Optimal Planning of Power Distribution Networks in Rural Areas PACE Seminar 2017 2 nd March 2017 James Fletcher Tyrone Fernando Shervin Fani Herbert Iu Mark Reynolds
Outline Research area – Power distribution network planning • Core goal • Complexity of the problems • Approaches and constraints when solving the problems Traditional network planning + Microgrid planning • Motivation • Objectives • Methodologies • Constraints • Results • Ongoing and future work
Power Distribution Network Planning (Western Power, 2016) To minimize the costs of constructing and operating the traditional power network according to technical and operational constraints. (SpidaView 2016)
Problem Complexity Base problem: • Nodes (load points and source) • Edges (poles and wires) • Radiality constraint Complexity: • NP-hard • Exponential (or worse) steps 1E+11 E.g. 1E+10 1E+09 • Brute-force for NP-hard 100000000 10000000 problem may be factorial Steps 1000000 100000 10000 1000 • NP-hard problems still solvable by 100 deterministic algorithms given the 10 1 problem size is small enough 0 5 10 15 Nodes
Optimisation Methodologies Deterministic Methods 6 x 10 6.43 • MILP 6.425 • MINLP 6.42 • Dynamic Programming 6.415 Y Location (m) 6.41 6.405 6.4 Heuristic Methods • 6.395 Simulated Annealing 6.39 • Ant Colony System 6.385 • Swarm Intelligence 6.1 6.2 6.3 6.4 6.5 6.6 6.7 6.8 X Location (m) 5 x 10 • Genetic Algorithm Three Phase Node Single Phase Node Complexity ~ NP-hard • Deterministic methods for small-medium problems • Heuristics for small-large problems
The Western Power Network • 100,000+ kms of network • Infrastructure aging • Continual organic addition of customers over lifetime of the network • Is rebuilding the existing network the best solution? (Western Power 2016)
Network Optimisation Methods Method 1 • Re-route the network connections to reduce costs while meeting constraints. • Use genetic algorithm (GA) Method 2 • Introduce stand-alone power systems (SPS) to supply electricity to isolated loads by generating power at the point of demand. • Develop network tracing algorithm to find candidates
Stand-alone Power Systems (SPS) What is an SPS? • An off-grid power system used to supply electricity to isolated loads by generating power at the point of demand. • Renewable (Solar, Wind etc.) and Non-renewable (Diesel) • Batteries banks are often used as backup to store energy. (Western Power 2016)
Network Routing - GA Genetic Algorithm What is the GA? How does the GA work? Initial • Optimisation metaheuristic which Population mimics the evolutionary process through the use of genetic operators Selection Why did we choose the GA? • Ability to optimise complex combinatorial problems New • Application in literature Population
Network Routing - GA Genetic Algorithm Objective Function Procedure
Network Routing – IP + Selection Initial Population • Minimum Spanning Tree • Random reformations Selection • Stochastic Tournament Gr Best 1 Gr Best Population 2 . . .
Network Routing – Operations Genetic Operations a. Direct Steiner point b. Indirect Steiner point c. Steiner point removal d. SPS addition e. SPS Removal f. Reconfiguration
Network Routing – Constraints • Voltage deviation limit • Radiality • Geographical weights
Network Routing – Case Study Reduction ~11% • Improvement over the existing network architecture
Network Routing – Case Study Reduction Reduction ~11% ~11% ~20% • • Stand-alone power systems provided optimisation beyond grid connections Stand-alone power systems provided optimisation beyond grid connections
Network Routing - Results
Network Routing - Assumptions Locked 3 phase Single phase network doesn’t need to be next to roads Fixed discount rate Outage costs are not yet incorporated No decommissioning cost
SPS Motivation
SPS – Bushfire Response Area 1 6 6 6 KDNZ340-SPS Network KDNZ340-SPS Network KDNZ340-SPS Network x 10 x 10 x 10 6.338 6.33 6.3365 6.336 6.336 6.334 6.328 Correlated poles Correlated poles Correlated poles 6.3355 6.332 Correlated DTs Correlated DTs Correlated DTs Y Location (m) Y Location (m) Y Location (m) Uncorrelated Poles Uncorrelated Poles Uncorrelated Poles 6.326 6.33 6.335 Uncorrelated DTs Uncorrelated DTs Uncorrelated DTs 3ph DT 3ph DT 3ph DT 6.328 6.3345 1ph DT 1ph DT 1ph DT 6.324 3ph Removed Network 3ph Removed Network 3ph Removed Network 6.326 6.334 1ph Removed Network 1ph Removed Network 1ph Removed Network 3ph Network 3ph Network 3ph Network 6.324 6.322 6.3335 1ph Network 1ph Network 1ph Network 6.322 6.333 6.32 6.32 6.3325 6.318 6.7 6.77 6.82 6.75 6.78 6.84 6.79 6.8 6.86 6.8 6.85 6.81 6.88 6.9 6.82 6.9 6.83 6.95 6.92 6.84 7 X Location (m) X Location (m) X Location (m) 5 5 5 x 10 x 10 x 10
