Mobile Edition
Print
Get the Mag
Weekly UpdatesAbstract
One of the challenges when optimizing the positioning of a log in front of the primary breakdown saw is the limited time available for software to reach a solution. In this study, the distribution of optimal log positions in front of the primary breakdown saw, some variables that influence the software solution time, as well as the effect of positioning errors were investigated. The study is based on sawmill simulation results using real log images to simulate cant sawing in a softwood dimension lumber sawmill. The distribution of optimal log positions shows that a higher frequency of optimal positions is spread around the conventional log positions used (i.e., "horns-up," "horns-down," centered) but that in most cases the optimal log position will be different than the exact conventional position. A significant gain in volume recovery can be achieved if only a relatively small range of positioning combinations around the centered, "horns-up," and "horns-down" positions is considered. The relationship between the positioning range and volume recovery is a curve of diminishing returns. Once the curve starts to flatten, a large increase in the positioning range (and subsequently number of positioning combinations to simulate) is required to obtain a small gain in volume recovery. When reducing the increment sizes within a positioning range, the average volume recovery increased marginally. Errors caused by the positioning equipment can negate any volume recovery improvements from optimizing the search process. The range of positioning errors simulated resulted in average volume recovery losses of between 0.77 and 3.50 percent.
**********
When trying to position a log optimally in front of the primary breakdown saw using a computer controlled positioning system, one of the main problems is the short time available to arrive at a solution. The number of different possible log positions can run into tens of thousands depending on the positioning range and increments considered. This paper explores the problem of optimizing log positioning of small-diameter pine logs when using a cant-sawing pattern.
The selection of optimum sawing patterns and positioning methods for logs has been studied extensively since the 1960s (Peter and Bamping 1962, McAdoo 1969, Hallock et al. 1976 1971, Steele et al. 1987, Maness and Donald 1994, Chang et al. 2005). In the earlier work, highly idealized log shapes were used taking only diameter, length, and taper into consideration. As computing speed and software developed over the years, more variables could be incorporated in studies with some sawing simulation models able to represent the actual three-dimensional shape as well as the internal quality of logs (Occena and Schmoldt 1996, Todoroki 2001, Pinto et al. 2006). Real-time log positioning systems, however, are still constrained by the lack of suitable commercial scanning systems with the ability to scan the internal structure of logs fast enough, so they only consider log shape.
Optimization of the position of a real log and cant in front of the breakdown saws is a complex problem. Even though computing speed has greatly advanced over the years, real-time positioning systems still make use of simplified assumptions in order to arrive at a solution in time. For instance, often the only rotational position considered for a log where cant-sawing is applied is the "horns-up" or "horns-down" position (where the plane of maximum log curvature is vertical), especially where curve-sawing is applied (Gibson and Pulapaka 1999, Vuorilehto and Tulokas 2007). In another approach used to arrive at a solution in time, the log is positioned and sawn according to an existing stored position found in a look-up table for a log of similar shape (Steele et al. 1989). Even though assumptions made in order to speed up the process to find an "optimal" positioning solution in real-time systems might generally enable good positioning solutions, it will not be optimal for each individual log. To the author's knowledge, no research results have been published in which real log shapes were used to analyze optimum log positioning in terms of the three positioning variables--rotation, offset, and skewing--in a cant sawing system.
The objectives of this study were:
1. To determine and analyze the positioning range within which optimum and close-to-optimum solutions for individual logs are found in a small-diameter pine dimension lumber sawmill and where the three positioning variables of rotation, offset, and skewing can be adjusted accurately;
2. To determine the effect of the increment sizes used for different positioning variables when searching for an optimal position;
3. To determine the benefits, in terms of volume recovery, if positioning for individual small-diameter pine sawlogs is optimized; and
