Performance Optimization of a PHEV/PEV Enabled Municipal Parking Deck in a Smart Grid Environment Wencong Su, and Dr. Mo-Yuen Chow Department of Electrical and Computer Engineering, North Carolina State University [email protected] [email protected]

Objectives:  To optimally allocate power as well as communication resources to a large number of PHEVs/PEVs to maximize customer satisfaction and minimize disturbances to power grids  To develop a digital testbed to facilitate a smooth integration between plug-in electric vehicles and power grids

Multi-objective Optimization Real-time Large Scale Optimization Power Plant

Solar PV

Smart Charging

http://www.freedm.ncsu.edu/ http://www.adac.ncsu.edu/projects/Roadmap/Project_Home.html

Challenges:

Wind Farm

PHEV Battery Model

 To achieve multi-objective optimal solutions in real-time for allocation of power and communication resources at a large-scale PHEV/PEV municipal parking deck  To enable low-cost and effective communication among vehicles, charging stations and energy management systems

Real-time Monitor Energy Storage

Fig 1. Envisioned Large-scale PHEV/PEV Charging Infrastructure in a Smart Grid Environment

Activity & Accomplishments:  Simulated the real-world transportation scenarios and the aggregate load demand at a large-scale PHEV/PEV enabled parking deck  Evaluated the impact of the integration of PHEVs/PEVs on power grid under a variety of charging scenarios (i.e., uncontrolled charging, normal controlled charging, and TOU-based charging)  Considered the real-world constraints Battery Charging Limit

SoC Requirement 0  SoCi (k )  SoCi ,max .

0  Pi (k )  Pi ,max  k  .

Utility Limit  Pi  k   Putility  k  . i

Method

Fitness Value

Computation Time

PSO

28.01

14.2 sec

EDA

Auction Theory

GA

IPM

34.30

34.57

28.07

34.26

2.98 sec

0.3 sec

16.7 sec

0.9 sec

Pros Fewer parameters Easy to handle constraints Good for multi-objective problem Good for complex system Little dimension limit Avoid premature convergence Good for multi-objective problem Fast Easy to Implement Simple Concept Built-in Matlab Toolbox Useful for loosely defined problems No need to compute derivatives Good for multi-objective problem Fast Relatively easy to implement Good for large-scale problems

Ramp Rate Constraint

Fig 2. Large-scale PHEV/PEV Charging Infrastructure Digital Testbed (Energy Management Module)

0  SoCi (k  1)  SoCi (k )  SoCmax .

Cons Relatively low quality solution Time-consuming Relatively slow convergence rate Moderate computation cost Moderate local search ability Need statistical background Hard to handle constraints Not good for the complex objective function Need to compute derivatives Progressive slower improvement Need many parameters to adjust Need mutation and crossover Computationally expensive Only search for local minima May fail to find global optima Need to compute derivations

 Developed computational intelligence based algorithms to achieve the optimal power allocation  Achieved multi-objective energy scheduling Minimize the charging cost Minimize the peak demand n

T

Min C j  Pi , j

Min[max( Pi , j )] j

i 1 j 1

i 1

Maximize the customer preference n T Min | ( SOCi ,desired  SOCi )  Ei   ( Pi , j  t ) | i 1

j Start/Stop Charging

GUI for Customer/Driver

iSpace

Parking Occupancy

Table 1. Comparisons on Computational Intelligence-based Optimization Algorithms

 Performed the sensitivity analysis on various PHEV/PEV battery models (e.g., a PHEV battery model considering relaxation and hysteresis effects)  Demonstrated a two-way communication network among plug-in vehicles, PHEV charging stations, and an intelligent energy management system (iEMS) using TCP/IP and ZigBee network

n

Arrival/Departure V, I, Temp

GUI for Parking Deck Operator

GUI for Charging Station

Power, Price

Optimization Algorithm Battery Model Statistical Analysis

V, I, Temp

Robot/PEV 3

Matlab/Simulink-based Optimization Module

Fig 3. Large-scale PHEV/PEV Charging Infrastructure Digital Testbed (Communication Module)

Next Steps:  Apply the control strategies (e.g., Distributed Control)  Performance evaluation with communication delay, packet drop, signal strength, and bandwidth constraints  Achieve the optimal allocation of communication resources between vehicles, chargers, and aggregator/utility  Develop large-scale Vehicle-to-Grid (V2G) algorithms  Integrate with FREEDM GreenHub digital testbed

Potential Impacts:  Provide solutions to enable a smooth interaction between the plug-in vehicles and power grids  The proposed technologies can be extended to other large-scale PHEV/PEV charging/V2G scenarios as well as largescale power system applications.