3 kW Solar Generation: Monthly Estimates for Five Cities

How many units will a 3 kW home solar system produce? A useful answer needs a location, month and set of system assumptions. The same nameplate capacity can produce different electricity in a clear March, a cloudy July and a shaded December afternoon. This guide gives a reproducible model for five cities in Gujarat, Madhya Pradesh and Maharashtra, then shows how to compare it with a real roof and electricity bill.

Direct answer. The archived PVWatts V8 results dated 30 October 2026 estimate 4,618 to 4,726 AC kWh a year for a hypothetical unshaded 3 kW roof system. They cover Surat, Ahmedabad, Rajkot, Indore and Sambhajinagar. That averages roughly 385 to 394 units a month, with lower July and August output. Actual generation depends on the roof, weather and equipment. These figures are model estimates, not a minimum output or savings promise.

What assumptions produced the city estimates?

The archived model uses the National Laboratory of the Rockies PVWatts V8 API documentation and is dated 30 October 2026. One electricity unit means one kWh. The stated inputs for each city are a 3 kW DC system, standard modules and a fixed roof-mounted array (array_type=1). Tilt is 20 degrees and the due-south azimuth is 180 degrees. Other inputs are 14% system losses, a 1.2 DC-to-AC ratio and 96% inverter efficiency. The weather dataset is nsrdb, using the nearest available grid point. The API reports monthly AC output for comparison with metered electricity use.

The location inputs were central-city coordinates rather than a surveyed rooftop: Surat 21.17° N, 72.83° E; Ahmedabad 23.03° N, 72.58° E; Rajkot 22.30° N, 70.80° E; Indore 22.72° N, 75.86° E; and Sambhajinagar 19.88° N, 75.34° E. The selected weather points were about 3.0, 3.8, 7.6, 3.4 and 3.4 km from those inputs, respectively. The archived model inputs and outputs retain the monthly AC values, annual totals and weather-station details. The table below has been checked against those saved outputs. The model date is not the weather year; the stored weather source is NSRDB SUNY METEOSAT IODC typical meteorological year data. The public Surat PVWatts API run illustrates the method.

PVWatts models typical weather and equipment behaviour. It does not inspect a specific roof, water tank, neighbouring building, cable run or cleaning practice. The 14% loss input is an explicit modelling assumption, not a measured loss for a Heaven Green project. The 20-degree south-facing configuration is also a comparison baseline. An east-west roof, steeper pitch or shade needs its own run. The home system-size guide explains how capacity should follow consumption and usable roof, not only a model yield.

Check the model, roof and meter together. First get a transparent model with location and inputs. Second survey the actual roof and shade. Third compare predicted AC units with the home’s import, self-use and export readings. The model gives a starting estimate. The roof and meter decide whether it describes a real installation.

Model input	Value in every city run	Why it matters
DC capacity	3 kW	Holds system size constant
Tilt and orientation	20° and south	Holds module geometry constant
Losses	14% assumed	Covers aggregate model losses
Weather	PVWatts V8 nsrdb	Makes monthly estimates location-specific

The values above are dated model parameters. They should be replaced when an EPC has a surveyed roof layout and current equipment specification.

How many units does a 3 kW system produce each month?

The table below shows rounded AC kWh from five archived PVWatts V8 API runs dated 30 October 2026. The same model settings were used in each city. Each number is a hypothetical model result for a south-facing, unshaded 3 kW array, not an observation from a customer system. Rounding individual months means the displayed monthly values may sum a little differently from the model’s unrounded annual output.