4. To determine the sensitivity of volume recovery to positioning inaccuracies or errors.
Methodology
Computer sawmill simulation of 60 real log images, randomly selected from a log database, were used for this study. A mix of plantation grown Mexican weeping pine (Pinus patula), slash pine (Pinus elliotti), and loblolly pine (Pinus taeda) log images from the Mpumulanga area, South Africa, was used for the study. Each species has its own shape peculiarities. Slash pine, for instance, tends to have reverse sweep. Many logs in the database have sweep in more than one plane. The sampling procedure was stratified to cover log diameters ranging from 180 to 219 mm, 240 to 279 mm, and 300 to 339 mm in small-end diameter representing three typical log sorting classes. The sample was also stratified to cover log lengths of 3.6 to 6.6 m. The sampling procedure was not stratified with respect to taper and sweep. Table 1 shows a summary of the study logs. These real log images were created by digitizing sample logs using a measurement system together with a specially manufactured log jig to ensure a high degree of accuracy. The digitizing of a log involved loading it onto the jig and marking positions along the length of the log where a digital cross-sectional profile was required. The X-Y-Z coordinates of two points on each cross section were determined by measuring the distance from two reference lines on the jig to each point. Each cross section was then removed using a chain saw and the exact profile of the cross section digitized. The two marked reference points on each cross section provided the means to position the cross-sectional profile correctly in three-dimensional space and creating an accurate log image. The sawmill simulation software package Simsaw5R was used for the simulation study. The first version of the Simsaw program was developed by the CSIR in South Africa in 1975, and the latest versions have the ability to simulate three-dimensional log images (Wessels et al. 2006).
The simulated sawmill represents a modern process where curve sawing and automated log and cant positioning are available. A cant-sawing process is followed where 27mm-thick side-boards (wet size) are first removed from each of two opposing faces of a log. The remaining slabs are sent to a resaw where, if enough material remains, additional 27-mm-thick boards are removed. The cant is processed by a multirip saw with curve-sawing ability following the centerline of the cant. The centerboards of the cant are cut into 41-mm boards (wet size) and the sides into 27-mm boards. All of the remaining slabs from the cant are also sent to a resaw where, if possible, more 27-mm-thick boards are removed.
Three sawing patterns were used for the three log classes (Fig. 1). All of the boards with bark were edged for optimal volume recovery and no wane was allowed on boards. Board widths of 80, 120, 160, and 240 mm were allowed with any length between 0.9 and 6.6 m in 0.3-m increments. The simulated log positioning variables include rotation, offset, and skewing (Fig. 2). Rotation was measured in degrees with zero degrees being the "horns-up" position. Offset is when the centerline of the log is shifted in the horizontal plane. It was measured in millimeters with zero being the center position. Skewing is when the front end of the log is kept at a constant position and the back end is shifted in the horizontal plane so that the log is fed through the primary breakdown saw in a "skew" position. It was measured in millimeters with zero being the centered position where no skewing is present.
The cant positioning method was kept constant in all of the simulations using curve-sawing and arris alignment. Arris alignment is the positioning of the cant so that the length and width of a specific board next to the "arris" blade is maximized. The arris blades in Figure 1 are marked as bold lines, and the board below the arris blade had the same length as the log and the full cant width. Note that this positioning method is non-symmetrical and the "horns-up" rotation will give different results than the "horns-down" method. This method of positioning a cant is often used with sash-gang saws and produces, in most cases, superior volume recovery results to centered solutions.
[FIGURE 1 OMITTED]
[FIGURE 2 OMITTED]
Evaluation of all of the results in this study is based on volume recovery--that is the product volume after drying, edging, and trimming as a percentage of the original log volume. The reason to use volume recovery is its universal application and independence of specific regional pricing policies. Although results may differ slightly when value recovery is used, the same principles will be applicable when searching for an optimum position based on value recovery.
Processing of each log was simulated for a wide variety of combinations of positioning variables. The simulation study was completed in two scenarios. The following annotation for the positioning variables is used: rotation; offset; skewing.
FEED