So, exactly how big a battery do I need to go 100% off grid
As the residential solar solution keeps gaining momentum, one of the questions often get asked is can Solar plus battery storage satisfy 100% of our electricity needs. And if so, then at what cost? There are enough cloudy days, especially during winter, to cast a doubt if that’s even possible without a garage full of batteries and at an exorbitant price. In this article I set out to get that answer.
And the answer is YES. It is easily possible and doable for an average American household to go completely grid free. All it needs is just a small assist from EV manufacturers and better utilization of existing battery technology. And I say this not on any theoretical proposition, but real-world data collected over the last decade.
My calculations are based on 10 years of hourly production data from a 5.59 Solarcity/Tesla PV system and 2 years of hourly meter data from electric provider Pepco Electric. The main result is to demonstrate that a regular American household which consumes about 11-kWh electricity/annually would need 40 kWh of battery storage, for less than $5,000 cost (at current manufacturing cost of LiOn battery and after tax incentive) along with 20 kW of solar panels to go completely off-grid. Based on current technology that would mean about 3 Tesla powerwalls in every home or, even better, tapping into the battery of any Electric Vehicle (EV), with a 70 kWh battery (as most new EVs do). The latter is not currently permitted and I hope that this article further bolsters a pitch I have made to the Biden administration to open customer access to EV batteries so we can move further along towards becoming grid-free.
And perhaps going 100% off-grid isn’t the right goal when batteries are still fairly expensive (about 1K $/kWh). How much coverage might be feasible with smaller batteries? My numbers show that you can get to 98% with — a 15 kW Solar PV system with 30 kWh. And the percentages (defined as # of days in a year when the solar+battery won’t be able to meet 100% of that day’s electricity requirement) aren’t shoddy even with smaller batteries as the following charts show:
1. Percentage of days when your daytime and nighttime electricity needs can be met by a 10 kW solar PV system and different of battery sizes:
% of days coverage by battery size for a 10 kW Solar system
2. Percentage of days when your daytime and nighttime electricity needs can be met by a 15 kW solar PV system and different of battery sizes:
3. Percentage of days when your daytime and nighttime electricity needs can be met by a 20 kW solar PV system and different battery sizes:
Data and Methodology
To understand the data I used, let’s start with the design of a typical residential rooftop solar system connected to home breaker panels and the grid as shown in the image below:
The 2 data points that I utilized obtained from inverter data (for solar PV production data) and the meter (for utility data).
The utility data is the sum of house consumption and solar production. For example, suppose in a given hour, a house consumes 3 kWh and a solar PV system produces 2 kWh in the same hour, then the utility meter would show 3–2 = 1 kWh, as the amount that you would buy from the utility. So, subtracting solar inverter data from the utility data gives you the house consumption data.
I performed the steps below to extract and transform the data to derive my results:
- Step 1. Analyze solar generation data
— 1a. download solar generation data from inverter
— 1b. Average out the data for a year
— 1c. Normalize this data for a 1kW system (i.e. divide is by 5.59) & extrapolate for perfect orientation (multiply by 1.27)**
- Step 2: Analyze meter data (date from utility)
— 2a. Download 2 years of hourly utility data
(The maximum period my utility company will let me download)
— 2b. Separate out the daytime and nighttime data
— 2c. Average out the daytime and nighttime data for a year
- Step 3. Calculate US household daytime and nighttime consumption
— 3a: Calculate annual daytime and nighttime consumption by subtracting 1a from 2a
— 3b. Multiply consumption data (step 3) by a factor of 1.44 to extrapolate this for an average American household of 10.715 kWh (against my annual usage of 7457.26 kWh).
- Step 4: Derive the results
1: Analyze solar generation data
The system size of my PV system is as follows:
- System Size: 5.59 kW DC
- Tilt: 30 degrees
- Azimuth: 115 degrees
Step 1a. For this I extracted 10 years of daily data (daily by hours) from my inverter.
Though the above data appears noisy, there are a few things that can be seen clearly:
- Since the annual graphs intersect each other several times, it proves that the panels performance has NOT degraded over time, at least not significantly enough to be noticeable — Good
- In the winter the production is low (week numbers 1–10 or 40–50 etc.) but it gradually increases as we move towards the summer (center of the graph) — Expected behaviour.
Step 1b. Next, I averaged out the daily solar production data for a year over 10 years
Step 1c & 1d. Next, I normalized this data for a 1 kW system in optimum orientation (ie divide is by my system size of 5.59 & multiply it by 1.27*)
2. Analyze meter data
Step 2a: My electric provider lets me download the usage/consumption data broken by day and time only for the last 2 years. So this is what I ended up downloading.
Step 2b. Next, treating 6am to 6pm as daytime and rest as nighttime, I separated out the usage data between those 2 sets.
Step 2c. Next, I averaged out the data for each day of a year over those 2 years
3. Calculate daytime and nighttime consumption
Referring to the very first graph, since the meter data shows the total usage which is solar generation plus household consumption, we get the consumption data by subtracting the solar generation. I multiply this data by a factor of 1.44** to arrive at a typical US household consumption pattern over a year for both daytime and nighttime.
4. Derive the results
Let us assume that we need a minimum of X kW system for solar and a minimum of Y kWh system for battery. Then in this step, our goal is to find out X and Y such that on any given day the battery has enough juice to last till the next morning when the sun starts shining again.
I did this calculation twice for each day: 6am in the morning and 6pm in the evening. The very first day, at 6am, the battery is assumed to be full to its capacity ie Y kWh. Then throughout the day, the battery will keep filling up as per the solar generation data for that day (step 1) and will keep depleting at the same time as per that day’s consumption data (step 2 above). This will leave the battery at a certain capacity, which will always be less than or equal to its full capacity i.e., Y kWh, that evening at 6pm.
Next, throughout the night, the house will draw only from that capacity as there is no sunlight. That will deplete the battery some to arrive at the next morning battery capacity.
Both battery capacities, if greater than 0, will denote that the battery could sustain the household demand during that period. If it is less than 0, then it didn’t. The percentage below shows the number of days in a year when this number was more than 0, implying the battery was able to sustain the load.
*Extrapolate for perfect orientation — Assuming if the system were set for maximum solar production which is 35 degrees pitch and 180 degrees azimuth in my area, how much it would have produced. per PV watt calculator (a respected calculator among solar installers, managed by govt organization) it turns out that I needed to multiply my system data by 1.127813.
** The step #3 gives a total annual consumption of 7457.26 kWh, hence this needs to be multiplied by 1.44 to mimic an average household consumption.