Uni®ed Power Ow Controller (UPFC) Model in The Framework of Interior Point Based Active and Reactive OPF Procedure

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 7

Electrical Power and Energy Systems 24 (2002) 431±437

www.elsevier.com/locate/ijepes

Uni®ed power ¯ow controller (UPFC) model in the framework of interior


point based active and reactive OPF procedure
M.V. Cazzol*, A. Garzillo, M. Innorta, M.G. Libardi, M. Ricci
CESI S.p.A., Via Rubattino, 54 20134 Milano, Italy

Abstract
As the deregulation process of the energy market is taking place, the operation planners are faced with heavily loaded networks and with
greater wheeling demand. Therefore, in order to increase the network ¯exibility, the exploitation of Flexible AC Transmission Systems
(FACTS) devices becomes more and more signi®cant. For this reason, the system operator needs an extension of the traditional Optimal
Power Flow programs capable of managing the models of these new devices.
In this paper, a model of the Uni®ed Power Flow Controller (UPFC) is described and introduced in the framework of an Interior Point
based OPF procedure and once again the good performances of this solver have been demonstrated. Finally, the paper presents some test
cases showing the gains obtainable by the exploitation of UPFC managing wheeling and/or congestion situations. q 2002 Elsevier Science
Ltd. All rights reserved.
Keywords: Uni®ed power ¯ow controller; Transmission systems; Interior point methods

1. Introduction technical (equipment capability, system losses, overall


reliability, security and quality of the power supply) and
Power electronic equipment, named FACTS (Flexible economic aspects [1±3].
AC Transmission Systems), has been applied on some Planning and management of production and transmis-
networks for the rapid control of power ¯ows on lines or sion of electric systems require sophisticated calculation
along corridors. The use of FACTS devices increases the instruments in order to optimise resource allocation and
electric power system ¯exibility under variable operating exploitation, while ensuring the requirements of service
conditions (power infeed and demand) and allows a rise in security and quality.
the power transfer capability of the network. The problem of performing optimised load-¯ow calcula-
A wider application of power electronic equipment in tions (Optimal Power Flow) gives rise to a non linear
transmission networks could be envisaged in the future of programming model which results to be particularly
Italian electrical power transmission system in a competi- complex because of the high number of variables and
tive electricity market with a more intensive wheeling constraints associated to the requirements of service
service required and in presence of increasing amount of security and quality that have to be guaranteed.
generation from Independent Power Producers (IPP), not Recently, the development of the interior point methods
under the direct control of the Transmission System Opera- and their extension to non linear optimisation models have
tor (TSO). The need of these devices also arises from the allowed to successfully face the problem of coupled active±
dif®culties to build new lines or to modify the existing ones, reactive dispatching, thanks to their ability to solve large
due to public concern on the environmental impact. Further- scale problems (thousands of variables and constraints)
more, the use of these devices could contribute to the within acceptable calculation times. An effective descrip-
opening of the European electricity market increasing the tion of this method and its application can be found, for
power transfer capability across international borders. example, in Refs. [4±6].
The power ¯ow control in a meshed transmission system A detailed description of the models for the different
implies the use of different devices in order to comply with kinds of FACTS devices (Static VAR Compensators, Series
the system requirements based on a trade off between Capacitive Compensators, Phase Shifters, Uni®ed Power
Flow Controllers (UPFC)) can be found in Refs. [7±8].
* Corresponding author. Tel.: 139-2-21255841; fax: 139-2-21255579. The paper presents a mathematical model of UPFC,
E-mail address: cazzol@cesi.it (M.V. Cazzol). which is used in the context of an interior point based active
0142-0615/02/$ - see front matter q 2002 Elsevier Science Ltd. All rights reserved.
PII: S 0142-061 5(01)00055-2
432 M.V. Cazzol et al. / Electrical Power and Energy Systems 24 (2002) 431±437

