CHAPTER 5 RESTORATION OF POWER USING MULTI-AGENT...
Transcript of CHAPTER 5 RESTORATION OF POWER USING MULTI-AGENT...
102
CHAPTER 5
RESTORATION OF POWER USING
MULTI-AGENT SYSTEM
5.1 INTRODUCTION
It is a known fact that the restoration of a power system after a
system blackout is a difficult task. It is a complex, delicate, and time-
consuming process. If power is not restored, there will be heavy loss to
consumers. At present, control system software has been developed primarily
to facilitate monitoring and control of the power system. Still the network
undergoes outages. Even at present restoration task is performed manually,
according to guidelines prepared in advance. The power system operators
should execute the restoration plan under a stressful situation. But the
restoration plans are prepared based on the predefined fault conditions and the
plans are complex and convoluted as to be very difficult to interpret for a new
fault condition.
In the era of deregulation on energy policies, the coordination of
independent power producers, generation companies and consumers is
difficult and cannot be solved easily by the conventional approaches such as
Heuristics, Expert systems, Mathematical programming, and Soft computing.
This drawback initiates this study to choose the MAS. This system has many
agents and is defined as one that interacts with external environments,
103
cooperates with other ones and processes to be solved autonomously. After
total black-out, power system restoration requires coordination among
generating units, load pick-up characteristics and transmission systems. Since
co-ordination is necessary for the power system restoration, the MAS
technique provides an interesting environment for evaluating agent behaviour.
5.2 ARCHITECTURE OF MULTI-AGENT SYSTEM
The technique, MAS divides the big complex processes into small
process which is assigned to agents, and then the agents interact with each
other to achieve the system goal. The distributed agents perform the task
simultaneously and are believed to increase performance of the system. MAS
technique has been successfully applied in many fields such as power system
and computer network.
The MAS with the two- level architecture consists of a number of
local managing agents (LMAs) and a global managing agent (GMA) which is
shown in Figure 5.1. GMA is an agent that facilitates the negotiation process
of the MAS approach. It classifies the de-energized buses into many groups
based on the power requirement. It sends the start message to one of the
LMAs which is at the higher power level. It repeats the same task
simultaneously with other de-energized groups. After the occurrence of fault,
LMA is developed to decide a suboptimal target configuration by interacting
with other LMAs. Based on the number of buses, the LMAs are allocated. It
has certain rules associated with it to restore the load and are given as follows:
• If there is more number of branches that can energize certain
bus then LMA is allocated to the particular bus which has the
largest amount of available restoration power.
104
• If the amount of available power for restoration is insufficient,
LMA tries to restore the bus by negotiating with its
neighbouring LMAs.
• If a load must be shed because of insufficient power, LMA
cuts off the load connected to its own bus to the minimum
possible.
Whenever there is a fault, the LMA corresponding to the
de-energized bus sends a request to the GMA. Many such messages are sent
to GMA if there are many de-energized buses. After receiving the first
message, the multi agent system initiates its function. The GMA then sends a
start message to the LMA, which starts its communication with the
neighbouring LMAs. In this way, MAS recovers the entire network.
Figure 5.1 Architecture of Multi-Agent System
LMA
GMA
LMA
LMA
105
5.2.1 Problem Formulation
The objective of the mathematical model (Nagata et al 2002) of a
power system restoration is to maximize the capacity of the served loads in
power system network.
Max Li . yi (5.1)i D
where Li is the load at bus i, yi is the decision variable or expressing its status
(yi=1:restored; yi=0:not restored) and D denotes the set of de-energized loads.
Typical constraints associated with the restoration model taken into
account are:
Limit on the capacity of available power source for restoration
Pb . xb Gq (q S) (5.2) b Fq
where Pb is the power flow in the directed branch b xb is the decision variable
of branch b (xb = 1:b is included in the restoration path; xb = 0:otherwise). Fq
is the set of branches with starting node q, Gq is the restoration power from
the energized bus q, and S is the set of energized buses that can be connected
to de-energized area.
Power balance between supply and demand
Pk - Pk – Li . yi = 0 (i N) (5.3) k Ti k Fi
where Ti is the set of branches incident to bus i, Fi is the set of branches with
originating from bus i, Li is the load at bus i, and N is the set of buses.
106
Limits on branch power flow
|Pk| - Uk 0 (k B) (5.4)
where Pk denotes the power flow of branch k, Uk is the capacity of branch k,
and B is the set of directed branches.
Constraint on radial configuration
The target configuration must be radial. To ensure a radial
configuration, the total number of branches incident to bus i must be almost
unity.
xk 1( i N ) (5.5)k Ti
This MAS technique is applied to 14 bus network and the obtained
results are compared with the work of Nagata et al (2002). The same 14 bus
sample network is extended to 30 bus network and is analysed with three
different fault conditions.
