Comparison of Surface Quality and Tool-Life of Glulam Window Elements after Planing

The quality of the surface of wooden elements, that have been planed, has a crucial importance in the whole production process, since the obtained effects affect the quality of wooden surface after fi nishing (painting). The occurrence of defects is usually the reason for qualifying a workpiece as scrap or for requiring additional work. This paper presents the selected results of research of the effect of the cutting tool wear on the surface quality of elements after planing. Research experiments were conducted on the SCM Superset Class machine tool. Glulam elements of pine wood (Pinus sylvestris L.) were researched. The raw material samples (semi-fi nished products), 6 m long before planing, had been machined by suppliers also by planing. These workpieces were selected according to the plant requirements, e.g. their moisture content, straightness, and other defects. This paper presents the measuring results of surface roughness and some examples of surface profi les, as well as the dependence of total length of the planed elements on the type of blade material. From an economic point of view, the results showed that the use of solid carbide blades were more cost effective.

Machinability of species of raw and thermally treated wood has been assessed based on quality of planed surfaces.Belleville et al. (2016) evaluated the potential of young eucalypts growing under short rotation planting conditions, as a resource for the manufacture of high quality furniture and furnishings, which were planed with tungsten carbide cutters.In this case, the number of defects on the sample was a measure of quality acceptance.
Aguilera and Muñoz (2011) stated that better surface quality could be obtained for dense Blackwood (Acacia melanoxylon), as ten point height of irregularities Rz (ISO) was equal to 15 m, in comparison to Redwood (Sequoia sempervirens), where Rz (ISO) was 22 m.The experiments were conducted with a tool of ∅124 mm in diameter, equipped with 4 knives made of high speed steel, with their geometry as follows: rake angle 25° and clearance angle 17°; number of revolution 4200 rpm, and feed speed in of range of 4-11 m•min -1 .
Heat treatment of selected Turkish wood species (soft and hard wood) does not affect surface roughness (Budakçı et al. 2013).On the other hand, Gűndűz et al. (2008) reported that surface roughness of modifi ed Camiyanı Black Pine wood (Pinus nigra Arn.subsp.pallasiana var.pallasiana) is lower.Similar fi ndings were stated by Kisselbach (2009) concerned modifi ed timber for window construction.Analogous results of decreased roughness were described by Kvietková et al. (2015) in case of birch wood after thermal treatment.Findings of the experiments by Škaljić et al. (2009) revealed that roughness Ra of thermally modifi ed beech-wood was very close to the same values of steamed beech-wood samples.However, the surface roughness of heat-treated beech processed by milling was slightly higher than that of untreated wood (as measured by Ra, Rq, Rt, Rk, and Rk+Rpk+Rvk) (Ispas et al., 2016).The latter fi ndings were confi rmed by Hacibektasoglu et al. (2017), who revealed that heattreating beech (Fagus sylvatica L.) for 1 h and 2 h had a negligible effect on the processing roughness after planning, measured across the grain by Rk.Moreover, the Rk increased by 15 % for 3 h and 4 h of treatment and with approximately 33 % for treating the beech for 5 h and 6 h.
The researches by Škaljić et al. (2009) have shown that the samples of planed surface of oak-wood (Quercus L.) had the best quality and the samples of fi r wood (Abies alba Mill.) had the highest values of surface roughness (Ra).In both cases the specimens were machined by planing in radial directions with two knives at 6, 12, 18 and 24 m•min -1 feed speed.The cutting depth of 2.0 mm was constant and knife rake angle was 15°.The machining experiments were carried out using a single cutter-block of a Weinig Powermat 400.The cutter-block with a diameter of Ø125 mm rotated at 6000 revolutions per minute (rpm) (Škaljić et al., 2009).Öhman et al. (2015) found that, after planing, the average surface quality of a whole batch of boards made of Scots pine (Pinus sylvestris L.) can be increased by adding water to the surface short before planing.However, a positive impact cannot be guaranteed for single boards (Öhman et al., 2015).
After planing with dull knives, the wood cells were greatly distorted and cell walls were extensively damaged, particularly at and near the glue line.On the other hand, sharp knives caused much less damage to cell walls and the cells had a normal or near-normal appearance (Singh et al., 2002).
Industrial development and international competitiveness impose higher demands on wood industry.New technologies and cutting materials are the key to successful productivity in the manufacturing process.From the industrial point of view, the frequency of tool changes should be minimised and simultaneously the production costs per each piece ought to be reduced (Aguilera et al., 2003).Therefore, the aim of this paper was to evaluate the effect of the cutting tool material (wear) on the surface quality of planed semi-fi nished pine beams for window frames.