and reactive OPF procedure. Therefore, the proposed OPF wherein N is the number of nodes of the network, Ngter the
procedure supplies the standard state and control network number of thermal generators of the network, Ng the number
variables (active and reactive power productions, voltage of generators of the network, Ntap the number of tap-
magnitudes and phases etc.) as well as the UPFC internal changers of the network and Nc is the number of controlled
variables satisfying their operation limits. Although the connections of the network.
most important task required to UPFC is to redirect the Variables relevant to the controlled current ¯ows on lines
active power ¯ows, their ability to control the voltage also and transformers are `dynamically' considered by the pro-
requires, as suggested, for example in Ref. [8], the adoption cedure. On each iteration of the algorithm, a check is
of the AC coupled active and reactive network model in performed to identify possible violated transit constraints.
order to take into account the in¯uence of the voltages on If the current on a connection is higher than the relevant
the network power ¯ows and losses. thermal limit, the associated constraints from the two sides
The paper is organised as follows: of the connection and the corresponding variables TI are
introduced in the model and controlled.
Section 2 presents the full set of equations that describe The size of the problem is, therefore, a minor one at the
the base coupled active and reactive OPF model. beginning of the solution process and may increase during
Section 3 introduces the adopted UPFC model by the various iterations, following the activation of current
de®ning its internal variables and its impact on the constraints.
power system. This type of procedure guarantees the satisfaction of
Section 4 describes the modi®cations to the base OPF security requirements, making explicit only a small percent-
required to take into account of the UPFC models in the age of current constraints on lines and transformers.
transmission system.
Section 5 presents some test cases showing the advant- 2.2. Objective function
ages that can be obtained by a suitable installation of
UPFC to cope with wheeling or/and congestion situations The objective of the procedure is to supply a feasible
as well as the excellent performance of the interior point operation point for the network while optimising the
method in handling the non-linear constraints introduced adopted operational criterion. In particular, it is realistic to
by the new devices. choose as objective function the total cost of thermal
generators.
From the analytical viewpoint, by assuming a quadratic
cost function for each generator, one must minimise the
2. Formulation of the OPF base problem following objective function:
Ngter
X
Coupled active and reactive power dispatching has been OF ˆ …c0i 1 c1i Pi 1 c2i P2i †
formulated by making load-¯ow equations explicit as equal- iˆ1
ity constraints, adopting therefore, a sparse type model.
In the following, the corresponding base optimisation wherein c0i ; c1i ; c2i are the cost coef®cients of the i-th
problem in its formulation in terms of variables, objective generator.
function and constraints is described.
2.3. Constraints
2.1. Variables of the problem The constraints of the problem may be grouped as
follows:
These are:
² Equations of nodal balance of active and reactive power
modules and phases of voltages in all nodes (V and q ) (load-¯ow equations) (2 N equality constraints)
(excluding node 1 phase, ®xed at 0), X X
active and reactive powers produced by the thermal TAi; j 1 IAi ˆ 0 TRi; j 1 IRi ˆ 0 i ˆ 1; ¼; N
groups and reactive powers produced by the hydraulic j[ai j[ai

groups and the synchronous compensators (P and Q), with


values of rated voltage in the control terminal of !
tap-changer transformers (VT variables) Vi Vj Ki; j sin di; j gi; j
TAi; j ˆ sin…qi 2 qj 2 di; j † 1 Vi2 1
current ¯ows on controlled connections (TI). zi; j zi; j 2

The vector of the problem variables is, therefore: !


Vi Vj Ki; j cos di; j yi; j
TRi; j ˆ2 cos…qi 2 qj 2 di; j † 1 Vi2 2
u ˆ bq2 ; ¼; qN ; V1 ; ¼; VN ; P1 ; ¼; PNgter ; Q1 ; ¼; QNg ; zi; j zi; j 2

