Thirteen years of rooftop solar power

We had solar panels installed on the roof of our home 13 years ago today. Those panels, which face South-West have been supplemented by additional solar panels facing South/South-East, installed in 2022/2023.

Solar power total production in kWh by our rooftop and garage roof panels since their installation in 2012 / 2022. There's still no sign of degradation: Production from the rooftop panels last year was 3% higher than an average year.

I think it's worth noting that the output of the 13 year old panels on the roof of our home is still as good as it was when they were new. In fact, last year's output was 3% higher than the average over their 13 year life so far. The highest single year output was two years ago when they were already 11 years old. Solar panels do not degrade quickly and unless they are physically damaged it's very unlikely that they need to be replaced.

I increasingly see people selling older solar power systems in order to replace their existing panels with new ones, and I do not understand why they do this. Newer panels do have a higher output than the old ones (e.g. our garage panels are rated over 400 Wp each while the rooftop panels are 270 Wp each), but this is mostly because they are larger. Even if the horrible effect on the environment of disposing of products early is disregarded, I can't believe that the slightly increased output per square metre with newer panels makes it financially viable to swap panels.

I suspect that the large number of single phase older equipment on the second hand market is being driven by a hard sell when people "upgrade" their electrical connection from single phase to three phase. But as it's possible to buy a string inverter for three phases to replace a single phase inverter if so required, it seems likely to me that people are being sold something they don't always need. In my opinion, installing a few extra panels alongside the older ones makes more sense than replacing everything.

When we installed our extra panels on the garage I considered buying second hand panels from one of the people who was for some mysterious reason replacing their existing set, but I couldn't find panels locally with prices low enough to make them attractive. The problem is that even if the panels are local and they have a very low price there are still two things working against them: 1. because they're smaller in capacity you need more material per kWh to mount them (and that costs almost as much as the panels do). 2. you don't know what someone has done with the old panels. Have they been removed from the old roof carefully ? There are too many unknowns.

Cleaning solar panels
I've never tried to clean the solar panels on the roof of our home. They're a long way off the ground and trying to clean them would be dangerous. Nevertheless, their output remains at the same level. My recommendation is that you never clean solar panels installed on a roof. If you try to clean them then you may damage them or injure yourself, and I think it's completely pointless anyway because they self-clean remarkably well each time it rains.

Dynamic tariff effect
The graph above will be less meaningful next year because we now deliberately turn our solar panels off for a few hours on some days to balance the grid. We've already been doing that for almost a month so the graph above is already affected slightly. Unfortunately this will make it more difficult compare the performance of our solar panels each year.

Inverters
As those who read this blog before will know, the ABB inverter for our rooftop panels had a five year warranty and failed after just over six years. The company who made it were absolutely no help at all, offering to do nothing more than sell us a complete new inverter. I took the inverter apart and fixed the problem. The factory soldering was poor, with dry joints which heated up and failed, so I replaced a damaged relay applied new solder. That inverter has now worked for more years since my repair than it did from new. But it's now been in use for 13 years, so I have some concerns about whether something else will fail.

The solar panels on our garage are connected using Hoymiles microinverters. These are only a couple of years old so I would hope they still have a long future ahead of them. Time will tell.

Optimising energy usage so we can do everything with a 25 A mains connection and minimally upset the grid

When our home was built in 1972, people had fewer electrical devices. But more importantly, they heated and cooked with gas, which at that time was not only a cheap energy source but also one which few people in the Netherlands seemed to have realized could eventually run out, or cause earthquakes, or cause sea level to rise which could make much of our country uninhabitable, or that it could warm the planet and make many places uninhabitable.

Anyway, when our home was built the need for electricity wasn't so great and a 25 A connection was considered to be enough for average homes. That's what we have. 25 A * 220 V = 5500 VA (which is not exactly the same as 5500 W). We can therefore only draw a maximum of 5500 VA from the mains before we risk blowing the fuse. It also of course means that we can't send more than that much electricity back to the grid from our solar panels without risking blowing the fuse.

These days many people upgrade their fuse boxes, often switching to a three phase connection (often referred to as "krachtstroom") at the same time. Many newer homes are built with 35 A fuses on all three phases, allowing them to consume more than four times as much energy from the grid as we can take. This is the case even when those homes still use gas for heating. If everyone actually started using so much electricity at once, we'd be in trouble: it would require the grid itself to be upgraded and more power stations to be available to supply the demand.

We've not gone down the path of increasing the supply. Instead, we're trying to limit the energy usage of our home and keep our consumption within 25 A, even though we no longer have a gas connection so all our heating and cooking demands are now met with electricity. To do this we need to match our energy production with our solar panels so closely as we can, and to make sure that heavy users of electricity don't all switch on at the same time.

Balancing supply with demand
The wholesale price of electricity varies per hour depending on the consumption vs. production. The shape of the graph varies per month, with solar panels producing electricity in daytime in sunny months greatly reducing the value of electricity generated during those hours compared with the value at the same time in winter.

Graph of average wholesale prices of electricity per hour (source). Just trying to match this is a good start, but I ended up doing more as you can read below.

The price of electricity is directly related to the demand vs. supply. During the night most people are asleep and inactive thus energy demand is lower. During the day the peak demand is at breakfast time and dinner time, while peak solar production is in the middle of the day.

13 years ago when we first installed our solar panels, we would generally try to wait until the sun was shining before switching on large energy users like our washing machine, but nothing in our home automatically responded to the demand curve. At that time not many homes had solar panels so any electricity that we exported was always useful to someone. Our energy company gave us the same credit for every kWh that we produced as each kWh that we consumed, regardless of the time of day or the season. The situation is quite different now because 35% of homes now have solar panels and there are a great many solar farms in the countryside. That is why the price of electricity now drops so obviously during bright days. There are also many wind turbines now and on windy days the price also drops.

Automatically matching demand to supply
When I first installed our electric water heater it ran on a simple and inexpensive mechanical timer which switched it on at about 10:00 and switched off at 18:00 (it would actually stop much earlier when it reached the set temperature, unless we took a shower in the afternoon).

A now retired simple mechanical timer which used to switch our water heater on and off. I still recommend this as a good first step. It's easily affordable and very easy to use.

