QUALITY OF TROPICAL HARDWOOD FLOORS MADEIRAS

– This study aimed to determine the basic density of the wood and to simulate the performance of ﬂ ooring produced with wood from ﬁ ve tropical species: Dipteryx odorata (Cumaru), Handroanthus spp. (Ipê), Hymenaea Courbaril (Jatobá), Astronium Lecointei (Muiracatiara), and Bowdichia virgilioides (Sucupira-Preta). Falling steel sphere, static and dynamic friction, indentation caused by loads applied in small areas, rolling load, and abrasiveness impact tests were simulated. The results were subjected to analysis of variance and Tukey’s test at 5% signiﬁ cance, and Pearson’s correlation was performed between the basic density and the indexes of each ﬂ ooring. The basic density of the wood ﬂ ooring made from the evaluated species ranged from 0.735 to 0.958 g.cm -3 . D. odorata , Handroanthus spp. and H. courbaril woods were classiﬁ ed as heavy, while those of A. Lecointei and B. Virgilioides as moderately heavy. It was possible to indicate D. odorata , Handroanthus spp., H. courbaril , and B. Virgilioides ﬂ ooring for environments with intense traﬃ c where there is dragging or falling objects such as industries and companies. The A. Lecointei ﬂ ooring can be used in residential environments with light traﬃ c, where the loads exerted are low. There was a correlation between the basic wood density and the ﬂ ooring use simulation tests.


1.INTRODUCTION
Wood is used prominently in the manufacture of higher value-added products (PMVAs) such as doors, frames, fl oors, and furniture, among others. Hardwood fl oors are widely used in residential, commercial and industrial environments with the aim of improving the appearance of surfaces, in addition to off ering thermal and acoustic comfort (ANPM, 2015). Rocha et al. (2014) emphasize that wood is a durable and resistant material, and it is synonymous with comfort and warmth when used for fl oor production, which makes it highly appreciated.
The physical and mechanical properties and aesthetic attributes should be evaluated to select the species to be used for producing fl ooring. The density of wood is one of the main factors to be considered in the selection of raw material, because dense woods result in more resistant fl oors (Padilha et al., 2006;Rocha et al., 2014;Blanco et al., 2015). Tropical species are the most used in fl ooring production (Oliveira et al., 2019) because they have higher densities and diff erent combinations of colors and designs.
Flooring should be suitable for use and the environment where it will be inserted, therefore aesthetics, quality, economy and safety should be considered. It is noteworthy that wood fl ooring is subject to risks when in use such as damage due to objects falling, wear by abrasive elements, wear by people traffi c, dragging of objects, among others (Blanco et al., 2015), which visibly aff ects its aesthetics.
Faced with the diffi culty of evaluating the mechanical performance of wood fl oors in the long term, tests which simulate the performance in use are used to evaluate and ensure their quality, obtain information for their best use and also to assist in the choice of raw material. According to Padilha et al. (2006), it is necessary to evaluate the fl ooring characteristics and their behavior in use to propose their proper use; this results in a diff erential in the market whether to add value to the fi nal product, or to ensure the quality of the product to the consumer.
In this sense, wood fl ooring use simulation tests are an important tool to evaluate their behavior. They also make it possible to select tropical species which have quality for fl ooring production or to indicate their best use. Therefore, the objectives of this study were to determine the behavior in use of the most commercialized tropical species fl ooring by the companies in Mato Grosso and to evaluate the correlation between basic wood density and fl ooring quality.
A total of 50 fl oor pieces with the dimensions 12.7 x 27.9 x 2 cm were selected for each species, which is in agreement with the company's marketing standard.

2.1.Basic density
The basic wood density of the evaluated fl oors was determined according to the standard ABNT NBR 11941 (ABNT, 2003), adapted for the size of the specimens. Twenty fl oor samples were used for each species, totaling 100 specimens.

