You are on page 1of 7

Measurement 165 (2020) 108133

Contents lists available at ScienceDirect

Measurement
journal homepage: www.elsevier.com/locate/measurement

The effect of granularity on surface roughness and contact angle in wood


sanding process
Bin Luo ⇑, Jian Zhang, Xue Bao, Hongguang Liu, Li Li
College of Materials Science and Technology, Beijing Forestry University, No. 35 Tsinghua East Rd, Haidian District, Beijing 100083, China
MOE Key Laboratory of Wooden Materials Science and Application, College of Materials Science and Technology, Beijing Forestry University, No. 35 Tsinghua East Rd,
Haidian District, Beijing 100083, China

a r t i c l e i n f o a b s t r a c t

Article history: Sanding is a common machining method to obtain good surface quality. The 2D surface roughness Ra pre-
Received 12 September 2019 sents high error due to the biological structure of wood materials. In this study, a new calibration method
Received in revised form 10 May 2020 of reference plane is proposed to measure 3D surface roughness Sa, especially for large format panels.
Accepted 18 June 2020
Besides, the effects of G (granularity) on Ra, Sa and h (equilibrium contact angle) are investigated on
Available online 30 June 2020
Korean pine (Pinus koraiensis Sieb. Et Zucc.), JATOBA (Hymenaea cunrbaril L.) and medium density fiber-
board (MDF) respectively. The experimental results show the standard deviations of Sa are much lower
Keywords:
than that of Ra. As G increased, the variation of Ra, Sa and h of Korean pine and JATOBA are distinct from
Wood sanding
Granularity
MDF. The results of Grey Relevancy Degree (GRD) indicate MDF shows an obscure correlation between Sa
Surface roughness and h. The correlation between Sa and h for Korean pine is the highest.
Contact angle Ó 2020 Elsevier Ltd. All rights reserved.

1. Introduction significance, what exactly the influence mechanism of the surface


roughness on the coating performance still keeps a hotspot in
Some advantages make wood sanding to be one of the most wood machining research.
important machining methods in wood processing. They are: Precisely measuring the surface roughness of sanded wood
majority cutting with negative rake angles makes it less likely to workpieces is fundamental to investigate the relationship between
create advance splitting and crack expansion for wood surface; surface roughness and coating performance. In previous published
the ‘‘elastic sanding” feature of belt sanding makes it easier to works, two-dimensional (2D) contact method by using stylus was
achieve self-adapting and dysmorphic sanding; the cost of sanding widely applied to measure the surface roughness of wood and
tools is lower than milling tools and sawing blades. It’s acknowl- metal, with primary parameters like Ra (arithmetical mean devia-
edged that the purposes of sanding are mainly about thickness tion of the profile), Rz (mean maximum height of the profile) and
calibrating, surface smoothing and pre-cured layer removing [1]. RSm (mean width of the profile) [6–8]. However, one of the typical
In wood products manufacturing, good surface quality is expected characteristics of wood is its internal structure, which has anatom-
and considered as a consequential matter, especially prior to gluing ical cavities such as vessels and cell lumens inside [9]. Some of the
and painting, which shows a significant effect on the coating per- cavities are visible to the eyes like vessels consisting of vertical
formance [2]. units doing the transportation, the diameter of which can be up
According to previous research, the surface roughness strikingly to 300 lm. Some are relatively small like the tracheids (cavities
affects the equilibrium contact angle (h) [3], which indicates the in the long parenchyma cells) in the range 10–20 lm [10]. Thus,
wettability of solid materials and is also used to analyze the paint- the stylus is prone to be interfered by these small anatomical
ing quality. Specifically, the smoother surface induces smaller h, cavities on machined wood surface during the process of contact
and consequently the adhesion of paint reduces [4]. So far, the rela- measurement, which could result in measurement error and low
tionship between surface roughness and painting quality has not accuracy. Besides, the sampling length is usually as small as
been clarified [5]. Given the vague relations and the practical 25 mm for the stylus, which shows complexity to obtain the value
of Ra for wood panels with large format.
⇑ Corresponding author at: College of Materials Science and Technology, MOE Key Three-dimensional (3D) non-contact measurement like optical
Laboratory of Wooden Materials Science and Application, Beijing Forestry Univer- profilometer can improve the 2D contact measurement of wood
sity, No. 35 Tsinghua East Rd, Haidian District, Beijing 100083, China. surface roughness for a better regularity and reliability of Sa
E-mail address: luobincl@bjfu.edu.cn (B. Luo).

