1 of 14

T H A L E S

Chargepoint Challenge 2: Using Data To Shape Electrification of Transportation

Grid-Level Adaptive EV Charging

1

Rio Richardson, Rebecca McCabe, Cindy Wu, Cynthia Garde

2 of 14

Current and Future EV Adoption

2

Regulatory�Requirements (increases)

EV fleet size (million)

2010 2030 2050

180

80

0

Low

Medium

High

Uptake scenario:

EV Demand

EV Production & �Variety

EV �Affordability

3 of 14

Grid overload → Unreliability

3

Time

Resource

(Source: https://thehub.agl.com.au/articles/2020/03/explainer-the-duck-curve)

Grid Load

Charging Stations

EV �Demand

4 of 14

The Problem

The Solution

As more EVs & charging stations are built, the grid will struggle to keep up with electricity demand.

Scheduling charging will allow charging stations to optimise which cars are charging at any one point.

While there are methods to do this a site-level, there is currently no tool to schedule charging across multiple sites or at the network-level.

Our algorithm looks at a number of individual & grid-level factors and decides the optimal power level and charge timing for a group of EVs across a range of sites.

4

5 of 14

How our solution works

Our adaptive charging algorithm would be used by a network of chargers to optimise which cars to charge when, allowing automated demand management across a range of sites.

Each driver would see a smaller impact on their charge time than with site-level adaptive charging, while adding up to a big impact on grid demand at any one time.

Car 1: 13.4 kW 13:07-15:02� 6.2 kW 15:03-15:43� ...

Car 2: 8.3 kW 13:19-14:25� 9.9 kW 14:25-15:02� ...

Car 3: 5.2 kW 12:42-13:06� 1.2 kW 13:07-14:11� ...

Car 4: 4.7 kW 11:57-12:32� 1.9 kW 12:33-14:24� ...

Car 5: 2.3 kW 12:27-13:04� 6.2 kW 13:05-13:07� ...

5

Site 1: Public charging station

Sites 2-4: Residential charging station

6 of 14

Our algorithm optimizes charge profiles

Grid power constraints

For each car / site:

  • Initial and desired energy
  • Hours plugged in
  • Maximum power

Maximize clean energy usage

Minimize power peaks

  • Grid level
  • Site level
  • Car level

Power profile in time

  • for each site (grid level optimization)
  • for each car (site / car level optimization)

6

INPUTS

OPTIMISATION

OUTPUT

7 of 14

We created a prototype of the algorithm in MATLAB

7

8 of 14

We used several data sources for our prototype - and would add more given availability/time

8

Data used:

  • Grid power load
  • Grid power available
  • CO2 emissions

Data identified to incorporate:

  • Statistical fleet charging data
  • Home energy usage
  • Energy pricing

9 of 14

Demand management is useful to both utility providers and charge point networks, among others

Utility providers

Charge point networks

Grid balancing is a critical problem for utility companies; many are already providing incentives to EV owners to charge during off-peak hours

Need to target great customer experience while helping clients (businesses, retailers etc.) optimise electricity costs

Other stakeholders

  • Workplaces
  • Logistics companies
  • Local bus operators
  • ISO
  • Grid operators & regulators

9

10 of 14

T H A L E S

Grid-Level Adaptive EV Charging

10

11 of 14

References

11

12 of 14

Appendix 1: Energy Demand Over Time

12

Source: https://thehub.agl.com.au/articles/2020/03/explainer-the-duck-curve

13 of 14

Appendix 2: Grid Modernization & Energy Continuity

13

Source: https://www.windpowerengineering.com/5-ways-ultracapacitors-operate-in-utility-grids-microgrids/

14 of 14

Appendix 3: US Grid

14

Source: https://www.eia.gov/realtime_grid/#/status?end=20201107T14