Is there really an efficiency loss with an over-sized minisplit?
I don’t think there should be with “proper” software! I certainly can see where “crude” software can cause a loss of efficiency in cyclic operation, here is one example, if every re-start ramps the compressor to full output, thinking every restart is an initial start. If the software simply remembers that the shut down was due to having reached the desired set point and ramps only to the minimum capacity the efficiency will be impacted minimally by extra energy during the start and the possibility of overshooting the upper set-point reduced. The efficiency under steady state minimum compressor speed should rise monotonically with rising outdoor temperature. The need for cyclic operation is due to the absence of need for the heating capacity at minimum compressor speed. “Proper” software should force a minimum run time at minimum capacity and allow set point overshoot then center the temperature variation around the set-point. With what I call “proper” software the “efficiency penalty” of over-sizing will be small to non-existent but over-sizing may result in larger temperature fluctuation.
Do current generation mini splits have proper software? Reports of ” short cycling” suggest otherwise.
GBA Detail Library
A collection of one thousand construction details organized by climate and house part
Replies
Any idling time has standby losses that are never recovered. It doesn't have to ramp to full output for it to be an efficiency problem. The bigger the mini-split in most cases means the minimum modulation output levels are larger, yielding more cycling and standby losses.
The on board firmware in most minisplits does the "right" thing. Actual short cycling problems are almost always attributable to how the room temperature is sensed, when the head is mounted with less than optimal clearances to ceilings & walls. Short cycling can also be caused by lack of air sealing on the refrigerant & control cable path, where wind, exhaust fans etc can draw outdoor air to the inside the mini-split head resulting in very rapid sensed temperature changes.
Dana,
Those "standby" losses are pretty darn small and a good part of them are in the indoor controls. Modulation works at both ends of the envelope ie a larger unit operating at reduced capacity will have a higher COP at lower temperatures than a smaller unit straining to meet the same load..
That theory works great from the armchair, but in-situ efficiency measurements indicated otherwise. Even spinning up a compressor from zero to minimum modulated output requires a significant amount of power, a unless it's running continuously at minimum output the measured COP at minimum speed isn't the most important factor.
Take a look at this:
https://www.greenbuildingadvisor.com/articles/dept/musings/ductless-minisplits-may-not-be-efficient-we-thought
In particular:
--------begin copy & paste-----------
Oversized units suffer from short cycling. The researchers wrote, “At Site 2, where the heat pump was oversized for the very small load, the heat pump cycled often. Daily COP peaked when average outdoor air temperatures were near 40°F. At average daily temperatures higher and lower than 40°F, COPs drop. When the unit was able to modulate without cycling at outdoor air temperatures of 40°F, it was common to see daily COPs near 2.5. When average outdoor air temperature rose above 50°F, frequent cycling resulted in COPs near 1.5.
---------end copy & paste---------
Modest oversizing can result in a slight efficiency improvement as long at the minimum modulation is low enough due to the larger coils and higher COP when running very long cycles. But they all take a serious hit when primarily cycling, rarely modulating. The bigger the minimum modulated output, the shorter those cycles will be, making spin-up power use a larger fraction of the total.
When looking at this 5-6 years ago based on NEEA in-situ data it seemed like the sweet spot was 1.2x-1.5x oversizing of capacity at the 99% outside design temp, and above that most vendors' lines would take a dive. The more recent NEEP studies seem to indicate that 1.5x may be too high.
When you have the time, pore over the 2015 study the GBA article was discussing:
https://www1.eere.energy.gov/buildings/publications/pdfs/building_america/inverter-driven-heat-pumps-cold.pdf
+1 on Dana's responses.
A couple of other things to consider regarding short-cycling:
#1: most in-field testing shows real-world minimum outputs are greater than mfr's specs. For instance, my Fujitsu AOU/ARU9 slim-duct system has a mfr spec'd minimum output of 3,000Btu/hr, but during my measurements at lowest compressor speed, I measured no less than 5,000Btu/hr.
#2 From the units I've worked with, it seems that most mfr's have a minimum compressor speed that varies with outdoor temperature. At colder outdoor temperatures, the minimum compressor speed is higher....which can lead to some short cycling at quite cold temperatures in very efficient houses. See attached example from the AOU/ARU series...minimum compressor speed at <23F speed.
I'm thinking that the biggest loss is the amount of money you paid for the oversized unit in the first place.
Thank you all, I've learned a lot!
