How the numbers are calculated

Your CSV from ESB Networks contains one row every 30 minutes — typically a year of half-hourly Active Import readings in kW. The analyser converts each row into kWh (kW × 0.5 h) and groups the same data four ways: by hour-of-day (the daily shape), by day, by month, and into three day-parts.

The day-parts are night (23:00–07:00), day (07:00–17:00), and evening (17:00–23:00). The split between these three buckets is what tells us how much of your consumption could realistically be shifted onto solar + battery, versus how much will always come from the grid.

If the file covers fewer than seven days the report is flagged as low confidence — a single week is not enough to capture the variation between weekdays, weekends, and weather.

The analyser does not have access to real PV data — it builds a simulated half-hourly generation curve from the system size and orientation you enter. The annual yield is based on a typical Irish kWh-per-kWp figure for the chosen orientation (south is best, east/west least), reduced if your roof tilt is far from the optimal ~35°.

That annual figure is then spread across the year using a fixed monthly distribution (very little in December, peak in June) and across each day as a half-sine between sunrise and sunset, with sunrise/sunset times that vary by month. East/west split arrays use two opposed half-sines so generation is flatter and longer.

The result is a plausible shape, not measured output. It is good enough to sketch self-consumption curves, but it ignores cloud cover, shading, panel degradation, and inverter clipping.

For each candidate capacity (3, 5, 7, 10, 13, 15, 20, 25, 30 kWh), the analyser walks through every half-hour of your data and simulates a battery: when generation exceeds load it charges (capped by a 5 kW charge rate and a round-trip efficiency of 90%), and when load exceeds generation it discharges. Anything that overflows is exported, anything the battery cannot cover is imported. The battery is also depth-of-discharge-limited to 90% of nameplate capacity.

That gives a self-consumption percentage for each candidate size. Plotted, it is a curve with diminishing returns — going from 5 kWh to 10 kWh adds a lot, but going from 15 kWh to 20 kWh barely moves the needle.

The report shows three recommended sizes, each picked as the knee of a different curve so you can weigh strategies side-by-side. The Solar-led size is the knee of the self-consumption curve described above — the smallest battery that captures most of your solar surplus. The Grid-independence size is the knee of the load-shift coverage curve — the smallest battery that covers most of a day's consumption from a single overnight charge. The Best-for-your-tariff size is the knee of an annual € savings curve computed against a specific supplier plan: we re-run the per-half-hour simulation for every candidate size with that tariff's arbitrage rules switched on (charge from the grid during the cheap band, discharge against the day/peak band), and pick the smallest size where the next step up adds less than €15/yr in savings. The dropdown on that card lets you switch tariffs — for flat-rate plans the answer tends to track the solar-led size, but for EV and smart night-rate plans the arbitrage value pushes the optimum noticeably larger.

For each recommended size we then pick the two products from a static catalogue whose nameplate capacity is closest to that number.

Everything above is an annual figure, and an annual figure hides the question most people actually want answered: what happens in December? The seasonal section on the report replays your whole year twice — once with solar alone, once with solar plus the recommended battery — and buckets the result by calendar month, so you can see the share of your usage you would cover in each one.

That share is deliberately a different number from the self-consumption percentage used to size the battery. Self-consumption divides by generation and asks "how much of my solar did I keep?". The seasonal view divides by consumption and asks "how much of my usage did I cover?". The distinction matters most in winter: a dark December scores well on self-consumption, because almost all of the little solar you produce gets used immediately — while covering very little of what you actually consume. Only the second number tells you anything useful about December.

The summer headline covers May to August and the winter headline November to February, with March, April, September and October treated as shoulder months belonging to neither. The gap between those two numbers is usually large, and it is the honest limit of the whole exercise: in an Irish December the constraint is daylight, not storage. A bigger battery cannot store sunshine that never arrived.

Two caveats. Your consumption comes from your real meter readings, but the monthly generation figures are simulated — the annual yield is spread across the year using a fixed seasonal weighting, so treat the month-to-month shape as the expected swing rather than a forecast for any particular month. And because a file that begins mid-month would otherwise show a false dip, every figure is plotted as an average day rather than a monthly total, with short months flagged in the chart tooltip.

The inverter rating is sized to the larger of two numbers: 80% of your PV array size (a typical safety margin — panels rarely deliver their full rating), and the battery's own power rating (so the inverter can keep up with discharge during peaks). That target is then rounded up to the next standard size for your phase: single-phase systems use 3.6 / 4.0 / 5.0 / 6.0 kW, three-phase systems jump to 5–15 kW.

Phase also sets the grid export limit — single-phase installs are capped at 6 kW of export, three-phase at 15 kW under the Irish microgeneration scheme.

A handful of battery brands (Huawei, SolarEdge, Sigenergy, Fox ESS, Alpha ESS) use proprietary protocols and must be paired with an inverter from the same brand — the catalogue filters accordingly. Open-protocol batteries (BYD, Dyness, Pylontech) work with any third-party hybrid inverter. The Tesla Powerwall 3 is an all-in-one — it includes its own 11.5 kW solar inverter, so no separate inverter is recommended.

