Electric Vehicle Grid Integration: Challenges and Solutions

20
Jianhui Wang Associate Professor Department of Electrical and Computer Engineering [email protected] November 2019 Electric Vehicle Grid Integration: Challenges and Solutions

Transcript of Electric Vehicle Grid Integration: Challenges and Solutions

Jianhui WangAssociate ProfessorDepartment of Electrical and Computer [email protected]

November 2019

Electric Vehicle Grid Integration: Challenges and Solutions

EV Market Grows Rapidly

IEA analysis based on country submissions, complemented by ACEA (2019); EAFO (2019); EV Volumes (2019); Marklines (2019); OICA (2019).

โ€ข In 2018, the global electric car fleet exceeded 5.1 million, up 2 million from the previous year and almost doubling the number of new electric car sales.

โ€ข In 2030, in the New Policies Scenario, global electric car sales will reach 23 million and the stock will exceed 130 million vehicles (excluding two/three-wheelers).

โ€ข In 2030, in the EV30@30 Scenario, EV sales will reach 43 million and the stock will be more than 250 million (excluding two/three-wheelers).

(Source: International Energy Agency, โ€œGlobal EV Outlook 2019 Scaling up the transition to electric mobility,โ€ May. 27, 2019)

Global EV Market 2013-2018

EV Electricity Demand Surges

Note: LDV = light duty vehicles; NPS = New Policies Scenario)

โ€ข In 2018, electricity demand from the global EV fleet is 58 TWh.

โ€ข In 2030, in the New Policies Scenario, electricity demand from the global EV fleet will exceed 640 TWh).

โ€ข In 2030, in the EV30@30 Scenario, electricity demand from the global EV fleet will exceed 1,110 TWh.

โ€ข Our current power distribution networks are not designed for such large scale mobile loads.

(Source: International Energy Agency, โ€œGlobal EV Outlook 2019 Scaling up the transition to electric mobility,โ€ May. 27, 2019)

EV Charging Scenarios

โ€ข AC level two charger and DCFCcharger with high voltage level and high charging power dominates the public charging facility.

โ€ข Concentrated public charging can cause distribution transformer overloading.

โ€ข In 2017, the Sacramento Municipal Utility District (SMUD) forecasted that 17% of its distribution transformers need to be replaced due to EV charging.

Energy demand in % of kWh

Source: McKinsey & Company Automotive & Assembly, โ€œCharging ahead: Electric-vehicle infrastructure demand,โ€ Aug. 2018

EV Charger by Voltage Level

Source: California Energy Commission, โ€œElectrical Vehicle Charging 101,โ€ 2019.

(Source: H. K. Trabish, โ€œElectric vehicles can be grid assets or liabilities. How utilities plan will decide,โ€ Utility Dive, May. 17, 2017.)

Voltage Voltage Voltage110V 1-phase AC 240V 1-phase AC 480V 3-phase ACAmps Amps Amps12 - 16 Amps 16 - 40 Amps <200 Amps (Typ. 60 Amps)Charging Speed Charging Speed Charging Speed3.5 โ€“ 6.5 miles/hour 14 โ€“ 35 miles/hour 178 miles/hour

Challenges and Opportunities

Challenges OpportunitiesUncontrolled charging of large-scaleEVs can affect the safe and economicoperation of power systems in thefollowing aspects: Increased electricity T&D losses

and higher peak load. Increased voltage deviations. Transformers loading. Business models. Increased price volatilities.

If appropriate charging/dischargingstrategies are adopted, EVs cancontribute to power system operationin various ways. Load valley filling. Line congestion management. Demand response. Frequency regulation. Increase renewable penetration. Vehicle-to-grid (V2G).

Harnessing Interdependency for Resilience: Creating an "Energy Sponge" with Cloud Electric Vehicle Sharing

โ€ข Objective: to model, test, and validate a transformative โ€œenergy spongeโ€ service that transforms the interdependency between power and transportation systems into extra resilience through sharing electric vehicles (EV).

โ€ข Funding Source: National Science Foundation

Figure 1 Cloud EV Sharing Service

โ€ข Collaborators: University of Wisconsin-Madison, University of South Florida, Arizona State University

โ€ข Broader Impact: The outcomes of this research will benefit infrastructure system planning and operation practices, particularly in the context of developing green transportation, smart grid, and green/smart cities.

Charging Management for EV through Tariff Incentive

EV Load Profiles by ScenarioUnconstrained charging: charge as soon as EVs arrivehome.Constrained charging: charging is delayed for 3 hoursafter vehicles arrive home.Smart charging: charge only during the off-peak time.Observations: EV charging is generally more flexible and easier to

manage than shifting multiple residential loads. Varied pricing programs can be designed by utilities

to better manage EV charging-induced spikes inresidential locations.