2.2.Flooring use simulation tests Preparation of specimens
To perform the wood fl ooring use simulation tests, 40 fl ooring pieces of each species were processed to manufacture 300 specimens, in the dimensions recommended by ASTM D 2394-05 (ASTM, 2011) of 24 x 12 x 2 cm, and by Martins (2008) of 9.5 x 9.5 x 2 cm, with adjustments made to the fl oor length, but with no change in thickness or faces.
The specimens were subsequently taken to the acclimatization chamber [T = (20 ± 2) ºC and UR = (65 ± 5) %], with the purpose of keeping them with equilibrium humidity equal to 12% to perform the fl ooring use simulation tests. The specimens did not receive any fi nishing products.
The following tests were performed according to ASTM D 2394-05 (ASTM, 2011): impact of falling ball, static and dynamic friction, fl oor surface, indentation from small area loads, and rolling load. For the fi rst, 100 samples were used per treatment and for the others, 50, all with dimensions equal to 24 x 12 x 2 cm. The abrasiveness test was conducted according to Martins (2008), in 50 samples per treatment, with dimensions of 9.5 x 9.5 x 2 cm.
The fl ooring of the evaluated species were grouped into classes according to the classifi cation proposed by Oliveira et al. (2019), which established wood fl ooring quality classes based on mean values of the fl ooring use simulation tests.

2.3.Statistical analyses
The experiment was installed according to a completely randomized design with fi ve treatments (species) and 50 repetitions (pieces of fl oors), totaling 250 sample units.
The data were submitted to the Shapiro-Wilk tests to assess the normality of the errors, and Bartlett's test to test the homogeneity of the variances. The data were transformed when they did not meet the assumptions of normality and homogeneity. Next, the results were subjected to analysis of variance (ANOVA) to verify the diff erences between the evaluated treatments. The Tukey test was applied at 5% signifi cance when there were signifi cant diff erences. Pearson's correlation was conducted with the pairs of basic density values and the indexes of each fl ooring use simulation test.
In the steel ball shooting simulation test, the regression analysis was performed with the pairs of values from the produced indentation and the height of the ball falling, obtaining an adjustment of the fi rst degree equation. Thus, it was possible to calculate the indentation index for the height of 180 cm for each evaluated species, according to the recommendation of ASTM D 2394-05 (ASTM, 2011).
Statistical analyses were performed using the R-3.5.1 software program (Fereira et al., 2018).

3.1.Basic density
The results obtained for the basic density of the solid hardwood fl oors of the fi ve evaluated species are shown in Figure 1.
The D. odorata and Handroanthus spp. wood presented the highest averages of basic density, and did not diff er statistically between them. The A. lecointei and B. virgilioides woods showed the lowest averages and were statistically similar, while the H. courbaril wood diff ered statistically from all the others (Figure 1). The variation among all species was low, and a general coeffi cient of variation of 7.91% was observed.

Falling ball impact test
The relationship between the depression caused by the impact of the steel ball and the heights of fall on the solid hardwood fl oors of the fi ve evaluated species are shown in Figure 2.
It was found that the release of the steel ball on the surface of the fl oor caused deformation for all species; as verifi ed by Blanco et al. (2015), with the increase in the ball release distance, the greater the depression (Figure 2).
The mean indentation index calculated for the height of 180 cm for the D. odorata wood was equal to 0.195, while for Handroanthus spp. it was 0.184, for H. courbaril 0.316, for A. lecointei 0.339, and for B. virgilioides 0.468. There was no signifi cant diff erence between the averages of the static friction test for the wood fl ooring of the diff erent species, presenting a variation of 22.44% between the highest and lowest mean value observed.

3.3.Static and dynamic friction tests
The mean dynamic friction coeffi cient showed a signifi cant diff erence, with the highest mean value for the D. odorata (0.323) fl ooring, and the lowest value for A. lecointei (0.130). The Handroanthus spp., H. courbaril and B. virgilioides woods were statistically similar. Averages followed by the same letter do not diff er statistically at 5% signifi cance by the Tukey test. Médias seguidas pela mesma letra não diferem estatisticamente a 5% de significância, pelo teste Tukey. Figure 3b shows the results obtained for the indentation of loads applied in small areas test after 50 and 100 cycles for the solid wood fl ooring of the fi ve evaluated species.

3.4.Floor surface indentation from small area loads
The D. odorata, Handroanthus spp. and H. courbaril wood fl ooring presented the lowest mean values with the application of loads in small areas in all cycles, which indicates greater resistance of the fl oors due to the lower depression. The A. lecointei wood presented the highest mean value, and therefore it was considered the least resistant, diff ering statistically from the others. With the exception of the hardwood A. lecointei fl ooring, all the evaluated woods were considered similar to each other according to the Tukey test at 5% signifi cance ( Figure 3b).  Table 1 shows the results obtained for the rolling load test after 10, 25 and 50 cycles for the solid wood fl ooring of the fi ve evaluated species.

