Areal surface texture parameters on surface

Additive manufacturing (AM) techniques enable the manufacture of components with free-form geometries and complex internal and external features. X-ray computed tomography (CT) is increasingly being used to inspect internal features of AM parts. An advantage of the CT process, compared to optical and stylus instruments with limited acquisition slope angles, is the ability to reconstruct reentrant features (undercuts). Processing reentrant features provides an advantage in the computation of surface parameters. If the surface includes many reentrant features, their elimination can lead to a biased estimation of parameters related to the height or the area of the scale limited surface. A unified framework capable of handling free-from surfaces, with generic form surface, reentrant features and unevenly
spaced points, such as those from CT reconstruction, will be proposed. Standard software instruments employed for roughness parameter require evaluation of height data on a rectangular grid. This allows the computation of areal parameters based on discrete methods with good approximation, dependent upon the sample size. The reconstruction from CT volume to mesh allows performance of an adaptive meshing based on the maximum allowable distance between the implicit function (implicit surface defined by a constant grey value) and the final triangular mesh [1]. With irregular meshes it is not possible to perform the integral with the discrete approximation and a bias on the parameters computation can arise. In this paper an approach that approximates a generic mesh based on locally refined (LR) Bspline is proposed [2]. The approach can be applied to a generic form surface because the local stretching of the surface is taken into account. Mesh parameterisation enables to handle undercuts, each acquired point is described as a function of two abstract parameters. In this paper the proposed method will be compared with the discrete (ISO 25178-2 compliant [3]) method implemented in standard software packages [4]. Since filtering techniques based on a general mesh are not yet defined in the standard, the primary surfaces, the surface after removing the form, will be analysed. The areal parameters of a Rubert sample (casting plate 334, nominal Ra of 25 µm) measured with a focus variation (FV) instrument will be evaluated. Two form surfaces will be taken into account: plane and cylinder. Robustness of the discrete method will be finally evaluated with the mesh reconstructed from two CT measurements: the Rubert sample and an AM part.