Own The Peak · Part 1 of 8
I Electrified My House. Then I Discovered What the Data Was Hiding.
In short
- I took the gas out, filled the roof with solar and added a 48 kWh battery.
- The money turned out to be in timing: a few rare half-hours carried most of the export earnings.
- The value was never in the hardware; it was in the data and the control logic.
- 62 %of five months of export earnings came from 205 half-hours
- ~90 %of grid import in the 3 to 9 pm peak after solar is one appliance: the cooker, wired on the grid side
- 8.7 kWhexported into a negative price on 5 April after a restart, about 50 cents lost
Everything in this series comes from one system in one place: a 19.4 kW solar array and a 48 kWh battery in Berwick, Victoria, on Amber’s passthrough pricing. These are my numbers, from my roof and my appliances. Yours will differ. This is personal experience, not advice.
Sixty‑two per cent of what I earned exporting solar over five months came from 205 half‑hours. The other 6,000 or so earned the rest. One half‑hour I didn’t capture, on 27 January, would have been worth more than several weeks of normal export combined. And on Sunday 5 April my system spent about an hour paying the grid roughly fifty cents to take electricity I had generated, because of a Home Assistant restart and a fragile trigger.
That is the story this series is about. I thought the value was in the panels and the battery. It turned out to be in the data and the logic.
From gas to electric
The house is a mid‑1990s suburban brick veneer. Tile roof, concrete slab, two‑car garage. I put the gas in myself in 2006: continuous‑flow hot water I genuinely liked, two‑zone ducted heating, and a gas cooktop I had cooked on happily for years. Almost twenty years later, I was taking it all out.
The renovation forced the call. Gas was not getting cheaper, the ducted system was getting old, and you cannot half‑electrify a brick veneer and expect the economics to land. So I went the whole way: six‑zone reverse‑cycle, heat‑pump hot water, induction cooktop, and a sparky who became very familiar with my driveway. I took the rebates on offer.
That left an all‑electric house and a power bill that was about to get interesting.
Filling the roof
With six zones of reverse‑cycle and a heat pump pulling 2 kW, you either accept the grid bill or generate your own. I chose generate.
The install went on at the end of November 2025: 19.4 kW of Jinko panels across west, north and east faces, feeding a Sigenergy SigenStor with a 20 kW three‑phase inverter and 48 kWh of battery. It is bigger than most households need. I sized it for the hard case, winter with six zones running and no data on what the HVAC would draw. Panels are the cheap part once you are paying for scaffolding, an electrician and an inverter, so I filled the roof.
In summer it is over‑provisioned. On a good day the battery is full by lunchtime and needs nothing from the grid until the next morning. Winter will be different. I will know more in July.
The price you don’t normally see
Most Australian solar owners are on a fixed‑rate plan: a set price for imports, usually high twenties to mid thirties of cents a kilowatt‑hour, and a fixed feed‑in tariff of three to eight cents. It hides the most important fact about household electricity: when you buy matters far more than how much you buy.
I went with Amber because it passes the half‑hourly price through, with the regulated charges added on top. Two things to know, because the rest of the series leans on them.
The first is that the buy price is wholesale plus a large stack of regulated charges. In my postcode the network component alone steps up from roughly ten cents off‑peak to around thirty‑two cents in the peak window between three and nine in the evening. When wholesale is five cents at six o’clock, the delivered price is closer to thirty‑seven. So avoiding peak‑hour import is the most valuable thing the battery can do. A kilowatt‑hour displaced at six in the evening saves forty to fifty‑five cents. A kilowatt‑hour exported at noon is worth, on most days, single digits. The two sides of the meter are not symmetric, and the battery has to be run like it knows that.
The second is that the feed‑in tariff is built differently. There are no network charges on exports. The export side is wholesale only, so the feed‑in price is lumpy and often negative. It goes negative through the middle of the day as a routine feature of the solar window, and selling solar into the grid then costs money. That is why the system is worth operating actively rather than leaving on default settings, and it is the subject of Post 2.
