The latest version of our dynamic heat pump simulator now supports:
Available: openenergymonitor.org/tools:
I’ve been using this tool over the last week to simulate what the effect of improving control settings could be on some of the Electrification of Heat systems, e.g this period for EOH2578: Emoncms - app view.
Here’s the real data:
and here’s the simulated version:
Tuning the weather compensation and running the heat pump all day could improve the COP and reduce running costs by 20%:
This heat pump is a bit over-sized hence continued cycling even at minimum modulation.
We might be able to save 10p with a bit of a setback (using auto adapt control mode here rather than fixed speed):
Cosy costs would be similar with this simple setback schedule but deliver savings with over-heating during cheaper periods, e.g almost a £1 saving here vs continuous mode (18%) or a £2.31 saving (34%) vs the initial poorly optimised weather comp base line.
Thoughts very welcome!
It’s possible to explore all sorts of things with this!
This is great, and very helpful for explaining how heat pumps work best to heat-pump-curious but gas-boiler-indoctrinated users who can’t wrap their heads around best-practice methods.
Would it be possible to add custom ToU tariff pricing to this tool, and to the broader heatpumpmonitor platform? Or at least add the Tomato Lifestyle tariffs, which are very popular with heat pump owners without EVs/solar etc.
Thanks @ectoplasmosis !
Yes it’s possible to enter custom ToU tariffs, it’s tariff pricing and timing is fully editable. There an option to load cosy as an example but you can change everything from there..
Great, thanks! Can this custom tariff comparison also be done when viewing monitored installations via heatpumpmonitor.org?
HeatpumpMonitor.org needs pre-set tariffs, as tariff cost processing is done as a background process over-night. The best thing there would be to add these further tariffs to that calculation process.
Understood. Would it be possible to add the Tomato Lifestyle tariff, with 01:00-06:00 @ 5p/kWh period, 09:30-11:30 and 20:30-22:30 @ 14p/kWh, and 23p/kWh at all other times?
Yes sure, Il make a note of that, it might be a few days
It’s very cool! Well done! To be able to instantly test a new scenario is quite valuable.
I said this before but it would be nice to have an MPC controller so you can make use of what you learned about the building. Not easy to implement though!
Speaking of, how are you determining the building fabric parameters? It doesn’t say explicitly but I assume these are the thermal masses and resistances in between?
Thanks Stefan
Definitely interested in this, While the PID works to a degree, it’s not producing exactly what we see on real world heat pumps, so I’d like to get to something better.
Yes just on simple 3 stage RC model on the building side of things dynamic/docs/dynamicmodel.md at master · emoncms/dynamic · GitHub
Even this simplicity can give great reproduction:
Thanks! I am mostly curious if you have come across any good method to identify the thermal parameters of a building. All the data to do so automatically should be present in heat pump monitor.
Second point and quite important is how to determine the initial conditions. What temperatures do the masses have when you start the simulaiton? In my own simulations I found that increasing the temperature of any significant thermal mass, mainly the envelope, will have a dramatic effect on the results. You will also see this in your own data: a day at -5 C preceded by a string of cold days will require more heating than a day at -5 C preceded by a string of mild days.
Not yet, agreed that would be nice! What I would like is the ability to drop an Emoncms MyHeatpump dashboard link into the simulator tool and for the tool to then automatically tune the model based on that data. Maybe something for the future.
Agreed, when you open the simulator it runs the simulation over 20 days (at 30s interval), the final result is based on the average over the last 4 days. This seems to be enough to remove the sensitivity to those initial starting conditions.
Agreed, whenever a change to conditions is made in the simulator, a further 3x 4 day = 12 days total are simulated. Even this can not be quite enough and that’s why there is a “refine” button, just to make it possible to run the same conditions for another 12 days per hit.
My own experience with tariffs (referred to in other posts) shows that smooth running of my heat pump, on a constant price tariff (benefitting HP, not rest of house) compared with a time of use (benefitting all usage during certain hours) leads to:
At the start of this year (end of Christmas holidays! time flies!) I did a little bit more work on the dynamic heat pump simulator but did not update this thread about it at the time. I did discuss some of the results from the new features in this thread here: What SCOP can you expect from a system that runs at 55C and 50C flow temperatures on the coldest days?
The latest version now includes:
The main features that are still really needed are a realistic defrost model and hot water cylinder simulation.
Check out the latest version here: https://openenergymonitor.org/tools/dynamic_heatpump_v1
I thought I’d share an interesting result on the topic of continuous vs. intermittent heating. It’s so difficult to test this in the real world, given how much outside conditions change on a day-to-day or week-to-week basis and how the thermal mass of a building will always be at varying levels of “thermal charge.” Apart from running such scenarios in a controlled environment, such as the Salford Energy House, I think simulation tools are probably our best option to explore this question. I think the heat pump industry would benefit from much more sophisticated simulation tools to explore these questions.
In this example I created a baseline scenario with a 20C all the time schedule and then compared it to a schedule that effectively turns the heat pump off over night between 10pm and 8am.
Another interesting variation is a schedule for folk that are out of the house all day until the evening, the simulation suggests that letting the heat pump turn off between 10pm and 2am and then come on at a lower level during the day before boosting up to 20C after 3pm can save almost 17% in terms of electric consumption and fixed tariff costs. The SPF drops from 4.55 to 4.08.
On Agile again however this is more expensive still (lower comfort is 5% more expensive than continuous heating this time), which we would expect given that consumption is more focused on peak time hours.
This is an interesting result as it contradicts the simple rule of thumb that continuous will always be better than intermittent heating for heat pumps. It looks like there could be a benefit for folk on fixed tariffs even with the drop in SPF.
In practice however the interaction with time of use tariffs, solar and batteries and operation that aims at minimizing grid load at peak times is likely to favor continuous heating or tariff and solar aligned boosting of indoor temperature even if electricity consumption is slightly higher.
Ooo there seems to be a bug in setting the schedule timing! Hmmm
Fixed, overall results to above is pretty similar.
A few sensitivities on the first example with schedule of 20C from 8am to 10pm:
I’ve now refined the PI auto adapt algorithm and added a couple of metrics to measure how close to set point the algorithm is (degree minutes below set point and above). With this refinement the saving from the above intermittent heating schedule on a fixed tariff drops to 6.3%, indicating that the higher savings above resulted from indoor temperatures not quite reaching the schedule set points.
The savings are now more limited but I can get fractionally higher % savings by increasing the set back temp from 15 to 17/18C, e.g 6.6%
Slight variations maybe 7% ±2% are possible with various different PI algorithm factors.
I can still get savings in the 15% range for 20C 3pm-10pm evening schedules (assuming an hour or so heat up below target temp).
All are more expensive on Agile of course!
I’m sure each scenario can be tweaked and optimised, but I think that is quite telling.
I suppose that another substantial disadvantage of highly intermittent running is that there isn’t really any opportunity for optimisation of flow temperatures via room influence or Havenwise or some other optimiser. The heat pump is has to run flat out every morning to recover the heat to the house, and there is no real way around that.
