4 ms·
Why no intelligent central HVAC yet?
Why has nobody designed an intelligent consumer central hvac system yet? We have a furnace and an air conditioner, they both pump hot (or cold) air throughout our house via a system of floor vents, each of which can be opened or closed.
What hasn't changed in decades, is that there is a single central thermostat on the main level in a hallway, that controls the whole house. So essentially the entire hvac system is heating or cooling based on measurements from a small section of a hallway.
Isn't this just a big system-identification problem? Why can't we have a thermostat in each room, and motorized controls on each floor vent (to open or close), so that any set of room desired temps can be achieved using a certain selection of room vent configurations?
Seems perfect for a machine learning system... to learn how a particular house's vents are to me opened or closed to achieve a certain set of room desired-temps, in combination with factors such as time of day, season, outside weather conditions, etc.
- angersock 12y agoSo, it's interesting, right? Two parts: control of the rooms, and control of the overall system. One the room side, what's the cost of installing those motorized registers? How do you run power to them? Where else in the room do you place sensors? You could probably fish lines through the ducts, but that won't always work and also you've got fire issues to consider. For new work, this is probably pretty easy, but for old work refits it could be a pain in the ass. Also, what's the cost? You're looking probably at another couple dozen bucks per register, easy. Also, how often do you have return vents per-room? A lot of time you have outlets in each room and then a central intake/filter. On the system side, this isn't that bad--a hardened little box or panel that talks to all these sensors and actuates the registers. ~ In a place like Houston or Texas we appreciate our air conditioning, so this makes sense--I'm not sure it'll be something everybody wants. Even then, the existing systems work well enough, and you can always find people to work on them. It's a pretty good example of a system that is obviously improvable and yet one which is probably not worth the effort to fix.
- gjmulhol 12y agoThere is a large cost to a system like this -- you would need to have sensors in different places around the building and sensors or motors on registers. You would need to give the system a layout of the building or run it through some sort of training regimen (because, I'm not sure about you, but I almost never adjust my registers). I worked on the design of a cleanroom. Even there, in a single room where we just wanted to control temp and humidity and the internal conditions never changed, it took an incredible amount of tuning to get the system to run properly. This is certainly coming, but today, I don't think most people would be willing to pay $000s for a system that still needs training and may or may not provide cost savings or comfort benefit.