https://doi.org/10.1016/j.measurement.2020.108133
0263-2241/Ó 2020 Elsevier Ltd. All rights reserved.
2 B. Luo et al. / Measurement 165 (2020) 108133

Nomenclature

G granularity H equilibrium contact angle


Ra arithmetical mean deviation of the profile
Sa arithmetical mean deviation of the surface

(arithmetical mean deviation of the surface) [11]. Li et al. [12] sample quantity. On the one hand, surface roughness varies with dif-
found that the mean variation of Sa was much smaller than Ra ferent sanding parameters [15]. On the other hand, there are many
when the sawing surface of Qurecus Coccofera was measured wood species showing complex cellulose structures. Therefore, Grey
through a 3D contact method, with the mean errors of Sa and Ra Relevancy Degree (GRD) analysis for Sa and h was conducted rather
all above 20%. But the 3D non-contact measurement technology than neural network predicting to evaluate the correlation between
of surface roughness has been so far developed for homogeneous Ra, Sa and h. The results would provide practical and theoretical
material like metal and glass. Wood materials’ fibrous texture value for wood materials sanding research.
and biological unevenness is inevitably to influence the determina-
tion of datum plane, which should be seriously considered. More-
over, machined wood surface always yields worse unevenness 2. Materials and methods
than metals, and wood products always have large surface areas,
which makes it inappropriate for existing 3D optical profilometers 2.1. Experimental details
with small scanning sizes.
To promote the precision and efficiency of roughness measure- As presented in Fig. 1, sanding experiments were conducted on
ment on sanded wood surface, a new method of setting a real ref- a single contact-roller sander. Details of equipment and materials
erence plane based on the sampling points has been proposed in used in this study were shown in Table 1 and Table 2, respectively.
this study. The calibration process of the reference plane can avoid Before the test, the wood workpieces should be double-sided
the overall non-uniform thickness of wood workpieces, and less planed and the MDF workpieces should be sanded to remove the
data processing makes it more suitable for large format wood pre-cue layer (double-sided for 4 mm). The workpiece was rigidly
panels. fixed on the workbench with a special fixture. The feed movement
Granularity (G) of the grits on abrasive belt shows a significant of workpiece was parallel to the wood texture during the whole
effect on Ra, but the dispersion of Ra is very large, which is attrib- sanding process. The specific sanding parameters were selected
uted to the biological and heterogeneous characteristics of wood at the sanding speed 5.35 m/s, the feed rate 1.49 m/min, and the
[13]. In fact, the surface roughness of sanded wood is mainly sanding thickness 0.2 mm.
caused by anatomical cavities’ structure and local deformation
derived from multi-grit cutting. Abrasive grits with different gran- 2.2. The measurement of surface roughness Ra and Sa
ularity induce various cells damage and machining waviness [10].
Using a blunt tool edge for sanding large-vessel wood species may To verify that traditional 2D surface roughness parameters can-
result in surface waviness due to compression. The wood structure not fully reflect and describe the integral sanded surface character-
however contains cavities and even cavities of smaller size are istics, Ra and Sa were selected to evaluate the sanded surface in
comparable to the radius of a sharp tool edge. In this sense, gran- this research, and the results of which were compared. The defined
ularity was emphasized and further investigated in this study to Ra and Sa can be better understood by Eqs. (1) and (2).
see its effect on Ra and Sa. Z lr
1
There are some studies about establishing a predictive model for Ra ¼ jzðxÞjdx ð1Þ
lr 0
surface roughness in wood sanding process. Bao et al. [14] used BP
neural network system to predict the surface roughness by using ZZ
1
2D contact method, but the results were not as good as expected Sa ¼ jzðx; yÞjdxdy ð2Þ
A
due to the large dispersion of original data and the insufficient A

Fig. 1. Experimental set-up.


