Large-Scale Adaptive Electric Vehicle Charging Zachary J. Lee , - - PowerPoint PPT Presentation

▶

Jan 21, 2023 236 likes •616 views

Large-Scale Adaptive Electric Vehicle Charging Zachary J. Lee , Daniel Chang, Cheng Jin, George S. Lee, Rand Lee, Ted Lee, Steven H. Low I dont have to convince you EV are coming https://i.ytimg.com/vi/tj6B489H_zg/maxresdefault.jpg We

SLIDE 1

Large-Scale Adaptive Electric Vehicle Charging

Zachary J. Lee, Daniel Chang, Cheng Jin, George S. Lee, Rand Lee, Ted Lee, Steven H. Low

SLIDE 2

I don’t have to convince you EV are coming…

https://i.ytimg.com/vi/tj6B489H_zg/maxresdefault.jpg

SLIDE 3

We assume EV charging will look like this…

http://o.aolcdn.com/hss/storage/midas/f31dd15c97d6237dd816c5d186980528/200403157/DP6V4737.jpg

SLIDE 4

But the future of EV charging in cities looks like this…

SLIDE 5

Capital Costs Prohibitively Expensive

SLIDE 6

The Need for Adaptive Charging

Transformer Line Currents

06:00 10:00 14:00 18:00 22:00 200 400 Current (A) 06:00 10:00 14:00 18:00 22:00

Uncontrolled Charging Adaptive Charging

SLIDE 7

Physical Charging Testbed

SLIDE 8

What good is a real testbed?

Working with real systems allow us to

understand their limitations.

Without a proper understanding of

these limitations our algorithms may look great on paper but be practically useless.

SLIDE 9

The Adaptive Charging Network

Main Switch Panel Garage Loads (Lighting, Fans, Elevators, etc.) EV Switch Panel Caltech Substation 480 V 800 A 3phase Transformer 150 kVA, 480V/208V 208 V 420 A Utility Company 50 kW 400 VDC t1 t0 x19

...

54 controllable level-2 EVSEs
50 kW DC Fast Charger.
Oversubscription of transformers, cables

and breakers.

Demonstration environment for demand

response, pricing schemes, and renewables integration.

SLIDE 10

charging stations

54+

SLIDE 11

kW of Capacity

150

SLIDE 12

MWh of energy delivered

585

SLIDE 13

million mile equivalent

1.8

SLIDE 14

tons of CO#

$% avoided

610

SLIDE 15

What can we do with this system?

SLIDE 16

Data Collection

SLIDE 17

Charging Sessions since April 2018

11,000

SLIDE 18

Charging Session Statistics

Average Number of Sessions Average Length of Sessions Average Total Energy Delivered Average Energy Delivered per Session Maximum Concurrent Sessions

SLIDE 19

Arrival Statistics

SLIDE 20

Session per Day

SLIDE 21

Simultaneous Sessions

SLIDE 22

Online Scheduling

SLIDE 23

Scheduling Problem

max r t < di, i t ≥ di, i t ∈ T , j Uk(r) 0 ≤ ri(t) ≤ ¯ ri(t) ri(t) = 0 di−1 X t=ai ri(t)δ ≤ ei fj(r1(t), ..., rN(t)) ≤ Rj(t)

SCH

No discharging. Maximum charging rate. Relaxation of allowable rate set. Infrastructure constraints. Total energy delivered must be less than energy requested. Maximizing profit. Charging quickly. Maximizing renewable energy use. Following demand response signals. No charging after departure.

SLIDE 24

Unbalanced Three-Phase Constraints

Garage Loads DC Fast Charger Ia DC Ic DC Ib DC Ib aux Ic aux Ia aux Ia0 ≤ R0,a = 640 A Iab P Ica P Ibc P t1 primary t1 secondary EVSEs Iab evse Ica evse Ibc evse Ia1 ≤ R1,a = 180 A Ia3 ≤ R3,a = 420 A a b c Ia1 ≤ R2,a = 180 A

|I3

= | Ievse

− Ievse

| ≤ R3,a

SLIDE 25

Unbalanced Three-Phase Constraints

We assume that we know/can measure the voltage phase angles at the EVSEs.
Since EVSEs can be modeled as constant current loads with unity power factor, we

thus know the phase angles of their currents.

Since the magnitude of the current phasor is the only variable, these constraints are

second-order cone constraints and the optimization problem is tractable. |I3,a|2 = |Ievse

− Ievse

|2 = (|Ievse

| cos φab − |Ievse

| cos φca)2 + (|Ievse

| sin φab − |Ievse

| sin φca)2 ≤ R2

3,a

SLIDE 26

Unbalanced Three-Phase Constraints

We assume that we know/can measure the voltage phase angles at the EVSEs.
Since EVSEs can be modeled as constant current loads with unity power factor, we

thus know the phase angles of their currents.

Since the magnitude of the current phasor is the only variable, these constraints are

second-order cone constraints and the optimization problem is tractable.

|I3,a| = |Ievse

− Ievse

| ≤ |Ievse

| + |Ievse

| ≤ R3,a

SLIDE 27

Phase Aware Constraints

0.0 0.2 0.4 0.6 0.8 1.0 Infrastructure Capacity (% Nominal) 25 50 75 100 125 Price ($) Affine Constraints SOC Constraints Unconstained

Profit ($)

U(r) := X

t∈T i∈V

(p(t) − c(t))ri(t)

SLIDE 28

Imperfect Actuation

Control is done via a pilot signal.
Pilot signal is only an upper bound on

charging current.

Battery management system is free to

charge at any rate lower than the pilot.

17:00 18:00 16:30 16:45 17:15 17:30 17:45 Time 10 20 30 Charging Rate (A) Pilot Actual

SLIDE 29

Model Predictive Control

We use model predictive

control to account for deviations.

Schedule is recomputed

periodically or when changes

ccur in the system.

no recompute? update energy remaining and remaining duration collect active charging sessions update pilot signals from most recent schedule compute new optimal schedule using SCH yes

SLIDE 30

Simple Battery Model

17:00 18:00 16:30 16:45 17:15 17:30 17:45 Time 10 20 30 Charging Rate (A) Pilot Actual Actual Charging Behavior 0.00 0.25 0.50 0.75 1.00 State of Charge 10 20 30 Current (A) Two-Stage Battery Model

SLIDE 31

Robustness

10 20 30 40 50 60 Maximum Recompute Period (minutes) 1.42 1.43 1.44 1.45 1.46 1.47 1.48 U(r) ×107 Robustness to Non-Ideal Charging Behavior Ideal Noiseless σ2 = 1A σ2 = 2A σ2 = 3A

U(r) := X

t∈T

(T − t) X

i∈V

ri(t)

SLIDE 32

Results

SLIDE 33

Profit Maximization

SLIDE 34

Conclusions

We should consider the unique challenges of large-

scale charging infrastructure.

Adaptive scheduling can significantly reduce the

capital and operating costs of large-scale charging systems.

Experience with real systems can inform how we

design practical algorithms.

Real time data from our testbed can be found at

caltech.powerflex.com.

SLIDE 35

Future Work

Demonstrating how large-scale EV charging can be

used to flatten the “duck curve”

Demonstrating the viability of large-scale EV

charging in demand response markets

Analyzing user behavior to design predictive

scheduling algorithms

SLIDE 36

Releasing Dataset and Simulator

email zlee@caltech.edu to be notified of the release

SLIDE 37

Questions