P50 is the middle of the range. It is the annual production you should expect in a typical weather year, after location, tilt, temperature and your loss assumptions are applied.
P90 is the conservative planning number. It is the figure production should reach or exceed in roughly nine years out of ten. It is always lower than P50, and the gap tells you how variable your site is.
Neither number is a guarantee. The model does not see shading, inverter downtime or module failure, so use P50 for planning and P90 for a cautious case, not for a bankable loan application.
What the yield calculator actually answers
Most solar calculators answer “how big should the system be?” This one answers a different question: given this system, at this place, facing this way, how much energy will it produce in a year? That makes it the natural second step after sizing. Sizing tells you the kWp you need; yield tells you whether the design you have in mind will actually deliver it.
The calculator returns four things that are easy to confuse:
- P50 annual generation (kWh/year) — the central estimate.
- P90 annual generation (kWh/year) — the conservative estimate for a bad-but-plausible weather year.
- Specific yield (kWh/kWp/year) — production per installed kilowatt, which lets you compare locations and designs.
- Monthly profile — the seasonality that decides whether your annual total is useful in the months you need it.
If you only read one number, read specific yield. A 10 kWp system in a poor location and a 7 kWp system in a strong one can produce the same annual energy; the specific yield tells you which is the better use of capital.
The inputs that move the answer
System size and location
Capacity is the DC nameplate in kWp. Location is not a cosmetic field: it selects the historical weather record used for the simulation. The calculator uses the last five complete calendar years of solar radiation and temperature, then runs a one-year hourly simulation for a reference year and corrects it to the long-term monthly averages.
Direction and tilt
Azimuth is measured from true north (0° = north, 180° = south). In the northern hemisphere, south is the annual optimum; in the southern hemisphere, north is. Tilt changes both the amount of irradiance the plane receives and how much of it arrives at a useful angle. “Auto” derives a tilt from latitude — roughly 0.87 × the absolute latitude — which is a good annual compromise, not a site survey.
Soiling
Soiling is the average annual loss from dust, bird droppings, pollen or snow. The default is moderate (about 2%). In dry, dusty regions, or near construction and unpaved roads, high soiling is more honest. Cleaning reduces the loss, but the calculator does not schedule cleaning for you.
Engineering settings
The advanced fields are the panel and installation values that shift production by a few percent each:
| Input | Typical value | What it does |
|---|---|---|
| System losses | 8% | Cables, inverter conversion, mismatch, availability and other fixed losses, combined into one derate. |
| Temperature coefficient (Pmax) | −0.35%/°C | Power lost per degree of cell temperature above 25 °C. A cooler cell gains a little; a hotter one loses more. |
| NOCT | 42–45 °C | The module’s operating temperature under standard conditions. It sets how hot the cell runs for a given irradiance. |
These are not cosmetic. A hot site with a poor temperature coefficient can lose several percent more than a mild site with the same sunshine, and the model applies that hour by hour rather than as a flat annual haircut.
The method, step by step
The important thing to understand is that the calculator does not multiply kWp by a single “sun hours” number. It builds an hourly energy balance and then corrects it for the long-term climate.
2. Fetch one reference year of hourly plane-of-array (POA) irradiance at the chosen tilt and azimuth.
3. Scale each month of that reference year to the 5-year monthly average.
4. Estimate cell temperature: Tcell = Tambient + (POA ÷ 800) × (NOCT − 20)
5. Temperature multiplier: 1 + (temp coefficient ÷ 100) × (Tcell − 25)
6. Hourly energy = kWp × POA ÷ 1000 × temperature multiplier × (1 − fixed losses) × (1 − soiling)
7. Sum the hours to annual and monthly kWh.
Two details are worth calling out. First, the temperature multiplier can be slightly above 1 in cool conditions, which is physically correct: a cell below 25 °C produces a little more than its nameplate. Second, the losses are multiplicative, not added. An 8% loss and a 2% loss leave 90.16% of the ideal output, not 90%.
How P90 is derived
P90 is not a second simulation with worse weather. It is a statistical planning estimate applied to the P50:
P90 = P50 × (1 − 1.282 × σtotal)
The inter-annual term comes from how much annual radiation varied across the available years. The system term is a 5% uncertainty allowance. The 1.282 multiplier is the one-sided normal value for 90% confidence. This is a reasonable planning convention, but it is not the lender-certified P90 that a bank’s independent engineer would produce from a longer record and a full uncertainty budget.