B. Luo et al. / Measurement 165 (2020) 108133 3

Table 1
Details of experimental equipment.

Equipment Type Manufacturer Working range Resolution Remarks


Belt sander Single contacted-roll Qingdao machinery Maximum sanding speed is Thickness Contacted-roll is made of rubber with
with reciprocating manufacturing Co. Ltd, 12 m/s; maximum feed speed is precision is spiral patter. The hardness of the rubber
workbench China 5 m/min ±50 lm is 75 HD.
Shore hardness TH210 Beijing TIME high 0–100 HD  1 HD —
tester technology Co. Ltd,
China
Stylus profiler Surtronic3+ Taylor Hobson Ltd, UK Z-axis (vertical): 150 lm; Z-axis Stylus diameter is 10 lm. The tracing
X-axis (horizontal): 17.5 mm (vertical): speed is 1 mm/s.
0.01 lm
3D Profile Scanner Gocator 2330 Daheng Imaging Co. Ltd, Measurement range: 80 mm; Z-axis —
China Field of view: 47–85 mm (vertical):
6–14 lm;
X-axis
(horizontal):
44–75 lm
Contact angle OCA20 Shanghai Innuo 0–180° 0.01° One droplet of 3 ll of distill water.
tester Precision Instruments
Co. Ltd, China
Abrasive belt YA 374 Tianjing Gaoli Grinding Granularity: P40, P60, P80, — Twill substrate. Brown fused alumina
Co. Ltd, China P100, P120, P150 oxide grit and open coated.

Note: The granularity of sanding belt is following the standard of ISO 6344-2-1998.

Table 2
Details of workpiece materials.

Materials Air-dry density (g/cm3) Hardness (HD) Countryof Origin Size (mm) (L  W  T)
Korean pine (Pinus koraiensis Sieb. Et Zucc.) 0.47 42.4 China 150  100  30
JATOBA (Hymenaea cunrbaril L.) 0.96 83.7 Brazil 150  100  30
Medium density fibreboard (MDF) 0.66 62.0 China 150  100  24

where A represents the measurement area, x and y refer to the the wood. One of the stylus testing traces was located near center
boundary of the measurement area, z(x,y) means the height value line of the sanded surface, and the other two were along the bilat-
of the sanded surface. eral sides of the sanded surface. Then an average value was calcu-
The stylus moved across the wood texture during the 2D con- lated to determine Ra.
tacted roughness measurement, and the sampling length (lr) was At present, 3D profilometer has features of low speed, high
25 mm. Prior to the measurement, the sanding dust was blown accuracy and small single scanning area (like KEYENCE VR series,
off from the sanded surface by an air gun. Then, the stylus moved with a single scanning area of 5 cm2), which makes it inappropriate
across the feed direction to obtain the 2D average surface rough- for large format panels. Moreover, the required surface quality and
ness Ra. Under the same testing condition, the 2D contacted rough- the measurement accuracy of surface roughness is not that high,
ness measurement repeated three times, covering different parts of compared with metal grinding. In this sense, a more efficient

Fig. 2. 3D morphology and 2D cross-sectional profiles of MDF sanded surface.


4 B. Luo et al. / Measurement 165 (2020) 108133

points under real reference plane. In this way, the calibrated sur-
face profile can reflect the real surface roughness with low thick-
ness error.
The least square median plane (LSMP) can be calculated by a
new scanned surface profile data according the Eq. (3) [17].
X
m X
n
    2
e¼ z xi ; y j  f xi ; y j ð3Þ
i¼1 j¼1

where (i, j) means the coordinates of the planar projection of the


surface profile, z(x, y) means the height from the scanned surface
profile, f(x, y) means the height from the calibrated surface profile,
and e means the sum of squares of height difference value between
LSMP and the calibrated surface profile.
In this study, the height of LSMP is a constant because it’s a
Fig. 3. The calibration of measuring reference. plane parallel to scanned reference plane, which is recorded as z.
The z(xi, yj) is recorded as z1, z2, z3, . . .. . ., zn. Thus, e can be calcu-
lated as Eq. (2).

