Why Isn't My Thermoelectric Cooling Module Achieving Expected Performance?

[jwells]

hI'm hoping someone here is well versed at using thermoelectric cooling modules. I'm not able achieve anywhere near the cooling capacity shown in the datasheets and I'm pretty sure I've covered all possible things that could be going wrong. All I'm trying to do is maintain chilled water recirculating into an insulated line bundle at about 38 deg F about 20' long. Ambient air temp is 75 deg F. I'm using off the shelf components that are normally used in PCs which includes a small water pump with reservoir, liquid heat exchanger and multiple TECs secured to the heat exchanger using thermal grease on the cooling side and a 1/8 thick copper heat spreader and heavy aluminum heat sink on the hot side using forced air fan. I've calculated the heat load at around 100 btu or about 30 watts per hour when it's reached it's target temp. The modules I'm using are rated at DT 70C with a Qmax of 60 watts. I've run them at a few current levels that would be considered reasonably efficient 8 watts up to what should be over 30 watts each. There are six of them. For whatever reason I'm not able to drop the temp below 50 deg. Even when I've cut out most of the heat load. I've used online calculators for TECs to determine what it should take and what I have should be way overkill. Even if I remove the heat load I can't seem to get the temps down. I assume it may have something to with the construction of the heat exchanger. Does anyone have experience with this?

 

[Bystander]

You're absolutely certain you've got them all oriented hot to hot and cold to cold?

 

[russ_watters]

Welcome to PF!

How did you calculate the required heat transfer? Do you have any diagrams or photos you could provide? My guess would be there is a heat gain somewhere that you aren't accounting for, but it is tough to know where.

 

[jwells]

I've tried a number of variations to this and very conscious of the direction of the modules as well as the polarity. I did confirm that each module was working by applying power to each and checking to make sure they were cooling.

I agree that I am picking up significant heat somewhere and I'm also not sure that my expectations may be too high. Attached are a couple pics showing my setup. One shows the liquid heat exchanger new as well as one which I had 4 modules attached, two on each side making it double sided. The one I'm currently trying is set up the same way except I'm using 6 of a TEC1-12706 module. In the second picture you see is this assembly sandwiched between two heavy aluminum heat sinks with a 1/8 copper plate that is 3x the area of the cells used as a thermal spreader.

All surfaces use Phobya thermal grease which claims to have a very high thermal rating, 16 W/mK. I have a fan blowing air thru the heat sink with a makeshift cowling around it to direct the air flow across the heat sinks. Using a small pump water is circulated thru the unit and into a 20' trunk that insulated with 1" elastomeric pipe insulation with an outside diameter of 2.75".

As I'm thinking through my calculations I see I made one error but still doesn't account for the total. I calculated the heat load based off the elastomeric insulation being 1" thick that had a .25 btu /hr/ft per degree F. There is roughly 15 sq ft and I was looking for a temp difference between outside ambient and temp within the insulation of 40 which is about 150 btu/hr or about 44 watts, not 30.

I know there is other transient heat coming in but I would think this accounts for almost all of it. For the delta T I've had to make some educated guesses. While running and the water temp around 50 deg I measure the surface temp of the copper plate at the junction of the modules which was approx. 110 degrees. Even though I'm checking the temp right at the junction I know the Thot is something higher so I decided to use 120 deg. My circulating water was 50 but estimated the Tcold to be 10 deg colder thru the thin aluminum heat exchanger. I don't know how accurate these estimates are but that's why I'm asking you ;-).

This puts my dT at about 80 deg or 27 C. There is typically a 10 deg difference between the water temp and the space inside the insulated trunk so my load at this point should actually only be less than 20 watts. At 12v the modules should be drawing 4.6a each and according to the chart this is a Q of almost 30 watts per cell. That's 180 watts total. Even if I was off on my load calculation by many times it should be pulling heat out at a couple times the rate it's coming in. Obviously there is something majorly flawed in my logic and that's why I'm asking for some help on this.

 

[Bystander]

Well, I can always dream about "easy" fixes. No load on the cold loop other than ambient through the insulation? Flow rate through what kind of pump?

 

[Bystander]

Have you tried a slow start? TECs off, circulating fluid, and gradually turn up current through modules from zero?

 

[jwells]

I have no load on the system other than ambient thru whatever surfaces and hardware there is. I intend to have other liquids enter additional lines in the trunk which are already down to temp so I'm just trying to keep them from warming up. I haven't got to that yet so the only liquids in the line right now is the coolant.

This is the pump and although I didn't get an exact flow rate it appears to be in spec. Roughly a gallon a minute.

https://www.amazon.com/dp/B00MB9EP3G/?tag=pfamazon01-20

 

[jwells]

I haven't tried gradually increasing the amperage. I have tried using 6v instead of 12v which effectively halves the amperage and roughly half the cooling but decrease the waste heat generated by 75%. Could only reach 53 deg.

 

[Bystander]

Looking at it the other way around, at 6 V, you pull things down 22 F; kicking it up to 12 V only pulled things down another 3 F.

.25 btu /hr/ft per degree F. There is roughly 15 sq ft and I was looking for a temp difference between outside ambient and temp within the insulation of 40 which is about 150 btu/hr

Double-check this leak calculation --- I just ran it and came up 1800 rather than 150.

 

[jwells]

How did you arrive at 22 F and 3 F? Using TECs is very new to me so I'm not sure what I should expect or watch out for. I appreciate your input.

Am I using the formula incorrectly? I have .25 btu x 15 ft x 1 hr x 40 deg = 150 btu/ hr. 150 btu / 3.41 = 44 watts/hr . A typical refrigerator leak is less than 500 btu/hr so I thought the calculation looked reasonable.

I did some modifications which only used about 3' of thinly insulated tubing to make the connections:

6v = 28 deg
12 = 23 deg

This is beverage industry insulated trunk line and obviously thru the manufacturing process it is loosing the R rating of the elastomeric so there definitely is much more heat load than what should be. You can see the insulation has been compressed.

I'm going to hand make the trunk for this application and retry. I also see a few thing I can do with the unit to make it more efficient. Mill out the liquid heat exchanger and thermal epoxy the TECs so the water is in direct contact with the cold side. This should reduce my delta T. Trim away the insulation on the side of the exchanger that is in contact with the hot copper thermal spreader because it would appear the heat is being recycled back to the exchanger thru the insulation. Is there a better insulation that could be used in an application where it must be very thin?

 

[Bystander]

How did you arrive at 22 F and 3 F?

Just took 75 as the "starting" temperature for the two final temperatures you reported.

1" thick that had a .25 btu /hr/ft per degree F.

This is where I picked up a factor of 12 in the heat leak calculation.