Time Domain Ultrasonic Signal Characterization for Defects in Thin Unsurfaced Hardwood Lumber

One of the major users of thin, unsurfaced hardwood lumber is the pallet manufacturing industry. Almost all manufactured products spend part of their life cycle on a pallet during transportation. This makes pallets a critical component of both the transportation and manufacturing sectors of the economy. Many newly constructed wooden pallets, however, are not currently manufactured to deliver the best performance (strength, durability, and safety)—despite interest by pallet users and pallet manufacturers—because manual grading and sorting of parts is impractical due to processing speeds and volume, labor costs, and laborer skill. This paper describes initial work aiming to create an automated grading/sorting system for hardwood pallet parts using ultrasonic. Experiments were conducted on yellow-poplar (Liriodendron tulipifera, L.) and red oak (Quercus rubra, L.) deckboards using pressure-contact, rolling transducers in a pitch-catch arrangement. Sound and unsound knots, cross grain, bark pockets, holes, splits, and decay were characterized using six ultrasound variables calculated from the received waveforms. Our scanning system shows good data-collection repeatability, and scanning rate has little effect on the calculated variables. For each defect type, at least one ultrasonic variable demonstrated significant capability to discriminate between that defect and clear wood. Energy loss variables exhibited the greatest sensitivity to many defect types. Based on the empirical relationships identified in this study, we are now developing models to classify defects using ultrasonic signal characteristics. Scanning properties of the prototype apparatus suggest that it can readily be translated into a commercial product