The effect of cutting speed on surface quality of black spruce (Picea mariana (Mill) B.S.P.) cants produced by a chipper-canter was evaluated. Four cutting speeds (18.9, 21.3, 24.2, and 27.1 m/s) were tested while feed speed was adjusted to obtain the same nominal chip length of 27.5 mm for each condition. For each speed, 15 logs were processed under frozen and unfrozen wood temperatures using 25 mm of cutting width at their small end. The surface quality was analyzed according to roughness and waviness standard parameters and to the depth of torn grain. The results showed that surface quality was not affected by the cutting speed. In contrast, surface was affected by the temperature of the logs and was better under unfrozen conditions. Quality of cants was also better at the small end of the log and generally at the upper part of the cant.ABSTRACT
Chipper-canters are the most common primary breakdown machines installed at sawmills in eastern Canada. These machines mainly convert small-diameter logs into cants at high speeds with very low sawdust production. The surface of cants is frequently produced by finishing knives, but this surface is unsatisfactory in certain cases. Some machine manufacturers install a thin circular saw in order to improve the quality; however, this saw increases the proportion of fines (sawdust), which is not desirable as raw material for the pulp and paper industry. Therefore, improving the surface of cants produced by the finishing knives is desirable.
Surface quality is very important in many areas of the wood products industry, including primary breakdown. Good surface quality from the first phase of wood processing mainly contributes to minimizing material losses due to the oversizing of wood pieces. It is also considered as a good indicator of the final product quality, level of wear of the cutting tools, as well as errors arising in the machining centers themselves (Lemaster and Taylor 1999). However, assessing the wood surface quality is very complex due to the wood anisotropy and property variations including anatomy, density, and moisture content (MC), along with the kinematics of the cutting process and machine conditions (Sandak and Negri 2005).
Traditional methods to measure surface quality include visual and tactile approaches; both of them are effective, but they are highly subjective. There is currently a variety of different surface texture tools available to measure wood surface geometry, including stylus profilometers, optical profilometers, ultrasonic optical light sectioning, and image analysis using video cameras (Hendarto et al. 2006). The surface geometry of wood can be considered as a superposition of several subgeometries related to various bases (Sandak and Negri 2005). Profile data from a nominally flat surface contain form errors, waviness, and roughness. Form errors are the long-wavelength deviations of the surface from the corresponding nominal surface, which usually results from an inaccurate alignment of the work-piece or uneven wear in machining equipment (Hendarto et al. 2006). Surface roughness is described by the texture effects due to the wood structure and by the effects produced by cutting the wood with a knife edge. Surface waviness is defined as the intermediate wavelength components produced by the machining process, including any deviations from the ideal waviness profile. The quality of a given surface is influenced by the type and condition of the wood work-piece, type and condition of the cutter equipment used, machine configuration, method of machine operation, and the engineering quality of the machine (Jackson et al. 2002).
Once the surface profile is obtained by any profilometer, filtering is applied to remove form errors and separate the waviness and roughness profiles. The cut-off length of the filter is the value that separates the wavelength that is within the range of interest for a particular feature from those that are not (ISO 11562 [International Organization for Standardization 1996]). The evaluation of the surface quality is a numerical characterization with parameters contained in general standards. These parameters allow comparisons between different surface textures. One of the most commonly used parameter is the arithmetical mean deviation of the assessed profile, Ra, defined in ISO 4287 (ISO 1997). This parameter, along with Rq, the root-mean-square deviation of the assessed profile (ISO 4287 [ISO 1997]), could give a good evaluation of the topography of wood surfaces (Khazaeian 2006). Similarly, Wz, the maximum height of the profile (ISO 4287 [ISO 1997]) can be calculated. Lemaster and Taylor (1999) noted that Rq and Wz are good descriptors of the fuzzy grain and torn grain in wood surfaces, respectively. However, the selection of a set of surface descriptors and of a suitable measuring technique to apply on wood surfaces is even difficult (Lemaster and Taylor 1999, Sandak and Negri 2005). For instance, Fujiwara et al. (2003) and Gurău et al. (2005) have suggested filtering techniques to eliminate the effect of wood anatomy from the estimates of roughness parameters. Sandak and Tanaka (2003) have indicated that accuracy of laser sensors for measuring roughness is dependent on the scanned wood properties (species, density, color).
