Profile measurements of the surface topography of engineering components are ofren used in attempts to assess the functional suitability of surfaces. This paper reviews the dejinitions and properties of a large number of parameters which are used to characterize measurements of this type.
CHARACTERIZATIONMethods that characterize surface roughness ideally need to embody an assessment of the height (amplitude) characteristics and the spatial (wavelength) characteristics of the topography of a specimen to describe it completely.Parameters that are used to characterize measurements of surface topography are defined almost entirely in terms of the profile signal generated by stylus-type measuring instruments. This arises for historical reasons. For several decades stylus instruments have provided the only reliable means of obtaining a qualitative record of the shape of a specimen surface. In itself this has influenced the development of parameters used to assess surface topography. Until recently any method adopted as a means of parameterizing the transducer signal needed to be relatively simple and based on mathematical expressions that could easily be evaluated by means of analogue circuits. Several surface pararneters evolved within this constraint and they were mainly related to the statistical properties of the amplitude distribution of an arbitrarily filtered version of the transducer output. Their suitability for assessing surface topography was limited although some remain in common use in the manufacturing environment (for example RJ. The advent of digital electronics has removed most computational restrictions so that parameters of virtually unlimited complexity can now be defined and evaluated for surface data. The availability and flexibility of this new computational power beckons investigations which develop more useful and relevant ways of characterizing surface topography. This paper reviews some methods used to characterize surface topography in both industry and research. The inherent advantages and weaknesses of these techniques are discussed.The length of a profile determines a long-wavelength limit for the features which may be observed within it. If the profiles measured from the surface of a specimen with a periodic structure are not sufficiently long toThe MS was received on 30 October 1986 and was accepted for publication on 2 March 1987. admit all the periodic content, parameters assessed using profiles recorded from different sites on the specimen will be inconsistent. Parameters assessed from profiles of differing lengths will also yield variable results. Even more intractable difficulties arise if the specimen surface contains an appreciable random component in its structure. Sayles and Thomas (1) have identified a non-stationarity which arises in manufactured surfaces that have a random structure. They find that the statistical properties of a profile depend on its length and that its power spectral density function, G(w), is described by the relationship 27cK G(w) = -wz where w = (2n/wa...
