TY - JOUR
T1 - Identification of the length distribution of lumber defect-free areas
AU - Mysyk, Mykhailo
AU - Matsyshyn, Yaroslav
AU - Mayevskyy, Volodymyr
AU - Ray, Charles D.
AU - Kurka, Roman
AU - Sopushynskyy, Ivan
PY - 2017/1/1
Y1 - 2017/1/1
N2 - Presented here are the results of statistical analysis of length distribution of defect-free areas (DFA) of pine, beech, and oak blanks. The investigated empirical distributions of the lengths of defect-free areas nearly always exhibit right-side asymmetry and “heavy tails,” with high coefficients of variation (30% c 110%). Therefore, the arithmetic mean of these lengths is not an appropriate measure for description of any of the investigated samples, and to describe the dimensional and qualitative characteristics of blanks, not only characteristics of location must be used, but also the relative characteristics of the dispersion. It is proposed that rather than use estimates of variability, to apply an assessment of stability—the value inverse to the squared coefficient of variation, which allows, with minimal computational cost, to correctly compare the lengths of DFA obtained from different lumber and in different operating conditions. It is shown that the distribution of lengths of DFA for pine, oak, and beech blanks can be only described entirely by two theoretical distributions—the Burr and log logistic, with different parameters for different wood species and various sizes of defect-free areas.
AB - Presented here are the results of statistical analysis of length distribution of defect-free areas (DFA) of pine, beech, and oak blanks. The investigated empirical distributions of the lengths of defect-free areas nearly always exhibit right-side asymmetry and “heavy tails,” with high coefficients of variation (30% c 110%). Therefore, the arithmetic mean of these lengths is not an appropriate measure for description of any of the investigated samples, and to describe the dimensional and qualitative characteristics of blanks, not only characteristics of location must be used, but also the relative characteristics of the dispersion. It is proposed that rather than use estimates of variability, to apply an assessment of stability—the value inverse to the squared coefficient of variation, which allows, with minimal computational cost, to correctly compare the lengths of DFA obtained from different lumber and in different operating conditions. It is shown that the distribution of lengths of DFA for pine, oak, and beech blanks can be only described entirely by two theoretical distributions—the Burr and log logistic, with different parameters for different wood species and various sizes of defect-free areas.
UR - http://www.scopus.com/inward/record.url?scp=85050344509&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85050344509&partnerID=8YFLogxK
M3 - Article
AN - SCOPUS:85050344509
SN - 0735-6161
VL - 49
SP - 396
EP - 406
JO - Wood and Fiber Science
JF - Wood and Fiber Science
IS - 4
ER -