e ¼ ðz1  zÞ2 þ ðz2  zÞ2 þ ðz3  zÞ2 þ    þ ðzn  zÞ2


¼ nz2  2zðz1 þ z2 þ z3 þ    þ zn Þ
 
þ z1 2 þ z2 2 þ z3 2 þ    þ zn 2 ð4Þ
The minimum e, referring to the average height values of the all
sampling points in the calibrated surface profile, is as follows:
z1 þ z2 þ z3 þ    þ zn
lim e ¼ ð5Þ
n!1 n
The average diameters of vessels, cavities of tracheid and resin
canals in wood are all above 15 lm [18]. So, achieving a very small
value of Sa makes no sense exactly. According to preliminary tests,
3000 height points per square centimeter was selected to calculate
Fig. 4. Determination of the real reference plane. Sa and that achieved a minimum biological roughness of 10 lm for
wood. In this study, 3000 height points was selected on the surface
method was proposed to be feasible for the Sa measurement of profile to calculate a precise Sa as below:
large format panels. The point-cloud data of the sanded surface,
1
which was used to calculate the 3D mean surface roughness Sa, Sa ¼ ðjz1  ej þ jz2  ej þ    jz3000  ejÞ ð6Þ
was collected by the 3D profile scanner (scanning area: 400 cm2). 3000
On the one hand, abrasive belt sanding presents an elastic
machining characteristic due to the elasticity of rubber 3. Results and discussions
contacted-roll, polymer glue and the substrate [16]. On the other
hand, wood materials show a complex biological structure, which 3.1. Effects of G on Ra and Sa
leads to diverse forms of cells damage and material removal. Con-
sequently, the thickness of removed material is not always uniform According to Fig. 5, the standard deviations of Ra were greater
in sanding processes. The tilted sanded surface needs to be cali- than those of Sa, especially for Korean pine and JATOBA. It indicated
brated by the algorithm of the 3D optical profilometer before cal- that Sa had better reliability for wood materials. Using 2D contact
culating the mean surface roughness Sa, as shown in Fig. 2.
Similarly, height data (Z-axis) of the workpiece surface is an
absolute height based on the scanned reference plane in the pro-
cess of 3D profile scanning. Generally, the scanned reference plane
is horizontal and includes the lowest point of the sanded surface,
as shown in Fig. 3. In this case, the Sa calculated with the collected
data will inevitably make errors. Therefore, the reference plane
must be determined at first to reflect the real surface profile and
assure the accuracy of height data from sampling points.
The real reference plane should be in accord with the gradient
of real sanded surface. As shown in Fig. 4, three areas were chosen
to calculate the real reference plane. Considering the existence of
biological unevenness, the lowest points on area 1, 2 and 3 were
selected to establish the real reference plane for wood materials.
Then the calibrated surface profile was composed of points with
coordinates (Z-axis) determined by the difference between scan-
ning height and the height of real reference plane. Some of the
sampling points may locate under the real reference plane, which
might be caused by some inborn defects. Compared with the center
of the surface, it is easier to remove the material during the sand-
ing process. Therefore, the coordinates (Z-axis) were set as 0 for the Fig. 5. Ra and Sa after sanding with different granularity (G).
B. Luo et al. / Measurement 165 (2020) 108133 5

measurement, the stylus may destroy the original surface structure


and drop into the anatomical cavities as it moves forward on the
measuring surface of wood and MDF workpieces. Moreover, the
sanded surface was prone to be compressed and flattened by the
stylus especially for soft wood materials, leading to lower Ra. Thus,
nearly all of Sa were greater than Ra. Compared with wood, MDF
shows more homogeneity, hence the effect of its microstructure
on surface roughness is weaker. But biological microstructure of
wood has a significant impact on Sa. And Korean pine, a soft wood
with low density and distinct early-wood and late-wood variation,
showed greater standard deviations of Sa than JATOBA and MDF
did.
The Sa curve slowed down and varied gently when G was above
P80 because the biological and the structural roughness made it
difficult to be further smoothed by the sanding process. The lowest
surface roughness depends on the anatomical structure of the
wood and MDF. Compared with the case of P80, Sa of Korean pine,
JATOBA and MDF were increased when G was P100, especially for
JATOBA with an overwhelming rise of Sa. This conforms with the
research result of Sun and Li [19] that the bigger granularity of Fig. 6. Effects of granularity (G) on equilibrium contact angle (h).

abrasive belt doesn’t cause a definite lower surface roughness.