5.3 NEGOTIATION PROCESS BETWEEN AGENTS FOR A 14
BUS SYSTEM
To restore the de-energized areas, the negotiation process is the
main key process. In order to demonstrate the MAS, it has been applied to a
14 bus network which consists of 8 substations and 14 buses as shown in
Figure 5.2. Loads are indicated by arrows together with their magnitude. The
numeral on the branch shows its line flow capacity and s/s refers the
substation. A square in the diagram represents the status of the circuit breaker.
[black square- closed; white square- open]. Simulations are carried out for the
14 bus system in three different cases of fault.
107
5.3.1 Case 1 - Fault Between Substations B and D in 14 Bus
It is assumed that the fault occurs between the buses BUS2 and
BUS5. Due to the fault, the buses BUS5, BUS6, BUS11, BUS12, BUS13, and
BUS14 are de-energized which is shown in Figure 5.2. Hence the loads, L2,
L5, and L6 are not supplied with power. The main objective is to re-energize
all the buses and to supply maximum load. Hence the lists of buses that are
de-energized are placed in the DEBLIST.
Figure 5.2 Post Fault Network of 14 Bus for Case 1
DEBLIST is not only used to store the de-energized buses in list,
but also to group the buses according to the power level. Hence, when the
fault occurs between the buses BUS2 and BUS5, LMA5 sends the first signal
to GMA to restore it. GMA sends the signal to restore BUS5. LMA5
108
communicates with LMA3 and checks the available power in BUS3 which is
in healthy network. Since the total load in the de-energized area (4.0) is
greater than the line capacity between the buses, BUS3 and BUS5 (2.0), D-G
line is being opened.
When D-G line is opened, the total load in the de-energized area to
be restored through BUS3 is reduced to 2.0. The total power now flowing
through C-D line will be 2.0 which is equal to the capacity of the line thereby
not violating the constraint of the line capacity. While energizing each bus,
the corresponding LMAs send the signal to GMA in order to remove the bus
from DEBLIST.
Now the buses BUS11 and BUS12 are in DEBLIST. Hence, GMA
sends the signal to BUS11 for restoration. LMA11 communicates with LMA9
and checks the available power in BUS9 which is in healthy network, whether
it can supply the load in G substation. The available power in BUS9 (2.0) is
sufficient to supply the load L5 (2.0) and it is less than the capacity of the F-G
line. Hence F-G line is being closed. The buses BUS11 and BUS12 are
energized and are removed from DEBLIST. Since the DEBLIST becomes
empty, the objective of energizing all the buses in the network is achieved. A
program is written in C# language to obtain the output for the 14 bus network
and it is shown in Figure 5.3. The diagrammatic representation of restoration
plan is given in Figure 5.4.
109
Figure 5.3 Switching Operation Sequence for 14 Bus Case 1
Figure 5.4 Target Network Configuration of 14 Bus for Case 1
110
5.3.2 Case 2 - Fault between Substations B and D by Changing Line
Capacity in 14 bus
Figure 5.5 Post Fault Network of 14 Bus for Case 2
In the second case the same fault network is considered as in the
first case, but here the capacity of the line between the buses, BUS9 and
BUS11 is changed from 4.0 to 1.0 as shown in Figure 5.5. As soon as the fault
occurs, DEBLIST is created. Hence the buses to be restored are identified.
The same procedure is followed to re-energize the load in the network for the
14 bus network. The switching operation sequence and the network
configuration are shown in Figures 5.6 and 5.7 respectively.
111
Figure 5.6 Switching Operation Sequence of 14 Bus for Case 2
Figure 5.7 Target Network Configuration of 14 Bus for Case 2
112
5.3.3 Case 3 – One Fault Between Substations B and D, Another
Fault Between C and E in 14 Bus
In the third case, two faults are occurred, one fault is between
BUS2 and BUS5, and the other is between BUS4 and BUS7. Hence buses,
BUS5, BUS6, BUS7, BUS8, BUS11, BUS12, BUS13, and BUS14 are de-
energized due to the faults and are stored in DEBLIST. In this case the loads
L5 and L6 are interchanged. Hence L5 has the load 1.0 p.u and L6 has the
load 2.0 p.u. All the other parameters are as same as in first case. The
restoration plan for this fault and the corresponding target network
configuration is given in Figures 5.9 and 5.10 respectively.