The timer was my first attempt at matching demand to supply, and if you can do nothing else I recommend it as it works quite well on average. The timer roughly matched the consumption of our water heater to when the electricity was flowing from our solar panels. However it didn't respond to other appliances using electricity at the same time (e.g. dishwasher, electric water heater in the kitchen, cooker, boiling a kettle) so we would quite often see our total consumption rise above what our solar panels were generating even on sunny days, while later on the same day we'd see the sun shining enthusiastically on our solar panels but the water heater would have stopped ages ago on reaching it set temperature, and we'd be exporting instead. A simple timer is much better doing nothing at all, but we can do better than that.

Reading from the meter
Dutch smart meters have a P1 port from which it's possible to obtain real time readings of the amount of electricity flowing in and out of your home. This provides electrically isolated serial data which it's safe and perfectly legal to use with a DIY project.

This is installed in our meterkast. An Arduino receives serial data from the P1 port, lights LEDs depending on energy usage (three green = 1500 VA export, orange = underfloor heating on) and passes the data to an ESP32 (the blue LED is the ESP32 activity light) which controls other devices in our home.

The first device that I attached to the P1 port was an Arduino programmed to display energy consumption on a set of LEDs so that we could see at a glance whether we were sending electricity to the grid (green LEDs) or consuming energy from it (red LEDs). It also included an alarm which would go off if we were approaching the 5.5 kW limit for consuming energy from the grid and automatically switched on underfloor electrical heating at any time that the temperature in the entrance hall was too low, or at any time that we were exporting a lot of electricity if the temperature was lower than we'd like it to be.

My intention at the beginning was to completely avoid wireless connections. The air filtration system uses a serial link between two Arduinos on the ground floor and the top floor (there was an existing conduit which made this easy). But for this project I realised quite quickly that it would require me to run a lot of cables. I was also going to need to know the time of day and the date. For those reasons I switched to using ESP32 controllers.

The ESP32C3
The ESP32C3 embedded boards that I'm using are very inexpensive, they contain a fast and powerful 32 bit processor, plenty of memory and even built in wifi. They're also smaller than my thumb.

An ESP32C3 controller board resting on my thumb.

The first, master, ESP32 is built into the same case as the Arduino, receiving the P1 data over serial from the Arduino. It couldn't replace the Arduino in that location because it doesn't have enough GPIO pins, but the two devices complement each other nicely. This first ESP32 works as a server and the other ESP32 powered device controllers connect to it to find out what they should do. I can see the status of all the devices on a simple web page at a static IP address hosted by the ESP32:

The single webpage hosted by the master ESP32 within the meterkast. It displays how much power we're importing, exporting, the current price, the rank (in the above this is the 4th cheapest daytime hour today), and also data about the hallway temperature and heating as well as data sent to it by various other connected devices. It's still quite cold outside which is why the temperature sensor by the front door went down to 12.7 C, and a few small heating elements were switched on in order to make sure no condensation can form. The "Tibber price" is the retail price including taxes etc.

 

The connected secondary devices all have a mains lead, two sockets (one switched with a relay) and use old phone chargers for their 5 V supply. Some of them use Dallas DS18B20 compatible temperature sensors. Some also have an override button which can be pressed to switch on the heater for a short period when it would otherwise not switch itself on.

The master ESP32 downloads the day ahead electricity prices at midnight and calculates a ranking of when the cheapest time is for any devices to operate (that's dayrank and nightrank on the status page above). Secondary devices can then  decide on which hours they can best consume electricity in order to reduce the price, and because these things are intrinsically related, best balance the demand on the grid.

For instance, we know that our main water heater heats its contents at a rate of about 7 C per hour. As a result, the controller can calculate ahead of switching on time how many hours it will need to raise the temperature from the currently measured temperature to the target temperature and it can optimise by only switching on for the daytime hours which rank below that value. At times of the year when our solar power can't cover all our demand it also automatically uses the cheapest nighttime hours using the same mechanism and the nightrank. This means in practice that in the morning when it has some value we allow our electricity to be exported to the grid, while around lunchtime when the price is at the lowest for the day the heater then uses the electricity rather than adding to the excess on the grid.

Note that these controllers also continuously monitor how much electricity is available either by importing from the grid or our exporting to it. Thus if we are producing just a little more electricity than the water heater needs and we put on the kettle to make a cup of tea, the water heater controller responds by switching off until the kettle is boiled, before switching itself back on. This means we humans get priority. The kettle, or anything else that we humans switch on and off, automatically gets whatever solar energy we are generating and we never have to manually switch off any of the automatically controlled devices so that we don't exceed the maximum available from our supply on a dull day. The water heater also automatically responds to such things as fluctuations due to clouds coming and going over our panels, and tries to optimise so that to the greatest extend possible we only consume electricity from our panels when we are generating it.

One of the heater controllers. A secondary ESP32 connects to the master and makes decisions about whether to switch on the heater plugged into the socket depending on the temperature measured by the probe, the time of day and the amount of electricity being imported or exported.

Whenever the energy price goes negative (which it did for three hours around lunchtime today), all the heaters are told they can go on and consume as much electricity as they want, pursuing a higher than usual target temperature. The extra solar panels on our garage are also switched off by another secondary ESP32 device when there are negative prices.

This is what is installed inside the controller for the solar panels on our garage. The stripboard includes two transistors used to safely switch the 5 V inputs to the double relay board from our 3.3 V microcontroller. There's also an input for power at the top and at the bottom a connection for the pulse output from an SDM120 energy meter in the garage. A 62 mm long piece of wire connected to one end of the antenna (red object marked C3) makes the ESP32C3 wifi more reliable.

Does it work ?
This system really does seem to do what I designed it to do. But as I've only just "finished" it (what hobby project is actually ever really "finished") I don't have any substantial data yet. When I do, it'll be added here as an update.

Screenshot from my energy supplier's app showing yesterday's electricity price and our consumption per hour (including taxes etc.). Nearly all the system worked yesterday. It should be better than this from now onward. The peak at just before 4 am is from the kitchen boiler, which comes on as a convenience for us during whichever is the cheapest hour at night in order to make sure we have some hot water in the kitchen in the morning. The peak at 12:00 is for an hour when we had negative electricity price, and the hours in the evening are still a bit of a problem even though most of our panels face South West, because when the sun goes down we don't have our own electricity.

An example day (update April 2025)
This graphic shows the electricity that we bought and sold to the grid yesterday (March 31st 2025):

Electricity import vs. export on March 31 2025. The high spikes in the blue on the left are on the hours when the price was negative (note that the price shown here includes tax so almost never goes negative). On those same hours our solar panels were switched off so we did not export any electricity. Note the different scales for kWh on the left of both graphs - that's an artifact of the mobile phone app of my supplier.

During the hours of negative pricing, devices automatically switch on while the solar panels are switched off. We avoid using our own solar electricity on the hours when it has a positive price and do as much work as we can with the negative price electricity, resulting in exports having a higher value than imports. In March we consumed a total of 120 kWh from the grid with a cost of €31, while our exported 230 kWh had a value of €51. Our total energy bill for March was therefore negative €20.

Downloading ?
Once I know that this thing really works, and if people turn out to be interested, I'll make the code available to download somewhere. It's probably easier than starting from scratch. There's nothing amazing or patent-able about any of this - I'm not expecting to make money out of it. Mine is just a small contribution built on the shoulders of giants, including those who wrote the library code for TCP/IP, temperature sensors etc. I used to make a living out of writing code like that, but in the Arduino environment all the low level code for pretty much any device you want already exists. The work of others has made everything really easy to do.

Changing energy supplier
We've changed energy supplier from Pure Energy to Tibber. For the last few years Pure Energy have been paying us about €200 per year more for our exported energy than we paid them. I'm expecting a payment of about €280 shortly to cover the last year + one month that we were with them. Unfortunately, this happy situation wasn't going to continue: Pure's renewal agreement stated that they wanted to start to charge a fine ("terugleverbijdrage") of 14 cents per kWh exported to the grid, regardless of our trying to optimise our exports to be of the most use to them. This would total more than €400 per year. I tried to get a quote from them for a dynamic tariff (with prices per hour) and couldn't seem to get any sense at all from their salespeople.

I'm not expecting to be paid for dumping unusable electricity on the grid at midday in summer when it has no value. I've actually put quite a lot of work into trying to make sure that we don't do that, and to ensure that we use as much of our own electricity as we can, but I couldn't get any sense from Pure. Instead I've had to cancel their contract, which resulted in a snotty email in which they said they were sad to hear that we were no longer interested in using green electricity and gas ("We vinden het erg jammer dat je geen gebruik meer zal maken van onze groene stroom en/of gas!"). There's such a thing as "green gas" ? I don't think so.

Tibber, on the other hand seem to offer a straightforward easy to use dynamic tariff. I'm expecting that our total energy bill for the year will still be negative this year, especially given the recent optimisations detailed above. If you switch to Tibber and use the uitnodigingscode "dyd4ick5" then both you and I will receive €50 in credit for use in their online store - not that I've yet seen anything there which I think I would actually want.

Footnote 1: The ludicrous power of cheap embedded computers in 2025
I've often heard people say things like "this mobile phone has more power than the computers that took Apollo to the moon", but many people have no idea just how much more powerful those things are.

The little ESP32C3 microcontroller boards that I've been using have 160 MHz 32 bit RISC-V processors (including hardware floating point) which execute between 40 and 160 million instructions per second. They have 400 kB of RAM, 4 MB of flash, and include WiFi on chip as well as lots of other neat features. They consume less energy than I can reliably measure and they cost me $1.60 each.

The Apollo Guidance Computer was a truly amazing device given it's design in the mid 1960s. But while the performance was amazing then, it's not so amazing now: The AGC processor was clocked at just 1 MHz, used words 15 bits wide, executed "50000 to 100000 instructions per second" there was just 70 kB of (hand woven) ROM and 4 kB of RAM. They consumed 55 W and cost the equivalent of about $2 M each in 2025 dollars.

So the boards that I'm using execute instructions roughly 1000 times as fast as the AGC, they have about 100 times as much RAM and 60 times as much ROM (flash). And as my code doesn't achieve anything remotely as interesting as landing on the moon, these processors spend most of their time twiddling their thumbs in standby mode waiting for something to happen. By 1960s, 70s, 80s or even 90s standards this is a horrible waste of computing power.

Footnote 2: The ludicrous inefficiency of electric cars
I recently cycled past a garage and saw them advertising "175 kW charging". Yes, while we maintain our entire home on 5.5 kW, charging a single car can pull 175 kW from the grid. That's enough to blow the fuse of 30 homes like ours all at the same time !

Twelve years of rooftop solar power

Our rooftop solar power system was installed 12 years ago today. In total they've generated 40900 kWh since installation. While last year (2022-2023) was a record year for output from these panels, this year (2023-2024) was not. The end of 2023 and beginning of 2024 were marked by particularly grey and dull weather so this  output is not surprising. But the output of 3362 kWh over the whole year is still only slightly below the 12 year average of 3378 kWh per year.

Total output of our solar power installation per year. The blue columns are the contribution of the 12 year old rooftop system. The red shows the additional power generated by the extra panels we've installed on our garage roof.

The garage roof panels added an additional 1416 kWh to the total for the year. These were installed as we wanted to to compensate for the consumption of the heat pump and electric water heater that we installed when we got rid of our gas connection.

This winter was the dullest that we've recorded, resulting in the blue bar for March 2024 being easily the lowest in the graph. Even the substantial contribution of the new panels, shown in red, didn't result in higher total output than we have seen in brighter March months in past years from the rooftop panels alone. But the extra panels still helped us to generate a higher proportion of our consumption this winter than we have done without them.

Our rooftop panels having been operating for 12 years also of course means that the inverter has been operating for 12 years. The inverter actually only lasted for six years and three months before it failed due to poor soldering and the manufacturer refused to fix it. I fixed it myself and the repair that I made has very nearly doubled the life of the inverter thus far. I am still very irritated that ABB, the inverter's manufacturer, preferred to tell us that the whole inverter had reached the end of its life and needed to be thrown away and replaced with a new one when they could have made the same simple repair in order to keep it operating.

Anyway, the system as a whole is still working very well. It's difficult to work out exactly by which date this system paid for itself because the electricity price has changed over time. But the cost of electricity to consumers, including all taxes etc., has always been higher than 19.5 c, so I think we can now reliably state that the original purchase price of €8000 has been repaid by the solar panels and inverter together. Since 2012 the cost of solar panels has dropped precipitously and I expect the extra panels on our garage will cover their cost in under 4 years.

If you've read this far you'll probably also be interested in my blog post from four days ago about how the heat pump, electric water heater and solar panels together have reduced our energy bill to less than zero.

More more solar panels. Do we now have enough energy for a gas free home?

We now have four solar panels on our garage roof. They're at an angle so that they face exactly toward the south.

Today with help from a friend we installed two more solar panels on our garage roof. This means we have four 400 W panels on the garage roof to work alongside the sixteen 235 W panels which are on the roof of our home.

The original two garage mounted panels were in the shade until about 9 am so you can see from this graph that they suddenly "wake up" at that time. The new panels placed today do better a few minutes earlier as they'll be earlier out of the shade.

The roof of our home is oriented south west, while the panels on the garage are oriented directly toward the south so as discussed a few days ago they compliment each other. The garage is shaded by our neighbour's home early in the morning but as the new panels are further to the south and will be shaded less (even though to arrange this we had to push the older pair slightly further north) we're hoping that we see a little bit more electricity early in the morning than was previously the case.

The new set of panels, closer to the camera, are mounted at just 12 degrees, vs. the 24 degrees of the set which we put up last year. This will mean they have slightly lower output overall, but they will shade the older set behind them less often due to being lower at the back and they will catch the morning sun from the east a bit better due to their lower angle creating less of a self-shadow.

The usual "back of an envelope" design process

Last time I couldn't get commercially made hooks as everything seemed to be sold out everywhere. This time I used commercially made hooks to hold the solar panels in place as they were available inexpensively. Otherwise the frame which these panels are mounted on is very similar to that of the last pair of solar panels except that they're at 12 degrees from horizontal this time instead of 24 degrees. This is to decrease the chance of the new set of panels putting the slightly older set behind them in shade and to hopefully increase their output early in the morning when the sun comes from the east. We'll see if that works out.

So far as possible I collected the parts required for this job by bike. Three meter long pieces of wood do make for a slightly unusual sight on the cycle-path.

In total the bill for the two new panels, all the parts required to make the brackets and all the parts required to make a safe connection to our electricity supply added up to about €550.

Helping a friend with his installation a few days ago. He then helped me today. Doing things for each other certainly helps to keep costs down !

We now should have enough energy

Our gas supply was removed last week so we need to have a heating solution for next winter which does not involve gas. As discussed a couple of weeks ago, we actually didn't use much gas at all, so replacing it shouldn't require too much electricity. Added to the overproduction of electricity which we already had before they were installed, the new panels ought to be enough to make our net electricity consumption very close to zero for the year.

As it stands right now, our energy company is asking us to pay €5 a month for energy, with an expectation that we will have overpaid by €290 at the end of the year. That seems to be working out quite well !

Over the summer we installed the heating system which the two extra solar panels will supply, a poor man's heatpump. This was too inexpensive to attract a subsidy but it should be enough for us.

Eleven years of rooftop solar power - and it's a new record year

Our rooftop solar panels have been in place for eleven years, and the highest output year was the most recent. In total the rooftop panels have delivered 37591 kWh to date.

When we had our rooftop solar panel system installed in April 2012 we were told to expect an output of no more than 3150 kWh per year due to the angle of the panels and the direction they face on our roof. We were also warned that output would drop slowly over time. In practice we actually saw an average of 3357 kWh over the first ten years. Until now the highest output year was the second year after they were installed with 3516 kWh, but that record was broken in this last year, 2022-2023, which is year eleven for our system. No less than 3614 kWh of electricity came from our panels last year, which is nearly 3% more than the previous record.

We actually generated a little more than this because we added a couple of extra panels in September. But because these have only been operating through the darker months until now, they've only added slightly to the total, bringing it to 3780 kWh.

The new peak output wasn't the result of a particularly sunny winter. March, was particularly cold and dark, with snow and hail and produced the third lowest amount of solar power from our roof top panels since they were installed. Luckily, April has brought far more pleasant weather so far.

March 2023 was one of the darkest ever and even the extra panels didn't bring our total for the month to a total which was as high as the average over the ten previous years

The effect of panels facing in different directions

The extra panels on the garage face are installed facing directly south while those on the top of the house face south-west as that's how our house is built. This means that the sun hits the extra panels on the garage earlier than those on top of the house and that we have significantly more solar power earlier in the day now than was the case when we only had the panels on the top of the house.

Proportion of theoretical maximum output achieved by the solar panels on our house roof and those on the garage roof on the day of writing. Having panels facing in different directions flattens out the production curve meaning that we can cover our own usage for a larger proportion of the day.

Early in the morning all our solar panels are in shade, only receiving indirect light. The output of the panels on the garage suddenly come out of the shade of our neighbour's home at about 9:15, giving a rapid rise in output, today seen as a rise from from 5% to 22% of their potential. On the other hand, the panels on the roof of the house don't see a sharp rise due to an obvious shadow, but because of the angle of the roof they don't reach 22% of their potential on the same day until more than an hour later, around 10:30. This difference means that while on a day like this the output of the roof top system alone wouldn't reach 1 kW until nearly 10:45, adding two extra panels on the garage have brought that forward by more than half an hour.

Our garage roof at just after 09:00 this morning. These panels are at a 45 degree angle because that means they face directly toward the south. The sharp shadow line is due to our neighbour's home. The sun has melted the ice off of most of one panel and output is increasing rapidly as the panels receive direct sunlight. When we install two extra panels these two will be pulled back by about half the width of a panel and the two new panels will see the sun slightly earlier each morning than these do.

Doubling the size of the installation on the garage should mean on a day like this we can reach an output level of 1 kW by about 9:45 and 2 kW by just after 10:30. As such, two extra panels will address a source of slight annoyance - ever since the roof top system was installed we've observed that turning on appliances like our washing machine in the morning meant that we drew energy predominantly from the grid instead of from our solar panels, but with four panels facing south on the garage to take up the slack while the larger array on the roof "wakes up" this will no longer be the case - at least in summer.

No more gas so we will probably need more electricity than before

Our gas supply is being removed next week. We've already not used it for some time. In the future we'll probably need a bit more electricity than now so extra capacity is of course helpful.

Effect on a possible future battery installation

If we install a battery in the future, which we are considering in the future, it will have less to do because we will already have improved our autonomy by covering more of our morning electricity usage directly with solar power, thus reducing stress on a battery as it won't have to cover such a large proportion of our energy usage during mornings. I had hoped to have figures for the year so far showing improved autonomy compared with last year, but due to the very dark winter it hasn't been possible to produce those, so that's something for a future blog post.

Let's have more solar power

On Wednesday we brought home some solar panels and an inverter and we finished installing them on the roof of our garage today.

These panels are aimed to the south to try to pick up the maximum sun that they can over the whole day, but nearby trees will cause some shading.

We bought this set of parts from a local supplier, Cedel, so that we could pick them up. The panels aren't extremely heavy, at about 18 kg each, but they are large at 1.8 m x 1.1 m each and of course they're made of glass so quite delicate. As a result, this turned out to be one of the very rare jobs which couldn't be done by bicycle but luckily our son-in-law volunteered to drive with a trailer to collect the panels. Cedel sells a set of two 400 Wp panels with a Hoymiles HM-800 micro-inverter for €800, including a plug and play cable which can simply be connected to an electrical outlet, but I paid an extra €50 for an HM-1500 which can run four panels so that we can expand this system economically in the future. An expanded system will require wiring directly to the meter cupboard so that's another thing that I'll had to do if we decide to add those extra panels

Due to supply chain issues, standard mounting parts were not available. However, that's not really a difficult job. A quick bit of SohCahToa remembered from school and some literally back of the envelope calculations allowed me to work out dimensions for the substantial impregnated fence posts which I'd decided to use to construct the supports.

It turned out that when I included the 68 mm width of the posts in the calculation I could make the entire support with wooden parts of 1 m in length for all the horizontal parts and 0.4 m in length for all the vertical parts, dimensions which resulted in the minimum of waste from fence posts of a standard 2.4 m length.

Our garden fence consists of 4 m2 panels each of which is held up by two fence posts of the same type shared with the next panel. i.e. there's about one post per 4 m2. This has stayed intact through every kind of storm for at least 20 years, and the only work I've had to do was to re-enforce a few panels  which gone rotten at the base where they're buried in soil. So I think the same type of wood is certainly strong and resilient enough in much shorter lengths when holding a 2 m2 panel at a 22 degree angle from horizontal.

It's recommended that 22 kg of ballast be used for each panel in our location and at the height of our garage roof. These supports much heavier than the commercial systems, already weighing nearly 20 kg for each panel, and I've added quite a lot of tiles from the garden, each of which weighs about 8 kg so in total including the supports we have about 70 kg of ballast per panel. This should be adequate. Note that while that sounds like a heavy thing to put on the roof, I weigh nearly as much as the ballast and panel combined and when I walk on the garage roof all the weight is concentrated in the area of one of my feet. The solar panels mounted in this way spread their similar weight over about 4 times the area that I do when I walk.

This took about two days to do, including all the measuring, lifting, cutting etc. We were also delayed by rain. It's a fairly easy DIY job. And if the panels generate around 800 kWh per year, then at next year's 60 cent per kWh price it'll take less than two years for them to earn back what they cost.

The parts were ordered last Friday and ready for collection on Monday. It then took me two days to arrange collection and today I finished installing this setup on the roof of our garage. It's almost impossible to get anything done quickly by someone else at the moment due to covid related staff shortages, but if you're willing to do some DIY most things can still be done quite quickly.

That's not all

These aren't our first solar panels. Actually, we had 16 235 Wp panels installed on our roof ten years ago which still operate perfectly.

Our original 16 panels oriented south-west and still working as well as when they were first installed 10 years ago.

When the 16 panels were originally installed we expected that they'd fall a little short of covering our annual consumption of electricity, but actually they turned out to produce more than had been estimated and almost exactly the same amount as we used. Over time, our consumption has dropped quite a lot while production of the panels has stayed exactly the same for ten years, so for the last few years we've exported more electricity to the grid than we have taken from it. The only problem we've had with that system in the last ten years is that the inverter failed four years ago, but even though the manufacturer wasn't co-operative at all I fixed it at minimal cost and it's operated perfectly since then.

So you might wonder why we are now installing more panels. The answer of course is that we want to get rid of all our gas appliances. It's been known for many years that fossil fuels are endangering all life on the planet. This has to stop. We can't control everything, but we can control our own consumption. By continuing to buy fossil fuel we empower the companies that produce it and we just can't keep doing that. Insulating our home has already reduced our gas consumption for heating to less than an average small apartment, and we now cook entirely with electricity (the cooker only ever used a fraction of the gas that heating uses). We didn't replace our central heating boiler, which is old and inefficient, because buying a new boiler would lock us in to continued gas usage. Our low consumption is entirely the result of it not having much to do any more because the house is so well insulated. That means we could now economically replace our gas central heating with electrical heating, and that's even more true as of today with the extra pair of solar panels.

Heat pumps which could replace the central heating and heat the entire house in winter are still quite expensive and a bit hard to justify on price grounds, but we could install a small air conditioning unit which can produce heat with about the same efficiency as a heat pump in the living room as that's relatively inexpensive so that may be what we choose. We barely use any heating in the upstairs anyway.

A third smaller, experimental system

We've also been experimenting with smaller scale solar power system. One of our daughters lives in a flat in Assen. Unfortunately we can't fit solar panels anywhere on the outside. The balcony might have been an option but it faces north west so only receives sun just before sun-down which means it's quite a pleasant place to sit on a summer evening, but doesn't make any sense at all for solar panels. The rear windows face south east, but there's no place outside on that side of the building. As there are no really good options I decided to experiment with mounting solar panels inside the double glazed windows. After all, if you don't try it you don't know how well it works.

From the outside it's quite difficult to tell which window is full of solar panels.

They're just the right size to fill this window. Luckily there are other windows in this flat so this doesn't block all light.

The panels for the flat are a flexible type ordered from Aliexpress. I couldn't find them anywhere else. Luckily, one supplier has a depot in Poland which because it's inside the EU meant they were delivered to us promptly and without any unpleasant customs related surprises. These panels are a quarter of the size of the new panels that we've installed on the garage, just 1050 mm long by 550 mm wide. They're also really light in weight. As a result they're much easier to transport, I took both of these panels to the flat packed in their cardboard box by riding my bike with the box wedged under one arm. The inverter in this case is a Hoymiles HM-400, which operates quite well with two of these panels connected in series, though as it's rated for 400 W it's working well under maximum capacity.

These panels are nominally rated at 100 W each. Two of them connected in series and propped up in our garden roughly aimed at the sun on the reasonably sunny day when they arrived together produced 140 W, which seems quite reasonable. In the window of the flat we've not yet seen more than 90 W due to losses caused by the glass and the requirement to mount them vertically. Update later in September: we've now seen over 100 W on several occasions from the panels in the flat window. This might make them more worthwhile.

It's still an experiment. We want to help our daughter with the ever rising energy costs, but will this be worth doing given that the output is quite low ? It seems that mounting the panels inside roughly halves their output. This will vary depending on the glass, of course, and also on the panels due to different panels reacting to different frequencies of light. One smaller 12 V / 20 W panel that I have at home loses only about 20% due to being held inside a double glazed window, but unfortunately these panels are affected to a larger degree.

In total this system cost about €360. Half of the price was for the solar panels and the other half for the inverter. In the first week that they were connected, in September, we saw about 0.25 kWh per day from the panels mounted inside the flat. At the end of the year my daughter's electricity will cost over 60 cents per kWh, so that's worth about 15 cents per day. If we estimate that the average over a year is about half that much the electricity generated by these panels will be worth about €27 a year, which means a payback period of over thirteen years which I think is too long by any way of looking at it.

How about if we add two more panels (connected serial/parallel - making the equivalent of one 400 Wp panel which is still within the specs for the HM-400) ? Obviously this would block another window. It would also increase the cost of the system by 50%, but output would increase by 100%, or perhaps even slightly more if the inverter wakes up earlier due to the increased output of four panels. The result is a would be a payback period of almost exactly ten years, if the electricity price stays at 60 cents per kWh, which is still not great, but it's not far off what I originally expected from the roof-top system. Of course if prices continue to rise then the payback period will become shorter, but they may also drop.

For now I don't desperately need these panels for another purpose so they can stay put for a longer experimental period, or until my daughter says she wants rid of them. They'd perhaps be more effectively deployed on the south west facing wall of our home where they'd pick up the early morning light without their output being reduced by being mounted behind glass, and pay for themselves in about 7 years. I think a set of four of these would make a fairly good easily transportable system for someone who lives in a rented apartment and has a south facing balcony as they'd then pay for themselves in about five years.

Another really small system

I've had a small flexible solar panel on my velomobile for over ten years. In summer it's kept the battery charged so everyday use didn't require plugging in a charger. Unfortunately, the circuit got damaged and I had to replace it. I decided that this time I'd use a much simpler circuit - a shunt regulator with a zener and a single transistor to limit the voltage to about 7 V. Current is limited by the panel itself, the internal resistance of the battery and because once the voltage gets close to the maximum the shunt starts to conduct a bit. This circuit does a reasonable job of keeping a NiMH battery pack topped up and it can't damage that kind of battery. Don't even think about trying this with a lithium battery. I built two of these so that Judy no longer has to charge the battery for her velomobile either:

Velomobiles are the most efficient vehicles on the planet, maximising the potential of human muscle power. Judy's now has a small solar panel, just like that on my velomobile, which keeps the battery used for lighting and indicators topped up.

Of course I've not even tried to work out whether this is economical compared with charging from the mains. I had the parts and it's convenient for us.

Ten years of rooftop solar power - no decline in output can be seen

Today is the tenth anniversary of the installation of our mains connected rooftop solar panels. In total the inverter reports that 33935 kWh of electricity have been generated in ten years, an average of 3394 kWh per year since they were installed.

After ten years we still can't see any decline in output

Output in kWh per year. Note that there's considerable variation depending on the weather. The lowest output was 3126 kWh in 2018-2019 and the highest 3516 kWh in the second year. Last year was also higher than average, within 0.1% of the highest.

A rough calculation before installation suggested that we could expect around 3478 kWh of electricity each year under ideal circumstances, but our supplier suggested that in practice given the angle of our roof we should expect around 3150 kWh per year. The guarantee with the panels said that a decline in performance of 10% over the first 10 years was within normal bounds. This year's 3392 kWh is only slightly under the ten year average, but as it's higher than four of the previous years (including the first year) it's also far from abnormal. I don't see evidence of a decline in output due to aging.

We still produce more electricity than we consume

Blue and red bars show production and consumption per month. The yellow line shows the cumulative difference which we've exported to the grid. We started consuming less electricity five years ago, but over the last year we've consumed slightly more than the previous four years.

I painted our house last year, but I didn't clean
the solar panels. That doesn't seem to matter at all

Over the ten year period we've produced about 10% more electricity than we've consumed, but actually for the first five years our consumption was close to the production (for reasons explained previously) so it's more accurate to say that we've produced about 20% "too much" over the last five years. In other words, our overproduction over five years is roughly the same as the annual consumption of a household like ours.

Cooking with electricity changes the pattern

You'll notice that the graph showing the cumulative difference between our production and consumption has actually leveled off a bit over the last year. The reason for this is that we've switched to cooking with electricity instead of gas, this being one of the ways we've been trying to further reduce the footprint of our already low carbon diet. There's no new kitchen, that'll have to wait for a while. We've been using a small portable electric hob resting on top of the gas appliance as an experiment, which has worked out very well. An induction hob would probably work better. Anyway, it's nice to see that despite this increase in electricity consumption the yellow line is still heading upward at a faster rate than it did on any year before 2016.

Thoughts about home storage

A smart meter was installed four years ago so we now have three full years of smart meter readings. Though we have a single tariff contract so pay the same for electricity any time night or day, the smart meter nevertheless separates out the low and high rates of electricity which roughly correspond to day and night. Overall through the year, 39% of the electricity that we use comes directly from our own panels while 61% comes from the grid and we use 38% of what is generated by our panels while exporting 62%. Obviously a battery comes to mind immediately when looking at these figures, but I'm still not convinced that it's worth the investment. How much could it change these figures ? I think by less than we might hope:

Home storage batteries have capacities of around half of a day to a whole day of typical usage. It's enough to reliably keep your refrigerator going overnight during summer, but not remotely enough to span the seasons.

In the summer months such a battery would fill up in the first couple of days and then it would stay nearly full for weeks. The battery would allow us to consume our own electricity overnight, so that for periods of nice weather all of the electricity we use would be "our" electricity from our panels, which is of course an attractive idea. During that best case scenario we would increase from the average 60% of our electricity that currently comes from our own panels during summer to 100%. For the six best months each year we produce on average between 1.7x and 2.1x as much electricity as we consume so even with a battery working as well as it possibly could we could still only capture some of our own electricity for our usage and doing this would only actually increase the percentage of our own electricity that we consumed from the current still 30-35%% to about 50-60%. Also, this is Northern Europe. There are periods even during the nicest Northern European summers when the solar panels don't produce as much as we'd like, so the battery will run down and we would still have to buy some of our electricity from the grid.

New, huge, windturbine in Drenthe

And then there's winter: In the winter months there's far less sunlight falling on our panels and we possibly produce enough electricity to cover our own usage without annexing our neighbours' roofs. Because there's so little sun in the winter what comes from our panels currently covers only about 10-15% of our consumption and during these dark months that means we are already consuming 70-80% of our own electricity just with normal daytime usage. During the winter a battery would therefore be empty or near empty most of the time. While a battery would mean we would reliably consume all of our "own" electricity during winter it wouldn't win us much because there's not much to store: It would reduce the proportion of our electricity that we buy from the grid from the approximately 90% during winter now to around 85% with a battery.

Financially I don't see the benefit of a battery at all. It's an expensive gadget which will change very little. Let's also remember that the electricity leaving our home is not wasted. It's used elsewhere, reducing demand for other, on average less green, generation. So what exactly is gained by trying to keep our electricity to ourselves ? It seems more likely to benefit the ego than the environment.

So to summarise, I'm still not convinced that there is actually any real point in home storage using batteries. I think we'd possibly get some real benefit from installing a few more panels on the roof as while these could of course not do anything about our nighttime usage they would mean that we'd cover more of our own consumption during the day and if we get around to installing an electrical heat-pump for heating (instead of the old gas boiler) then balancing the resultant greater electricity consumption with more panels would make sense. I also have doubts about how wise it is to encourage people to install large and potentially highly flammable lithium batteries in their homes. They might be fine when new, but what happens in 20 years time when they've not seen maintenance engineers in many years and they're failing in various unexpected ways? I think it is a better idea to install batteries as large scale devices at substations or next to large solar parks or wind turbines.

Our business is also powered by these solar panels. We sell practical bike parts which help people, especially those in other countries where such parts are not so easy to find, to reduce their impact by cycling. Bicycles are the most efficient vehicles on the planet. We don't use motor vehicles so every shipment begins with at least the first few km on my bicycle, using nothing but human power.

Nine years of solar power. 30 MWh generated.

Today is an anniversary. Our rooftop solar power setup was installed on this day nine years ago. When the system was first installed I had two concerns, about the inverter and about the panels themselves.

Inverter
The biggest concern was about the inverter. I expected that this would have a limited life because an inverter is inevitably a box of power electronics which has to work quite hard. Power supplies, especially in my experience switched mode supplies, can be quite unreliable after a few years. I've fixed quite a few of them in the past, and switched mode supplies can be quite tedious to work on. An inverter is like a large switched mode power supply which works in reverse so I didn't have huge expectations for longevity and sadly the inverter has actually failed twice, first in 2018 and then in 2020. In both cases it was quite simple to fix and because I know which end of a soldering iron to hold onto I did that myself.

Panels
The other concern was the less well known longevity of the solar panels. There are two main ways in which solar panels degrade. The first is due to corrosion should they become damp and the second due to the sun light falling on the panels degrading them. I've had solar panels on homes that I lived in since the mid 1980s, starting with several 30 cm x 30 cm panels which I used to charge batteries. Those panels were not sealed from the weather so they got damp, there was visible corrosion in some places, and their output dropped markedly over time. But around the turn of the century I bought a 12 V sealed panel to replace them. I still have that on my garage roof and it's output is still close to the specification so I hoped that the panels for the roof be similarly long lasting.

The guarantee said that the output of the rooftop panels would still be at 90% of the initial level after 10 years of use. I realised soon after the panels were installed that their peak output was significantly higher than it the specification suggested. The rated peak output for the installation was 3760 W and the installers suggested that we'd probably not see more than about 3600 W because of the angle of our roof. But within a few days I was seeing ~3990 W. I had some concerns at that time that perhaps there had been a little slight of hand on the part of the manufacturer, who perhaps under specified the panels in order to protect themselves from guarantee claims. i.e. Perhaps we'd see the panels degrade by more than 10% over ten years but that they'd still be within 10% of 3760 W or even within 10% of 3600 W in ten years time, meaning that a loss of almost 20% of output would be possible without being able to claim on the guarantee. But this has not turned out to be the case at all. There is no degradation. I can happily report that the observed peaks last summer were still around 3990 W. What's more, we're not quite at ten years yet, but the total output from the system also seems not to have dropped even slightly. After the first three years passed I noted that the system had generated 10200 kWh in total. Six years further on the total is almost exactly three times this, at 30556 kWh. What's more, if not for the inverter glitch in 2018 which costs us about 250 kWh of output, we'd actually be ahead of the first three years by now.

Graph of output over time

The orange line shows our excess production over time. It goes up in summer and down in winter, and by March we just about break even which is why the last part looks flat. When the panels were installed we had teenage children at home and our consumption was similar to our rate of generation. You can tell how long ago our children left home from this graph (some lower power appliances also made a difference). The glitch three summers ago was caused by the inverter failing and my taking a few days to fix it. The second failure was a year ago in February and lost only one day's winter output so that isn't visible on the graph.

Finance
When the panels were installed I estimated that it'd take about ten years for them to pay for themselves. Because electricity prices change, we've changed our electricity meter once and supplier more than once, it's quite difficult to work out exactly where we are now so far as the return on investment is concerned. However it's quite easy to make an approximate calculation so that's what I'll do. I know that we'd generated 30442kWh up to the end of March, and of that we'd consumed 27767 kWh and exported 2589 kWh. Electricity costs us just under 20c per kWh so we've saved about €5300 from our electricity bill. With our current tariff we get pretty much the same for what we export so that's worth about €500. This leaves us with about €5800 of our original €8000 investment returned, and at the same rate the system will have paid for itself in about 13-14 years after installation. This is about the same result as I came to about three years ago. Just four or five more years to wait, then.

Note that in reality over a whole year we use about 40% of our own generation directly from the panels and export the rest of what we produce. The electricity company doesn't bill based on this, though, so for a purely financial calculation we don't need to take that into account.

Conclusion
I'm happy with the system. The inverter fault was disappointing and the manufacturer's response even more so, but I fixed that. The panels are faultless. Solar panels with lower performance were much more expensive when I first started experimenting with them in the 1980s but when we installed this system it was clear that they'd pay back within a reasonable amount of time. They're now not far from half the price that they were then so someone considering installation these days can expect to see their money back very quickly indeed.

Eight years of solar power. How much of our own generation do we consume ?

As of today our rooftop solar panels are eight years old. Here's the latest graph of electrical generation vs. consumption. You can see the eight years which have passed, the data neatly making sinusoidal shapes due to changing seasons caused by the rotation of our planet around the sun:

The blip in the graph is due to the inverter failing in 2018. It also failed in January but I got my soldering iron out sooner this time and because it was January and not the middle of summer far less was lost. Therefore no visible second blip.

The red bars show our consumption of electricity each month. The blue bars show the generation of electricity from our solar panels each month and the yellow line shows our total generation vs. consumption. Since the beginning we've generated more electricity each year than we used and in the last four years we've reduced our consumption making this difference between generation and use much greater. This led to the yellow line rising steadily higher. If you live in the Netherlands or anywhere in Europe connected to the same grid then it's possible that during the last eight years you've used some of the electricity generated on our roof as what we don't use goes into the grid.

Over the last eight years the system has produced 27039 kWh of electricity. That's 3380 kWh per year, very close to the estimated 3478 kWh per year predicted before they were installed. But this total over eight years was impacted quite severely by the period when the inverter failed two years ago. Over just the last 12 months we generated 3438 kWh, which is remarkably close to the prediction.

But while it's nice to see that we generate much more electricity than we use and the system is working very nicely, the graph also exposes an obvious problem. We never manage to generate even close to enough electricity in the winter, and of course there's never any output from our panels over night, which is exactly when we turn on our light bulbs. The installation of a smart meter a year ago has given me more data to work with and so now I can see how much of the electricity that we use ourselves comes from our panels and how much of our own electricity we use (these are not the same thing...). When we're not using our own electricity we're using what comes from the grid, and the Dutch grid is far from 100% renewable (we have of course signed up to a 100% renewable contract, but the effect of that is mostly only to make the paperwork say its renewable).

Data from the beginning of January 2019 until the end of March 2020.
Blue: our generation vs. consumption. Any month where this is above one we generated more electricity than we consumed.
 Red: the proportion of our consumption which was from our panels, varying between 0.12 in winter and 0.6 in summer.
Yellow: proportion of our generation which we consume, varying between 0.25 in summer and 0.75 in the winter.

This new graph which makes use of the smartmeter data shows a more complicated picture. Though each year we generate far more electricity than we consume (blue bars are often well above 1), our pattern of usage doesn't match the generation very well. We over-produce in summer, under-produce in winter and we keep putting our lights on when it's dark outside rather than in the middle of the day. We come closest to being independent of the grid in the summer when our panels generate more than twice what we consume, and though we don't use much of our own electricity (yellow bars) it does make up nearly 2/3rds of our consumption (red bars). On the other hand, while in the winter we consume most of the electricity that we generate, there is very little of it so it accounts for only about an eight of our consumption, the other 7/8th coming from the grid.

As you can see, having solar panels on the roof, even if they generate far more electricity than we use, does not make us even close to being independent from the grid. What's more, there's no way we could make ourselves so.

Batteries
It is of course possible to buy a relatively small domestic battery which has enough capacity to store about one day's supply. If we had one of those it would allow us to consume exclusively our own electricity for March through to September, pushing the red bars for those months up to 1. However in the winter months it would be nearly idle because we then generate on average only about a quarter of our consumption, falling to a sixth in the worst month, and we are already using a high proportion of our generation (the yellow bars reach as high as 0.9), so the battery could do almost nothing to help in winter.

To be independent from the grid through the winter we'd need to store electricity from June, July, August and September to fill the trough in our production between October and February and there is no possible storage system which could do this. At present we could buy a small unit which we could fit into our garage, costing a not inconsiderable sum, to cover ~24 hours usage but to store enough to go through winter we'd need something about 120 times as large and 120 times as expensive. It's just not a sensible idea on any level.

Quite apart from the cost and I have reservations about the reliability of batteries for home storage. We already know that, as I expected in advance, inverters are not entirely reliable. It's a box of power electronics which works hard and so we should expect less than perfect reliability. A domestic back-up battery adds far more complication and is almost certainly less reliable. Why ? Because such a battery installation would require not only its own inverter circuitry similar to what we already have in association with the solar panels, but also a battery charger and the battery itself. More power electronics means more to go wrong.

Doubling the number of panels on our already full south facing
roof isn't an option. North facing panels won't work so well.

Such a battery would more than double the price of the solar panel system, more than double the embedded carbon cost, more than double the chance of it going wrong and requiring costly maintenance or replacement as a result. None of that is good news for us or for the planet. Even given all those constraints, it seems that the manufacturers' somewhat optimistic assumptions still suggest that a battery could only provide us 100% usage of our own solar power if we also at least doubled the number of solar panels that we have. As our roof is already almost full this would require us to annex our neighbour's roof entirely for our own use. That's not going to happen.

Let us remember that the excess that we generate and do not use is not lost. Investing too much in trying to use our own electricity ourselves is almost certainly counterproductive. A good proportion of our excess is consumed by other electricity users, both domestic and industrial. This leads to those consumers having lower emissions from their electricity consumption.

Reduce consumption!

A domestic battery makes no practical sense and doesn't necessarily make sense for our environment either. The best thing for us to do, and I would suggest for everyone else as well, is to reduce energy consumption as much as possible. We're continuing on that path, making our home more efficient every year. The less we consume, the easier it will be for our grid to operate without fossil fuel input. Any additional loads will keep the fossil fuels burning for longer.

Domestic and commercial
Our domestic solar panel setup is not actually 100% domestic. We operate our business, Dutch Bike Bits, out of our home, so the business is also powered from our solar panels. We try to run our business in the most ethical way we can. Therefore there is no extra energy consumption in another building and we don't use any form of motorized transport so our parcels, all of which contain goods to support people who cycle and are therefore assisting other to also use the most efficient vehicles on the planet, begin their journeys by bicycle.