3.5.Rolling load test
The D. odorata, Handroanthus spp. and H. courbaril hardwood fl oors showed greater resistance to dragging of objects, because they presented the lowest averages of the depressions obtained in the rolling load test, and did not diff er statistically among each other in all cycles (Table 1).  3 -a -Valores médios e desvio-padrão dos coefi cientes de atrito estático e dinâmico para os pisos das madeiras avaliadas. b -Valores médios e desvio-padrão das endentações causadas por cargas aplicadas após 50 e 100 ciclos em pequenas áreas para os pisos das madeiras avaliadas.
Averages followed by the same letter do not diff er statistically at 5% signifi cance by the Tukey test. Médias seguidas pela mesma letra não diferem estatisticamente a 5% de signifi cância, pelo teste de Tukey.
It can be verifi ed that only the abrasive assay of the fi ve fl ooring use simulation tests performed in the present study did not show signifi cant correlation with wood density (-0.08).

4.1Basic density
The D. odorata, Handroanthus spp. and H. courbaril woods were classifi ed as heavy, while the A. lecointei and B. virgilioides woods were classifi ed as moderately heavy, according to Rocha et al. (2014), who elaborated a classifi cation regarding the basic density of wood for use in paving.
Considering the results found for the basic density (Figure 1) of the woods evaluated in the present study, an emphasis on D. odorata (cumaru), Handroanthus spp (ipê) and H. courbaril (jatobá) woods are indicated for fl ooring production because, according to Rocha et al. (2014), timbers considered very heavy and heavy are ideal for use in fl oors, since they are more resistant and have greater durability over time.

4.2Falling ball impact test
This index expresses that the lower its value, the better the wood resistance for this type of force. Therefore, the fl oors which presented higher resistance to the impact of the steel ball were D. odorata and Handroanthus spp., with values lower than 0.200.
In a study with native commercial timber for manufacturing fl oors, Oliveira et al. (2019) found an indentation index of 0.186 for D. odorata fl ooring, 0.241 for Tabebuia impetiginosa and 0.390 for Bowdichia nitida, being similar to those observed in this study.  D. odorata and Handroanthus spp. wood presented the lowest indentation indexes and had the highest density ( Figure 1). In contrast, B. virgilioides presented the highest indentation index, and had the lowest density ( Figure 1). Blanco et al. (2015) worked with Tectona grandis medium density wood of 0.540 g.cm -3 and found a mean value equal to 0.490 for the indentation index, being higher than that observed for the fl ooring of all the species evaluated in this study. The higher indentation index is related to lower density of the T. grandis wood compared to the tropical species of this study.
According to the classifi cation of Oliveira et al. (2019) in considering the shooting steel ball test, the D. odorata and Handroanthus spp. wood fl oors fall into the high quality class (< 0.180), while the H. courbaril, A. lecointei and B. virgilioides fl oors are classifi ed as low quality (> 0.301). Therefore, as the impact resistance to the steel ball was low for H. courbaril, A. lecointei and B. virgilioides, it is necessary to protect the fl ooring of these woods from eventual object falls to avoid further deformation and damage.

4.3.Static and dynamic friction assay
Lower static coeffi cient values indicate that the surface is smoother and less force is needed to initiate movement on the fl oor. In practical terms, this fact represents the slip resistance.
A lower value for the dynamic friction coeffi cient indicates less force required for the continuity of the movement. Therefore, the D. odorata fl ooring was considered the least slippery, which constitutes a safety factor in the risk of people falling during traffi c.
It is important to mention that the specimens did not receive any fi nishing products, and were only sanded. Therefore, the coeffi cients of friction found represent the superfi cial resistance of natural wood. These values were lower than those observed for the same species in the present study due to a layer of varnish that the specimens received.
In evaluating hardwood fl oors of diff erent Eucalyptus urophylla clones, Padilha et al. (2006) found static and dynamic friction coeffi cient values equal to 0.376 and 0.230, respectively. In addition, Martins et al. (2013) found static and dynamic friction coeffi cient values for Eucalyptus fl oors equal to 0.250 and 0.150, respectively. In studying T. grandis wood, Blanco et al. The coeffi cients of friction are determinant for the choice of the wood species depending on the use and traffi c of the site where the fl oor will be installed, and it is important that they do not provide a slippery surface. In other words, the higher the value of the friction coeffi cient, the less slippery and the safer the tread surface (Oliveira et al., 2019), which can help avoid falls caused by slipping and allows safe transit (Martins, 2008).

