Solar panels

This week we got solar panels installed! This isn’t exactly a novel thing, lots of people have solar panels including some of my friends. But I’m stuck on a flight and wanted to write something, so this is what I shall write about.

I had a couple of requirements for solar:

1. A reasonably quick repayment time. I don’t know how long we will stay in this house. I believe in solar enough that if there’s a modest loss when we come to sell up then so be it, the world is an iota better for a few more solar panels on a roof; but if it can pay off financially then that’d be good.
2. Local control and monitoring. Clearly I’m a certain kind of person… The kind of person who enjoys Home Assistant and prefers things to not be in the cloud. So to that end, I wanted a solar system that would be something I can hook up to home assistant and control myself and geek out about. All the better if that connection to home assistant doesn’t involve wireless.

I’ve of course done research on the technology itself. I did modelling of different panels/battery configurations using the EU PVGIS tool; I set up an EasyPV account to do similar modelling there, as well as a bit of cost estimation. The universal thing is that solar panels are substantially more cost-effective with a battery. Up-front costs increase quite a lot, but the repayment time decreases significantly. EasyPV by the way is a great tool to do some “window shopping”. It’s got nice features for drawing roofs using satellite imagery, getting their aspects correct, picking different panel sizes/layouts/brands, and projecting repayment time. Highly recommended if you’re considering solar yourself.

I did some research online and lots of the energy companies; EDF, British Gas, Octopus, … will sell a very competitively priced solar setup, but often with very little choice over specifics. I weighed them up, and also got a quote from a local installer who had a fairly polished looking (clearly AI-built) website. Despite my misgivings about an entirely AI-built website, I chose them and got a quote. It was £14,500 for 9.8 kW of solar panels on two planes, a 15 kWh battery, a 10 kW hybrid inverter¹, scaffolding and all the paperwork. This is certainly a higher cost than the energy companies, but gave me a level of control I liked. Amusingly, the cost estimate and system design came out of EasyPV (the solar company didn’t try to hide it, I’d already discussed EasyPV with him during the initial phone call). The company in question only stocks one brand of battery (and panel, for all that matters) so I checked over the technical documents and spoke to the installer and found it supports local MODBUS over TCP either by Wi-Fi or by Ethernet. The battery/inverter system is made by SigEnergy, and it seems reasonable. The battery chemistry is LiFePo4 which I understand is now typical for home batteries, except the Tesla Powerwall system which still uses Li-ion cells (perhaps the same sort that Tesla uses in its cars, so NMC-based?). I prefer the less spicy higher cycle count chemistry, and the cells that don’t contain conflict minerals like Cobalt.

Installation

The only work I had to do myself was to pour a concrete footing for the battery. I hate handling concrete and other forms of magic mud which harden when mixed with water, but it’s a simple enough task, so I sorted that out. It took me most of an afternoon regrettably, mostly because it’s not possible to buy a straight 2x4 from a hardware store so getting it vaguely level was harder than it should have been. I’m sure someone more skilled would have knocked this out faster, but oh well. I built a 910x310 mm footing at about 150 mm deep for the battery and inverter.

We coordinated with the electrical company and scaffolders to conspire to have the scaffolding up a couple of weeks earlier than the installation date. This meant we could get some roof work done before the solar installers came. We got some roofers in to repair the chimney, ridgeline, and a handful of cracked tiles. This added to the installation cost of the solar system, but probably wasn’t optional anyway. Sooner or later we’d have needed this work doing.

We have a network rack in the cupboard under the stairs, as you do, which is also where the gas and electric come into the house. This is extremely convenient for installation of the mains equipment inside the house, including network access. There’s just about an angle the electrician could drill at to get the SWA cable out of the house and around to the battery. Sadly, due to some miscommunication the electrician didn’t realise we wanted the hardwired Ethernet up to the inverter, so we don’t have a run of fancy PV Ultra cable which integrates a CAT5 UTP cable into the SWA mains cable. Instead, our Ethernet cable has been clipped in underneath the SWA mains, which is totally fine.

The actual installation took 2 days. The first day was electrician and roofers working on site together. Roofers installed the mounting system on the roof, the panels themselves, and fished cables into the loft space. The electrician did first-fix work routing cables and preparing a new DB for the cupboard under the stairs. On day two it was just the electrician, and by 13:00 he had shut off the power to connect the DB to the incoming mains for the house. The shutdown lasted about 45 minutes, and went very smoothly. The UPS for the server rack (which powers the ONT and Wireless access point by PoE) lasted just long enough to keep me online throughout. After that he ran a bunch of tests for the electrical safety requirements, and packed up at about 3pm to finish off the paperwork.

Once the electrical installation was complete, and the out-the-box software for the inverter was ready the real setup could begin.

PredBat

The out-the-box software for the inverter seems fine, although I’m a little untrusting of it on my home network. As a home assistant user, and being friends with at least one other energy nerd, I was aware of PredBat. PredBat is a Battery Prediction software which models household energy usage patterns, time of use tariffs, and forecast solar production to optimise the charging of a home battery. To do this, one needs control of the battery within home assistant – which is easy with the MODBUS Sigen add-on in HACS (TODO: link) – and some solar forecasting service set up. Tariff information is provided by the Octopus Energy API, via BottlecapDave’s add-on. Home assistant has forecast.solar out-of-the-box, which requires a latitude/longitude and panel capacity/orientation information, but reportedly that’s a little inaccurate. forecast.solar uses the EU Solar prediction model (TODO: name), and I’m a little surprised it’s considered inaccurate. PredBat recommends Solcast, which is an Australian service for solar prediction which is free for hobbyists/home users. Given how long Australians have been using home solar it’s not too surprising they’ve got good tools for this.