Materijali
The samples in the research were wooden beams made of glued laminated Scots pine lamellae (Pinus sylvestris L.), generally used for the production of window frames.Semi-fi nished products were provided in the form of 6-meter-long beams initially machined by suppliers.It should be emphasised that the total number of samples was not limited because beam planing was conducted until there were no defects on the machined surface.The working drawing of the semi-fi nished product includes dimensions, wood species, surface and main features as shown in Fig. 1.

Metode
Tests were carried out on the moulding machine for four-side machining type of Superset mt (the planer moulder, SCM Group SPA, Italy).This test was con-ducted on the fi rst out of three operations in the production process and was associated with shaping of the workpiece (Fig. 2).
The test was carried on until no defects associated with the tool wear (material pulled out, cracks) could be observed on the planed surface of the element.The element was classifi ed as unfi t for further production process based on visual inspection.One of the important indicators of the process evaluation was the number of metres of machined wood meeting the quality requirements, meaning that wood surface of the remaining machined parts were of poor quality.The length of the properly cut beams could be considered as a wear factor (Aguilera et al., 2016).
The occurrence of defects in the material and on the surface can be divided into those dependent and independent on humans.The fi rst group, as previously mentioned, was associated with the occurrence of cracks and knots of the material.In the second group, there are mostly prints and scratches in the material, as well as fi bres torn away from the surface.
The occurrence of any of the above mentioned defects usually disqualifi ed the material for further machining.Some of them could be corrected but it required additional machining (re-work) using sanding machines.

Tool and machine tool 2.3. Alat i stroj
A set of blades necessary to equip the planer moulder Superset mt consisted of 8 knives 80 mm long, 2 knives 130 mm long and 2 knives 150 mm long.Such equipment provided the opportunity to implement all the necessary processes in one machine setting.Table 1 presents the main woodworking information.
In the research, a Tersa ® system planer head was used of 140 mm in diameter (Fig. 3).The knife was made of the standard uncoated solid carbide (HW, K a group of application according to ISO (Tersa 2014)) of micrograin quality, and additionally for comparison blades were made of chrome steel (CR, HR13%Cr (Tersa 2014)).The knives are also characterised by clearance angles (α) in the knife-in-hand system equal to 50° (CR) and 35° (HW) (Tersa 2014).Moreover, the Tersa ® system is characterised by great fl exibility resulting from the system's modular design of identical aluminium washers.
The quality of machined parts was evaluated taking into account various aspects, including dimensions, presence of defects of the raw material, fi nishing with the surface texture.As shown in the working drawing, dimension deviations were approximately ± 0.2 mm (tolerance of 0.4 mm).For this reason, they were measured with a calliper.
Before further machining, semi-fi nished wood elements were subjected to a thorough inspection.Moisture content MC, straightness of workpiece, as well as the presence of material defects, such as cracks and knots, were tested among other things.
The accepted defl ection of an element of 1 m in length was 0.7 mm at the highest point.The density of the wooden element made of, for example, two or more lamellas could not be less than 500 kg/m 3 .An impor- tant parameter under control was the moisture content that was to be MC 12 % ± 2 %.
The surface texture measurements (P -profi le, R -profi le, W -profi le) were carried out with the surface roughness tester Hommelwerk Standard 1000.Each 1-m of the sample, where defects were observed, was cut-off from the 6-m specimen, and then measured.At each measurement point with a visible defect, three measurements were done perpendicularly to the side surface of the workpiece for averaging the results.The results of the measurement of the testing set were recorded in the measuring table, along with the obtained total length of the machined wood.
Additionally, the surface topography measurements of the processed products were made with the 3D Optical Profi ler S Neox (Sensofar 2017) with objective 20× magnifi cation and Z scanning range of 83 micrometres.This system is one of the methods of 3D optical measurements applied in the study of surface topography, which allowed us to scan the samples with a confocal technique.