Smart charging strategies can significantly delay theneeds for new generation capacity.

Ref.:

M. Biviji, C. Uรงkun, G. Bassett, J. Wang and D. Ton, "Patterns ofelectric vehicle charging with time of use rates: Case studies inCalifornia and Portland," ISGT 2014, Washington, DC, 2014, pp.1-5.L. Poch, M. Mahalik, J. Wang and A. Vyas, "Impacts of plug-in hybridelectric vehicles on the electric power system in the western UnitedStates," IEEE PES General Meeting, Providence, RI, 2010, pp. 1-7.

Improve the Wind Penetration through EV and Demand Response

Wind power contains strong variability and uncertainty

Distribution of vehicles by last-trip ending timeWind power is inherently uncertain and variable. High penetration of wind power in the power system threats the security operation of the grid.However: EVs can shift the charging demand from

critical hours to non-critical hours by changing the (1) charging time; (2) charging amount; (3) charging power.

EVs and DR can quickly adjust the end-use consumption.

Batteries in EVs can contribute to load leveling by injecting power to the grid during on-peak hours.

Batteries in EVs can absorb wind power during off-peak times.

Improve the Wind Penetration through EV and Demand Response

Four scenarios investigated in the case studies:1. Unconstrained charging of EVs. 2. 3-hour delayed-constrained

charging of EVs. 3. Smart charging of EVs. 4. DR and smart charging of EVs.

DR programs are assumed to be in place for managing non-EV load. EV hourly charging power under the four scenarios.

Ref.: J. Wang, C. Liu, D. Ton, Y. Zhou, J. Kim, A. Vyas, โ€œImpact of plug-in hybrid electric vehicles on power systems with demand response and wind power,โ€ Energy Policy, vol. 39, no. 7, Jan. 2011, pp. 4016-4021.

Improve the Wind Penetration through EV and Demand Response

EV hourly charging power under 4 scenarios.

Case Study: Running unit commitment program simulations on the four scenarios. The running results for one of the studied weeks are presented.

Result:The cooperation of DR and EV significantly reduces the use of peaking units and achieves the lowest operating costs to the generation units.

Hourly total thermal generation

Start-up cost for each day in scenarios

Charging Load Adjustment through Optimal Pricing

Idea: adjust prices of electricity at public charging stations to influence the spatial distribution of EV charging loads tomitigate their impacts to the grid.

Objective one: minimize the EVsโ€™ total traveling time.

min๐‘ฅ๐‘ฅ,, ๏ฟฝ๐‘ฃ๐‘ฃ,๐‘ž๐‘ž

๏ฟฝ๐‘Ž๐‘Ž๏ฟฝ0

๐‘ฃ๐‘ฃ๐‘Ž๐‘Ž๐‘ก๐‘ก๐‘Ž๐‘Ž ๏ฟฝ๐‘ค๐‘ค ๐‘‘๐‘‘ ๏ฟฝ๐‘ค๐‘ค +

1๐›ผ๐›ผ๏ฟฝ๐‘Ÿ๐‘Ÿ

๏ฟฝ๐‘ ๐‘ 

๏ฟฝ๐‘ž๐‘ž๐‘Ÿ๐‘Ÿ๐‘ ๐‘  ln ๏ฟฝ๐‘ž๐‘ž๐‘Ÿ๐‘Ÿ๐‘ ๐‘  โˆ’ 1 +1๐›ผ๐›ผ๏ฟฝ๐‘Ÿ๐‘Ÿ

๏ฟฝ๐‘ ๐‘ 

(๐›พ๐›พ๐‘ฆ๐‘ฆ๐‘ ๐‘ ๐‘๐‘๐‘ ๐‘ ๐‘’๐‘’๐‘Ÿ๐‘Ÿ๐‘ ๐‘  โˆ’ ๐›ฝ๐›ฝ๐‘ฆ๐‘ฆ๐‘ ๐‘  โˆ’ ๐œƒ๐œƒ๐‘ ๐‘ )๏ฟฝ๐‘ž๐‘ž๐‘Ÿ๐‘Ÿ๐‘ ๐‘ 

s.t.

