Cooling with an economizer

[Tim Peterson]

Hello,

I've recently moved into a house in upstate NY that is heated by natural gas via hydronic baseboard radiators with two zones, and cooled by an air conditioner with no zoning. Some of this equipment is 30+ years old, so I'm researching what the best replacement option will be when the time comes. I would like to go with a heat pump system, either air-source or geothermal, and it seems that currently the greatest efficiencies can only be had by installing a fully proprietary communicating system, of which I am not a big fan, hence my interest in this project.

The area in which I currently experience the greatest discomfort is also the area in which the current residential systems don't seem to have a good out of the box solution available. We prefer to have the master bedroom cold at night, ideally around 64F. During the summer, we could obviously add zoning to the air conditioner and cool that room in the evening, but our cooling season only lasts 2-3 months per year, and for at least half of the year it's too cold to run the air conditioner to bring the bedroom temperature down. Currently I just open the window by whatever amount seems appropriate for the outside temperature, and sometimes I get it wrong. There is also a large whole-house exhaust fan, which I have tried using to rapidly exchange the air in the bedroom with outside air.

It seems like the best solution would be to install an economizer, so that the air handler can switch to pulling in outside air and push that into the bedroom to provide cooling during the colder months. It seems like the best control algorithms for this entail calculating the moist air enthalpy of both the inside and outside air supplies and choosing accordingly, to use energy as efficiently as possible when additionally cooling the air using the A/C. This also helps avoid over-humidifying the house using the outside air. Does your system do any enthalpy calculations at present? Here is the best resource I've found so far on this subject:

https://www.cmfnh.com/documents/ASHRAE_Journal_QA.pdf

My understanding is that this software doesn't yet do automatic heat/cool changeover, which seems like it would be problematic for my scenario, where I'm not sure if there is in fact a distinct changeover time in the spring or fall that would always satisfy the bedroom's needs simultaneously with the remainder of the house. During much of the year the house overall would require heating during the day, including the bedroom on cold days, but then the bedroom would call for cooling in the evening, beyond what would happen through allowing the house to cool naturally. The only alternative I see would be to keep the bedroom permanently cold with the door closed, which would not be ideal and perhaps not even practical given the lack of insulation in the interior walls.

What I wish existed is a system where one could describe the HVAC equipment's physical layout and connections, control parameters, and energy consumption, then define a temperature/humidity/enthalpy target per zone, and have the system estimate based on its own recorded historical data what parameters would yield the most efficient route to the target conditions. For instance, a multi-speed air handler when cooling can run at a lower speed to produce greater dehumidification relative to the room temperature drop, so the outcome could be tracked for each fan speed option at all observed starting conditions to produce a vector field in the temperature-humidity plane. This would make it possible to estimate which fan speed would yield the best approach toward the target conditions. Perhaps there is a better method, but this is as far as I've gotten in pondering the situation.

How would you approach this problem?

Thanks,

--Tim

[Vadim Tkachenko]

Hello Tim,

> It seems like the best solution would be to install an economizer,

You're not going to believe this, during the last month or so people
around me are asking me questions about the things I'm doing in
private just as I'm doing them :)

Here:

https://github.com/home-climate-control/dz/wiki/HVAC-Device:-Economizer
- a pretty old outline of what the device will do
https://github.com/home-climate-control/dz/milestone/13 - a cryptic
message inside that I referred to in my post two days ago (search the
forum for "Early Access: Z-Wave switches to control HVAC hardware")

All right, I'll have to come clean - Feature #2 is the economizer.
Like I said in that message, I don't like patent trolls snatching my
work out of my mouth like it happened in the past, so the work will
stay undefined and no implementation details will be published until
the whole thing is ready and can be referred to as "prior art". At
this point, an initial draft is done, I'm writing test cases and am
about to start dogfooding the system in my workshop. As soon as it is
ready for consumption, I'll push the code and make an announcement.

