A Healer Reinforcement Approach to Smart Grid Self-Healing by Redundancy Improvement

Alireza Shahsavari, Alireza Fereidunian, Hamid Lesani

f School of Electrical and Computer Engineering, University of Tehran, Tehran, Iran.

2 Electrical Engineering Faculty, K. N Toosi University of Technology and School of ECE, University of Tehran, Tehran, Iran.

Abstract- Smart Grid is expected to heal the electricity distribution system to improve its reliability, known as self

healing capability. Self-healing can be performed in system

level (like restoration and local generation), in component level,

or by reinforcing the healer system reinforcement (called as

healer healing). Smart Grid literature chiefly focuses on the

two former approaches; while, the latter has gained less

attention despite merit. A proper healer reinforcement method is expanding redundancy, in both control and protection

subsystems. The redundancy expansion and the consequent

self-healed distribution network is anticipated to express lower

outage times, thus higher reliability and increased social

welfare. This paper investigates the effect of redundancy

expansion on reinforcement of the Smart Grid healer, by

reliability modeling of protection and control subsystems. The

positive effect of the proposed healer reinforcement approach

on the overall Smart Grid reliability is shown on RBTS4, and

the results are discussed then.

Keywords-Smart Grid, Self-healing, Healer Reinforcement;

Reliability Modeling; Redundancy Expansion; Non-Dominated

Solutions.

NOTATION

The notation used throughout this paper is reproduced below for quick reference.

nust. ! Aline

J Atrans. J Abusbar J Lline J Ltrans. J LBusbar J r.line !)

r.trans. !)

r.busbar !)

In!

Life cycle of study;

Number of load points;

Average annual time that the load point i is out of supply (h/yr);

Average load per customer in load point i (kW/cust.);

Number of customers in load point i (kW/cust.); Lines failure rate of section j (f/km.yr);

Transformers failure rate of section j(f/yr);

Busbar failure rate of section j (f/yr);

Length of section j lines (km);

Number of transformer in section j ;

Number of bus bars in section j ;

Average outage time of section per fault in

section j lines (hIt);

Average outage time of section per fault in

section j transformers (hIt);

Average outage time of section per fault in

section j busbar (hIt);

Inflation rate;

Int wfes

wfomm

wfnd

w[C

wfPC

JCRes JCcomm JCind JCPC JCCPC ENS CENS TCENS

TCR TPJ RB

Interest rate;

Residential load coefficient;

Commercial load coefficient;

Industrial load coefficient;

Public customer load coefficient;

Critical public customer load coefficient;

Residential customers interruption function;

Commercial customers interruption function;

Industrial customers interruption function;

Public customers interruption function;

Critical public customers interruption function;

Energy Not Supplied per year (kWh/yr);

Cost of Energy Not Supplied per year($ US/yr);

the Total Cost of Energy Not Supplied during

life cycle of study;

Economic Factor;

the Average Interruption Cost for load point i ($ US/kW);

Total Cost of Redundancy expansion;

Total Planning Investment;

Redundancy expansion Benefit.

I. INTRODUCTION

H IGH reliability requirement is considered as the main challenge in modern grids; specifically in electric power distribution systems which are in charge of delivering

the electrical energy to the consumers. Smart grid satisfies

this reliability requirement this obstacle by the self-healing

capability [1].

A self-healing system uses information, sensing, control

and communication technologies to handle issues and unwanted events by eliminating or minimizing their

disadvantageous effect for maximizing reliability [2, 3].

Self-healing can be performed in system level (like local

generation and capability of automatic fault detection, isolation, and service restoration), in component level, or by

healer system reinforcement which we named it as healer

healing. In [4], we introduced a novel framework for self

healing methodologies.

Healer reinforcement method is developed into wide area approaches. A pragmatic approach is improving its

redundancy, in both control and protection subsystems.

From the Sfstem configuration point of view, the

redundancy improvement leads to achieve a self-healer

distribution network. The consequent self-healed

distribution network anticipated to express lower outage

times, in both frequency and duration of interruptions, thus

higher reliability and increased social welfare.