 VT1 ; ¼; VTNtap ; TI1 ; ¼; TINc c where TAi, j is the transit of active power on i, j connection,
M.V. Cazzol et al. / Electrical Power and Energy Systems 24 (2002) 431±437 433

positive if from i to j; TRi, j is the transit of reactive power reactive power and the DC voltage by exchanging active
on i, j connection, positive if from i to j; IAi and IRi are power.
the injections of active and reactive power in node i In this model, the losses are neglected, therefore the
(positive if coming out of node i); a i is the set of nodes active power Pu from node j is entirely transferred in to
connected to node i; zi, j is the module of impedance of the node i.
connection i, j; gi, j is the shunt conductance of connection The equations which describe the steady state working
i, j; d i, j is the loss angle of impedance of connection i, j; point and link the internal variables (Vt , Iq , Psh ) to the
yi, j is the module of shunt susceptance of connection i, j; voltage magnitude and phase angle of nodes i and j
and Ki, j is the transformation ratio of the transformer on …Vi ; Vj ; ui ; uj † and to the active and reactive injections
connection i, j(Ki;j ˆ 1 for the lines). …2Pu ; Qui † and …Pu ; Quj † are:
² Equations of current transits for security (Nc equality q
constraints) Vt ˆ …Vj sin…uj 2 ui ††2 1 …Vi 2 Vj cos…uj 2 ui ††2 …1†
1 q
TIh ˆ p TA2i;j 1 TR2i;j h ˆ 1; ¼; Nc
3 Vi
Q Quj P
with: TIh ˆ variable relevant to the current transit on the Iq ˆ 2 pui 2 p cos…uj 2 ui † 1 pu sin…uj 2 ui † …2†
3 Vi 3V j 3 Vj
connection h between nodes i and j.
² Upper and lower bound constraints on problem variables
2…N 1 Ngter 1 Ng 1 Ntap 1 Nc † Vi V
Psh ˆ 2Pu 1 Pu cos…uj 2 ui † 1 i Quj sin…uj 2 ui † …3†
Vmin;i # Vi # Vmax;i i ˆ 1; ¼; N Vj Vj

Pmin;i # Pi # Pmax;i i ˆ 1; ¼; Ngter The ®rst equation derives easily from the following
vector expression: Vj ˆ Vi 1 Vt
Qmin;i # Qi # Qmax;i i ˆ 1; ¼; Ng The second and the third of the previous equations are
obtained by the explicit formulation of reactive and active
VTmin;i # VTi # VTmax;i i ˆ 1; ¼; Ntap power balance equations on i-bus (see Appendix):

TImin;i # TIi # TImax;i i ˆ 1; ¼; Nc Qsh ˆ Qui 1 Quj 1 Qseries

Psh ˆ Pui 1 Puj 1 Pseries


3. UPFC model
where:
The basic con®guration of UPFC device installed p
between i±j busses is presented in Fig. 1. This device, by Pserie 1 jQserie ˆ 2 3Vt Ijp :
using two gate turn-off solid state converters, allows to
insert a series voltage of any phase angle and magnitude
(Vt less than Vtmax) used to control the power ¯ow on the Pui ˆ 2Pu ; Puj ˆ 2Pu
transmission line and to insert on node i a shunt compensa-
tion (Iq ) used to control the AC voltage by exchanging with the hypothesis of no losses in the UPFC device.

Fig. 1. UPFC model.


434 M.V. Cazzol et al. / Electrical Power and Energy Systems 24 (2002) 431±437

4. Modi®cations of base OPF for a network with UPFC Table 2


devices UPFC characteristics

Anom (MVA) Vnom (kV) Vtmax (kV) Ashmax (MVA) Pshmax (MW)
Some modi®cations are needed to the OPF formulation
previously described for managing UPFC devices. 2000 400 70 400 200

