EcoRenovator (
-   Geothermal & Heat Pumps (
-   -   My DIY heatpump attempt (

ThMihov 01-29-18 08:33 AM

My DIY heatpump attempt
3 Attachment(s)
Hi all,congratulations for the great forum first :)

In general I'm experimenting with an old split air conditioner which i would turn in to some kind of homemade heat-pump.The basic idea is to replace the indoor unit with some kind of heat exchanger so the AC heats up water instead of air ,which in turn circulates over radiators in house-some kind of central air conditioning.
Since I'm not a HVAC or AC guy by occupation (I'm an electrician),but it's just my hobby,like electronics,i need some help and advices.

I'll try to explain what i have done so far,and what are my concerns and thoughts about the next steps.I forgot to mention that since that is still just a project with unknown result,i don't want to invest much funds in it,and installation i made it's just for testing,that's why it's not so good-looking:D

So the AC unit is a split type,12000BTU,DC Inverter,running on R410A.It's model is HAIER HSU-12HD03/R2(DB).My first idea was to buy a refrigerant to water heat exchanger,but it turned out,that R401A heat exchangers are way too expensive.For reference-new AC of this make and model costs about 250$,the heat exchanger costs nearly 220$,and the minimal salary in my country is about 230$...So it's not an option.The next idea which popped-up in my head was to use a container full of water in which i will wound a copper tube coil for the refrigerant.A friend of mine gave me an rusty old 60 liter water tank,which is just fine for that purpose.As "side" effect it will work as buffer,for the days when not all energy of the AC is required to heat the house-the AC heats up the tank to the specified temperature,and stops,and when the temperature of the tanks drop to some level,it turns on again,instead of running all the time.

The next question was-how much copper tube is needed for the coil?
Looking at some Danfoss and KAORI heat echangers on the web,for non R410A exchangers,about 12000BTU,the surface is about 0.3m2,and for R410A exchanger,slightly more,but about that value.
The original AC indoor's unit tubing was 7mm diameter,so the next bigger standard copper tube is 8mm diam with 1mm thick wall. I bought 8m of that tube,which surface area is (2*pi*r)*l or S=6.28*0.004*8=0.2m2.
I know it's less than the original ones,but my idea was to get higher discharge temp/water temp respectively.It should also rise the suction temp,which should make the outdoor unit build up frost more slowly or not?.I know it is at the cost of greater compressor working temperature,power consumption and wear,but if everything works fine,the tank will be from stainless steel and i'll put more copper tube inside.
Here are some pictures of the tank with the coil:

I'll continue in the next post,because this will become too long.

ThMihov 01-29-18 09:11 AM

4 Attachment(s)
Since it will be a low-temperature heating system,it will require very large radiators to compensate for the lower temperature of the water.That means a lot of money for radiators and a lot of space for them.So i decided to convert AC indoor units to be used as radiators because of it's fans (forced convection) and because i can get them almost for no money.I got one of these indoor units (luckily its 12000BTU also) in my basement with ripped off electronics,so the fan doesn't have speed control-it's just running at max speed,or not running ,but that is not a problem-I'll make simple VFD later.The only thing that i have made is just soldered 2pcs 1/2 NPT brass bushings on it's pipes to connect it to to the tank and the pump.
As i said pump-i found a cheap GRUNDFOS UP 25-50 130 80W single speed,so i bought it.The connecting pipes are PEX-AL-PEX 20mm diameter (about 3/4") insulated with 9mm rubber piping insulation.The tank is insulated with 5cm of mineral wool.Temporary i've covered the wool with nylon bags,to protect it from soaking with water,but later i'll wound it with aluminum tape.
I was wondering to buy a expansion vessel but the pressure rises only slightly when the water is 40C,so to volume also,and i don't think it's really necessary to have one.
So sipmplified the setup is water tank-pump-indoor unit.
Finally some pictures,and in the next post i'll explain my idea about electronics and about custom control for the AC (which i tested and works) and the results from the first run.

Piwoslaw 01-29-18 02:15 PM

Hi ThMihov, welcome to ER:)

I envy how far you've gotten with this project:cool: I can't wait to see the results!

I see that what you have made is a small heat buffer. I'm not sure of the heating needs that your home will have, but a 60-liter tank may not be much. A long time ago I started a thread on heat accumulators (which by now has lost its photos), but to cut it short:
  • With a slightly larger, and well insulated, tank you will have heat stratification - warmer water at the top and cooler water at the bottom
  • If you keep the copper coils in the lower part of the tank, transferring heat to the cooler water will be more efficient
  • Taking the warmer water from the top of the tank will make heating your space more efficient, possibly allowing you to run the HP less often
  • Instead of 1 coil from 8mm tube, connect 2 or 3 coils from thinner tubes in parallel - lower pumping losses and possibly more surface area.
  • Keep the tank indoors - the higher ambient temperature will reduce heat loss, and whatever heat does escape will warm the house, not the outdoors.

