OpenEnergyMonitor Community

Optimisation of Solar usage across the whole house - how to?


I have been kicking around how to sequence/schedule loads in the house to match the predicted solar output for the next day.

I can capture/profile each of the major loads in the house such as

Washing Machine
Clothes Dryer
Hydronic HeatPump
Pool Pump Etc

I run each of these major appliances either through a Sonoff POW or on a dedicated circuit with an IOTAWATT capturing what they are doing.

My idea is to firstly take loads that can be variable and/or started stopped and/or automated.

So the Pool Pump and the HeatPump both fall into this category

Then we would have mandatory loads that need to run every day - the dishwasher falls into this category.

Then we have optional loads that are needed but are flexible - so the washing machine falls into this category as does the clothes dryer.

If i take each of these devices i can create a simple energy “profile” by storing their power draw at say one minute intervals and recording this.

Lets say the standard dishwasher cycle takes 3 hours - i know that is uses short bursts of intense power (whilst heating the water) during the first two hours and then in the last hour more sustained bursts whilst it is rinsing and then drying the dishes. Once the dishwasher has started it can not be stopped (without resetting the cycle) - so powering this on and off as power is available is a non-starter.

This is similar to the washing machine which can not be stopped once it has started or it forgets its place in the cycle.

I know the pool pump draws a constant load and can be turned on and off throughout the day as excess solar is available - similarly the HeatPump.

How would people approach creating a framework for the sequencing of these devices across a a day to enable optimum self consumption and/or minimal demand tarriff/TOU sequencing ?

In my mind i see something like an excel spreadsheet with (say for the dishwasher) 180 columns for the 3 hours of operation - each column would have an energy value

I would then do the same for all the other devices

You then want to somehow slide each of the devices sideways until you end up with the optimum sequence/overlap so at any point in time the energy total is less than the solar output whilst still achieveing the days goals

Ideas anyone ?



If any UK forum user has successfully/accurately predicted solar gen by time of day then it would be great if they were to post details on this forum.

I spent many days earlier this year trying to do that incl using very elementary machine learning but failed and gave up – beaten by UK weather variability.

If yr objective is to maximise use of yr solar gen then have you considered investing in a battery? In default mode, a battery charges when there is excess solar gen and discharges when the load exceeds the solar gen. So this greatly simplifies the load scheduling problem.

If yr objective is also to minimise cost then have you considered using a Time of Use tariff eg: Octopus Agile?

And optimising operation is complicated by virtue of the fact that average winter solar gen is approx 25% that of summer solar gen.

These comments are based on my experience with an installation – 114 solar panels (30kWp), batteries (40.5kWh), ASHP heating an outdoor pool and two EV’s.

1 Like

Thanks for the reply.

  1. Yes we are investing in a battery - 3 of them actually - they are on the way
  2. I am in Australia not the UK so slightly different - but we have a similar company to Octopus - called Amber - they however expose us direcly to the wholesale prices - which is the only way that a battery makes economic sense - however they do not do TOU as such but constantly variable prices based on the wholesale rates.
  3. We have size our 25KW solar array for the Winter season specifically
  4. I have started to use Stephens Solar Prediction scripts on here and they are showing a lot of promise for my setup

I am working with another developer on a different forum using Node Red to try and come up with a scheduling and optimization algorithm and was intersted to see if anyone on here had done anything on the automation and optimization front.


1 Like

I would suggest you take a look at this group on facebook - Redirecting...

Its for battery systems and they’ve built a Pi based flow to charge batteries based on expected sunlight

I wasn’t aware of Amber - I’ve been looking for an offer like this in Australia, as I have a 10 kWh battery. I’ve just signed up! I have been looking at next day solar forecasts but at the moment they are too approximate for my use. Currently I manually overide the settings if it looks like solar+battery won’t see me through a day/night cycle. Usually I will start a grid trickle charge overnight to bring the battery up to say 40-50% then rely on solar to supply the load with support from battery. This is rarely needed in the sunnier months though.