There has been little research on the impact of chipper-canter parameters on cant surface quality. Most of the studies have been focused on size distribution of pulp chips produced by this machine (Hernández and Quirion 1993, 1995; Hernández and Lessard 1997). In a recent study, Hernández et al. (2010) investigated the effects of the cutting height and cutting width on the surface quality of black spruce (Picea mariana (Mill) B.S.P.) cants produced by a chipper-canter. Frozen and unfrozen logs were processed. The results showed that the surface quality was better for unfrozen logs processed at lower cutting width and cutting height. Furthermore, surface quality varied within the cant, being generally better at the small end of the log and at the upper part of the cant. This emphasized the importance of the diameter of the cutterhead. Thus, the surfaces would be more uniform if processed with a large rather than a small diameter cutterhead. Additionally, the authors indicated the advantage of feeding the log into the machine just below the center of the cutting circle or cutterhead. In good agreement with Koch (1964), these results suggest that the wood machining process involves interactions between the workpiece, cutting tool, and machine to produce the desired surface quality on the final product.
Few studies have investigated the impact of cutting speed on wood surface quality. Results from these studies are somewhat contradictory. For example, Kivimaa (1950) found that an increase of cutting speed (from 2.5 to 50 m/s) has no effect on cutting forces in a 0°–90° process. Franz (1958) concluded that the surface quality (determined by the chip formation) was independent of cutting speed in a 90°-0° cutting direction. McKenzie (1960) also found that an increase of cutting speed from 0.0002 to 142 m/s caused a small increase in cutting forces, with no distinguishable change in surface quality. However, for a 90°–90° cutting situation, Ohta and Kawasaki (1995) reported that cutting speed affects the cutting mechanism of wood and observed that this difference came from the change in the apparent stiffness of specimens under the different cutting speeds. Using scanning electron microscopy, they found that surfaces cut at a very-low cutting speed (0.08 mm/s) were very rough, with cell wall deformation and splits along the fiber. At high-speed cutting (5 m/s), the surfaces were rather smooth with less cell wall deformation. Similarly, Costes and Larricq (2002) concluded that cutting speed affects the routing across the grain (90°–90°) and that MC seems to play a significant role. They conducted a routing experiment with constant mean chip thickness (0.2 mm) and cutting speeds varying from 3 to 62.2 m/s. The surface quality of ovendry pieces increased as cutting speed increased. However, for green pieces (55% MC) changes in cutting speed did not affect surface quality.
The aim of this work was to study the effect of cutting speed on surface quality of cants produced by a chipper-canter processing black spruce. Logs were processed using four cutting speeds under frozen (winter) and unfrozen (summer) wood conditions.
Materials and Methods
A total of 120 stems of black spruce were selected for this study. The stems were crosscut into 2.44-m logs and were freshly debarked. The crosscutting position of the stem was chosen to yield logs with a small end diameter inside the bark of 152.4 mm and a mean taper of 8 mm/m. The logs were without crook or visible decay and had a minimum of knots, straight grain, and concentric growth rings.
Log processing was done with a Swecan chipper-canter with two side opposed end-milling cutterheads, which have the shape of shallow truncated cones (Fig. 1). Each cutterhead was fitted with eight uniformly distributed knife holders, each of them with a bent knife and a knife clamp. The bent knife had two cutting edges, which were joined at an angle; the longer edge severed a slice to make chips and the shorter edge smoothed the cant. Basically the shorter edge of the knife cut nearly across the grain at the point of entry on the log and more obliquely to the grain as the knife exited from log (Hernández et al. 2010).
Citation: Forest Products Journal 63, 1-2; 10.13073/FPJ-D-13-00016
The experiment consisted of processing black spruce logs using four cutterhead cutting speeds with the feed speed adjusted to obtain a nominal chip length of 27.5 mm for each cutting speed value. In addition, the seasonal effect on log process was evaluated by conducting the experiment during winter and summer seasons. Fifteen logs were used for each cutting condition, yielding a total of 60 logs for each season.
Sawmill experiments
The log temperature was measured using a digital thermometer to the nearest 0.1°C at two uniformly spaced points at a depth of 20 mm. During processing, the log was always fed with the small end first. The feed system of the chipper-canter includes a rugged steel frame with an automatic self-centering belt mechanism. The log was machined flat on only two sides in order to produce a 101.6-mm-wide cant. The maximum cutting width at the small end was, therefore, 25 mm under all speed conditions.
The cant obtained was immediately weighed, painted on both ends, and wrapped in polyethylene to maintain its initial MC until measurement. The knife angle of the shorter edge of the knife (finishing knife) was set constant at 28°, with a rake angle of 60°. The four cutting speeds used were 18.9, 21.3, 24.2, and 27.1 m/s. Feed speed was adjusted, respectively, to 119, 134, 154, and 172 m/min, which gave a calculated feed per knife (or chip length) of 27.5 mm. The knives were freshly sharpened before the experiment to minimize the effect of tool wear on the surface quality. The processing was done between −15°C and −5°C in the winter and between 11°C and 16°C in the summer.