These would be for bumping the efficiency if you ever go to a new version. It might turn out that the tank you have already covers your needs.

ThMihov 01-30-18 05:12 AM

Hi,Piwoslaw :)
This,setup is just for testing with single radiator,that's why is so small :).As i said earlier,if the setup remains the same i'll replace it with bigger one from stainless steel :)
Honestly,I'm very interested on experimenting with homemade shell and tube heat exchanger,but since i don't like to do something without any theoretical background,I'm still reading about them,and trying to make some math,before i grab the tools :).
As for the points that you mentioned-splitting the coil in few smaller diameter is a good idea which didn't come to my mind,although almost every AC heat exchanger is split in two circuits.
I'm getting the warm water from the top already,and i think when the pump runs (meaning constantly) it should circulate the water at the tank ,and the temp should be somewhat even in all the volume or i am wrong?.For example if the water at the top is 35C and the retuning water is 30C ,around the center of the tank (excluding the top most and the bottom parts ) it should be about 32-33C or not?.If there is no pump,to circulate the water it really should be much colder at the bottom:confused:
As for the placement of the tank-unfortunately i don't have a choice,since if everything is ok this HP will go to my parents apartment where there is no appropriate place to install it,except on the balcony,but i'll make a wooden cover with door for easier servicing, insulated on both sides (inside and outside) with XPS plates for example.
As for the heating needs of the apartment,i can't find the file with heat loss calculations i made,but according to my memories with 22C indoor temp and -5C outside,heat losses were about 2500W or something about that number. Basically,it's a warm apartment,because it's in the middle of the building (it's on the 6th floor from 12),and is in the middle between other two apartments,so we only have 1 outside wall,about 12m long,and 2.5m high,which greatly minimize the losses :)
P.S-As I'm writing this reply,it comes to my mind,that on the bottom of my tank,there's a flange with small tube to the middle of the height of the tank (no surprise,it was a boiler in the past ;) ) where the thermostat bulb is normally sitting so i can insert a temp sensor in it,to get an idea about the temp in the middle of the tank :)

