Galling wear severity quantification
5.6 Case study: galling progression
5.6.4 Galling progression case study discussion
Again, the presented DWT wear severity measurement methodology provides a measure of galling wear severity that maintains the accuracy of visual assessment, in this case study for extensive galling wear progression over thousands of parts. Based on correlation with visual assessment rankings the Ra surface roughness parameter was found to give the most accurate measure of galling wear severity in this case study. As was discussed in Section 5.6.3.2, this result for the Ra parameter was due to the significant difference in scale between galling wear and the surrounding base surface roughness. Other standard surface roughness parameters were also found to be noticeably more accurate for the galling initiation case study (Section 5.5.4.2). In addition to the difference in scale of the galling wear features it
99
is also possible that the number of parts assessed could have influenced the observed improvement in accuracy in these surface roughness parameters. Approximately 1 in 150 parts were assessed due to the large number of parts formed in the trial, as opposed to every part in the small scale galling initiation case study (Section 5.5). As a result, in this case study the wear damage could progress significantly between sampled parts. Spearmanβs rank Correlation is used to assess
the accuracy of the parameters, and distinct differences in the quantity and severity of wear damage between sampled parts could help to clarify the order of relative severity and hence give stronger correlation for all tested parameters. More frequent sampling of formed parts could possibly capture the progression of wear damage in finer detail, which would highlight the strengths of the DWT galling wear methodology over other standard surface roughness parameters, as demonstrated in section 5.5.
Both the Daubechies 2 wavelet and Haar wavelet derived ππ·ππ parameters have a
very strong correlation with the visual assessment ranking. The MDL criterion suggested the Haar wavelet as the βbestβ wavelet function for describing the isolated
2D profile wear sections (Section 5.4.2). There is a noticeable difference in strength of correlation between the visually selected db2 wavelet results and the MDL selected Haar wavelet. However, Figure 5.18 showed that the Haar wavelet detail coefficients had a more significant response for the smaller scale damage features than the db2 detail coefficients, for example at x = 4 mm (Figure 5.18b) and x = 14 mm (Figure 5.18d).
The square-wave shape of the Haar wavelet meant it was overlooked based on visual comparison to 2D profiles of galling wear features. However, its shape does make the Haar suitable for edge detection in other applications. Depending on the scale of the wear damage, Haar wavelets may approximate the distinct drop from the gouge shoulder to below the reference surface seen in the 2D cross-section of galling wear damage. This makes the Haar wavelet suitable for the detection of small scale wear damage features in the aggregate 2D surface profiles.
The 2D roughness parameter Rdq had a strong correlation with the visual assessment
rankings. Rdq was the best performing standard 2D surface roughness parameter
across both case studies (sections 5.5 and 5.6). Rdq gives a measure of the variability
100
the distinct features of galling wear damage on the part surfaces. GSI showed a moderate correlation with the visual assessment rankings, an improvement upon what was observed in section 5.5, but still not as effective as the ππ·ππ parameter or
even some of the 2D roughness parameters.
As discussed in section 5.5.5, it should be noted that the DWT detail coefficients capture important information about the location and severity of the wear features on the part sidewalls, and the ππ·ππ parameter reduces this information into a form
that is suitable for statistical analysis. However, by reducing the detail coefficients important information is lost, and so, for a comprehensive assessment of wear the
ππ·ππ parameter should be considered in conjunction with the raw detail
coefficients.
Galling wear progression can occur over the course of individual strokes, resulting in varying levels of galling wear severity on a single part. This was noticeable, in particular, for parts later in the forming trial, with more severe galling adhesive damage occurring late in the stroke, outside of the measurement and visual assessment region. The presented methodology is based on 2D profilometry, and so the resultant measure is dependent upon the surface profile location. Selecting a measurement location that corresponds to later in the stroke or taking multiple measurements will help to ensure an accurate measure of the wear severity is captured.
5.7Summary
A new methodology and parameter for measuring the severity of galling wear has been presented that utilises DWT to target the distinct features of galling wear damage. The DWT methodology provides a targeted, repeatable and non-subjective measure of galling wear severity from the formed part that reflects visual assessment of part damage in SMF. This was developed to address the requirement for such a measure of galling severity in tracking galling wear development in sheet metal forming operations.
The DWT methodology has been applied in two case studies. In the first case study, the method was applied to a galling initiation trial (Section 5.5). In the second case
101
study, the method was applied to a long galling progression trial (Section 5.6). In each of these case studies the DWT methodology was compared to a number of other parameters that have been used to quantify wear. Of the parameters tested, the presented optimised ππ·ππ values in sections 5.5 and 5.6 were found to have a very
strong correlation with visual assessment rankings, and the highest correlation of all the tested measures across both case studies, see Table 5.5 and Table 5.9. In the section 5.6 where part sampling was infrequent, the standard roughness parameter Ra appeared to perform well for quantifying wear severity, having a very
strong correlation with visual assessment rankings. However, this was not the case for section 5.5 where every part was collected and measured.
Of the existing wear measures assessed, Rdq was found to have a strong correlation
with visual assessment rankings, and this performance was consistent for sections 5.5 and 5.6. Rdq was only outperformed by ππ·ππ, and that had a consistent very
strong correlation for sections 5.5 and 5.6. GSI, the only other targeted galling wear measure, did not perform particularly well in either of the case study situations. The DWT methodology was found to be appropriate for SMF tests that focused on galling wear in industrial style conditions. The method allows the quantification of multiple localised galling wear features over large surface areas, as opposed to fixed single tracks that are often studied in experimental wear trials. Therefore, this method will be ideal for quantification of galling wear severity in the industrial style channel forming trials, like the trial described in section 5.6, and will be used for testing galling detection and monitoring methodologies. Such a methodology will be presented in the subsequent chapters, and the DWT wear measurement methodology will be utilised to track the quantity and severity of galling wear throughout those trials.
102