Reliability requirement specifies the degree of

redundancy enhancement. On the other hand, it is conceivable that overall budget is run over by ting up high degree of ra:lundalt pcths. Thus, it is necessary to do a compromise between reliability requirement and investment

of redundancy expansion. Economically, the utility cost will

generally increase as consumers are provided with higher

reliability. On the other hand, the consumer costs associated

with supply interruptions will decrease as the reliability

increases. The total costs to society will therefore be the sum

of these two individual costs. This total cost exhibits a

minimum and so an optimum or target level of reliability is

achieved.

From the reliability aspect, enhancing the redundancy in

control subsystem is expected to influence on duration of an

interruption and by enhancing it in protection subsystem expected to influence on both duration and frequency of an

interruption. The redundancy enhancement in control

subsystem decreases SAIDI (System Average Interruption

Duration Index) and in protection subsystem decreases both

SAIFI (System Average Interruption Frequency Index) and

SAIDI.

All in all, healer reinforcement approach by using of

redundancy approach possesses some technical and

economical benefits. For instance, improving overall

reliability indices and decreasing total cost of energy not supplied. However, it incurs more installation cost

maintenance. Hence, it is necessary to economically analyz; the healer reinforcement by redundancy approach.

In [3], reliability models for protection and control

subsystems are presented and different redundancy schemes

are checked. However the impact of redundancy expansion

?n overall reliability did not discussed. In [4, 5], redundancy

m some parts of the protective system is examined and

A.Abbarin and Fotuhi-Firuzabad extended a Markov model

and examined redundancy and protective components

effects. In [7], a technical novel framework is introduced for

redundancy approach in both control and protection

subsystem.

This.

paper .investigates the effect of redundancy

expansIOn on remforcement of the Smart Grid healer, by reliability modeling of protection and control subsystems. The positive effect of the proposed healer reinforcement approach on the overall Smart Grid reliability is shown on RBTS4. The effect of redundancy expansion is checked in both subsystems individually on RBTS4. Also the economical and technical effects of redundancy expansion on subsystems are investigated and best solution is selected by using non-dominated solution, the results are discussed then. In section 2, by modifying these models different redundant paths have been defined. Next in section 3 reliability achievements of implementing redundanc; approach has been presented and possible redundant sets have been introduced as a Non-Dominated solution and final redundant set has been selected through max-min approach.

2

II. METHOD

A. Research Methodology

For evaluating the overall smart grid reliability and the

positive effect of redundancy expansion on reinforcement of

the Smart Grid healer, the reliability of protection and

control subsystems should be modeled. Reliability modeling

for protection and control systems and different redundant

schemes are presented, in section 2.3.

Figure. l shows the flowchart of the overall and load point

base reliability evaluation. In this flowchart, ARPM and

ARTM represent Auto-Restoration Probability Matrix and

Average Restoration Time Matrix, respectively. ARPM is

calculated by conditional probabilities and depends on the

allocated automatic and manual switches, protection devices

and probability of their successful operation [8]. In [3]

ARPM is calculated by using RadPow software base on

event tree, while in this article, reliability evaluations are

simulated in a home developed program. ARTM is

calculated by conditional probability and depends on

switching time and the calculated ARPM. Redundancy

influences the values of protection and control block in the

flowchart; therefore, redundancy affects the system reliability.

Feeders' Evaluation

Figure.I.Flowchart of the reliability evaluation

B. Problem Formulation

As mentioned before, the economical and technical

effects of redundancy expansion on subsystems are

investigated in this paper by using prepared models and

evaluation flowchart. For investigating technical impact of

redundancy expansion some reliability indices are

examined, for instance, SAIFI, SAIDI, CAIDI, AENS and

ASAI, which are expressed in [9, lO]. To evaluate the

economical the impact of the proposed healer reinforcement

by components redundancy expansion on the system

reliability, total cost of energy not supplied (TCENS) during

the life cycle is examined, which is expressed as:

ny LP TCENS = L L CENSi x (EF) n (1)

n=l i

Where n is the number of years, i is the load point numbers and EF is the economic factor for evaluation of present worth/cost factor which changes the cost of time

study to current cost. The CENSi and the EF are written as:

AlC-(U) CENS = ENS x ! ! ! ! Ui

E F = _1

_+

_I

_n.:....f

1 + Int

(2)

(3)

(4)

In equation (2), AICi(Ui) is computed by equation eq.(5), j is the faulted section.