Laboratory experiments
The cants were used to prepare boards for wood surface quality tests and smaller specimens for physical property tests. The cants were cut into boards 840 mm long from each side and each log end. The boards from the two ends allowed the evaluation of the effect of the log taper on the wood surface quality. Also, the effect of the finishing knife edge orientation with respect to the wood grain on the surface quality was analyzed. The round-edge or wany part of the cant was used to assess the mean specific gravity (reported as the ovendry weight–to–green volume ratio) as well as the MC of both sapwood and heartwood at the time of transformation.
Surface topography evaluation
Roughness and waviness of the cants were measured using a MTI Microtrack system 7000 provided with two MT-250 sensor laser heads. The data were collected with LabView software using an acquisition frequency of 50 Hz and a scanning speed of 15 mm/s. Roughness and waviness of the cant were assessed according to two directions: along and across the grain. For the evaluation along the grain, two profiles in the upper part and two others in the lower part of the board were assessed following the grain direction. These profiles were 220 mm long and corresponded to the second complete rotation of the cutterhead (boards coming from the small end of the log) or the next to last revolution of the cutterhead (boards coming from the large end of the log).
For the assessment across the grain, eight profiles (one per knife) corresponding to one rotation of the cutterhead and following the cutting direction were taken per board. The length of the profiles corresponded to the width of each board. This assessment covered the same surface (cutterhead rotation) used for the evaluation along the grain (at the small end and at the large end of the log).
For the evaluations along and across the grain, 12 surface quality parameters were determined according to ISO 4287 (ISO 1997) using a task software developed with LabView software (Table 1). A cut-off length of 2.5 mm and the robust Gaussian filter ISO 16610-31 (ISO 2002) were applied for calculations.
The maximum depth of the torn grain present in each board was measured with a Micromeasure confocal microscope. Such depth represents the distance between the cant surface and the lowest point at the bottom of the defect. This measurement was made within the same region (corresponding to the same cutterhead rotation) used for roughness and waviness evaluation. The data were collected with Surface Map 2.4.13 software using an acquisition frequency of 300 Hz and a scanning speed of 12.5 mm/s.
Statistical analyses
Statistical analyses were performed by means of SAS package version 9.2 (SAS Institute 2007). The raw data were first transformed using the Box and Cox method. All surface parameters were transformed using the logarithmic transformation. Given the number of surface parameters studied, a principal component analysis (PCA) was then applied to data in order to regroup them in common factors and facilitate their analysis.
A split-plot analysis of variance was used to evaluate the variation in surface quality of the processed cants at 99 percent confidence level (mixed procedure of SAS Institute). The cutting speed and season were the sources of variation as principal plots and the log end was the source of variation as subplot. For the assessment along the grain and for torn grain, the position (upper and lower part of the cant) was added as a source of variation as subplot. The normality of the data was verified using the Shapiro-Wilk test (SAS Institute 2007).
Correlation analysis among surface topography and torn grain were performed with a mixed procedure from the SAS package. When required, means were compared with the least squares means statement from SAS GLM procedure at a 95 percent confidence level (SAS Institute 2007).
Results and Discussion
The mean specific gravity was 0.437 for sapwood and 0.428 for heartwood. This difference was not statistically significant at the 0.05 probability level. However, the difference in MC of sapwood and heartwood during cutting was high. In winter, the sapwood MC was 143 percent and heartwood MC was 37 percent. While in the summer, MC was 108 percent for sapwood and 45 percent for heartwood.
Surface topography evaluation
Principal component analysis (PCA)
The purpose of a factor analysis is to determine the number of common factors and their factor loading (Tabachnick and Fidell 2007). The factor loading, which is obtained for each component within the factors generated by the PCA, is a type of correlation coefficient, for which a higher value is associated with greater significance. The number of factors was defined according to the Kaiser criterion (Kaiser 1960), which retains only the factors with an eigenvalue greater than 1 (Table 2). In addition, a varimax rotation was needed for the measurements along the grain.
The PCA results are shown in Table 2. The PCA of the profiles measured along the grain showed that 92.8 percent of the variance of the scaled data was explained by two factors. The first represents the surface roughness and explained 49 percent of the total variance. The second factor accounted for 43.8 percent of the total variance and represents the surface waviness. The PCA of the profiles measured across the grain revealed a different structure underlying the variables. The surface parameters were grouped into one principal factor, which explained 78.4 percent of the total variance. This factor adequately describes the roughness and waviness of the wood surface.
