• Tidak ada hasil yang ditemukan

Proposed Distributed Control Model for Service Restoration

Dalam dokumen Smart Trends in Computing and Communications (Halaman 98-103)

Agents-Based Service Restoration in Electrical Secondary Distribution

8.4 Proposed Distributed Control Model for Service Restoration

further take advantage of the renewable generation sources installed at the customer sides for increased capacity.

8.4 Proposed Distributed Control Model for Service Restoration

The SDN is large in size and complex with a number of branches and laterals as seen in Fig.8.3. Adding intelligence to this part of the network requires robust algorithms and may lead to large amount of data as a result, centralized control approaches may not be appropriate. This is due to the requirements of large processing unit and may suffer from single point of failure. The aim is to take advantage of computer technologies which has inverted distributed control approaches with more than one decision-making units for deploying in complex network. In this study, distributed control based on multi-agent system is used.

The proposed distributed algorithm for service restoration comprises of four agents, namely control agent (CA), grid agent (GA), load agent (LA) and switch agent (SA). The grid agent once resides in the area under fault, becomes the con- trol agent. Table8.1describes the input, actions to be performed, conditions to be met and output by different agents. Features and capability of these agents will be implemented in the Raspberry Pi in which all decision making will be done and then send control commands to the actuators. In built processors for the smart meters

88 R. J. Mwifunyi et al.

Fig. 8.3 Tanzania secondary distribution network Table 8.1 Agent operation

Control agent Grid agent Load agent Switch

agent Input • Forecasted load

demand

• Switch status

• Load demand

• Available capacity

Fault information Switching request Actions • Receive request for

power supply from load agent

• Send power request to the nearby GA

• Run optimization problem

• Sends switching request to the SA

• Receive request for power supply from nearby GA

• Forecast the available capacity based on loading history

• Request the power from the CA

• Forecast load for the estimated fault duration

Sends control command to the circuit breakers and smart isolators Condition Pgrid Pload Pavailable Pload Psupply Pload Pre-fault

status Output • Load to be restored

• Voltage levels

• Load to shed

Remaining capacity • Forecasted load

• Shedding load

Switch status

8 Agents-Based Service Restoration in Electrical Secondary … 89

SENSORS

Raspberry Pi Contactor

Transformer

Communication Infrastructure

Fig. 8.4 Actuation node

throughout the secondary distribution network will be used as the load agents, and microprocessors connected to intelligent switches as the switch agents. The system design architecture for the actuators to be used in the secondary distribution network is shown in Fig.8.4in which sensors sense voltage and current levels and then send information to the Raspberry Pi which processes and sends the control commands to the contactors or relays which performs actuation during service restoration.

8.4.1 Objective Functions

Service restoration problem in power distribution system is a multi-objective, multi- constraint and nonlinear optimization problem. Among others, three objective func- tions have been found to be key for the service restoration in the secondary distri- bution network due to the overall physical nature of the network and overall needs for the service reliability. These objectives are maximizing the number of restored

90 R. J. Mwifunyi et al.

customers [10,19], minimizing power losses [4] and minimization of the cost asso- ciated with load shedding [12] as stated in (8.1–8.3). Together with the mentioned objective functions, voltage limits, line current limits and available capacity limits for the load transfer constraints need to be satisfied. The cost associated with load shedding will merely be attributed by the amount of load demand and/or the priority of load in which high priority customers will have higher cost as compared to lower priority customers.

Maximizing the number of restored loads based on their priorities.

maxf(x)=

N

i=1

wi×Li×yi (8.1)

whereLi: the load at busi,yi: status of the load at the busi,wi: priority level of the load at busi,x: network configuration undergoing service restoration represented by status of switches andN: total number of branches.

Minimization of power loss.

Ploss=

N

i=1

Ii2Ri (8.2)

where Plossis the total power loss, Ii is the current through branchiand Ri is the resistance of branchi.

Minimizing cost (C) associated with Load Shedding (LS). The cost associated with load shedding is directly related to the amount of load shed and the criticality of the shedding load.

min

iΓLS

CiLS×LSi (8.3)

whereCiLSis the cost associated with shedding load at busiand LSi is the amount of load to be shed at busi.

8.4.2 Recommendations for the Improvement of the SDN

For realization of the self-healing in the Tanzanian secondary distribution networks, the following issues need to be taken into consideration:

• Motorized air circuit breakers should be installed at the distribution transformers to allow the protection of the transformers as shown in Fig.8.5. The circuit breakers will comprise of the contactors for making and breaking purposes, as well as microcontrollers with decision-making capabilities.

8 Agents-Based Service Restoration in Electrical Secondary … 91

MCB MCB

ATS

Smart Isolator Switch

Transformer 1 Transformer 2

Fig. 8.5 Proposed network topology to support SR

• Smart switches along the branches also need to be installed to facilitate the load shedding in case the power supply is not sufficient enough to supply all loads in the network.

• Making use of the renewable energy sources to allow restoration of more loads apart from high priority loads.

• Open wire jumpers at the point of interconnection between two secondary distribution networks can be replaced with the automatic load transfer switches.

8.4.3 Design for the Equipment Specifications to Be Installed to Support Service Restoration in the Pilot Site

Specifications for the low voltage automation equipment are mainly based on IEC- 60947 standard. General specifications for the equipment to be used in Tanzanian electrical SDN are as follows: AC electrical properties, applicable at electrical dis- tribution level, support remote control operation, 400 V rated voltage and operate at 50 Hz frequency. Moreover, other parameters like the rated current in (8.4) and short circuit current in (8.5) vary with transformer capacity. Table8.2summarizes the design specifications for three distribution transformers in terms of breaker ratings.

92 R. J. Mwifunyi et al.

Table 8.2 Circuit breaker

specifications Item Application Specifications

Motorized circuit breaker

315 kVA transformer

Rated current: 455 Short circuit current: 12 kA 200 kVA

transformer

Rated current: 289 Short circuit current: 7.2 kA 100 kVA

transformer

Rated current:

144 A Short circuit current: 3.6 kA

Rated_Current= S

√3×V (8.4)

Isc= Rated_Current

Z% (8.5)

whereSis the rated transformer capacity,Vis the rated voltage andZis the impedance of the transformer.

Dalam dokumen Smart Trends in Computing and Communications (Halaman 98-103)