ThMihov 01-30-18 01:34 PM

1 Attachment(s)
Now the test results,from the first run.
I forgot to mention,that i have mounted a flow sensor after the indoor unit.It's like this one in the attachments.It's big,but it was about another project,so i have connected it with reducing bushes.I have tried to calibrate it,because it's default constant of 4.5 was not correct for my setup (maybe because of the reducers,maybe because of the working pressure-who knows?...),but i still get an error of about 200g.Way after that i took a look at it's specifications (shame on me:rolleyes:) that says it's minimum flow is 10L/min or 600l/h which i can't achieve with that pump and such a narrow tubes of the indoor unit.After some calculations i have made after the test i roughly figured out the real flow and next time i will reclibrate it to get a real flow rate,and than to be able to make calculations automatically in software.
By the way-for the tests I'm using Arduino (although I'm not a fan of it),because it has million of libraries for it and is ideal for quick test,without writing much code.Usually I'm writing my own code in C and just using the board,because it's ready made and chinese clones on e-bay are at ridiculous
price,and are working fine :)
Back to the tests.At the time of the run,outside temp was between 1-2C at about 85% humidity.Water pressure in the system was 2bar (any recommendations about the working pressure by the way?),water temp 12C.Hooked up my refrigerant gauges and i started the machine.Since initially all the parts of the system needs to be heated (the tank,the plumbings,and etc.),I waited the water to heat up to 32C,for that time they should be heated up already,and started to measure the time needed to heat the water in the tank with 4C to 36C,and logging current consumption at that time.That was done with turned off indoor unit fan.The time was 6minutes and 32 seconds,and the average current consumption was 5.7A.
If we assume (ignoring the temperature difference) that specific heat capacity of water is 4.178kJ/,1 liter ow water is 1kg,and the total mass of heated water is 60kg,so total energy transferred to the water is
Q=cp*m*dT=4.178*60*4=1000kJ or in kW=1000/380sec =2.63kW
Since the average current consumption at that time was 5.7A this gives a power consumption of P=230*5.7=1300W or 1.3kW
So the COP must be 2.36/1.3=2:confused:
As i said earlier I'm not an HVAC guy,and i cant judge if that it normal or not,but according to my intuition it's not good,because according to the AC's manual it's COP at 7C outside temp is 3.6.If we assume that it's a cheap model and the real COP is lower than the stated above,say 3,that is still about 30% reduction for just a 5-6C reduction in the outside temperature,which i think is not normal:confused:.
Few words about the pressures-since there is only one service valve available,only one pressure can be measured at a time-either discharge in heating mode,or suction in cooling mode.
So the highest discharge pressure i got was about 35bar or about 55C temperature of condensation.When i turn the AC into defrost mode,the suction pressure is very low-about 5-6bar or -10C,although the suction coil is now the one that is into the tank,and the surrounding temp is greater than 35C....Since the defrost takes about 2 minutes,if i wait some time after that,the suction pressure will rise,but still both pressures are low,especially the suction:confused:.I think the AC is little undercharged,and after adding few grams of refrigerant,everything will be ok:confused:
As i said defrost-it needs defrost once an hour.For that time all surface of the outdoor unit is covered with frost.The time taken to melt everything was 2 minutes,which i think is OK:confused:
The last thing i'll mention in this post,because it became too long(again) is the heat output of the indoor unit.As i said earlier,since i don't believe on the flowmeter's readings,i decided to do the same test as earlier-as the water became 36C,i stopped the AC and turned on the indoor unit's fan and waited to see how long it takes to cool the water with 3C.The resulting time was 9min and 30sec or 570sec,so the output is Q=4.178*60*3=752kJ or 752/570=1.3kW-honestly,i was expecting a bigger number...anyway.
Initially my calibration procedure showed a flow rate of 140L/h,but that's way the power would be 0.7kW,but the real output is 1.3kW,so the real flowrate is ~260L/h.I'll use it to check if the flow rate has dropped by some reason-clog,or something else.
To double check the ouput i measured the difference beteween the input and return air temp from the indoor unit.The input was 14C and the output was 25C.According to the sticker on the unit,the flowrate of air at maximum fan speed is 420m3/h.Again simplifying that 1m3 of air is 1kg,than Q=1*420*11=4620kJ or divided by 3600 (1 hour) Q=1.28kW-very close to the first value,so it makes sense :)
I'll continue in the next post.

Piwoslaw 01-30-18 03:00 PM

Amazing. I mean - the math:thumbup: I like the depth that you are going into (I can see myself doing it exactly the same way).

The COP=2 looks OK, in the sense that 3.6 from the specs is probably a best case scenario, with everything working ideally. You are using a different setup than was used for determining the COP, and you have introduced a lot of new variables which you have not yet completely identified and quantified. But you are getting close with the methodology you are using.

I think that first you will have to find a way to better confirm your flow rates and pressures. With that you'll see much more accurate value of COP.

You can also take an approach which requires much less detailed calculations, but will show how well this setup will work in real life conditions: Test it in a wide range of conditions (low and high outdoor temp, humidity, run time up to a few hours) and see whether the heat pump is providing too much or not enough heat for the space heating needs. Monitor the temperature in the tank, if it goes too high you need more flow to the indoor fan, if it stays low then reduce flow. In this case you don't care about pressures and actual flows, only about adjusting the water pump's speed to get the end result. Of course, You won't know your actual COP, but you will know it works.

jeff5may 01-31-18 08:06 AM

Nice rig. It's looking to me that you have proof of concept. Hopefully the experiment convinced you to continue refining the beast.

The calculations you have done do far don't really lead to anything useful or accurate yet. Too much uncertainty and too short on run time. To get very accurate measurements and calculations out of these things they have to be run for a lot longer than a few minutes. I'm sure your worm coil heat exchanger is doing its job, and so is the outdoor unit, but that's about it. Run the thing for a week with a kilowatt hour meter attached and let us know how much better it works than an electric space heater.