4.4.Assay of load-fi lling applied in small areas
The behavior of D. odorata, Handroanthus spp. and H. courbaril woods, classifi ed as heavy by their basic density (Figure 1), occurred as reported in the literature. According to Oliveira et al. (2019), high densities provide greater strength and hardness, which refl ect the resistance to applied load. For the A. lecointei hardwood fl ooring, which has the smallest density (Figure 1), the depression observed in the assay was higher.
In evaluating the behavior of fl ooring use in simulation, the smallest diff erence between the depressions obtained in the fi rst and the second measurement interval (50 and 100 cycles) was observed for the Handroanthus spp. fl ooring (8%), which shows that the increase in the number of cycles did not signifi cantly increase wear on the fl oors of this species. For the H. courbaril, A. lecointei and B. virgilioides fl oors, diff erences in depressions between 50 and 100 cycles of 30, 25 and 26% were observed, respectively. In contrast, the greatest strain variation of 121% was observed for the D. odorata fl ooring, which indicates that this fl ooring will present greater deformation over time of use. Oliveira et al. (2019) found average indentation values after 100 cycles, equal to 0.016 mm for B. nitida fl oors (sucupira) and 0.024 mm for Mimosa scabrella (Amendola) fl oors, constituting values close to those observed for the D. odorata, Handroanthus spp., H. courbaril and B. virgilioides woods evaluated in this study (Figure 3b). In contrast, the authors did not observe depressions after 100 cycles for T. impetiginosa (Ipê) or D. odorata (cumaru) wood. Padilha et al. (2006) found indentation values for E. urophylla wood after 100 cycles of between 0.049 and 0.092 mm; Martins et al. (2013) found values of 0.050 mm after 100 cycles for E. microcorys; and Blanco et al. (2015) found 0.10 mm of depression for T. grandis fl oors. In evaluating Hovenia dulcis wood for fl ooring production, Marchesan (2016), found 0.11 mm of depression after 100 cycles. The values found in the cited studies were higher than in the present study, and this diff erence was due to the lower density of the wood used by the aforementioned authors.
According to the classifi cation proposed by Oliveira et al. (2019), the D. odorata, Handroanthus spp., H. courbaril and B. virgilioides hardwood fl oors were classifi ed as intermediates (0.006-0.030) and the A. lecointei fl ooring was classifi ed as low quality (>0.031). Therefore, the resistance to indentation by loads applied in small areas was satisfactory for D. odorata, Handroanthus spp., H. courbaril and B. virgilioides hardwood fl ooring. These species are considered suitable for use in residential and local environments which have intense traffi c of people with high heels.

4.5.Rolling load test
According to Marchesan (2016), the lower the depression caused in the wood, the greater its resistance will be, meaning the better its performance in relation to the eff orts made against the wood fl ooring in use. The best obtained results of resistance may be related to higher values of basic density for the cited species (Figure 1), because the higher the wood density, the lower the depression caused in the fl oor.
The A. lecointei and B. virgilioides hardwood fl oors showed the highest averages of depressions and were considered statistically similar in each of the cycles. The lower results obtained for these species are related to lower basic density values (Figure 1). Therefore, the variation in the values of the depressions caused by the drag was generally inversely proportional to the variation in wood density.
The mean values obtained from depressions after 50 cycles for the wood fl ooring of the species evaluated in this study were lower than the average reported by other authors (Padilha et al., 2006;Santos et al., 2010;Blanco et al., 2015;Marchesan, 2016), including for B. virgilioides fl ooring, which presented inferior behavior to that of the others (0.2460 mm). Oliveira et al. (2019) found mean depression values after 50 cycles equal to 0.110 mm, 0.105 mm and 0.205 mm for D. odorata, T. impetiginosa and B. nitida woods, respectively. These values were higher than those observed in this study for the D. odorata and Handroanthus spp. fl oors, and close to that observed for B. virgilioides (Table 1).
It was found that the hardwood fl ooring evaluated in this study showed increased deformation with the increase in the number of rolling load cycles (Table 1). However, the increase was higher in the interval of 10 and 25 cycles when compared to the range of 25 to 50 cycles.
For the D. odorata fl oors, the proportion of increase in depression was 38% in the range of 10 to 25 cycles and 29% in the range of 25 to 50 cycles, while the increase was 82 and 24% for the Handroanthus spp. fl ooring; 21 and 12% for H. courbaril; 40 and 13% for A. lecointei; and 51 and 14% for B. virgilioides. This fact is explained by the compaction suff ered by wood after successive rolling load passages, which confers greater mechanical strength (Martins, 2008).
According to the quality classifi cation of Oliveira et al. (2019), the D. odorata, Handroanthus spp. and H. courbaril wood fl oors were classifi ed as high quality (< 0.120), and are therefore indicated for installation in places which receive intense traffi c and frequent drag of heavy objects such as machines occurs. The A. lecointei and B. virgilioides fl oors were positioned in the intermediate quality class (0.121-0.300) and are indicated for installation where traffi c is not intense, as in residential environments.
When comparing the results of depressions obtained by abrasive assays with other potential species for fl ooring production, it appears that the results obtained in the present study were superior. Martins et al. (2013) found medium depressions equal to 0.055 mm, 0.073 mm and 0.076 mm for E. Cloeziana, E. Microcorys and Corymbia maculate fl oors, respectively. In evaluating fl oors manufactured by a combination of bamboo blades and (edge-glued panels) EGP of pinus wood, Cortez-Barbosa et al. (2014) (2016) obtained 0.08 mm of depression in the abrasive assay in evaluating H. dulcis wood for fl ooring production, constituting higher results than those observed in this study.