๐ธ๐ธ๐ธ๐ธ ๐‘ก๐‘ก๐‘ก๐‘ก๐‘ก๐‘ก๐‘ก๐‘ก๐‘ก๐‘ก๐‘ก๐‘ก๐‘ก๐‘ก ๐‘ก๐‘ก๐‘“๐‘“๐‘“๐‘“๐‘ค๐‘ค ๐‘ก๐‘ก๐‘“๐‘“๐‘๐‘๐‘๐‘๐‘ก๐‘ก๐‘ก๐‘ก๐‘ก๐‘ก๐‘ก๐‘ก๐‘๐‘๐‘ก๐‘ก๐‘๐‘

๐‘ฃ๐‘ฃ๐‘Ž๐‘Ž = ๏ฟฝ๐‘Ÿ๐‘Ÿ๐‘ ๐‘ ๏ฟฝ๐‘ฅ๐‘ฅ๐‘Ž๐‘Ž๐‘Ÿ๐‘Ÿ๐‘ ๐‘  + ๏ฟฝ

๐‘Ÿ๐‘Ÿ๐‘ ๐‘ ๏ฟฝฬ…๏ฟฝ๐‘ฅ๐‘Ž๐‘Ž๐‘Ÿ๐‘Ÿ๐‘ ๐‘  ,โˆ€๐‘ก๐‘ก โˆˆ ๐ด๐ด

โˆ†๏ฟฝ๐‘ฅ๐‘ฅ๐‘Ž๐‘Ž๐‘Ÿ๐‘Ÿ๐‘ ๐‘  = ๐ธ๐ธ๐‘Ÿ๐‘Ÿ๐‘ ๐‘  ๏ฟฝ๐‘ž๐‘ž๐‘Ÿ๐‘Ÿ๐‘ ๐‘ โˆ†๏ฟฝฬ…๏ฟฝ๐‘ฅ๐‘Ž๐‘Ž๐‘Ÿ๐‘Ÿ๐‘ ๐‘  = ๐ธ๐ธ๐‘Ÿ๐‘Ÿ๐‘ ๐‘  ๏ฟฝ๐‘ž๐‘ž๐‘Ÿ๐‘Ÿ๐‘ ๐‘ 

โ‹ฎ

๐‘ƒ๐‘ƒ๐‘“๐‘“๐‘ค๐‘ค๐‘’๐‘’๐‘ก๐‘ก ๐‘ก๐‘ก๐‘“๐‘“๐‘“๐‘“๐‘ค๐‘ค ๐‘ก๐‘ก๐‘“๐‘“๐‘๐‘๐‘๐‘๐‘ก๐‘ก๐‘ก๐‘ก๐‘ก๐‘ก๐‘ก๐‘ก๐‘๐‘๐‘ก๐‘ก๐‘๐‘

๐‘ƒ๐‘ƒ๐‘˜๐‘˜,๐‘ง๐‘ง = ๐‘ƒ๐‘ƒ๐‘š๐‘š,๐‘ง๐‘ง โˆ’๐‘ข๐‘ข๐‘š๐‘š,๐‘˜๐‘˜ ๐‘ƒ๐‘ƒ๐‘š๐‘š,๐‘ง๐‘ง

2 + ๐‘„๐‘„๐‘š๐‘š,๐‘ง๐‘ง2

๐ธ๐ธ๐‘š๐‘š2โˆ’๏ฟฝ

๐‘˜๐‘˜,๐‘ง๐‘งโ€ฒ โˆˆ๐พ๐พ๐พ๐พ,๐‘ง๐‘งโ€ฒโ‰ ๐‘ง๐‘ง๐‘ƒ๐‘ƒ๐‘˜๐‘˜,๐‘ง๐‘งโ€ฒ โˆ’ ๏ฟฝ๐‘ƒ๐‘ƒ๐‘˜๐‘˜ โˆ’ ๏ฟฝ๐‘ƒ๐‘ƒ๐‘˜๐‘˜ ,โˆ€ ๐‘š๐‘š, ๐‘˜๐‘˜, ๐‘ง๐‘ง โˆˆ ๐ต๐ต๐ต๐ต

๐‘„๐‘„๐‘˜๐‘˜,๐‘ง๐‘ง = ๐‘„๐‘„๐‘š๐‘š,๐‘ง๐‘ง โˆ’๐‘ค๐‘ค๐‘š๐‘š,๐‘˜๐‘˜ ๐‘ƒ๐‘ƒ๐‘š๐‘š,๐‘ง๐‘ง

2 + ๐‘„๐‘„๐‘š๐‘š,๐‘ง๐‘ง2

๐ธ๐ธ๐‘š๐‘š2โˆ’๏ฟฝ

๐‘˜๐‘˜,๐‘ง๐‘งโ€ฒ โˆˆ๐พ๐พ๐พ๐พ,๐‘ง๐‘งโ€ฒโ‰ ๐‘ง๐‘ง๐‘„๐‘„๐‘˜๐‘˜,๐‘ง๐‘งโ€ฒ โˆ’ ๏ฟฝ๐‘„๐‘„๐‘˜๐‘˜ โˆ’ ๏ฟฝ๐‘„๐‘„๐‘˜๐‘˜ ,โˆ€(๐‘š๐‘š, ๐‘˜๐‘˜, ๐‘ง๐‘ง) โˆˆ ๐ต๐ต๐ต๐ต