According to previous work, there are four stages in wood sanding relationship between Ra, Sa and h. The basic principle of this theory
process, which are rubbing, compressing, ploughing and cutting is to evaluate whether the link of curves is close according to the
[20]. The cutting edges become short when G increases. To remove similarity degree of their shapes [28]. Conventional data analysis,
deformed material, grits with discernable cutting edges must pen- like regression, variance or principal component analysis, requires
etrate into wood surface at a certain depth. When the cutting a large amount of data and the quantitative results may not
depth is not big enough, elastic and plastic material deformation conform with the results of qualitative analysis. In this study, the
take a larger proportion. At rubbing stage, the elastic material evaluation of relationships between Ra, Sa and h was based on
deformation will recover after the grits pass. And plastic pile-up, the Grey Relevancy Degree (GRD), which makes up the defects of
especially for high-density wood species, predominantly occurs conventional analysis methods like Spearman’s correlation
at ploughing stage [21]. This is the main reason for the sudden coefficients which requires smooth and derivable data, as well as
increase of Sa when G increased form P80 to P100. When G was large sample [29–31]. Besides, it can be analyzed according to
above P100, most of the grits were at compressing stage accompa- the changing trend of data. Specific calculation procedures of
nying with rubbing, the anatomical cavities were compressed and GRD are shown from Eq. (7) to Eq. (15).
presented surface densification, which almost leads to a continu-
ous decrease of Sa. (1) Formation of the system behavior sequence.

3.2. Effects of G on h X i ¼ ðxi ð1Þ; xi ð2Þ;    ; xi ðnÞÞ ð7Þ


 
As shown in Fig. 6, Korean pine and JATOBA are hydrophilic X j ¼ xj ð1Þ; xj ð2Þ;    ; xj ðnÞ ð8Þ
materials (h < 90°) and MDF is a hydrophobic material (h > 90°),
where Xi(n) means the parameter array, and Xj(n) means the com-
which is in accord with previous study [22]. Looking at Fig. 6, the
paring array.
variation trends in Sa and h are similar for sanding Korean pine
and JATOBA. Water drop can penetrate the interior of wood through
(2) Calculation of initial phase for each sequence.
the gaps among the fibers, the pits or the vessels [23]. During the
sanding process, original surface is removed and new surface is
X 0 i ¼ X i =xi ð1Þ ¼ ðxi 0 ð1Þ; xi 0 ð2Þ;    ; xi 0 ðnÞÞ ð9Þ
generated with some gaps and channels being opened [24], which
promotes the permeability. Meanwhile, the sanding dust fills into  
the gaps and channels, resulting in decreasing permeability and X 0 j ¼ X j =xj ð1Þ ¼ xj 0 ð1Þ; xj 0 ð2Þ;    ; xj 0 ðnÞ ð10Þ
increasing h. When G increases, the shorter cutting edges and smal- where X0i (n)
means the initial phase of parameter array, and X0j (n)
ler chip spaces increase the sanding temperature [25]. The higher
means the initial phase of comparing array.
temperature makes the water and the chemical group of the wood
surface more active, which can also account for the improved per-
(3) Evaluation of sequence of difference (Dij ðkÞ).
meability of wood surface [26]. The penetrated process is a syn-
thetic action, and the influencing weight of each effect  
Dij ðkÞ ¼ x0 i ðnÞ  x0 j ðnÞ ¼ ðDð1Þ; Dð2Þ    ; DðnÞÞ ð11Þ
mentioned above varies in different wood species. While for
MDF, the adhesive (chemical impact) has a dominated impact on h.

