RP-8-25 Solar Edition | By Piotr Mikus
RP-8-25 Solar Edition (Series): Browse the series: https://solarlightingnightshift.com/category/rp-8-25-solar-edition/ | Previous: The Fourth Column: Why Veiling Luminance Is the Metric Solar Submittals Refuse to Show
(What to demand before accepting a solar lighting sizing sheet):
- Daily or hourly irradiance for the full record, not a monthly mean, with the dataset and years named
- The irradiance plane stated: horizontal, or the tilted panel plane with tilt and orientation
- The system derate itemized: soiling, wiring, controller and battery efficiency
- The load in watt hours for every programmed profile, controller and communications draw included
- Battery chemistry, voltage, usable depth of discharge and charging temperature floor
- A battery state of charge simulation through the whole record, not days of autonomy
- The unmet load percentage and the longest recovery time after the worst stretch
- The worst year in the record, not the worst average month
Put NASA on the page and the reviewing stops there.
A solar lighting sizing sheet crossed my desk. It cited 20 years of NASA data, and it promised about 5 days of backup and an array to load ratio of 1.2. I rounded the values here, because the pattern matters more than the page. NASA is an excellent data source. The way its name gets used is the problem.
A Name on the Page Ends the Review
NASA stands behind the measurements. It has signed off on nothing the sheet does with them.
I asked colleagues how they would treat a spec that names NASA or NOAA as its source. Most said they would never challenge it, because those names read as the last word. That reflex is the whole mechanism. The name certifies the input, and the reader assumes it certifies everything built on top. The sheet then makes its averaging, derating and battery choices under a borrowed halo, and nobody reviews them because the review ended at the first line.
A Monthly Average Contains No Days
The sun delivers days. The sheet bought an average.
The sheet sizes on one December number, the average of 20 years of December days. A December with two gray weeks and two clear ones gives the same average as thirty mediocre days, and a battery treats those two Decembers very differently. Whether it runs empty on day 10 depends on the order of the days before it, and an average has no order. The sheet calls December the month with the worst average sun. That is the lowest of twelve averages, and the worst December in the record sits below it.
The number is also hard to trace. The December bar on the page’s own chart reads lower than the number printed beside it. The page never says whether the value is for a flat surface or the tilted panel, and in winter that difference is big.
One Average, One Guess, One Promise
Reliable performance for 365 days a year, courtesy of one bar on a chart.
I rebuilt the math behind the page, and it is short. Take the energy the panel makes on an average December day. Take away the losses. Compare what is left to what the light uses overnight. The sheet lands at 1.2, which means the panel makes about one fifth more than the light needs.
Look at what went in. The panel is counted at its rated watts, the light at its nominal watts, the battery at its label capacity. Those are datasheet numbers, and a datasheet is a lab result. The page never lists the real world losses: snow, dirt, wiring, controller losses, a battery that is cold or getting old. The math only closes if one loss factor is hiding in there, and nothing shows how it was built. The page also calls its backup figure worst case winter, while the input behind it is an average.
A 20 Percent Surplus Is Thinner Than It Sounds
A 20 percent surplus sounds generous until it is the only energy the battery ever sees.
A ratio of 1.2 sounds comfortable. That leftover energy is also the only thing that refills the battery after a cloudy stretch. If a December comes in about 15 percent below average, the leftover is gone, and the panel just covers the light with nothing to spare. The 20 years of record the sheet cites already show how often that happened. The sheet does not say.
Days of Backup Is a Battery Sum
Five days of backup, and nobody has to explain how it gets refilled.
Five days of backup sounds like safety. It is a storage number, and storage only helps if the battery gets refilled. With that small leftover, refilling five days of backup takes about four weeks of average December sun. The sheet shows the five days and skips the four weeks.
