| Although automatic and computerized equipment has been incorporated into modern gang-rip first rough mills, many of the process steps still rely on human decision making. Examples include choosing the appropriate lumber grade for processing; designing the optimal gang-rip saw arbor; and defining part priority values for the chop saw. This research was based on the hypothesis that these decisions can be improved through optimization strategies without extra capital investment.; This research first developed a software program, the Gang Ripsaw Optimizer (GRO) written in C++ language. GRO generates an optimized fixed-blade arbor by reiteratively searching and comparing optimal part combinations provided by the Romi-Rip 2.0 simulator. Results showed that GRO provides better solutions than two other available arbor design programs.; Second, this research created a system that generates static priority values through a 20-factor face-center central composite design. Various formulas were formed based on part sizes. Results indicated that the static value mode could produce a yield comparable to that given by the dynamic modes for representative cutting bills. On average, the yield from the static value mode was 0.99 percent lower than that from the complex dynamic exponential mode, 0.97 percent lower than the simple dynamic exponential mode, and 0.87 percent lower than simple dynamic mode, respectively.; Third, the simple linearity assumption required to apply linear programming to solve the least-cost lumber grade-mix problem was examined. The study statistically proved that the simple linear relationship between yield and two- and three-grade lumber combinations does not hold for 90 percent of the industrial cutting bills. To solve the least-cost lumber grade-mix problem without violating the assumption, a five-factor mixture design was applied. Upper bounds were applied to 3A Common lumber, according to the cutting bill's level of difficulty, because of the limited availability of larger parts. By locating the lowest cost point from the lumber grade—cost response surface, the corresponding lumber grade mix was obtained. This statistical method allows the user to pre-specify the lumber grades that are available. |
