#Improving Heat Pump Efficiency

1 messages · Page 1 of 1 (latest)

granite anvil
#

I'm seeking some help from the experts here as I've done some research and become more confused.

We have eight Mitsubisbi mini splits, all controlled by Mitsubisbi2MQTT. Our local power costs are high, so I thought I could save money by setting automations to reduce fan speed and target temperature (or even turning units off) during the night.

Yet I'm looking into this and am seeing a lot of people state that this is dead wrong - that the energy used by a unit to heat a room up from a low setting to a normal setting is greater than setting and leaving the unit on one temperature setting the whole time.

Is anyone familiar with automations to minimize wasted energy on heating/cooling with heat pumps? I'd hate for my work here to prove to have been counterproductive.

#

Bonus: these HVAC units are connected directly to their own circuit breakers, but I'd love to find a way to measure my energy use by compressor or by HVAC unit, even if I'd need to hire an electrician to do it.

#

Bonus 2: Ive also had mixed signals on the extent to which I should heat the (small) basement - unoccupied but shares a thin ceiling/floor with many occupied areas.

queen glen
#

A little of column A, a little of column B
It really depends on how you run your heating - to take it to an extreme - it's far less energy to turn the system off for a month and then heat everything up to 20C than it is to keep it at 20C for a month
Heat pumps also have additional complications - as depending on external temperature it varies not only the amount of heat it can generate, but also how efficiently it can do so - At night is when electricity can be the cheapest, but is also when your air temps are going to be the lowest, so your efficiency will be down, so your cost per unit heat might be lower than if you'd run them during the day
What is best will vary depending on your room temperatures, external temperatures, insulation quality, electricity tariff...

#

For example - in a very well insulated home, one of my friends just turns on his heat pumps at full whack between 2 and 6am when he gets cheap electricity to get 2C over his nominal "comfort" temperature and leaves them off the rest of the day while it drops to about 2C under that comfort temperature - reduces the whole day's heating bill to ~£1 iirc

#

however, if your house drops 10C over a day instead of 4, such an approach would be unpleasant

viral sigil
#

What I would recommend in general is to stick what the manufacturer supplies you, and dont try to do it yourself. They know best

#

For example my heat pump has a 'night mode' which can lower the setpoint at predefined times, until it no longer makes sense to do so. If your heat pump has good controls it can also run on dynamic electricity tariffs

#

(but in the end this is a physics / optimization problem, and so that's how I solved it)

#

and I agree with everything @queen glen says btw

hollow stirrup
#

If you live in an area where you get freezing weather, and your pumps are not RW but something older, you will benefit from a Finnish system called Mitsurunner. It takes care of defrosting with much more optimal control than original Mitsubishi. Highly recommended.

#

Also, if you get freezing weather and you have FD series, you will benefit from controlling the bottom tray heating coil with Mitsurunner as well. The FD series keeps it on constantly when below zero, which is obviously not necessary.

granite anvil
#

@viral sigil Wow, that's really impressive experience!

To the pumps - alas, the Mitsubisbi heat pumps have been satisfactory hardware but their controls are minimal. They just offer a simple "on at HH:MM, off at HH:MM" control on the remote.

The installer ambiguously said to keep them on all the time.

If you mean I can check the manual or ask the manufacturer specifically, I can do that.

viral sigil
#

I think there are some blueprints and integrations for this kinda stuff as well

granite anvil
#

Big drive home! Sorry for the delay.

granite anvil
viral sigil
#

Can you directly control thermal power output or only temperature?

granite anvil
#

But in this context the system is neither optimized by Mitsubisbi nor the local HVAC organization, meaning it's no one's job.

granite anvil
#

I'd come into this thinking the question of whether setting an unoccupied living room to 61 degrees and fan-auto at night versus turning the unit off entirely was more efficient considering you'd turn the temp up in the morning would be relatively straightforward to determine.

#

That's really all I want here, not the type of fine tuning that an integrated thermostat and ducted and zoned furnace system could provide

#

Overall this conversation has planted some substantial seeds of buyer's remorse considering that I thought all the zones would let me better tailor power usage to where people actually are, instead of the ancient system that we replaced that heated the entire house in total.

#

If the difference in power consumption between 7 pumps running and 3 running is essentially unknowable then that's not an advantage at all.

#

I thought it was obvious that automatically turning temperature targets and fan modes down in rooms known to be unused for many consecutive hours would use less electricity than keeping them hot.

viral sigil
#

The most straightforward way is to just try it, preferably with equal conditions, and measure the difference

granite anvil
granite anvil
viral sigil
#

To be honest I did not really consider multi room control. They are not so common here, our houses are small

#

The thermostat is the master

granite anvil
#

8 different heat pumps, 3 different compressors. Each heat pump has an independent control. All but two are in a different room - 2 are in one main room.

#

Temps have to be set high as their sensors report temps way higher than targets but that's a different issue

viral sigil
#

Damn

#

You live ina mansion?

granite anvil
#

Lol no. Small rooms, small units hah. Weird setup in very old house. The opposite of "open plan," hence the several units.

