Analysis procedures for determining sawmill efficiencies

Typescript (photocopy).SAMTAM--Sawmill Analysis Model from Texas A&M was developed in three versions: (1) in Fortran for a mainframe computer, (2) in Applesoft for the Apple II, and (3) for the TI-59 calculator. SAMTAM is better than other models currently available for several reasons. First, the model provides two quality control checks, three recovery efficiencies, and a profit or loss (marginal log) analysis. Second, the residue volumes used in the analysis for a given log are based on the actual lumber sawn from that log. Third, the log density used in calculating residue weights is randomly selected for each log from a distribution of log densities found in a sawmill. The model is valid, operational, easy to use, and provides many of the answers that sawmill managers need to make production decisions. Prior to developing the decision model, four preparatory studies had to be conducted. The first study was conducted to determine an equation relating sawdust to the sawing pattern. The data from 65 logs resulted in a significant equation relating sawdust volume (VS) to the board feet of lumber produced (BF) and the number of pieces of lumber (PC). VS = -2.0 + 0.01 BF + 0.5 PC.The second preparatory study was conducted to determine the density of southern pine sawlogs. The data from 292 sawlogs were extremely variable. Analyses of variance indicated that mill, butt or upper log, and diameter were significant variables. Because of the variability, a distribution function relating density to log diameter and position in tree was incorporated in the decision model. A third study was conducted to determine a regression model for estimating sawing time. The data from three sawmills indicated that sawmills use more than one sawing method for the same size log and that the sawing time data should be segregated by sawing method. Cubic volume by Smalian's formula provided to be a significant predictor of sawing time. The fourth study was conducted to determine the effect of sample size and precision on the sawing variation estimate. Samples of 115 to 150 pieces of lumber were measured at three sawmills. Each piece was measured five times to the nearest 0.01 inch. The sample size was reduced in stages to 25 and the number of measurements per piece was reduced to three. The precision was changed in stages to 1/16 inch. There was no appreciable change prior to the 1/16 inch precision. It was recommended that for a given class, 25 pieces be measured at the distance of two feet from each end and in the middle of the board to the nearest 1/32 inch