SPS – Bushfire Response Area 2 6 6 6 6 KDNZ341-SPS Network KDNZ341-SPS Network KDNZ341-SPS Network KDNZ341-SPS Network x 10 x 10 x 10 x 10 6.33 6.303 6.322 6.312 6.3025 6.325 6.3115 6.321 6.302 6.311 6.32 Correlated poles Correlated poles Correlated poles Correlated poles 6.3015 Correlated DTs Correlated DTs Correlated DTs Correlated DTs 6.3105 Y Location (m) Y Location (m) Y Location (m) Y Location (m) 6.32 Uncorrelated Poles Uncorrelated Poles Uncorrelated Poles Uncorrelated Poles 6.315 6.301 Uncorrelated DTs Uncorrelated DTs Uncorrelated DTs Uncorrelated DTs 6.31 3ph DT 3ph DT 3ph DT 3ph DT 6.3005 6.31 1ph DT 1ph DT 1ph DT 1ph DT 6.3095 6.319 3ph Removed Network 3ph Removed Network 3ph Removed Network 3ph Removed Network 6.3 6.305 1ph Removed Network 1ph Removed Network 1ph Removed Network 1ph Removed Network 6.309 6.2995 3ph Network 3ph Network 3ph Network 3ph Network 6.318 6.3085 6.3 1ph Network 1ph Network 1ph Network 1ph Network 6.299 6.308 6.295 6.2985 6.317 6.3075 6.298 6.29 6.9 6.995 7.155 7.01 7.16 7 6.95 7.005 7.165 7.02 7.01 7.17 7 7.03 7.015 7.175 7.05 7.02 7.18 7.04 7.025 7.185 7.1 7.05 7.03 7.19 7.035 7.15 7.195 7.06 7.04 7.2 7.2 X Location (m) X Location (m) X Location (m) X Location (m) 5 5 5 5 x 10 x 10 x 10 x 10
SPS – Bushfire Response Area 3 6 6 6 KATZ44-SPS Network KATZ44-SPS Network KATZ44-SPS Network x 10 x 10 x 10 6.238 6.228 6.2245 6.236 6.224 6.227 6.234 Correlated poles Correlated poles Correlated poles 6.2235 Correlated DTs Correlated DTs Correlated DTs 6.232 Y Location (m) Y Location (m) Y Location (m) Uncorrelated Poles Uncorrelated Poles Uncorrelated Poles 6.226 6.223 Uncorrelated DTs Uncorrelated DTs Uncorrelated DTs 6.23 3ph DT 3ph DT 3ph DT 6.2225 1ph DT 1ph DT 1ph DT 6.225 6.228 3ph Removed Network 3ph Removed Network 3ph Removed Network 6.222 1ph Removed Network 1ph Removed Network 1ph Removed Network 6.226 3ph Network 3ph Network 3ph Network 6.2215 6.224 1ph Network 1ph Network 1ph Network 6.224 6.221 6.222 6.223 6.2205 6.22 5.25 5.48 5.4 5.485 5.3 5.49 5.41 5.35 5.495 5.42 5.5 5.4 5.505 5.43 5.45 5.51 5.44 5.515 5.5 5.52 5.45 5.55 5.525 5.53 5.46 5.6 X Location (m) X Location (m) X Location (m) 5 5 5 x 10 x 10 x 10
SPS – Bushfire Outcomes Cost Simulation Current Number of Scenario Reduction Time Network (km) SPS (per unit) (seconds) KDN/Z340 68.5 12.70 9 11.6 KDN/Z341 71.0 4.56 3 12.4 KAT/Z44 76.0 2.90 2 17.8 • SPS solutions do exist and are quickly found SPS Selection • Operational costs Planned Works • Visual Relationship to a Financial Relationship
SPS Identification – Modelling For SPS identification it is necessary to know: 2. Customer Load 1. Network Formation F EEDER Network Formation B ACKBONE (Grid) • Pole IDs & D ISTRIBUTION Locations S UBSTATION • Transformer IDs S PURS Database & Locations • Bay IDs & Connections • Poles • Transformers Equipment Meter SPS Customer Load Mapping Usage Size • Meters (SPS) (kWh) (kWh) Trans ID Meter ID • Switches • Protection 123 10 12.0 20 • Equipment Devices Mapping 123 11 35.3 40 • Power Quality • Historical Meter 123 12 1.2 2 Devices Consumption Etc… Data
SPS Identification - Path Tracing
SPS Identification - Path Tracing • Finds all individual edge of network opportunities • Can miss groups of SPS candidates
Application of SPS Algorithm Network Details: • 191 customers • Loads: 0 – 50 kWh/day • 580 km of network • 60 km x 80 km Results: Base SPS SPS with Constraints Number of SPS 99 (52%) 61 (32%) Pole Reduction 39 % 28 %
Conclusions and Future Work Conclusions: • SPS identification model has identified numerous SPS opportunities • Some SPS candidates provide significant financial benefits Future Work: • Data quality issues • Consideration of risk and reliability • Customer clustering mechanic • Shift focus to GA
SPS Assumptions Investment period is 50 years Fixed discount rate Cost parameters are correct Outage costs are not yet incorporated No decommissioning cost Geographical constraint not taken into account
Recommend
More recommend
