A Framework for Generating Synthetic Distribution Feeders using

A Framework for Generating Synthetic Distribution Feeders using Open. Street. Map Shammya Saha, Eran Schweitzer, Anna Scaglione, Nathan Johnson Arizona State University 51 st North American Power Symposium

Why do we need synthetic distribution feeders? • Confidentiality and security policies governing critical infrastructure data can limit researchers from accessing real power systems data needed for scientific development. • For decades, researchers have extensively used a small set of standardized networks such as the IEEE transmission and distribution test cases. A Framework for Generating Synthetic Distribution Feeders using Open. Street. Map 2

Prior Work Transmission System ACTIVS (Birchfield et al. , 2017) Feeder Generation Using Statistical Analysis (Schweitzer et al. , 2017) Synthetic Feeder Generation Distribution System Reference Network Model (Domingo et al. , 2011 ) Smart DS (Hodge et al. , 2016 ) A Framework for Generating Synthetic Distribution Feeders using Open. Street. Map 3

Our Contribution ACTIVS Statistical Information from prior literature + Developed Geo-Embedded Synthetic Feeder Network Information from Open. Street. Map • Data requirement is less than RNM. • Includes geographic information in the graph generation process. • A general framework to adapt any statistical information for load distribution. A Framework for Generating Synthetic Distribution Feeders using Open. Street. Map 4

Preliminaries Assumptions Input Output • The power distribution network is radial and follows the road network. • The framework produces positive sequence synthetic feeders. • Geographic coordinates and apparent power demand of substations with positive net real power from ACTIVS • Boundary of zip codes and population information for census blocks • A cable database with impedance per unit length and MVA capacity. For a chosen substation, the attributes of the distribution feeder that includes line parameters (length, resistance, reactance), nodal MVA demand node voltages A Framework for Generating Synthetic Distribution Feeders using Open. Street. Map 5

Process Flow 1. Retrieve zip code information 2. Find all substations under the zip code from ACTIVS case 3. Retrieve the 'drive' network using Open. Street. Map 4. Create Voronoi regions 5. Distribute the load MW and MVAR within chosen region following any distribution 6. Update distribution network topology with load information 7. Run power flow using linearized power flow equations 8. Convert to Open. DSS model for non-linear AC power flow solution A Framework for Generating Synthetic Distribution Feeders using Open. Street. Map 6

• The chosen substation belongs to the zip code 85212 which belongs to the city of Mesa, Arizona. A Framework for Generating Synthetic Distribution Feeders using Open. Street. Map 7

Zip code map of 85212 A Framework for Generating Synthetic Distribution Feeders using Open. Street. Map 8

A Framework for Generating Synthetic Distribution Feeders using Open. Street. Map 9

Two substations under zip code 85212 results in two Voronoi regions. A Framework for Generating Synthetic Distribution Feeders using Open. Street. Map 10

Substation marked by blue Substation marked by black A Framework for Generating Synthetic Distribution Feeders using Open. Street. Map 11

Algorithm for connecting isolated nodes A Framework for Generating Synthetic Distribution Feeders using Open. Street. Map 12

A Framework for Generating Synthetic Distribution Feeders using Open. Street. Map 13

1. Minimum spanning tree algorithm is applied to convert the graph to a radial network. 2. The node closest to the location of the substation is chosen as the slack bus. 3. The total demand of the substation is then distributed among the nodes. A Framework for Generating Synthetic Distribution Feeders using Open. Street. Map 14

• A Framework for Generating Synthetic Distribution Feeders using Open. Street. Map 15

Population information excluded Population information included A Framework for Generating Synthetic Distribution Feeders using Open. Street. Map 16

Lossless Branch flow equations for radial network M. Farivar, L. Chen, and S. Low, “Equilibrium and dynamics of local voltage control in distribution systems, ” in 52 nd IEEE Conference on Decision and Control, Dec 2013, pp. 4329– 4334. A Framework for Generating Synthetic Distribution Feeders using Open. Street. Map 17

Selecting appropriate cable for each line What we know so far • Graph (location of nodes and edges) • Nodal real and reactive power demands • Edge length A Framework for Generating Synthetic Distribution Feeders using Open. Street. Map 18

Steady state voltage profile from Open. DSS A Framework for Generating Synthetic Distribution Feeders using Open. Street. Map 19

Extension to multi-phase model 1. Select a total number of three-phase lines as a percentage of the total number of lines. 2. Incrementally step through the power lines from the source to end-loads. 3. Sum the real power demand across individual singlephase lateral to create the set of single-phase nodes. 4. Assign phases to the single-phase nodes by solving an optimization problem to achieve balance among total real power load value per phase. A Framework for Generating Synthetic Distribution Feeders using Open. Street. Map 20

Future Work 1. Generating unbalanced three-phase synthetic distribution feeder with multiple voltage level 2. Incorporating shunt capacitors and voltage regulators into the model based on statistical analysis 3. Validation of generated models against actual network by statistical analysis 4. Attaching load profiles to individual node by dispersing the aggregated load data A Framework for Generating Synthetic Distribution Feeders using Open. Street. Map 21