4.1. The new variables of the problem


allow to take into account the effective regulation capability
The vector u of problem variables increases for the of the devices.
presence of the following six variables for each UPFC
installed on the network
5. Some results of applications
internal variables (Vt ,Iq ,Psh ) in the UPFC devices;
active power ¯ow in UPFC devices (Pu ) (the internal The proposed automatic OPF procedure incorporating the
losses are neglected); UPFC model has been tested on a network of 1300 nodes,
reactive power at the nodes i and j (Qui ,Quj ). 1800 links (lines and transformers) and about 300 thermal
and hydro generation plants.
This network represents the Italian transmission system
4.2. New constraints and a meaningful part of the electrical European system so
that the whole set of the active and reactive power
The set of constraints must include the following three exchanges with foreign countries can be evaluated by an
new equations linking the internal variables of UPFC with OPF procedure. In the examined base situation without
the other ones: UPFC devices, the total active power imported by the Italian
q system is about 6000 MW.
Vt;k ˆ …Vj sin Du†2 1 …Vi 2 Vj cos Du†2 In this base situation, three different pairs of busses, the
®rst one in the Italian system (receiving node) and the
Q Quj;k P second one in a foreign country (sending node), have been
Iq;k ˆ 2 pui;k 2 p cos Du 1 pu;k sin Du selected and the related active injections, simulating an
3´Vi 3´Vj 3´Vj
active production in the foreign node and a consumption
k ˆ …1; ¼; Nupf † in the Italian node, have been increased more and more
until the maximum feasible wheeling has been reached. For
Vi V this task, the base OPF procedure has been employed and
Psh;k ˆ 2Pu;k 1 Pu;k cos Du 1 i Quj;k sin Du only the Italian thermal generators have been rescheduled.
Vj Vj
Subsequently, for each one of the three different wheeling
where i, j are the indices of the k-th UPFC buses; situations, an UPFC have been installed on a suitable site of
Du ˆ uj 2 ui ; Nupf is the number of UPFC. the network in order to obtain a further active power transfer
Upper and lower bound constraints on the new variables from the foreign countries. For this second task, the OPF
are: procedure including the UPFC model has been used. The
analysis of power ¯ows on the interconnection lines has
0 # Vt;k # Vtmax;k suggested as site for UPFC installation the interconnection
2Iqmax;k # Iq;k # Iqmax;k line that, in the examined test cases, presented the lowest
2Pshmax;k # Psh;k # Pshmax;k k ˆ …1; ¼; Nupf † p.u. current.
2Pumax;k # Pu;k # Pumax;k For the three test cases, the relevant sending and receiving
2Quimax;k # Qui;k # Quimax;k nodes, the UPFC installation site and the resulting maxi-
2Qujmax;k # Quj;k # Qujmax;k mum wheeling (without and with UPFC) found by OPF
program are described in Table 1.
The upper and lower bound on the internal variables The main characteristics of the UPFC installed in all the

Table 1
Test cases

Case test Wheeling nodes Without UPFC With UPFC

Sending Receiving Maximum wheeling (MW) Site Maximum wheeling (MW)

Case 1 Albertville (France) Rondissone (Italy) 450 Rondissone 1000


Case 2 Laufenburg (Switzerland) Baggio (Italy) 650 Rondissone 1000
Case 3 Podlog (Slovenia) Ferrara (Italy) 1550 Redipuglia 1850
Table 3
Active and reactive power ¯ows and currents in interconnection links for maximum wheeling without UPFC

Interconnection links Active power ¯ow (MW) Reactive power ¯ow (MVAr) Current ¯ow (A) Maximum current ¯ow (A) Utilisation of the link (%)

Padriciano Divaccia 2 287.007 76.898 718.862 720 99.84


Redipuglia Divaccia 2 1196.92 35.584 1659.817 2400 69.16
Soverzene Lienz 2 266.715 21.656 686 686 100

Table 4
Active and reactive power ¯ows and currents in interconnection links for maximum wheeling whit UPFC
M.V. Cazzol et al. / Electrical Power and Energy Systems 24 (2002) 431±437

Interconnection links Active power ¯ow (MW) Reactive power ¯ow (MVAr) Current ¯ow (A) Maximum current ¯ow (A) Utilisation of the link (%)

