A Probabilistic-based PV and Energy Storage Sizing Tool for Residential Loads

Xiangqi Zhu, Jiahong Yan, and Ning LuObjective:

Technical Approach:

Accomplishments:

Next Steps:

Potential Impact:

This work presents a probabilistic-based sizing tool for residential homeowners, load serving entities, and utilities to select energy storage (ES) andphotovoltaic (PV) based on historical load characteristics and load managementoptions.

Inputs: historical load profiles and solar radiation data. Outputs: home energy consumptions, monthly energy payment, self-consumed

solar and backfeeding energy, etc. Demand-side Energy Management (DSM) is also involved.

Data DescriptionVarious patterns of house load models and solar power output models could be extracted from this data set.

Modeling and Sizing Methodologies

target load PVP P P= −

target max_d target max_c _ _d

max_d target max_d _d

max_ target max_c _c

, ( ) & E ( 1)

, & ( 1)( )

, & ( 1)

0

Lmt c ES Lmt

ES LmtES

c ES Lmt

P P P P E t EP P P E t E

P tP P P E t E

otherwise

> > − > − >

> − >= − > − <

(t) ( 1) ( )ES ES ES

E E t P t T•= − −

targetPloadP

PVP( )

ESE t

( )ESP t

max_ cPmax_ dP

max_ cE

max_ dE

:Target battery power output, kW; : Load power consumption, kW: PV power output, kW; : Power output of the battery, kW: Battery energy level at t, kWh;

:charging and discharging power limits, kW

:charging and discharging energy limit, kWh

Battery Control Logic

Battery Sizing MethodologyStep 1: Create seasonal load groups.

Winter load group November, December, January, February, MarchSummer load group April, May, June, July, August, September,

October

TABLE I. Load Categorization

Step 2: Generate the net load ensembles.For each load group, an ensemble of the net load scenarios is generated by subtractingcorresponding all possible solar profiles from each load profile ( as shown in Fig.5).

int int int151 96 151 96 151 96w ergroup w ergroup w ergroupnetload load solarP P P

× × ×= −

214 96 214 96 214 96summergroup summergroup summergroupnetload load solarP P P

× × ×= −

Step 3: Calculate the statistics of the backfeeding power.When the battery power or energy limit is reached, the excess solar power will be backfed to themain grid. After modeling the battery operation for all scenarios, we obtain a probabilitydistribution of the daily reverse power for each 15-min interval that can help the user to evaluatethe occurrence of reverse events. As shown in Fig. 6, the black line represents the 80% probabilityline for a 5kW PV operating in conjunction with a 1kW/1kWh battery.

Fig.1. Fig.2.

Fig.3. Fig.4.Fig.5. Fig.6.

Sizing PV and ESD at the Home Level As shown in Fig. 7, Reverse power is not anissue in the winter load group but can causeproblems in the summer load group This showsthat it is not wise to just use one size of ES forthe whole year because of the seasonal loaddifferences. Also, the users can run different ESand PV combinations to find the optimal one thatmeets the power limits of backfeeding.

Fig.7.

Sizing PV and ESD at the Transformer Level Assume that a transformer supplies three homes,each of which has a 5-kW rooftop PV. As shown inFig.8, installing a 2kW/2kWh battery at thetransformer level will receive the same result asinstalling one 1kW/1kWh battery at each home.

Fig.8.

Next Step:

Sizing PV and ESD at the Feeder LevelFig.9 shows the reverse power for penetrationlevel of 50%, 70% and 90% respectively, alsoshows the trend of the maximum reverse poweron the 80%-line of various penetration levels.To give a comparison, the red line in the trendfigure is obtained using a 10kW/3kWh ESD atthe feeder head.

In the future, sizing analysis involving DSM will be studied in detail, to see to what extend DSM will benefit the residential loads with renewableenergy.

This method can give the users a clear comparison of the tradeoffs among different PV and ES options and assist them make more informed decisions.

Fig.9.