AICi(Ui) = wfesICRes(Ui) + wfomm I Ccomm (Ui)

+w{nd ICrnd(Ui) + wfc ICpc(Ui) + wfPC ICcpc(UJ (5)

The IC represents interruption cost function for each kind of consumers. In (6), the total cost of redundancy approach and upcoming costs is computed. Also by using (7) benefit

of implementing redundancy in distribution system is

computed.

TCR = TCENS + TPI

RB = TCRnon-redundant - TCRredundant

C. Reliability Modeling for Redundancy

(6)

(7)

Figure.2 shows smart grid subsystems including control,

protection and IT infrastructure interacting with legacy

system [12]. Protection and control systems in contingencies

conditions are responsible for isolating faulted zone and

supply restorable zones, for minimizing the duration of

interruption and to limit the impact of fault [7, 11], thus the

overall smart grid reliability depends on successful

operation of protection and control systems. In this section, protection and control subsystems reliability models are

presented and different redundancy extensions are

implemented.

Control and Protection Power Distribution System System (The Legacy System)

IT Infrastructure of Smart Grid Figure.2. Smart Grid Sub-systems [12]

J) Protection System Reliability Modeling

Protection system is the most important factor to the

secure operation of the electrical power networks, induding

3

di stri buti on systems. A rei i ci:lI e prota::ti on system improves the overall network reliability. The probctlility for a prota::tion system to operate responding to a fault depends on the rei i aI:li I i ty of its components One etfa::ti ve approcrll for improving the prota::tion system's relial:lility is

enhancing its ra::tundancy.

a. Feeder Protection (Breakers) Reliability Modeling

For a normally dosa::l breaker the operaing and failure states are [3]:

1- Passes currents from in its closed state; 2- Opens successfully when requested to do so; 3- Fails to open when requested to do so; 4- Opens inadvertently when not requested. State 1 and 2 are the desirable conditions and 3 and 4 are

the unwanted ones. 1 and 4 refer to dependability and

security of the protection system [13]. In [9] these two aspects of reliability in redundant protections system have

been discussed.

Figure3 represents the protection system in term of its

components block diagrams. These blocks related to feeder

protection in which the digital relay unit (DRU) includes the

electronic circuits of signal conditioning and the digital

processing relay system. The auxiliary relay unit (ARU)

contains trip relay (TR) and the associated power supply

unit (PSU).

1

................................ : PR :ARt:j ..

1---F"Ti-.

Figure.3.Block diagram of the base protection scheme, adapted [3]

When a permanent fault occurs, all components should

operate correctly, i.e., failure in each block causes fault in

protection operation. From the reliability point of view, all

components are series and the reliability block diagram

(RBD) of the basic protection scheme is shown in figure.4.

Figure.4. Reliability block diagram of the basic protection scheme.

In the basic protection scheme, any failure in each block

causes malfunction in protection operation. Consequently,

back up protection operates and both faulted and faultless

feeders experience outage. The probability of malfunction in

protection scheme reduces by redundancy enhancing in

components. In order to see different redundant protection schemes, for preventing from unnecessary repetition,

figure.5 shows only sample of reliability block diagram

cases. Cases 2-5 comprise one redundant component in

sequence for Breaker, DRU, ARU, and CT.

Case 6 comprises two DRUs and two ARUs which

prepare two paths; the system is operating if at least one

path is functioning. Case 7 has two breakers and two CTs and prepares 4 paths; the system is functioning if at least one

of breakers and one of CTs are functioning. Figure.6 shows

the even tree of case 8, it consist of two DRUs and two

ARUs and prepares 4 paths; the system is functioning if at

least one of DR Us and one of ARUs are functioning. Case 9

and case 10 are equal to (n-l) index of reliability, case 10 in

components layer and case 9 in system layer. Assuming the

failure events of the components blocks are independent,

also the probability of each basic component operation is

equal to tablel and the probability of each redundant

component operation is 0.95% of the basic component.

Table2 shows the probabilities of base case and redundant

protection cases operation.