๐ธ๐ธ๐‘˜๐‘˜2 = ๐ธ๐ธ๐‘š๐‘š2 โˆ’ 2 ๐‘ข๐‘ข๐‘š๐‘š,๐‘˜๐‘˜๐‘ƒ๐‘ƒ๐‘š๐‘š,๐‘ง๐‘ง + ๐‘ค๐‘ค๐‘š๐‘š,๐‘˜๐‘˜๐‘„๐‘„๐‘š๐‘š,๐‘ง๐‘ง +๐‘ข๐‘ข๐‘š๐‘š,๐‘˜๐‘˜2 + ๐‘ค๐‘ค๐‘š๐‘š,๐‘˜๐‘˜

2 ๐‘ƒ๐‘ƒ๐‘š๐‘š,๐‘ง๐‘ง2 + ๐‘„๐‘„๐‘š๐‘š,๐‘ง๐‘ง

2

๐ธ๐ธ๐‘š๐‘š2, โˆ€(๐‘š๐‘š, ๐‘˜๐‘˜, ๐‘ง๐‘ง) โˆˆ ๐ต๐ต๐ต๐ต

โ‹ฎ

Traveling time of EVs to their destinations

link flow distributions of PEVs

link flow distributions of regular vehicles

Charging Load Adjustment through Optimal Pricing

Objective two: minimize the EV ownersโ€™ payment for the charging services.

min๐‘ฅ๐‘ฅ,๐‘ฃ๐‘ฃ, ๏ฟฝ๐‘ž๐‘ž,๐‘๐‘,๐‘ƒ๐‘ƒ,๐‘„๐‘„,๐‘‰๐‘‰,๐œŒ๐œŒ,๐œ๐œ,๐œ†๐œ†

๐‘๐‘๐‘š๐‘š ๏ฟฝ ๏ฟฝ0,๐‘ง๐‘ง โˆˆ๐พ๐พ๐พ๐พ

๐‘ƒ๐‘ƒ0,๐‘ง๐‘ง โˆ’ ๏ฟฝ๐‘˜๐‘˜โˆˆ๐พ๐พ

๏ฟฝ๐‘ƒ๐‘ƒ๐‘˜๐‘˜ + ๏ฟฝ๐‘ƒ๐‘ƒ๐‘˜๐‘˜ + ๐œ‹๐œ‹ ๏ฟฝ ๏ฟฝ๐‘Ž๐‘Žโˆˆ๐ด๐ด

๐‘ก๐‘ก๐‘Ž๐‘Ž(๐‘ฃ๐‘ฃ๐‘Ž๐‘Ž)๐‘ฃ๐‘ฃ๐‘Ž๐‘Ž

s.t.๐ธ๐ธ๐ธ๐ธ ๐‘ก๐‘ก๐‘ก๐‘ก๐‘ก๐‘ก๐‘ก๐‘ก๐‘ก๐‘ก๐‘ก๐‘ก๐‘ก๐‘ก ๐‘ก๐‘ก๐‘“๐‘“๐‘“๐‘“๐‘ค๐‘ค ๐‘ก๐‘ก๐‘“๐‘“๐‘๐‘๐‘๐‘๐‘ก๐‘ก๐‘ก๐‘ก๐‘ก๐‘ก๐‘ก๐‘ก๐‘๐‘๐‘ก๐‘ก๐‘๐‘๐‘ƒ๐‘ƒ๐‘“๐‘“๐‘ค๐‘ค๐‘’๐‘’๐‘ก๐‘ก ๐‘ก๐‘ก๐‘“๐‘“๐‘“๐‘“๐‘ค๐‘ค ๐‘ก๐‘ก๐‘“๐‘“๐‘๐‘๐‘๐‘๐‘ก๐‘ก๐‘ก๐‘ก๐‘ก๐‘ก๐‘ก๐‘ก๐‘๐‘๐‘ก๐‘ก๐‘๐‘

