We commissioned our colleague Jon Shafer to help us with a simple infrared study of several different designs for heat transfer plates now on the market. Jon, owner of PowerHouse Integrated Conservation Solutions, is an accredited building energy auditor and experienced thermographer. We wanted to demonstrate the comparative heat transfer characteristics of the different designs under identical operating conditions.
Our study consisted of placing five different samples of heat transfer plates on separate copper tubes that all carried the same water temperature (120° F) and same flow in each tube (1.25 gallons per minute). This relatively high flow rate through the relatively short copper tubes insured that each tube supplied each sample with identical temperature along the length of each tube. We let the samples all come to an equilibrium temperature while convecting in free air at 68° F. Hot water was supplied by a 5 gallon electric water heater set at 120° F. When the samples reached equilibrium with the air, the surface of the copper tubes was measured at 118° F.
Then, for comparison, we used an infra-red camera to evaluate the high, low and average equilibrium temperature of each heat transfer plate. Because the low emissivity of (shiny) aluminum causes the infra-red camera to lose accuracy, all of the aluminum samples were painted flat black. The graphite sample was naturally black. The graphite and its polyethylene support returned high emissivity values. (The graphite is light and rather fragile, somewhat like a thin sheet of wet cardboard, it does not hold a shape and requires the polyethylene support to give it a shape.)
Copper also has a low emissivity and the copper tubes were also painted flat black so the IR camera could show that all the tubes reached an equal temperature. Copper tubing was used instead of the PEX tubing normally used in radiant floors in order to minimize the temperature drop across the tube wall and accentuate the difference in heat transfer between the samples.
When evaluating these data, it's important to bear in mind that free convection in 68 °F air represented the entire heating load on these samples. At equilibrium, the temperature of these samples is a balance of the rate at which the sample gains heat and the rate at which it loses heat. The actual heat flux in btu/hr from any of these samples is dependent on the ability of the sample to retain temperature under load. Any increased load as in any actual heating application, especially where the heat transfer plate is in direct contact with a lower temperature solid material would serve to decrease the the temperature values seen in these data. Only the Ultra Fin is intended for use as a freely convecting device. Any freely convecting device is dependent on a strong temperature difference to move the mass of the air between surfaces and will lag any direct contact device. The others are all intended for direct contact.
With the orientation of the samples vertically as shown, heating water was circulated through the tubing at 120 °F. Jon's IR camera recorded a high temperature of 118.1° F at the exposed copper elbows. It also noted a low temperature of 67.7° F, which is reflective of the air temperature in the test room. It should be noted that the plane of the camera lens was most perpendicular to the samples at the far left while the samples at the far right were presented to the camera at a more oblique angle. Theoretically the oblique angle reduces the accuracy of those readings (likely tenths of a °F or less in this case). To a very small extent this skews the results, but in favor of the samples which demonstrated the poorest performance.
To show each plate at its best, Jon evaluated the average temperature of each sample according to a "best average" based on the varying geometries of the different samples.
Discussion of the results
Graphite Sheet - The graphite sheet - #1, measured 6" wide x .020" thick and clearly demonstrated the worst heating performance achieving an average temperature fully 32 °F less than the circulating water temperature. The graphite sheet is significantly oversize relative to the outer dimension of the copper tube, creating an air gap as large as .040". It can be seen in the photo below that it is not really possible to compensate for this oversize fit by altering its attachment to the substrate. If the PE support is compressed, air gaps open at the top of the tube; if it is pulled apart, air gaps open at the sides. Both the IR photograph and the natural photograph were taken with the sample natural to its over-sized form. Finally, it appears that heat received by the plate from the tube in the center is not being conducted to the edges, calling the conductivity of the graphite into question. From this we can conclude that even if the graphite sheet were somehow firmly fixed to the tube, the sheet would heat in the middle but not at the edges.
