[OpenTRV-interest] Adaptive comfort

[Tim Small]

On 12/07/16 16:51, Simon Hobson wrote:
> Tim Small <tim at buttersideup.com> wrote:
> 
>> > Difficult to get right
> I suspect impossible !

Depends on your definition of "right".

If your definition is to make 100% of people feel comfortable 100% of
the time, then yes, I agree that's impossible.  The models I've seen
explicitly acknowledge that by giving a "90% of occupants happy" type
outcome in their analysis.

On the other hand, if you're "just" happy with doing better than every
other domestic system in existence (that I know of), then I suspect it's
actually reasonably easy, as they all completely ignore such issues
entirely, and are thus almost certain to be getting it more wrong.

The bar is currently set very low.

There is quite a large body of academic work studying perceived comfort,
so you can get quite a few of the items pretty-much right reasonably
easily (using indoor and outdoor temperature recent history for
example).  Then given enough data and the correct centralised analysis,
you can start to respond to the others pretty well too I suspect.

The job of a heating system is to keep the occupants comfortable using
the minimum amount of heat.

In the past this has (justifiably) been solved using a simple
temperature set-point and time program (perhaps with a few refinements
like adapting ensuring you attain the set point by firing up the heating
system a bit early).

I suspect it's now possible to do a lot better than that.  I'm sure it's
possible to do it better than the usual suspects do.

Get it right, and you will start to make the existing systems look a lot
like circa 2005 Nokia and Ericsson etc. mobiles would in today's mobile
phone market.

Tim.