๐‘ƒ๐‘ƒ๐‘“๐‘“๐‘ค๐‘ค๐‘’๐‘’๐‘ก๐‘ก ๐‘š๐‘š๐‘ก๐‘ก๐‘ก๐‘ก๐‘˜๐‘˜๐‘’๐‘’๐‘ก๐‘ก ๐‘ก๐‘ก๐‘๐‘๐‘‘๐‘‘ ๐ธ๐ธ๐ธ๐ธ ๐‘ก๐‘ก๐‘“๐‘“๐‘“๐‘“๐‘“๐‘“ ๐‘ก๐‘ก๐‘“๐‘“๐‘๐‘๐‘ก๐‘ก ๐‘ก๐‘ก๐‘“๐‘“๐‘๐‘๐‘๐‘๐‘ก๐‘ก๐‘ก๐‘ก๐‘ก๐‘ก๐‘ก๐‘ก๐‘๐‘๐‘ก๐‘ก๐‘๐‘ ๏ฟฝ๐‘ก๐‘ก๐‘Ž๐‘Ž โˆ’ ๏ฟฝ๐œŒ๐œŒ๐‘–๐‘–๐‘Ÿ๐‘Ÿ๐‘ ๐‘  + ๏ฟฝ๐œŒ๐œŒ๐‘—๐‘—๐‘Ÿ๐‘Ÿ๐‘ ๐‘  ๏ฟฝ ๏ฟฝ๐‘ฅ๐‘ฅ๐‘Ž๐‘Ž๐‘Ÿ๐‘Ÿ๐‘ ๐‘  = 0,โˆ€๐‘ก๐‘ก = ๐‘ก๐‘ก, ๐‘—๐‘— โˆˆ ๐ด๐ด, ๐‘ก๐‘ก โˆˆ ๐‘…๐‘…, ๐‘๐‘ โˆˆ ๐‘†๐‘†๐‘ก๐‘ก๐‘Ž๐‘Ž โˆ’ ๏ฟฝ๐œŒ๐œŒ๐‘–๐‘–๐‘Ÿ๐‘Ÿ๐‘ ๐‘  + ๏ฟฝ๐œŒ๐œŒ๐‘—๐‘—๐‘Ÿ๐‘Ÿ๐‘ ๐‘  โ‰ฅ 0,โˆ€๐‘ก๐‘ก = ๐‘ก๐‘ก, ๐‘—๐‘— โˆˆ ๐ด๐ด, ๐‘ก๐‘ก โˆˆ ๐‘…๐‘…, ๐‘๐‘ โˆˆ ๐‘†๐‘†

โ‹ฎ

Reformulate the objective one and two as an equivalent bi-level programming model:

๐‘ˆ๐‘ˆ๐‘๐‘๐‘๐‘๐‘’๐‘’๐‘ก๐‘ก ๐‘“๐‘“๐‘’๐‘’๐‘ฃ๐‘ฃ๐‘’๐‘’๐‘“๐‘“ โ†’ min(๐‘ฃ๐‘ฃ, ๏ฟฝ๐‘ž๐‘ž,๐‘๐‘,๐บ๐บ)

๐‘๐‘๐‘š๐‘š ๏ฟฝ โˆ‘ 0,๐‘ง๐‘ง โˆˆ๐พ๐พ๐พ๐พ ๐‘ƒ๐‘ƒ0,๐‘ง๐‘ง โˆ’ โˆ‘๐‘˜๐‘˜โˆˆ๐พ๐พ ๏ฟฝ๐‘ƒ๐‘ƒ๐‘˜๐‘˜ + ๏ฟฝ๐‘ƒ๐‘ƒ๐‘˜๐‘˜ + ๐œ‹๐œ‹ ๏ฟฝ โˆ‘๐‘Ž๐‘Žโˆˆ๐ด๐ด ๐‘ก๐‘ก๐‘Ž๐‘Ž ๐‘ฃ๐‘ฃ๐‘Ž๐‘Ž ๐‘ฃ๐‘ฃ๐‘Ž๐‘Ž + ฮ“ ๏ฟฝ max๏ฟฝ0,โˆ‘ 0,๐‘ง๐‘ง โˆˆ๐พ๐พ๐พ๐พ ๐‘ƒ๐‘ƒ0,๐‘ง๐‘ง ๏ฟฝ ๐‘๐‘๐‘š๐‘š โˆ’

Charging energy costs

Vehicle traveling toll costs

Charging Load Adjustment through Optimal Pricing

Solution framework

Result๏ผšThe travel cost is reduced by 2.6%The real power losses are reduced by 4.5%

Conclusion:Electricity pricing can be an effective tool for addressing the challenges that EV charging loads impose on power distribution grids, and integrated pricing of electricity and roads has potential for better managing and operating thecoupled transportation and power networks.

Ref.: F. He, Y. Yin, J. Wang, Y. Yang, โ€œSustainability SI: optimal Prices of electricity at public charging stations for plug-in electric vehicles,โ€ Networks and Spatial Economics, vol. 16, no. 1, pp. 131-154.

