by Mark Bomberg
Heat transfer though wet thermal insulations is complexed and difficult to understand. So I will show you through results. Figure 1a shows tests, where the relation between thermal conductivity coefficient and moisture content of the material depends also on the duration of the measurement process. Undoubtedly, the presence of thermal gradient changes water distribution in the material and some engineers (Figure 1b) decided that doing a rapid test will be a good solution. They assumed that as the rapid test does not include movement of moisture, the result would give us a real effect of water on thermal conductivity.
Figure 1a (left) Thermal conductivity of fiberboard measured in 1932 as a function of moisture content and the duration of the test. Figure 1b (right): Comparison of rapid test methods with equilibrium testing at the same moisture content in different laboratories
Figure 1b shows that using the rapid measurements does not reduce the effect of moisture. On the contrary (Figure 1b), an increased thermal conductivity has been obtained with a probe method and while somewhat similar is effect was obtained the constant thermal gradient method (Guarded Hot Plate), the results are different from the expected. Thus, the issue warrants much more careful examination.
Figures 2a, 2b, and 2c show the preparation of mineral fiber insulation (MFI) specimens and Figure 3a and 3b show the test results.
Figures 2a (left), 2b (middle), 2c (right): Preparation of MFI specimens for testing.
Figure 3a (left): Schematics of moisture redistribution in a specimen tested for the thermal conductivity coefficient. Figure 3b (right): thermal conductivity coefficient measured at CRIR on two wet MFI specimens.
Figure 3a highlights the moisture movement to the cold plate and Figure 3b shows the continually measured thermal conductivity coefficient. These tests were performed by CRIR (Center of Industrial Research in Rantigny) of the Saint Gobain company. To explain what is happening, we are reproducing Figurers 4a and 4b showing similar tests performed in Canada.
Figure 4a and 4b: A comparison of MFI (left) and CFI specimens (right). Figure 4a shows heat flux versus the test duration and Figure 4b moisture distribution profiles as they change until reaching the dynamic equilibrium. Test performed at NRC Canada laboratory in Ottawa.
Figures 4a an 4b show a similar pattern of phenomena, yet the effect of water on heat transfer through the material are different. In mineral fiber insulations, water becomes quickly collected in a narrow layer at the cold surface and the overall effect on the heat flow is small. The water transfer in the cellulose fiber insulation is slower and water is spread over a wider range of the material making the effect on thermal conductivity somewhat larger. The difference is caused by the cellulose fibers being more hygroscopic than mineral fiber.
Cause and Effect
The effect of water redistribution on the heat flow depends on the rate of water vapor diffusion through the material and the value of thermal gradient because the phenomenon includes two independent components. When water vapor (WV), driven by the thermal gradient, reaches a condensation point, the liquid water starts a movement towards the dryer material, whichever side it may be. In this case, it is on the warmer side and the liquid flows in opposed direction to the vapor flow. These two processes last several hours, until reaching the final dynamic equilibrium. We are talking about dynamic equilibrium (or a quasi-steady water profile) because two different transports occur at each place at each moment of time. The thermally driven water vapor transport goes towards the cold side and liquid water equalization goes in the opposite direction. Therefore, one of the names of this phenomenon is a thermal syphon.
Intensity of the condensation depends on the temperature in the point of condensation so the S-shaped curve will be different if the mean material temperature is different. With other words, the same test performed on specimen with different thickness may give differently shaped curves. Furthermore, the rate of water vapor transports (WVT) are measured differently in the Europe and the North America. In the EU one uses so called miu-value as a function of moisture content (i.e., ratio of the actual WVT to a diffusion through air layer), while in the North America one uses actual values of the transport coefficient.
Effectively, one may consider water as the phase changing material (PCM) and the two independent effects: (a) presence of immovable water as a disperse phase in the matrix of tested material, and (b) movement of water at the rate approximated by the WVT coefficient.
Figure 5a and 5b: Comparison of a Spray Polyurethane Foam (SPF, left) and Extruded Polystyrene (XPS, right). Figure 5a shows heat flux versus test duration and Figure 5b moisture distribution at the dynamic equilibrium. Test performed at NRC Canada laboratory.
As typically foam products are not uniform, the profiles of the dynamic moisture equilibrium within the specimen (Figure 5b) are further affected by individual factors. The extruded polystyrene has a uniform material core but different density on the surface and as the rate of water vapor flow is much slower at the surface layer one can see water accumulation there. We do not see water in the material core because the polystyrene foam has a negative wetting angle and repels water accumulation.
On the other hand, the moisture profile in the SPF shows a beginning of the expected S-shape curve but somewhere in the middle changes its character. One must remember that the SPF specimen has mostly (above 90%) closed cells and that the specimen is sealed in a polyethylene bag on both sides. Water vapor proceeding to the cold side encounter a growing a growing resistance to the flow and the gradient of water content gradient is getting steeper while it moves further towards the low temperature.
