Response-based Control through Dynamic Optimization

in Large-scale Power Systems

Sergio Bruno, Matteo D’Aloia and Massimo La Scala

Dipartimento di Ingegneria Elettrica e dell’Informazione (DEI), Politecnico di Bari, via Re David 200, 70125, Bari, Italy

Keywords: Power Systems, Wide Area Measurement Systems, Phasor Measurement Units, Transient Security,

Dynamic Optimization, Real-time Power System Simulation.

Abstract: Modern real-time measurement equipments and associated communication/computing networks allow

developing advanced power system control architectures able to identify dangerous states of power systems,

and, when necessary, evaluate and apply remedial control actions. An approach, derived by a dynamic

optimization methodology, for evaluating response-based control actions and enhance power system

security, is presented. In this paper, studies performed for wide-area control of transient phenomena are

reviewed. Despite severe computational efforts and time requirements, the authors’ position is that the

nowadays technology can make the approach feasible.

1 INTRODUCTION

New smart technologies are the current drive for

achieving profound modifications in the way power

transmission system are built and operated. The

pervasive diffusion of power converters and flexible

systems, large storage facilities, adaptive relays,

smart meters and sensors allows the achievement of

improved levels of automation, intelligence,

monitoring and control in power systems (Momoh

2009); (Bose 2010).

A key enabling technology for such change is to

be found in Phasor Measurement Units (PMUs)

(Phadke and Thorp, 2006) whose deployment in

power systems has been possible thanks to the on-

going process of substitution of old

electromechanical relays with new digital relays,

started in the early 90s and under completion in most

advanced countries. PMUs are devices that allow to

measure synchronously electric variables (currents

and voltages, in module and phase) in several nodes

of the network, giving a real-time realistic snapshot

of system conditions and state.

According to the vision presented in (Bose et al.,

2004a; 2004b), data coming from PMUs can be

collected and elaborated in order to achieve real-

time control of power systems. The diffusion of

IEC-61850 protocol-based equipments, will

facilitate the integration of such devices into

SCADA/EMS data base and simulation tools, and

will achieve, at the same time, interoperability of all

active elements of the network (for example digital

protection relays, power converters and flexible AC

transmission systems - FACTS, remote controlled

switches, etc.).

Such real-time control architecture can be based

on a revolution of classical dynamic security

assessment (DSA) functions that are traditionally

based on off-line models and simulations and on

“event-based” predetermined control rules.

The idea is to develop powerful real-time tools

for on-line updating of power system parameters

database (usually stored in SCADA and updated

only occasionally), power system behaviour

prevision and real-time control assessment, leading

to the definition of a “response-based” real-time

control approach to power system security (Taylor et

al., 2005).

This approach is based on the integration of

dynamic optimization methodologies with modern

real-time measurement equipments (PMUs and wide

area measurement systems - WAMS) and is aimed at

evaluating real-time corrective control actions, as

soon as degraded dynamic trajectories are detected.

Since unstable transients must be controlled

within hundreds of milliseconds from the insurgence

of the fault, in order to implement “response based”

remedial actions, fast actuators are needed.

Load/generation shedding can be fast enough to

210

Bruno S., D’Aloia M. and La Scala M..

Response-based Control through Dynamic Optimization in Large-scale Power Systems.

DOI: 10.5220/0004593202100216

In Proceedings of the 3rd International Conference on Simulation and Modeling Methodologies, Technologies and Applications (SIMULTECH-2013),

pages 210-216

ISBN: 978-989-8565-69-3

 2013 SCITEPRESS (Science and Technology Publications, Lda.)

correct the undesirable transient behaviour (Bose et

al., 2004b) but more advanced fast control actions

can be implemented through FACTS devices (Bruno

et al., 2012a), line switching techniques (Bruno et

al., 2012b) or fast tuning of adaptive relays (Bruno

et al., 2012c). In addition, real-time tuning of

distance relays settings performed considering

actual, or forecasted, configurations of the power

system will allow to overcome the classical conflict

between dependability and security (Bruno et al.,

2011) and to avoid improper operation of such

devices in vulnerable conditions and during major

blackout events.

