Assessment of Wood Surface Quality Obtained During High Speed Milling by Use of Non-Contact Method

This paper presents the method of surface quality evaluation at high-speed milling of wood. The effects of specifi c parameters of machining on the surface quality are described and results are presented of experiments carried out on selected tree species, where the effects of specifi c parameters on the surface quality were noted. The surface quality is evaluated by a non-contact method using a confocal sensor (CLA), which processes the results of the roughness measurement, Talysurf CLI 1000. It can be concluded that this method could actually be used for the evaluation of the surface quality.


INTRODUCTION 1. UVOD
Milling is a technological process used for making the surface of certain shape, dimensions and quality cutting material particles (chips) or wood fi bre bundles (cutting operation).Fig. 1 shows the kinematics of cutting a chip in standard milling (conventional milling).In practice, however, the actual chip cross-section can differ from the nominal cross-section due to blunting, inaccurate spindle run, deviations of cutting edges from the cutting circle, irregularities of the feeding device cycle and, particularly, due to splitting off and heterogeneity of the workpiece material.

DRVNA INDUSTRIJA 62 (2) 105-113 (2011)
The tool edge trajectory creates a cycloid curve, cutting speed being, however, very high at higher diameters of milling tools in proportion to the feed rate.Thus, at the section of the edge cut, it can be supposed with high probability that its cutting track will form a circle.
The problems of milling and machining surface quality have already been thoroughly investigated.The paper is based on fi ndings of authors and their publications (Lisičan, 1996;Prokeš, 1965) and some other foreign and Czech authors.
The process of chip production shows a fundamental effect on the workpiece surface quality.So, it can be quite well described by relationships ( 1) and ( 2) resulting from Fig. 2.
Problems of machining are dealt with by Koch (1964).Problems of high-speed machining have been discussed recently.Lichtág et al. (2003) lay out in their research report an opinion of the general division of machining to normal, high-speed and extreme (Fig. 3) specifying a limit between normal and high-speed machining according to the relation (3).For the fi eld of high-speed machining, they mention the rate v c = 70 -130 m/s.
According to Prokeš (1965), decisive importance for the quality of a machined surface is shown by the size of feed per tooth (see equation 4).Problems of measuring the surface roughness using a contact method are dealt with by Škaljić et al. (2009).The authors give results of measuring surface quality with samples of four wood species.Svoboda et al. (2009) deal with the application of a non-contact method and lay out that the method can be used for the evaluation of roughness and waviness of the wood surface.The paper of Kowalski et al. (2008), where modelling the surface geometrical structure originating in the process of wood milling is discussed at a theoretical level, is rather inspiring.Sandak (2008) dealt with the problem similarly.After a theoretical analysis, he used the non-contact method of measurement to evaluate the surface quality.Using the laser sensor for surface inspection is discussed in a paper by Barcík et al. (2009), where differences determined between the surface quality of juvenile and mature wood are presented.A laser sensor of the surface profi le was also used in a paper by Siklienka et al. (2008) at surface roughness quantifi cation in cutting by horizontal band saw.
The theoretical mean chip thickness (h m ) is the function of feed per tooth (f z ).For its determination the following relations can be used: where: h m -mean chip thickness / srednja debljina strugotine, mm f z -feed per milling head edge / zahvat po oštrici alata, mm ϕ m -average angle / srednji kut zahvata, °R -cutting edge radius of a milling head / polumjer kružnice koju opisuju rezne oštrice, mm e -cutting depth / dubina blanjanja, mm Cutting speed: Feed per tooth: where: v c -cutting speed / brzina rezanja, m/s n -tool revolutions / frekvencija vrtnje alata, min -1 D -cutting edge diameter of a milling head / promjer kružnice koju opisuju rezne oštrice, mm v f -feed speed / posmična brzina, m/min z -number of edges of a milling head / broj oštrica Considering the fact that the milling cutter edge creates a circle, it is possible to determine the theoretical depth of the surface roughness (y) of milled surfaces up to hundredths of mm.Then, a known relationship can be accepted to determine the wave theoretical depth y: where: y -depth of the surface kinematical roughness / dubina površinske kinematičke hrapavosti, mm The relation (5) applies to absolutely accurate setting of the milling head knives.In practice, it is often rather diffi cult to set up milling heads into a cutting edge with absolute accuracy.Thus, a situation shown in Fig. 4 can occur, if we know that R = D/2.
Characteristic factors determining the surface quality of a milled workpiece are as follows: tree species, volume weight, mechanical properties, moisture content, and orientation of wood fi bres with respect to the direction of the cutting edge movement.For the purpose of this paper it is suffi cient when these factors are expressed by the tree species and its moisture content.The optimum moisture of wood is 8 ± 2%.
Kinematic roughness is the result of mutual motion of a milling head and a workpiece.The rotational motion of a milling cutter and usually the straight-line shift of a workpiece cause typical surface roughnesscycloid waves.The waves are characterized by their depth and the distance of their tops.This type of regular inequality is termed as waviness.Roughness arising from the wood structure results from the non-homogenous structure of wood and different behaviour of early and late wood during machining.It is assessed as irregular waviness.Picking-up the wood fi bres is also related to the wood structure and behaviour.Torn bundles of wood fi bres and microscopic and macroscopic cavities on the wood surface are characterized by their roughness (smoothness).The occurrence of this roughness is mostly irregular on the tooled surface.