REZULTATI I RASPRAVA
The results of six tests of total lengths of machined frame beams (together with corresponding number of beams) are presented in Table 2.In order to perform a correct analysis, results obtained from machining with blades made of solid carbide (HW) were compared with results from the same test performed with blades made of chrome steel (CR).It should be emphasized that the productivity of the process in the fi rst case is nearly 3.5 times higher than in the second one (length Median (HW) = 468 m and length Median (CR) = 131 m) (Fig. 4).The analysis took into account the cost of production and especially the difference in the cost of equipping the machine with a given type of knife.The total cost of equipping the heads with knives made of HW was almost twice higher than with knives made of CR.Nevertheless, the average production cost of 1 meter of material in the fi rst case (the head equipped with HW knives) was about 30 % less than the second one (CR knives used).
Measuring the surface texture enabled the comparison of surface structure processed with the new blades and those that showed wearing.In 2D measurements (Gurau and Irle, 2017; Zhong et al., 2013;Sandak and Tanaka, 2003;Sandak et al., 2004;Sandak and Negri, 2005), the evaluation length amounted to 12.5 mm.During the measurements, the following parameters were adopted: evaluation length ln = 12.5 mm, cut off value λc = 2.5 mm, cut off ratio λc/λs = 300, sampling interval 1.5 μm and fi lter type ISO 11562(M1).The stylus end was conical (taper angle of cone: 60°) with a spherical tip (tip radius r tip = 2 μm).These values were adopted since the main objective was to assess the surface condition on maximum possible length, which was limited by maximum measuring length of the used device.Comparing the profi les between the sample manufactured with a set of new knives and the sample obtained after manufacturing n-meters of the product, signifi cant changes can be noticed in the appearance of the primary profi le (the profi le of waviness was added for better visualisation).The profi les of the fi rst samples, i.e. after manufacturing 6 m long beams, were smooth and the knives wear was not observed (Fig. 5).The surface texture parameters for the primary and roughness profi les were: Ra = Measurements of the last defect-free element were made in two areas that had been assessed as the most damaged by visual inspection.Three parallel measurements at 7-mm intervals were done perpendicularly to the side surface of the workpiece (perpendicularly to the feeding direction).The results of the measurement of the testing set, i.e. the last samples, are presented in meas-   18) 79 ( 13) 299 (50) Figure 4 The mean value and value dispersion of total length of machined frame beams vs. selected types of blade material (box plot presents: the mean, median, interquartile range box and data range) Slika 4. Srednja vrijednost i rasipanje podataka o ukupnoj duljini obrađenih lameliranih elemenata s obzirom na vrstu materijala noža (na grafu su prikazane srednje vrijednosti, medijani, interkvartilni raspon i raspon podataka)  3 and 4).
The maximum value of the total profi le height-Rt (68.54 μm) was observed in Sample #2 for the sec-ond area (HW) and it was about 25 % greater than the maximum value for CR blade (Sample #6 second area).In both cases, the range of Rt parameter was signifi cant and amounted to 33.31 -68.54 μm (HW) and 26.74 -54.97 μm (CR).Generally, mean and median were about 30 % greater for HW blade (Fig. 6).Similar situation occurred for maximum height of the profi le (Rz).a) b) Figure 6 The mean value and value dispersion of total height Rt (a) and maximum height Rz (b) of the profi le vs. selected types of blade material (the box plot presents: the mean, median, interquartile range box and data range) Slika 6. Srednja vrijednost i rasipanje podataka: a) o ukupnoj visini Rt i b) o maksimalnoj visini Rz profi la obrađene površine pri obradi dvama različitim noževima (na grafu su prikazane srednje vrijednosti, medijani, interkvartilni raspon i raspon podataka) However, the inter-quartile range was 8.92 μm (HW) and 7.66 μm (CR), respectively, whereas these ranges were twice as high as parameter Rt.This indicates a greater concentration of results around the median, which is related to the method of determining this parameter based on a sampling length and not on an evaluation length as in the case of Rt.
Therefore, both minimum and maximum heights are characterised by particular maximum height of the profi le (Rz) in the range of 20 -45 % for HW blade, and 12-45 % for CR blade, which is smaller than corresponding values of profi le total height (Rt).The maximum difference of 45 % is presented by the example of the profi le of Sample #2 second area (HW) (Fig. 7a).Moreover, the profi le for the minimum value of 12 % observed in Sample #5 second area (CR) is pictured in Figure 7b.Scratches, pull-outs and other damages to the machined wooden material could be distinctly ob-served on profi les, especially for HW blades, (the last element that does not fulfi l quality requirements).