Figure 5.8 Post Fault Network of 14 Bus for Case 3
113
Figure 5.9 Switching Operation Sequence of 14 Bus for Case 3
114
Figure 5.10 Target Network Configuration of 14 Bus for Case 3
5.4 NEGOTIATION PROCESS BETWEEN AGENTS FOR A 30
BUS SYSTEM
The negotiation process is the main method of restoring the
de-energized areas. In order to demonstrate the MAS, it has been applied to a
30 bus network which consists of 16 substations and 30 buses as shown in
Figure 5.11. Simulations are carried out for the 30 bus system in three
different cases of fault.
5.4.1 Case 1- Fault Between Substations B and D in 30 Bus
It is assumed that the fault occurs between the buses BUS2 and
BUS5. Due to the fault, the buses BUS5, BUS6, BUS11, BUS12, BUS13,
BUS14, BUS23, BUS24, BUS25, BUS26, BUS27, BUS28, BUS29, and
115
BUS30 are in the de-energized bus list (DEBLIST) which is indicated in
dotted background and the loads, L2, L5, L6, L11, L12, L13 and L14 are not
supplied with power. The main objective is to re-energize all the buses with
maximum load.
Figure 5.11 Post Fault Network of 30 Bus for Case 1
DEBLIST is not only used to store the de-energized buses in list,
but also to group the buses according to the power requirement. Hence, when
the fault occurs between the buses BUS2 and BUS5, BUS5 sends the first
signal to GMA to request for restoration. GMA sends the signal to restore
BUS5. LMA5 communicates with LMA3 and checks the available power in
BUS3 which is in healthy network. Since the total load in the de-energized
area (8.0) is greater than the line capacity between the buses 3 and 5 (5.0),
D-G line is being opened.
116
When D-G line is opened, the total load in the de-energized area to
be restored through BUS3 is reduced to 5.0. The total power now that need to
flow through C-D line will be 5.0 which is equal to the capacity of the C-D
line thereby not violating the constraint of the line capacity and hence C-D
line is closed. While energizing each bus, the corresponding LMAs send the
signal to GMA in order to remove the bus from DEBLIST. Now the buses
BUS11, BUS12, BUS23, BUS24, BUS25, and BUS26 are included in the
DEBLIST. Hence, GMA sends the signal to BUS11, for restoration. LMA11
communicates with LMA9 and analyses the available power in BUS9 which
is in healthy network, whether it can supply the load in G substation. Since
the total load in the de-energized area (3.0) is greater than the line capacity
between the buses, BUS9 and BUS11 (2.0), G-M line is opened.
When G-M line is opened, the total de-energised load to be restored
through BUS11 is reduced to 2.0. Since the power that needs to flow through
F-G line and capacity of F-G line are equal, the same line is being closed. The
buses BUS11, BUS12, BUS25 and BUS26 are energized and are removed
from DEBLIST.
Still buses BUS23 and BUS24 are in DEBLIST. LMA23 sends the
signal to GMA in order to restore BUS 23. GMA checks the available power
in LMA21. Since the available power is enough to restore the load and the
power needs to flow through L-M line (1.0) is less than the capacity of L-M
line (6.0), L-M line is closed. Hence buses BUS23 and BUS24 are energized
and are removed from the DEBLIST. Since the DEBLIST becomes empty,
the objective of energizing all the buses in the network is achieved. The
restoration plan and the target network reconfiguration for case 1 is given in
Table 5.1 and Figure 5.12 respectively.
117
Table 5.1 Switching Operation Sequence of 30 Bus for Case 1
S.No Station Operation Content
1 B S/S B-D Line Open
2 D S/S B-D Line Open
3 G S/S D-G Line Open
4 D S/S C-D Line Close
5 M S/S G-M Line Open
6 G S/S F-G Line Close
7 M S/S L-M Line Close
Figure 5.12 Target Network Configuration of 30 Bus for Case 1
118
5.4.2 Case 2- Fault Between Substations B and D By Changing Line
Capacity in 30 Bus
In the second case, it is assumed that the fault is between the buses
2 and 5 but the capacity of the line between the buses, BUS3 and BUS5 is
changed from 5.0 to 3.0 which is shown in Figure 5.13. As soon as the fault
occurs, DEBLIST is created. Hence the buses to be restored are identified.
Figure 5.13 Post Fault Network of 30 Bus for Case 2
119
GMA sends the signal to restore BUS5. LMA5 communicates with
LMA3 and checks the available power in BUS3, which is in healthy network.
Since the total load in the de-energized area (8.0) is greater than the line
capacity between the buses, BUS3 and BUS5 (3.0), D-G line is opened.
