Adding a Balcony Solar Kit To Our Capacity
Our solar install is not far off three years old.
We've got two east facing panel arrays providing 3.2kWp of PV capacity, supported by a `6kWh` battery.
However, only having an easterly exposure has always been a _little_ bit of a bugbear. Although the panels work throughout the day, they start to wane around lunch time, wasting some of the potential that a sunny afternoon can bring.
The house itself doesn't have a usable roof with a westerly (or southerly) exposure. The _garage_ , however, does.
Although it obviously won't generate as much as the existing install, I decided to order a balcony solar kit for the garage, so that we could mop up some of the afternoon sun.
This post describes the install process along with the initial results (despite the weather being a bit... British).
* * *
### Balcony Solar In The UK
These kits sometimes get referred to as "plug in" solar because, in Europe, they're widely available with a plug allowing them to be plugged into a spare socket. _Balkonkraftwerk_ kits are, apparently, hugely popular with renters in the German market.
In the UK, however, it's **not** OK to simply plug into a socket and a dedicated circuit is required instead. It's not that plugging it in won't work, but that wiring regulations (and therefore home insurance) don't currently support doing so.
However, that **is changing**: following a study, the government announced that it will legalise plug in solar, with an amendment to BS 7671 expected to land this year.
This is, overall, a great outcome as it will make solar more readily available to renters.
However, **my** system isn't connected via a plug: design and setup pre-dated the Government's announcement so I've had it hardwired in to comply with the existing regulations1.
* * *
### The Kit
My kit consisted of
* 2 Trina 440w panels
* A Hoymiles HMS-800w-2T 800w grid tied micro inverter
* Various Fixings
To suit my planned layout I also ordered a couple of extra bits
* 2 pairs of DC Solar extension leads
* A couple of pairs of hinged solar mounts
* * *
### Install Topology
Although I ordered the kit with the garage roof in mind, I was actually jumping the gun a bit.
Our garage has an asbestos concrete roof, which we're planning on getting replaced this year.
It's obviously not possible to mount panels on there without risk of disturbance, so I either needed to wait, or to install the panels into suitable temporary locations.
I'm not _great_ at waiting, so I went for the second option.
We've got a small lean-to shed attached to the end of the garage, and a pergola style walkway2 up the side, so I planned to place one panel on the shed and the other over the walkway:
The microinverter would then sit inside the garage, protected from the elements.
Although the shed sits a little lower than the garage roof (causing shade when the sun's to the south), it seemed like a good location to catch the waning sun and deliver electrons when we need them most: during peak pricing4 at the end of the day.
* * *
### The Install
With the challenges of working at rooftop height removed, installing panels isn't particular difficult. They **do** weigh a little more than you might expect, but are still within the bounds of a one-person job.
To sit the panel at an angle and create a suitable mountpoint, I screwed a bit of wood across the top of the pergola and then used hinged mounts to fix the panel to it:
So that the panel wouldn't flap in the wind, I also used hinged mounts to secure the other end to the cross beams:
Before starting, I had thought that the shed mount would be the easiest of the two but, at install time, ran into an issue.
So that I could make it tilt and face west, I'd ordered hinge mounts, assuming that they could be fitted along any edge of the panel.
However, the panels only have fixing holes on the _long_ sides.
Although it'd probably have been fine, I didn't fancy drilling new holes into a brand new panel, so decided to use flat mounts instead
The downside of this is that it increases the likelihood of the panel being in shade. It's not a great _long term_ placement for it but, I hoped, would be OK for a while.
DC cables run from the panels into the garage and back to the micro-inverter (which supports two strings, allowing me to track the two panels separately).
The inverter's AC output is wired back to a fused isolation switch:
To protect future sparkies, the warning sticker in our main fusebox has also been updated to list the garage as a generation location.
* * *
### Initial Start-Up
We had a _little_ bit of a false start.
It was time for the grand switch on, so I connected the panels and flipped the isolation switch.
The inverter's LED started blinking indicating that it was starting up. Because it can take up to an hour to sync with grid frequency, I walked away with the intention of checking back a little later.
However, before that, I got a somewhat alarming message from my partner:
I went downstairs and, as soon as I stepped out of the back door, was greeted with a very definite smell of burning.
I went into the garage, flipped the isolator switch and then realised that it didn't smell inside, indicating that it was _probably_ coming from around the panels or their connectors.
I checked the connections for each, then climbed up on a ladder to check the panels themselves. The shed panel didn't have any smell around it, the walkway one had more of a _hint_.
