Pro Ligno (Sep 2015)
REPLACING OUTLYING WOOD ANATOMY IN THE EVALUATION OF PROCESSING ROUGHNESS DATA AT SANDING
Abstract
Anatomical irregularities should be removed from the evaluation if a reliable processing roughness is to be evaluated. In order to get only measures of processing, wood anatomy can be removed with a method based on the Abbot-curve, which separates the core data from outliers by means of an upper and a lower threshold. Although researchers agreed on the need of separating the processing roughness from anatomical roughness, there was no study on the most appropriate method of replacing the missing data in the roughness profiles. This paper examined three methods of replacing the oulying data and their effect on roughness parameters calculated on sampling lengths, as instructed by ISO 4287 (1998) and on evaluation length. The Zero method replaced outliers with zero values disregarded in further calculations, the Predicted method replaced the missing data by cubic spline interpolation and Total removal method removed completely wood anatomical features up to the profile mean line.The results showed that Zero method is the best choice when roughness parameters are calculated on the evaluation length. Compared to Predicted method it has the advantage of using real data giving similar results.Total removal method dramatically reduces the number of profile valleys in the evaluation biasing the roughness parameters.