In this paper, the authors’ experience, devoted to

assess the feasibility of advanced power system

control architectures, is reviewed. The paper

presents the authors position about the possible

implementation of response-based control

architectures for controlling transient phenomena on

a wide geographical scale. Despite the severe

computational effort required by the response-based

simulations and the tight time requirements, the

authors’ position is that the nowadays technology

can make the approach feasible.

2 EVOLUTION OF CONTROL

ARCHITECTURES

In this section, an advanced control architecture

oriented at evaluating and implementing control

actions as soon as power system dynamic behaviour

worsens is proposed. This architecture, basing on

real-time measurements and fast simulation tools,

guarantees a substantial evolution with respect to

event-based control approaches.

2.1 Event-based Control Strategy

The control of power system stability is mainly

operated by means of automated control systems that

feedback local signals to generation unit control

systems (automatic voltage regulators, power

systems stabilizers, governor, etc.) (Kundur, 1994).

This kind of control enhance power system stability

behaviour (especially small-signal stability) but it is

not fit to ensure stability when large and severe

transients are experienced. Moreover, local control

schemes cannot provide enough information on

phenomena arising on wide geographic areas and

detectable only on wide-area spatial scale (for

example inter-area oscillations, voltage instability,

etc.).

Large oscillations and transients are usually

controlled by means of another class of controls that

includes both preventive and corrective control

approaches. These controls are operated via

SCADA/EMS by local relays or by means of Special

Protection Systems (SPS) (Taylor, 2000).

Corrective schemes operate following an event-

based control approach (Figure 1) that requires no

intervention of system operators: defence schemes

react automatically when an event is detected (event-

driven actions). SPS working principle is based on

the pre-arming of actuators that react automatically

following pre-defined control strategies at the

detection of a certain disturbance. The arming of

SPS is based on off-line (preventive) simulations,

performed on the basis of a predefined set of

postulated contingencies (event-based approach).

Since this approach is based on off-line

simulations, off-line models and on a set of

postulated contingencies, it might fail anytime

unexpected events or anomalous conditions are

experienced. The failure of the automatic load

shedding scheme called EDA (Elaboratore Distacchi

Automatici) in avoiding disaster during the major

Italian 2003 blackout is a clear example of such

problematic (Berizzi 2004); (Berizzi and Sforna

2006).

Figure 1: Event-based control strategy.

2.2 Response-based Control Approach

Given the actual availability of advanced

measurement and communication systems, it is

possible to imagine a new control architecture for

response-based control of power systems. Real-time

measurement equipments and associated

communication systems (i.e. PMU and WAMS) can

be exploited for developing such advanced control

approach.

In the proposed control scheme, power system

trajectories, acquired in real-time, allow the

identification of threats to system security and of

degraded dynamic states. If necessary, through

simulation or sensitivity analysis, suitable corrective

control actions can be evaluated and implemented

sending in real-time corrective signals to any fast

Response-basedControlthroughDynamicOptimizationinLarge-scalePowerSystems

211

actuator device (e.g. load/generation shedding

schemes, FACTS devices, line switching, adaptive

relaying).

A schematic representation of a response-based

control system is given in Figure 2. Such scheme

reacts to static and dynamic constraints violations by

means of control actions calculated and

implemented on transient time-scale.

Figure 2: Response-based control approach.

A response-based control strategy, integrated

with WAMS, has the potential to grasp the whole

picture of dynamic system behaviour and perform

calculations on-the-fly on the basis of actual

dynamical state, taking also into account possible

sudden changes in grid topology of operating

conditions as usually experienced during severe

cascading events leading to major blackouts.

The proposed approach when integrated into a

WAMS architecture gives rise to what is defined

Wide Area Measurement and Control system or

WAMC.

2.3 Wide-area Measurement

and Control Architecture

The proposed WAMC architecture requires the

existence of a measurement and communication

network that collects and distributes real-time

information about system state. PMUs can be easily

located in strategic points of the system (substations,

generation points or important interconnection

points), giving real-time measurements of voltage

and current phasors, suitably synchronized through

GPS time signals.

