voltage control

means clustering algorithm,” in 2017 IEEE regulation: dynamic simulation and 26th International Symposium on Industrial continuation power flow results,” in 2001 Electronics (ISIE), 2017, pp. 106–110. Power Engineering Society Summer Meeting. [10] International Council on Large Electric Conference Proceedings (Cat. Systems-CIGRE, “Coordinated voltage No.01CH37262), 2001, pp. 791–796. control in transmission networks,” 2007. [14] P. Lagonotte, J. C. Sabonnadiere, J.-Y. Leost, [11] N. A. Daher, I. Mougharbel, M. Saad, and H. and J.-P. Paul, “Structural analysis of the Y. Kanaan, “Comparative study of electrical system: application to secondary partitioning methods used for secondary voltage control in France,” IEEE Trans. voltage control in distributed power Power Syst., vol. 4, no. 2, pp. 479–486, May networks,” in 2013 IEEE International 1989. Conference on Smart Energy Grid [15] F. Milano, Power system modelling and Engineering (SEGE), 2013, pp. 1–7. scripting, Illustrate. London: Springer [12] Bo Hu, C. A. Canizares, and M. Liu, Science & Business Media, 2010. “Secondary and Tertiary Voltage Regulation [16] M. K. Verma and S. C. Srivastava, based on optimal power flows,” in 2010 IREP “Approach to determine voltage control areas Symposium Bulk Power System Dynamics considering impact of contingencies,” IEE and Control - VIII (IREP), 2010, pp. 1–6. Proc. - Gener. Transm. Distrib., vol. 152, no. [13] N. Martins, J. C. R. Ferraz, S. Gomes, P. E. 3, pp. 342–350, 2005. M. Quintao, and J. A. Passos, “A demonstration example of secondary voltage

[78] TecnoLógicas, ISSN-p 0123-7799 / ISSN-e 2256-5337, Vol. 21, No. 42, mayo-agosto de 2018, pp. 63-78