Month	Surat	Ahmedabad	Rajkot	Indore	Sambhajinagar
January	420	412	426	403	423
February	402	387	403	408	410
March	466	450	460	458	449
April	438	426	443	432	428
May	438	435	445	434	427
June	362	373	365	352	355
July	296	317	302	300	315
August	316	297	306	298	320
September	346	368	379	364	365
October	411	420	425	410	406
November	375	373	375	379	396
December	390	385	397	381	406
Modelled annual AC units	4,660	4,642	4,726	4,618	4,699

The Ahmedabad run, Rajkot run, Indore run and Sambhajinagar run show the request parameters. A later public DEMO_KEY replay was rate limited; the table uses the saved outputs, which are available in the archive above. PVWatts documentation defines ac_monthly as monthly AC system output in kWh. A modelled July is not a forecast for every July.

An independent European Commission JRC PVGIS 5.3 model was checked on 30 October 2026 using the same city coordinates, 3 kW capacity, 20-degree tilt and 14% loss input. It uses ERA5 weather for 2005 to 2023 and a building-integrated thermal model. Its annual estimates range from about 4,377 to 4,546 kWh across the five cities. The saved JRC results and parameters show why model choice matters. This is a separate cross-check, not an exact reproduction of PVWatts: weather, thermal treatment and horizon handling differ. Do not combine monthly values from the two models into one forecast.

Across the five runs, March was one of the strongest months and the monsoon period lowered output. City rankings were close and can change with roof geometry or weather-year selection. Do not choose an installer or system size because one city column is a few units higher than another. The state generation guide gives broader geographic context; a rooftop survey is needed for a specific address.

The annual output is useful for a first electricity budget. The monthly pattern is more useful for a household with seasonal use. If air conditioning raises summer consumption or travel reduces it, the value of solar depends on whether those loads occur while the panels generate and how exports are settled.

What does that equal per day, and why is the average misleading?

Divide annual modelled output by 365 to get a calendar-day average. Surat’s 4,660 kWh is about 12.8 units per day; Ahmedabad’s 4,642 is about 12.7; Rajkot’s 4,726 is about 12.9; Indore’s 4,618 is about 12.7; and Sambhajinagar’s 4,699 is about 12.9. These are averages of the stated 2026 PVWatts runs, not daily minimums or targets. An actual clear day can exceed the average, while a rain-heavy day can be far below it.

Monthly averages show the variation better. Surat’s modelled March output of 466 units is about 15.0 units per March day, while its July output of 296 units is about 9.5 units per July day. The arithmetic uses 31 days for both months and rounded monthly totals. It does not imply a real system will make 15 units each day in March. Weather, cloud, heat and maintenance create day-to-day spread that a monthly table hides.

The common claim that a 3 kW system always makes a fixed number of units daily is too rigid. A nameplate is measured under standardized test conditions. A roof receives changing sun and temperature. The inverter also has operating limits and electrical losses. PVWatts simulates those effects, but only for the specified system. If a roof is shaded by a water tank from 10 am to noon, a generic city average misses a major constraint.

The 3 kW system guide explains equipment and space considerations. When comparing quotes, ask each EPC for a month-by-month AC production estimate, not just a single annual number. Check whether the assumed tilt, shade, losses and inverter capacity match the proposed layout. A forecast with no input sheet is difficult to audit after installation.

Illustrative home example: assume a household uses 420 units in March and 300 in July. A Surat model run shows 466 and 296 solar units in those months. It would be incorrect to subtract the two numbers directly from the bill and call the remaining amount payable. Some solar is used instantly; some may export; some household use occurs at night; and fixed charges remain. The example labels the bills as hypothetical and exists only to show why timing matters.

The household should use the daily number as a rough sense of scale. For sizing and bill planning, monthly and hourly patterns are more informative.

How should you use the five city numbers when sizing?

The five city columns are useful for planning a first roof survey, but they are not a ranking of where solar is worthwhile. Annual modelled outputs are close because the same equipment and settings were held constant. A household’s daytime use, export settlement and roof suitability can create a larger difference in bill value than the small gap between these city model totals. Start with twelve months of actual consumption, then choose the nearest weather run as a reference.