Padriciano Divaccia 2 238.874 131.211 661.484 720 91.87


Redipuglia Divaccia 2 1498.52 116.227 2156.023 2400 89.83
Soverzene Lienz 2 277.6 43.738 683.21 686 99.59
435
436 M.V. Cazzol et al. / Electrical Power and Energy Systems 24 (2002) 431±437

three cases are summarised in Table 2 where Anom and Vnom 6. Conclusions
are the UPFC nominal apparent power and nominal voltage,
Vtmax is the maximum value of the series voltages and Ashmax, An UPFC model has been introduced in the integrated
Pshmax are the maximum values for apparent and active shunt active and reactive OPF procedure and the modi®cations of
powers. the variable and constraint sets have been described. The
The performed tests have shown a very interesting new proposed procedure is able to fully exploit the UPFC
increase of transfer capability of the interconnection lines control capabilities and to automatically handle the
when the UPFC are installed. The same site and size have co-ordination of the UPFC internal control variables with
been used both for `case 1' and for `case 2' while for `case the other control variables of the electrical system, such as
3' a new site has been chosen. The problem of selecting the generator active and reactive power productions. Therefore,
location for UPFC devices in order to face a set of different the procedure allows a more accurate evaluation of the
operative scenarios minimising the investment cost has not UPFC impact on the objective function, on the active and
been considered in this paper, therefore, the location in the reactive power losses and on the voltage levels in different
test cases derives from heuristic considerations on the operation scenarios. Computer simulation results have
different load levels of the interconnection lines. con®rmed the effectiveness of the proposed approach both
The active and reactive power ¯ows on the interconnec- in relieving the load of interconnection lines and in increas-
tion lines mainly affected by the wheeling scenario of the ing the wheeling capability.
`case 3' are described in Table 3 and in Table 4. The
currents (%) in Table 3 show that the wheeling limit is
due to the saturation of the 240 kV line Soverzene±Lienz,
Appendix A. De®nition of the constraints required by
while the link of the same corridor Redipuglia±Divaccia
UPFC devices
contributes to the power transmission only with the 69.1%
of its capability because of the network topology and the
The following considerations are referred to the symbols
power ¯ow control capacity of schedulable generators. This
and diagram presented in Fig. 1 (Section 3).
asset does not fully exploit the transmission capability of
The balance active and reactive power equations on bus i
the 400 kV Redipuglia±Divaccia line and underlines a
are:
bottle-neck in the transmission system.
The advantages obtainable by the installation of an UPFC (
Qsh ˆ Qui 1 Quj 1 Qserie
device are described in Table 4, which shows a remarkable
increase in the active power import and a better utilisation of Psh ˆ Pui 1 Puj 1 Pserie
the interconnection lines.
In fact the Redipuglia±Divaccia line appears more loaded where:
and the new operation point exhibits more equilibrate power p
¯ows. Pserie 1 iQserie ˆ 2 3Vt I pj …A1†
The advantage of using UPFC devices in the transmission
system is measured not only by the increase of the real I pj represents the conjugate complex of the current on link
power transfer capability but also by the improvement of j±i, that is the current in the high voltage circuit of the series
the objective function and voltages pro®le with a con- transformer. p
sequent reduction of the active power losses. To evaluate By using the relation, Pu 1 iQuj ˆ 3V j I pj , we get:
this improvement, the situation corresponding to the wheel-
ing limit without UPFC for `case 3' has been optimised also P Quj
I pj ˆ pu 1 i p …A2†
with the UPFC installation. 3Vj 3Vj
The results of this test are presented in Table 5 where the
signi®cant reduction both in the total thermal cost and in the and Eq. (A1) can be reformulated as:
network active losses represents a saving not negligible in
the assessment of the economic advantage obtainable by a Pu V t Q V
proper UPFC installation. Pserie 1 iQserie ˆ 2 2 i ui t …A3†
Vj Vj

Table 5
Moreover, since V t ˆ V j 2 V i , it becomes:
Optimisation results (wheeling ˆ 1550 MW) with and without UPFC
Pu …V j 2 V i † Qui …V j 2 V i †
Case 3 Wheeling ˆ 1550 MW Without UPFC With UPFC Pserie 1 iQserie ˆ 2 2i …A4†
Vj Vj
FO (k£/h) 2,69,99,426 2,66,76,800
Active losses (MW) 868 765
Exploiting the polar notation for the voltages on bus i and
Reactive losses (Mvar) 19,449 14,371
j, by easy algebraic manipulations, the following complex
M.V. Cazzol et al. / Electrical Power and Energy Systems 24 (2002) 431±437 437