A worked example
Take a 10 kWp system in Lahore, facing true south, at the auto tilt for latitude 31.5° (about 27°), with moderate soiling and default losses. Suppose the simulation returns a specific yield of 1,550 kWh/kWp/year:
Monthly average = 15,500 ÷ 12 ≈ 1,292 kWh/month
If inter-annual variability is 6%: σtotal = √(0.06² + 0.05²) = 0.078
P90 = 15,500 × (1 − 1.282 × 0.078) ≈ 13,950 kWh/year
The 10% gap between the two is the point. If you are comparing a solar loan against a grid bill, use the P90 figure for the conservative case and the P50 for the expected case. A proposal that quotes only the higher number is not lying, but it is not showing you the risk either.
Seasonality matters as much as the annual total. In most of Pakistan the summer months produce far more than December and January. If your expensive consumption is evening-heavy, annual kWh alone overstates what the system will offset; pair this result with the battery ROI or grid vs battery analysis before assuming a straight bill reduction.
Where the estimate stops being reliable
Use the result for three decisions it is good at: comparing two orientations, checking whether an installer’s production claim is plausible, and deciding whether a design is worth a full engineering study. Do not use it as the sole basis for a loan, a power-purchase agreement or a warranty claim.
How to read the monthly chart
A system can hit its annual target and still disappoint in the months you care about. Look for three things:
- The worst month. If December production collapses, a grid connection or battery is doing more work than the annual number suggests.
- The peak-to-minimum ratio. A high ratio means more seasonal storage or grid dependence; a flat profile is easier to self-consume.
- Plane-of-array irradiance versus generation. If generation falls faster than irradiance in summer, temperature and losses are eating the difference.
If you want the electrical design that goes with the production estimate, the next step is string sizing. If you have not chosen the array size yet, start with the system size calculator and come back here with a real kWp.
Frequently asked questions
What is the difference between P50 and P90 in solar?
P50 is the median or expected annual production: roughly half of plausible weather years should be above it and half below. P90 is the conservative planning figure that production should reach or exceed in about 90% of years. P90 is lower because it subtracts 1.282 standard deviations of the combined weather and system uncertainty from P50.
How much does a 10 kWp system produce per year?
Multiply 10 kWp by the specific yield for your location and design. In much of Pakistan that is roughly 1,300–1,700 kWh/kWp/year, so about 13,000–17,000 kWh/year. Hotter, dustier or east/west-facing systems sit at the lower end; cooler, cleaner, south-facing systems at the higher end. The calculator returns the figure for your exact coordinates rather than a national average.
Why is my estimate lower than kWp × 24 × 365?
Because the sun does not shine at full strength for 24 hours. A 10 kWp array is rated at 1,000 W/m² irradiance and 25 °C cell temperature, conditions that occur for only part of a sunny day. Night, low sun angles, clouds, dust, high cell temperatures and conversion losses all reduce the total. A specific yield of 1,300–1,700 kWh per kWp per year is normal, which is 15–19% of the theoretical 24-hour maximum.
Does the calculator account for shading from trees or buildings?
No. It models the irradiance at your coordinates and tilt, not the obstructions around your specific roof. A chimney, parapet, water tank or tree can remove far more production than a few degrees of tilt. Use the roof space checker to study obstacles and shading, and ask an installer for an on-site shade survey before signing.
Can I use this P90 for a bank or investor?
Not on its own. The page labels it a planning estimate, not a lender-certified P90. A bankable study normally uses a longer weather record, a full uncertainty budget, an on-site shade measurement, confirmed equipment specifications and an independent reviewer. This tool is designed to tell you whether a proposal is in the right range before you pay for that study.
Why does the result change when I change the tilt by a few degrees?
Tilt changes the angle at which sunlight strikes the plane, so it changes both the amount of irradiance received and its seasonal distribution. A few degrees near the optimum usually moves annual yield by well under 1%, which is smaller than the uncertainty in the weather data. Large changes — a flat roof versus a steep pitch, or east versus south — matter much more.