#

Long story there.

#

1450 sq ft or so.

#

Odd tax rebate situation also meant switching to the units at that time worked very well as we replaced an ancient oil furnace. Oil furnace replacement programs ultimately covered about 85% of the base quote for the system.

#

That one did only had one zone and ran through 1k gallons of heating oil or more in a single season. Gross.

#

Now of course, they came with nothing but the stock remotes

#

My project to control them remotely was the entire reason I set up Home Assistant to replace a pretty simple light bulb only Samsung hub

#

So the news that was essentially not that valuable was unwelcome news.

granite anvil
hollow stirrup
# granite anvil I'll check the documentation but the pumps were installed in 2022 and the manual...

20F is -6c, so you'll definitely have some defrostings. Manufacturer statement of no loss of efficiency is irrelevant, they definitely don't work optimally at those temperatures due to the problematic defrosting cycle. You can see it yourself, at freezing temperatures the pump sometimes goes to a cycle where it defrosts every hour, or even more frequently, like every 45 minutes. This is a totally unnecessary waste of house heat energy (used for defrosting) and time. You'll find that the condenser unit has only a little frost on it when the pump already defrosts. Our findings with the mitsurunner show that the delta between outside temperature and condenser unit pipework can be used to manage defrosts much more optimally. This allows to do defrosts only when actually needed, which greatly improves the overall efficiency. I bet the manufacturer lists and tests the pump efficiency without consideration on defrosts. Manufacturer test is probably short enough so that defrosting does not happen. Based on the date of install, your pump may be LN25 or LN35, which have this issue. Only the new RW25 and RW35 units are ok. However, your pump is new enough so that the bottom tray heater is properly managed, and you do not have to touch it.

viral sigil
#

Efficiency numbers consider defrosts. According to EN14825. The number of days below 0 here (NL) can be counted on one hand

queen glen
#

"Manufacturer test is probably short enough so that defrosting does not happen" - Shockingly, the people who write the test standards have given it some thought

hollow stirrup
#

Well, then there must still exist some test consideration they omitted, because for example LN25 effectively sucks in freezing weather, and it's manufacturer spec does not reflect that 🙂

#

It could be that the the manufacturers mostly reside in countries where the issue is not so prominent. Such as NL, as stated. Anyway, experience shows that in a cold country like ours, many Mitsubishi models such as LN25 take a while to get to a bad state regarding defrosts, and they do not recover from it. Maybe this is what fools the standard test. The root cause for bad defrost behavior is not clear, but some have suggested that the temperature sensor area in the pump gets frost, and this causes incorrect data to be fed to the logic. This is unfortunate, since the colder the country, the more heating is required, and the overall pump efficiency has more effect on the total heating cost. If your country is not as cold, then you can disregard the issue.

#

Something could also be said regarding standards and RW35, whose production batches up until ~half a year ago failed to produce significant heating power below -20c, while the manufacturer states that it can heat with maximum stated power at a temperature as low as -25c. Local Mitsubishi dealers have been busy replacing logic boards this year. RW25 was not affected.

tacit void
#

I'm having similar questions but instead for gas burning central heating (opentherm).

#

i guess custom components to the resque: magic_areas for area aggregates & presence, versatile_thermostat for zone control & automation, smart autotune thermostat for global/central control. (Slightly different for heatpumps). Then on the supply side there's load-shedding/scheduling components that can make use of dynamic tarrifs / solar / battery etc. (also interesting if you have heatpump heating a tank of dhw)

#

Now just how to drive those 2 'worlds' (comfort vs cost) ... I was hoping something like "thesillyhome" would be able to solve things like this for us in the future

viral sigil
#

you can solve that using multi-objective optimization (that might not mean anything to you)

tacit void
#

I'd love to do things like that, solve systems of equations or control plant design in HA, any pointers?

viral sigil
#

I can send you my thesis if you're interested. But it assumes a control system that can directly control the thermal output / water temperature. So it's not fit for implementation in HA

tacit void
#

Maybe pyscript/appdaemon like things?

tacit void
viral sigil
#

hmm could work

#

boiler is easier to control because its efficiency is more or less static

tacit void
#

i'm working towards replacing gas boiler for dual heatpumps (hpwh + multisplit air-air, looking also into combining those with air-to-water but not sure yet (vrf/heat-recovery)) this is why I'm diving into gas boiler control (spec/configure it like heatpump capabilities to test that out)

viral sigil
#

yeah SAT is pretty much what most manufacturers do by default nowadays. You can go much further than that by using model predictive control

tacit void
#

I'm def interested in any work you have (done) on this 🙂

#

I was looking into deepc

viral sigil
granite anvil
#

In the meantime, and based on the feedback here - I suppose I can work to make some test automations and report findings here on different days. While that isn't the best experimental environment, I could fire different automations on different days and collect a few samples of each? Unless anyone has any better recommendations (please!)

