Short answer: for most sites in East and West Africa and Southeast Asia, three nights of battery autonomy is enough, but only if the panel is about 30 % larger than the worst month strictly needs. In our 10-year simulation, a panel sized exactly to the worst month left lights dark 5 to 16 nights a year even with three nights of battery; with 30 % more panel, every city fell to 2 nights or fewer.
Tenders often specify “3 days of autonomy” or “5 rainy days” and leave the panel to the bidder. That is the wrong way round: a battery only bridges cloudy days if the panel can refill it afterwards.
Key numbers
- We used NASA POWER daily solar irradiation for 2015–2024, 3,653 days per city, for 10 cities.
- Worst-month averages ranged from 3.6 kWh/m²/day (Yangon, August) to 4.8 (Addis Ababa, August).
- With the panel sized exactly to the worst month, going from 3 to 5 nights of battery still left up to 7 dark nights a year (Yangon). Adding 30 % panel with 3 nights of battery cut that to 1.5.
1. The model
Each simulated light uses 1 unit of energy per night. The panel delivers its rated power × the day’s irradiation × 0.75 for temperature, dust, wiring and charging losses. The battery holds a number of nights of usable energy. Each day the panel charges the battery; each night the light draws its energy, and if the battery cannot cover the whole night, that night counts as dark (in practice the controller dims early). Panels are horizontal, which is conservative away from the equator.
2. Results: dark nights per year
| City | Worst month (kWh/m²/day) | Panel 1.0×, 3 nights | Panel 1.0×, 5 nights | Panel 1.3×, 2 nights | Panel 1.3×, 3 nights |
|---|---|---|---|---|---|
| Nairobi | 4.71 (Jun) | 7.0 | 1.8 | 0.8 | 0 |
| Kampala | 4.74 (Jul) | 7.0 | 0 | 0 | 0 |
| Dar es Salaam | 4.52 (Apr) | 10.3 | 5.4 | 4.5 | 1.9 |
| Addis Ababa | 4.78 (Aug) | 8.4 | 4.7 | 1.0 | 0.1 |
| Lagos | 3.82 (Jul) | 4.6 | 1.4 | 0.8 | 0 |
| Abuja | 4.06 (Aug) | 4.7 | 1.3 | 0.1 | 0 |
| Yangon | 3.63 (Aug) | 15.7 | 7.1 | 4.5 | 1.5 |
| Mandalay | 4.29 (Dec) | 7.1 | 0.8 | 2.1 | 0.1 |
| Kuala Lumpur | 4.33 (Dec) | 9.7 | 4.5 | 1.8 | 0 |
| Vientiane | 4.61 (Aug) | 15.4 | 4.9 | 3.0 | 0.3 |
Our own simulation. “Panel 1.0×” means a panel that exactly covers one night’s energy on an average day of the worst month; 1.3× is 30 % larger. Satellite data average out local showers, so real sites see somewhat longer dull spells; treat these as relative results, not guarantees.
3. Why panel margin beats battery days
After a cloudy spell, the battery has to be refilled while the light keeps running every night. With a panel sized exactly to the worst month, there is no surplus to refill with: in the rainy season the battery is never fully recharged, so the next dull spell finds it half empty. Extra battery helps only if there is extra energy to put into it. That is why five nights of battery on a tight panel performed worse than two nights on a panel with 30 % margin in most cities. Our sizing guide shows the recovery calculation.
4. Our recommendation by climate
| Climate | Examples | Panel margin over worst month | Battery autonomy |
|---|---|---|---|
| Highland and inland East Africa | Nairobi, Kampala, Addis Ababa | +30 % | 2–3 nights |
| West African savanna and coast | Abuja, Lagos | +30 % | 3 nights |
| Monsoon coast | Yangon, Dar es Salaam | +30–50 % | 3–4 nights |
| Inland Southeast Asia | Mandalay, Vientiane, Kuala Lumpur | +30 % | 3 nights |
Autonomy here means usable energy. A LiFePO4 pack should be specified so that, after the depth-of-discharge limit and the capacity lost by the end of the warranty, it still holds those nights; see our warranty guide. A dimming profile (for example full power until midnight, then 50 %) reduces nightly energy and is the cheapest way to add autonomy.
5. How to write it in a specification
- State the worst-month irradiation for the site and its source (NASA POWER or Global Solar Atlas).
- Require the energy balance: nightly Wh by dimming step, panel Wh in the worst month, and the margin.
- Specify autonomy in nights of usable energy at end of warranty, not nameplate amp-hours.
- Ask for the recovery time after the autonomy period: how many average days to recharge fully.
FAQ
Is 3 days of autonomy enough?
For most of the cities we tested, yes, with a panel about 30 % larger than the worst month needs. On monsoon coasts, use 3–4 nights and more panel margin.
Why not just specify 5 or 7 days of battery?
Without extra panel, the battery cannot be refilled in the rainy season, so the extra capacity is rarely full when it is needed. It also adds cost and weight, and theft value.
Which month should I size for?
The month with the lowest average irradiation at the site. It is not always the rainiest month; in Mandalay and Kuala Lumpur it was December in our data.
Where can I get daily solar data for my site?
NASA POWER provides free daily irradiation for any coordinates; the World Bank Global Solar Atlas gives long-term monthly averages.
How we help
Send us the site coordinates and the required lighting hours. We will run the same daily simulation for your location and size the panel, battery and dimming profile of our solar street lights to it, with the energy balance in the offer.
Sources: NASA Langley Research Center, POWER project, daily all-sky surface shortwave irradiance (ALLSKY_SFC_SW_DWN), 2015–2024; World Bank Global Solar Atlas. The simulation, loss factor and recommendations are our own engineering analysis. Related: all-in-one vs split, road lighting classes.