A Phasor Data Concentrator (PDC) collects data

sent continuously by PMUs and exports

measurements as soon as they have been correlated

and normalized. Real-time data sent from all PDCs

to the Control Centre provide a coherent picture of

the system state. Data are elaborated through real-

time power system simulators reproducing power

system trajectories and detecting possible threats to

system security.

The proposed approach is based on creating a

dynamic replica of power system dynamics,

permitting the assessment of system dynamic

performances and a rapid calculation of real-time

control actions through sensitivity approaches or

through the solution of dynamic optimization

problems.

In smart transmission grids, both innovative IT

infrastructure and the standardization process have

been implemented. This scenario enables the

achievement of important objectives, such as

seamless interoperability and fast communication

and information exchange. The International

Electrotechnical Commission IEC has produced the

IEC TC57 standards reference framework that

defines two important standard families: IEC 61970,

also known as Common Information Model (CIM),

and IEC 61850 for substation communication.

Control centres, thanks to IEC TC57 standards,

can monitor and control any field device. It is

foreseeable that control centre operators will be able

to gather real-time information about any installed

device; in this scenario, a real-time updated database

of field devices can be made available for simulation

tools. Response based control can take advantage of

this possibility by improving system representation

and introducing new control functions. At the same

time, for example, remote control of field device

will made possible to implement fast actions such as

changing protection settings or remote line

switching.

3 AN ALGORITHM FOR

RESPONSE-BASED

CORRECTIVE CONTROL

The power system behaviour on transient time-scale

is generally described through a set of nonlinear

differential and algebraic equations (DAEs) (Kundur

1994):

0uVxg

uVxfx



))(),(),((

))(),(),(((t)

ttt



(1)

where x is the state vector, u is the control variable

vector, and V is the vector of nodal voltages. The

control variable vector u is given by the set-point

value (Bose et al., 2004a; 2004b) or set-point time

varying trajectory (Taylor et al., 2005) of

controllable devices and control action actuators.

The DAEs set (1) can be discretized through a

trapezoidal rule and written in implicit form as:

SIMULTECH2013-3rdInternationalConferenceonSimulationandModelingMethodologies,Technologiesand

Applications

212

0uyH ),

(

(2)

where

Tii

ni ,,2,1,0),( 



 0uyH

(3)



iii

Vxy 

(4)



TTT

yyyyy 

(5)



TTT

HHHHH 

(6)

with y

representing the composition of all state

variable and voltage vectors at the generic i

time

step and ŷ representing the discretization of the

whole trajectory of the system; H

is the

discretization of the DAEs set (1) at the generic i

time step; n

is the total number of time steps

relative to the integration interval [0,T].

The methodology proposed in this paper is based

on (La Scala et al., 1998); (Bruno et al., 2002).

Corrective control actions are evaluated by solving a

dynamic optimization problem for non-linear

systems, in the presence of static and dynamic

inequality constraints. The dynamic problem is

formulated in terms of static optimization and solved

by applying the Lagrangian multipliers method.

The main difference with regard to (La Scala et

al., 1998); (Bruno et al., 2002) is that, instead of

formulating and solving a single optimization

problem, this method aims at finding the optimal

value of the control-vector u

that solves the

optimization problem given for each time window t

by the equations:



)

()

()(min

wPwOwU

CCC

yyu



(7)

subject to

0uyH ),

(

(8)

and

w min

 u



w max

(9)

In (7), ŷ

represents the system trajectory in the time

window t

, C

represents an objective function

aiming at the minimization of the controlling effort;

is the objective function whose scope is to

improve the dynamic behaviour of the system; C

a penalty function that takes into account inequality

constraints. Inequality constraints are usually

referred to technical and operational constraints and

define time-varying domain where system

trajectories should be contained.

The problem is solved by applying the

optimization method of Lagrangian multipliers:

(

)

()

()(

wPwOwU

CCCL uyHγyyu 

(10)

)(















(11)













(12)

0),

(





uyH

(13)

From (11) and (12) derives:

CCC

yuu 





























)(

(14)

Usually, in off-line applications, the optimization

problem is solved through an iterative algorithm,

updating at each iteration the control-vector

wneww







(15)

simulating a new dynamic trajectory, calculating a

new control vector update in a recursive approach

that stops when the sensitivity term

L u is lower

than a specific tolerance limit.