ThMihov 01-31-18 03:12 PM

Hi again from me :)
Thanks for the kind words :)
Piwoslaw,I know that I can't be 100% accurate,because unfortunately we all can't have a laboratory in our homes,but it would be nice:D.As you said,real life results are more important,because the real world is not an ideal place :).As for the flow rate and pressures-I'm measuring them with a refrigerant gauge,that's the best i can do.Whether or not,they are only useful to judge about the charge,or some kind of abnormal condition.The best scenario,which unfortunately is not the case with these mini-splits is to have two service valves,so one can measure both pressures and superheat/subcool simultaneously,in that way,knowing motor parameters (cylinder displacement and RPM) and using refrigerant properties,can see the real refrigeration effect,but unfortunately it's not possible only with one of the pressures...
The only way I can measure better water flow rate is by using a smaller flow sensor,for which I'll have to wait a month or so,because I can't find some in mo country,so I'll have to buy it from e-bay...I hate to wait :(.But that doesn't mean that until I'm waiting ,I can't try to figure out how to calibrate and use that one:rolleyes:.
As for the heating needs-maybe i said earlied,i dont remember,but this setup is in my garage under my floor.Paradoxically i hated math in school,but now i like to do math,and to see numbers:rolleyes:So i made some calculations about heat losses of the garage,and after the bottom line,the result was 2400W to maintain 22C indoor with 0C outdoor.I doubt that these 1.3kW from the indoor unit will manage to heat it,unless the outdoor temp rise significantly.Even more,it's on the ground floor,and it's full of "junk" (as can be seen from the pictures:o),so initially everything in this room should be heated first,then the concrete floor,walls...To get more clear idea about the room-it's 5 by 6m at 2.2m height.Concrete floor,with nothing on it,45cm outer stone wall with plaster and 30cm brick wall with plaster.Windows are old,not good sealing with area of 2.3m2 and a wooden garage door 2.5x2.2m
with some gap between the floor.So it really isn't an energy efficient garage:D
As for the speed of pump-it's only a one speed one..Excluding VF drive,the only way,for which I have prepared a T-joints while i was connecting the plumbings,is to bypass some of the water with a tap after the pump to the return pipe of the tank,so that way only part of the water goes to the indoor unit,and the other goes directly to the tank.At the moment,they are sealed,but i'll experiment with this too :)
jeff5may,I'll buy this one tomorrow,or in Friday-BRENNENSTUHL PRIMERA-LINE PM231E.Sorry about,not post a direct link,but i still don't have permission to post links.The next thing which i have to do is to write some code to collect data for the indoor unit (either more inaccurate way by measuring deltaT and time,until i got the new flow meter,or if i manage to calibrate the current meter),to be able to estimate the COP in different conditions as you and Piwoslaw mentioned :).The last thing to do is to write some code for automatic defrost,because the defrost's i made was done by me,so that everything can run without my intervention.
The major problem is the lack of free time,because i have a little boy who wants to play with me after i get back from work,and in general spare it's time with me :),and the only time in week in which i can do something is at the weekend if there isn't something urgent to do,and when he is with my parents.So the progress is very slow,but the family is the most important,science and crafting is next ;)
Thanks for the participations,and every correction and good ideas are welcome :)
Tomorrow,because now it's getting late i'll write the simple method which came to my mind to operate the AC ,in particular the defrosts,since these models of Haier are known with their odd defrosts-either it defrosts when there is no need,or they defrost once a year:D

ThMihov 02-04-18 01:41 PM

Hi again :)
Unfortunately,in the past few days i had no free time to update my project,except that today before i went to the other garage to repair my car,i turned the heatpump,to see if it can heat up the garage,since today was warmer then the last time-it was 7C at 77% humidity.The result-2 hours after i turned it up without waiting the water to heat up before i turn on the indoor unit fan,and turn it when the AC was turned on,the room was 25C.The starting temperature was 14C.So it works,it only needs some patience for the initial room heat up :).If there is some free time this week after work,i'll finish the electronics for the control of the AC and for logging and I'll finaly post them here :)

Ron342 02-06-18 07:56 AM

3 Attachment(s)
Great work, guy!!!
First though, spend the time with your son. You and he will be glad you did regardless.
As to the defrost, with the humidity and temps you have, you'll need it! Older heat pumps had a simple spring loaded vane in the airflow over the coil which, when the coil iced up, senses the lesser flow and triggered a microswitch to start defrost but I think the vane itself would ice up. we Newer ones I think use ambient air temp and humidity sensors but our wizard from Spain, Actaurio, did it with micro pressure sensors to detect blocked flow by sensing before and after the coil. I think it was simple and inexpensive - check out his post on that.
On your pump regulation, you can simply throttle the output of that type of pump to reasonable amounts with out much loss to test it but a bypass or vfd would be better long term.
For a shell and tube exchanger, I have good results with HVAC copper coiled into 3/4 or 1 black poly pipe. If you have the copper and poly in the same diameter coils, you can easily back coil the copper into the slick poly pipe and it's easy and cheap. Here's some pics.
When you do the Arduino hardware and software, please use beginner descriptions! Some of us love it but didn't grow up with it!

All times are GMT -5. The time now is 10:18 AM.

Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2021, vBulletin Solutions Inc.
Ad Management by RedTyger