Talking Myself Out Of A New Solar Battery
Our solar battery wasn't a great financial choice: although it yields _some_ savings, the steep upfront cost means that it will probably never break even.
Recently, however, I found out that the Pylontech US2000C modules that it's made up of have come down in price quite significantly - I could add `2.4 kWh` of battery capacity for a little under £500.
That's about a third of the price of the originals, which _feels_ like an absolute bargain.
Such a significant reduction in price would mean that a new module would break even much sooner. After that, any savings that it generated would go towards the amortisation of its more expensive counterparts.
So, I sat down and ran some numbers to see if it would make sense.
* * *
### Can We Fill The Capacity?
The first question to answer was: if we had the capacity, could we actually fill it?
There'd not be a lot of point in paying to expand capacity only for it to sit empty.
* * *
#### Grid Charging
The battery is currently1 set to charge at 25 amps.
At that rate, it goes from 30%2 to full in about 4 hours:
If we added a fourth module, we'd need a third longer: 80 minutes.
Our battery is full by 04:00 and peak pricing starts around 06:00, so there'd be time for the charge to complete before the morning peak starts.
* * *
#### Solar Charging
For solar charging, it's largely just a question of whether we export enough energy that we could fill the module instead.
That's somewhat more marginal, but is generally achieved over summer:
This, incidentally, is the _non financial_ benefit of the battery: whatever break-even looks like, the battery enables us to consume more of the clean, green energy coming from our panels rather than pulling from the, still mainly fossil fuel powered, grid.
* * *
### Calculating Break-Even
So, we've established that we could fill the battery, whether charging from the grid or the panels.
The next questions to answer are:
* When would the new module break-even?
* How much (if at all) a new module improves break-even of the originals?
* * *
#### Current Weekly Savings
To answer those questions, we need to know what savings the battery currently unlocks.
I queried my InfluxDB instance, limiting the data to this year (because savings in prior years were inflated by savings sessions and Powerups):
select
mean(val) as "mean",
median(val) as "median",
max(val) as "max",
min(val) as "min"
FROM (SELECT
difference(last("battery_total_generated_savings"))/100 AS "val"
FROM Systemstats.rp_720d.solar_amortisation_new
WHERE time > '2026-01-01T00:00:00Z'
GROUP BY time(1w)
)
Mean | Median | Max | Min
---|---|---|---
£4.29 | £4.23 | £7.86 | £0.72
Those savings are achieved with 3 modules, so we can take the median (the range between min and max is large, so median feels safer than mean) and turn that into savings per module:
£4.23 / 3 = £1.41
Adding a module _should_ mean increasing the median weekly saving by £1.41, to £5.64.
Technically, we also need to account for inflation in energy prices but the future is quite hard to model and we'd need to predict:
* The average rate of inflation
* The average daily delta between low and high electricity prices
The second is essential because the delta is where battery savings come from: if prices dropped, but the gap between charge & discharge prices didn't change, our battery savings would remain _exactly the same_.
It makes as much sense to not introduce uncertainty and, instead, to simply not account for either.
* * *
#### Individual Break-Even
There are 2 ways that we can think about break-even: we can roll the cost into the original capital, or we can calculate individual breakeven with subsequent impact of break-even on the original.
In _theory_ both figures should come out about the same, so we're going to calculate both as a cross-check to catch mistakes in my maths.
So, first, how long would it take a new module to break even?
# Cost divided by weekly saving
£500 / £1.41 = 354 weeks (6.8 years)
Just under 7 years. That's already more than half of the warrantied lifetime, ouch.
After that 7 years, the £1.41/week saving would help pay down the original modules. So we need to calculate the impact that that has:
# Figure out how much of the original would have
# amortised in the leading 7 years
(£4.23 * 52) * 7 = £1539.72
# Current capital outstanding, minus 7 years amortisation
£4999 - £1539.72 = £3459.28
# Remaining amortisation would be at the combined rate
£3459.28 / £5.64 = 613 weeks (11.79 years)
So, break even would occur in about 19 years (the initial 7 years, followed by the 11.79).
If we consider that the original modules are now 3 years old, that's more than double their warrantied lifetime.