With the thermal diffusion bringing more water a peak is reached somewhere in the middle and at a certain moment, both vapor and liquid phase flows unite to carry more water vapor towards the cold side. As SPF contains mostly closed cells with a small volume of open cells, the small water content gradient is sufficient to distribute the free capillary water to the cold side.
Figure 6a (left): Test set-up where specimen is set in the air between hot water and water absorber on the cold side. Figure 6b (middle): Moisture profile in the Spray Polyurethane Foam (SPF) with both surfaces opened and thermally driven water vapor entry. Figure 6c (right): Mass of water entering form the hot water side versus time. Test performed at NRC Canada laboratory.
To demonstrate this phenomenon, we have performed measurements of moisture accumulation in foams with two open surfaces (Figure 6a) while the water content distribution during these tests is shown in Figure 6b. Indeed, we have obtained an asymmetric bell curve, because the water vapor entering the specimen (on the left side) has to overcome increasing resistance while that on the right side has the resistance becoming smaller as the layer becomes thinner.
There are a few, important conclusions that can be drawn from the above reported experimental information. First of all, in the moment when one changes the boundary conditions for the test specimen, the energy status of this specimen chnges instantaneously. The rapid test method measures the same physical phenomena as the so-called steady state method when applied to transient conditions. Any good test method has some but small and accountable effect on the conditions being measured. Thus, the conclusion of the first issue being examined is actually opposite to that what the engineers hoped for.
Figure 1b shows the highest possible thermal conductivity coefficient was measured with a rapid test method while the dynamic equilibrium obtained for these boundary conditions is at last a magnitude lower. As these results are confirmed with different test methods and different laboratories, it is now well established that the measurements of thermal conductivity are related to the actual energy status of the material.
Figure 3b shows two continuous curves – showing significant change in thermal transmission vs time; observe that in the most of currently used energy models, we use a limiting case of dry material with a constant thermal resistance. Figures 3 and 4 also explain performance of phase change materials, except there the range of changes in PCM is not relates to the space in which water moves but to the efficiency of melting and solidifying of the PCM as a function of cycling temperature.
Figures 5 and 6 provide an understanding of moisture transport in foams and explains the resistance to freeze-thaw damage in winter conditions. Figure 6c shows the inflow of water to SPF/ polyisocyanurate foams. The rate of water entry to the SPF and polyisocyanurate are slowing with time and may even stop within a several-week period, indicating that an increase of moisture during the winter exposure will not result in a material water saturation.
Now, we need to broaden the scope of our examination (Figure 7), and show that changing the boundary conditions can also modify the response of different foam types to water entry. Measured results are shown in the actual scale, yet without the numerical values ascribed to water content. It is done because the engineering tests used in Europe or USA do not show the real moisture content inside the material. These tests show an apparent water absorption (i.e., water inside the specimen and water attached to the surfaces of the specimen).
As the results depend on the test method used, we cannot use it in discussion on material performance. Yet, these charts represent the apparent material characteristics. For the SPF, the worst-case was the constant thermal gradient and thanks to the slow process, the moisture pick during the daily oscillation is much slower and the closed cell nature makes it even smaller for isothermal conditions.
Figure 7: Effect of changing boundary conditions: (a) constant thermal gradient, (as Figures 4- 6, (b) freeze-thaw, oscillating gradient, (c) isothermal conditions (constant temperature). Materials: EPS low density (PS 1), medium density (PS2) and SPF (PU).
For an expanded polystyrene, these results confirm what we have already discussed in the first column, namely that in any exposure to wetness, the low-density material performs best. The denser material picks more water under thermal gradient than under isothermal water intake. There is no need to test water absorption of open cell polyurethane foams. With a fine, open pore structure water response of these foams is identical to the denser polystyrene. If water is not a concern, these foams can function as air flow retarder. If you are not sure about possibility of becoming wet, you must require results of dimensional stability test when open cell foam is exposed to water condensation under thermal gradient (do not mix it with a dimensional stability after isothermal water intake).
Unfortunately, understanding the field performance of materials also requires an understanding of their interactions with other materials in the assembly and with prevailing climatic factors. As most of our standards were first developed for purchase speciation for the sake of fair business practices, they may use any apparent test method. To characterize material for a use in buildings, we need to know its compatibility with other materials, its thermal, moisture, and air barrier performance, its dimensional stability, and its durability.
Mark Bomberg is a research professor at Clarkson University in New York. He’s also an honorary member of the Building Enclosure Technology and Environment Council (BETEC)/National Institute of Building Sciences (NIBS).