[From Bill Powers (960921.0600 MDT)]
David Wolsk (20 Sep 96) --
Years ago, my uncle Jiggers, invented and started manufacturing (Heat Timer
Corp, in N.J.) a system that we have in our home: Its an electric hot water
furnace (quite small) that connects with baseboard hot water radiators.
There is a sensor mounted on the outside of the house. When the outside
temperature is cold, the circulating water is quite hot; for warmer days,
it's lukewarm. It incorporates the standard inside thermostat that we set
according to our own desires. Jiggers made a lot of money since it saved
much in the heating bills for New York City apartment buildings and we have
been happy here for 22 years of troublefree temperature control of our
house. I have no idea if this is what Hans was describing.
David Wolsk
Thanks for that, David. In that system, as I understand your description,
the primary control of temperature is feedback control of the usual sort.
The temperature of the house is maintained at the reference temperature
(that you set at the thermostat) by turning the hot water circulation on and
off, and this is not affected by the outside temperature sensor. What the
Uncle Jiggers Thingamajig does is to vary the temperature of the water that
the control system circulates to oppose losses of heat. This changes the
loop gain of the system, because the amount of heat transferred when the
water pump is on is greater when the water temperature is higher. But the
system can still control temperature when the loop gain is low, because then
the heat losses to the outside air are also low.
You indicate that the main function of the feedforward loop was to cut
heating bills. This makes me think that the system worked because of
reducing unnecessary heating of the water -- that is, heating the water
beyond the temperature needed to maintain control. When circulation stops,
the heat stored in the water simply drains away through the walls to the
outside air or through the basement walls, not serving to heat the house. In
warmer weather, the water is simply losing heat to the outside more of the
time because it circulates less often, so a greater proportion of the heat
would be wasted if the water were still heated to the maximum temperature
needed in very cold weather.
This is a nice application of feedfoward effects, since precision isn't
required, and any reduction of wasted heat is a benefit even if the effect
varies. In my experience, feedforward is useful mainly in conjunction with a
feedback control system; by itself it can't cope with disturbances that
affect the "controlled" variable without affecting the sensor. A purely
feedforward home temperature regulator would require, in addition to the
outside air temperature sensor, an anemometer and wind vane, because wind
velocity and direction greatly affect heat losses from a house; it would
also need a solar flux detector because the solar input also acts directly
on the house, through roof, walls, and windows. On a cloudy day you need
more heating than on a sunny day, at the same outside air temperature and
wind velocity. The recovery from transient disturbances is also very slow,
for example from a door or window being left open for a while, because the
internal heating does not increase just because the inside temperature
decreases. And if you build a fire in the fireplace or cook something in the
kitchen or invite a lot of people to a party, the feedforward system will
simply overheat the house because it's not sensing the _inside_ temperature.
I should point out that Uncle Jiggers' system would have saved even more
money if the outside air temperature compensation had been replaced by a
feedback control system sensing the heat losses to the outside air and
holding them below some reference level. A more sophisticated contemporary
(I won't say "modern") control system could perceive the ratio of electrical
input energy to energy delivered to the room air, and vary the water
temperature to maintain the efficiency at the highest practical level. Even
on a cold day, such a system could take advantage of solar heating or low
wind velocity, without having to sense them. Such a system might have been
more expensive to develop, but with pocket calculators costing $4 and
thermistors costing 50 cents, the production system would not have cost much.
But perhaps the system Hans is describing has some way of getting around
these problems. I can only wait and see what he can tell us about the design
of his system.
Best,
Bill P.