Thermal Mass in an older property - observations on heating

[KenB]

As some will be aware I am in the process of monitoring my domestic heating system, with the intention of developing a heating controller that offers better performance than the usual timer/thermostat one s normally fitted.

I believe that the thermal demands are unique to each property, which coupled with widely different lifestyles and levels of occupancy mean that there is no "one size fits all" solution to domestic heating control.

The first stage is to monitor the thermal profile of the house, and the boiler behaviour and gas usage, for different external temperatures and generate a simple model of the thermal behaviour of the property.

My house has solid 9" walls and 4" internal brick walls.  The living room has a large chimney breast which also retains a fair amount of heat.

What I hope to establish is a series of warm-up times, based on a given starting room temperature and setpoint versus the outside temperature.

Once the thermal mass of the room has been warmed up, the house will take much less heating power to keep it at the comfortable temperature.

For example, yesterday lunchtime between 12 and 1pm, it took 27kWh of gas to raise the room temperature from 17C to 20C.  Once warmed up, the 20C room temperature could be maintained for the rest of the day and overnight with just 3kW heat (57kWh energy).  During this time, the outside temperature varied between 12C and 9C - so not particularly cold.

The intention is to repeat these measurements for a range of outside temperatures.  What is likely is that the 1 hour/ 27kWh "boost" heat will become longer in duration and use more kWh as the outside temperature falls.

I should mention that the heating system at the moment consists of 3 room radiators - with TRV, a bathroom towel radiator and 25m2 of underfloor heating.  This heats the living room, kitchen, bathrom, my work room and 1 bedroom on TRV set to low. These rooms we consider to be the core of the house,  leaving hall, and unused spare bedroom not directly heated.

More later - when I have some graphs.



Ken

[StBarnabas]

Hi Ken
interesting work. StBC is also thermally massive and I am not convinced algorithms developed for modern housing work well.
Sean

[KenB]

Sean,

Older houses do take a lot more heat to get them comfortable - but once they are up to temperature the heatsource only needs to supply heat at the rate it's being lost through the walls.

A good analogy is a small stream filling a reservoir. It takes a long time to fill from empty, but once full, the stream need only supply water to match the rate it's being drawn off.

Continuing with the analogy,  in cold weather, the "draw off" rate increases, and so the time to fill increases accordingly.  What I want to do is characterise this behaviour, so that the controller has a model to work from - knowing how long and how hard the boiler has to work in order to meet the heating demands.

The 90kWh I used yesterday thoroughly heated the property through - such that the boiler didn't fire again til 4pm this afternoon.

My weekend datalogging has come to a premature end - Mrs. B has just lit the woodstove in the living room ;-)




Ken

[Baz]

Wind speed and direction, and humidity might play a significant role too.
A service that the met office could consider providing (free!) is a form of weather prediction that looks ahead 24hrs to produce an 'energy factor' for each say 10 mile block on a grid taking into account wind, rain, fronts etc. Then a simple web request would get this to control input to storage heaters, underfloor, and similar things that need to know well in advance what the conditions will be.

[StBarnabas]

Guys
I had best declare my hanf her. I became interested in this some years ago through the EU JOULE programme and have been director of studies for 2 PhDs - produced c 12 journal/conference papers and won a national medal for the best technical paper in 2004  "Thermal Comfort Based Fuzzy Logic Controller" from CIBSE. But naturally happy to chat....

[StBarnabas]

Ken
raw data is very useful but doing proper building research ia very difficult. You either have two identical houses in the same location with identical usage patterns and try your different plant on both houses or develop a good enough mathematical model of the building using say TRNSYS or MATLAB/Simulink. For example at StBC I have an excellent solar position model (doing a PhD in Astronomy in an earlier life helps here) a variety of fairly decent atmospheric models,  a decent Virtual Reality model for calculating the interaction of light with my 40 tube 47mm panels and a state based water cylinder model amongst other things.

You also need very good instrumentation - not always easy to get. My own Kipp and Zonen Splite-2 irradiance sensor (c £300) supposed accurate to 2% absolute reads about 20% high despite it's guarantee. I have about 15 temperature sensors; a flow meter and am working on various power meters.

