SRP EV Adoption Propensity and Transformer Load Management Jeff - - PowerPoint PPT Presentation

Jeff Loehr Senior Engineer National Electric Transportation Infrastructure Working Council October 24, 2018

SRP EV Adoption Propensity and Transformer Load Management

2

Project Goals

• Propensity to Adopt

– Translate system wide corporate EV forecast into locations – Identify possible problem areas

• Transformer Load Management (TLM)

– Speed up analysis to identify transformer overloads caused by EV charging clusters

• Both are focused on residential charging

3

Metering

• Over 1.07 Million meters

– 60% have 15-minute interval kWh data – 38% have 15-minute interval kWh, kVarh and average voltage

Service Transformers

• 163,000 banks comprised of 173,000 units
• 78% are pad-mounted
• 33% are 50kVA single-phase transformers
• 33% are 75kVA single-phase transformers

SRP System Information

4

Background

Expected Growth of DER and EVs Maintain System Reliability Strategy

5

Predictive Analytics

Corporate Forecast + Locations Customer Propensity Model

6

Customer Propensity Model

Known owners of EV and PV Demographic data Model Machine learning Random forest Find customers with similar attributes as known adopters Score each customer’s likelihood of adoption Data

7

EV Adoption in the Service Territory

4,985 6,682 8,742 10,917 13,200 15,684 18,251 20,892 23,706 26,652 3% 4% 5% 7% 8% 10% 11% 13% 15% 16% 0% 2% 4% 6% 8% 10% 12% 14% 16% 18% 5,000 10,000 15,000 20,000 25,000 30,000 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026

% of Service Xfmrs Fleet # Service Xfmrs

Service Transformers with EVs

2,000 4,000 6,000 8,000 10,000 12,000 14,000 1 2 3 4 5 6 7 8 9 10

# Service Xfmrs # Electric Vehicles

A more granular look….

8

Validating EV Growth Forecast

9

EV Model Performance

SRP has ~ 1,400 feeders in the service territory

10

EV Model Performance (cont’d)

• 153 EV adopters (Oct 2017 – Dec 2017)
• 76% have scores in the top 25%

Scored ~1 million existing, residential customers in the service territory Top 25%

11

Lessons Learned

• No immediate concerns for system reliability
• Limitations with propensity modeling
• Collaboration between data scientists and business areas
• Need to prepare for the future

12

Prior to interval data

• First implemented at SRP in 1960

– Run annually until 1993 – Run weekly 1994-2015

• Methodology

– Monthly billing kWh energy

• Estimate peak kW demand

– kW demand for some commercial / industrial customers – Coincidence factors – Assumed 85% power factor

Transformer Load Management (TLM) at SRP

13

Interval data

• Started in 2015
• Use interval data where available

– Increasing over time

• Improved power factor estimation
• Fall back to old methodology
• Run daily
• Overloads evaluated annually

Transformer Load Management (TLM) at SRP

14

Thermal modeling

• No longer fall back to old methodology
• Identify bad interval meter reads
• Estimate interval data where missing
• IEEE C57.91
• Simple model

– Need model for each size/type

• New overload criteria

– Hotspot temperature

• Overloads evaluated within 10 days

Transformer Load Management (TLM) at SRP

15

Transformer A

• Peak Load 164%
• Peak Hotspot 143 °C

Transformer B

• Peak Load 182%
• Peak Hotspot 142 °C

Transformer Load Management (TLM) at SRP

16

2017 study

• Thermal modeling
• Thermal overload criteria
• Sample load profiles
• Various charging start times
• Assume uniform distribution of EVs across transformer fleet

Electric Vehicle Impacts

17

Scenario A

• No EVs
• Peak Load 116%
• Peak Hotspot 97 °C

Scenario B

• 6 EVs charging at 4pm
• Peak Load 172%
• Peak Hotspot 127 °C

Scenario C

• 6 EVs charging at 11pm
• Peak Load 126%
• Peak Hotspot 97 °C

Electrical Vehicle Impacts

18

75 kVA Transformers

2017 Study

19

50 kVA Transformers

2017 Study

20

Combine propensity and thermal modeling

• Best/worst case analysis

– On peak charging – Off peak charging

• Detailed studies

– Vehicle type – Charging start time based on arrival time and TOU likelihood – Charging duration based on miles driven – Multiple vehicles per household

Ongoing and Future studies

21

Thank you for your attention Any Questions? Contact Information: Jeff.Loehr@srpnet.com