How do you design a solar PV system?

Establish the project's starting point first. If the consumption target is known, calculate the first system capacity range; if the usable area is the binding constraint, choose the module and build the layout; if the panel count or installed capacity is fixed in advance, verify that target against the site. In every flow, check the final layout's yield, its share of consumption covered, the inverter limits and the economics together.

What does annual yield per kWp tell you?

It is the energy the system is expected to deliver on the AC side each year for every 1 kWp of DC capacity installed, expressed in kWh/kWp per year. The industry also calls it specific yield, specific production or energy yield. It depends on location, orientation, tilt, temperature, shading, equipment and system losses, so it must never be carried over directly from another project.

How many kWp does a household consuming 4,200 kWh a year need?

In a first pass that starts from consumption, at an annual yield of 1,030 kWh per kWp, 4,200 ÷ 1,030 gives a 4.08 kWp capacity target. That figure is not the final panel count. The chosen module and the real roof layout may land on nine panels at 3.96 kWp or ten at 4.4 kWp, and the right option is settled by weighing yield, area, tariff and connection limits together.

What is a good DC/AC ratio?

There is no single ratio that holds for every project. PVWatts V8 uses 1.2 as its default for preliminary modelling, and that default is not an equipment selection rule. The final ratio has to be verified against an annual yield and clipping simulation, the manufacturer's permitted DC power, the maximum voltage, the MPPT voltage range, the input current and the short-circuit current limits.

How much energy is lost between the panels and the meter?

There is no single loss percentage that holds for every project. Soiling, shading, temperature, module mismatch, DC and AC wiring, inverter conversion, clipping and availability are separate assumptions. In a simple annual calculation the percentage losses can be combined by multiplication; in a detailed model, temperature, shading and inverter behaviour are computed at each time step instead.

Should the system be sized to cover annual consumption completely?

Not always. Annual generation matching annual consumption does not mean the bill goes to zero or that the payback period is the best available. The outcome depends on how well generation and consumption coincide in time, on the netting period, on the import price, on the value of exported energy and on the capital cost.

How does shading change the panel layout and the yield?

Shading affects both the usable area and the energy yield. Shading one module in a series-connected string can move the operating point of the whole string; bypass diodes, string layout and MPPT distribution determine how large the effect is. Treat shade as a layout and electrical design question rather than as a fixed annual loss percentage.

Does a battery change the PV design?

It can. A battery raises the self-consumption share by making daytime generation available on site later. The result depends on the battery's capacity, power, round-trip efficiency, operating strategy, the tariff and the load profile. The PV system and the battery have to be designed together against generation and consumption data at the same time resolution.

Back to all posts

How to design a solar PV system: from consumption to panel layout

Start a PV design from consumption or from usable area, then verify the panel layout, the system capacity, the yield and the inverter selection together.

Key takeaways

PV design has no single starting point. A project can begin from a consumption target, from the usable area, from a fixed system capacity, or from a panel count.
When the project starts from consumption, annual yield per kWp gives the first capacity target. When it starts from the area, the panel layout is what produces the panel count and the installed DC capacity.
Panel count is a design input on some projects and an output of the layout on others. Which one it is has to be settled before the design work starts.
Inverter selection does not rest on the DC/AC ratio alone. Open-circuit voltage in the cold, MPPT voltage in the heat, input current and short-circuit current all have to be verified together.
The largest system that is technically buildable is not always the system with the best economics. Self-consumption, tariff, exported energy and future loads have to be modelled together.

An EPC project does not always start in the same place. Some customers arrive with their annual consumption and a savings target. On other projects the usable roof area is what decides the answer. On tender or revision work the system capacity or the panel count may already be fixed. None of these inputs is a finished design on its own.

The job is to compare consumption, site, equipment and connection limits on the same project. Once the panel layout is complete, the panel count and the installed DC capacity become visible. The yield simulation shows how much energy that layout produces, and the consumption and tariff model shows what the result means for the customer.