The differences in the PCA results between the two assessments can be explained by the fact that the profiles along the grain were more irregular than those across the grain. Thus, parameters were more variable and less likely to be correlated with each other. Similar conclusions were reported by Hernández et al. (2010) in a recent experiment.
The variance in the topography along the grain was explained by two factors as mentioned in the PCA. The first factor represents the roughness and the second, the waviness of the wood surface. Results of the analysis of variance showed that season and position had an influence on both factors (Table 3). The effects of the main factors season and position on waviness were highly significant while the effects of the interaction season × position on roughness was significant. In addition, waviness was affected by the log end. However, the cutting speed did not affect either roughness or waviness along the grain. Roughness and waviness exhibited greater values when processing frozen logs. A similar effect of log temperature on surface quality produced by the chipper-canter has been previously reported (Hernández et al. 2010).
The mean values of roughness (Ra) and waviness (Wa), 2 of the 12 parameters included in the PCA, are presented in Table 4 as an example. Table 4 shows the effects of the temperature, position, and log end on both factors. Waviness presented greater values at the lower position compared with the upper position of the cant. Roughness had the opposite behavior, but the difference between both positions was very small (Ra, 0.4 μm). The finishing knife of the chipper-canter cuts the log nearly across the grain at the point of the entry and more obliquely to the grain as the knife exits from the log. As the angle between cutting edge and the grain becomes more oblique, the surface quality will decrease (Stewart 1969). In this particular case, for a 152.4-mm log diameter and 25-mm cutting width, the angle of the finishing knife increased from about 2° (orientation, 2°–88°) at the entrance to 25° (orientation, 25°–65°) at the exit. This could explain why the surface quality was lower at the lower part of the cant compared with its upper part.
On the other hand, the diameter of the cutterhead (or diameter of the cutting circle) should also affect the surface quality of the cant. The cant surface will be more uniform when processed with a large diameter cutterhead than with a small cutterhead. For a given log, orientation of knives with respect to the grain will be more constant in larger cutterheads than in smaller cutterheads. Furthermore, logs should be fed into the machine such that knives can cut nearly across the grain rather than more obliquely.
Finally, waviness along the grain increased as the cutting advanced toward the large end of the log (the logs were always fed with the small end first; Table 4). As the log cutting advanced toward the large end, cutting height and cutting width became greater due to the log taper. It is expected that the cutting forces will increase as more material is transformed, which happens when cutting height and width increase. Consequently, waviness increased from the increase of the cutting forces (Hernández et al. 2010).
As mentioned in the PCA, the variance for across the grain analysis was explained by one principal factor. This factor adequately represents roughness and waviness observed over the cant surfaces. Similar to waviness along the grain, roughness and waviness across the grain presented a statistically significant simple effect of season and log end. Also, cutting speed did not affect roughness and waviness across the grain (Table 3). These parameters had greater values when cutting in frozen conditions, and they increased as the cutting advanced toward the large end of the log. This behavior is shown in Table 5 for Ra and Wa.
Tables 4 and 5 show that mean roughness was on average 2.4 times higher (summer, 2.5; winter, 2.3) across the grain than along the grain. Alternatively, waviness was on average 1.5 times higher along the grain than across the grain (summer, 1.3; winter, 1.7). Similar results were reported by Hernández et al. (2010). Higher waviness along the grain was expected because this direction was affected by the variation in the projection of the eight knives mounted on the cutterhead. This confirms that quality control of the knife projections on the cutterhead plays an important role on the quality of surfaces produced by the chipper-canter.
Torn grain evaluation
Means of the maximum depth of torn grain for all cutting conditions are shown in Table 6. Values ranged from 0.8 to 2.7 mm, depending on the cutting conditions. According to Table 3, only the season of machining showed a statistically simple effect on the maximum depth of the torn grain. The other sources of variation were not significant. The maximum depth of torn grain presented greater values when processing logs under frozen conditions. Torn grain was 1.5 times deeper in winter than in summer. In addition, a higher incidence of torn grain was observed at the lower part of the cants during both seasons. Seventy-six percent of the boards had the deepest torn grain at the lower part in contrast with 24 percent at the upper part of the cant. Hernández et al. (2010) explained this behavior as a result of the change in the finishing knife edge orientation with respect to the grain. It becomes more oblique as the knife exits the log. Consequently, the incidence of torn grain at the lower part of the cant could increase.