> so that the air handler can switch to pulling in outside air and push that into the bedroom to provide cooling during the colder months. It seems like the best control algorithms for this entail calculating the moist air enthalpy of both the inside and outside air supplies and choosing accordingly, to use energy as efficiently as possible when additionally cooling the air using the A/C. This also helps avoid over-humidifying the house using the outside air. Does your system do any enthalpy calculations at present?

No, but that's pretty easy to adjust - there's a "strategy pattern"
callback in the code that decides when the economizer is kicking in,
it'll be a simple matter of just taking humidity into account along
with temperature (needless to say, humidity sensors are supported).

> Here is the best resource I've found so far on this subject:
>
> https://www.cmfnh.com/documents/ASHRAE_Journal_QA.pdf

2010 :) Implementing the economizer was something I wanted to do since
day one, but it never went above the pain threshold - until last year.
We've installed the hardware and turned it on and of manually, it
performed exactly as we thought it would.

> My understanding is that this software doesn't yet do automatic heat/cool changeover, which seems like it would be problematic for my scenario,

I wouldn't be so sure about that.

I don't know how much of the forum you've read; I covered this issue
in detail several times. tl;dr: limitations of existing zoning systems
(hardware, usability, UX) dictate using the same sensors in the same
configuration for both heating and cooling. Since sensors are dirt
cheap for HCC, and the complexity of configuration is only determined
by how far you want to go, you will at some point realize that heating
and cooling configurations of the house are drastically different in
all aspects - sensor placement (especially if there's stratification
or multi-story house), PID tuning, schedule, and so on.

Still, auto changeover is on the list, I know how it can be done, but
so far - the demand for this feature hasn't exceeded the pain
threshold. If you install the system and after configuring it and
going through at least one changeover period still think that
automatic changeover is what it needs - this is the place where
desires materialize :)

> where I'm not sure if there is in fact a distinct changeover time in the spring or fall that would always satisfy the bedroom's needs simultaneously with the remainder of the house. During much of the year the house overall would require heating during the day, including the bedroom on cold days, but then the bedroom would call for cooling in the evening, beyond what would happen through allowing the house to cool naturally.

I think this can be configured without the need for the changeover. A
fact about HCC that is not obvious - you can have multiple HVAC units
configured in one HCC instance, and can share control over different
HVACs and different zones, and the economizer can be treated as a HVAC
unit as well.

> What I wish existed is a system where one could describe the HVAC equipment's physical layout and connections, control parameters, and energy consumption, then define a temperature/humidity/enthalpy target per zone, and have the system estimate based on its own recorded historical data what parameters would yield the most efficient route to the target conditions. For instance, a multi-speed air handler when cooling can run at a lower speed to produce greater dehumidification relative to the room temperature drop, so the outcome could be tracked for each fan speed option at all observed starting conditions to produce a vector field in the temperature-humidity plane. This would make it possible to estimate which fan speed would yield the best approach toward the target conditions. Perhaps there is a better method, but this is as far as I've gotten in pondering the situation.
>
> How would you approach this problem?

Here's a bit of a sideways answer - "what you cannot measure, you
cannot improve". At the beginning of the project (a couple of decades
ago) I also went all ballistic and studied the HVAC theory, but then
it turned out that much simpler things matter more than theory does -
say, how someone feels today has more effect on the temperature they
want to set than a 50% change in humidity. So I'd suggest getting the
sensor network out (ESPHome is probably the cheapest and fastest way,
with esphome2influxdb on top of it) to see what the things *really*
are, and then making a decision of where you want to go next.

All the system code is available
(https://github.com/home-climate-control/dz/tree/reactive), if you're
up to it, I can tell you where to go to make the changes you want. If
not, I can try to do it - just like with the economizer, re-tuning the
system from just plain dry temperature to enthalpy will require a
pretty localized change (in the virtual thermostat).

Now, about historical data and learning... That's a bigger piece I've
done the groundwork to take on, but that will take more bandwidth than
I have at the moment. To give you some food for thought - just about
every parameter of the system is fed into InfluxDB, so you can run
machine learning on data collected - and you can easily store data for
many years in it. Either way, if you plan to ever do this, you need to
start data collection as soon as possible.

Hope this helps.

> --Tim

--vt