In the response-based approach presented in [4-

5], the trajectory ŷ

is acquired through WAMS and

cannot be modified. The control-vector u

w+1

evaluated optimizing ŷ

trajectory, can be applied

only after that the trajectory itself was acquired. The

control variable is therefore updated after each time

step with a simple sensitivity analysis as shown in

Figure 3. After the implementation of control

actions, new corrective actions are evaluated after

that a new piece of trajectory is acquired, and so on.

Figure 3: Schematization of the proposed response-based

approach.

Response-basedControlthroughDynamicOptimizationinLarge-scalePowerSystems

213

The approach in [4-5] is feasible having

considered that suboptimal solutions are still suitable

if stability is ensured in response to large

disturbances and severe threats to system integrity.

A further evolution of this approach, can be

achieved if trajectories ŷ

were forecasted through

powerful simulating tools before they are actually

experienced and acquired. In such case, a real-time

optimal control of power system trajectories would

be achieved. This paper embraces the opinion that

through real-time dynamic simulators, or simulators

even faster than real-time, such approach might be

possible.

An important observation on this regard is that

the approach is feasible as long as control actions

can be applied with a reasonable time delay δ. The

overall time delay δ takes into account the time

necessary for data acquisition from WAMS, data

transmission to the Control Centre, data

synchronization, CPU simulation time, data

transmission to actuators, triggering of the corrective

control actions.

4 TIME REQUIREMENTS

Some experiences have been carried out considering

a representation of the Italian power system and part

of the interconnected UCTE network. The model is

characterized by about 1333 nodes, 1762 lines, 273

generators and 769 transformers, implying that state

variables are 1638 and voltages are 2666. At each

time step 4304 variables are involved in a typical

index-1 problem. Due to the discrete time approach,

the trajectory of the system have to be evaluated for

about 10s i.e. for about 500 time steps:

Consequently the trajectory vector consists of about

1.3 millions of variables and the overall problem

(eqns. 11-13) can have the size of more than 2.6

millions of equations depending also on the number

of control variables and time discretization.

In order to assess the feasibility of the proposed

response-based control architecture, three different

tests have been carried out considering different

actuators: FACTS devices (Bose et al., 2004b) and

load shedding schemes (Bose et al., 2004a), line

switching (Bruno et al., 2012b) and adaptive

relaying (Bruno et al., 2012c); (Bruno et al., 2011).

In tests, corrective control is applied with a variable

time delay δ after the onset of the fault, in order to

understand how far it is possible to delay the

application of control actions, and still ensure

system stability. The evaluation of maximum time

delays is a key issue, permitting to assess

communication channels technical requirements,

computational time and actuators’ response speed. In

such tests, it has been assessed that time delays

around 300-500 ms are still compatible with stability

requirements for most simulated contingency cases.

4.1 Actuators

Actuation for load shedding needs about 100ms and

can be considered negligible for FACTS devices and

adaptive relaying. Our position is that delays of few

hundred milliseconds (let us say 300 ms) are enough

for response-based control to be able to correct most

of unstable behaviours. The next step is to assess if

computational and communication timings fit this

challenging requirement.

4.2 Computational Time Delay

Assessment

Since in the proposed approach, trajectories are

assumed to be known by WAMS, corrective actions

can be derived through the mere solution of the

linear system (11-12) and the consequent application

of eqn. (14). It should be considered that, in the

evaluation of the lagrangian multipliers  it is not

necessary to adopt the same time step discretization

as for the evaluation of the trajectory. It was

observed that time steps equal to 0.1s are sufficient

to provide a good approximation of the trajectory of

Lagrange multipliers .

The solution of such equations, having

considered the abovementioned system model (1333

buses) and a time windows of 0.1s, requires about

0.32s on a standard PC equipped with a Intel Core2

Quad CPU Q9650 processor, 3 GHz, 4 GB RAM.