The smell _was_ definitely dissipating though, so it was hard to ignore the coincidence of it coming and going around the time that things were turned on and off.
I did a few extra checks
* Pointed my thermal camera at _everything_ : no sign of anything being hot
* Put my multimeter across the end of the DC extension leads - both panels were giving a good 37V, suggesting no issue with the leads or the panel end connectors
I was **sure** that everything looked fine and yet, the smell...
I had other things that I needed to be doing, so I decided to leave things disconnected until the next day, when I'd be more able to monitor it.
Later that night, though, I was reading the local news and spotted this:
We're far enough away that the smoke wasn't visible, but close enough (with the help of the wind) to have caught the smell.
The next day, I connected things back up and, sure enough... no burning smell.
* * *
### Monitoring
Hoymiles inverters support both remote and local monitoring.
Once I'd figured out getting the inverter onto our wifi it started reporting into the S-Miles Cloud:
I haven't played around with it much, but it seems OK.
* * *
#### Connecting Telegraf
I _already_ monitor our existing solar install so wanted to get metrics into InfluxDB.
Happily, it turned out that someone has already created a telegraf plugin to connect to the inverter and retrieve metrics.
I cloned the repo down, built the plugin and then moved the resulting binary to the directory that I tend to drop telegraf plugins into:
git clone https://github.com/liwde/telegraf-hoymiles-wifi.git
cd telegraf-hoymiles-wifi
go build -o hoymiles ./cmd/hoymiles
sudo mv hoymiles /usr/local/src/telegraf_plugins/
The plugin requires a separate config file to define inverters that it should connect to:
cat << EOM > /usr/local/src/telegraf_plugins/hoymiles.conf
[[inputs.hoymiles_wifi]]
hostname = "192.168.13.227"
EOM
With everything in place, I configured `telegraf` by adding the following to `/etc/telegraf/telegraf.conf`:
[[inputs.execd]]
command = ["/usr/local/src/telegraf_plugins/hoymiles", "-config", "/usr/local/src/telegraf_plugins/hoymiles.conf"]
signal = "none"
I reloaded the telegraf service and points like the following started being written into InfluxDB:
hoymiles_dtu,dtu_serial_number=xxxx dtu_energy_daily=52i,dtu_power=70.6 1774600109000000000
hoymiles_inverter,dtu_serial_number=xxxx,inverter_serial_number=xxxx inverter_power=70.6,inverter_temperature=13,inverter_voltage=246.4,inverter_frequency=50.03,inverter_current=0.28 1774600109000000000
hoymiles_pv,dtu_serial_number=xxxx,inverter_port_number=1,inverter_serial_number=xxxx pv_energy_daily=27i,pv_energy_total=84i,pv_voltage=31.5,pv_current=1.23,pv_power=38.8 1774600109000000000
hoymiles_pv,dtu_serial_number=xxxx,inverter_port_number=2,inverter_serial_number=xxxx pv_voltage=32.5,pv_current=1.09,pv_power=35.7,pv_energy_daily=25i,pv_energy_total=38i 1774600109000000000
* * *
##### Reporting Gaps
One thing that I hadn't really thought about before, is that the inverter becomes unreachable overnight.
It's powered by the panels themselves, so as the yield drops away the inverter goes offline:
Happily, though, the telegraf plugin copes with this just fine.
* * *
#### Grafana
As a start point, I added cells to my existing Grafana dashboard:
I also created a new dashboard dedicated to the Hoymiles inverter:
* * *
#### Output Level Alerting
Although it doesn't _really_ add much beyond some psychological safety, I created a Grafana alert to warn me if the inverter's AC output ever pushes towards its max capacity:
The alert is driven by a simple query to get the maximum reported current in the queried period:
SELECT max("inverter_current")
FROM "Systemstats"."autogen"."hoymiles_inverter"
WHERE $timeFilter
The maximum output listed on the back of the inverter is 3.7A. However, 800W at 240v is around 3.3A so I decided to flag if it went much above that.
* * *
### Limiting Export Levels
In order to export electricity and receive payment for it, solar installs have to be registered with the local electricity distributor (the DNO). The idea behind this is that it allows the DNO to maintain an understanding of what might be fed back into to the local network.
Following install of our main system, we received approval to export up to 3.6kW.
I track our metered electricity usage via our glow IHD, so queried this data to check whether we've ever been anywhere near that level:
from(bucket: "Systemstats")
|> range(start: 1)
|> filter(fn: (r) => r["_measurement"] == "smart_meter")
|> filter(fn: (r) => r["_field"] == "export_now")
|> aggregateWindow(every: 1d, fn: max, createEmpty: false)
|> group()
|> max()
Our highest recorded export rate was `2.827kW`3.
