Ultrasound for defect detection and grading in wood and lumber

Ultrasonic inspection of solid wood and lumber has been investigated for over 20 years, with limited commercial impact. Potential applications run from logs to finished boards, with the goal of either characterizing internal defects (knots, splits, rot), or assessing overall strength. This paper reviews recent efforts to apply ultrasound technology to the Forest Products industry. A practical ultrasound system must operate continuously, unattended, over a range of temperatures, and provide fully automatic defect detection. This required new transducer configurations, signal processing methods, and detection algorithms. Compounding the problem, wood is a highly anisotropic material, with sound speeds from 1200 to 5000 m/s depending upon wood fiber (grain) orientation. However, rather than being an obstacle, this anisotropy can be exploited as part of the detection process. Because many defects involve either a change in grain orientation (knots), or a disruption of the fiber integrity (splits, pitch inclusions), we developed novel signal processing indices to accentuate the detection of different defects. For example, knots can be distinguished from rot by using different combinations of basic waveform parameters (i.e. intensity, time of flight, etc.). Time of Flight may also be used to estimate the Modulus of Elasticity (Y). Because of the highly anisotropic nature of the material, mode conversion can be used to measure longitudinal sound speeds (along the fiber direction) from the sides of the boards. Systems have been built using multiple ultrasound transducers in through-transmission mode; these operate from 120 to 400 kHz depending upon the resolution required and the thickness of wood. Specific examples are described, including: automatic classification of "cants" (partially processed logs) for grading; detection of sawing induced splits in logs; determination of bond strength in laminated products, and board defect imaging for automated saw control. Efforts are continuing towards automated lumber grading, which would combine defect imaging with strength estimation in a single measurement system. One limiting factor is the transit time through a board, while it is travelling past the transducers at over 2000 linear feet per minute.