  • shifting unused rooms to a very low temp during non-use times and increasing heat an hour or so before use (more or less the current scenario)
  • Keeping unused rooms at normal temp and using the "auto" fan function to keep room temperature stable at all times
  • Trying to build in a "bell curve" that gradually reduces temp targets after non-use times and increases them when use times are approaching. In most cases, this corresponds with bedtime and sunrise respectively.

Currently, my electricity plan has constant rates for all hours of the day, unfortunately. The provider, ConEdison*, does provide a time-of-use program, but it's just being phased in as a pilot program and currently seems like an exceptionally bad deal for wintertime.

*insert chorus of jeers and bottle breaking sounds

#

@viral sigil, does a heat pump have significant power use when it's turned on? I've been curious about, and do not have information on, the difference in power use between a heat pump being in an "inactive" mode but on and a heat pump being turned on and off entirely.

Or to simplify the scenario: is there a difference in power consumption between a heat pump that has "hit its target" (sensor temperature > temperature target) for N hours, and a heat pump that is shut down manually when the target is reached and turned back on when the temperature drops?

viral sigil
viral sigil
#

shutting on/off is not something the user should have to do. The heat pump itself can control that process

granite anvil
#

I suppose I don't understand why I'd add a master thermostat if there's not much of a scenario in which I'd want to ever increase temperature across the whole house.

granite anvil
#

Or to put it another way: I did not set this up system to be intentionally complex. I did so as the stock functionality has no system or compressor wide controls whatsoever - it's all pump specific.

#

I would understand if you meant that each of the small compressors' heat pumps had a central control I could set up in Home Assistant.

#

But I ask this in honesty and ignorance: what is the advantage of activating all pumps to target a specific temperature across 3+1 floors and eight "rooms?" Is the point that a house heated to 15 degrees C all over and 20 C in some rooms is more efficiently heated than a house with 20 C in some rooms and 10 C in others?

viral sigil
#

I just meant there is typically one master thermostat, which controls the temperature in the living room / kitchen and the other rooms can open/close their own radiator according to their own wishes (this is a master-slave setup). You can have a master-master setup with multiple independent thermostats across the house, one per room. That seems to be what you have? However you have from what I understand independent heat pumps, wich I have never seen before

#

Not saying its wrong, might be common where you live

#

it's in dutch but should make sense

granite anvil
#

More or less. Based on this I called mitsubishi's US subsidiary's support like (Mitsubishi Electric Trane).

The support rep claims that the BTUs and pressure ramp up in the comprsssor on use, so only using some of the pumps or pumps on a lower setting on the same circuit saves power.

I.e. there are three outdoor compressors with a heating capacity of 36k BTUs, each attached to two or three units. I'm looking for the specific heating capacity of each unit, but I think most have a capacity of 18k.

Service rep stated that lower settings or turning off certain units on the same compressor reduces overall compressor utilization and so reduces energy draw in that scenario.

#

Having all units active will raise compressor use to maximum, increasing BTU use and system pressure, increasing energy use and decreasing compressor lifespan.

#

As you said, it didn't seem like there was much difference between a unit that was "on and inactive" and "off."

#

I was looking over the voluminous documentation online to find a diagram similar to yours. So far this is the closest thing I could find.

Screenshot_20241202_130923_OneDrive.jpg
viral sigil
#

that's an electrical diagram haha, not a thermal one

granite anvil
#

Yeah I automate financial reporting lol

#

Having trouble going through the hundreds of pages on my phone here across different manuals.

#

Does this help?

#

I did also leave a message with the local installation group to ask them but have yet to hear back.

#

Interestingly Mitsu-Trane's on hold messages even include a short commercial blurb similar to this: "why cool your whole house when you only use a portion of it? Mitsubishi room controls..." blah blah efficiency.

granite anvil
#

In doing research for this I'm also looking into a device line the Emporia Vue 2 energy monitor that I could get an electrician to hook up to the breaker box to get more direct energy use info.

granite anvil
#

Postscript: leaving units on auto and temperatures up instead of reducing temperature targets and fan speeds overnight used an enormously larger amount of energy compared to what I was doing before.

viral sigil
#

Were the ambient temperatures comparable?

granite anvil
#

Yes. The issue is the "auto" mode is much more aggressive than you'd think and so burnt up a lot of energy on overy aggressive heating.

viral sigil
#

Funny

#

Needs better controls 🤣

granite anvil
#

Postscript: this interface ultimately gave me everything I wanted: https://github.com/echavet/MitsubishiCN105ESPHome. Mainly as it provides options to have the control of the heat pump use a more reliable external temperature sensor rather than the erratic in-unit sensors.

granite anvil
viral sigil
granite anvil
#

Right now I don't think it'll be that hard to beat the implementation of Mitsubishi Kumo Cloud! That's the commercial implementation all these integrations have been developed to replace.

I didn't even consider it as each wifi controller is around $200 per unit and it's tough for me to even buy them directly rather than work through the local HVAC techs.

My whole smart implementation across eight units, hardware side, probably cost about $80 tops.