PV design has no single starting point

Depending on which information is driving the project, three flows are available:

Consumption-led project. Annual consumption, the hourly load profile, the self-consumption target and the tariff set a first system capacity range. Whether that target actually fits on the site is checked afterwards.
Area-led project. The module is chosen and laid out across the usable area. The layout produces the panel count and the installed DC capacity. Yield, consumption and economics follow from it.
Target-led project. A panel count, an installed capacity or an annual generation figure is taken as the design input. Site, string and inverter checks then confirm whether the target is buildable.

Whichever flow you take, inputs and outputs can trade places over the life of the project. In consumption-led work, system capacity is first a target and then a result confirmed by the layout. In area-led work, panel count and installed capacity are outputs of the layout. Detailed system models likewise allow equipment quantities to be entered directly or the array configuration to be calculated^[7].

Turn consumption data into a design target

In a consumption-led project, twelve months of bills give the annual energy requirement but not the timing of that consumption. Hourly or shorter interval meter data reveals how much of the PV generation can be used on site as it is produced. Self-consumption and the share of consumption covered are different results, and both are affected by system capacity and load profile^[8].

Where interval data does not exist, build a representative profile from working hours, shift patterns, heating type and large loads such as electric vehicles. If an EV, a heat pump or a battery is planned, add that near-future consumption to the model as well.

The design target is not annual consumption alone. Where exported energy is worth little, a smaller system with high self-consumption can produce a better economic result. Usable area, structural capacity, connection capacity and budget can all move the target too.

If you start from consumption, calculate the first capacity range

Industry sources use several names for the same quantity: specific yield, specific production or energy yield. This article uses the plainer annual yield per kWp. The value is the energy the system is expected to deliver on the AC side each year for every 1 kWp of DC capacity installed:

Annual yield per kWp = annual AC generation (kWh) ÷ installed DC capacity (kWp)

From a consumption or generation target, the first capacity range follows:

First DC capacity target (kWp) = annual generation target (kWh) ÷ annual yield per kWp (kWh/kWp per year)

This result is not the final project capacity or panel count. It is a design target used before the layout exists.

Annual yield per kWp and irradiance are not the same quantity. Annual irradiation in the plane of the array is expressed in kWh/m² per year, while the yield result is expressed in kWh/kWp per year. The yield result depends on location, orientation, tilt, weather data, temperature, shading, module, inverter and system losses^[5].

PVGIS and PVWatts build an energy yield estimate from these variables^[4]^[1]. For more detailed time-series analysis, models such as SAM can be used^[6]. If the tool you use returns an AC yield that already includes losses, do not deduct the same losses a second time. Record explicitly which output your calculation used.

Once the module is chosen and the layout is built, project capacity is read directly from the layout:

Installed DC capacity of the layout (kWp) = panels placed × module rating (W) ÷ 1,000

The first capacity target derived from consumption and the real capacity the layout produces may not match. Rather than forcing the layout to one panel more or one panel fewer, compare the buildable arrangements against their yield and economic results.

If you start from the roof or site area, build the layout

Usable roof area is smaller than total roof area. Take module dimensions from the current product datasheet, and handle edge setbacks, maintenance walkways, rooftop obstructions, fire access and local practice separately. Until the structural check is complete, do not treat a layout that merely fits geometrically as buildable.

Where there is more than one roof face, model each face as its own sub-array with its own orientation, tilt and shading. Collapsing different generation profiles into a single average orientation distorts the hourly result, especially where east and west faces are used together.

Shading affects both the panel layout and the electrical design. Shading one of the series-connected modules in the same string can change the string current and the operating point, and bypass diodes change the shape of that effect^[9]. So do not treat shade only as a fixed percentage applied at the end of the year. Check the MPPT and string distribution alongside an annual geometric shading analysis. The mechanism is worked through in solar shading and string design, and the method itself is covered under shading analysis.