Do not size the array to the highest month. March output in the table may exceed a home’s March consumption while a lower-output month still requires imports. The system produces electricity by hour, and the household uses it by hour. A proposal should estimate how much generated energy meets load directly, how much is exported, and how much grid energy remains. If the home is empty on weekdays but uses more electricity after sunset, its direct-use share may be lower than a similar home with daytime occupancy.

The next check is usable roof area. An EPC should draw modules around shade, access and equipment, then confirm structural suitability. If the roof cannot safely hold the assumed 3 kW layout, do not force the city model into the proposal. A smaller, unshaded array may be more useful than a larger one placed behind a water tank. If there is abundant roof but limited sanctioned load or export permission, the electrical approval may constrain capacity before area does.

Use two cases in the proposal: one with the intended system size and one smaller size that closely follows daytime load. Keep the weather and tariff assumptions identical. Compare expected direct use, export and remaining bill charges. The solar calculator can start the discussion, while a site-specific design should settle it. This method also explains why two houses on the same street may reasonably choose different capacities.

Finally, retain the model as a dated baseline. Weather datasets and inverter specifications can change. If the EPC changes modules or mounting after the initial quote, request a revised forecast. The homeowner should be able to tell which estimate describes the built plant when generation is reviewed next year.

Why can the real roof produce less than the model?

Shade is often the first reason. A parapet, adjacent building, water tank, dish or tree can shade modules during valuable sun hours. The effect depends on which modules are shaded, inverter design and the time of year. A survey should show shade by hour and season. The PVWatts runs above do not include site-specific obstructions. A shaded roof needs a separate loss input or more detailed model, not a promise that the city table applies unchanged.

Orientation and tilt are next. The table assumes 20 degrees and due south. Many homes have flat terraces where mounting can be chosen, but setbacks and terrace use may force another layout. A pitched roof follows its existing direction unless it is modified. A west-facing array may shift output later in the day even when its annual units differ. The right design is one that is structurally safe, buildable and useful to the household, not simply the one that maximizes a model total.

Dust and cleaning practices matter, particularly near roads, construction or industrial areas. Monsoon rain may clean modules at times while clouds lower irradiation. Heat reduces module electrical output relative to cooler conditions at the same sunlight level. Inverter conversion, cable resistance, mismatch and occasional faults also affect AC units. The model’s 14% aggregate loss assumption stands in for several effects, but a real site can be above or below it.

The roof and structure must be checked before adding panels. A system moved away from a weak or shaded area can have different output from the first sketch. The roof inspection guide and mounting-structure guide help frame those checks. A safe, maintainable system should be the starting point for an energy estimate.

Equipment selection can change the result, but brand names alone do not explain a difference. Ask for module data, inverter DC-to-AC ratio, string layout, clipping estimate and warranty conditions. A proposal that assumes premium modules in its yield model but quotes another bill of materials is inconsistent. The solar module selection guide gives a buyer checklist.

Weather also varies from year to year. A typical-year model is a planning tool; one rainy season can differ from the data pattern it represents.

How do generated units differ from bill units saved?

Solar generation, household consumption, grid import and export are four different quantities. The inverter reports how much AC electricity the system produces. The home may use part of that production immediately. If production exceeds load at that moment, the excess can flow to the grid under the approved metering arrangement. At night or when solar is low, the home imports electricity. A single annual generation figure does not tell you the split.

Here is a hypothetical two-hour example. Each unit means one kWh. In the first hour, solar makes 3 units. The home uses 2 of them and exports 1. In the second hour, after sunset, the home imports 2 units. Total home use is 4 units. Solar made 3, of which 2 were used at home. The meter records 1 unit of export and 2 units of import. Export credit follows the approved billing rule. This simple example leaves out meter-interval details.

The net-metering glossary explains the settlement mechanism. The electricity-bill-slabs guide shows why tariff structure also matters. Even if a household’s net energy charge falls sharply, fixed charges, duties or other bill lines may remain. The why a solar bill is not always zero guide addresses that common expectation.

