lupine/capsule
1
0
Fork 0
Code Issues 1 Pull Requests Releases Wiki Activity

Add solar blog post

Browse Source
This commit is contained in:
Nick Thomas
2022-09-30 23:43:19 +01:00
parent 851e378c7e
commit a84054452c
1 changed files with 67 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files

67
content/blog/2022-09-30-solar-installed.gmi Normal file
View File

@@ -0,0 +1,67 @@
---
title: "Solar Installed"
---
Well, almost. Yesterday they put the scaffolding up, delivered the panels, and got to putting them on the roof. Six are up there, but the remaining three are sheltering in the garage because they didn't bring enough of the railings that connect panels to roof.
Of course they didn't.
The scaffold is now up until at least Wednesday, which is the first day they can come back with more railing to complete the job.
The six panels are, however, functional, and comprise 2.4kWp of generation capacity. Despite only being there from 2pm, they generated ~3.4kWh yesterday. Today was foggy and overcast, but they managed another ~1.4kWh. Tomorrow and Sunday are looking pretty sunny, but October is hardly peak generation time. Still, according to PVCalc, I can expect an average of 5.23kWh/day in October once the remaining panels are up - nearly half our average daily usage. This lines up with the generation projection the solar survey came up with too.
=> https://joint-research-centre.ec.europa.eu/pvgis-photovoltaic-geographical-information-system/getting-started-pvgis/api-non-interactive-service_en PVCalc
Using the generated energy is the challenge. Of that 3.4kWh, I exported 2kWh - over half. Since there are no batteries right now, I need electricity loads in the house to soak up the solar as it's being generated, or some means of storing it. Energy in the UK is bucketed into half-hour billing periods, so generation and use don't have to match exactly, but it's frustrating to see a big energy use peak right next to a bucket where 0kWh was imported.
## Regrets
One of the electricity loads I have is an "iBoost", which is meant to turn on the immersion heater, bypassing the heat pump entirely, if there's excess generation. Charging the thermal store (hot water cylinder) at a CoP of 1 is pretty rubbish, but - economically speaking - better than exporting to the grid if you're getting paid peanuts to do so. For me, the device is pointless, but it's installed. At least I can turn it off.
It seems that there might be room for fifteen panels on the roof, but the installers are refusing to install capacity over the G98 limit of 3.68kWp, even if the export is inverter-limited to that amount. So the scaffold will probably be going back up next year, but adding six, rather than three, panels, makes it more worthwhile. In retrospect, there was time to get the G99 application in to have more installed in a single go - although permission is not guaranteed, so perhaps this was the correct route after all.
It's also occurred to me that my shiny new heat pump is plumbed in wrongly for a house with solar PV. Leaving aside the fact that it would have been wise to get solar thermal as well, it would be better if the heat pump exclusively heated the hot water cylinder, and the cylinder was used to drive the central heating as well as domestic hot water. This would allow me to avoid switching the pump on at night, when there is no solar, but would probably have required the cylinder to be much larger than it is. It's all good learning for the next house, though.
## Buffering
The cylinder is a thermal store, but not a large one, and not plumbed for maximum utility. However, the "grid as battery" idea that I talked about in "Off-grid in Suburbia" is looking better and better. I'm switched over to Octopus, from E.on (who were giving 3p/kWh for export). Once all the paperwork is done, I can either choose a guaranteed 15p/kWh (just increased!), or use "Agile Outgoing", with prices ranging anywhere from nothing to > 60p/kWh, depending on demand and time of use.
=> /blog/off-grid-in-suburbia/ Off-grid in Suburbia
=> https://www.energy-stats.uk/octopus-agile-outgoing-export-yorkshire/ Octopus export tariff prices
Since I can't shift my generation (via batteries) to the highest-priced times, the fixed rate is probably the right one for me. It's still 5x the price my previous supplier was offering, and is high enough to obsolete the iBoost. The heat pump has a CoP of ~3.5; when it's running on imported energy (~33p/kWh), this means each kWh of heat costs ~9.4p. The iBoost, running on generated solar with CoP=1, has a notional cost of 15p/kWh, so using it is the same as throwing away ~5.6p/kWh.
