Wood is heterogeneous and has a specific anatomical structure that is independent of any processing. Anatomical irregularities can distort the measuring results so that surface appears rougher than it really is.
The main aim of this thesis was to develop a set of recommendations that can provide reliable data measurements and roughness evaluation of solid wood surfaces after sanding.
The performed study compares two different types of measuring techniques: laser and stylus scanning. The results indicate that stylus scanning is the more effective method for research purposes. It is recommended that, the measuring direction should be across the grain, the measuring resolution should be 5 mm and the evaluation length should be at least 50 mm.
It was found that the shape of the primary profile “P” is affected by the presence of deep pores under a smoother plateau. A number of methods for excluding shape deviations were examined, but a method contained in ISO 3274: 1996, which employs a regression fit, was selected and applied within the present research.
The efficiency of a number of roughness standard filters was examined to see if they were suitable for wood. Filters in current standards introduced distortions, but a robust Gaussian regression filter (RGRF), contained in draft standard ISO/DTS 16610‑31: 2002(E), appeared to give roughness profiles free of distortions. An alternative method was marginally less accurate, but computationally less expensive. A more time‑efficient derivation of RGRF was implemented in C++. The processing roughness and anatomical roughness, contained in the filtered profile, were separated with an algorithm based on a threshold defined in the Abbst curve.
Further work based on the measuring and evaluation method suggested in this research could test values of the roughness parameters for different combinations of sanding variables, in order to predict output results for each processing set. Predicted roughness values can serve as references for automating the sanding process.