* * *
#### Amortisation As One Block
That wasn't a particularly happy result, but let's see if it's corroborated by rolling everything together
£4999 / £4.23 = 1181 weeks (22 years)
# Add the capital cost of the new module
£4999 + £500 = £5499
# Calculate the breakeven of that at the combined
# weekly savings rate
£5499 / £5.64 = 975 weeks (18.75 years)
Once again, the result is about 19 years from now (or 22 years from the original purchase).
* * *
#### Break-even conclusion
A new module **would** help reduce the break-even period of the original. However, the reduction isn't _nearly_ enough to bring ROI back within the battery's warrantied, or likely, lifetime.
Unfortunately, despite the significant reduction in the price of new modules, all that we'd actually be doing is _pissing more money away_.
* * *
### It Gets Worse: Time Of Use
The calculations above rely on an important assumption: that the new module would be able to yield the same savings as its counterparts.
However, that may not be the case: if the existing capacity is already sufficient to see us through peak pricing periods, an additional module would discharge when prices are lower (and therefore achieve lower savings).
If we graph the last few days out, we can see that the battery almost always hits minimum charge (the bottom of the blue troughs) after peak prices have passed:
Even on the occasions that the battery did run flat a little early, it wasn't so early that there'd be time to discharge another 1/3 of the energy.
This is actually a dynamic that I've noted as an issue before:
> In the time covered, one of primary determining factors seems to be whether or not we use the oven in the evening: If we instead cook with the airfryer (or a gas hob), we don't draw enough energy at peak time. The battery's charge is then released gradually over the course of the night when prices are lower, leading to negative savings.
So, if we did add capacity, it's unlikely that it'd actually increase current savings by 1/3. We'd still save _something_ but the break-even would be more distant, even, than the already gloomy numbers above.
* * *
### Change Tariff?
A huge part of the reason that the yielded savings are so low is that the energy tariffs we're on are already _extremely_ competitive:
* Export: `£0.12/kWh` (down from £0.15 last year)
* Import: Weighted average over last year, `£0.21/kWh`
Any solar charged into the battery hits an `£0.12/kWh` opportunity cost. The average unit cost being just £0.21 limits our scope for savings to about £0.09.
Big spikes at peak do accelerate our savings, but the larger daily spikes tend to be accompanied by higher minimum prices:
There **are** other tariffs which offer greater pricing deltas though. For example, Octopus Go currently has the following prices
Time | Price
---|---
00:30 - 05:30 | £0.094
05:30 - 00:30 | £0.347
Compared to not having a battery, each unit charged overnight would yield a saving of £0.253 when discharged in the day.
But we **do** have a battery, so assuming that the price delta stayed the same, we'd be saving about `£0.11/kWh` more than we do now.
Go's higher pricing period is much longer than on Agile and we'd need enough capacity to run off battery during those times, otherwise things could quickly go south:
# Our consumption sits at about 30kWh a day
# ~4 kWh is consumed 00:30 - 05:30
# with another 6 charged into the battery
#
# The remaining 20 would be at full price
(10 * £0.094) + (20 * £0.347) = £7.88/day
# Current cost at weighted average
30 kWh * £0.21 = £6.30
# £1.58/day worse off. Over a year:
£1.58 * 365 = £576.70
So, switching to Go without upgrading capacity would leave us worse off by almost £600/year.
Winter configuration sets a minimum charge of 30%, so the usable capacity of each US2000C module is `1.68 kWh`. We'd need 16 modules in total to cover `26 kWh` of usage.
We've got 3, so we'd need 13 more. Purchased at the new price, that would be an investment of (`13 * 500`) £6500.
Running a lazy break-even calculation against that:
# £0.11 saved per kWh, we'd need to charge
# (and discharge)
£6500 / £0.11 = 59,091 kWh
# If we're averaging 26kWh supplied by battery/day
59,091 / 26kWh = 2273 days (6.23 years)
Break-even of the upgrade could happen in just over 6 years.
Unfortunately, there are important costs that we haven't accounted for:
* To fill `26 kWh` in 5 hours we'd need a charge rate of about 6 kW, which means a new inverter, adding ~£1500 to the price
* Because of the increased capacity, we'd need a G99 application
It would be a big capacity upgrade, so we'd need to pay an installer for all of the above. That would also increase the per-module price: the most recent installed price I've seen from our installer is more like £1000/module.