Even then it will be a year or two before I feel confident enough to predict output figures. It might be worthwhile investing in a proper heat meter - like this one from Sontex as suggested by Ken for my GSHP system

http://www.dmsltd.com/SONTEX/Supercal_539_Compact_Heat_Meter?

Sean

[Billy]

What Baz says about wind direction/speed/humidity is a valid point.  Certainly makes a difference to me.   

What does amaze me is the rate the heat is lost  when it is windy.  I thought that I was pretty well insulated and sealed up tight but I think not.  Thermal bridges on a windy damp morning sap heat like it's going out of fashion.

How's that condense patch on yer window Sean?

Billy

 

[StBarnabas]

Billy
Baz is exactly right. On old buildings such as StBC which used to leak like a sieve wind can have an enormous influence. Rain can also  change the U values of walls sigificanty through evaporative cooling, change in emissivity etc.

Thanks the window seems to be OK but will keep an eye on it
Sean

[KenB]

Here's the first of the plots of the "cool-down" of the living room. The plot shows the time period from 00:30 to 12:30 on 21/10/2010.

This was the early morning of 21/10   where the mean outside temperature was -2C.

Maximum temperature drop is about 1 degree per hour between 00:30 and 01:30.  

The two kinks in the curve are at 04:30 when I got up to let the cat out - before it clawed the furniture anymore, and then when I get up on a weekday about 05:45.  Both of these events involved going into the living room, which probably caused a mixing of the stratified air layers and thus a brief rise. Perhaps a temperature probe with more thermal mass, or sheltered from air currents would eliminate these "draught" effects.

Also shown is the outside temperature over the same time period - on the first frosty day of the Autumn this year.   Interesting to see how the outside (air) temperature can rise by 3 degrees per hour - the temperature sensor is on the NE side of the house and does not see any direct sunlight.

I'm now doing a repeat "cool-down" plot based on today's temperatures  between 2.6C and 8.6C.



Ken

[StBarnabas]

Ken
I would suggest trying some system identification techniques to see if you can a decent dynamic model of the room. Probably a state space based approach would be a good place to start. I would be happy to run it through a system such as http://www.mathworks.com/products/sysid/ for you if you have nothing suitable.
Sean

[beelbeebub]

would a standard industrial PID controller with auto tune not work?

[StBarnabas]

I my professional capacity I would used a tuned PID controller as a baseline. If whatever "fancy" control algorihm you are using can't do better than a PID then it is not worth implementing. The one we developed was specifically was for passive  solar buildings - defined as those who got 40% or more of their heat energy from the sun. The big problem with these is that a PID will run them too hot early morning and by afternoon the occupants are opening all the windows to dump surplus heat we were running with a fairly crude neural network predictor to estimate solar input and a fuzzy logic controller. And yes it did significantly better than a PID saving c 20% of energy over a heating season.
Sean
 

[Rupert]

Ken they say that 35% of heat is lost through the walls of a house acording to the stuff you find on the internet.  I should imagine if you insulated your loft and stopped the heat escaping upwards then you would just increase the heat escaping sidewards through your walls, and the only way to stop this is to reduce or stop the heat escaping.

Just a thought Ken as it may be quite shocking to discover while you do these calculations that you could be losing 50%? of your heating money into the sky through your walls, but in this present economic climate would it be possible to build an outside skin/wall to your house so the cavity could be insulated or even maybe replace the walls of your house in a staged way.  I know this may seem extreme but it may pay you to do such a thing the way fuel is rising?...I have no idea of the type or construction of your house

[KenB]

Rupert,

We all lose 100% of our heating money through our roofs and walls - it's just a matter of how fast we lose it.

In 2009 I spent £576 on 17800 KWh of gas

I spent £1500 on council charge,   £3200 on commuting fares, £7050 on Mortgage,  £3300 on pension. £215 on broadband, £720 on phone  - That's £16K before I have even considered food.

Flaggellate yourself as much as you wish about every last kWh - but in the grand scheme of things, energy costs are small change.


Ken

[Rupert]

I understand Ken i didnt mean to be offensive and i appologise and you really only spend a small amount on heating compared to some.