When D-G line is opened, the total load in the de-energized area to
be restored through BUS3 is reduced to 5.0. However, the load to be met (5.0)
through the bus, BUS5 from BUS3 is greater than the capacity of C-D line
(3.0). Hence, H-O line is opened which has the load, 2.0 and thereby the total
load that has to be restored by closing C-D line becomes 3.0. Now C-D line is
closed. The total power, now flowing through C-D line becomes 3.0, which is
equal to the capacity of the line, thereby not violating the constraint of the line
capacity.
Since the buses BUS11, BUS12, BUS23, BUS24, BUS25, BUS26,
BUS27 and BUS28 are in DEBLIST, GMA analyses the available power in
BUS9, which is in healthy network. The available power in BUS9 (2.0) is not
sufficient to supply the load (3.0) that has to be restored through BUS11.
Hence, G-M line is opened thereby reducing the load to be met through
BUS11 to 2.0. Hence, the power to flow through F-G line (2.0) in order to
supply the load is equal to the capacity of F-G line. Now F-G line is closed.
The buses BUS11, BUS12, BUS25 and BUS26 are energized and are
removed from DEBLIST.
Still buses BUS23, BUS24, BUS27 and BUS28 are in DEBLIST.
LMA23 sends the signal to GMA in order to restore. GMA checks the
available power in LMA21. Since the available power is enough to restore the
load and the power needs to flow through L-M line (1.0) is less than the
120
capacity of L-M line (6.0), L-M line is closed. Hence buses BUS23 and
BUS24 are energized and are removed from the DEBLIST.
Buses BUS27 and BUS28 yet to be energized are in the DEBLIST.
BUS27 can be energized only through BUS25. Hence LMA27 sends the
signal to GMA. GMA checks the availability of power in BUS25. Since the
load in O substation (2.0) cannot be met with the available power in BUS25
(0.0), the load L13 is cut and then the BUS27 and BUS28 are energized by
closing the N-O line and are removed from DEBLIST. Now, the DEBLIST
becomes empty. Though L13 load is cut, all the de-energized buses are re-
energized. The restoration plan for case 2 and the diagrammatic representation
of the network is shown in Table 5.2 and Figure 5.14 respectively.
Table 5.2 Switching Operation Sequence of 30 Bus for Case 2
S.No Station Operation Content
1 B S/S B-D Line Open
2 D S/S B-D Line Open
3 G S/S D-G Line Open
4 O S/S H-O Line Open
5 D S/S C-D Line Close
6 M S/S G-M Line Open
7 G S/S F-G Line Close
8 M S/S L-M Line Close
9 O S/S Load Cut
10 O S/S N-O Line Close
121
Figure 5.14 Target Network Configuration of 30 Bus for Case 2
5.4.3 Case 3 – One Fault Between Substations B and D, Another
Fault Between C and E in 30 Bus
In the third case, it is assumed that two faults occur between the
buses BUS2 and BUS5, and BUS4 and BUS7. In this case the loads L13 and
L14 are interchanged compared to the previous case. Hence L13 has the load
1.0 p.u and L14 has the load 2.0 p.u. All the other parameters are same as in
the second case as shown in Figure 5.15. Hence buses, BUS5, BUS6, BUS7,
BUS8, BUS11, BUS12, BUS13, BUS14, BUS15, BUS16, BUS17, BUS18,
BUS23, BUS24, BUS25, BUS26, BUS27, BUS28, BUS29, and BUS30 are
de-energized due to the faults and are stored in DEBLIST.
122
Figure 5.15 Post Fault Network of 30 Bus for Case 3
GMA sends the signal to restore LMA5. LMA5 communicates with
LMA3 and checks the available power in BUS3 which is in healthy network.
Since the total load in the de-energized area (8.0) is greater than the line
capacity between the buses, BUS3 and BUS5 (3.0), D-G line is opened. When
D-G line is opened, the total load in the de-energized area to be restored
through BUS3 becomes less (5.0). However, the load to be met (5.0) through
the bus, BUS5 from BUS3 is greater than the capacity of C-D line (3.0).
Hence, H-O line is opened which has the load 1.0 and thereby the total load
that has to be restored by closing C-D line becomes 4.0.
Since E substation is under de-energized area due to the fault
between buses BUS4 and BUS7, GMA sends the signal to restore BUS7.
LMA7 communicates with LMA9 and checks the available power in BUS9 to
123
restore BUS7. Since all the constraints are met, E-F line is closed and hence E
substation, I substation and J substation buses are energized.
Now the buses BUS5, BUS6, BUS11, BUS12, BUS13, BUS14,
BUS23, BUS24, BUS25, BUS26, BUS27, BUS28, BUS29 and BUS30 are in
DEBLIST. The total load under BUS5 that is to be restored is 4.0, which is
greater than the power that can be obtained from BUS3 (3.0). Hence H-P line
is opened. When the C-D line is closed, the buses, BUS5, BUS6, BUS13 and
BUS14 are energized and the loads L2 and L6 are supplied with power. Now
these buses are removed from DEBLIST.
