Residual Stress Evaluation in Vitreous Enameled Steel Sheets by Digital Images Analysis of Microstructures - Visual Computing: Scientific Visualization and Imaging Systems - page 172

Table 9.4 Material properties

BRT a Material a E m

(10 -5 C -1 ) (C) (MPa)
Low carbon steel 1.35 - 180,000 0.3
Wet enamel 1.18 400 70,000 0.27
Dry enamel 1.23 400 70,000 0.27
a The Break Rate Temperature (BRT), as mentioned in the Sect. 9.2 , is the temperature below
which, during the cooling phase, the residual stress starts developing

9.4 Methodology

The analysis of the residual stress was performed by means of the digital images of
cross-section micrographs of the material. These images highlight the local
characteristics of the material such as the bubbles and the graded interface between
the two constituents.
In particular, the presence of bubbles in the coating is mostly evidenced, as can
be observed in Fig. 9.7 , where one of the analyzed images is reported.
The procedure followed for image analysis can be as summarized. A median
filter (3 9 3) is first applied to the image [ 14 , 15 ]. The next step consists in
thresholding the image, so as to expose the geometries of interest. The threshold
value is set with an iterative procedure based on the histogram image distribution.
Morphological operations are then applied to the binarized image. In particular,
closing was employed in order to fill in small holes, while opening was used for
clustered pores.
The binary image obtained highlights the bubbles in the coating and the graded
layer between the coating and the substrate as dark objects on a light background,
as evidenced in Fig. 9.8 .
The discretization of the coating requires the identification of the borders of the
objects in the binary image. To highlight the sharp changes in intensity, an edge
detection filter is applied to the image. For this task, a Sobel operator is used, based
on convolving the image with a small, separable, and integer value filter in hor-
izontal and vertical direction. The edge detection allows to produce an image
where, in particular, the geometry and location of the characteristics of interest is
well identified.
As known, the binary interface image uses the unit gray level to represent the
interface pixels. The closed interfaces are a cluster of the pixels with the unit gray
level. In order to carry out the mechanical analysis, finite element meshes must be
generated for the material distribution and geometry shown in the refined binary
image. The finite element meshes cannot be generated directly from the binary
interface image. The discrete interface pixels have to be transformed into their
geometry vector data format. This conversion is the preliminary procedure for
automatic generation of the finite element meshes for the material.