Do not multiply annual generation by the highest slab rate and call that guaranteed savings. Some units are used on site and may offset a different marginal charge. Exported units may receive a different treatment, and there can be settlement limits. The electricity connection, state rule, licensee and commissioning date all matter. A quote should state those assumptions and show an example bill, not only annual kWh.

The PVWatts model also outputs AC production, not the home’s meter data. An inverter may show generation even while the bill reflects import and export separately. To reconcile them, read both the inverter and the bidirectional meter for the same period. A difference is not automatically a fault: direct use is invisible to the import register because it never came from the grid.

The most useful question for a homeowner is how much modelled generation overlaps daytime load. A household with people at home during the day may directly use more than one that is empty until evening. The correct system size should account for both patterns and the approved export rule.

How can you check your own system against the estimate?

Keep the proposal’s original monthly AC forecast, its assumptions and the commissioning date. Then record the inverter’s monthly generation reading. Compare the same calendar months, not a full month against a partial first month. Note any outages, cleaning, unusual shade from construction or equipment faults. One low week during heavy rain is not enough to diagnose a problem; a persistent difference across comparable periods deserves investigation.

The generation-report guide explains inverter data. The monitoring guide covers routine owner checks. Also record when the system was available to generate. Note any time the grid caused the inverter to reduce output or disconnect. Check alarms, breaker trips and lost app connections before blaming the weather model.

Read the bidirectional meter’s cumulative import and export registers at the same dates as the bill. The inverter’s generation should not be expected to equal export, because much generation may be consumed inside the home. If monthly generation looks normal but the bill remains high, check nighttime consumption, seasonal appliances, meter settlement and fixed charges. The high solar bill diagnosis gives a broader checklist.

Use a measured baseline when possible. Twelve pre-solar bills show how consumption changes through hot months and holidays. After installation, compare similar months and account for new appliances or occupancy changes. A new air conditioner can raise household consumption even when the solar system is performing as modelled. A tariff change can alter rupee savings without changing generated kWh.

Ask the installer for the final as-built system capacity and inverter details, not only the initial quote. If modules were moved or a string was changed during installation, the original PVWatts inputs may no longer describe the built array. Keep a revised model and handover record. The homeowner can then separate design changes from performance changes.

The model is most useful when its inputs can be replaced with the as-built roof. A transparent, repeatable estimate gives the homeowner a better reference than a single unqualified daily-unit claim.

What should you ask before buying a 3 kW system?

Ask whether 3 kW fits the home’s sanctioned load, roof, annual use and approved metering route. A household with a small daytime load may export a large share of a 3 kW system’s output. Another household with daytime cooling and pumping may use more directly. The sanctioned-load glossary and solar calculator provide useful starting points, but the proposal should use the actual bill and site.

Request a roof layout with module count, orientation, tilt, setbacks, water-tank and shade map, cable route and maintenance access. Ask for the PVWatts or equivalent modelling inputs. If the forecast is materially higher than the city table, the EPC should explain why, such as a different module capacity, weather dataset or loss assumption. If it is lower, there may be legitimate shade or roof geometry. The key is an auditable explanation.

Check the bill model on its own. It should show generation, direct use, export, import and charges that remain. An annual energy total cannot guarantee a monthly rupee saving. Ask which regulator and DISCOM rule sets the export value. Check whether approval needs a change to the sanctioned load. For PM Surya Ghar, check eligibility and the application steps on the MNRE PM Surya Ghar portal. This guide does not assume that every home is entitled to a subsidy or a set amount.

Read the equipment schedule and warranty documents. Confirm module model, inverter model, mounting type, electrical protection, monitoring and service responsibilities in writing. Ask how the system will be tested and what readings will be handed over. The solar quote guide helps compare proposals, while the solar commissioning guide explains the handover stage.