Since the buses BUS11, BUS12, BUS23, BUS24, BUS25, BUS26,
BUS27, BUS28, BUS29 and BUS30 are in DEBLIST, GMA analyses the
available power in BUS9, which is in healthy network. The available power in
BUS9 (2.0) is not sufficient to supply the load (3.0) which is to be restored
through BUS11. Hence, G-M line is opened thereby reducing the load to be
met through BUS11 to 2.0. Hence, the power to flow through F-G line (2.0) in
order to supply the load is equal to the capacity of F-G line. Now F-G line is
closed. The buses BUS11, BUS12, BUS25 and BUS26 are energized and are
removed from DEBLIST.
Still buses BUS23, BUS24, BUS27, BUS28, BUS29 and BUS30
are in DEBLIST. LMA23 sends the signal to GMA in order to restore. GMA
checks the available power in LMA21. Since the available power is enough to
restore the load and the power needs to flow through L-M line (1.0) is less
than the capacity of L-M line (6.0), L-M line is closed. Hence buses BUS23
and BUS24 are energized and are removed from the DEBLIST.
Buses BUS27, BUS28, BUS29 and BUS30 that are yet to be
energized are in the DEBLIST. BUS27 can be energized only through
BUS25. Hence LMA27 sends the signal to GMA. GMA checks the
124
availability of power in BUS25. The load in O substation (1.0) cannot be met
with the available power in BUS25 (0.0). Hence the load L13 is cut and then
the BUS27 and BUS28 are energized and are removed from DEBLIST.
Finally, BUS29 and BUS30 are to be energized only through
BUS27. Since the available power in BUS27 is not sufficient to meet the load
in P substation (2.0), the load is being cut and then BUS29 and BUS30 are
energized by closing O-P line and are removed from DEBLIST. Now, the
DEBLIST becomes empty. Thus the objective function is satisfied. Table 5.3
includes the switching sequence for case 3 and Figure 5.16 explains the
implementation of restoration plan.
Table 5.3 Switching Operation Sequence of 30 Bus for Case 3
S.No Station Operation content
1 B S/S B-D Line Open
2 D S/S B-D Line Open
3 C S/S C-E Line Open
4 E S/S C-E Line Open
5 G S/S D-G Line Open
6 O S/S H-O Line Open
7 F S/S E-F Line Close
8 P S/S H-P Line Open
9 D S/S C-D Line Close
10 M S/S G-M Line Open
11 G S/S F-G Line Close
12 M S/S L-M Line Close
13 O S/S Load Cut
14 O S/S N-O Line Close
15 P S/S Load Cut
16 P S/S O-P Line Close
Generally, the restoration plan is obtained for the predetermined
condition of the electrical network, and the optimal path for the transmission
125
of power is not discussed in the available literature. But there are several
reports about MAS-based approaches for power system restoration. Takeshi
Nagata et al (2002) proposed MAS which consists of number of bus agents
and a single facilitator agent. In this process, the restoration model considers
the transmission capacity, power balance between the supply and demand,
available generation and finally the radial constraint. But the process does not
consider the safe operating voltage limit and the shortest path for power
transmission that may be helpful for loss reduction.
Figure 5.16 Target Network Configuration of 30 Bus for Case 3
Takeshi Nagata et al (2004) proposed a new decentralized multi-agent
simulator for a bulk power system restoration. The proposed MAS is
constructed with four-level hierarchical architecture with agents in each level.
After negotiating with agents, the restoration plan is obtained.
126
This initiates the need for modified MAS with the separate
restorative path search guidance tool which suggests the optimal path through
which the power can be transmitted. The optimal path considers the shortest
path by distance (minimum losses), capacity of the transmission line, power
balance between source and demand, priority of loads and allowable voltage
limits of the network. So, there will not be cascaded outages which help the
MAS to be more effective.
5.5 ARCHITECTURE OF MODIFIED MULTI-AGENT SYSTEM
The multi-agent approach consists of several Demand Agents
(DAs), Source Agents (SAs), Load Dispatching Agents (LDAs) and a single
Management Agent (MA) as shown in Figure 5.17. The system is constructed
with three-level architecture. MA is located in the upper level. LDAs are
located in the second level and several DAs and SAs are implemented at the
third level. DAs correspond to the individual load while the SAs include the
generator which has the different generator characteristics such as ramp rate
and capacity. LDAs are equipped with the restorative path search guidance
tool to identify the optimal path. The LDAs are given suggestion for
connecting the electrical subsystem to the actual network by the MA.
