Minimizing the Integration Costs of Wind Using Curtailment and Electric Vehicle Charging

1

MINIMIZING THE INTEGRATION COSTS OF WIND USING CURTAILMENT AND

ELECTRIC VEHICLE CHARGING

October 11, 2011

Allison WeisAdvisors: Paulina Jaramillo and Jeremy Michalek

Department of Engineering and Public PolicyCarnegie Mellon University

The integration of large amounts of wind power is an increasingly important issue in the United States.

Required by Renewable Portfolio Standards (RPS) 29 States Up to 40% of produced electricity must come from

renewable sources Complicated by the variable and intermittent

nature of wind power

2

Grid flexibility must increase to cope with the fluctuations in wind output

Ramp existing plants Build additional ramping plants, such as

gas turbines Build extra wind plants and allow for

curtailment Variably charge electric vehicles

3

Model Goal

4

Find the optimal combination of new plants, plant operation, and controlled electric vehicle charging in a high wind penetration scenario to minimize systems costs (grid and vehicle)

Current Focus: 20% RPS standard for a given mix of fuel types

Model Components

5

Optimization model with: Capacity Expansion

– what new plants to build, including wind plants

Unit commitment – plant operation

Choosing the number of charging-controlled electric vehicles and how they should be charged

Model includes existing and new power plants, electric vehicles, and non-vehicle load

6Electric Vehicles

Conventional Power Plants

Wind Plants

Non-vehicle Load

Grid Energy Balance

Wind Plants

7

Eastern Wind Integration and Transmission Study dataset EWITS identified sites necessary to meet a

30% RPS in the Eastern Interconnect On-shore and off-shore wind Capacity factors and 10 min. modeled

production data from 2004-2006 Added by capacity factor (high to low) until

plants capable of meeting RPS All remaining EWITS plants can be built if

cost effective

Load data spatially and temporally matched to wind data

8

5 minute load data for NYISO from 2006

Wind and load data averaged to create hourly time series

Continuous 5 day sample used for computational feasibility

Power Plant Fleet

9

Power Plant Fleet Composition:

Plant size and heat rate distributions for conventional plants were matched to NYISO

Total capacity of system

0%

20%

40%

60%

80%

100%Required WindGas Combustion TurbineGas Combined CycleOil/Gas Steam TurbineCoalNuclear

Electric Vehicle Type

10

Characteristic

Value

Battery Size 16 kWh

Charging Power 9.6 kW (Level 2)

Charging Efficiency

88%

Vehicle Premium

$8000*

Vehicles modeled as plug-in hybrid vehicles with the following characteristics:

*Argonne National Lab “Multi-Path Transportation Futures Study : Vehicle Characterization and Scenario Analyses” (2009) , estimate for 2015 PHEV-40

Sample vehicle driving profiles chosen to match aggregate characteristics of all NHTS data.

11

Weighted sample taken from the National Household Travel Survey to match aggregate characteristics: Percent of vehicles of vehicles at home,

work, driving, or elsewhere at every time step

Average number of miles driven in every time step

Average number of cumulative miles driven in every previous time step

Optimize the Mixed-Integer Linear Problem Using the Cplex Solver

12

Objective:

Choice Variables: Number of new wind

and conventional plants to build

Operation of every plant in every time step

Number of electric vehicles

Vehicle charging in every time step

Constraints: Load = Generation Meet RPS Standard Power plant operating

constraints Ramp rate Minimum on and off

times Minimum generation

levels Electric vehicle constraints

Charging rate Battery capacity

premium gas savings

Preliminary Model Output

13

5 Day Sample Schedule

Preliminary Results

14

Very few controlled charging vehicles can help reduce system costs in the current model (0-25)

Build extra wind capacity is reducing system costs Without curtailment: $28 million With curtailment: $20 million

Current Model Limitations

15

Hourly time step Perfect knowledge of wind and load (no

forecasting)

Both reduce the need for grid flexibility

No transmission constraints No emissions costs

Sensitivities to be investigated

16

Vehicle characteristics Charging Scenario

Home only Work and home

Other regions See the effect of different correlations

between wind and load Fleet composition RPS Level

17

Questions and Feedback

Related work has value in incorporating electric vehicle charging

18

Sioshansi and Denholm calculated the value of controlled charging and vehicle-to-grid services with a unit commitment model of ERCOT (Texas)

Wang et. al. calculated the benefit of a set number of electric vehicles with 20% wind power in Illinois with a set number of power plants

Pacific Northwest National Lab calculated the number of electric vehicles necessary to provide

Policy Implications How critical is it to include a grid-to-

vehicle communications protocol in the standards for electric vehicle chargers?

Will the shift of DOE funding to electric vehicle research away from stationary technology still improve grid management?

Consequences of different cost structures under different RPS standards

19

20

EWITS Data Set

21

Model

22

Explanation of Vehicle Profile choosing