Towards the Hackable Boiler?

This post is the result of some discussions I had last Thursday at the Homecamp Christmas Party, with Paul Tanner and Mark regarding the feasibility of hacking the home heating system (in a safe, benign kind of way) to see if it is possible to make it better featured and return improved efficiency over the existing time of day and thermostat type controls that almost every central heating boiler employs.



Whilst a lot of people are now monitoring home electricity consumption - because it's cheap and easy to do, I thought I'd take on the challenge of monitoring my gas consumption and to try and link it in with a thermal model of this 1905 brick-built house.

Home heating is becoming increasingly expensive, and will continue to become more-so as we get even shorter of natural gas reserves. What I want to do here is analyse my domestic heating system to see if there are scopes for big savings. My electricity consumption is already fairly low - so there may well be an better opportunity in focussing on gas.

I have suspected for some time that it is more efficient to leave the heating on 24/7 at a low heat in an old house like this, and get the thermal mass of the brickwork thoroughly warmed through. It was now a good opportunity to try and prove that theory. I now have some temperature sensors and know enough Arduino code to be dangerous, and with the snow lying around outside, now was a good weekend to get some data to try to prove the point.

The first thing to assess was the availabilty of suitable inputs and monitoring points on the boiler to allow it to be monitored and controlled. Most boilers have a simple on/off control that is wired out to the timeswitch and thermostat. The timeswitch brings it on at a certain time of day, it keeps heating the rooms until the set point on the thermostat is reached, and the thermostat turns it off. Simple on/off control.

About 10 years ago, boilers started to get a little more sophisticated, - they had to, in order to cope with the control demands of the condensing mode heat exchanger. Boilers now have thermistors to monitor the flow and return temperatures and a variable speed fan and modulating gas valve, to ensure that the boiler can react to sudden demands and always combust the gas at the best air/fuel mixture. When your burner, rated at 24kW is only the size of a 5 litre paint can - you have to be fairly much in control of the air/fuel mix. Additionally you need to turn the power down such that the water returned to the boiler after heating the radiators is sufficiently cool enough to condense the water vapour in flue gases, and thus recuperate the heat of vapourisation/condensation. This is how condensing boilers manage to achieve efficiencies of about 90% when running correctly in their condensing mode.

So firstly a quick monitor of the boiler flow and return temperatures. I had some pipe clip thermistors from Rapid Electronics, which I attached to the flow and return pipes. As it's close to freezing this weekend, and the boiler is on almost 24/7 right now - it's a good time to observe the heating cycle. The plot below shows flow and return temperatures over a 3 hour period. The boiler had been struggling to get the living room much above 17C, so the first thing to do was see what the respective temperatures were before and after I turned up the water temperature control knob on the front of the boiler.

The traces are the raw count output from the thermistor ADC channels - but for the purpose of understanding, just divide by 1000 to get approximate water temperatures in degress C.

First I let the temperatures stabilise for about 50 minutes before turning up the heat control on the boiler. The red trace (the flow), shows a near immediate rise, as the boiler adjusts its combustion power to increase the flow temperature. The blue trace is the return water temperature that has circulated around the radiators and underfloor heating and its response to more heat, was a more sluggish response. This is because it takes time to heat up all those rads and pipework to their new elevated temperatures and this is characterised by the rounded, almost exponential rise in the blue trace in response to a near instantaneous step change in the red.

After the traces had stabilised at their new levels, I then decided to turn on a radiator in the front room that had previously been shut off. This corresponds to an increase in heat demand or thermal load, and quite expectedly, the return water temperature fell by a few degrees as the cold water in the dead rad, was now circulated back to the boiler. What I hadn't fully appreciated was the accompanying drop in boiler flow temperature - as I thought that the boiler would quickly modulate it's power output up to match the increased thermal load. Instead it was the flow temperature that showed a slow rise back up to the 51000 asymptote. Finally I decided to go for another boost in the heat control and this is shown by the final step up in the red trace, with the accompanying slower exponential rise in the blue trace.

So what can we learn from this chart.