The default settings aren’t doing what you think
The system went live in late November. I spent December doing what any reasonable person does with a new solar installation: checking the app every twenty minutes to see what the battery was doing. This is not a productive use of time, but it is apparently universal.
By January I looked at the billing data. The system was generating more than the house consumed on most days. The battery filled by early afternoon. Grid import was small. Five months in, export credits had out‑earned the import charges by about $57, before the fixed daily charges, and four invoices checked against the raw half‑hourly data match to the cent. More on that in Post 4.
What I had not noticed was that the default firmware mode, “Maximum Self Consumption”, does exactly what it says and nothing else. It discharges into household load whenever there is load and charges from solar whenever there is surplus. It does not know what anything costs. It does not know the kilowatt‑hour it puts into the HVAC at two in the afternoon, at fifteen cents, could be worth fifty if held until six. It does not know that exporting at noon on a sunny Sunday in April will cost money.
On a fixed tariff that is fine. On wholesale passthrough it leaves most of the value on the table.
The thing that actually surprised me
The first surprise was how concentrated the money was. About sixty‑two per cent of my export revenue came from 205 half‑hours in which I exported at a feed‑in price above ten cents. Seven of those were above thirty cents and earned twenty dollars between them.
On 27 January a wholesale spike took the feed‑in price past a dollar eighty a kilowatt‑hour, and it went past me. The automation that should have caught it did not exist yet. That one half‑hour would have been worth a meaningful slice of the whole period’s earnings. I put a number on it in Post 7.
The second surprise was the Belling. About ninety per cent of my post‑solar grid import in the 3 to 9 pm peak window is one appliance: the induction range cooker, still wired on the grid side of the battery from the original install. The battery cannot help because it cannot see the cooker. That is a wiring problem waiting on a sparky, not a control problem. I flag it throughout the series so the numbers stay honest.
The third surprise, and the one that made me take restart resilience seriously, was 5 April. Home Assistant restarted for an unrelated reason. The export curtailment automation, which clamps exports to zero when the feed‑in price goes negative, used a trigger that needs the sensor value to cross a threshold. After a restart the value is restored without crossing anything, so the automation woke up convinced the world was fine. Over the next hour or so the system exported about 8.7 kWh at between negative four and negative eight cents. I paid the grid roughly fifty cents to take my electricity. Fifty cents is small money. The lesson is not. Post 6 has the fix and the gap that remains.
The fourth surprise was the forecast I was paying for. The commercial solar forecasting service I subscribe to overestimates output by 117 to 414 per cent on cloudy days, in my data. Damping it does not fix it. What does is the irradiance sensor on my own weather station. The local signal is the trustworthy one. That is Post 5.
What this series is
Eight posts. One homeowner working out what the system is actually doing and why.
- Post 2 is the feed‑in price distribution and the spike‑capture story.
- Post 3 is the battery as a peaker plant, run against a four‑tier hierarchy: cover household load, cover monthly fixed costs, earn back battery wear, and only then call anything profit.
- Post 4 is the data layer, and how I checked four invoices to the cent.
- Post 5 is forecasting, including why the battery’s charge at seven in the morning is not the cloudy‑day predictor I once thought it was, and what the real local signal is.
- Post 6 is the automation stack, with the 5 April incident as the worked failure. Curtailment netted out at about minus a dollar across the period, so the system did its job.
- Post 7 is the receipts, structured against the hierarchy from Post 3, with the 27 January miss on the table.
- Post 8 is what I am watching, what I would do differently, and what is unknown. Winter, mainly.
I did not set out to learn any of this. I set out to electrify a house and stop paying a gas bill. The hardware is easy: you write a deposit and someone turns up with a ladder. The interesting part, where the system stops being expensive furniture and becomes an asset, is the part the brochures leave out.
That part is the rest of this series.