Optimum photovoltaic system sizing

Scope

This is mostly a rough guide on sizing a PV system without breaking the bank. There are many variables and vendors, component pricing varies around the world, geographical position matters as well, electricity pricing makes a huge difference.

The prices are quoted for 2024/2025, currency is avoided as EUR, USD and GBP are roughly the same, local taxes might have a bigger impact than exchange rate.

The goal is to have the best price-to-performance ratio as possible without the need to go off-grid. It also means that small outages (<24h) can be mitigated. I am presenting the calculations for my needs but you can scale up for your home.

Preparation

First you need to determine your house base load. This means the hourly average for 24h, in my case this is around 130W/h, so fridge, servers, cameras, central heating, around 3-5kWh per day when nobody is home. A laundry load takes about 2kWh, one dishwasher cycle 2kWh, heat-pump dryer again around 2kWh. Basic cooking with electricity around 1-2kWh per meal for induction stovetop, 2kWh for circulating oven. So the house needs can vary between 3kWh and maybe 12, per day.

It's best if you plug all the numbers into a spreadsheet, including electricity price, which in my case is about 0.2 (EUR/USD/GBP) per kWh. This will help you later into figuring out in how much time the system pays for itself.

You should already replace demanding consumers with higher-efficiency ones: using laptops instead of desktops whenever possible, induction cooktop instead of ceramic, AC for heating (above 3C) instead of space heaters.

Inverter

This is arguably the most important piece of the system. I would advise it's better to spend more here and wire your entire house through a true-hybrid inverter. I'm using a 6kW single-phase Deye inverter, my house wiring and grid connection is also rated for 6. So the inverter needs to be able to handle the full house load. For 3-phase consumers, it's best to decide on an inverter that can handle unbalanced lines, so that energy is always used from the solar or battery whenever possible. Deye is again a good option here.

Not least, I want my system to handle grid outages and perform the switch to battery in an instant, Huawei for example requires a special (costly) unit for this. This feature replaces the need for a UPS or line conditioning unit.

As it stands right now, low-voltage (~48V) batteries are about three times cheaper than high-voltage (~300V) ones. Huawei, Fronius and other inverter vendors can only handle HV batteries.

If your available roof surface has one angle to the South, you can choose any inverter. If your only available roof space is facing East and West, you need to choose one with 2 MPPT controllers. If you have all three orientations available and want to take advantage of them, you should choose something that has 2+1 MPPT(s).

Each MPPT controller on the inverter can take a "string" of photovoltaic panels. Reading the datasheet, your maximum string size (i.e. panels wired in series) should not exceed the maximum voltage. Each panel manufacturer lists the maximum no-load voltage, e.g. 42V. If you have 10 panels in series per string, the no-load voltage will be 420V. The inverter should list a higher number (at least 450V) for the maximum string voltage. The amperage does not matter, you can put 8kW of panels per string on a 6kW inverter, as long as it does not exceed the maximum voltage.

Panels

As discussed above, the panels should fit below the maximum total voltage. You can wire them in series, or two strings in parallel, but that's already going a bit overboard. Older-style panels that you can buy cheap in bulk are not worth your time, mostly the ones below 300W. You should not mix-and-match different panel makes on a string, but you can. Newer panels are almost always more efficient. Unless you have a special need, you shouldn't spend a lot of money on specialized panels (bi-facial, peroskovite, ...). The most cost-effective panels as of right now are the monocrystalline-type in the range 450-550W. Best to shop around and figure out the best price per watt.

Mounting structure

This can vary for each roof type, If you have a ceramic-shingle roof, those are one of the most expensive but trouble-free systems. Expect to pay around 200 for each 4kW of installed panels for materials alone. The more panels you can fit per string, the better.

For metallic sheet roofs it's much cheaper but the big downside is the need to make absolutely sure that every roof hole is completely watertight. A small leak will lead to having to replace the entire roof structure.

For flat roofs, there are special structures that can be mounted at any angle. Without real experience, I don't have any pointers on that option.

Orientation and angle

By far, the best orientation is South. The East+West orientation shaves the mid-day peak and provides a better production average, at least in summer. In winter, the Sun mostly rises at SE (9 o'clock) and sets at SW (16 o'clock), so East+West does not provide a lot of gain.

The angle is dictated by your position but considering the global population density you should aim for around a 35 degree vertical angle. This is dictated mostly by your roof but it's better to have a smaller area that's steeper than a larger one that's more flat. Case in point, in winter, a 3100W string at 35 degrees elevation produces 15-40% more power than a 3800W string at 13 degrees.

Number of panels

As written above, as many as you can fit without exceeding the maximum voltage. More on the South face is better for year-round performance. As many as you can fit on steep inclines.

More power isn't always better, unless you have a way to sell it or store it. A 3kW string can produce 25kWh per day in the middle of summer. The same string can probably make 10kWh on a sunny winter day.