1. The flow temperature is controlled so that it runs at a few degrees hotter than the return temperature.
2. The boiler responds rapidly to demand changes made on the control knob on the front panel of the boiler.
3. There is an appreciable thermal lag associated with having to raise the temperature of the pipework and rads up to the new temperature.
4. A sudden change in thermal load such as a cold rad being turned on (or a zone valve or DHW valve) will cause the boiler flow temperature to drop and slowly climb back to the original level.
5. For a given position of the front panel water temperature control, and a given circulation pump speed, the boiler can only produce a certain heat output. If the house is losing heat to the outside (on a cold day), the boiler might not be capable of reaching the room temperature as demanded by the thermostat - unless you turn the wick up a bit, by manual intervention on the heat control. This was proven over the last couple of days when the boiler struggled to get the living room above 17C.

Other Observations - made over the last few years whilst messing with the heating controls.

1. Older houses leak heat very easily. The amount of heat needed to keep them at a constant comfortable temperature, is directly proportional to the temperature difference between the outside temperature and the temperature you set the thermometer to.
2. it has taken the boiler about 2 hours to get the living room up to temperature (19C) and this was only as a result of turning up the heat control. The boiler has now entered a kind of cycling mode, where it ticks on and off about every 10 minutes. This is bad, because it gives the water in the pipework a chance to cool down, and the boiler has a 4 minute start up sequence. This is the boiler equivalent of stop-start driving in traffic - wasteful in fuel. So it must be possible to hack into the boiler heat control potentiometer so that it can be set exactly at the right setting to balance the thermal load and heat loss from the house - so the boiler doesn't keep cycling.

In the case of my house, where I have nearly 10 years of gas readings, I have found the constant of proportionality to be about 6kWh of gas consumption per day, for every degree difference. So if I set the thermostat in the living room to 19C, and its averaging 9C outside, the boiler will consume approximately 60kWh of gas in a 24 hour period.

If like over the last few days it's close to freezing outside, the temperature difference is closer to 20C, and so you will consume more like 120kWh per day in order to keep the house at the same comfortable temperature.
2. Somewhat contrary to expectations, I believe that in an older property that it is better to leave the heat on low 24/7 as this uses less gas than trying to heat the house for a few hours in the morning and at night.

So the next thing to do is to capture some data of room temperature and outside temperature and tie this in with the boiler behaviour. This will need another couple of thermistors wiring into my sensor board - and probably a move to using the Arduino as the sensing and control platform - mainly because its easier to try things out quickly than having to edit PIC assembly code.

Future Requirements.

I'm currently using a little PIC16F88 to read the two thermistors and send the readings plus a time stamp out to the PC using an FTDI serial lead. I built the PIC board onto some stripboard a few years ago as it was intended to be a wireless temperature sensor, so it has a Telecontrolli RT4 433MHz transmitter module on it and a couple of connectors to add thermistors and radio control servos - as I thought I would be using this board to open and close radiator valves. It's a handy little board, which predates any of my Arduino stuff by 5 years.

So in order to progress further with this investigation there are some extra things needed. I already have the flow and return sensors logging on a second by second basis. I now need to extend this to include a room temperature sensor in the living room and an outside air temperature sensor. This could easily be done with the Arduino, which could probably be embedded inside the control box area of the boiler, in such a way to share the boiler's low voltage DC supply. A protoshield would include the interfaces for the 4 thermocouples, an on/off relay for firing the boiler on and off and an analogue PWM voltage output to over-ride the heat control potentiometer. If I include 433MHz wireless Tx and Rx modules on the protoshield, I can get wireless temperature data out of the boiler and use the receiver to intercept the boiler on/off signals from the Drayton Digistat wireless thermostat - but I suspect that will shortly will be redundant. A USB socket on the bottom of the boiler will allow me acess to data output and to re-hack the Arduino code as new ideas arise.

According to the digital thermometer outside, it's currently -2.7C out there - so no wonder the heating has been struggling these last couple of days.

I can probably extend my little PIC board to read another couple of thermistor channels, so that the serial output of this board can be read by hyperterminal and graphed in XL - as how I have been doing it so far. However it would be good to get more practice in on C programming the Arduino, and this sort of project is ideally suited to it.

There is also the question of hacking into the water temperature control pot on the boiler. Recent boilers use either an analogue voltage control (0 - 10V) or an open protocol called Open Therm (OT) to allow them to be controlled remotely from other systems or devices. I suspect that I will first use an Arduino to generate an optically isolated PWM signal, which drives the potentiometer input.