Experimental study on insulation performance of structured-core transparent vacuum insulation panels for different core materials

. The conversion of existing buildings to ZEH and ZEB requires increased thermal insulation, in which the insulation of window surfaces is particularly important. This study focused on measuring and comparing the changes in thermal conductivity of transparent vacuum insulation panels (TVIP), which could be used as simple and easy to install insulation to building windows. These changes were later compared with the results obtained from previous studies conducted by other research group members. The measured thermal conductivity values ranged between 0.017 W/(m*K) of the best performing frame-type with the thickness of 5.1 mm, and 0.011 W/(m*K) of the best performing double-layered peak-type core with the thickness of 7.8 mm. The calculated U-values of the samples ranged between 4.1 W/(m 2 *K) and 1.4 W/(m 2 *K). From the experimental results it could be seen that changing the core composition had a larger effect on the overall thermal conductivity of the sample than the added parameters of coating, getter agent, and heating had. These parameters, however, have had a substantial effect on the longevity of the vacuum inside the sample envelopes, and will thus be experimented further.


Introduction
As time goes on, our knowledge on the human effect on our planet's climates increases. The larger strokes on how the scale of this change could be diminished are done by national governments, multinational corporations and so on, but individual people do have things that can be done to help with the effort.
Reports have shown that in the EU and the US [1] [2] a large portion of a nation's energy consumption is due to private household energy needs, from which majority portion is due to the need for space heating during colder, and cooling during hotter seasons. The concepts of net zero energy houses, also known as ZEH or ZEB, are likely to become the norm in construction in the future, but older buildings will still be present even then.
The authors' laboratory has been conducting research on transparent vacuum insulation panels (TVIP) for window insulation, constructed from different kinds of 3D-printed cores vacuum sealed inside a transparent gas barrier.
The objective is to create an insulation panel that is cheap to manufacture, easy to install, lets light through, and improves the thermal insulation of regular windows to better meet modern standards for insulation.

Outlines of TVIP core materials
The printed cores proposed in previous studies [3] [4] were divided to "frame" and "peak" types. A peak-type sample consists of a core made of a 150 by 150 mm core plate (L1), shown in figure 1. The peaks are 1 cubic mm in size (L4), spaced 10 mm apart from each other L2. The plates have anchor points set to two opposing corners and matching holes set to the others to keep the plates from sliding.  A frame type core's basic structure is pictured in figures 2 and 3. Frame-type experiments in previous study [4] were conducted with multiple values for the lattice dimensions, presented in table 1. Cores A-M are classic frame-types as seen in figures 2 and 3, N is a frame-type made of hexagons instead of squares, and O is a peak-type core. Additionally, core K includes only the bottom lattice and pillars, making it essentially similar to a peak-type core.
The printed cores were modeled using Autodesk Fusion 360 and printed with a Stratasys F170 printer using ABS as printing material, polypropylene (PP) and polyamide (PA) cores were manufactured with injection molding. Figure 4 shows a cutout of a base sample. The basic composition of the samples includes from top to bottom: an acrylic or glass pane, low-emissivity PET film, core layer, and a glass pane. Low emissivity film is added in preparation for further studies on TVIP cores.
The authors noted that the effect the film has to the sample's thermal conductivity is non-negligible, which is why it was kept as a layer to the samples at this stage of research. The samples are heat-sealed inside a gas barrier envelope, from which gases can be vacuumed out during testing. This paper focused on testing a double-layered peak-type core to see how its effectiveness compared to the results showed in previous research, as well as to see what effect the additional conditions outlined in chapter 3.1 would have on the results.

Experiment outlines
The testing of the samples consisted of measuring the samples' thermal conductivity under near-vacuum conditions. The measuring equipment used in the experiments had a lower limit of 0.01 Pa, which was thus set as the pressure to be reached during the preparation of the samples.

Experimental conditions
Testing was conducted based on a combination of three separate conditions. These were namely coating with a polysilazane solution, adding a getter agent tablet consisting of calcium oxide tricobalt tetroxide lithium inside the envelope, and heating the sample to a temperature of 70 °C during the preparatory vacuum period. These conditions will later be referred to as C or coating, G or getter, and H or heating.