3.3. Grey Relevancy analysis (GRA) for Ra, Sa and h (4) Calculation of maximum and minimum difference of Dij ðkÞ.

The chemical composition of wood is not uniform, so the Wenzel M ¼ maxi maxk Dij ðkÞ; ð12Þ
and Cassie theory about h can’t be applied in this study. However,
in some cases, the relationship between Sa and h is obvious. So m ¼ mini mink Dij ðkÞ; ð13Þ
regardless of the complex influencing process on h, Grey Relevancy
Analysis (GRA), an important part of the Grey System Theory where M means the maximum difference of Dij ðkÞ, and m means the
founded by Deng Julong [27], was applied to analyze the minimum of Dij ðkÞ.
6 B. Luo et al. / Measurement 165 (2020) 108133

Table 3 (3) According to the results of GRD, MDF showed a closest rela-
The results of GRD analysis. tionship between Ra and Sa. The correlations between Sa and
GRD (c) Korean pine JATOBA MDF h, Ra and h for Korean pine were the highest.
Sa and h 0.815 0.742 0.499
Ra and h 0.953 0.851 0.588
Ra and Sa 0.887 0.911 0.987
CRediT authorship contribution statement

Bin Luo: Conceptualization, Methodology, Writing - original


draft. Jian Zhang: Data curation, Software, Writing - review & edit-
(5) Calculation of correlation coefficient (cji ).
ing. Xue Bao: Visualization, Investigation. Hongguang Liu: Soft-
ware, Validation. Li Li: Supervision, Funding acquisition.
m þ nM
cji ðkÞ ¼ ; n 2 ð0; 1Þ; k ¼ 1; 2;    ð14Þ
Dij ðkÞ þ nM
Declaration of Competing Interest
where n is the distinguishing coefficient of the grey relevancy which
is usually set as 0.5. In this study, n = 0.5 was selected.
The authors declare that they have no known competing finan-
cial interests or personal relationships that could have appeared
(6) Calculation of GRD (c)
to influence the work reported in this paper.