If the layout changes, update the yield calculation

Rerun the energy simulation whenever the panel count, the distribution across faces or the string structure changes. The generation figure in the proposal has to come from the same design version as the layout the customer is shown.

Select the inverter against the layout

Inverter selection is made up of three separate groups of checks. Verifying only one of them is not enough.

The DC/AC ratio divides the array's DC nameplate power by the inverter's nominal AC power. PVWatts V8 uses 1.2 as its default for preliminary modelling^[2]. That value is not a design requirement. A higher ratio can keep the inverter closer to nominal power more often, and it also affects clipping, thermal load and manufacturer limits. Verify the final ratio against an annual simulation and the equipment datasheet^[12].

The voltage check examines two temperature extremes. At the lowest design temperature, the corrected string open-circuit voltage must not exceed the inverter's maximum permitted DC input voltage. At the highest cell temperature, the string operating voltage must stay inside the range of the relevant MPPT. Check the start-up voltage and the MPPT range at nominal power separately from the manufacturer's datasheet as well.

The current check weighs the number of parallel strings per MPPT, the maximum usable input current, the permitted short-circuit current and the connector limits together. On high-current modern modules in particular, a voltage calculation alone is not sufficient. The current IEC 62548-1 series covers DC design and protection requirements for arrays; for product selection, the manufacturer's current datasheet and design tool remain the reference^[10]^[11].

If any of these checks fails, change the panel count, the string length, the MPPT distribution or the inverter model, and rerun the simulation.

Check the loss boundaries in the yield calculation

Module rating is a DC nameplate value at standard test conditions. What the customer uses or exports is the AC energy that survives hourly irradiance, temperature, shade, electrical losses, inverter conversion, clipping and availability.

The values below are example assumptions chosen only to show the method. They are not typical values and they are not a project recommendation:

Loss item	Used in the example	How to verify it in the project
Soiling	2%	Site conditions and cleaning schedule
Shading	3%	Annual geometric shading model
Cell temperature	6%	Hourly cell temperature and module coefficient
Module mismatch	2%	Module and array assumptions
DC wiring	1.5%	Cable route and voltage drop
Inverter conversion	3%	Efficiency curve and operating points
AC wiring	1%	Cable route and voltage drop
Availability	0.5%	Operations and downtime assumption

These are the worked example's assumptions. On a real project each item is re-established against the site and the equipment.

In this simplified example, which uses fixed annual percentages, the losses are combined by multiplication. The PVWatts technical reference aggregates percentage system losses the same way, while the current PVWatts V8 takes system loss and inverter efficiency as separate inputs^[3]^[2]. In a detailed simulation, temperature, shading and inverter behaviour are computed at the hourly operating point instead. So do not use the simple loss chain as a substitute for a full time-series model.

The performance ratio expresses how much of the solar resource survives real system effects such as module temperature, shading, wiring and the inverter. The simplified relationship is:

Equivalent full sun hours = annual irradiation in the plane of the array ÷ 1 kW/m²

Annual yield per kWp = equivalent full sun hours × performance ratio

Keeping these apart prevents reading irradiation as if it were a yield result, and prevents applying the same losses twice^[5].

How the same project runs from two different starting points

This example is not a site estimate. It shows how a consumption-led flow and a roof-led flow meet on the same project. Assumptions: 4,200 kWh annual consumption, a 4,200 kWh annual generation target, a south-facing roof at 35° tilt, 1,250 kWh/m² annual irradiation in the plane of the array, 440 W modules, and the simplified loss calculation above.

Simplified annual yield chainkWh/kWp per year

Fixed percentages are combined by multiplication in this example. In a real simulation, time-dependent effects are computed at their own operating point.