Alongside self-consumption, each candidate gets a load-shift coverage figure — the share of total household consumption a fully-charged battery could cover each day. The model assumes the battery starts every day full (e.g. topped up overnight on a cheap night-rate window) and discharges into that day's load, capped per-day at the day's usage so partial days do not over-count.

This is the right number to look at if your goal is tariff arbitrage rather than solar self-consumption: charge from the grid at the night-rate price, discharge during the day to avoid the day-rate. The load-shift sizing panel inverts the calculation — pick a coverage target (say 80%) and an off-peak window length, and it suggests the smallest candidate that hits the target while still being chargeable inside the window.

EV charging is included in load-shift coverage by default, but can be excluded — EV charging usually happens overnight from the grid directly, so counting it in "load you want to shift to night-rate" double-counts.

If you're planning changes that will reshape your consumption — adding an EV, swapping an oil or gas boiler for a heat pump, working from home more, or moving load to a night-rate tariff — the report can re-run on a synthetic version of your data with those changes layered on top of the real CSV.

EV charging adds 0.18 kWh/km spread across either a night window (02:00–08:00) or evening (18:00–22:00), capped by the charger's power rating. Heat pumps add winter-weighted heating load divided by the chosen COP; replacing gas/oil also accounts for the boiler efficiency you're removing. WFH adds a flat daytime kWh figure on the chosen number of weekdays. Night-rate shift moves a percentage of evening consumption into the cheap-rate window.

All adjustments operate on the stored half-hourly rows server-side and re-run the full pipeline (analysis, solar sim, battery sizing, inverter sizing) — so every chart and recommendation reflects the adjusted profile, not the original.

The "Compare tariffs" page reruns the full grid-flow simulation against every Irish supplier in the catalogue and shows your estimated annual electricity bill across three scenarios: today (no solar), with your planned solar array but no battery, and with solar plus the recommended battery. Three scenarios let the table surface both the total saving and the marginal contribution of the battery on top of solar.

Each scenario is a per-half-hour simulation — identical to the battery-sizing model — but the battery charging strategy is tuned to the specific tariff being scored. On a flat-rate tariff the battery simply charges from solar surplus and discharges into the load. On a smart (time-of-use) tariff with a meaningful night-to-peak price spread, the battery also tops up from the grid during the cheap night band so it can displace more expensive peak-band imports during the day. This means the same battery can save materially more on a smart tariff than on a standard tariff.

Costs are built from four components: unit-rate spend (kWh in each time band multiplied by the band's per-unit rate), annual standing charge, ongoing supplier discounts (percentage or flat), and microgen export credits (MSS rate applied to surplus generation). Welcome discounts that apply only in the first twelve months are shown as small text on each row — they reduce your first-year bill but are not included in the main "ongoing" cost column, so the savings arithmetic reads cleanly for year two and beyond.

Battery utilisation (kWh/day and cycles/year) is shown for each tariff. A cycle is one full usable capacity discharged; typical home batteries are designed for 250–365 cycles per year. A low cycle count means the battery has more capacity than your evening consumption pattern actually needs — sliding the battery size down until the cycle count approaches 365 is the most cost-efficient choice for that supplier. Smart tariffs with three-hour peak windows generally warrant smaller batteries than flat-rate tariffs because the peak window is easier to cover in full.

Dynamic (wholesale-linked) tariffs such as Electric Ireland's Dynamic Price Plan are deliberately excluded from the comparison. Their variable component is re-priced every 30 minutes against the day-ahead market, so any annual cost figure we computed would depend on a year of historical wholesale prices we don't currently model. Battery owners with flexible loads are the natural fit for those plans — we surface them on the Tariffs page instead, with the caveat that switching is a judgement call rather than a calculable saving.

The payback estimate combines three inputs you provide — solar panel cost, battery cost, and installation cost — with two unit rates: your grid import price (typical Irish residential is €0.30–0.40/kWh) and the SEAI Microgeneration Support Scheme export rate (currently €0.135/kWh).

Annual saving is the sum of avoided grid import (your total consumption minus what the recommended battery still has to pull from the grid, multiplied by your import rate) and MSS export earnings (whatever still spills over after the battery is full, multiplied by the export rate). Simple payback is total cost ÷ annual saving; the 25-year figure is annual saving × 25, with no inflation, tariff rises, or battery degradation modelled.

If you toggle the smart-tariff option, a third stream is added: night-rate arbitrage. The solar simulation already uses some of the battery's yearly cycle budget for self-consumption (`cyclesPerYear` in the candidate). Whatever cycles are left over (365 − cyclesPerYear) are assumed available to charge from the grid at the night rate and discharge during the day. Annual arbitrage saving = spare cycles × battery capacity × 0.9 (depth-of-discharge) × (day rate − night rate), capped by the kWh you still import after solar+battery so we never claim more shifted energy than you actually consume from the grid.

The defaults populate from your system size (€1/Wp for panels, €500/kWh for the battery) but are placeholders — replace them with real installer quotes for a meaningful number.