Process capability

The workpiece of the molder was milled at four sides with four spindles. The width of the workpiece was measured at 5 points (Fig. 5), the thickness of each 5 points on the right and the left side.

Figure 5: Workpiece molder.

The measured values were shown in an original value chart and a histogram (Fig. 6). The 200 pieces were manufactured in two blocks a 100; an offset between the two blocks with 100 parts can be seen. Next an emergency stop occurred, this influence is partly to recognize as a minimal offset of one axis.

Figure 6: Original value chart and histogram of a machined board thickness.

Here, the capability indices were calculated according to the methods, which are described above. As expected, the capability indices calculated with the range, were mostly smaller than those of the factor of 6 , but not always.

The distributions were normally distributed under no circumstances, partly bimodal and skewed; the calculation by 6  should therefore not be used.

Calculated with the range capability indices were as expected always

smaller than those determined by the quantile (Fig. 7). However, the differences were small and added up to average 4.8% and maximal 18.6%

based on the correct calculation using the quantiles.

Figure 7: Capability indices for a machined board thickness.

For the characteristics of the workpieces produced by the machining center the results were similar. The evaluations were partly more complex. In addition to pure distance measures also squareness, parallelism and position tolerances of various holes were analyzed (Fig. 8). Positional

tolerances are not defined by DIN 68100. Alternatively a factory specification of the machine manufacturer was used, the tolerance range is smaller than HT 15. The distributions were rarely approximately normally distributed, mainly towards trapezoidal distribution shaped and sometimes skewed. The validity of the capability indices calculated with a factor of 6  is therefore low and the values were both larger as well as smaller than the values obtained with the two other methods. With the exception of one value the capability indices calculated using the range were also smaller than that obtained with the quantile (Fig. 9). The differences here were in the average 4.0% and a maximum of 16.9% based on the correct calculation using the quantile.

Figure 8: Workpiece machining center.

In both cases of the capability analysis no outlier tests were made, even if they had been authorized in part by the current standards. The focus of the investigations was clearly on the suitability of capability tests for machine acceptance procedures. For no machine buyer, it would be understandable in this context that single measured values of produced workpieces are not statistically taken into account.

Figure 9: Capability indices for a position of a serial bore hole.

4 CONCLUSIONS

The limits of the capability indices are normally established in steps of a third. This is because of the correlation to 6  or 3 . Usually a value of 1.33 for normal and 1.67 for most relevant features are fixed in contracts as limits for process capability indices. Values for short-term skills (machine capabilities Cm, Cmk) can be raised to one level to have a reserve during the final inspection. Higher values are rarely sought, since the cost of the machine and the process would rise. The resolution required for the capability indices is therefore regarded as relatively low. As already explained in the last conference, the RV-factor or its inverse gives significant advantages over other methods in its suitability for attributive characteristics.

The measurement capability of sensory tests, despite methodological procedures, special equipment and intensive training (see VDI 3414-2 [9]), cannot be significantly improved so that requirements such as the ISO 22514 would met. From a lot of 50 workpieces, the search out each best and worst and assign a grade on a scale with respect to tolerances is possible.

For 200 parts the evidence should be shown. However, the effort and costs would probably increase significantly or it would lead to an overwhelming of the auditors.

Statistically, the difference in the calculation using the range or the quantiles is given and it turns out - as evidenced by the tests - in practice. It is depending on the form of distribution but it often has a low dependence. This raises the legitimate question whether the acceptance procedures of

woodworking machines analogous to VDMA 8669 using the RV-factor or its inverse should be raised to a standard. Because of the high importance of the attributive, only sensory testable characteristics would allow an equal treatment for measuring and testing. Their determination and of the indexes itself would be simple and comprehensible but the use of them in capability analysis determined parameters in a subsequent SPC is currently rather rare. Because of possible outliers and other reasons, the method may not be 100% mathematically correct but practical.

REFERENCES

[1] ISO 11462-2 (2010) Guidelines for implementation of statistical process control (SPC) – Part 2: Catalogue of tools and techniques.

[2] ISO 22514 (2014) Statistical methods in process management – capability and performance. part 1. part 8.

[3] VDMA 8669 – VDMA-Einheitsblatt (1999) Fähigkeitsuntersuchungen zur Abnahme spanender Werkzeugmaschinen. Verband Deutscher Maschinen- und Anlagenbau e. V. (VDMA).

[4] VDA (ed.) (2010) Qualitätsmanagement in der Automobilindustrie 5 – Prüfprozesseignung … 2. Auflage. Verband der Automobilindustrie:

Berlin.

[5] Hanrath, G. (1997) Fähigkeitsuntersuchungen an spanenden Werkzeugmaschinen. Diss. RWTH Aachen.

[6] Kortüm, C., Riegel, A., Solbrig, K. (2015) Process Qualification in the Wood Industry. In: Villmer, F.-J., Padoano, E. (Editors): Proceeding of the 5^th International Conference, Production Engineering and Management. Trieste.

[7] Dekomien, K., Huxol, A., Schulz, S., Riegel, A. (2014) Learning from geography – Topography as a basis for quality assessment of high gloss surfaces, In: Villmer, F.-J., Padoano, E. (Editors): Proceeding of the 4^th International Conference, Production Engineering and Management. Lemgo.

[8] Robert Bosch GmbH (ed.) (2003) Schriftenreihe Qualitätssicherung in der Bosch-Gruppe Heft 10 – Fähigkeit von Mess- und Prüfprozessen.

Stuttgart.

[9] VDI 3414 (2014) Blatt 2: Beurteilung von Holz- und Holzwerkstoffoberflächen – Prüf- und Messmethoden. s.o.

[10] Mindlin, A. (2016) Machbarkeitsstudie für eine Prozess-fähigkeitsanalyse nach DIN ISO 22514-2 für ein Bearbeitungszentrum.

bachelor thesis HS-OWL.

[11] Schniedermann, F. (2013) Beitrag zu Prozessfähigkeitsanalysen in der Holzindustrie. bachelor thesis HS-OWL.

DRAFT DYNAMIC SEAT WITH A USE OF UNCONVENTIONAL