MATERIJAL I METODE
Samples of Norway spruce (Picea abies) and beech (Fagus sylvatica) were milled on an experimental stand.
Parameters of milling on the experimental stand were selected in two regimes: -sampling time: 10 ms -measurement time: 15 s.For the surface evaluation, an optical method was used based on the surface topography by a non-contact confocal sensor CLA (Chromatic Length Aberration), which is part of a system for the surface texture evaluation (Talysurf CLI 1000).This system can carry out evaluation of the 2D (in total 117 parameters) and 3D (in total 40 parameters) surface structure.
Laser beam (intensive laser point source of light) is focused on a diaphragm, and it goes through the object lens up to a sample, where the diaphragm image is focused into a point whose diameter corresponds to the diffraction limit (differential limit).
Also, light refl ected or scattered on the sample goes by return through the same object lens or fl uorescence occurs.
Secondary light going back passes again through a diaphragm whose point image is localized by the beam divider through a photomultiplier (PMT), where the second confocal point diaphragm is placed, which fi lters light coming from an area outside the microscope focal plane.
An image of the whole focused plane can be obtained through stepwise scanning using some of the following procedures: wobbling a laser beam, transverse feed of a sample before object lens, the object lens feed over the sample.
Methods of the surface quality evaluation by Talysurf CLI 1000: 1. Levelling the surface -levelling the surface according to the selected area 2. Zoom -the selection of an area from the measured surface for further adjustments and analyses 3. Form removal -used for the geometrical shape separation in measuring the actual surfaces 4. Thresholding -including the corresponding spectrum of data into analysis 5. Obtaining the basic area 6. Filtration -obtaining the area of waviness and roughness 7. Depiction of the 3D area 8. Determination of 3D parameters of the area surface structure The surface of milled samples was evaluated using the following parameters: S parameter -the surface amplitude parameter R parameter -a parameter calculated from the roughness profi le W parameter -a parameter calculated from the waviness profi le Gaussian fi lters of 0.8 and 2.5 were used for roughness and waviness parameters.

REZULTATI I RASPRAVA
The results of the surface quality measurement were processed using the Talymap program in Talysurf system (see Fig. 8).An example is given of the 3D image of a beech sample, data on roughness and waviness and a selected section of the 2D basic profi le, waviness profi le and roughness profi le.
Through the processing of data, waviness parameters can be obtained that can be compared with theoretically calculated values.Fig. 9 illustrates that the actual length of waves is higher than the theoretical value.Thus, an assumption is proved that all knives are not in gear.
The measured parameters of the surface quality were then processed in the form of well-arranged tables.Only changes of the waviness parameters Wz (Fig. 10) were processed in diagrams.Data on roughness were processed in the same way.
By parameters of waviness it is possible to monitor kinematic unevenness, i.e. the process of high-speed milling, changes during machining, machine and tool instability, changes in the operation of a cutting tool.By roughness parameter it is possible to monitor the effects of particular species on the surface quality or chipped (torn) grain.
The mean arithmetic deviation of parameters of waviness Wa ranges from 4.27 μm at 9000 min -1 to Annual ring -summer wood / god -kasno drvo

Wave top / vrh vala
The actual distance of waves -2D profi le of waviness (mm is the unit of axis x and µm of axis y)