Although we've never hit our approved level, I still needed to make sure that the addition of the new panels wouldn't risk taking us past it.
Because there's a data cable between our main install and the fuse box, our Solis inverter is able to read import and export rates from a meter which sits between our leccy supplier's meter and our consumer unit.
The energy provided by the new panels feeds into the consumer unit side, so any export that it generates would be included in the Solis inverter's readings.
I double checked that the Solis inverter was configured appropriately:
As a fail-safe, I also added an alert to Grafana so that I'll be paged if our Smart Meter ever reports greater than 3600W of export:
SELECT
max("export_now") AS "export_now"
FROM "Systemstats"."autogen"."smart_meter"
WHERE $timeFilter
GROUP BY time($__interval)
FILL(null)
I don't expect that this should _ever_ fire but, if for some reason our export does spike, I should find out in good time rather than relying on eventually being contacted by the DNO.
* * *
### Performance
Unfortunately, the install missed the recent spate of really sunny weather and most days have suffered from prolonged cloudy spells. However, as I'd hoped, the panels **do** help us produce electricity later into the day, taking advantage of the afternoon sun.
However, the horizontal panel on the shed roof is outputting **much** less power than I'd been hoping for:
That's partly a consequence of it being horizontal, but is also probably a time of year thing - it covers a part of the sky that the sun won't be high in until later in the year.
Still, it's probably a sign that that location is a pretty poor use of a panel, so I'm considering relocating it to the walkway instead.
What matters **most** , though, is that the panels are delivering energy some way into peak, supporting the battery and shaving some small amount off the power that we draw from grid as peak pricing starts.
* * *
### Calculating Break-Even
I've put quite a lot of effort into accurately tracking break-even of our main install, so the introduction of the new system presented a number of questions:
* Should I track break-even for this system separately?
* Should I update the existing calculations to factor in new system output (and cost) or only report separately?
For now, I've decided to track it entirely separately - the kit cost hundreds, rather than thousands of pounds, so it didn't _quite_ seem worth risking (inevitable) mistakes updating existing calculations.
Because there's no battery and no (separate) export, the calculations are quite a bit simpler, feeding into a new dashboard that's similar to my existing one:
* * *
#### The Downside Of Cheap Electricity
It's a "nice" problem to have, of course, but being on Octopus Agile means that, on average, we pay less per unit:
Break-even time calculations are based on the cost of the electricity that we'd have to have bought if solar didn't provide it.
But, realistically, if we didn't have Solar (or more accurately, the battery) we _probably_ wouldn't actually be on Agile in the first place (and certainly wouldn't achieve such a low average unit price).
To get an idea of the difference, we need to look at Fixed price tariffs instead:
* British Gas: 25.817 p/kWh5
* Octopus: 25.86 p/kWh
* EDF: 27.69 p/kWh
Splitting the difference between those prices gives a unit cost of 26.73 p/kWh6. Although only a matter of pence more, it's still a 40.7% increase: if I used fixed rates in the calculations, the time to break-even would almost halve!
With that in mind, the current projected break-even doesn't look quite so bad, especially given that the figures are currently based on a single cloudy week.
* * *
### Conclusion
Although the balcony solar kit only provides 25% of the capacity provided by our existing install, the capital cost was **significantly** lower (helped, of course, by not needing scaffolding). The break-even point should, therefore come quite a bit sooner.
Once the garage roof has been replaced, the panels will move to a location where they'll be better exposed to the sun (and for longer). The fact that they'll be most active during times of peak pricing should further accelerate them down the path to payback.
Time will tell, but I wouldn't be overly surprised if the break-even period ends up being just a few years (which will be all the more likely for others if these kits ultimately end up in the Aldi middle aisle).
* * *
1. This isn't a bad thing really, even if the timing is a _little_ annoying β©
2. Which I've always hated anyway β©
3. Which, rather than being panel driven, is almost certainly from me telling the battery to dump to grid during a savings session or similar β©
4. Essentially, this panel's role will be to support the battery in supplying our peak usage. The battery already doesn't make economic sense so adding capacity to it isn't really an option. β©
5. In keeping with Centrica's awful customer service, you need to provide usage details to see tariff prices. Which is only part of why friends don't let friends use British Gas. β©
6. Though, between writing this and publication, prices are expected to come down a bit as a result of Ofgem lowering the pricing cap β©