From the second equation, neglecting the internal UPFC


Vi losses …Pui ˆ 2Puj †; we get the Eq. (3) of Section 3:

If Pu V i Quj Vi
Psh ˆ 2Pu 1 cos…qj 2 ui † 1 sin…qj 2 ui † …A8†
Vj Vj
p
Dividing the ®rst equation by 2 3Vi , we can easily
Ish obtain the expression of the shunt compensation Iq taking
into account the following relations:
p p
Psh 1 iQsh ˆ 2 3V i I psh ˆ 2 3Vi Ish cos…ui 2 w†
ϕ p
θi 2 i 3VIsh sin…ui 2 w† …A9†
I sh ˆ I f 1 I q If ˆ Ish cos…ui 2 w†
Iq
Iq ˆ Ish sin…ui 2 w†
where I f and I q are the components of I sh de®ned in Fig. 2.
In this way we obtain:
Fig. 2. Components of I sh .
Q P Quj
Iq ˆ 2 pui 1 pu sin…qj 2 ui † 2 p cos…qj 2 ui †
equation is obtained: 3 Vi 3V j 3 Vj
that is the Eq. (2) of Section 3.
P V
Pserie 1 iQserie ˆ 2Pu 2 iQuj 1 u i cos…qi 2 uj †
Vj
References
PV Quj Vi
1 i u i sin…qi 2 uj † 1 i
Vj Vj [1] Ge SY, Chung TS. Optimal active power ¯ow incorporating power
¯ow control needs in ¯exible AC transmission system. IEEE Trans
Quj Vi Power System 1999;14(2):738±44.
 cos…qi 2 uj † 2 sin…qi 2 uj † …A5† [2] Grif®n J, Atanackovic D, Galiana FD. A study of the impact of FACTS
Vj
on the secure-economic operation of power systems. Proceedings of
that is: the 12th PSCC, Dresden, August 1996.
[3] Garzillo A, Innorta M, Migliardi P, Nicola G, Cazzol MV, Ricci M.
8
> Pu Vi Quj Vi Location and sizing of power electronics based equipment for a more
>
> P ˆ 2Pu 1 cos…qj 2 ui † 1 sin…qj 2 ui † ¯exible and intensive use of meshed transmission networks, CIGRE
< serie Vj Vj Symposium, Tours (F), 1997.
>
> Pu Vi Quj Vi [4] Blanchon G, Dodu JC, Renaud A, Bouhtou M. Implementation of a
>
: Qserie ˆ 2Quj 2 sin…qj 2 ui † 1 cos…qj 2 ui † primal±dual interior-point method applied to the planning of reactive
Vj Vj power compensation devices, PSCC, Dresden, 1996.
…A6† [5] Wu YC, Debs AS, Marsten RE. A nonlinear programming approach
based on interiorpoint method for Optimal Power Flows, IEEE/NTUA
Using the last equations, the active and reactive power Athens Power Tech Conf, Athens, 1993.
balance equations for the bus i can be rewritten as: [6] Garzillo A, Innorta M, Ricci M. The problem of the active and reactive
8 optimum power dispatching solved by utilising a primal±dual interior
> Pu Vi Quj Vi point method. Int J Electl Power Energy Systems 1998;20(6).
>
> Q ˆ Qui 1 Quj 2 Quj 2 sin…qj 2 ui † 1 cos…qj 2 ui †
< sh Vj Vj [7] Gyugyi L. Solid-state control of electric power in AC transmission
> systems, EECPS-CAPRI±May 1989.
>
> Pu Vi Quj Vi
: Psh ˆ Pui 1 Puj 2 Pu 1 cos…qj 2 ui † 1 sin…qj 2 ui † [8] Keri AJF, Mehraban AS, Lombard X, Elriachy A, Edris AA. Uni®ed
Vj Vj power ¯ow controller (UPFC): modelling and analysis. IEEE Trans
…A7† Power Delivery 1999;14(2):648±54.

You might also like