RE: multi-zone central heating

["Neil Ball" <neilball@xxxxxxxxxxxxxxx>]

I don't quite understand how the process is too long unless you are only
considering proportional control with no integral action to add in a
function that relates to rates of change - the SK approach uses modified
PID
type control loops (it's actually a fuzzy logic, so I'm told) but is based
on quite fundamental thermodynamics, and the type of room control you're
considering allows us to get temp control maintained to within +- 0.5
degree
C in most cases. Where this does not work is if one aspect of the control
is
unstable or outside normal parameters. The very fast heat input to your rad
would need to be dealt with as the degree of control from the TRV would not
have the relevant finesse to cope - most of these valves are designed to be
on/off so are not too linear. Some even give 90% of flow at 10& open
for
example. If you're interested I can look out some stuff on the kinds of
control strategies you need to consider and the formula used. If this is
any
help let me know.

Neil B.

-----Original Message-----
From: Kevin Hawkins [mailto:lists@xxxxxxx]
Sent: Friday, January 21, 2005 14:55
To: ukha_d@xxxxxxx
Subject: Re: [ukha_d] multi-zone central heating


Kevin Hawkins wrote:

I think the best way to
do this is with some form of proportional valve or similar - and playing
with the 'feedback' system will be reasonably fun I think.


Based on a very helpful (and informed) offlist reply I am advised that
the feedback process 'time' on such an implementation would be too long
and that no useful gain would be achieved in such a setup.  I know you
can get thermostats that give a proportional signal out too - ie give a
larger adjustment value out the further you are away from setpoint but I
can see the process time will still be long.  Maybe this feedback of
'room temp' to proportional valve is wrong and it would be better to
have a temp sensor actually on the radiator and modulate my valve based
on that - ie effectively be able to set a radiator temperature. I am
guessing this feedback process could be much tighter (quicker) and more
likely to be stable. Then I could have 'smarts'  that allowed me to
select a radiator temp that maintained the room temp based on my
modelling of the room tehrmodynamics and say the outside temp.   Where
I'm trying to get to is a balanced system where the heat into the
radiator from the boiler exactly balances the radiation and maintains my
set point.

K