Non-destructive wood density distribution measurements using computed tomography

Abstract

Two computed tomography scanners have been used to acquire density profiles from five wooden test pieces. These test pieces had annual growth ring widths varying between about 0.6 mm and 3.5 mm. The two scanners employed were a medical scanner, a GE 9800 Quick scan system, and a purpose built microscanner constructed around a Rigaku D-Max II diffractometer. The results demonstrated that the annual growth rings could be resolved and density measurements could be reliably determined using the microscanner when the annual growth rings were about 0.9 mm or greater in width. The medical scanner could not satisfactorily measure density in carlywood or latewood resions even in samples where the annual growth ring width was of the order of 3.5 mm. A small difference in density measurements could be observed between the two scanners for the same specimens. The difference is attributed to calibration differences, the reconstruction algorithms used and the different geometries of the two systems; particularly in relation to the detector apertures. The larger detector size in the medical system results in too high or too low density values in the carlywood or the latewood respectively due to spatial averaging. However, the medical scanner provides satisfactory density accuracy in specimens where the density is very uniform or as long as measured areas include both carlywood and latewood. An accuracy of ±2–6 kg/m3 at 95% confidence level can then be expected.