Batteries can be added next year, but definitely not this year. I'll probably attempt to DIY it on the DC side; the batteries can then discharge when there isn't any solar, or when I want to export, through the existing inverter, without DNO involvement. I need to start being more hands-on with this stuff. Only without getting electrocuted.
Buffering is about moving the energy to time of use; taking the same problem from the opposite side, we can move the time of use instead, which leads to:
## Reasonable Home Automation
It feels like that house is far more connected to the elements now it has solar panels. Suddenly, it's beneficial to take local conditions into account when using energy.
For example, the heat pump is currently programmed to heat the cylinder between 12-1pm each day. When running directly off of solar, we can divide the export price by the CoP to arrive at a notional cost: 4.2p/kWh - less than half the cost of running it off imported energy. If it's cloudy at midday but sunshine is forecast for 4pm, it makes sense to delay running the pump until then, so it can take advantage.
Similarly, we have "set-back" crudely configured into the thermostat at the moment - the temperature is allowed to drop from 21'C to 18'C between 9pm and 7am. These are very arbitrary, though, and I often find the pump is running at 8:30pm because temperatures have dropped to 20.8'C or so. Having the thermostat take sunrise and sunset into account, and having it progressively ramp setback up and down rather than it being a cliff edge, makes sense. So does allowing the temperature to drift a little (above or below 21'C) depending on current and predicted solar availability.
There are more conventional examples too, like running conventional appliances while it's sunny, but for 4-6 months of the year, the heat pump is the single biggest energy user, so I'm focusing on that. (The second-largest appears to be cooking, which I'm ignoring for now, but changing the time you cook dinner is entirely feasible without any automation).
All this is solved with "smart home" gadgetry, but sadly, those gadgets are of the "upload everything to the cloud" variety. I'm dipping my toes into more-reasonable approaches, and have the bones of a plan:
=> https://shop.electrolama.com/collections/usb-rf-sticks/products/zzh-multiprotocol-rf-stick ZigBee zhh coordinator module in the home server
=> https://shimmeringbee.io/ ShimmeringBee coordinator software (ZigBee + MQTT) running on the home server
=> https://www.develcoproducts.com/products/meter-interfaces/external-meter-interface/ ZigBee pulse counter on the generation meter
=> https://shop.glowmarkt.com/products/display-and-cad-combined-for-smart-meter-customers Glow "CAD" add-on to get (local, not cloud) MQTT feeds for the import + export meter
=> https://github.com/aerona-chofu-ashp/modbus Scrape the heat pump's Modbus interface and get it into ShimmeringBee somehow
That should be enough to get me the data I need to calculate scenarios like the ones above. "All" I have to do then is enrich it against internet-available data (weather predictions, say), make decisions automatically, with the potential for some human input too, and have them be implemented by hardware.
These things need coordinator software. ShimmeringBee is a bit of a find, being a tiny Go daemon without any kind of UI. Trivial to deploy, trivial to modify, I glommed onto it as soon as I found it.
Lots of other options, but I'm going to stick with it to begin with, even though I'm going to have to write code to get most of the items above talking to it. At least the zzh coordinator works out of the box. It has an MQTT port open too, so hopefully the import/export meter will work fine. The decision code can just go straight into the daemon, I think.
=> https://ubidots.com/blog/open-source-home-automation/ Other options
I don't have hardware for implementing decisions planned out yet. So much of it is going to be around the heating system. A thermostat I can control from ShimmeringBee will be the first requirement, possibly along with ZigBee-connected TRVs, so rooms can be heated or not individually.
It's a techy dream (nightmare?), but the underlying goal is to get the house using less energy. Of course, all this tech *uses* energy, but I'll be doing my best to keep that aspect under control, mostly through careful selection of low-power components and software. At least the glue is code I'm looking forward to writing!