The sizing page lists the battery size and nothing else. The type sits on a different page of the package: lead acid. To deliver those five days, that battery would be drained about 70 percent, counted against its label capacity, with no allowance for cold or age. Lead acid ages fast when it is drained that deep. Stop the drain at 60 percent and the same battery lasts a little over four nights. The package also lists a charging limit at freezing, and December days can sit below it.
Storage cannot fix an energy shortage. If the panel makes less than the light uses through the worst stretch, a bigger battery only delays the blackout.
What to Run on NASA Data Instead
NASA publishes the daily and hourly series. The sheet picked the bar chart.
NASA POWER publishes daily and hourly sun data for any location, and a sizing run should start there. Put in the real panel tilt, the full nightly load including the controller and communications, and the real battery. Then simulate it hour by hour through every winter on record with HOMER Pro. Report how many hours a year the light misses its schedule and how long it takes to recover after the worst stretch.
Asking for that run leaves NASA’s reputation untouched. It checks the part the name never covered, and the count of missed hours is the real reliability claim. A ratio and a day count cannot stand in for it. My earlier post, Solar Street Light Energy Budget: Monthly Harvest vs Load, Autonomy, and Recovery, walks through the monthly side of that balance.
What to Require in a Specification
If the sheet cannot show a bad winter, nobody ran one.
Keep it enforceable:
- Irradiance source named down to the dataset, resolution and years used, so the citation can be checked as well as admired
- Simulation method named: a sequential hourly run in HOMER Pro, with the input files attached
- Plane stated for every irradiance value, with tilt and azimuth given
- Battery chemistry, nominal voltage, usable depth of discharge and the low temperature charging limit printed on the sizing page
- Load in Wh for each programmed profile, controller and communications included
- Derate itemized line by line
- Unmet load in hours dark per year against a stated target
- The worst year in the record and the longest recovery time after the worst stretch
Three Questions That Expose a Borrowed Authority
Polite questions. The answers tend to run long or not at all.
Which dataset and which years fed this value, and was it taken on the horizontal or the tilted plane?
What usable depth of discharge produced the days of backup figure for that chemistry, and what is the charging temperature floor?
In the worst year of the record, how many hours does the light miss its profile, and how many nights does recovery take?
If the answer is a ratio and a day count, nobody ran the days.
Closing Thought
A famous source name has never kept a streetlight on in January.
If a sizing sheet cannot name the worst day it survived, it never looked for one.
Sources and Where to Verify
- NASA POWER, Prediction of Worldwide Energy Resources, daily and hourly solar irradiance series: https://power.larc.nasa.gov/
- NREL, PVWatts Calculator, monthly production by tilt and orientation: https://pvwatts.nrel.gov/
- IEEE 1562, Guide for Array and Battery Sizing in Stand Alone Photovoltaic Systems
- HOMER Pro documentation, time series simulation and imported resource data
- Battery manufacturer datasheets for the stated chemistry (usable depth of discharge, cycle life, charging temperature range)
- Solar Lighting Night Shift, Solar Street Light Energy Budget: Monthly Harvest vs Load, Autonomy, and Recovery: https://solarlightingnightshift.com/2026/02/15/why-should-i-care-about-energy-budgets/
- A solar lighting manufacturer’s sizing methodology package (name withheld)
Piotr Mikus is a roadway lighting designer and specifier focused on solar powered street lighting and controls.
Quick FAQ
Does citing NASA make a solar sizing sheet reliable? It makes the irradiance input credible. Reliability depends on what the sheet did with that input: the averaging, the derates, the battery depth, and whether anyone simulated a bad winter.
Why is a monthly average not enough to size a solar light battery? It contains no sequence. Two gray weeks and two clear weeks give the same mean as thirty mediocre days, and battery drawdown depends on the order of the days.
What should a sizing sheet report instead of days of backup? Unmet load in hours dark per year from a sequential simulation on daily or hourly data, the worst year in the record, and the recovery time after the worst stretch.
Continue reading the series: https://solarlightingnightshift.com/category/rp-8-25-solar-edition/
