### Abstract:

To compete with other building materials, the wood products industry must
find a way to increase value and lower costs. Wood stiffness is important for most
wood uses. Value can be increased and costs can be lowered by sorting logs for
stiffness in the woods using acoustic devices mounted on harvesters and processors in
order to properly allocate logs to their processing destination. This decreases shipping
costs because logs do not have to be reshipped if they are delivered directly to their
final processing location and increases value recovery because only logs that are fit for
the end use are processed.
The goal of this study was to determine if four increasingly more difficult-to measure
variables - length from the butt, acoustic velocity, bark percentage, and wood
density - can be used to predict the acoustic velocity variation along the length of a
tree in Douglas-fir and thus improve the utility of acoustic devices as tools for the
optimal bucking of trees into logs which have been sorted according to stiffness.
Research was undertaken in five stands. Six trees were selected from each stand.
Time of flight (TOF) acoustic velocity measured across the bole of the
tree had little value in predicting the longitudinal resonance acoustic velocity of a
section of the tree. Although TOF across the bole was statistically significant it had a
very low correlation coefficient and lacked the ability to accurately and precisely
predict resonance acoustic velocity. Wood density, moisture content, and bark
percentage were not significant predictors of resonance acoustic velocity of a section
of a tree.
The use of tree length resonance acoustic velocities was found to be a
statistically significant and strongly correlated predictor of acoustic velocity of tree
section. Distance of a log section from the butt of the tree and distance of a section
from the butt squared were statistically significant and strongly correlated predictors
of acoustic velocity of tree sections.
Final models including resonance acoustic velocity measurements, distance
from the butt, and distance from the butt squared were developed with high
coefficients of determination (R² 0.849 using all tree length acoustic velocities
(velocities taken every 3 meters up the tree that measured acoustic velocity of the
entire tree after the previous 3 meter section), R² 0.827 using initial tree length
acoustic velocity (acoustic velocity of the entire tree), and R² 0.678 using initial 3
meter log acoustic velocity). The models developed were found to be stand
dependent, indicating a possible need for model calibration for each stand to be
harvested.
Lower acoustic velocity in the butt of a tree due to high microfibril angle
inhibits the predictive capability of models based on acoustic velocity measurements
taken from the butt of the tree to predict acoustic velocity of 3 meter sections after the
first 6 meters of the tree. Acoustic velocity models based on a 3 meter log acoustic
velocity or tree length acoustic measured after the first 6 meters of the tree have been
removed are the best predictors of the acoustic velocity of 3 meter sections after the
first 6 meters of the tree. Final models based on measurements taken after the first 6
meters of the tree included resonance acoustic velocity measurements and distance
from the butt had high coefficients of determination (R² 0.860 using third tree length
acoustic velocity and R² 0.887 using third 3 meter section acoustic velocities). These
models were also found to be stand dependent.
The use of either a single tree length acoustic velocity measurement or a single
log acoustic velocity measurement with distance from the butt and/or distance from
the butt squared has the potential to increase value recovery from a log by predicting
the stiffness in that log and effectively matching it to its end use.