Additional costs pose a _big_ problem for break-even:
# Go Daily savings vs now
26 kWh x £0.11 = £2.86/day
# Turn it into a year
£2.86 * 365 = £1044/year
Every additional £1000 of cost pushes breakeven back a year. Adding just a couple of grand takes us uncomfortably close to the warrantied period again.
If the installer's price _is_ still double the self-installed price, we're back to break-even being more than the battery's warrantied lifetime:
# Modules at installed price
# plus inverter
(13 * £1000) + £1500 = £14,500
# At £1044/year of savings
£14,500 / £1044 = 13.9 years
There's another issue, because the calculations above are all sized with _very_ little room for manoeuvre:
* Mistakes are _guaranteed_ to cost `£0.34/kWh` (almost four times the desired rate)
* If our daily consumption grows, everything not covered by the battery would cost `£0.34/kWh`
* The battery's usable capacity reduces as it ages, gradually pushing more of our usage into the `£0.34/kWh` period.
There _is_ an upside: the tariff would give a guaranteed margin between low (£0.094) and export unit prices (£0.12), significantly reducing the opportunity cost of charging the battery from solar.
Even ignoring that the export price may change in future, the gain wouldn't make much difference during the winter months and so doesn't really change calculations much.
So, although Octopus Go offers a higher pricing delta (accelerating ROI), switching to it would either cost us £600/yr or mean paying £14,500 up front and hoping nothing goes wrong.
* * *
### The Problem
The problem is actually the electricity prices that we're paying: the savings generated by the battery rely on being able to charge at a low price and sell at a higher one. But, odd spikes aside, the range of prices on Agile is fairly compressed.
Switching to Go widens that gap and so made the original calculations look better.
But a swtich would also extend peak pricing from ~4 hours a day to 19 and we'd need to invest in significant capacity increases to avoid paying `£0.34/kWh`. We'd essentially be spending 15 grand to insulate ourselves from the consequences of our own decision.
So, on paper, the battery looks like a bad investment but that's because the prices that it avoids are still comparatively low. Unfortunately, it's not just me: Money Saving Expert seems to have reached the same conclusion:
> even if you slash that solar battery price (and therefore payback time) by 50%, the typical user will still struggle to to pay it off through savings within its 15ish-year lifespan.
Ultimately, I'd rather have cheap electricity and good export rates than have them change and make the sums look better.
* * *
### What We Could Do
Extending the battery isn't a viable financial option, but there are things that we _could_ do:
* **Reduce consumption** : every kWh of consumption saved is one less unit to pay grid prices for (and in summer might also mean another kWh of export)
* **Add multi-purpose storage** : buying a battery might not make sense, but at some point I _am_ going to have to get a new car. With a charger that's able to let an EV supply the house, we'd gain a large storage increase as a side effect of something that doesn't need to justify its own cost via electricity savings alone.
* **Add an islanding switch** : It'd add a couple of grand to our costs, but we'd derive _some_ benefit from the house continuing to operate during power cuts. It's hard to assign a value to that, and power cuts are rare, but the benefit is definitely not 0.
The final two options have something in common: they introduce a significant area of fuzziness into ROI equations. How do we attach an accurate value to operating through a powercut or maintaining mobility/independence?
The answer is almost certainly that the fuzziness provides psychological insulation from what might still be a questionable investment choice. I'm **totally** up for that.
* * *
### Conclusion
It's a shame.
I was really hoping that the cheaper module price might allow me to shift from "the battery was a mistake" to being able to talk about making gains.
I had thought that the original purchase price might have been weighing us down, but the Go calculations suggest that, even at 1/3 of the original price, household batteries still don't really make financial sense.
For the battery to work financially, we'd need there to be a bigger difference between low and high prices each day. But, whilst that might help the battery deliver ROI, we'd probably still be worse off overall.
I think the answer is to view the battery differently: what we've done is paid upfront to reduce our monthly bills by ~10%, with a premium attached3. It's a little like buying an annuity, but one that doesn't actually guarantee the level of income4.
The path forward for us is not to add capacity but to do the things that were already likely: reduce consumption and, at some point, replace the car.
* * *
1. See what I did there? ↩
2. We're currently on Winter settings, so min SoC is set to 30% rather than 20% ↩
3. Over the course of the battery's lifetime, inflation may eat that premium away. ↩
4. That's not really making it better, is it? ↩