Designing a solar system isn’t complicated, but there are enough decisions to make that getting one wrong can cost you years of suboptimal performance. This is the exact process I use — the same one I applied to my own system in Whitchurch, Cardiff.

Step 1 — Understand the roof

Everything starts with the roof. The questions that matter:

Direction (azimuth): The best is due south (180°). Each degree away from south reduces output slightly. South-west and south-east are fine. East or west arrays work but produce around 20–30% less annually and generate at different times of day — which isn’t always bad if you have specific afternoon or morning loads.

Pitch (tilt angle): The optimal for the UK is roughly 35–40°. Most standard roofs are 30–45°, which is close enough. Flat roofs need pitched mounting frames. Steep roofs (50°+) lose some output but not dramatically.

Shading: This is the killer. A chimney or tree that shades even one panel for part of the day can drag the whole string’s output down to the shaded panel’s level (with string inverters). I map shadows across the roof at winter and summer solstice — worst case and best case.

Available area: A 455W panel is roughly 1.72m × 1.13m. Measure the usable roof area, subtract any no-go zones (velux windows, soil pipes, ridge, eaves clearance), and you have your panel count limit.

Step 2 — Analyse the household’s usage

The roof tells you what you can generate. The usage data tells you what you need.

I ask for:

12 months of half-hourly smart meter data (downloadable from Octopus, British Gas, OVO etc. — usually a CSV)
Or at minimum the annual kWh and quarterly bills

From the half-hourly data I extract:

Total annual consumption
Peak demand times (when is the house using most electricity?)
Summer vs winter pattern
Overnight baseline (always-on loads)

A house using 4,000 kWh/yr with heavy afternoon usage benefits differently from one using 3,000 kWh/yr but mostly in the evenings. The usage shape matters as much as the total.

Step 3 — Size the solar array

With roof capacity and usage data in hand:

Array size = min(roof capacity, financial optimum)

For most UK homes the financial optimum is somewhere between 3–6kWp. Bigger arrays hit diminishing returns once you’re generating more than you can use or store.

Rule of thumb: size the array to generate roughly the same as your annual consumption. With a battery and Octopus Agile, self-consumption rates of 70–80% are achievable even on a well-sized system.

Example: my own system

Annual usage: ~3,000 kWh/yr (before solar)
Roof space: 8 panels max south-facing
Array: 8 × 455W = 3.64kWp
Estimated annual generation: ~3,200 kWh/yr in Cardiff
Self-consumption with 10.3kWh battery: ~72%

That’s a well-matched system. Going to 12 panels would generate more but export the surplus at the lower SEG rate rather than self-consuming at the full avoided-rate value.

Step 4 — Choose the inverter and size it correctly

The inverter is the brain of the system. For battery systems you want a hybrid inverter — it handles solar, battery, and grid all in one unit.

DC:AC ratio (oversizing): Inverter manufacturers allow you to connect more solar capacity than the inverter’s rated output — typically up to 133%. So a 3kW inverter can handle up to ~4kWp of panels. This is deliberate: panels rarely hit rated output in UK conditions, so slight oversizing costs nothing and captures more energy at low light levels.

My setup: 3.64kWp panels on a 3kW inverter (121% ratio — conservative and clean).

MPPT inputs: This is where it gets interesting. MPPT (Maximum Power Point Tracking) inputs allow the inverter to optimise different strings independently. A two-MPPT inverter can handle:

South array on MPPT1
East or west array on MPPT2

…each optimised for their own sun angle. This is how I’m adding 4 west-facing panels to my existing system using my spare MPPT2 — no new inverter needed.

String sizing for each MPPT: Each string must stay within the MPPT’s voltage and current limits.

For the Fox ESS H1 3kW (my inverter):

Max string voltage: 600V Voc
MPPT operating range: 70–550V
Max current per MPPT: 15A

For AIKO 455W panels (Voc 44.8V, Isc 13.2A):

Minimum 2 panels in series (89.6V > 70V minimum)
Maximum 13 panels in series (583V < 600V limit)
4 panels in series on MPPT2: 179.2V Voc, 13.2A — perfect

Always check the actual panel Voc at minimum temperature too (panels produce higher voltage in cold conditions). Use the temperature coefficient to calculate worst-case cold-day voltage.

Step 5 — Size the battery

Battery sizing depends on two things: what you’re trying to store, and what tariff you’re on.