In both cases in question, the means and medians of maximum profi le valley depth (Rv) were on similar level, whereas in case of maximum profi le peak height (Rp), the difference between them was about 7 μm (Fig. 8).The maximum values of Rv and Rp were 20 % and 40 % greater for HW blades, respectively.The range of the quartile box for HW blades was nearly two times higher than for CR blades.Likewise, the dispersion of measured values was doubled.However, the consideration of 8.42 μm outliers leads to the conclusion that ranges of Rv can be comparable.
The Skewness (Rsk) values for most HW samples and all CR samples indicate that the height distribution was deviated upwards.In the case of one of Samples #2 and both Samples #6 (HW), the height distribution was deviated downwards.Meanwhile, in all examined  cases, the tip geometry of peaks and valleys was sharp, Kurtosis (Rku) > 3.
For a more detailed analysis of the damage to the surface, some 3D measurements of the surface condition were made with 3D Optical Profi ler S Neox (f.Sensofar 2017) (Fig. 9 and 10).Moreover, the 3D measurements have given a complete image of the surface condition as a critical element of the entire machining process.The obtained results are in agreement with fi ndings by Zongh et al. (2013).The results of selected 3D surface texture parameters are shown in Table. 5.
In most of the samples, root mean square height (Sq) values were about 30 -40 % higher than arithmetical mean height (Sa) values.Such differences are characteristic for irregular random profi les, where individual maximum and minimum height values, observed in the profi le, have greater effect on Sq than on Sa.This was particularly noticeable in Sample #2 and Sample #6, when these differences increased to about 55 -65 %.The unevenness evaluated using the above height parameters Sq and Sa refl ects general texture of the machined wood having considered scratched areas caused by worn blades.
For Samples #1, #4 and #5, maximum height (Sz) values were in the range of 70 -90 μm.However, in other cases the surface damages were very large with high scratches, values of the Sz parameter even up to 146 μm.The valley's depths were predominant and maximum pit depth (Sv) represented 62 -82 % of maximum height (Sz).In Samples #3 and #6, the distribution among peak heights and valley's depths was more equal and maximum peak height (Sp) represented 52 % and 45 % of the maximum height (Sz).
Presented Skewness parameters (Ssk) were used to evaluate deviations in the height distribution.Nega-  Since deep scratches are the key factor for surface quality before painting, their ΔZ heights were established (Table 5).This height (ΔZ) was calculated as a height between the points (if there is 1 extreme value -Samples #1 ÷ #4 -pink line, Fig. 9b) or central point of the mean line (if there are more extreme values -Samples #5 and #6 -pink area, Fig. 10b) and central point of the mean line of the plateau (pink areas).For the above points, the angle A (Table 5), which gives the information about distance between them, was also determined.Height (ΔZ) and maximum pit depth (Sv) in Sample #1, #3 and #4 were comparable, while differences among the remaining samples were signifi cant.In the case of Sample #5 and Sample #6, the ΔZ parameter was even 2.5 times smaller, meaning that height of the deep valley was smaller than Sv suggested.It should be emphasised that 3D surface texture parameters (ISO), together with additional parameters, are essential for defi ning irregular surface features.

ZAKLJUČAK
The conducted analysis of the planing process revealed that: The obtained total length of machined beams was signifi cantly higher for HW knives than for CR knives.
Even though the total cost of equipping the heads with knives made of HW was almost twice higher than those with CR knives, the production cost of 1 meter of glued pine beams was about 33 % less.
It can be stated that the surface characteristics based on 2D parameters are defi nitely insuffi cient.It is important to analyse the results of the 3D measurements, which gives a complete picture of the surface condition as a critical element of the entire machining process.
Total length of the material after machining, m ukupna duljina materijala nakon strojne obrade, m

Machine tool Stroj No. of tool spindles Broj vratila Tool type Vrsta alata Cutting depth, a p Visina
dodatka za obradu mm Feed per tooth f z Posmak po zubu mm