It can be concluded that CPU timings should be

more or less compatible with response based control

if reduced by a factor 4. This possibility is at hand

considering that more powerful computational

resources (supercomputers, parallel computing

environments) can be adopted. As an example, a

speed-up around 9 can be obtained through

vectorization (Granelli et al., 1993) or speed up

around 6 with message passing machines or

distributed machines equipped with 32 CPUs as

proved in (Aloisio et al., 1997).

4.3 Communication Time Delay

Assessment

Potentials of WAMS architectures and advanced

stability control architectures were investigated

during a full-scale experiment (La Scala et al. 2006).

SIMULTECH2013-3rdInternationalConferenceonSimulationandModelingMethodologies,Technologiesand

Applications

214

The experiment was carried out benefiting from the

collaboration of PMU manufacturing companies and

dealt with the installation and testing of PMU

devices, and with the assessment of SPS

performances, including telemetry, monitoring and

wide-area detection.

On the basis of data acquired during the above

mentioned experiment, time performances of the

communication infrastructures have been monitored.

The overall time delay for acquiring, transmitting (to

the Control Centre) and re-transmitting (to actuators)

data has been estimated in the range 70-100ms

(25ms for each one-way data transmission) (La

Scala et al., 2006); (Naduvathuparambil et al.,

2002). This result is also consistent with results of

an Italian WAMS project (Cirio et al., 2011).

4.4 Final Observations

Performances of a centralized wide-area monitoring

architecture are not yet compatible with a response-

based control approach, since the delay associated to

data acquisition and control action implementation

may exceed the maximum delay assessed in the

previous section.

On the basis of the results obtained so far, it can

be estimated that the overall architecture would need

around 400ms for computations, 100ms for

communications and 100ms for actuating remedial

actions as an example. Thus we are close to the goal

but we did not score yet.

Our position is that computation is not the

bottleneck. Since the elapsing time related to the

dynamic sensitivity calculation can be drastically

reduced through high performance computing, we

believe that the bottleneck in the time response of

the control chain is still associated to the

communication system and to the actuators.

The position is that, fast actuators such as

FACTS devices, but also less expensive ones such as

adaptive relays, can meet the time challenge with

regards to actuation delay. More investments in the

high-speed communication infrastructure can

provide the right answer to meet the strict

requirements imposed by a centralized response-

based control architecture.

5 CONCLUSIONS

In this paper, a centralized wide-area control

architecture for evaluating and implementing

response-based corrective control actions has been

illustrated. Feasibility studies of the integration of

dynamic optimization methodologies with advanced

monitoring and control technologies have been

carried out. Moreover, the maximum acceptable

overall delay, ranging from 300 to 500 ms, has been

assessed to stabilize the Italian power grid with a

response-based control strategy.

The position is that the presence of WAMS, fast

actuators, high performance computing and high-

speed communication infrastructure can meet the

challenge of a response-based control for large scale

power systems.

REFERENCES

Aloisio, G.; Bochicchio, M. A.; La Scala, M.; Sbrizzai, R.,

1997, A distributed computing approach for real-time

transient stability analysis, IEEE Transactions on

Power Systems, Volume: 12, Issue: 2, pp. 981-987.

Berizzi, A., 2004, The Italian 2003 blackout, IEEE Power

Engineering Society General Meeting, Denver (CO),

USA, June 6-10.

Berizzi, A., Sforna, M., 2006, Dynamic security issues in

the Italian deregulated power system, IEEE Power

Engineering Society General Meeting, 18-22 June.

Bose, A., Bronzini, M., Bruno, S., De Benedictis, M., La

Scala, M., 2004a, Load Shedding Scheme for

Response-based Control of Transient Stability,

Proceedings of IREP Symposium 2004 - Bulk Power

System Dynamics and Control, Cortina D'Ampezzo,

Italy, August 22 – 27.

Bose, A., Bruno, S., De Benedictis, M., La Scala, M.,

2004b, A Dynamic Optimization Approach for Wide-

Area Control of Transient Phenomena, Proceedings of

CIGRÈ General Meeting, Paris, 29 August – 3

September.