Fault

CT

Case 6

Case 8

Non-redundant Protection system 1

Non-redundant Protection system 2

Case 9

Case 10

Figure.5. Reliability block diagram of different

redundancy schemes in protection systems

B

o

i ARUI i ARU2 i 0 i

Outcome

i->'---:-----: I: Trips 2: Trips

3: Fails to Trips

-"";""----i 4: Trips 5: Trips

o 6: Fails to Trips

I-'-F _--+-__ +-_-1 7: Fails to Trips F _-+ __ +-_-+ __ +-_-i 8: Fails to Trips

L..:...F __ ---'----'---""'"""---'-----' 9: Fails to TrillS

F

Figure.6. Event tree of case 8

b. Fuses Reliability Modeling

Fuses in automated distribution systems have direct

impact on systems reliability. Fuse in low voltage branches

decreases duration and frequency of interruptions. From the

reliability point view, fuse is independent device and its

reliability block diagram is one block.

2. Control System Reliability Modeling

Distribution automation and control systems consist of

three functions; line's facilities or secondary substations,

primary substations, and distribution control centers (DCC)

4

[3, 14, 15, 17]. Communication systems are non-negligible

and

requisite infrastructures for data exchanges in these three

layers [16]. In this paper, the reliability of automation

control systems is investigated into two layers, local control

system and central control system.

The central control system performs functions, including

control and monitoring of all substation equipments,

accumulate and process local controls data, dispatch control

commands and receive results. The local control system

consists of remote terminal units (RTUs), these devices

exquisite various types of information, execute commands

from the central center and reporting status after

implemented commands to the central center.

A rei i cD! e control systan improves the overall smai gri d rei icDi I ity. The probabi I ity for the control systan to operate responding to a commald depends on the reliability of its

components. One effective approcdl for improving the

control systan rei i cDi I i ty is enhalo ng its redundalCY. Figure.7,8 represents control system consists of local and

central layers in term of component block diagrams. Central

control includes Power Supply Unit (PSU), Central

Processing Unit (CPU), Memory Unit (MU), controller and

Communication Interface (CI). The main components in

local control system are Switching Device (SD), battery and

charger or Power Supply Unit (PSU), power actuator (Drive), Current transformer and Voltage Transformer

(CTIVT), Remote Terminal Unit (RTU) , Fault Passage

Indicator (FPI) and Communication Interface (CI). The

RTU includes central processing unit (CPU), Input / Output

interface (I/O). A typical RTU possess eight digital inputs

(D!) and eight digital outputs (DO), also may equipped with

six analog inputs [3, 11, 14, 15].

Ccntral Computer [ cPu I [ Controller I

I Power Supply I Urnt I Communication I Interface

(a)

Figure.8.(a). Block diagram of the base central control system

(b). Block diagram of the base local control system; adapted [3, II, 14, 15]

In order to achieve a proper operating control system,

each component in both local and central control layers

should operates correctly, i.e., any individual event that

causes failure of each component in local/central control

layers fails the local/central control system. From the

reliability point of view, all components are connected in

series. Assuming failure events are independent, equations

(8) and (9) represent the probabilities of non-redundant local

and central control operation.

P(cent ral cont rol) = P(PSU) P( Cont roller) ' P( CPU) . P(MU) . P( CI)

P(local cont rol) = P(SD) . P( CT) ' P(PSU) P(FPI)

(8)

. P(PA) . P(RTU) . P( CI) (9)

a. Central Control Reliability Modeling

In order to see different redundant central control

schemes, figure.9 shows reliability block diagrams of four

different redundancy schemes in central control system. For

preventing from unnecessary repetition, figure4 shows only

sample of reliability block diagram cases. Cases 2-6 comprise one redundant component in sequence for CI,

PSU, CPU, MU and controller.

Case 7

Case 8

Central control system 1

Central control system 2

Case 9

Case 10

Figure.9. Reliability block diagram of different

redundancy schemes in central control systems

Case 7 comprises two CPUs, MUs, and controller which

prepare two paths, and Case 8 is alike case 2, which

prepares 8 paths; the system is functioning if at least one of

paths is functioning. Case 9 and case 10 are equal to (n-l)

5

index of reliability, case 10 in components layer and case 9

in system layer. Assuming independent failure events for the

components, also considering the probability of each

component operation is equal to table 1 and the probability

of redundant component operation is 0.95% of the basic

component. Table 2 shows the probabilities of base case and

redundant cases operation.

b. Local Control Reliability Modeling

In order to see the effect of redundancy expansion in local

control system, figure. 10 shows reliability block diagrams of

four different redundancy schemes in local control system.

Also case 2-6 comprise one redundant component III

sequence for CI, PSU, RTU, power actuator, and FPI.

