P50, P90, and P99 are probability labels engineers use to decide which weather a design has to survive. Read them as odds. In any given year there is a 50 percent chance conditions are at least as severe as the P50 case, a 10 percent chance they are at least as severe as P90, and a 1 percent chance they reach P99. In battery storage design, "severe" means hot, because heat is what limits a BESS: it raises the auxiliary power the thermal system draws and it derates the capacity the cells can deliver at the same time.
This guide explains what each level means, why BESS sizing works from hot-year cases instead of averages, how the same P-labels mean something different in solar yield estimates, and where the profiles come from. It is written to be readable without a statistics background.
The P stands for probability of exceedance. Rank many years of real weather for a site from mildest to most severe and pick points along that ranking. P50 sits in the middle, P90 sits near the severe end, and P99 sits at the edge of the record. None of them is a forecast for a specific year; each is a statement about how often a year at least that severe shows up over the life of the project. A plant designed to a 20 year term should expect to meet its P90 year roughly twice, which is exactly why that case matters.
The P50 case is the median: a year hotter than half the record and milder than the other half. It tells you how the plant behaves in a normal year, which makes it the right case for understanding expected auxiliary consumption and day-to-day operating margin. It is the wrong case for a guarantee, because half of all years are worse.
The P90 case is hotter than roughly 9 out of 10 years. For most battery storage projects it is the binding case: the year where thermal management works hardest, available capacity derates furthest, and the design either holds its usable energy target or does not. Sizing the guarantee against P90 means the plant delivers through the hot years it will actually meet, not just the average ones. The guide on how to size a utility-scale BESS walks through why the hot case usually binds.
The P99 case is hotter than 99 out of 100 years: near the worst the site record contains. Most projects do not size their base guarantee to it, but it earns its place in two situations. Some contracts require the target to hold even in near-worst conditions, and some owners want to know exactly how far delivery falls in an extreme year rather than being surprised by it. Checking the design against P99 answers both without changing the equipment count unless the contract says it must.
Temperature hits a battery plant twice on the same afternoon. The HVAC and controls draw more auxiliary power as it gets hotter, and the cells deliver less usable capacity at temperature extremes. Those are also the afternoons when the grid leans on storage hardest. A design sized to average weather underdelivers precisely when it is expected to perform, so the hot cases set the margin and the typical case describes the ordinary days in between. Degradation compounds the effect over the years, which is why weather cases and the augmentation schedule belong in the same model.
The same labels appear in solar production estimates with the opposite feel, and mixing them up is a common source of confusion in solar-plus-storage work. In a solar yield report, P90 is a low-energy case: the production you expect to exceed in 90 percent of years. In weather design conditions, P90 is a severity case: a year hotter than 90 percent of years. Both are conservative in their own context. When a document says P90, check whether it ranks energy or weather before comparing numbers.
A percentile is only as good as the record behind it, so the profiles should be built from long, site-specific history rather than a regional average. FluxPilot pulls 25 years of real weather for the project location and derives the P50, P90, and P99 design conditions from it in one click, then uses them to drive auxiliary load and capacity derate inside sizing. Change the site and the cases change with it; change the battery or the target and the same weather re-evaluates the new design in seconds. See the full capabilities, or book a demo to watch the cases built for a site you are working on.
P50 is the median case: a typical year. In any given year there is a 50 percent chance conditions are at least that severe. Engineers use the P50 case to understand expected behavior, not to set a guarantee.
Both are severity percentiles. A P90 year is hotter than roughly 9 out of 10 years, and a P99 year is hotter than 99 out of 100. P90 is the usual sizing case for battery storage; P99 covers projects that must hold their target in near-worst conditions.
No. In a solar production estimate, P90 is a low-energy case: the yield you expect to exceed in 90 percent of years. In weather design conditions, P90 is a severity case: a year hotter than 90 percent of years. Both are conservative in their own context, but they point in opposite directions, so always check which convention a document is using.
Model the P50 typical year to see expected behavior, size the guarantee against the P90 hot year because heat raises auxiliary load and derates capacity on the same days, and check the P99 extreme year when the contract requires delivery in near-worst conditions.