P Rocesses T Echnical N Ote

A versatile microcomputer-based interactive graphics sawing program has been developed as a tool for modeling various hardwood processes, from bucking and topping to log sawing, lumber edging, secondary processing, and even veneering. The microcomputer platform makes the tool affordable and accessible. A solid modeling basis provides the tool with a sound geometrical and topological foundation. Owing to the simulation’s flexibility, it can be used in a variety of wood-processing operations. Within any operation, it can be used to analyze processing alternatives, or to aid in the training of personnel. In the area of production operations, computer simulation has proven to be a versatile analytical supplement to physical testing. In lumber production, computer simulation is of special interest because it enables the repeated sawing of a log sample, in essence a means for examining the effects of several log breakdown patterns and sawmill variables. There have been many previous simulation studies on hardwood log breakdown (e.g., 2,4,6-8,10, 12,13 ). This type of simulation is of a physical nature that involves geometric representation, in contrast to systems simulation studies (3) that observe the operation of a system over time. In the late 1970s, Pnevmaticos et al. (7) introduced the first graphic simulation of log sawing using a hybrid graphics terminal. Graphic simulation has the advantage of a visual feedback on spatial relations vital to hardwood processing. Pnevmaticos used truncated cones and cylinders to approximate log shapes, and rectangular boxes to approximate defects. Finding the intersection of the log with the saw was treated as a linear programming problem. In 1988, Occeña and Tanchoco (4) reported the development of a graphic log sawing simulator as an analytical tool 40 for automated hardwood log breakdown. The log and its defects were represented as nonregular polyhedra and sawing was treated as a Boolean operation between closed solids. The polyhedral model more closely approximated the true shape of the log and its defects than previous log/defect graphics models. It was implemented on a minicomputer platform using device-independent graphics and user-developed programs. A subsequent paper by Todoroki ( 12) reported the development of an automated sawing simulation program also aimed at evaluating sawing strategies. Logs were represented as a series of polygon cross sections and defects as cross-sectional whorls. The use of nonsolid models emphasized speed over precision and flexibility in the representation of the log and defects. The continued interest in developing operation simulators attests to the significance of such modeling tools. Concurrently, researchers have been pursuing studies on non-invasive internal defect detection (1, 11, 14) with the intent of developing the capability to “see” internal log defects. This capability will lead to improved hardwood log breakdown decision making that will yield higher value lumber. A logical consequence of the capability to “see” inside a log prior to sawing is the need to resolve the issue of how such information can best be used to arrive at a higher yield. Studies have shown that irrespective of log grade, the optimal orientation of internal defects with respect to sawing pattern increased value yield over 10 percent on the average, and that a precise knowledge of internal defect location was required (9, 10). Given the importance of a realistic representation of the log and its internal defects, and the usefulness of a tool that will enable repeated interactive sawing of the same log sample, an interactive graphics sawing program for wood processing that runs on a microcomputer has been developed. This paper describes a prototype microcomputer-based interactive graphics sawing program for wood processing that integrates graphics rendering, solid The authors are, respectively, Associate Professor, Dept. of Industrial Engineering, 113 Engineering Bldg. West, Univ. of Missouri-Columbia, Columbia, MO 652 11; Research Forest Products Technologist, Southern Res. Sta., Blacksburg, VA 24061-0503. This research was supported in part by the USDA Forest Serv., Southern Research Sta., Primary Hardwood Processing, Products, and Recycling Res. Unit SE-4702, Blacksburg, VA. This paper was received for publication in September 1995. Reprint No. 8427. ©Forest Products Society 1996. Forest Prod. J. 46(1 1/12):40-42. NOVEMBER/DECEMBER 1996 Figure 1. — Reconstructed log. Figure 2. — Reconstructed knot defect. modeling, and data representation. An earlier version of the program founded on the same modeling principles only ran on a much larger computer (minicomputer platform) and did not have a tightly integrated environment (4). The current graphics sawing program, which we call GRASP (for GRAphic Sawing Program) is unique in its flexibility to model just about any sawing operation, from bucking, topping, log breakdown, quartering, and veneering, to edging, trimming, secondary processing, even extracting and representing furniture components. It is founded on solid modeling principles, which endows it with a robust foundation in geometry and topology. The implementation on a microcomputer platform makes it an affordable and accessible tool for many users. D ATA RECONSTRUCTION The input data for GRASP can be an object from any stage of wood processing, e.g., timber, log, quarter-log, flitch, board, blanks, etc. They have to be repre1 Logs were obtained from the USDA Forest Service Southern Research Station, Unit SE-4702, Blacksburg, Va. Figure 3. — Wire frame rendering with internal knot defects revealed. Figure 4. — Zoomed-in view of a section of the wire-frame image. sented as closed polyhedral solids, i.e., a set of concatenated polygonal patches or faces that fully envelopes a region describing the object. In this current development, we focused on logs. The sample data consisted of three approximately 12foot-long red oak logs. The data came in the form of both CT scan and digitized coordinates, representing the cross-sectional profiles of the log and its internal defects. The log profiles were sampled at 2-inch intervals, and the defect profiles at l/4-inch intervals, along the length of the log. For these raw data to be usable by GRASP, solid representations of the log and of the defects had to be reconstructed from the cross-sectional profiles. The log and its defects were reconstructed as separate solids, while retaining their original relative spatial orientations and locations. GRASP itself was used as a tool for reconstruction, uniting the profiles into polyhedral solids. To reduce the magnitude of the resulting data files, the log profiles that did not exhibit significant variation from their adjacent profiles were excluded from the log reconstruction. The same was done for the defect profiles in the defect reconstruction. This data reduction procedure still maintained the integrity of the solid Figure 5. — Quartersawn log for veneering process. representation(5). Figures 1 and 2 illustrate a reconstructed log and knot defect, respectively. Note the irregular intervals of log cross-sectional polygons resulting from the data reduction.