Nondestructive evaluation of end-jointed lumber using ultrasonic techniques

Abstract

End jointing short pieces of lumber is a promising way to reduce wood waste and better utilize wood resources. Maintaining quality of end joints in structural lumber requires intensive quality control (QC) efforts due to the complex interaction of adhesive, wood property, and process variables. Nondestructive evaluation (NDE) techniques need to be developed to facilitate real-time quality control inspection. One of the more promising NDE methods for inspecting wood products is stress wave propagation. Stress wave NDE techniques are based on the hypothesis that stress wave energy storage and dissipation are controlled by the same mechanisms that determine the static behavior of wood materials. The goals of this research were to: 1) characterize the mechanical properties of endjointed lumber, 2) explore ultrasonic techniques to identify the best predictor variables for end joint tensile strength, 3) compare ultrasonic techniques with other NDE techniques, and 4) evaluate multivariate statistical and neural network modeling approaches for predicting the tensile strength of end joints. Over 500 Southern Pine 2x6 end-jointed lumber specimens were randomly sampled from four manufacturers and across five lumber grades. Extensive material properties data were collected for the end joints, and ultrasonic pulse characteristics were analyzed in the time and frequency domains. Key ultrasonic features were extracted and multivariate statistical and neural network models were developed to predict the tensile strength of end-jointed lumber. Also, the ultrasonic method was compared with other NDE methods including static bending modulus of elasticity, impact stress wave, and transverse vibration methods. The linear correlation coefficient of the end joint bending MOE with ultrasonic wave MOE was 0.72, that with impact stress wave MOE was 0.91, and that with transverse vibration MOE was 0.95. However, three NDE methods such as static bending E, impact stress wave, and transverse vibration methods had relatively poor correlation with tensile strength. The linear correlation between actual and predicted tensile strengths with these three methods ranged from 0.40 to 0.64. By comparison, the linear correlation with ultrasonic waveform parameters ranged from 0.63 to 0.74. The ultrasonic method was the best predictor of end joint tensile strength. Differences between the specific gravity on either side of the joint had negative correlation with ultimate tensile stress (UTS). It would provide an incentive for using machine graded lumber for end joint applications. The neural network models and regression models gave similar results. In summary, the data collected in this study will be useful in models of engineered wood components and assemblies. Ultrasonic techniques appear promising for evaluating the tensile strength of the end joints. Future work should focus on refinement of dry-couplant transducers, and advanced signal processing and classification algorithms.