Figure 5.17 Architecture of Modified Multi-Agent System
MA
LDA LDALDA
SA DAASA SADAA DAA DAA DAA DAA
127
Source Agents (SAs) have the characteristics of each generator, the
capacity of the generator and the type of plant. They offer this information to
the LDAs and manage the supply of power to the load. Table 5.4 shows the
generator characteristics. It is assumed that the uniform ramp rate as
10MW/min.
Table 5.4 Generator Characteristics
Name of the
Power Station
Unit
ID
Type of
Plant
Capacity
(MW)
NCTPS
(station 1)
G1 Gas 210
G2 Steam 210
G3 Steam 210
ETPS
(station 2)
G4 Gas 110
G5 Steam 110
G6 Steam 110
G7 Steam 60
G8 Steam 60
MTPS
(station 3)
G9 Gas 220
G10 Steam 220
MAPS
(station 4)
G11 Steam 220
G12 Gas 220
NEYVELI
(station 5)G13 Steam 220
Demand Agents (DAs) are implemented in the third level. They
know the characteristics of each load, such as load demand. It sends the
request to LDAs to restore the load. The important loads are given priority
among the de-energized loads. An important step to pick up the load is to
connect the load nearest to the power sources in a power system network. A
128
load is an integration model of more number of customers. The load pickup
characteristic is shown in Figure 5.18.
Steps of Restoration
0 1 2 3 4 5 6
loa
d(%
)
0
20
40
60
80
100
120
Figure 5.18 Characteristics for Load Pick Up
Load Dispatching Agents (LDAs) are allocated to each electrical
zone. Power system network is divided into a number of electrical zones in
order to enhance the restoration effectively. The SAs and DAs within that
zone send the message to the corresponding LDA. LDA is equipped with
restorative path search guidance tool which is solved by the SPF algorithm
which was discussed in Chapter 3. The generator characteristics and load
characteristics are also stored here as data base. Based on this algorithm after
checking the constraints the optimal path for the power flow is decided.
Management Agent (MA) is located in the first level. LDAs send
all information to MA about the electrical zone such as the amount of load
restored, the available generation, optimal path through which power is
transmitted. On the whole, the MA will check the constraints by conducting
the load flow in order to validate the restoration plan.
129
5.6 EXECUTION PROCESS BY MULTI-AGENT SYSTEM
In this section, the restoration process using multi agent approach is
explained.
1. The black start generator starts in each electrical zone and the
current status of the generators are updated to LDAs by SAs.
2. LDAs calculate the serving electrical power and reply
message forwarded to SAs to supply power to the non-black
start generator.
3. At the same time, the DAs send information to LDAs such as
required power to restore the load. During black-out
condition, an important rule is to pick up the load nearest to
the power source.
4. LDAs acquire the available power of started SAs and figures
out a scheme of power assignment of each DA. The load
restoration process should proceed with the constraint and the
restorative path search guidance tool suggests the optimal path
for transmission of power. Then LDAs update this information
to MA.
5. MA validates the restoration plan by conducting the load flow
and sends the reply message to LDAs. If the limits are
violated, then the LDA will instruct the DA to shut down the
load. Only after getting the reply message, LDAs send the
information to SAs and DAs.
130
6. In each electrical zone more and more SAs are cranked, the
skeleton transmission lines are energized. In order to stabilize
generation and maintain normal voltage, sufficient load must
also be restored. The same process is continued until 60% of
the load in each electrical zone is supplied. Then the MA
decides about the interconnection of subsystem in order to
share the available source of power and the reconfiguration of
the entire power system network.
Different kinds of agents have different tasks during restoration.
The above mentioned process is only a frame work for power restoration.
The key process for the MAS is negotiation only. With the proper negotiation
between the agents, an efficient restoration plan is developed. In order to
enhance the restoration, the restorative path search guidance tool is used.
This will check the constraints like capacity of the transmission line, shortest
path, priority of loads and power balance between the source and the demand.
Based on the above mentioned procedures, the restoration plan is
obtained. Each LDA provides the values of generation and load to SAs and
DAs respectively. SAs increase its output power according to its
characteristics. It is to be noted that the transfer of power does not violate the
constraints since the optimal path is obtained from the restorative path search
guidance tool and the complete plan is validated by the MA.
5.6.1 Restoration Process for the Practical Network
The MAS is applied and a restoration plan is obtained for the 19
bus practical network. Initially, none of the generators are functioning and all
circuit breakers are open. Considering crew constraints, in every power
131
station it is assumed that the generators are turned on one after another with
the ramp rate of 10 MW per minute. The SAs and DAs are numbered
corresponding to their respective bus numbers. The SAs send information
about the available power and DAs send message about the requirements of
power to the zone LDA. The available power in the generating station is
allocated to the load based upon the load characteristics. The optimal path
through which power can be transmitted is given by the restorative path
search guidance tool.