Reduce Energy Loss through a Prediction-based Power DispatchChallenges: The mobile EV charging load causes extrapower loss due to imperfect power dispatch planning.

Idea: we propose a model predictive control (MPC)-basedpower dispatch approach. The proposed objectivefunctions minimize the operational cost whileaccommodating the EV charging uncertainty.

๐‘€๐‘€๐‘ก๐‘ก๐‘๐‘:๐น๐น =โˆ‘๐‘ก๐‘ก โˆ‘๐‘—๐‘— ๐‘ก๐‘ก๐‘—๐‘— ๐‘ƒ๐‘ƒ๐‘—๐‘— ๐‘ก๐‘ก +โˆ‘๐‘ก๐‘ก โˆ‘๐‘˜๐‘˜ ๐‘ก๐‘ก๐‘˜๐‘˜ ๐‘ƒ๐‘ƒ๐‘˜๐‘˜ ๐‘ก๐‘ก +โˆ‘๐‘ก๐‘ก ๐‘ก๐‘ก๐‘”๐‘”๐‘Ÿ๐‘Ÿ๐‘–๐‘–๐‘”๐‘”(๐‘ƒ๐‘ƒ๐‘”๐‘”๐‘Ÿ๐‘Ÿ๐‘–๐‘–๐‘”๐‘”(๐‘ก๐‘ก)) +โˆ‘๐‘ก๐‘ก โˆ‘๐‘—๐‘— ๐‘ก๐‘ก๐‘—๐‘—๐‘ˆ๐‘ˆ ๐‘ก๐‘ก

Subject to:โˆ‘๐‘š๐‘š๐‘ƒ๐‘ƒ๐‘š๐‘š ๐‘ก๐‘ก + โˆ‘๐‘›๐‘› ๐‘ƒ๐‘ƒ๐‘›๐‘› ๐‘ก๐‘ก + โˆ‘๐‘š๐‘š๐‘ƒ๐‘ƒ๐‘š๐‘š ๐‘ก๐‘ก + โˆ‘๐‘š๐‘š๐‘ƒ๐‘ƒ๐‘š๐‘š ๐‘ก๐‘ก +๐‘ƒ๐‘ƒ๐‘”๐‘”๐‘Ÿ๐‘Ÿ๐‘–๐‘–๐‘”๐‘” ๐‘ก๐‘ก = ๐‘ƒ๐‘ƒ๐ฟ๐ฟ๐ฟ๐ฟ๐‘ ๐‘ ๐‘ ๐‘  ๐‘ก๐‘ก + ๐ท๐ท๐‘๐‘๐‘Ž๐‘Ž๐‘ ๐‘ ๐‘๐‘ ๐‘ก๐‘ก + ๐ท๐ท๐ธ๐ธ๐ธ๐ธ ๐‘ก๐‘ก ,โˆ€๐‘ก๐‘ก

๐‘ƒ๐‘ƒ๐‘—๐‘—,๐‘š๐‘š๐‘–๐‘–๐‘›๐‘› โ‰ค ๐‘ƒ๐‘ƒ๐‘—๐‘— ๐‘ก๐‘ก โ‰ค ๐‘ƒ๐‘ƒ๐‘—๐‘—,๐‘š๐‘š๐‘Ž๐‘Ž๐‘ฅ๐‘ฅ ,โˆ€๐‘—๐‘—

๐‘ƒ๐‘ƒ๐‘˜๐‘˜,๐‘š๐‘š๐‘–๐‘–๐‘›๐‘› โ‰ค ๐‘ƒ๐‘ƒ๐‘˜๐‘˜ ๐‘ก๐‘ก โ‰ค ๐‘ƒ๐‘ƒ๐‘˜๐‘˜,๐‘š๐‘š๐‘Ž๐‘Ž๐‘ฅ๐‘ฅ ,โˆ€๐‘˜๐‘˜

0 โ‰ค ๐‘ƒ๐‘ƒ๐พ๐พ ๐‘ก๐‘ก โ‰ค ๐‘ƒ๐‘ƒ๐‘™๐‘™๐‘๐‘๐‘ฃ๐‘ฃ๐‘๐‘๐‘™๐‘™ ,โˆ€๐‘ก๐‘ก๐ธ๐ธ๐‘š๐‘š๐‘–๐‘–๐‘›๐‘› โ‰ค ๐ธ๐ธ๐‘ฆ๐‘ฆ โ‰ค ๐ธ๐ธ๐‘š๐‘š๐‘Ž๐‘Ž๐‘ฅ๐‘ฅ ,โˆ€๐‘ฆ๐‘ฆ

Procedure of the proposed MPC-based power dispatch method

Ref.: W. Su, J. Wang, K. Zhang, A. Q. Huang, โ€œModel predictive control-based power dispatch for distribution system considering plug-in electric vehicle uncertainty,โ€ Electric Power Systems Research, Vol. 106, Jan. 2014, pp 29-35.

