Hello there

Let me introduce myself for a bit: my name is Bart Hommels, and I live in Coton, a village just west of Cambridge. I am married to Rose, and we have a daughter named Ronja.

I work as an applied physicist at the Cavendish Laboratory, working on instrumentation for particle physics experiments.

I guess the Mr. Maker part of me is never switched off, as I like to make, mend and create things, both at work and in my free time.

Other things I like are outdoors activities: running, cycling and (almost) anything mountains - unfortunately not much of those around in Cambridgeshire.

When we bought our house, we knew it would require a lot of work to bring it up-to-date. After getting an electricity consumption meter, I got very interested in ways to reduce the energy demand of the house.

Soon after realising the potential of renewable energy, we decided to do what many others did at the time and we got solar panels installed: optimising the supply instead of reducing demand.

The thermal performance of house was addressed later: a year or two ago we decided to extend the house, as it was (is) generally quite small. I wanted this to be a starting point of a house-wide thermal upgrade instead of an add-on that would increase energy usage.

As I easily take many things too seriously, we embarked on a long eco-upgrade project. This is by no means completed, and hopefully I get to writing a few blog posts on what it all entails, and my random related ramblings on the subject.

Improving weather compensation for heating controls?

With the winter now only a distant memory it is time to make up the balance about the new boiler.

The system has worked very well. Boiler operation has been very silent, and due to the large thermal mass of the 80mm thick screed surrounding the underfloor heating (UFH) pipes there has been no short-cycling or intermittent firing whatsoever, and temperature in the house was very constant. According to the specifications, the boiler efficiency should be around 95% in this configuration!

The separate zone for the upstairs worked fine as well. I have to sort out the best location for the Nest thermostat so it does always register people upstairs and asks for heat if needed.

The thing that needed twiddling was the heating curve, and I realised that this is not easily done in a highly insulated, high thermal mass building. According to an online simulator, the temperature profile, and thus the rate of heat loss, for a wall build-up with U=0.15 W/m2K or better, takes in the order of 5-6 hours to settle. In addition to that, from experience it takes 2-3 hours to heat up the screed floors by a couple of degrees, so the time delay is about 8 hours in total! Even in the midst of winter, outside temperatures are hardly ever constant on that timescale, so firstly adjustments to the heating curve have to be small and on a best guess basis, and secondly the weather compensation is always out of step with reality.

The other side effect, and this is most prominent in the shoulder months when there is already considerable solar gain heating the inside of the home, is that when the boiler gets switched on in the morning, it senses a low outside temperature, fires up and starts running the UFH needlessly as the internal temperature is already at the desired level. The boiler then usually switches off an hour later when outside temperatures have risen sharply.

The above could be mitigated by a weather compensation system that does not read the actual temperature, but reads the forecasted temperature (of an on-line weather forecasting service), by a (adjustable) number of hours ahead, reflecting the heating & settling time of the building. The look-ahead time is governed mainly by the degree of insulation and amount of thermal mass present, and could in principle be estimated.

Of course it would be better to have a more advanced solution where a full thermal model of the building is simulated to estimate the current heat demand based on the thermal masses present, insulation values, current internal and forecasted external temperatures, however this might prove too complex and thus unaffordable as a solution for domestic heating. An internet search revealed two student projects of this kind, trying to arrive at better heating controls for two buildings housing faculties of computer science using machine learning and artificial intelligence.

The proposed solution using a fixed "look-ahead" time would benefit all types of heating systems that currently can be controlled using weather compensation controls. It would be particularly beneficial for heat pumps as these are most efficient when run continuously, with a slowly varying load.