Stage 6: Survey, sizing and basic design · topic 23 of 53

Sizing, yield estimates and losses

Turn a bill and a roof into an array size and a believable yield estimate, follow the chain of losses, and compare the estimate with what the system actually makes.

  • Core
  • Intermediate
  • about 12 hours

Part of the Solar PV technician roadmap: from first safety to mid level

Written by the GreenHired editorial teamSources last checked 20 November 2026How we research roadmaps

Sizing runs in one direction: use, size, yield, savings. Use comes from twelve months of bills, size from the roof and the budget, yield from a tool fed the roof’s tilt, orientation and shade, and savings from how much of that yield the household consumes itself. Starting from yield gives a number that looks precise and means little.

From 4,000 kWh a year to a payback range

A household uses 4,000 kWh a year, and its roof takes ten 450 Wp modules, which is 4.5 kWp. Every figure in this example is made up, including the household’s use, the yield, the prices and the cost. None is a real tariff or a real PVGIS result, and prices are in plain currency units.

The tool is assumed to give 900 kWh per kWp a year, so the array makes 4.5 x 900 = 4,050 kWh. Run the sum twice, because payback depends on how much the household uses itself.

If 40% is used on site: 4,050 x 0.40 = 1,620 kWh used and 2,430 kWh exported. With an import price of 0.25 per kWh and an export rate of 0.05, that is 1,620 x 0.25 = 405 plus 2,430 x 0.05 = 121.50, so 526.50 a year. An installed cost of 6,500 gives 6,500 / 526.50 = about 12.3 years.

If the household is out all day and uses only 30%: 1,215 kWh used (303.75), 2,835 kWh exported (141.75), 445.50 a year, and 6,500 / 445.50 = about 14.6 years.

Ten points of self-consumption move payback by more than two years. That is why MIS 3002 wants the occupancy pattern stated and the figure marked as guidance for the first year, not a guarantee, and why you quote a range. Once the size is chosen, string, inverter and cable design checks it against the inverter.

What PVGIS’s 14% covers and what it misses

The PVGIS manual describes its 14% default as the losses between the modules and the grid, and names cables, inverters, dirt (sometimes snow) and the slow loss of power as modules age. Shading is not in that list, and the shading option the manual describes is horizon data for hills and mountains. A chimney or a tree beside the roof is something you must account for yourself.

The MCS method does it with a shade factor: SF = 1 minus the estimated shade loss, so an estimated 11% loss gives SF = 0.89 (the standard’s own example in MIS 3002 issue 6.0, Appendix B). That is why the shade check in the site and roof survey changes the estimate, not just the drawing. Running the same house through a second tool is how you check a first one; design and estimating tools shows how.

What you will learn

  • Work out a customer's annual and daily use from bills and choose a sensible array size in kWp for the roof and budget
  • Run a yield estimate in PVGIS, understand its inputs, and see what its default 14% system loss covers and what it leaves out
  • Follow the loss chain: temperature, soiling, shading, wiring, mismatch and inverter limits
  • Separate generation from self-consumption, and explain why an estimate is guidance and not a guarantee
  • Compare an estimate with measured output using annual yield (kWh per kWp) and performance ratio

The parts of this topic

Load analysis from billsabout 2 hours

Read twelve months of bills to find annual use, seasonal swing and daytime use. These set the size and the savings far more than the roof area does.

Total the kWh on 12 monthly bills; if a smart-meter profile exists, note how much is used between 9 and 5. A household at work all day uses little of its own solar; a home office uses more. Example: 4,000 kWh a year is about 11 kWh a day (4,000 / 365 = 10.96). A single summer bill hides winter use, so size from the full year. The NABCEP guide shows reviewing electric bills as part of the site survey.

Array size and a PVGIS estimateabout 3 hours

Choose kWp from roof space, budget and usage, then estimate annual yield in PVGIS from location, kWp, mounting, tilt and azimuth.

kWp is the sum of the module label ratings at standard test conditions; ten 450 Wp modules, for instance, make 4.5 kWp. PVGIS asks for location, PV technology, installed peak power, mounting type, slope and azimuth, and reports annual and monthly energy, year-to-year variability and a loss breakdown; its default system loss is 14%. Before you accept that default, check that the roof, wiring and inverter match it.

Temperature, soiling and shading lossesabout 2 hours

Heat, dirt and shade each remove some output. Temperature depends on air temperature, sunlight, wind and mounting; soiling varies by place and cleaning.

The PVPMC pages say module temperature depends on air temperature, irradiance, wind speed and module materials, and that soiling depends on the particle type and time since cleaning, and varies even between sub-arrays. One loss figure for every site hides the places where cleaning or shade removal would pay: a flat roof in a dusty area loses more to soiling than a steep, rain-washed one.

Wiring, mismatch and inverter lossesabout 2 hours

Cable resistance, modules working at a shared compromise point and inverter limits all cost energy. Wiring loss grows with the square of the current.

The PVPMC pages say DC wiring losses come mainly from cable resistance and scale with the square of current; mismatch arises because series strings share one current and parallel strings one voltage, so devices sit off their best point; and clipping happens when array power exceeds what the inverter accepts. Two strings of unequal length with different cable sizes, for instance, have uneven voltage drops, and on long runs these losses are not small enough to ignore.

Self-consumption, payback and tariffsabout 2 hours

Generation is not savings: only energy used on site avoids a bill, and exported energy is paid at a different rate. Payback compares cost with yearly benefit.

MIS 3002 requires a self-consumption estimate based on an occupancy pattern and annual use, and says it is not a guarantee. Savings come from energy used on site at the import price, plus any export payment, and simple payback is installed cost divided by yearly benefit. With 3,500 kWh generated and 35% used on site, 1,225 kWh of import is avoided; multiplying all generation by the import price would overstate savings. A university course covers solar economics in more depth.

Estimate versus measured outputabout 1 hour

Annual yield in kWh per kWp and performance ratio let you compare a system with its estimate and with other systems, allowing for weather and temperature.

PVPMC defines annual yield as yearly energy divided by the DC rating at standard test conditions, in kWh per kWp, and it varies with the weather from year to year. Performance ratio (IEC 61724 style) is AC yield divided by reference yield and is sensitive to temperature, so it reads lower in warm months. MIS 3002 calls the estimate guidance for the first year only, and one dull month proves little about a system.

Practice task

From a real or sample 12-month electricity bill, choose an array size for a south-facing 35-degree roof, run PVGIS for your location and write a one-page estimate: annual kWh, kWh per kWp, a loss table, a self-consumption assumption you state openly and a plain-English caveat paragraph. Then change tilt and azimuth by 30 degrees and note what changes.

Skills you build

kWp sizingPVGIS yield estimateLoss chain reasoningEstimate versus measured output

Resources for this topic

Every source here is free. Each link opens the provider's own site in a new tab.

Back to the whole roadmap

Live jobs

Latest Renewable Energy jobs

No title matches the role right now, so these are the latest jobs in the same category.

All Renewable Energy jobs

No live jobs to show right now. Check back soon, or browse every green job.

Spot a problem with this topic?

Tell us what is wrong, missing or unclear here. Your note is attached to this topic.

Notes are sent from this page with JavaScript on. Without it, write to us on the contact page.