John,
We simply must accept the losses of some cycling, even a unit that modulated right down to zero would cycle because the load may drop to zero due to solar heating. Probably the best we can hope for is one cycle per day, and that may well be achieved with current units. What does it take to get to one cycle per day? From my long post above using the building heat loss " load line" and minimum output one can predict the ODT that will eliminate "extra" cycling. Your chart of minimum compressor speeds suggests that in any situation where the "cycling prevention temperature" is above the ODT that forces minimum compressor speed we are as good as can be. I hope the Fujitsu data applies to the Toshia compressors.
Kieth,
The cost of over sizing can be quite small, almost insignificant in today's market. I'm shopping for slim duct units that can heat my house which has a design load of 24,000 BTU at 0f. I'm willing to accept " off brands" and do my own installation, with the help of a licensed tech. I've done it before with an early conventional mini-split and even got repair parts both under warranty and out of warranty parts were much cheaper than what I have to pay for the Carrier stuff in my current house. After 10 years it's still running. So here are three units I'm considering: 90000 BTU $898, 12000BTU $923,180000BTU $1193. Probably more important for efficiency is the minimum output spec. Which more closely tracks the price 90000/4500, 12000/5000, 180000/5500. I've gotten extended performance data from the manufacturer so I know the capacity at -22f,-13f,,5f,+5f,+17f. BTW all three produce rated output at +5f and above. I also have The AHRI data as they are all "energy star". The AHRI data shows similar capacity reduction at 17f (90000/62%,12000/60.1%,180000/66%.) None will meet my heating needs with .a single unit. However, if I chose any pair where one is an 18000 or 2@12k my needs will be met all the time, if they are both working. So the difference between the most efficient (pair, a 90000 and 18000) and the least efficient (two 18000's) is all of $300. However if I choose that most efficient pair, or the lowest cost, a pair of 12k's and the 18k or a 12k craps out I'll get pretty cold if the temperature is 35f or lower compared to the back up offered by the most expensive pair where the hurt starts at 17f. Looking at it another way, with my heat load 9k ,or a 12k will raise the outdoor temperature by 26degf while an 18k will raise it 35 degf. which will keep it above freezing at the design temperature, pretty uncomfortable but I choose to spend the extra $ 300 to be ale to keep the place above freezing on one unit at a little below the design temperature. Choosing the more over sized unit does also affect the temperature below which cycling may occur which is 32f if both 12k units are on or 52f with either one or if I had a 9k unit and it alone were operating cycling would not occur below 54f . To run at max efficiency, I'll need to make sure only one is on whenever the ODT is between 32 and 52. If I stuck with Fujitsu or Mitsubshi I would e forced to use a multi-split giving up redundancy AND PAY MORE! not to mention gain no efficiency either..
Jerry,
Which off brands are you looking at?
I am looking at doing 4 mini duct units, one 7k unit on each floor.
I figured I'd be stuck with one multi outdoor unit to which is a big hit on efficiency. Based on what you wrote, you're saying there's a better option financially going with several outdoor units.
if you don't mind, Please explain for me.
Jamie,
First: beware! Here is the problem with "applying" mini ducts. They don't tolerate very much , "sloppy" ,flex, or long duct or small duct. The rating that says this is static pressure in inches of water column (WC). I'd say below 0.15" is near useless, 0.20-0.4 " is difficult and above 0.5" is user friendly. I'd strongly recommend doing a duct layout and preliminary analysis before serious shopping.. That said I looked only for systems/units that had maximum static pressure ratings above 0.18. My other criterion was that they allow operation below an ODT of -13f and offer significant heat capacity at 0f
I first found Carrier, then Midea, then Pioneer. They all appear to be the same Chinese machines wearing different names. Each has it's own nomenclature, model numbers and brochures but the ratings and specifications are almost identical they each have there own AHRI sheet and those test results have the widest differences, which are nevertheless small. The pricing, I believe suggests that Carrier gets them from Midea who gets them from ? in China In other words Carrier is proudest of their brand, and they won't honor a warranty on an internet purchase, The same thing wearing Midea paint is 15%-20% less and they give the warranty to anyone that had a "licensed" tech sign off on the installation. And for 15% less than Midea you can get the same deal if you'll tolerate the label "Pioneer", to find Pioneer search for Klimare, I believe the Gree brand has the same units but they have NO online outlets and are only sold through "contractors" I also believe Hair brand is available on the same units but I haven't found them in the US. How do I know about the last 2? I went shopping on Alibaba and found New Vision, a Chinese trading company that offers OEM services and the same specs with Gree paint or Hair paint for guess what about 25% less than Pioneer but without a detectable warranty or support and with an MOQ of 100 typically. So if you're building a chain of motels and will need a few thousand systems you can get them with the "TRUMP" brand through New Vision. BTW the 18K system comes two ways one allows 0.2"SP the other 0.4" One other caution there well may be firmware/ operational feature differences that don't show up in data sheets. Also I was shopping exclusively for single zone. However all the mentioned brands offer "flexible" multi zone systems in 18K,24K, 36K that use the same compressor family and have similar cold weather performance. You'l need to look at the multi's but the steps in minimum capacity are about 10 % of the rating step ie a 36k will have minimum twice as high as an 18k (which is 5.5K). Have fun!