Local control system 1

Local control system 2

Case 9

Case 10

Figure.lO. Reliability block diagram of different

redundancy schemes in local control systems

As discussed in protection and control modeling, table2

shows the probabilities of base and redundant local control cases operation, assuming independent failure events for the

components, also considering the probability of each

component operation is equal to table 1 and the probability

of redundant component operation is 0.95% of the basic

component.

Tablel. Components reliability, adapted [3]

Control system Protection system

Central Control Local Control

Component Reliability Component Reliability Component Reliability

CT 0.98 5 PSU 0.995 SO 0.990

DRU 0.990 CPU 0.994 PSU 0.995

ARU 0.98 8 MU 0.993 PA 0.992

B 0.993 Controller 0.98 7 CTIVT 0.990

CI 0.997 FPI 0.990

RTU 0.993

CI 0.997

Table2 antICipates the redundancy improves the

probability of successful protection and control operation.

Comparison between case 6 and 8 of protection system (or

case 7 and 8 in control system) illustrates redundant

components connection is extremely important, parallel

connections make more reliable than series. As expected,

redundancy in components layer is more reliable than the

redundancy in system layer, case 9 and 10 indicate that.

Table2. Probabilities of each considered cases operation

Protection System Control System

Case Feeder Fuse

Central Local Control

Protection Control

Base 0.956704 0.90 0.966431 0.943378

Case 2 0.963066 0.91 0.969185 0.946066

Case 3 0.965792 0.92 0.971021 0.947859

Case 4 0.967610 0.93 0.971940 0.949651

Case 5 0.970337 0.94 0.972858 0.950547

Case 6 0.975597 0.95 0.978366 0.952340

Case 7 0.976789 0.96 0.98 7800 0.959456

Case 8 0.976802 0.97 0.990486 0.961405

Case 9 0.990442 0.98 0.991534 0.980680

Case 10 0.997310 0.99 0.998027 0.996561

III. Case Study

A. Example Case

The effect of the proposed healer reinforcement approach

on the overall Smart Grid reliability is shown on RBTS4.

The RBTS has 5 buses (Bus2-Bus6), two buses (Bus2, Bus4) are defined as distribution network. Figure. 1 1 shows

RBTS4 which has three supply points, seven feeders, 38

load points, 4770 customers, and the peak and average load

40 MW and 24.58 MW, respectively. Customers data and

lines length are taken from [18]. Table3 shows components

and lines failure rates and average repair or replacement

time per failure. For evaluating the reliability, auto

switching time and manual switching time are assumed 40 seconds and l.5 hour, respectively. Life cycle of study is

considered 25 years, also inflation rate and interest rate 6%

and 7%, respectively. Figure.I2 shows the Ie s of different customer types.

S3W lIkV I

control facilities prices are exist, component's reliability

cost curve is an upward concavity exponential function [9,

10], Thus the reliability-cost curve for subsystems

consequence of components reliability-cost curve and can be

calculated.

Figure15 shows total cost of redundant planning in central

and local control, as shown minimum TCR consequences of

case 8 for both local and central control, it shows that by

implementing more redundant for instance cases 9 or 10,

planning costs increases more than decreased TCENS.

1.826

1.824

1.822

Vl 1.82 1.818 '" 1.816

1.814

1.812

X 106

L81-..J.--=-,"-- Case 3 Case I

Case 1 ease2 ca';"3 3 -:C"C:C

'4i-

C:;C

"'='5

C:;C

""=6S$s;'::::::::'c Case 5 \ CO\\tt"O

\ , .,:dunda High Redundant LOCal C::, Case 8 Case 9 Case 10 Case 9

Figurel5. Total cost of redundancy of redundant control cases

C. Impact of redundancy in protection system

This section presents impact of redundancy in protection

system base on explained modeling in part "C" of section II.

For investigating the positive impact of redundancy in

protection system on network, the control system is modeled

as the base case. By comparing 100 possible cases

determines that the CENS is reduced about 8% also the

SAID! is reduced about 10%, as shows in figure 16, 17.

7.9

7.7

7.6 8 7.5 U7.4

7.3 7.2 /- Case I ease 2 Case 3 cas;'gh Case 5 Case 6 Cac 7 Case 8 Case 9 Case 10 0.98 I ' Re