Pallet Part and Cant Evaluation for Grading and Processing Using High-Speed Ultrasound

Mark Schafer President and Principal Scientist, Sonic Tech Wayne, PA This paper presented the results of several years of testing the use of ultrasound to find structural defects in pallet parts and pallet cants used to produce pallet parts. To determine the magnitude of unsound defects, we inspected full length cants from seven saw mills from Virginia and West Virginia. Split, wane, shake, holes, decay, unsound knots, bark pocket, and mechanical defects were all considered as unsound defects. Regardless of mill and species, split accounted for the highest percentage of defect volume per cant. Decay: bark pockets, shake, and holes also contributed significantly to the total defect volume. The scanning results of deckboards, stringers and cants showed that ultrasound can be used effectively to locate, identify, and quantify the various pallet parts defects. Significant losses of ultrasound signals/energies were observed through these defects. Other ultrasound parameters, such as time of flight, pulse length can be used for characterizing defects. Reconstructed two-dimensional images were able to show the exact position and surface area of the defects. Defects can be classified/distinguished using classifying tools like multi-layer perceptron, probabilistic neural network, and k-nearest neighbor. Results demonstrated that real-time, online inspection and classification of defects in wooden pallet parts are possible by ultrasound scanning. Results also showed that pallet cants can be graded by unsound wood content and that processing full length cants to cut-to-size cants can be optimized with the knowledge of defect size and location provided by the ultrasound scanning. Introduction Grading and sorting of pallet parts are important for manufacturing high quality and longer lasting pallets. Cants are the primary raw materials for the production of pallet part components, such as stringers and deckboards. The stringers form the structural center members that support the load, and the deckboards are the top and bottom members that provide dimensional stability and a platform for product placement. Usually, pallet parts are produced from the low grade lumber or cants. These cants have a higher percentage of defects, and 133 Proceedings of Scan Tech 2003 International Conference, Seattle, Washington, U.S.A. therefore have less or no market value for other solid wood products. Use of higher quality cants results in higher parts yields, providing longer pallet life cycle. Higher quality pallet parts will ensure reuse of discarded pallets. Individual pallets are constructed from a variety of wood species and from parts with differing strength properties, resulting in pallets with random and unknown strength durability. Descriptions of allowable defects for minimum pallet component quality are shown in Table 1 (Anon 1994). These data are incorporated from the Standards for Wood Pallets, National Wooden Pallet and Container Associations (NWPCA). It is obvious from this table that defect information is important for manufacturing quality and durable pallet. Presorting and optimized sawing of pallet parts can reduce processing cost and will produce higher grade pallet parts. Schmoldt et al. (1993) also showed that increased pallet durability and performance impacts in a positive way the cost of using a pallet. Manual grading and sorting of pallet parts is a slow, inaccurate, and difficult process, which depends on the individual skill of the grader. Moreover, the presence, location, and extent of the defects in pallet parts are often difficult to determine accurately, making manual grading complicated. In the last decades, ultrasound became a popular tool for nondestructive evaluation of wood and woodbased products. The great advantage of using ultrasound is the direct relationship with the density, and elastic stiffness of wood. Generally, the defects found in pallet parts are sound and unsound knots, decay, splits, bark pockets, wane, etc. The location, extension, and type of defects must be characterized for grading and sorting of pallet parts (Table 2). The severity and extent of these defects often depends on the species as shown by Araman et al. (2003). Figure 1 illustrates how ultrasound scanning can be used for optimized sawing and increased yield of pallet parts from cants. The most common way of detecting defects is the measurement of transmission time through clear and defected wood. The measurement of transmission time is useful when there is only a single type of defect in board. Some defect types may not respond well to the transmission time, but may respond the other ultrasonic parameters, e.g., peak amplitude, centroid time, frequency domain energy, etc (Halabe et al. 1994, 1996). The recognition and classification of defect types in a single board is a very difficult and complicated task, since changes in the measurement parameters for many of these defects are similar. For the last few years, research has been conducted to develop an automated pallet part inspection system (Schmoldt 1997, Kabir et al. 2002, 2003). This paper discusses the possibility of ultrasound scanning for cant evaluation and pallet part grading. Unsound Defect Evaluation of Cants Seven sawmills were selected in Virginia and West Virginia to inspect pallet cants. Mill selection was based on the convenience of travel, type of sawmill operation, and willingness to participate. The species available and studied were red oak, white oak, yellow-poplar, and basswood. Cants were 10 to 16 feet length, 6 inches in width, and 4 inches in thickness. Defect data was then collected for any occurrence of splits, wane, shake, holes, rot, decay, unsound knots, bark pockets and mechanical defects. Defects were identified and classified according to the NHLA grading rules (Anon 1994). An imaginary rectangular solid was visualized around each defect, such that the entire unsound area was encompassed. Length, width and height dimension measurements were taken of the imaginary solid. All the measurements were taken from the wide face of the cant, even if the defect was located on the narrow face. If a defect was