Initially, the practical network is divided into two electrical zones
and it is separated based on the available generation in each electrical zone so
that the generation matches the demand. After the zone recovers 60% of the
load, it can be connected together. The power station NCTPS (bus no 1) and
ETPS (bus no 2) along with the load buses 3, 4, 5, 6, 8 and 9 are grouped as
one zone. The remaining load buses 7,11,12,13,15,16 and 17 along with the
generating stations 14,18 and 19 are grouped as second zone. The load bus 10
will act as a boundary bus between the zones. The GAs at Bus1, Bus2, Bus14,
Bus18 and Bus19 start communicating the amount of power generated by the
respective generating stations to the corresponding zone LDA. At the same
time, all the DAs send a request to the LDA to start the restoration process.
On receiving this request, the LDA decides to energize the buses that are
directly connected in one step to the generating stations.
In the first step (1-4 minutes), the power generated by the stations is
given in Figure 5.19. The load bus which is nearest to the generating stations
is picked up first. So the generating sources in NCTPS (bus no 1) supply the
load to DA10, DA4, DA5, DA6 and DA9. The generating sources in ETPS
(bus no 2) supply the load DA3 and DA4. Thus in electrical zone 1, the
largest power consumer bus 4 (Mylapore) is supplied by two generating
132
stations. Similarly, in the other electrical zone, the three generating stations
MTPS (bus no 14), MAPS (bus no 19) and NEYVELI (bus no 18) energize
the load DA10, DA11, DA12, DA13, DA15, DA16 and DA17. At the end of
the first step, all the loads are supplied 20% of their respective load and is
shown in Figure 5.19.
Figure 5.19 Reconfiguration of the Network for I Step
In the second step (5- 8 minutes) all the load buses recover 40% of
their load. The power generated by each station is given in the Figure 5.20. In
the fifth minute, the load buses DA3, DA5, DA7, DA13, DA15 and DA17
recovers their 40% of load. After energization the DAs inform the LDA about
the updation of load. The load buses DA4, DA6, DA12 and DA16 obtain 40%
of their load in the sixth minute.
133
Figure 5.20 Reconfiguration of the Network for II Step
In the third step (9-11 minute), all the loads are supplied 60% of
load so the corresponding LDA decides about the interconnection of electrical
zones at the end of the 11th minute and it is shown in Figure 5.21. The entire
network is reconfigured but the partial load is supplied.
134
Figure 5.21 Reconfiguration of the Network for III Step
In the fourth step (12-18 minute) the load buses are energized 80%
of load and it is shown in Figure 5.22. Before updating the network, the MA
conducts the load flow to validate the plan and then only the SAs and DAs are
informed about the energisation of load.
135
Figure 5.22 Reconfiguration of the Network for IV Step
In the final step (19-22 minute) all the loads are supplied the full
load and the entire network is reconfigured through the optimal path which
considers the shortest path by distance, priority of loads, power balance
between generator and demand, voltage limit and capacity of the transmission
line. The complete reconfiguration of the network is shown in Figure 5.23.
With the integration of optimal path guidance with the MAS enhances the
network restoration and reduces the cascaded outages.
136
Figure 5.23 Reconfiguration of the Network for V Step
The optimal path is given in Table 5.5, and it is clear that some of
the load buses are not supplied from a single generating source. For example,
the load bus 3 is energised by the sources 1 and 2. For all the load buses the
optimal path for the power transmission is given in the same table. The
restorative path search guidance tool could suggest the other available paths
also for the transmission of power. If the power system operator is not
available or the equipment is not working, then the other paths could be
considered for power transmission which is also given in the same table. For
load buses 13, 15, 17 there is only one path for the supply of power. Hence
that line is in fault condition, then the area around that station could not be
restored.