The surface quality of black spruce cants (roughness, waviness, and maximum depth of the torn grain) produced by a chipper-canter was not affected by the cutting speed (Table 3). This result is in good agreement with previous findings (Kivimaa 1950, Franz 1958, McKenzie 1960). However, this result contradicts a previous study (Ohta and Kawasaki 1995). These differences could be explained by several factors including the MC of wood, modes of machining, as well as the cutting speed levels studied. Working with dry wood, Ohta and Kawasaki (1995) noted differences in roughness in a comparison of quasistatic cutting (0.08 mm/s) with high speed cutting (5 m/s). In contrast, cutting speeds in the present study were higher and the range of speeds (18.9 to 27.1 m/s) was probably too small to detect any effect on cant surface quality. MC of logs was also at the green state. According to Costes and Larricq (2002), increasing routing speed had a positive impact on wood surface only on ovendried wood samples. On saturated samples, surface quality was not significantly affected as cutting speed increased.
There are also operational conditions that could cover up an eventual effect of cutting speed on the wood surface quality. Undesirable vibration is generally generated during the process and becomes more serious with increasing cutting speeds. The vibration could introduce a strong variability in wood surfaces that would eventually minimize the effect of cutting speed compared with the high variability encountered by vibration. For example, the coefficients of variation for all surface parameters along and across the grain obtained at the different cutting speeds were always superior to 45 percent.
On the other hand, the temperature of the log had an important effect on wood surface quality produced by this machine. The surface quality was better when processing the logs under unfrozen conditions. Hernández et al. (2010) reported a similar effect of log temperature on surface quality produced by the chipper-canter. Several researchers have detected opposite behavior but for other wood machining processes. For example, Yu et al. (1997) reported that surface finish of band sawn frozen wood was a little better than that of unfrozen wood. Orlowski et al. (2009) also found that mini gang sawn surfaces were smoother in unfrozen than in frozen wood. Equally, Lunstrum (1985) observed that frozen wood is more brittle than unfrozen wood and can therefore be sawn cleaner. The efficiency of the feeding system of the chipper-canter could again explain why machining frozen logs produced a somewhat worse surface than unfrozen logs. Cutting forces increase in winter conditions, generating undesirable vibrations that will affect cant surfaces.
Correlation analysis
A correlation analysis was made between the maximum depth of torn grain and roughness and waviness along and across the grain. The factors of the PCA of both assessments were used in the calculation given that the individual parameters have a strong correlation between them, as shown in Table 2. The PCA revealed two principal factors for the along the grain assessment. These factors showed a statistically significant correlation with the maximum depth of torn grain. The corresponding regression model explained 30.8 percent of the variation in torn grain:
Similarly, a correlation analysis was performed between the maximum depth of torn grain and the principal factor of the across the grain assessment. A statistically significant correlation was also found; as the principal factor (roughness and waviness across the grain) increases, the torn grain will be deeper. The corresponding regression model explained 53.4 percent of the variation in torn grain:
Therefore, waviness and roughness across the grain satisfactorily explained the variation in the maximum depth of torn grain. Similar results were found by Hernández et al. (2010). Because of its higher magnitude, torn grain could be favored as a good predictor of surface quality. Thus, any optimization of the cutting conditions for reducing torn grain depth should also reduce the level of waviness and roughness.
Conclusions and Recommendations
This work has shown that surface quality (roughness, waviness, and depth of torn grain) of black spruce cants produced by a chipper-canter was independent of cutting speed (within the range studied). The temperature of logs played an important role in cant surface quality. Moreover, roughness and waviness increase at the large end of a log because of the constant increase of cutting width and cutting height due to the log taper. Finally, the surface quality of the cants was in general rougher at the lower part of the cut due to the change of the finishing knife orientation with respect to the wood grain. The correlation analysis showed that roughness and waviness across the grain positively explained the variation of the maximum depth of torn grain with an R2 of 53.4 percent. Any reduction of torn grain would also have a positive impact on the roughness and waviness of the cant surfaces. Finally, results confirmed the benefit of feeding the logs into the machine just below the center of the cutting circle.
View of the end-milling cutterhead (left side) of the Swecan chipper-canter showing some knife holders.
Contributor Notes
The authors are, respectively, Professor and PhD Candidate, Centre de recherche sur le bois, Département des sciences du bois et de la forêt, Univ. Laval, Québec, Canada (roger.hernandez@sbf.ulaval.ca [corresponding author], Svetka.kuljich-rios.1@ulaval.ca); and Former MSc Student and Professor, Univ. du Québec en Abitibi-Témiscamingue, Rouyn-Noranda, Canada (Oussema.naffeti@uqat.ca, ahmed.koubaa@uqat.ca). This paper was received for publication in February 2013. Article no. 13‐00016.