However, you can compensate for dark and overcast days, especially important during the winter. Panels are pretty cheap (around 60 for a 480W unit right now, structure around 20 perhaps) compared to the expense of the inverter and setting everything up. You WILL have excess power in the summer and not enough in the winter.

Battery

This is one of the most expensive components right now and might not even be worth purchasing, depending on your grid pricing. But you can also use it if you have a day-night tariff (I don't) and as a UPS. With the current contract, I am getting around 1/4 of the price when I sell power to the grid versus when I buy it, you should expect this scenario or worse (0 selling price) as more people install PV systems.

I aim for 12-24h of grid autonomy so in my case 7kWh seemed like a good compromise. The hybrid 6kW inverter uses 60-70W/h by itself so already 1.4-1.7kWh per day is wasted as heat. Most hybrid inverters will use ~1% of their kW capacity as idle wattage.

The only chemistry worth considering right now is LiFePo4 (LFP). Forget about LiIon/LiPo (NMC) as those have much shorter endurance and are prone to fires. Lead-Acid (Pb) ones also have a very short life and hover around the same price. AGM, Solar or other names are mostly just another marketing gimmick to sell you lead batteries. LiPo has a usable life of 500-1500 cycles, Lead 400-800, LiFePo4 around 6000 cycles.

You can buy ready-made LV (48V) batteries for a price but you can also build them for a fraction of the price. Ready-made ones start from 600 for a 5kWh battery. You can build a 14kWh one for 700-1000. 

Regardless of the chemistry, all DIY batteries need a BMS, fuses, decent housing and monitoring.

I won't go into the details of building a battery for now but it's quite easy if you have some basic handyman skills. The most important thing is to ensure that all connections are secure as very high currents are involved (100A) and imperfect contacts are known to start fires.

Wiring

The inverter should be as close to the place where the grid wiring enters the house. This is because, in the scenario described above, all the house loads will pass through the inverter. A longer distance will mean either thicker (more expensive) or longer (expensive, higher losses) wires. The panels should also be close to the inverter. The battery should be right next to the inverter. So, in order of importance, for a modest house, the items rank like this:

• inverter: maximum 2 meters away (125A current, 35sqmm section)
• house: maximum 20 meters away (32A current, 6-8sqmm section)
• panels: maximum 35 meters away (18A current, 4-8sqmm section)

If the battery is too far away the inverter will believe the battery is empty when a load is present and will draw from the grid or even shut down for a moment. If the house is too far away, the consumers will see a low voltage when any big consumer starts up, for example lights will dim when the oven starts. If the panels are too far away, you will lose PV efficiency via heat losses.

All circuits have to be protected via breakers. There would be one breaker box for the house and inverter, at the point of entry, another box for the same purpose near the inverter (AC), a box for PV panels (DC) and at least one fuse + DC switch for the battery. There needs to be proper grounding, electrical (storm) discharge protection, DC disconnect switches, bypass breakers (MTS) but that's getting beyond the scope.

Monitoring and automation

Not to be neglected is a way to monitor the inverter which will give insight into all the other variables. This way you can also know how much your house is using, how much your panels are producing, what's the battery state.

The next level is to be able to automate consumers. Since PV power is highly variable and has a low resell value, you need to be able to use as much as possible from it. So you can start the AC in the summer in the morning and shut it off at dusk, or start the pool heater when there is excess production or some space heater in the winter.

It's good to have some remote control over the house, so, even if you are away, you can turn consumers on and off. The washing and drying can be timed to start just after the sunrise, with some delay between them.

Return-on-investment

For my initial system, which is 3kW of panels, 6kW inverter, 7kWh of battery, the total cost was around 3100. In one year, the power consumed from it was 3500kWh which, for a price of 0.2/kWh, means 700 currency units saved per year. So the ROI duration is around 4.5 years. This was a perfectly sized system for my house with very little excess power but I had to pay for additional consumption.

The same system was upgraded to 7kW of panels and 21kWh of battery for an additional cost of 1500. While I don't have a real estimate, I expect the ROI period to go above 8 years. Although the spreadsheet says it's 5 years (assuming a 4500kWh yearly production) there will be a lot unused excess power.

Without a battery, you can expect at most 25-50% usage of a decently sized PV system. A battery raises that usage to 70-90%. Unless you can babysit all your consumers and predict the sunshine, those are the realistic numbers.

Final thoughts

As of January 2025, this post wasn't really thoroughly prepared, so I might make changes and additions according to the feedback. I realize it's mostly dry text but it will eventually be amended with some graphs and links to studies.

South, mid-winter, 3100W panels at 35 degree elevation in blue, versus 3800W panels at 13 degrees elevation in yellow:

Yearly production with 3.1kWp: