Author Archives: admin

Solar output – 2017 Jan – Jun

2017 H1

Now that we’re half way through 2017 it seemed appropriate to have a look at energy usage from the solar panels – especially as those six months reflect the first six months with the battery storage system.

The graph is taken from my ImmerSUN smart controller which automatically diverts surplus solar electricity to the car charger or immersion heater.  The battery storage system has independent controls but its benefits can be seen via the ImmerSUN.

The purple line shows the consumption of electricity (excluding the immersion heater) and is relatively stable month by month.  Consumption is relatively large due to my electric car and cooking with electricity.

The green line shows the generation of electricity from my solar panels.  Not surprisingly output is lower in the winter, but from April we generate more electricity than we use despite our relatively high consumption.  In principle we could be electricity independent during those months but for the time of consumption not matching the time of generation.

The red line complements the green line as it shows the import or purchase of electricity from the grid, and thus reduces as the generation rises.

The blue line shows the diversion  of electricity to heat water via the immersion heater when neither the battery storage system nor the car charger can absorb the available electricity.

Finally on the graph the turquoise line shows export of electricity to the grid when all smart capability within the house to use electricity is exhausted i.e. battery storage system at maximum power or full, electric car battery full or absent, and water in cylinder is hot.

Among the numbers:

‘Savings” at £80 refer only to the value of the water heating achieved from solar electricity versus buying electricity (although our backup is mains gas).

“Self consumption” at 86% refers to the proportion of solar panel output used i.e. not exported to the grid.

“Green contribution” at 59% refers to the proportion of total electricity consumption (excluding water heating) derived from the panels rather than from the grid.


Battery storage – the first 6 months

I’ve now completed 6 calendar months with the battery storage system.  Those six months cover January – June so might be considered representative of the year as a whole.  I have real-time monitoring of solar panel output and house consumption / export so I have clear visibility that the battery is working and storing energy, and then discharging energy through the evening as I see the house with near zero electricity consumption sometimes through into the early hours of the morning (less if we’ve cooked our evening meal on electricity).  So, environmentally it’s doing its stuff, but what is it saving financially?

There are at least three ways to assess that, so let’s see how they look.

Firstly, from the capacity of the battery (4 kWh), and assuming that it’s filled daily one can calculate a benefit.  That benefit would tend to overestimate benefit in winter when there may not be enough energy to fill the battery, but equally could understate saving some days when the battery goes through repeated periods of charging and discharging during the day.  One could imagine the heating cycle of the dishwasher, for example, perhaps causing some discharge of the battery during the day if the washing machine load isn’t met from the solar panels, but then then re-charging before the evening and thus its daily throughput being higher than it’s capacity as some of that capacity is used more than once per day.

So, if the battery stores 4 kWh and there are 365 days in a year where each kWh not bought is worth 11.5 p/kWh then the saving might be 4 kWh/day x 365 days/year x 11.5 p/kWh =  £167.90.

Secondly, let’s look at electricity savings.   If I compare the first 6 months of 2017 (with a battery) to the first 6 months of 2016 (without a battery) then electricity purchase has reduced by 824 kWh (38%).  Thus the saving could be 824 kWh/six-months x 2 six-months/year x 11.5 p/kWh =  £189.52.

Of course, as my chart shows, there are other changes that potentially impact electricity use between those 2 time periods, so that might be an overestimate.

The third and final way that I’ve analysed this is to look at the data about how the output of the solar panels has been used.  In 2016 I used 44% of the output of the panels to replace bought electricity, and a further 28% of output to replace gas consumption.  In the first half of 2017 however I used 63% of the output of the panel to replace bought electricity, but only 24% of output to replace gas consumption.  The reduction in use for water heating reflects the prioritisation of the battery over water heating as a kWh of electricity purchase avoided is much more valuable than a kWh of gas purchase avoided.  Against 2016’s full year generation of 4,192 kWh that gives a saving of only £87.84 which is much the lowest of the 3 figures.  However this only accounts for more efficient use of the solar panel output, and not winter savings from shifting energy purchase from day to night time when it’s cheaper.

Whether any of these figures represents a saving over the life of the storage system entirely depends on the lifetime of the system, the life of the batteries inside it, and the replacement costs of those batteries

Adding smart boiler control

In my previous post I described replacing conventional thermostatic radiator valves (TRVs) with smart valves (sometimes called eTRVs).  The smart valves include both temperature set points and a schedule which allows me to operate shorter on times in rooms not used so much – such as the playroom heating turning off after my daughter’s bedtime.

My latest update is to link the valves to the boiler so that heat demand from a smart valve fires up the boiler, regardless of the settings of the older central timer and hall thermostat.  That would mean, for example, that if my wife want to watch a late film then commanding heat in the lounge would restart the boiler even if outside normal heating hours.

The effect of this change can be seen in the attached image which shows three days of valve position information for the two radiators in the lounge: two days where the boiler was enabled by the conventional timer and central thermostat, and the third day with smart boiler control.

For the first two days you can see the valve open wide for an extended period during some of which time the boiler won’t be pumping hot water as the hall is up to temperature.  However on the third day, with the link to the boiler, the valve closes very quickly from its initial position and then modulates to maintain the temperature since the boiler is running all the time when any valve is open.

The system is controlled by two rules through Apple Home:

  1. If any valve moves off closed (triggers)  then enable boiler.
  2. If any valve moves to closed (triggers), and all valves are closed (conditions), then disable boiler.

The picture shows the actual mechanism to turn the boiler on or off via the Elgato Eve Energy (which is a switchable mains outlet and energy meter) in the right side socket outlet.  I use the Eve Energy to operate a mains relay (in the black box) which in turn closes a contact between two terminals of the heating wiring box, which bypasses the heating timer and hall thermostat, sending a mains control signal to the dual port valve for the heating thus opening the valve thereby enabling the boiler through the existing controls.

The first evening’s operation showed two issues:

  1. Room temperature was reported as overshooting in some rooms, but not in one room with a newly installed valve. It may thus be that the older valves have self-tuned their controls and need to re-tune to the new more dynamic system characteristics.
  2. I needed to manually turn down the radiator in the hall which was getting too hot.  If that persists then I may need to add a TRV or smart valve to the hall.

Internet of Things

Last night I made a small update to our heating system by adding some smart radiator valves. I’d been thinking for some time that there were efficiencies to be made regarding what rooms were heated when. Until now we’ve had a 7 day heating timer (so we have different heating schedules on weekdays and weekends) and thermostatic radiator valves (so we can set specific temperatures in each room) but now we’ve gone a step further.

Radiator valve

Radiator valve

These days there are in my opinion three types of radiator valves:

  1. Traditional proportional valves – these valves allow the flow to a radiator to be set manually, but there’s no control to maintain a set temperature. So if for example a room is south-facing it may get too hot on a sunny day as no account is made of the solar gain, or a relatively exposed room may get too cold on a windy day as no account is made for the extra heat loss.
  2. Thermostatic valves – here the user can set a temperature for each valve, and then internal expansion or contraction of the thermostat reduces or increases the flow through the radiator to maintain the set temperature; but all radiators heat at the same times as set by the heating timer.
  3. Smart valves – smart valves add the ability to schedule temperature and/or on and off periods in different rooms at different times.

Valve Schedule

In my case I identified 3 rooms (5 radiators total) in which I thought that the typical usage was different enough from the house as a whole to warrant smart valves. For the purposes of illustration only, I’ve also added the schedule for the boiler timer although this is programmed independently of the radiator valves. The three rooms are:

  1. Lounge – we don’t use the lounge on termtime weekday mornings.
  2. Master bedroom – we don’t use the room during the day, so the heating can stay off until towards bedtime.
  3. Play room – my daughter doesn’t use her playroom before nursery or after her bedtime.

The chosen smart valves are Elgato Eve Thermos which are Apple HomeKit compatible but are also configurable via Elgato’s own App; but not configurable via non-Apple devices. I initially tried setting up the required sequences via timed scene changes, but couldn’t see an easy way to schedule both workdays and days off; so I ended up downloading schedules directly into the valves via Elgato’s own App. This allows different schedules to be established easily for both working and non-working days with a schedule to identify non-working days set in as a calendar in my iPad. It’s easy to set up a recurring schedule for weekends and then other dates like bank or school holidays can quickly be added. Vacations when we’re away from home are still set up via timed scene changes.

Latest software update

Part of the ladder logic program in the PLC

This afternoon I’ve spent a little time working on the charger. The charger stopped working on Monday after the power was interrupted for an hour or so while other electrical work was completed. A quick check at the time concluded that the radio link between the immerSUN and the charger had stopped working, and that the problem wasn’t a fuse, but other time commitments prevented further investigation until now.

The radio link carries the control signal from the immerSUN in the airing cupboard to the car charger in the detached garage.  The control signal is generated either when the immerSUN has detected that a suitable surplus of PV electricity is available to want the car charger enabled or its output current increased; or when a timed boost is programmed in the immerSUN typically to do overnight charging on cheap rate Economy 7 power; or indeed when a manual boost is demand via the immerSUN front panel, app or web portal.

Fortunately lack of the control signal doesn’t prevent car charging as I can disable the integral PLC (which leaves the charger on continuously) and allow the car’s internal timer to control when the car charges.

This afternoon I’ve successfully re-paired the receiver to the transmitter to restore normal charger operation and uploaded my latest software to the PLC.

The capabilities of the charger are now as follows:

Operating currents: 6, 10 and 16 Amps
Maximum continuous current (continuous grid load current):10 Amps
Peak current (from PV + battery only):16 Amps
Range switch interval:6 minutes
Minimum on time:10 minutes
Control signal watchdog (not during Economy 7 hours):15 minutes
Control source:ImmerSUN

Battery profile through the day

January 5th was a good day to demonstrate the battery’s contribution to the home on a day on which 9.75 kWh was generated. The battery at this point in the year is configured to charge on cheap overnight electricity as well as store excess PV electricity (when there is any). The day also included evidence of water heating and car charging after I returned home at lunchtime.

Over all only 2 kWh of power was purchased during the day at full price, while 12 kWh of cheap night time electricity ran the dishwasher and washing machine, did some car charging (I charge for an hour which leaves headroom for later solar charging), and contributed 5 kWh to charging the fixed battery.

During the relatively sunny December day, as PV output rises, the battery starts to recharge and then, once the battery is charging at its maximum rate, remaining PV output is used to heat water (although described as ‘Grid Power Out’ in the chart above).  At around 12:30 I arrived home and put the car on charge (the blue spikes) and made lunch (the brown spikes).  The blue spikes of car charging occur as the chargers turn off intermittently to allow available power to be assessed.  As PV output falls EV charging starts to require limited support from the battery (the rising brown line through the blue spikes) until eventually all EV charging stops and the system returns to a combination of battery charging and water heating.

As the sun sets the brown line rises again as the battery takes on the load of the house. Green spikes indicate boiling the kettle before 18:00, cooking dinner from shortly after 19:00, and a further kettle boil at around 23:00.  Apart from these import events totalling less than 2 kWh, the house continues running on stored energy from the battery through into the next day when the cycle restarts.

At some point I assume that it will be necessary to curtail the overnight charging of the battery so that it doesn’t miss out on day time charging from PV as a result of being full. That may be responsible for some of the blue cycling with the car charging associated with the rising brown line – if the battery is already full then the car charger will be enabled at lower levels of power generation potentially causing the battery to partially support the car charging.

Bought Electricity 2016

Yesterday I produced a graph of the number of units of electricity that I’d bought or imported each month since we changed to Economy 7 metering back in October 2015 – a change that I made to reduce the costs of charging my electric car.  I rather surprised myself.

Key features that caught my attention were:

  • If I ignore the months of significant solar generation where significant electricity is available that isn’t bought, then from November 2015 to March 2016 and November 2016 to December 2016 there are 6 months of consecutive month-on-month reduction in bought electricity.
  • Comparing November and December between 2015 and 2016 (the only 2 months available for direct comparison) there’s a reduction of a round a third in bought electricity.
  • Comparing December 2016 with January 2016, bought electricity is down by a quarter.

That seems to be a compelling case for a significant reduction in import having taken place. Such a reduction could be combination of 2 sorts of things, firstly fundamental reductions in electricity use such as through having a more energy-efficient appliance, or secondly shifting use from paid-for electricity to free solar electricity (albeit that the months in question are winter months where less solar is available).

Potential contributors to this effect are as follows:

  • November 2015 to March 2016 – Increasing availability of the garage for car charging. In November 2015 I demolished an internal partition within the garage that had rendered the remaining space too small for a car.  That created an open double garage which allowed a car in for the first time since the partition was created in the 1980s. In March a suspect garage door was replaced making routine access much easier since the old door was badly corroded, required considerable effort to lift, and had an odd locking arrangement.  Thus during this period garage use went from 0 to 100% utilisation for car parking/charging. This might seem an odd item to include but my rationale is that if the electric car is in the garage then it will be warmer than if outdoors and charging will be more efficient as there will be less heating and more charging. Charging also has potential to be at higher current in the garage due to the availability of a Mode 3 charger which I think is also more efficient.
  • December 2015 – gas boiler replacement. Again perhaps not obvious at first sight, but if the new boiler gets the water hotter then less work needs to be done pumping water round radiators to take place to deliver a certain heat output at the radiators.
  • January 2016 – fridge freezer replacement. In January we replaced the former separate fridge and freezer with a single combined fridge-freezer. Although considerably larger in combined volume, the new fridge freezer replaced a freezer that I’d had for over 25 years so I anticipate an energy saving there.  The new fridge-freezer is rated A+ at 496 kWh/annum.
  • April 2016 to September 2016 – charger control project. Development of the charger control project shifts some electric car charging from overnight bought electricity to daytime free electricity even in winter. Previously even on a sunny winter’s day I probably wouldn’t have charged in the daytime due to a risk of import ruining the economics, but now I just leave the car plugged in virtually all the time and let the charger run automatically if and when sufficient free power is available.
  • July 2016 – double oven replacement. In July the failure of the fan on the oven prompted a oven replacement as the old fan was inaccessible for replacement due to corrosion of the surrounding bolts. I’m not sure how old the prior oven was (although I have evidence of the kitchen being remodelled in 2005) , but anticipate an efficiency gain due to replacement.  The new double oven is rated A/B – i.e. the smaller oven is efficiency A and the larger one efficiency B.
  • December 2016 – storage battery. Installed too late in December to have much of an influence here, the ability of the battery to capture electricity that would have been exported as surplus to requirements for later use should reduce bought electricity. The secondary ability of the battery to storage cheap night time electricity for later day-time use is likely to be of interest only in the winter months, and then won’t reduce demand (indeed it will increase demand slightly due to its round-trip efficiency) but it will reduce costs.

It will be interesting to see how this develops through 2017.

Electricity use 2016

The immerSUN provides a useful app showing electricity use which includes an annual option.  The graph below shows the 2016 annual data:
Although some of the data was only collected from mid-March 2016, the graph still shows useful information. I think that the graph overstates bought / imported electricity in January to March but understates generated electricity proportionately in the same period.

The purple line shows monthly electricity consumption and is broadly consistent month-to-month.

The green line shows the generated electricity from the solar PV system.  Its seasonality is clearly visible.  Solar generation exceeds electricity use in four summer months, and is very close in a fifth.

The red line shows bought electricity.  It’s generally a mirror image of the green line reflecting more purchased electricity in winter and less in summer, but is not zero even in months where generated electricity exceeds used electricity due to time of day issues – cooking and car charging often occur at times when solar output is low such as cooking in the evening and charging at night.

The blue line shows surplus day-time electricity being used to heat water, and thus saving gas.

The turquoise line shows surplus day-time electricity being exported once the water has reached its set point.

It will be interesting to see how this changes in 2017 as a result of a full year of solar car charging in its current mature condition and with the new battery storage that should help get more of the generated electricity used by saving it for evening use.

Solar PV Generation 2016

The end of 2016 marks the end of the first full calendar year of solar PV operation in this house since the system was installed in late September 2015. In 2016 we’ve generated 4,129 kWh – 12.5% more than the 3,668 kWh annual output estimated for the system.

Income for 4,129 kWh generation with 50% deemed export is £641, with further savings made from the self-use of the electricity rather buying electricity or gas.   Detailed monitoring of these savings only started in March 2016, so there’s less than a full year of data, and my ability to use generated electricity improved through the year – particularly with my electric car charger control project.  From March to December 2016 we used 44% of the generated electricity replacing bought electricity, and a further 28% of the generated electricity replaced bought gas for hot water.  Back in September at the anniversary of the solar PV installation I estimated these energy savings as a combined £200 (link).

The chart above shows the minimum, mean and maximum daily generation for each calendar month since installation.

As one might expect, significant seasonal variation is evident month-to-month and well as day-to-day variation within any month. June 2016 looks a bit disappointing compared to May and July. The last few days of September 2015 immediately after the panels were installed were obviously quite good for September, but within the spread for September 2016. October looks quite similar for 2015 and 2016, while November 2016 is rather better than November 2015.  December again looks quite similar for both years.

Solar panel power outputs deteriorate with time.  My panels are supposed to have at least 90% of the original minimum power output after 10 years, and 80% after 25 years.  Generally mine seem to be performing well with annual generation significantly higher than estimated, and no evidence of performance deterioration in the late September to end of December period for which 2 years of data exists.

Battery capacity

As you may recall my storage battery has a usable storage capacity of 4 kWh which, with the quoted round trip efficiency of 80%, would mean it takes in 5 kWh of energy to make 4 kWh available for output later.

We’ve been away for a couple of days, leaving the house drawing a low amount of power in our absence, however yesterday the battery reports having discharged 5.9 kWh despite its 4kWh capacity.

I’ve already had some days when throughput was greater than capacity, but that’s generally been when the battery charged fully overnight and then partially recharged from solar PV during the day. However this was not the case yesterday where, although there’s a dip in demand associated with limited daytime generation on a dull winter’s day, there’s no evidence of battery charging.

There is a characteristic of battery behaviour that may describe some of this which is described by Perkert’s law. Peukert’s law describes a relationship where battery capacity reduces as current increases according to a power relationship. Since my battery is discharging at about 350 VA that much less than its 1200 VA maximum capability so it may appear to have more capacity.

However it seems to have both greater input and output energy than its rated usable capacity suggests (input energy remains higher than output energy) so it would seem that there’s something else going on here; whether that’s an early life effect, a genuinely larger capacity than nominal, or some other effect is unknown.

The battery has a nominal capacity of 8.8 kWh to provide is usable capacity of 4 kWh, so it’s possible that some characteristic of the control software is making more of the nominal capacity available than would normally be the case.