1X n
c¼ c ðkÞ; i ¼ 1; 2;    ; n ð15Þ
n k¼1 ij Acknowledgements

Generally, the relationship between two factors is not corre- This research was supported by the National Key Research and
lated when the GRD is smaller than 0.5. According to the results Development Program of China (No. 2018YFD0600304) and China
shown in Table 3, the GRD between Sa and h of Korean pine and Scholarship Council (NO. 201806515038). The authors are grateful
JATOBA are much higher than that of MDF. This is partly because for the support of MOE Key Laboratory of Wooden Material Science
most of the adhesive polymers applied in MDF presents and Application, Beijing Key Laboratory of Wood Science and Engi-
hydrophobicity. Compared with the effect of surface roughness neering at Beijing Forestry University.
on h, the force between water molecules and chemical groups
play a more decisive role [32]. The different surface structures
of wood affect the surface tension of water (capillary phe- References
nomenon) [33,34]. As a kind of diffuse-porous wood, there are
[1] C. Csiha, J. Kirsch, Vessel filtration-a method for analyzing wood surface
many vessels and other inborn cavities in JATOBA which tend to roughness of large porous species, Drevarsky Vyskum 45 (1) (2000) 13–22.
be exposed on wood surface after sanding. These cavities and [2] P.D. Evans, I. Cullis, Effect of sanding and coating with UV-cured finishes on the
fractured cells caused by sanding make the permeation of water surface roughness, dimensional stability and fire resistance of oriented
strandboard, Holz Roh Werkst 66 (2008) 191–199, https://doi.org/10.1007/
easier. While for Korean pine, the impact of biological characteris- s00107-008-0238-4.
tics on h is not significant, thus, a higher GRD between Sa and h [3] M. Akgül, S. Korkut, O. Camlibel, Z. Candan, T. Akbulut, Wettability and surface
presented. The results showed a closer relationship between Ra roughness characteristics of medium density fiberboard panels from
Rhododendron (Rhododendron ponticum) biomass, Maderas Ciencia Y
and h, however, the relatively high dispersion degree of Ra was Tecnología 14 (2) (2012) 185–193, https://doi.org/10.4067/S0718-
depicted in Fig. 5. So, the GRD of Ra and h cannot accurately 221X2012000200006.
describe and reflect the real relations. As mentioned above, Kor- [4] M.R.S. Amorim, P.G. Ribeiro, S.A. Martins, M.R. Souza, Surface wettability and
roughness of 11 Amazonian tropical hardwoods, Floresta Ambient 20 (1)
ean pine is a soft wood with lower density, the noticeable biolog- (2013) 99–109, https://doi.org/10.4322/floram.2012.069.
ical unevenness of which makes the lowest GRD between Ra and [5] M. Arnold, Effects of planning and sanding on wood surface properties and
Sa. Compared with the two materials, MDF presents a more coating performance, Surf. Coat. Int. 94 (5) (2011) 170–176.
[6] M. Danish, T.L. Ginta, K. Habib, D. Carou, A.M.A. Rani, B.B. Saha, Thermal
homogeneous composition and is less influenced by the biological analysis during turning of AZ31 magnesium alloy under dry and cryogenic
unevenness. That accounts for the fact that the relevancy degree conditions, Int. J. Adv. Manuf. Technol. 91 (2017) 2855–2868, https://doi.org/
of Ra and Sa is the highest for MDF. 10.1007/s00170-016-9893-5.
[7] B. Hendarto, E. Shayan, B. Ozarska, R. Carr, Analysis of roughness of a sanded
wood surface, Int. J. Adv. Manuf. Technol. 28 (2006) 775–780, https://doi.org/
4. Conclusion 10.1007/s00170-004-2414-y.
[8] C.M. Yousuff, M. Danish, E.T.W. Ho, I.H. Kamal Basha, N.H.B. Hamid, Study on
the optimum cutting parameters of an aluminum mold for effective bonding
Wood sanding as a widely used machining process can provide strength of a PDMS microfluidic device, Micromachines (Basel) 8 (2017),
better surface quality, which exerts great influence on surface coat- https://doi.org/10.3390/mi8080258.
ing performance. To promote the previous methodology of surface [9] A.C. Murray, S. Woodward, In vitro interactions between bacteria isolated from
Sitka spruce stumps and Heterobasidion annosum, Forest Pathol. 33 (2003)
roughness measurement, a new calibration method was proposed. 53–67, https://doi.org/10.1046/j.1439-0329.2003.00307.x.
The method can be applied to calculate 3D surface roughness Sa for [10] E. Csanády, E. Magoss, Surface roughness, in: E. Csanády, E. Magoss (Eds.),
larger format panels, which successfully eliminates the influence of Mechanics of Wood Machining, Springer, Berlin Heidelberg, Berlin, Heidelberg,
2013, pp. 167–194.
biological surface and sanded thickness error. The relationships
[11] B.Q. Li, Study on rule of 3D roughness average and root mean square, Tool Eng.
between Ra, Sa and h for wood materials were experimentally 42 (09) (2008) 107–110.
investigated. And the main conclusions are as follows: [12] W.G. Li, Z.K. Zhang, K. Zhang, Comparison between 2D contour and 3D
topography for testing surface roughness on sawn lumber, China Wood Ind. 28
(5) (2014) 13–15.
(1) The standard deviations of Sa were remarkably lower than [13] L. Gurau, N. Ayrilmis, Effect of raw material composition of wood plastic
those of Ra for the three kinds of wood materials with differ- composites on surface roughness parameters evaluated with a robust filtering
ent G. Compared with MDF, the sanded surface roughness of method, J. Thermoplast. Compos. Mater. (2018).
[14] X. Bao, J.H. Ying, F. Cheng, J. Zhang, B. Luo, L. Li, H.G. Liu, Research on neural
Korean pine and JATOBA is more greatly influenced by biolog- network model of surface roughness in belt sanding process for Pinus
ical structure like anatomical cavities. koraiensis, Measurement 115 (2018) 11–18.
(2) For Korean pine and JATOBA, h generally decreased as G [15] B. Luo, L. Li, H.G. Liu, M.J. Xu, F.R. Xing, Analysis of sanding parameters, sanding
force, normal force, power consumption, and surface roughness in sanding
increased. For MDF, h kept almost a constant as G increased. wood-based panels, BioResources 9 (4) (2014) 7494–7503.
B. Luo et al. / Measurement 165 (2020) 108133 7

