Page 4 - Test e-book yru envi

N.M. Garz on-Barrero et al. / International Biodeterioration & Biodegradation 115 (2016) 266e276 269
double faced carbon conductive tape, and the micrographs were 2.2.7. Evaluation of mechanical and physical properties
obtained by backscattered electron image (BSE) using an acceler- After 12 months of exposure, an evaluation of mechanical
ated tension of 15 kV, at a work distance between 5.0 and 5.7 mm. properties of modulus of rupture (MOR) and modulus of elasticity
(MOE) under three-point bending test was performed on the par-
2.2.5. Image recording by digital scanner ticle boards (BCP and MDP), with the goal of identifying the critical
Recordings of the test samples were performed by an HP Pho- mechanical support characteristics of the material subjected to
tosmart Plus scanner for both materials before the mold growth natural exposure. After the exposure time the test samples were
tests and after 7, 14, 21, 28 days of exposure to accelerated test. For stored at 20 C in an environment of 65% relative humidity for one

the samples submitted to the natural aging test, images were ac- week. The equipment used for the mechanical test was the uni-
quired by the digital scanner after 12 months of exposure. versal test machine EMIC, model DL30000, following the recom-
mendations of the (ABNT NBR 14810-3, 2006) Brazilian Standard
2.2.6. Statistical analysis of mold fungi colonization (major span equal to 220 mm and test speed equal to 7 mm/min).
Statistical analysis was performed for two scenarios. Initially a Six replicates were analyzed for each material. The thickness re-
comparative analysis of fungal colonization induced by natural sidual swelling of the specimens was measured in five different
exposure and by the accelerated test was performed using a points along each side of the specimen using a digital caliper, ac-
regression model for the described data. Following this, a longitu- cording to ASTM D 1037 (2006) Standards. This measure was per-
dinal analysis of mold colonization in different periods during the formed before and after natural exposure aging and accelerated
accelerated test was performed. mold growth test. After that, statistical study was conducted by the
SAS program, version 9.3 for the variation analysis (factorial
2.2.6.1. Regression model for described data for comparative analysis ANOVA) for the following sources of variation: material (BCP and
of mold growth by natural and accelerated exposure. The variable MDP), surface (coated and non-coated), and fungal growth test
response analyzed in this experiment (degree of colonization) is a (accelerated and natural). To evaluate the effects of the means an F
polytomous, multi categorical and ordinal variable, that is, it can test was adopted with a significance level of 0.05.
add values in the set {0, 1, 2, …, 8, 9, 10} and is influenced by: a) the
type of material (BCP and MDP), b) the castor oil based poly-
urethane resin surface (coated or non-coated), c) test sample face 3. Results and discussion
(front or back surface related to natural exposure; face 1 and 2
related to accelerated exposure), and d) the type of exposure 3.1. Mold growth in natural weathering
(natural or accelerated aging) in mutually exclusive categories. Data
of this nature are referred to as discrete and cannot be treated in the After 12 months of exposure, some superficial changes were
analysis as continuous variables and, for this reason, the proper observed in relation to the surface appearance such as, loss of
analysis applied a regression model for discrete data (Agresti, brightness, darkening, loss of color uniformity, and mold growth;
2012). the identification of fungal species was not the objective of the
To indicate the degree of fungal colonization in the materials, current study.
the Proportional Odds Model (POM) was used, which is a multi- Table 1 presents the results according to the grades established
variate extension of the Generalized Linear Models (McCullagh, for the degree of colonization visualized in each test sample, the
1980), and it allows the estimation of probabilities associated BCP and MDP materials, after 12 months of exposure to natural
with each of the response categories. weathering. The grades obtained after 12 months in the materials
are lower on the front face in relation to the back face indicating for
2.2.6.2. Longitudinal analysis of mold colonization in the accelerated both materials a greater degree of colonization on the sides directly
test. In the accelerated test (ASTM D 3273-12, 2012), a longitudinal exposed to the weathering degradation agents.
evaluation of mold growth in the specimens was performed after 7,
14, 21, and 28 days on each of the faces (1 and 2). In these situations, d. Mold growth in accelerated test (ASTM D 3273-12, 2012)
a possible dependency between the observations must still be
considered (as the observations were treated on the same test Fig. 2 represents the effect of exposure time on the BCP and MDP
sample over time). Ignoring this correlation when it actually exists,
could lead to distortions in the standard errors of the model co-
Table 1
efficients, which are also indicators of analysis accuracy. Thus, an Mold growth based on evaluation grades from the ASTM D 3273-12, 2012 Standard,
alternative to consider this dependency is to incorporate the after 12 months of natural weather exposure in Pirassununga, SP Brazil.
random effect in the model, i.e., to adopt the methodology of mixed
Material Surface Colonization rating
models, which includes within its structure another random term,
in addition to errors. Accordingly, the correlation between obser- Front Back Front Back
vations is the result of this random variable's necessary inclusion, BCP Coated 1 0 MDP 0 4
especially in longitudinal studies. Thus, the marginal probabilities 1 3 0 4
are determined not only by the fixed effects of covariates, but also 0 3 0 2
0 4 0 4
by a random component.
0 3 0 2
The applied technique was the maximum likelihood via squared 0 3 0 3
Gaussian (Molenberg and Verbeke, 2005). Additionally, the random Non-coated 1 8 4 4
effect was evaluated, that is, a test was carried out to check its 1 8 2 4
statistical significance. In case it was not significant, a fixed effects 1 9 4 5
1 8 3 7
model was adopted. In this work, both the data analysis and the
1 8 2 5
model adjustments were performed with the help of the ordinal 1 8 4 7
package (Christensen, 2011) available in the R software, version
Rating (degree of colonization) ASTM D 3273-12: 10 (0%); 9(1e10%); 8(11e20%);
2:14 (R Development Core Team., 2011), and the significance of the 7(21e30%); 6(31e40%); 5(41e50%); 4(51e60); 3(61e70%); 2(71e80%); 1(81e90%);
adopted tests was a 0.05. 0(91e100%).