Use a simple decision table for proposals:

Question	A reviewable answer includes
Why 3 kW?	Bills, roof and export assumptions
How many units?	Monthly AC estimate and model inputs
What is the bill effect?	Direct use, export and remaining charges
How is it checked?	As-built data and generation report

If the EPC cannot explain why the proposed system is 3 kW instead of another size, pause the capacity decision. The goal is a system matched to the roof and household load, with a forecast that can be checked later.

Plan a home site visit: Share your electricity bills and a roof photo through the residential solar enquiry. Ask for a monthly generation model, shade review and bill assumptions for your actual connection.

What to send with a home solar enquiry

Bring twelve months of bills if you have them. Add the consumer number, a roof plan or clear photos. Note water tanks, nearby trees and future terrace use. The proposal should state the system size, usable roof and estimated monthly AC units. It should explain how that estimate was made. Heaven Green Energy files PM Surya Ghar and DISCOM paperwork for its customers. Official approval still depends on eligibility and the homeowner’s documents.

The relevant routes are:

• Residential solar for the primary home site-visit enquiry.
• Solar calculator for an initial size exploration.
• Home system-size guide for the sizing method.

The model should be updated after the site survey. If the roof faces a different direction, has shade or needs a different mounting angle, those inputs should replace the generic city run. A bill model should use the household’s tariff and actual day-versus-night use. It should leave fixed charges visible and separate direct use from export.

Ask the proposal to attach its model inputs and the roof layout. That lets the homeowner check whether the published city table was used as a starting point or copied as a promise. A measured shade adjustment and a named weather source make later monitoring more useful. If an approval changes the final system size, update both the generation and bill estimates before installation.

Use this sequence:

1. Collect bills. Note monthly units and seasonal changes.
2. Survey the roof. Measure usable area and shade by time.
3. Run the model. Record location, capacity, tilt, losses and weather source.
4. Model bill flows. Separate generation, direct use, export and import.
5. Check approvals. Confirm sanctioned load and metering route.
6. Keep the baseline. Compare as-built generation with the saved monthly forecast.

After commissioning, use inverter data and meter readings together. A modelled month is a reference, not a promise. The residential solar route is the main place to start with a site-specific estimate.

Frequently asked questions

How many units does a 3 kW solar system make per day?

The five PVWatts V8 city runs dated 30 October 2026 average about 12.7 to 12.9 AC kWh per calendar day over a year. A real day can be much higher or lower. Roof shade, orientation, weather and equipment change the result, so use a monthly site-specific model for planning.

How many units will it make in a month?

In the stated model, monthly output changes by city and season. Surat ranges from 296 rounded units in July to 466 in March. Other city columns show a similar seasonal pattern. These are typical-weather model outputs for an assumed unshaded array, not guaranteed monthly meter readings.

Why are July and August lower in the table?

The PVWatts weather data and modelled solar resource produce lower output in the monsoon period for these locations. Actual rainfall and cloud conditions vary by year. Cleaning, shade and grid interruptions can also affect a real system, so one low month should be interpreted with its operating record.

Does 4,600 generated units mean 4,600 bill units saved?

No. Some generation is used directly and some may export. The home still imports at night or during low solar output, and fixed or other bill charges can remain. The export settlement rule and household timing determine the bill effect. Keep generation and billing calculations separate.

Can I use the same estimate for an east-facing roof?

The table assumes south-facing modules at 20 degrees. An east-facing roof needs a new run with its actual azimuth and tilt, plus measured shade. Its output may have a different daily profile even if annual units are close. Ask the EPC for the input sheet and loss diagram.

How do I verify my system after installation?

Compare the inverter’s monthly AC output with the proposal for that month. Read the grid meter’s import and export registers too. Note outages, cleaning, shade and alarms. Generation will not equal export: the home uses some solar power directly.

The figures above are model outputs from one dated parameter set. A site survey and approved metering arrangement are needed for a proposal to a particular household.