#### Installation — business terrible - 1 part

September 8th, 2015

Figure 8(a) displays the block plot targeting temperature (i. e., 30°C, 40°C, and 50°C) over 22 distinct combinations of RH, cyclic movement, and UV radiation. In this plot, temperature is the target factor (as denoted by the plot character), and UV radiation, cyclic movement, and RH are all robustness factors. The vertical axis shows the magnitude of modulus ratio, and the horizontal axis comprises various combinations of the three robustness factors. In order to facilitate comparison among different exposure conditions, the mean modulus ratio is determined by averaging the modulus ratio at the same temperature within each bar and is displayed in Fig. 8(b). For example, the first bar shows the effect of temperature on the modulus for the 0 % RH/no cyclic movement/ no UV condition. Note that a temperature of 30°C yields a modulus ratio of «1.25 and thus an increase of 25 %, whereas 50°C yields a 13 % decrease. For the second bar (i. e., the combination of 0 % RH and UV radiation without cyclic movement), the sealants exhibit modulus increases, but the magnitude of increase is comparatively less than those in the first bar. However, a lower decrease in the modulus of 25 % is observed for exposures at 50°C.

Scanning across the various robustness factor settings (horizontally), it can be seen that 40°C or 50°C is almost always located at the bottom of each bar, and that the corresponding modulus ratios are almost always below unity. Indeed, 14 of the 16 robustness factor settings show that elevated temperature exposures of either 40°C or 50°C result in a greater modulus reduction than those at 30°C. In order to quantify whether temperature is statistically significant over all robustness factor settings, the chance for 14 of the 16 robustness factor settings showing the importance of the temperature effect involving randomness is calculated using a binomial model.

P(x, n, p)=( n V(1 – p)n~x (4)

where:

x = number of successes in n trials, and p = probability of success in a single trial.

The probability of obtaining at least 14 of the 16 robustness factor settings under the null hypothesis of p = 0.5 is ~0.2 %. Such a low probability event is rejected as unrealistic, allowing the conclusion that elevated temperature is statistically significant in decreasing modulus irrespective of the robustness factor settings. The importance of the temperature effect is supported further by Fig. 9(a), which shows the values of the modulus ratio for different temperatures arranged in order of increasing magnitude for 54 combinations of exposure conditions. It is evident that modulus data for 40°C or 50°C are always

Copyright by ASTM Int’l (all rights reserved); Tue May 6 12:07:08 EDT 2014

Downloaded/printed by

Rochester Institute Of Technology pursuant to License Agreement. No further reproductions authorized.

located at the bottom left comer of the plot, which indicates a lower modulus. Moreover, there is a large “local" (i. e., for that particular combination of robustness factors) temperature effect on modulus reduction, which is manifested in the large within-block difference (i. e., tall blocks). The existence of (a) consistently large block heights and (b) consistent temperature arrangement within blocks demonstrates that the temperature effect on modulus reduction is important. Note that less important factors will have only one of these two properties, and unimportant factors will have neither.

The decrease in sealant modulus as a function of exposure temperature suggests that chain scission is more likely than cross-linking as the dominant degradation mode. Thermally enhanced chain scission may be attributed to an increase in the average kinetic energy of polymeric chains and other reactants with increasing temperature, thereby leading to faster sealant degradation. In addition, temperature contributes directly to degradation by increasing the diffusion rates of oxygen and radicals, further enhancing the accessibility of oxygen and radicals for the degradation process.

Copyright by ASTM Int’l (all rights reserved); Tue May 6 12:07:08 EDT 2014

Downloaded/printed by

Rochester Institute Of Technology pursuant to License Agreement. No further reproductions authorized.

Effect of Moisture

The block plots of raw and averaged data targeting RH levels (i. e., 0 %, 25 %, 50 %, and 75 %) over 14 different combinations of robustness factors are shown in Figs. 10(a) and 10(b). Although the bar heights for some robustness factor settings are considerably large, the local arrangement of RH levels within each bar depends on the settings of robustness factors on the horizontal axis. Unlike temperature, RH exhibits fairly consistently large block heights but inconsistent local RH level arrangements over all settings of robustness factors. This implies that RH is a less important factor. From Fig. 10(b), RH levels of 50 % and 75 % resulted in the greatest modulus reduction in 7 of the 14 settings of robustness factors. Based on binomial considerations, the probability of this happening by chance is ~60 %. The usual cutoff of 5 % has not been achieved here. Therefore, this observation, coupled with the inconsistency of the effect of RH over all of the robustness factor settings, suggests that RH is statistically less important. This observation is supported further by Fig. 9(b), which shows that many, but not a majority, of the data points associated with 50 % and 75 % RH are located at the bottom left corner of the plot. Although the RH effect on modulus decrease is not robust or universal over various robustness factor settings, the role of RH in modulus decrease cannot be dismissed for specific exposure conditions, i. e., RH might interact with other environmental factors. The inconsistency in the local RH arrangement in each bar across all of the robustness factor settings [Figs. 10(a) and 10(b)] lends support for such a hypothesis; this is discussed in more detail shortly.