Solar storage only (no smart tariff): Size to store a typical summer day’s surplus. For a 3kWp system generating 15–20 kWh on a good summer day and a house using 8–10 kWh/day, you need 5–8 kWh of usable battery capacity.

With Octopus Agile + Predbat: The battery serves two purposes — solar storage AND cheap overnight charging. Here you want more capacity. 10–15 kWh is ideal. You fill it with cheap overnight Agile electricity (often 5–8p) and discharge it through the expensive daytime peak.

My Fox ESS 10.3kWh usable capacity has been the right size for a 3-bed semi. It’s rarely full for long in summer (we’re exporting surplus) and it covers our evening needs in winter.

Step 6 — Tariff recommendation

The tariff is not an afterthought — it can add 30–40% extra value to the same hardware.

With a battery: Octopus Agile is almost always the right answer. Half-hourly variable pricing means Predbat can time charge and discharge decisions to the cheapest and most expensive slots. On my system, my average effective rate is around 12p vs 28p on a flat tariff.

Without a battery: Octopus Go (cheap overnight window) or a standard variable tariff. Agile without a battery is harder to benefit from unless you can manually shift loads.

With an EV: Intelligent Go gives you a cheap overnight window plus smart EV scheduling — great if your car supports it.

I always run the numbers for the specific household before recommending. Usage pattern matters enormously.

Step 7 — Smart home integration

This is where the system moves from good to excellent.

Home Assistant + Predbat is the combination I recommend:

Predbat reads Agile prices 48 hours ahead
Reads Solcast solar generation forecast
Plans every battery charge and discharge slot automatically
Handles grid export when prices go high (if your inverter supports it)

The Fox ESS H1 integrates with Home Assistant via the Predbat add-on with minimal setup. Once it’s running it’s hands-off.

I’ll be blunt here, because after designing this many systems I’ve seen what performs well in the real world and what falls apart the moment something goes wrong.

🏆 Tesla Powerwall — number one for customer care

If after-sales support matters to you — and it should, because you’re going to own this system for 15–20 years — Tesla is my top recommendation. Their customer care is genuinely excellent. When something goes wrong you can actually reach someone who can fix it. Warranty claims are handled properly. The app is polished and gives you clear data.

The Powerwall is a proven, well-engineered product with a strong track record. It’s not the cheapest option but the peace of mind is worth the premium for many customers.

✅ GoodWe — best all-rounder, especially the ERA range

For most homeowners I design systems for, GoodWe is my go-to recommendation. The GoodWe 8kW ERA all-in-one is particularly good — hybrid inverter, battery, and management system in a single unit. Clean installation, solid hardware, good Home Assistant integration, and a capable app.

GoodWe’s ERA range hits a sweet spot between price, performance, and reliability. The 8kW model handles most family homes comfortably, supports both solar input and Agile-style smart charging, and the customer support is responsive compared to many rivals.

If you’re looking at a system with a generous battery and a capable inverter without going full Tesla pricing, GoodWe ERA is where I’d point you.

I have a Fox ESS H1 myself — it works, and with Home Assistant and Predbat it does the job. But I’d be dishonest if I recommended it to customers without caveats.

The app is poor. Monitoring is limited and unreliable. You’re essentially flying blind without setting up Home Assistant yourself — which most homeowners won’t do.

Customer care doesn’t exist in any meaningful way. If something goes wrong outside of what your installer can fix, you’re on your own. The UK support structure is thin and response times are poor.

My Fox setup works because I’ve taken control of monitoring and automation through Home Assistant. That’s not a realistic expectation for most people. For customers who want a system that just works and has proper backup if it doesn’t — Fox isn’t the right choice.

This is my honest opinion based on experience. Every system is different and I’ll always give you a recommendation based on your specific situation, budget, and what matters most to you.

What I offer

If you want help with your system design, get in touch — it’s free.

System design:

Roof assessment (send me photos + orientation + pitch)
Usage analysis (send me your smart meter CSV or annual kWh)
Panel count and placement recommendation
Inverter specification and string sizing
Battery size recommendation
Tariff recommendation
Estimated annual generation, self-consumption, and payback period

This isn’t a generic calculator output — it’s a proper design based on your actual data. And once you have a design, I can put you in touch with The Solar House — the South Wales team who fitted my own system. They know Fox ESS, GoodWe and Predbat setups inside out.

Get in touch to discuss your project →

How I design a home solar system — the process I use for every project