137
Table 5.5 Optimal Path for the 19 Bus Practical Network
Generat
ing
station
Loads to
be
connected
Bus Optimal path Other available path
From To
1, 2,
4,18,19
3, 4, 5, 6,
7, 8, 9, 10,
11, 12, 13,
15, 16, 17
1 3 1 3 3
2 3 3
1 4 1 4 4
2 4 4
1 5 1 5 5
1 6 1 6 6
14 7 14 15 17 7 14 15 13 17 7
19 19 13 17 7 19 13 10 7
1 8 1 8 10 8
1 9 1 9 10 9
1 10 10 10
14 11 14 15 13 10
11
14 15 17 10 11
19 19 13 10 11 19 13 17 10 11
18 18 17 10
11
18 17 16 10 11
19 12 19 13 10 12 19 13 17 10 12
19 13 19 13 Only one path
18 18 17 13 18 17 10 13
14 15 14 15 Only one path
19 19 13 15 19 13 17 15
14 16 14 15 17 16 14 15 13 10 16
19 19 13 17 16 19 13 10 16
18 18 17 16 18 17 10 16
18 17 18 17 Only one path
138
Based on the optimal path guidance tool, the restoration plan is
obtained for the 19 bus practical network. In Table 5.6 the load bus is given in
numerals without the bracket and the numerals within the bracket represent
the generating station which is supplying the load. For example, 10(1) denotes
that load bus 10 is supplied by the generating station 1. For all the generation-
demand pair, the optimal path is given in the Table 5.5.
Table 5.6 Restoration Plan for 19 Bus Practical Network
Time (minute) Major Events
1 Energise load bus 10(1), 5(1), 3(2), 16(14),
15(19), 13(18)
2 Energise load bus 6 (1), 4(2), 7(14), 13(19),
17(18)
3 Energise load bus 4(1), 9(1), 11(14), 7(19),
12(19), 11(19), 16(18)
4 Energise load bus 8(1), 7(14)
All the load buses attains 20% of their load
8 All the load buses attains 40% of their load
11 All the load buses attains 60% of their load.
MA decides for interconnection of electrical
zone
18 All the load buses attains 80% of their load
22 All the load buses attains 100% of their load
The load flow is conducted for each minute in MATLAB and the
voltage profile of all the buses for the five steps are included in the Table 5.7.
The voltage of all the buses lies in the safe operating limits.
139
Table 5.7 Voltage Profile of All Buses in the Practical Network
Bus NoVoltage Profile (p.u)
Step 1 Step 2 Step 3 Step 4 Step 5
1 1.000 1.000 1.000 1.000 1.000
2 1.000 1.000 1.000 1.000 1.000
3 0.998 0.997 0.996 0.995 0.994
4 0.997 0.993 0.992 0.989 0.988
5 0.999 0.997 0.995 0.991 0.988
6 0.998 0.996 0.993 0.986 0.981
7 0.997 0.991 0.985 0.966 0.950
8 0.996 0.991 0.985 0.976 0.968
9 0.996 0.991 0.986 0.977 0.968
10 0.996 0.991 0.984 0.968 0.955
11 0.992 0.982 0.970 0.948 0.927
12 0.996 0.990 0.983 0.965 0.952
13 0.995 0.983 0.971 0.929 0.901
14 1.000 1.000 1.000 1.000 1.000
15 0.994 0.991 0.967 0.916 0.951
16 0.994 0.996 0.976 0.950 0.953
17 0.995 0.995 0.974 0.937 0.966
18 1.000 1.000 1.000 1.000 1.000
19 1.000 1.000 1.000 1.000 1.000
140
Table 5.8 Computation of Losses for the 19 Bus Practical Network
Time
(minute)
Total Losses (MW)
MAS with Optimal
Path
Power World Simulator
1 1.29 29.40
2 6.56 33.83
3 17.07 42.91
4 30.72 54.71
5 43.89 66.24
6 65.81 91.31
7 93.91 109.96
8 117.53 130.86
9 149.72 158.99
10 186.88 192.20
11 229.75 248.55
12 259.47 275.56
13 304.15 317.84
14 361.09 371.95
15 418.17 426.57
16 469.15 475.19
17 528.27 530.41
18 597.89 602.71
19 673.75 675.64
20 764.38 766.84
21 859.75 869.44
22 935.83 937.82
141
The important criteria while transmitting power is to select the path
with minimum losses. Since MAS is integrated with the optimal path search
guidance tool, the shortest path is selected for the power transmission which
automatically minimizes the losses. The total loss in the network is computed
using the MAS with optimal path search tool and the Power world simulator.
From the Table 5.8, it is signified that the losses in the MAS is very much
reduced.
5.7 CONCLUSION
The three-level architecture MAS along with the restorative path
search guidance tool is explained with the 19 bus practical network. In
today’s power system operation, the restoration plan is prepared only for the
predetermined fault conditions of the network. Even for a slight change in the
status of the power system network, the restoration plan has to be updated
under the stressful condition. Under such situation this guidance tool can be
used effectively for changing the network topology and it has the provision to
update the present status of the network.Outages and faults in interconnected
power systems may cause cascading sequences of events, and catastrophic
failures of power systems. So that the cascaded outages can be greatly
reduced with the help of this technique since it considers the entire essential
constraints.
The SPF algorithm based MAS technique provides the guidance to
the power system operators to solve the power system restoration problem for
any type of network topology.