4.7.Pearson Correlation
There was an inversely proportional relationship in the shooting steel ball impact test between the indentation index and wood density, which was confi rmed by the strong negative correlation of -0.92. This same correlation was verifi ed by several authors (Martins et al., 2013;Blanco et al., 2015;Oliveira et al, 2019;Marchesan, 2016).
The static friction and dynamic friction coeffi cients showed a moderately positive signifi cant correlation of 0.58 with the wood density. Thus, dense woods result in safer fl ooring for people traffi c.
It was verifi ed that the resistance to loads applied in small areas was aff ected by wood density, because there was a moderately negative signifi cant correlation of -0.58, which corroborates the results obtained by Blanco et al. (2015). Therefore, denser timbers result in more resistant fl ooring to loads applied in small areas.
In the rolling load test, the variation in the depression values caused by the drag was generally inversely proportional to the variation in the wood density, and a strong negative correlation of -0.93 was verifi ed. The resistance to this type of eff ort is the result of the hardness and density of the material, with the latter being the main factor to be considered in the classifi cation of wood uses (Blanco, 2016).
There was no signifi cant correlation between the basic wood density and the results obtained for the abrasive assay. In performing abrasion tests in bamboo mosso (Phyllostachys pubescens) for manufacturing solid fl oors, Berndsen et al. (2014) determined that higher densities do not always result in greater resistance to wear. This fact corroborates the results obtained in this study, in which the wood of the D. odorata and Handroanthus spp. species which have the highest densities (Figure 1) did not present the greatest resistance to wear caused by the abrasive test (Figure 4).

5.CONCLUSIONS
The D. odorata, Handroanthus spp. and H. courbaril wood fl oors were classifi ed as heavy, and those of A. lecointei and B. virgilioides as moderately heavy.
There was a signifi cant correlation between the basic wood density and the steel ball, friction, the fi lling of loads applied in small areas and the rolling load impact tests.
It was possible to indicate the D. odorata, Handroanthus spp., H. courbaril and B. virgilioides fl oors for environments with intense traffi c and/or where there is the drag or fall of objects, such as for industries and companies. The A. lecointei fl ooring can be used in residential environments with light traffi c where the exerted loads are low.

AUTHOR CONTRIBUTIONS
The manuscript is part of the master's thesis of the fi rst author. Aylson Costa Oliveira, Bárbara L. C. Pereira and José Reinaldo M. da Silva contributed to the study conception and design. Material preparation, data collection and analysis were performed by Ana Carolina Silva Costa, Mônica B. de Oliveira and Carolina N. Xavier. The fi rst draft of the manuscript was written by Ana Carolina Silva Costa and all authors commented on previous versions of the manuscript. All authors read and approved the fi nal manuscript.

7.REFERENCES
American Society for Testing and Materials. Simulated service testing of wood and wood-base finish flooring: ASTM D 2394-05. Philadelphia, 2011.