DG generation costs

Energy storage discharging costs

Costs for energy from the grid

DG start up/shut down costs

Power balance

Generation limits

Power transfer limits

Voltage deviation limits

Reduce Energy Loss through a Prediction-based Power Dispatch

System Setup: A modified IEEE 37-node distribution test feeder. DG, wind generator, solar generator, DES, and PEV are connected with nodes 701, 722, 730, and 737, respectively.

Case Study:Scenario 1: Non-MPC based Day-aheadScenario 2: MPCScenario 3: Perfect Forecasting

Compared with the non-MPC based day-ahead approach the MPC method reduces the charging costs by 13.44% and power loss by 15.78%.

Risk Management for EV Aggregators

Challenge: the information gap between the forecasted and actual electricity prices posts revenue risks for the EVaggregators.

Strategy: manage the revenue risk of the EV aggregator through an information gap decision theory (IGDT)-based approach.

Goal: guarantee the predefined profit for risk-averse decision-makers or pursue the windfall return for risk-seeking decision-makers.

Framework of EV aggregator participating in electricity markets

Managing the Profit Risk of the EV Aggregator Using Information Gap Decision Theory

Objective: maximize the total profit of EV aggregators

max๐น๐น ๐ธ๐ธ๐‘ƒ๐‘ƒ๐ธ๐ธ๐‘‰๐‘‰,๐‘ƒ๐‘ƒ๐ถ๐ถ๐ถ๐ถ ,๐‘ƒ๐‘ƒ๐ท๐ท๐ถ๐ถ๐ถ๐ถ = โˆ‘๐‘ก๐‘ก=1๐‘‡๐‘‡ (๐œ†๐œ†๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐‘ƒ๐‘ƒ๐ธ๐ธ๐‘‰๐‘‰,๐‘ก๐‘ก โˆ’ ๐›พ๐›พ๐‘ก๐‘ก๐‘ƒ๐‘ƒ๐ธ๐ธ๐‘‰๐‘‰,๐‘ก๐‘ก โˆ’ ๐œ†๐œ†๐ต๐ต๐ท๐ท๐‘ƒ๐‘ƒ๐ท๐ท๐ถ๐ถ๐ถ๐ถ,๐‘ก๐‘ก)

Payments from EV owners Energy costs battery degradation costs

๐‘๐‘. ๐‘ก๐‘ก.

๐ธ๐ธ๐‘ก๐‘ก = ๐ธ๐ธ๐‘ก๐‘กโˆ’1 + ๐ธ๐ธ๐‘Ž๐‘Ž๐‘Ÿ๐‘Ÿ๐‘Ÿ๐‘Ÿ,๐‘ก๐‘ก โˆ’ ๐ธ๐ธ๐‘”๐‘”๐‘๐‘๐‘๐‘,๐‘ก๐‘ก + ๐œ‚๐œ‚๐ถ๐ถ๐ถ๐ถ๐‘ƒ๐‘ƒ๐ถ๐ถ๐ถ๐ถ,๐‘ก๐‘ก โˆ’๐‘ƒ๐‘ƒ๐ท๐ท๐ท๐ท๐ท๐ท,๐‘ก๐‘ก๐œ‚๐œ‚๐ท๐ท๐ท๐ท๐ท๐ท

,โˆ€๐‘ก๐‘ก

๐ธ๐ธ0 = ๐ธ๐ธ๐‘‡๐‘‡๐‘ƒ๐‘ƒ๐ธ๐ธ๐‘‰๐‘‰,๐‘ก๐‘ก = ๐‘ƒ๐‘ƒ๐ถ๐ถ๐ถ๐ถ,๐‘ก๐‘ก โˆ’ ๐‘ƒ๐‘ƒ๐ท๐ท๐ถ๐ถ๐ถ๐ถ,๐‘ก๐‘ก ,โˆ€๐‘ก๐‘ก

โ‹ฎ

Uncertainty Model for the electricity price:ฮ“ ๐›ผ๐›ผ, ๏ฟฝฬ…๏ฟฝ๐›พ = ๐›พ๐›พ: (๐›พ๐›พ โˆ’ ๏ฟฝฬ…๏ฟฝ๐›พ)๐‘‡๐‘‡ ๐ถ๐ถโˆ’1(๐›พ๐›พ โˆ’ ๏ฟฝฬ…๏ฟฝ๐›พ) โ‰ค ๐›ผ๐›ผ2 ๏ผŒ๐›ผ๐›ผ โ‰ฅ 0