Sample preparation
The cores were kept in a hot cabinet for at least 24 hours between separate experiments to reduce the amount of moisture absorbed from the air, and thus keeping different core samples' default condition as even as possible. The sample cores were taken from the holding cabinet, heat sealed inside the envelopes and connected to the vacuum pump hosing with due haste for the same reason. The vacuum pump was then started, with the goal of reaching a pressure of under 2 Pa inside the envelope. The recording application used to pressure monitoring and value recording was created by ULVAC Technologies. Initial vacuum was pulled with a rotary vacuum pump with a displacement rate of 50 l/m, and a turbo molecular pump with N2 of 50 l/s and H 2 of 40 l/s was started in tandem after reaching a pressure below 30 Pa. The preparatory vacuuming period started after reaching 2 Pa. In previous research conducted by Katsura et al. [3] it was noted that a preparatory vacuuming period of eight hours brought the best results, which was thus set to be the initial vacuuming period. The samples mostly reached 0.01 Pa well within this time, except for some heated samples. These differences will be discussed later in the paper.

Thermal conductivity measurement under low pressure
The thermal conductivity Ȝ of the samples was measured while simultaneously vacuuming. The thermal conductivity measurement device used consisted of two heated plates, which applied a temperature difference and heat flux to the sample until a steady state was reached. These plates were set to 10.5 °C and 35.5 °C. After a steady state was reached, the device then calculated the sample's thermal conductivity in watts per meter kelvin [W/(m*K)] using equation 1, where Q u and Q d are the heat flows from above and below in watts per meter squared [W/m 2 ], d is the sample's thickness that the device automatically measures in millimeters at the start of the measurements, converted to meters [m], T u is 35.5 °C, and T d is 10.5 °C, converted from 308,65 K and 288,65 K respectively. The setup for thermal conductivity measurements was done according to figures 5 and 6.

Additional heating
It was noted after initial testing that the increased volume of the double-layered printed core caused an increase in the amount of gases that would need to be pulled from the sample bag, to which the initial 8 hour heating period was not sufficient. The preparatory heating period for every condition permutation was thus increased to 24 hours, which gave more promising results. Another test, named 3+24-hour heating, was then conducted with the CGH-condition, where the sample was first heated for 3 hours without the vacuum pump and then another 24 hours with the pump on. This 3+24-cycle produced results that were more uniform with the expected results as well as the results from previous studies.

Experimental results
The samples' thermal conductivity results from the earlier study conducted by the authors and the doublelayer experiments are collected in figure 7. Frame-type samples are shown as the top section, single-layer peak type sample O as the middle section, and the doublelayered peak-type samples as the bottom section. Figure  8 is formatted to also reflect this order.
A plain double-layered peak-type core had expectedly lower thermal conductivity than plain frame types, owing to the difference in vacuum layer thickness. The core type with the lowest thermal conductivity in previous experiments was seen to be a frame-type core with a thickness of 1.8 mm and a span of 11.5 mm, which achieved an overall thermal conductivity of 0.0171 W/(m*K).
The double layered peak-type core with the lowest thermal conductivity used in this paper's research was found to be the CGH (3+24h) variant with a thickness of 7.8 mm and an overall thermal conductivity of 0.0115 W/(m*K). It is however notable that the frame-type core achieved an effective thermal conductivity of 0.006 W/(m*K) within its vacuum layer, whereas the peak-type CGH (3+24h) variant only reached around 0.007 W/(m*K). This is likely caused by either by the relative vacuum volume within the sample cores or by increased thermal bridging through the "peaks". As seen when comparing the lambda and U-values of samples F and K, both frame-type cores, the measured lambda values of both cores are quite close to each other. Due to the difference in thickness however, 7.9 mm and 5.1 mm respectively, their calculated U-values are vastly different. This pattern follows the conclusion from equation 2, where increased thickness of insulation decreased the U-value of the sample, thus increasing its insulation performance. The calculated U-values of the double-layer peak type cores further enforce this notion, as the added core layer approximately halved the Uvalues in comparison to the single-layer experiments. A notable data point here is that even though the differences between different experimental conditions were not excessively large, the lowest U-values between single layer TVIP cores were reached with samples F and O, showing that similar values could be reached with both frame and peak type cores. The lowest U-value with a double-layer was achieved with the G-sample instead of the later ones with more conditions added. This could likely point towards the added conditions being more important to a TVIP's long-term longevity than to its initial values. Additional experiments will likely need to be conducted to test the long-term pressure rise inside the sample envelope.

Conclusion
Experiments were done to measure the thermal characteristics of a 3D-printed structured-core TVIP. Cores in different sizes and shapes were heat sealed inside vacuum envelopes with differing additional conditions, after which their thermal conductivity values were measured while simultaneously pulling a nearvacuum inside the sample envelope. These values were further converted to U-values, which were then compared with each other.
Peak-type TVIP cores were shown to be a viable alternative to frame-type cores when it comes to insulation performance, if slightly less aesthetically pleasing one. The cores in this study were mostly made of ABS, making them semi-translucent in high lighting conditions. Forming them from transparent materials with similar properties should produce an effect rather similar to a semi-transparent wire mesh glass.