Stvarna udaljenost valova -2D profi l valovitosti (skala na osi x iskazana je u mm, a na osi y u µm)
The theoretical distance of waves -2D profi le of waviness (mm is the unit of axis x and µm of axis y) Teorijska udaljenost valova -2D profi l valovitosti (skala na osi x iskazana je u mm, a na osi y u µm) Figure 9 The profi le of waviness of spruce at 9000 min -1 , feed 20 m/min Slika 9. Profi l valovitosti smrekovih uzoraka pri frekvenciji vrtnje alata 9 000 min -1 i posmičnoj brzini 20 m/min Legend / Legenda: S a , R a , W a -the mean arithmetical deviation of a profi le / srednje aritmetičko odstupanje profi la, μm; S q , R q , W q -the mean quadratic deviation of a profi le / srednje kvadratno odstupanje profi la, μm; S t , R t , W t -the total profi le height / ukupna visina profi la, μm; S z , R z , W z -the largest height of a profi le / najveća visina profi la, μm; RS m , WS m -the mean width of a profi le / srednja širina profi la, mm 2.831 μm at 10000 min -1 in spruce and from 3.50 μm at 9000 min -1 to 4.56 μm at 10000 min -1 in beech.The mean quadratic deviation of the waviness profi le Wq ranges from 4.76 to 3.20 μm in spruce and from 3.93 to 5.18 μm in beech.With the decreasing feed per tooth these values diminish and, vice versa, with the increasing feed per tooth these values increase.The average arithmetic deviation of parameters of waviness W a is the most common parameter in industry.It refers to a stable parameter not affected by accidental changes.The average quadratic deviation of the waviness profi le W q shows a foregrounding effect on particular or accidental projections and depressions.Therefore, it is a parameter that enables the differentiation between very fi ne surface and similar surface with traces and defects that are not typical.In our case, it rather determines extreme values of waviness.This parameter corresponds to the theoretical calculation of the waviness depth.However, theoretical values of waviness are about 2 μm smaller.The total height of the waviness profi le W t is the maximum distance between the profi le projection and hollow, i.e. their extreme limits.Particular measurements failed to show the dependence of W t on parameters of high-speed milling.The measured average values W t show extreme limits of the profi le; however, these are not reliable.The largest profi le height W z ranges from 8.61 μm at 9000 min -1 to 7.46 μm at 10000 min -1 in spruce and from 9.41 μm at 9000 min -1 to 7.62 μm at 10 000 min -1 in beech.The average width of the waviness profi le WS m determines the mean width of unevenness.It refers to a very important parameter because its values are often numerically equal to feed per tooth.
Roughness parameters are particularly affected by wood type/species and properties of the machined material.Of course, RS m parameters can partly determine the distance where torn grain occurs.The average quadratic deviation of the roughness profi le R q differentiates between a typical surface and a surface with non-typical changes, which is theoretically characterized by torn grain caused by specifi c knives.The average width of elements of the roughness profi le RS m is numerically equal to the distance of surface waviness tops -the cut of six knives.The measured values R q range from 3.49 μm at 9000 min -1 to 2.38 μm at 10 000 min -1 in spruce and from 2.38 μm at 9000 min -1 to 2.61 μm at 10 000 min -1 in beech.Based on the comparison between the results of measurements made by two different Gaussian fi lters of 0.8 and 2.5 mm, with the use of the 0.8 mm fi lter, roughness is suppressed in favour of waviness and, on the contrary, the 2.5 mm fi lter released the higher wave length of waviness parameters at the expense of the roughness parameter.It can be concluded that the 2.5 fi lter is more suitable for assessing waviness than the 0.8 fi lter.

ZAKLJUČCI
The measurement of surface quality was carried out by 3D spatial characteristics that enabled processing of a large number of data.These data then described the assessed profi le of the monitored surface.More samples should be measured so as to prove or disprove their depen dence.Parameters of high-speed milling were selected with respect to maximum revolutions of ordinary machines.The feed of cut material was chosen in accor dance with the requirement to achieve the best possible quality surface.The average quadratic deviation of the waviness profi le W q corresponds to the theoretical calculation of the waviness depth that is, however, about 2 μm higher, which can be attributed to the imperfect torn grain that causes "hairy surface".Differences in the W q pa ra meter among particular species are negligible and the growth of W q with the feed speed has been confi rmed.
The average width of the waviness profi le elements WS m corresponds to the theoretical calculation of the actual distance of the wave tops (one knife cut) particularly for beech; in spruce, the value of WS m is greater by 0.2 mm, which is affected by the microscopic structure of wood.Samples of spruce showed clearly how differences among early and late wood affected waviness.The growth of WS m with the increase of feed speed has been proved.With parameters of high-speed milling, 9000 and 10 000 min -1 , tool edges set to 0.001 mm and feed of 18 m/min, it was not possible to fully differentiate the parameter of waviness WS m because it partly interfered with the roughness parameter RS m .It follows that this surface corresponds to the quality of the ground surface.However, on the milled surface, waviness is partly preserved.
With the other samples, it was also possible to observe waviness caused by the local increase of the wear of knives.With the selected parameters, the fi nal surface quality is particularly affected by the condition and accuracy of knife setting.
Generally, it is possible to determine the following characteristics of the method: -Parameters used in mechanical engineering can also be partly used for wooden materials.-It is possible to observe changes during machining.
-It is possible to observe changes in the tool geometry.-The method enables processing of a large number of data.

Figure 10
Figure 10 Waviness parameters W z and WS m in a function of feed speed while milling spruce (a) and beech (b) samples with the head rotational speed of 9 000 min -1 Slika 10.Parametri valovitosti W z and WS m u ovisnosti o posmičnoj brzini pri blanjanju smrekovih uzoraka (a) i bukovih uzoraka (b) uz frekvenciju vrtnje alata 9 000 min -1