Judging by the commit history the majority of users/contributors to PredBat are in the UK. I don’t know what the energy situation is like in other countries, are we ahead of our time in giving end users lots of power to contribute to the grid themselves (and get paid for it) or are we behind in that we need these schemes to try and avoid our otherwise ruinously expensive energy prices. Who knows. I’m glad PredBat exists.

Anyway with all the data sources plumbed into Home Assistant, it was very easy to get PredBat managing the battery.

One thing that PredBat does which I’ll definitely keep an eye on is aggregate savings of the system. It uses your actual energy use and patterns to determine costs without a PV+Battery system, and also models what it would be with PV+Battery but without PredBat battery control. I’ll try and remember to update this time next year…

Our Grid Connection

Our grid connection starting all of this was 60A. Our grid connection at the time of writing is also 60A. This is a little low. Ideally it’d be 80 or 100A. The solar company noted the same thing, and needed to provide the DNO with evidence that we’d have a suitable grid connection for the G99 application. To that end, our DNO (UK Power Networks) agreed to come out and upgrade our grid connection. Yippee! Sadly… our main fuse is in a “biscuit tin” fuse box, so-called because it’s made of metal. Built like things from the 40s/50s, clearly made to last. Sadly, metal is conductive, and this box has unfused mains going into it – which makes replacing it sound quite spicy.

The UKPN are quite good at what they do, and in particular they (mostly?) avoid blowing up their electrical engineers with unexpected arc flash incidents. This is bad news for our upgrade though, it means that the fuse box can only be changed while dead. They sent someone out to double-check, and yep – we need to be disconnected from the grid in order to upgrade the service head. To do the upgrade they must cut the mains cable outside the house, replace the fuse, then reconnect the house. In this process they traced roughly where the mains cable crosses the property line, and that location is under our drive quite close to the house. The front of the house covered in scaffolding. So UKPN can’t dig under the scaffolding, so we have to wait for that to be gone before they can begin. A bit of a Polava, but it seems this hasn’t ended up being an impediment to getting a G99.

What’s next

We’re waiting on the DNO to issue us an MPAN (meter number) which will let us get paid for our exported energy – currently we’re simply giving it away to the grid. Once we have that we can pick some import/export tariff. Sadly the best tariffs are currently unavailable. PredBat’s internal modelling suggests we’d get a 3.7yr payback time with Octopus Intelligent Go import/export – which is crazy! Their non-intelligent Go is estimated for 5.9yr payback.

We’ll also set up Axle, which is a VPP (“virtual power plant”). VPP is a terrible name, but when grid wholesale costs are extremely high Axle will pay users to force discharge their batteries into the grid on top of their normal export tariff. Currently, they advertise £1/kWh. With some flexible import/export tariffs Octopus offer hitting £0.34/kWh for export this will hopefully cut the repayment time down a bit as well. Axle is also supported in PredBat, so hopefully it can plan around the Axle sessions to continue maximising savings. I’ll report back…

Is the name ‘Axle’ a pun? Are the battery owners the new flywheel and Axle is the thing which connects to those flywheels to transmit power? I’d enjoy that…

What about a heat pump?

Sigh. What about a heat pump indeed. I’d love a heat pump. Our house is probably reasonably well suited for a heat pump too. But the costs. The costs. Our estimates are around £8,000 after the government £7,500 grant, and the savings are potentially very small. So environmentally it’s the right thing to do, but economically it’s hard to justify unless we are planning to stay in this house for a long time. If/when the boiler packs up I’ll consider it harder. Thankfully the $6 mm^2$ SWA to the inverter/battery should be plenty for a reasonably sized heat pump as well – the inverter is limited to 9 kW export and 8 kW import so will draw about 37A under full load. A very large heat pump might be 4 kW, which is about 16A under full load – so should be in-spec with some safety margin on the wiring. Iirc the allowed power over clipped $6 mm^2$ SWA is about 67A.

In the spring I’ll review what our actual energy cost was per kWh and re-run the calculations on repayment time for a heat pump conversion. Our current gas heating is using a modern (2019) condensing gas boiler so has a COP of approximately 0.9 on gas which costs 8.3p/kWh, or a cost per kWh of heat of 9.222p. A decent heat pump will probably get us an average COP of about 4 over the winter season, which using just off-peak electric at 14.3p/kWh would mean a cost per kWh of heat of 3.575p. We use approximately 11,000 kWh per year of gas – so that times the difference in price means a saving of £621/yr, or a repayment period of 12 years. Plausibly sensible if we were to live here indefinitely, but even then it looks a bit dubious against inflation on the same amount of money.

I think until heat pump install costs drop by a lot it is going to be hard to justify the expense. I am also slightly suspicious that a heat pump upgrade doesn’t actually cost close to the £16k that government+customer are spending. It seems this is a market with more demand than supply, and one that’s being juiced by government subsidy. Perhaps that’s the tinfoil hat speaking, though.

Policy aside

As an aside to an aside… The government policy to hand out £7,500 grants for heat pump installation is, to me, quite short-sighted. In theory, it works as investment in that industry which will help drive down prices and shape the UK grid in the way the government would like. In practice, it’s hard to imagine a more inefficient way to do things. The same £7500 could totally pay for a battery installation for a home, which would save the user substantially more money and provide grid resilience as we move to more and more renewables. I can’t help but feel like this is a £7,500 bung to already quite wealthy middle class homes that don’t need the help and props up higher prices for heat pump retrofits.

It’s hard to test the counterfactual of course, so who knows. Hopefully with Octopus’s own-brand heat pump now available at a pretty low price that competition will actually manifest and the total cost of installation will drop precipitously. Time will tell.