Ultra Fin - The aluminum alloy "Ultra-Fin", sample #2, measured .015" thick and 5.5" wide. The Ultra-Fin product consists of two, 180° parts that are designed to be clipped or riveted together to surround the tube. In our experiment the two sides were fastened together with a piece of twisted copper wire. The left side of the sample was represented by a shorter piece of an older generation of the product. It should be noted that, compared with the graphite sheet and the sheet metal plate, the Ultra-Fin conformed closely to the copper tubing. Despite this the shortened plate on the back of the tube was observed to be in somewhat poorer contact with the copper tube. Possibly fastening the two halves of the Ultra-Fin together with rivets or the special clips provided by the manufacturer would result in better contact, especially of the back half.
It can be seen in the photos that the front half of the Ultra-Fin, (facing the camera), clearly gains heat from the tube and this results in a strong temperature rise in the center. Despite this, it does not appear much of this heat in the center is conducted to the edges. We can conclude from this that at .015", in free air, the Ultra-Fin is of insufficient thickness to carry as much heat from the center, directly below the tube, as it loses to the edges. The path length at the center directly below the tube is equal to the thickness of the Ultra-Fin, .015", where the path length from the plate at the outer edge of the tube to the edge of the Ultra-Fin is 2.25".
Omega Sheet Metal Plate - Sample #3 is a generic representative of a formed sheet metal plate. It measures 4.75" wide and .O15" thick. The inner diameter at the ends of the plate as measured with digital calipers is .0640", leaving an air gap of only .0075" between the aluminum plate and the .625" od copper tube. The tolerances with which such a thin plate can be formed are not that tight, however. At least one point can be seen in the IR photo where the air gap is considerably less and the tube is likely contacting the plate. The temperature measurements at this presumable point of contact resemble those of the Ultra-Fin, at the edge and the center of the plate. Both the sheet metal plate and the Ultra-Fin are formed of .015" thick aluminum alloy. It appears that even if the sheet metal plate were more firmly fixed to the the tube, it would gain heat in the center but is of insufficient thickness to conduct heat to the edges without a significant temperature drop.
ThermoFin U -Sample #4 is Radiant Engineering's ThermoFin U. This is an extruded heat transfer plate which measures 4" wide and .050" thick. This plate takes a U configuration, preferred for some applications or particular installation techniques. We recommend this product for use in radiant walls, ceilings, and over the sub-floor applications. The infra-red photo speaks for itself. Even this product shows some small variation with which good contact is made with the radiant tubing. Here we think the variation is an experimental artifact actually caused by the painted copper tube where the paint is acting as an insulator.
ThermoFin C - Sample #5 is Radiant Engineering's original ThermoFin C high performance heat transfer extrusion. This shape is 4" wide and .0625" (1/16" thick). The infra-red photograph speaks for itself. The infra-red camera can distinguish between the performance differences between the ThermoFin C and U shapes. This distinction between the two is due to the increased thickness of the extrusion and the rigidity with which the C shape grips the radiant tubing. This illustrates that both contact between the tube and the plate, the area of contact and the pressure with which the contact is made are all important factors in heat transfer.
The relatively thick and tempered aluminum extrusions allow consistently precise manufacturing tolerances. In these photos it can be seen how the normally round copper tube retains it's cylindrical shape even as it is firmly gripped by the tubing. The carefully designed snap channel uniformly grips the heat transfer tubing along the entire length of the finished product. These extrusions are offered in 4' and 8' lengths.
Compared with the other samples, both of the extruded aluminum shapes show differences in heat transfer performance that are clearly related to intentional design and clear specification. These results are easily predicted from elementary heat transfer theory. When interpreting these results, it is important to understand that heat transfer has not "increased" so much as resistance to heat transfer has been reduced.
As resistance is reduced, more heat can be moved at lower water temperatures. In this simple examination, convecting freely in room temperature air, none of the samples were under the heating load they would experience in any panel heating application. The temperature loss across the width of all these plates will increase under load. Heat that is not lost from the water stream in the tubing to the plate, will return to the heat source.
This experiment is simply intended to explore and demonstrate the relative differences in heat transfer between several different heat transfer plate designs. Heat flux is a term used to describe the rate at which heat can be moved through a system. The demonstration does not address heat flux except to note that flux will be higher with the better performing samples. Future experiments will address how the rate of heat transfer through different samples changes with load and with water temperature.
For more information on the reasoning behind the ThermoFin design see, Why ThermoFin?