Quality Assessment in Industrial Production of Plywood by Stiffness and Strength Properties in Bending

This research aimed to investigate, by means of the strength (MOR) and stiffness (MOE) properties in bending, the homogeneity in the manufacture of plywood ( Pinus sp wood and phenolic resin) produced by an industry located in the state of Paraná (PR - Brazil). The factors and levels investigated in this study were the choice of panels (A, B, C and D), the extraction position of the specimens (Center; Side) and fiber orientation (Parallel, Perpendicular), providing a full factorial design of 4 1 2 2 type, with 16 different treatments. On the panel, were extracted 20 specimens (500×75×18mm) for bending tests. ANOVA results showed that the choice of panel and fiber orientation was significant in the properties of strength and stiffness in bending, this is not occurring with the extraction position of the specimens. The MOR and MOE of the specimens obtained in the parallel direction were 87% and 141% higher than the MOR and MOE of the specimens evaluated in the perpendicular direction, respectively, and with respect to non-equivalence found between the mechanical properties by the choice of the panel, for results with greater reliability it is necessary the development of new research with a significantly higher number of plywood.


Introduction
The plywood is a panel based on wood veneers bonded with adhesive, pressure and high temperature for cross-lamination technique, that is, the direction of the grain of a veneer is perpendicular to grain veneer adjacent.
The arrangement of the wood veneer on panel and the use of adhesive provide to plywood greater dimensional stability and mechanical properties mo re uniform co mpared to wood, decreasing the anisotropy of solid wood.
Adhesives for plywood most used industrially are urea-formaldehyde, melamine-formaldehyde and phenolformaldehyde. Determinant for life of p lywood is its correct use in accordance with its recommendation for use. Generically, for internal, intermed iate and external uses are recommended, respectively, adhesives urea-formaldehyde, melamine-fo rmaldehyde and phenol-formaldehyde [1].
Currently there is a scarcity of tropical wood for various purposes, including for use as raw material in industries that produce veneer and plywood. A good alternative to meet the demand of this sector providing good quality woods are the reforestation of Pinus and Eucalyptus [2,3].
The researches about the properties of strength and stiffness of plywood using Pinus wood are important for purposes that may be earmarked, including civil construction and industry [2, 4, and 5].
Among the mechanical properties that should be known to characterize a lot of plywood, we have the modulus of elasticity (MOE) and modulus of rupture (MOR). In Brazil, these characteristics are determined fro m mechanical tests normalized by ABNT NBR 9533 [6] "Determination of strength to bending plywood".
Research about values of MOE and MOR of p lywood produced with Pinus taeda and Pinus oocarpa with three different formulat ions of adhesive phenol-formaldehyde revealed that the anatomical characteristics of the wood (adult and juvenile wood, width of growth rings and early and late wood) have influenced the mechanical properties of the plywood [7].
Others researches has shown that plywoods of Pinus sp. with three different formulat ions of PVA adhesive and two combinations of temperature and time of pressing, finding lower values of M OE and MOR co mpared to panels made with urea-formaldehyde and phenol formaldehyde adhesives, which are the most used by the industrial sector [8].
Research with ply wood made with wood fro m a hybrid Pinus elliottii var. caribea x Pinus elliottii var. hondurensis analyzing two points removed of the tree trunk, two adhesive grammages (380 and 420 g /cm²) and two t imes of pressings (8 and 12 minutes). Among the authors' conclusions, the wood of this hybrid presented good results for making plywood for purposes such as civil construction, furniture and packaging [9].
The aim of this research consists to analyze the variation of the values of MOE and MOR at different points of same plywood panels produced industrially for evaluation of homogeneity of the industrial manufacturing process of plywood.

Material and Methods
For this research we used four panels of plywood collected on the line production of an industry of Paraná (PR -Brazil). The panels were manufactured with wood veneers of Pinus sp. with dimensions of 1.25m×2.50m×18 mm, with 13 veneers in each panel bonded with phenolic resin.
Were fabricated and tested 80 specimens for bending tests according to the procedures of the normative document ABNT NBR 9533 [6]. With the purpose of determining the influence of the withdrawal position of the specimens in the MOE and MOR, 40 specimens were removed fro m the center and the others 40 removed fro m the edge of the panel in parallel and perpendicular d irections to the grain of the external veneer ( Figure 1). The factors and levels investigated in this study were the panels[Bo] (A; B; C; D), the extract ion position of the specimens[EP] (Center (C); Side (S)) and fiber orientation [FO] (Parallel (Pa); Perpendicular (Pe)), provid ing a fu ll factorial design of the 4 1 2 2 type, with 16 distinct treatments (Tr), exp lained in Table 1. The factors and interactions of the factorial design were evaluated via analysis of variance (ANOVA ) at the 5% level of significance (α), assuming the equivalence between the treatment means as null hypothesis (H 0 ) and the non-equivalence between means as alternative hypothesis (H 1 ). P-value less than the significance level imp lies rejecting the null hypothesis, accept it otherwise. To validate the ANOVA, Anderson-Darling (normality) and Bart lett and Levene (ho mogeneity of variances) test were used, both formulated at a significance level of 5%. For the Anderson-Darling test, the null hypothesis assumed was the normality of d istributions, and non-normality as the alternative hypothesis. P-value greater than 5% imp lies accepting H 0 , rejecting it otherwise. For Bartlett and Levine's tests, the null hypothesis assumed was the homogeneity of variances between treat ments, and the non-equivalence as the alternative hypothesis. P-value greater than 5% imp lies accepting H 0 , reject ing it otherwise. Factors considered significant by ANOVA, the sequence was used the Tukey test for mult iple co mparisons. Table 2 presents the average values ( x ) and coefficients of variation (CV) of the MOR and MOE of the panels evaluated.  Figures 2 and 3 show the results of normality and homogeneity of variance, respectively. For P-values found are both greater than 5%, g iven that the distributions are normal and the variances between treatments are equal, validating the model ANOVA.  Table 3 presents the ANOVA results for the MOR and MOE of plywood, with underlined P-values considered significant (P-value <.05). Fro m Tab le 3, it appears that the choice of ply wood and fiber orientation to obtain the modulus of elasticity and strength in bending were considered significant by ANOVA, which d id not occur with the extract ion position, which provided equivalent values for both properties, revealing the uniformity achieved in the manufacture of the panel.

Results and Discussions
By fiber orientation was expected that the MOE and MOR in parallel direct ion were significantly higher than the MOE and MOR in the perpendicular direction. As to the choice of the panels, it was expected that the results of the propert ies of stiffness and strength were equivalent, which has not occurred, indicated inho mogeneity in the manufacture of the panels. These results may be different for a considerable number of samples, highlighting the need for further studies. Table 4 presents the results of Tukey's test for grouping (Gr) the levels of the factors considered significant by ANOVA. Fro m Table 4, the same letters imp ly average with equivalent treatments. The MOR and MOE of the specimens obtained in the parallel d irection were 87% and 141% higher than the MOR and MOE of the specimens evaluated in the perpendicular direction, respectively.

Conclusions
The results showed that the choice of the panels used in the development of this research was significant in the properties of strength and stiffness in bending, this is not occurring with the extract ion position of the specimen panel, which provided equivalent results.
The values of MOE and M OR fo r the fiber orientation factor was significant, with those extracted fro m the fibers in the direction parallel the best results, as expected.
Even with a significant choice of panels in obtaining the properties of strength and stiffness in bending, so one cannot conclude on the homogeneity achieved in the manufacture of the panels by the company it is necessary the development of new research with a significantly higher nu mber o f panels, thus increasing the reliability of the results.