Intro
Everybody knows them: solar garden lamps. They work quite well during summers, when there is a lot of sun, days are long and nights are short. But come winter nights and you barely get anything from them. So i thought, what would it take to make a solar lamp that can always run when it is dark, winter or summer?
In fact, this ties into a more general question: how do I size a solar panel and battery to insure a certain device can work all the time only from solar energy? This question first popped into my mind when I wanted to make a solar powered monitoring station with a camera and weather station.
Getting some data
It’s easy to find maps that show the average sun hours per year, but that is not enough, since these hours differ a lot from day to day and from season to season. Since I don’t live in a house with solar panels, I looked for data elsewhere and found that one of the energy suppliers in Belgium shares historic data. Armed with 2 years worth of solar generation data, I crunched the numbers to understand a bit what is going on. The data is reported for the whole country, but from big solar installations, not home made ones. The capacity also changes over time, so I normalized the data to the installed capacity and also grouped it per day.
So, two years worth of historic solar generation looks like this:
I got some nice insights from looking at this data:
-in the best days, solar panels will produce the equivalent of 7 full sun hours (so if you have 2kW peak power installed, you would get 14 kWh of energy from them)
-in the worst days, solar panels will produce the equivalent of 0.1 full sun hours, or just 6 minutes.
-on average, we get 2.7 full sun hours of energy
-the ratio of best day / worst day is about 67, so the best summer day produces 67 times more energy than the worst winter day.
-the total per year is the equivalent of 1010 full sun hours, while Belgium is reported to have 1500-2000 hours of sun per year. The difference is coming from the fact that panels are not optimally oriented towards the sun all the time.
Now, this data will have some difference when translated to a single, local solar installation, so consider it as an approximation only. If you have a source for similar data from a single solar installation that you can share, I would be interested in it. One expectation I have from single solar system is that there are actually a lot of days with 0Wh production, which will skew the results.
Making a simple simulator
Armed with daily solar generation and 2 years worth of history, I made a simple simulator based on excel. The simulator works like this: at the end of each day, the energy in the battery is calculated as the battery energy from the previous day, plus the solar energy generated minus the energy consumed. The energy in the battery is capped up to the battery capacity or down to zero for each day. For a load to remain always powered, there should be no point when the energy in the battery dips to zero or below. For simplicity, I am assuming the load is running 24hours per day.
Below you can see two examples of the simulator in action, of what would be possible ways for a solar powered lamp. The load is assumed to be 20mW (typical for an LED) and the battery capacity is set to 5Wh. The comparisons below are identical, with the exception of the solar panel size (power). On the left side, we observe that a 0.6W panel is insufficient, as there are 57 days where the light will not be on permanently, per year. On the right side, we see that a 0.9W solar panel will be enough to power the LED on all the time.
Limitations: obviously the simulator is very simple, assumes everything is 100% efficient, assumes a constant daily load etc. In practice, things will need to be even more oversized.
Exploring more versions
Now, intuitively, you might observe that there are different combinations of solar panel size and battery size that would end up powering the load. One extreme is with a solar panel so oversized, that it would produce enough energy to run the load even in the worst days (when we get the equivalent of just 0.1 sun hours). On another extreme, we could use a battery so big that it can still provide energy in the winter from what the panels generated during the summer. And there are plenty of in-betweens. But how do these look like?
To make things simple, I decided to “run the simulation” with different values of storage and solar power panel for a load of 1/24kW or 42W, since this totals 1kWh of energy per day. This will help easily scale and understand these needs in the context of household energy, as many people would be familiar with their average kWh energy consumption. First, lets find the approximate edge cases: when the battery is really small and we oversize the panel, and when the battery is really large and we minimize the panel. Note that these may not be the absolute extremes, but they are close.
Extreme 1: Big solar panel, minimal battery capacity is 0.5kWh, or 12 hours worth of storage. From here we see the minimal solar panel power needed is 5100W. This panel will produce a lot more energy than the load needs, in fact only 7% of the energy ends up in the load in this situation, the majority of energy captured by the solar panel remains unused.
Extreme 2: Minimal solar panel (500W) and big battery capacity. Here the solar panel will generate exactly the total energy needed in a year, but the battery is required to carry this from summer to winter, resulting in a battery capacity of 74kWh, or 74 days worth of storage. In this case, 72% of the energy generated by the solar panel ends up in the load.
The in-betweens: I ran the simulation for various values of solar panel size between 500 and 5100W and looked for the minimal battery that can sustain the system. The value pairs of solar vs storage are shown below, together with the estimated cost:
What we learn from here is that there is a continuous trade-off between solar panel power and energy storage. However, when trying to reduce the size of the solar panel, the battery size goes up faster than the size of the solar panel goes down.
If we consider the cost of the solar panel and battery (1000 eur per kWh of storage and 1000 EUR/kWp solar panel installed) we see that the optimal is somewhere in the middle. Specifically, it is achieved with a battery storing about 2 days worth of energy consumption and a solar panel with an average energy production of about 5.4 times the daily energy consumption (i.e. 2kWp * 2.7 hours = 5.4 kWh for a 1kWh daily consumption).
What if I wanted to power my home like this?
We are 2 people currently living in an apartment, with electricity used for cooking, but not for heating and cooling, with an average consumption of 4kWh per day or about 1500 per year. While both the cost of solar panels and batteries are going down, my guestimate is that with these prices, we are still about 2-3X more expensive than grid generated electricity (assuming the solar panels last 20 years and I need 3 batteries during this time). This means we are not far, but still more expensive to be off grid around here. Toss in a heat pump which needs more energy during the winter exactly when you get less sunshine and I expect the equation to shift even further.
In practice, people’s homes will still remain tied to the grid. When adding the fixed cost of connection and the cost of energy from the grid and accounting for real installation costs, I expect the optimal will skew towards less battery and solar panels and more energy from the grid. The size of battery and solar panels grows fast when you want to cover absolutely every day and never run out of energy.
So, thanks to the curiosity of what it would take to power a solar lamp continuously throughout the year, I have a rough estimation of what it would take to power something else, like a home, from solar.
Ok, back to the lamp now
In part two of this, I will build the solar lamp that can get me through the winter. There’s already a spoiler here about what sort of battery and solar panel is needed, but construction is coming in the winter, so stay tuned for Part 2.