EDIT:
After re reading Jamie's post I found he said something I should correct.! I do not want imply that a single multi-split system is more expensive I decided, regardless of cost I wanted, independent systems to achieve a "fail soft" situation then I noticed I could also get an efficiency boost at light load by using miss matched units. My goal was always least catastrophic out come from a single system failure where I'm normally using two and "sharing" the load. Clearly an even division of capacity offers the most "reserve" should either portion fail.So I should have, early on rejected using the uneven split of capacity because it didn't satisfy my goal it turns out the difference in ODT below which cycling WILL NEVER happen is a paltry 2.5f. The next choice was to use 2@ 12k or 2@18K a clear over size choice
It comes down a choice between 2.5f of max NO CYCLE temperature or freeze immunity or not with a cost premium of 28% or $516 for a 50% capacity increase..
While shopping for single indoor head per outdoor unit systems I must conclude quite the reverse. a multi split is always cheaper about 3/4 the cost of an even split. The cost of capacity increments is really small one can get a 100% capacity increment for a 33% cost increment
Danna,
Based on 3 nearly identical systems, one an exact match of design load, one 50% over size, one 75% of need and a linear "load line" the difference in the upper ODT at which cycling stops is +/- 2.5f. These are based on available systems. based a " family" of similar compressor designs from one maker. Within this family doubling the capacity affects the onset of cyclic behavior by only 5f of ODT. I postulate the cause of excessive cycling is inadequate coupling of thermal sensors to building mass! Daily cycles, due to solar effects, should be typical, not exceptional. Perhaps measuring air temperature is not good enough! isn't this measuring the "input" to a VERY low pass, and complex, filter and using it to predict the output? Surface temperature, measured either by conduction or radiation may be much better, sort of measuring the state a little way inside the filter. I still do not see where excessive capacity causes more cyclic operation. All capacity can change is the ODT @ " transition" , from modulation to cyclic What am I missing?
Thanks Jerry,
I find it hard to get pricing on units so it's hard to make informed decisions or value engineer the system. Hey its even hard to get a technical rep (other than a contractor) out to my site. I've been trying to get a Fujitsu rep out for a month now.
I get your push of redundancy ( I'm a big fan of redundant systems) For me, since I want to do 4 units, it becomes a question of doing 4 outdoor units compared to 1 multi outdoor unit. This includes doing a cost/benefit anaylisis of the total unit costs and the difference in operating costs (since the multis are less efficient)
Have you looked into some of the other big brands that aren't so popular with minisplit like Samsung?
One thing that caught My eye about Samsung is that they have a thermostat adapter, to convert the controls into standard hvac wires. This lets you use a better Tstat. I was thinking of playing with the Lennox S30 which is pretty feature rich, including usage graphs and reports.
Jerry: "I still do not see where excessive capacity causes more cyclic operation. All capacity can change is the ODT @ " transition" , from modulation to cyclic What am I missing?"
What you are missing is the average lengths of the cycles and numbers of cycles when cycling, and the fraction of the cycle time at which the system is operating well below it's steady state efficiency at minimum modulation. Most heat pumps won't hit their steady state efficiency until 8-10 minutes of operation. With an oversized heat pump the overall duty cycle and cycle times are lower, and often the numbers of cycles are higher too. All of that adds up to yielding an overall lower average efficiency.
Even a 2F difference in the crossover point from modulating to cycling is actually bigger than it might seem at first glance, when it comes to the seasonal number of spin-up cycles, or the overall duty cycle when it's well into the cycling temperature zone.
Dana,
Thank you! I certainly see that higher " transition temperature" results in fewer days of cyclic behavior so seasonal average performance will improve. Also I know that larger units will always have a shorter temperature rise time resulting in shorter, therefore more cycles. What can be done to reduce the number of cycles? Accepting, or enforcing, a larger temperature variation will reduce,or cause reduction of, the total number of cycles per year. In cyclic operation the larger the temperature swing the higher the efficiency!!!!!!!