Robustness function:๏ฟฝ๐›ผ๐›ผ ๐ต๐ต,๐‘ก๐‘ก๐‘Ÿ๐‘Ÿ = max

๐พ๐พ,๐›พ๐›พ๐›ผ๐›ผ: min

๐›พ๐›พโˆˆ๐‘ˆ๐‘ˆ(๐›ผ๐›ผ,๏ฟฝ๐›พ๐›พ)๐น๐น(๐ต๐ต, ๐›พ๐›พ) โ‰ฅ ๐‘ก๐‘ก๐‘Ÿ๐‘Ÿ

Opportunity function:

๏ฟฝฬ‚๏ฟฝ๐›ฝ ๐ต๐ต, ๐‘ก๐‘ก๐ฟ๐ฟ = min๐พ๐พ,๐›พ๐›พ

๐›ผ๐›ผ: max๐›พ๐›พโˆˆ๐‘ˆ๐‘ˆ(๐›ผ๐›ผ,๏ฟฝ๐›พ๐›พ)

๐น๐น(๐ต๐ต, ๐›พ๐›พ) โ‰ฅ ๐‘ก๐‘ก๐ฟ๐ฟ

Ref.: J. Zhao, C. Wan, Z. Xu and J. Wang, "Risk-Based Day-Ahead Scheduling of Electric Vehicle Aggregator Using Information Gap Decision Theory," IEEE Transactions on Smart Grid, vol. 8, no. 4, pp. 1609-1618, July 2017.

Managing the Profit Risk of the EV Aggregator

Ref.: J. Zhao, C. Wan, Z. Xu and J. Wang, "Risk-Based Day-Ahead Scheduling of Electric Vehicle Aggregator Using Information Gap Decision Theory," IEEE Transactions on Smart Grid, vol. 8, no. 4, pp. 1609-1618, July 2017.

Case studyThe numerical solution to the optimization problem: max๐น๐น ๐ธ๐ธ๐‘ƒ๐‘ƒ๐ธ๐ธ๐‘‰๐‘‰ ,๐‘ƒ๐‘ƒ๐ถ๐ถ๐ถ๐ถ,๐‘ƒ๐‘ƒ๐ท๐ท๐ถ๐ถ๐ถ๐ถ is a neutral risk profit as shown as point O.

A risk-averse decision maker can choose a robust profit target that is below the max๐น๐น ๐ธ๐ธ๐‘ƒ๐‘ƒ๐ธ๐ธ๐‘‰๐‘‰ ,๐‘ƒ๐‘ƒ๐ถ๐ถ๐ถ๐ถ ,๐‘ƒ๐‘ƒ๐ท๐ท๐ถ๐ถ๐ถ๐ถ . A decision maker with positive attitudes towards risk can choose a opportunity target profit.

ConclusionThis IGDT-based scheduling methodology allows EV aggregators to consider different risk attitudes towards the uncertainty of electricity market prices. The proposed IGDT-based model provides an effective way for EV aggregators to pursue a predefined profit target.

Robustness price curve for profit target ๐‘ก๐‘ก๐‘Ÿ๐‘Ÿ = $3900 and opportunityprice curve for profit target ๐‘ก๐‘ก๐ฟ๐ฟ= $8100 compared with nominal price.

Electric vehicle charging and discharging power by robust scheduling strategy for profit target ๐‘ก๐‘ก๐‘Ÿ๐‘Ÿ = $2700 compared with neutral risk scheduling.

Electric vehicle charging and discharging power by opportunisticscheduling strategy for profit target ๐‘ก๐‘ก๐ฟ๐ฟ = $9900 compared with neutral risk scheduling.

Robustness price curve for profit target ๐‘ก๐‘ก๐‘Ÿ๐‘Ÿ = $3900 and opportunityprice curve for profit target ๐‘ก๐‘ก๐ฟ๐ฟ= $8100 compared with nominal price.

Neutral risk

- The development of EVs and their challenges to the grid โ€ข EV Technology and Marketsโ€ข EV Charging Demand and its Impacts to the Grid

- Tackling the EVโ€™s Challenges to the Gridโ€ข Charging Management for EVs through Tariff Incentiveโ€ข Charging Load Adjustment through Optimal Pricingโ€ข Reducing Energy Loss through a Prediction-based Power Dispatchโ€ข Managing the Profit Risk of an EV Aggregator

- The contribution of EVs to the Gridโ€ข Improve the Wind Penetration through EV and Demand Responseโ€ข Valley Filling through EV Fleetsโ€ข Fast Frequency Regulation through Battery Swapping Stations

Conclusions