Sources: Worked example assumptions; IEA PVPS definitions and the PVWatts loss method

The calculation steps:

Yield before losses: 1,250 kWh/m² per year of irradiation in the plane of the array corresponds to 1,250 equivalent full sun hours at a 1 kW/m² reference. For 1 kWp that means a theoretical 1,250 kWh per year before losses.
Simplified performance ratio: multiplying the loss coefficients leaves about 0.824.
Annual yield per kWp: 1,250 × 0.824 = about 1,030 kWh/kWp per year.
Consumption-led flow: 4,200 ÷ 1,030 gives a 4.08 kWp first capacity target. That number is the approximate value the layout will aim at, not the final panel count.
Roof-led flow: a 440 W module is chosen and laid out on the real roof. Nine panels give 3.96 kWp installed, ten panels give 4.4 kWp.
Yield check: at the same yield per kWp, the nine-panel layout produces about 4,079 kWh and the ten-panel layout about 4,532 kWh of AC energy a year.
Area check: at an assumed 1.95 m² per module, ten panels are 19.5 m² of module surface alone. The buildable roof area is set by the real layout, the edge setbacks, the maintenance walkways and the obstructions.
Preliminary inverter selection: for a 4.4 kWp array, a 3.6 kW inverter gives a DC/AC ratio of 1.22 and a 4 kW inverter gives 1.10. The choice is made after the clipping simulation, the manufacturer's power limit, the cold and hot voltages, the MPPT current and the short-circuit current have been verified.

In the example the consumption calculation points to a 4.08 kWp target, while the real layout lands the site on a buildable nine-panel or ten-panel result. Hourly consumption, the self-consumption share, the tariff, the value of exported energy, the connection limit and the customer's future plans decide between the two options. The calculation does not dictate the layout, and the layout is not judged independently of the consumption target.

What changes for a commercial system

The method does not change, but the binding constraints differ by project type. On tariffs with a demand charge, the intervals of highest consumption matter as much as the annual kWh. Grid connection limits and structural capacity can be more decisive than roof area on some projects.

Do not assume commercial consumption is always daytime-heavy. Verify shift patterns, weekends, seasonality and production shutdowns against a measured load profile. Where a battery is planned, design the PV system and the storage together from generation, consumption and tariff data at the same time resolution. The effect of time-based tariffs is worked through in peak shaving and time-of-use battery economics.

Five design mistakes that survive into the proposal

Confusing irradiation with yield per kWp. kWh/m² and kWh/kWp are not the same quantity.
Applying losses twice. Deducting losses again from a model's AC yield result understates the generation.
Not updating the yield after the layout changed. The panel arrangement in the proposal and the energy estimate then rest on different designs.
Choosing the inverter on the DC/AC ratio alone. Voltage, MPPT current, short-circuit current and manufacturer power limits get missed.
Treating annual generation as savings. Self-consumption, the netting period and the price of exported energy go unaccounted for.

How this runs in solarVis

The general design method runs in three connected steps in solarVis 3D design and simulation. Consumption Details defines monthly consumption, the daily load profile and the tariff^[13]. Site Modeling builds the site geometry and the obstructions. Panel Placement then lets you set panel count, annual generation, installed capacity in kWp or the share of consumption covered as the target^[14].

Once the layout is complete, the panel count, the installed capacity, the shading, the strings and the inverter selection are all checked on the same design^[15]. Generation and financial results carry into the solar proposal software from the same project data. The panel arrangement the customer sees and the generation figure in the proposal therefore do not rest on different versions.

Four boundaries are worth stating before you meet them. The yield and loss calculations described above run on an hourly simulation, so very short irradiance and shading transients inside an hour are smoothed into its average rather than resolved individually. The inverter, battery and heat pump sizing offered in Panel Placement is a suggestion built from the layout and the consumption model: solarVis proposes a starting point, and the engineer still confirms the final choice against the manufacturer's datasheet, using the same voltage, current and DC/AC ratio checks worked through earlier in this article. SolarVis does not grant code compliance or structural sign-off, submit a permit or interconnection application, meter production once the system is installed, or settle exported energy with the utility. Those steps stay with the certifying engineer, the installer and the grid operator.