MrKopow1
8th July 2009, 08:19 AM
I am looking for the best method for determining the capability from a carburizing process. In particular I’m concern about hardness data. I understand that the data would produce a non standard distribution but how would you determine the Cpk and Ppk? :thanx:
brahmaiah
8th July 2009, 08:49 AM
I am looking for the best method for determining the capability from a carburizing process. In particular I’m concern about hardness data. I understand that the data would produce a non standard distribution but how would you determine the Cpk and Ppk? :thanx:
The data need not necessarily be non-standard. Cpk calculation may not be possible but Ppk calculation can be done in the normal way. If the carburizing process is under control Cpk caln.can also be done. But we have to see the nature of variation in the data. Here the method picking samples.and the location of samples in the furnace is critical.
V.J.Brahmaiah
MrKopow1
8th July 2009, 10:17 AM
The data need not necessarily be non-standard. Cpk calculation may not be possible but Ppk calculation can be done in the normal way. If the carburizing process is under control Cpk caln.can also be done. But we have to see the nature of variation in the data. Here the method picking samples.and the location of samples in the furnace is critical.
V.J.Brahmaiah
I appreaciate the information, however, how do you handle things like surface and core hardness? They will only get so hard and a normal distribution shows that the parts can achieve a higher than maximum hardness which cause problems with the calculations. Let me know.
Thank you
brahmaiah
9th July 2009, 02:10 AM
I appreaciate the information, however, how do you handle things like surface and core hardness? They will only get so hard and a normal distribution shows that the parts can achieve a higher than maximum hardness which cause problems with the calculations. Let me know.
Thank you
Since core hardness checking is distuctive test it will not fall into normal Cpk calculation as we need large number of readings. You may go for X-MR control chart which requires ploting at longer intervals normally one per batch.In any cse the key for your problem lies in proces validation and strict control process parameters.If the process is not validated to achieve required results it is not acceptable by the standard.H.T. being a special process 100% inspection is out of question.After validating your process you may use following X-MR control chart to asses the process capability.:
V.J.Brahmaiah