Bose, A., 2010, Smart Transmission Grid Applications

and Their Supporting Infrastructure, IEEE

Transactions on Smart Grid, Vol. 1, No. 1, pp. 11-19.

Bruno, S., De Tuglie, E., La Scala, M., 2002, Transient

Security Dispatch for the Concurrent Optimization of

Plural Postulated Contingencies, IEEE - Transactions

on Power System, Vol. 17, No. 3, August, pp. 707-

714.

Bruno, S., De Benedictis, M., La Scala, M., Rotondo, G.,

2011, Balancing Protection Dependability And

Security In Large Transmission Grids, Cigré

International Symposium - The Electric Power System

of the Future. Integrating supergrids and microgrids,

Bologna, Italy, September 13-15.

Bruno, S., D’Aloia, M., De Carne, G., La Scala, M.,

Sbrizzai, R., 2012a, Controlling transient stability

through dynamic series compensation, IEEE

International Energy Conference & Exhibition

ENERGYCON 2012, September 9-12, Firenze.

Bruno, S., D’Aloia, M., De Carne, G., La Scala, M.,

2012b, Controlling Transient Stability through Line

Switching, IEEE PES Innovative Smart Grid

Response-basedControlthroughDynamicOptimizationinLarge-scalePowerSystems

215

Technologies Europe 2012, October 14-17, Berlin,

Germany.

Bruno, S., D’Aloia, M., La Scala, M., Sbrizzai, R., 2012c,

Protection system monitoring for the prevention of

cascading events in smart transmission grids,

Complexity in Engineering (COMPENG) 2012,

Aachen, Germany, June 11-13.

Cirio, D., Lucarella, D., Giannuzzi, G., Tuosto, F., 2011,

Wide area monitoring in the Italian power system:

architecture, functions and experiences, Euro. Trans.

Electr. Power, vol. 21, pp. 1541–1556.

Granelli, G.; Montagna, M.; La Scala, M.; Torelli, F.,

1993, Relaxation-Newton methods for transient

stability analysis on a vector/parallel computer, Power

Industry Computer Application Conference,

Conference Proceedings, pp. 387 - 393

Kundur, P., 1994, Power System Stability and Control,

McGraw Hill.

La Scala, M., Trovato, M., Antonelli, C., 1998, On-line

dynamic preventive control: an algorithm for transient

security dispatch, IEEE Trans. on Power Systems, vol.

13, No. 2, pp. 601-610.

La Scala, M., De Benedictis, M., Bruno, S., Grobovoy, A.,

Bondareva, N., Borodina, N., Denisova, D., Germond,

A. J., Cherkaoui, R., 2006, Development of

Applications in WAMS and WACS: an international

cooperation experience, IEEE PES General Meeting

2006, Montreal, Canada, June 18th-22th.

Momoh, J. A., 2009, Smart grid design for efficient and

flexible power networks operation and control, Power

Systems Conference and Exposition (PSCE), Seattle,

15-18 March.

Naduvathuparambil, B., Valenti, M. C., Feliachi, A., 2002,

Communication delays in wide area measurement

systems, Proceedings of the Thirty-Fourth

Southeastern Symposium on System Theory, pp. 118-

122, 18-19 March.

Phadke, A. G., Thorp, J. S., 2006, History and

Applications of Phasor Measurements, Proc. IEEE

Power Systems Conference and Exposition, Oct. 29 –

Nov. 1, pp. 331-335.

Taylor, C. W., 2000, The Future in On-Line Security

Assessment and Wide-Area Stability Control, IEEE

Power Engineering Society Winter Meeting 2000, vol.

1, pp. 78–83, January 23-27.

Taylor, C. W., Erickson, D. C., Martin, K. E., Wilson, R.

E., Venkatasubramanian, V., 2005, WACS-Wide-Area

Stability and Voltage Control System: R&D and

Online Demonstration, Proc. of IEEE, Vol. 93, n. 5,

pp. 892-906, May.

SIMULTECH2013-3rdInternationalConferenceonSimulationandModelingMethodologies,Technologiesand

Applications

216