[16] B. Luo, L. Li, M.J. Xu, H.G. Liu, F.R. Xing, Analysis of static coefficient between [26] R.R. Khasanshin, R.R. Safin, E.Y. Razumov, High temperature treatment of birch
workpiece and rubber belt in sanding wood-based panel, Bioresources 94 (4) plywood in the sparse environment for the creation of a waterproof
(2015) 7372–7381. construction veneer, Procedia Eng. 150 (2016) 1541–1546.
[17] P.J. Rousseeuw, Least median of squares regression, J. Am. Stat. Assoc. 79 [27] J.L. Deng, Introduction to grey system theory, J. Grey Syst. 1 (1) (1989) 1–24.
(1984) 871, https://doi.org/10.2307/2288718. [28] X.M. Leng, W. Wang, H.H. Gao, Y.Y. Gong, Materiel development program
[18] Z. Wang, M.X, Research progress of multi-scale pore structure and technical risk evaluation method based on grey relevancy degree, in: The
characterization methods of wood, Scientla Sylvae Sinicae 50 (10) (2014) Proceedings of 2011 9th International Conference on Reliability,
123–133. Maintainability and Safety, IEEE, 2011 – 2011, doi: 10.1109/
[19] Y. Sun, L. Li, Effect on bonding strength of surface roughness of sanding wood, ICRMS.2011.5979334.
Wood Process. Mach. 03 (2010) 45–47, https://doi.org/10.13594/j.cnki. [29] J.W.K. Chan, T.K.L. Tong, Multi-criteria material selections and end-of-life
mcjgjx.2010.03.008. product strategy: Grey relational analysis approach, Mater. Des. 28 (5) (2007)
[20] J. Zhang, J. Ying, F. Cheng, H. Liu, B. Luo, L. Li, Investigating the sanding process 1539–1546.
of medium-density fiberboard and Korean pine for material removal and [30] Y. Kuo, T. Yang, G.W. Huang, The use of grey relational analysis in solving
surface creation, Coatings 8 (2018) 416, https://doi.org/ multiple attribute decision-making problems, Comput. Ind. Eng. 55 (1) (2008)
10.3390/coatings8120416. 80–93.
[21] J. Zhang, B. Luo, L. Li, H. Liu, Cutting characteristics for sugar maple using single [31] J. Hauke, T. Kossowski, Comparison of values of Pearson’s and Spearman’s
grit with spherical cone and triangular pyramid geometries, Materials (Basel) correlation coefficients on the same sets of data, Quaestiones Geographicae 30
12 (2019), https://doi.org/10.3390/ma12213621. (2011) 87–93, https://doi.org/10.2478/v10117-011-0021-1.
[22] Q.D. Xie, G.Q. Fan, N. Zhao, X.L. Guo, J. Xu, J.Y. Dong, L.Y. Zhang, Y.X. Zhang, [32] N.M. Stark, L.M. Matuana, Characterization of weathered wood–plastic
Facile creation of a bionic supe-hydrophobic block copolymer surface, Adv. composite surfaces using FTIR spectroscopy, contact angle, and XPS, Polym.
Mater. 16 (20) (2004) 1830–1833. Degrad. Stab. 92 (10) (2007) 1883–1890.
[23] X.Y. Wei, D.H. Tao, D.X.L. Sheng, M.H. Zhang, Modification of poplar wood [33] J. Cool, R. Hernande, Improving the sanding process of black spruce wood for
using polyhexahydrotriazine and its effect on hygroscopicity, J. Wood Chem. surface quality and water-based coating adhesion, Forest Prod. J. 61 (5) (2011)
Technol. 38 (3) (2018) 214–223. 372–380.
[24] B. Luo, S.Y. Jin, F. Cheng, J.H. Ying, X. Bao, H.G. Liu, L. Li, Study on sanding force, [34] H.Y. Jiang, Y.X. Zhang, A.G. Liang, H.Y. Qi, Influencing factors and prediction
power consumption and quality of calibrating sanding recombinant bamboo, model of material surface wettability, Surf. Technol. 47 (1) (2018) 60–65,
Wood Res. 62 (6) (2017) 983–994. https://doi.org/10.16490/j.cnki.issn.1001-3660.2018.01.010.
[25] B. Luo, L. Li, H.G. Liu, M.Z. Wang, M.J. Xu, F.R. Xing, Effects of sanding
parameters on sanding force and normal force in sanding wood-based panels,
Holzforschung 69 (2) (2014) 241–245.

You might also like