I have a process for qualifying raw materials prior to manufacturing of a product using those raw materials. Does the qualification of the raw materials need to be validated? or can I just validate the large scale production process which incorporates those raw materials that are pre-qualified?
Thank you for your help!
I'm usually on the design side of things so take this with a grain of salt. The risk of only testing at the end of the process is that you have already sunk costs into the product. My first thought is that this should be evaluated against the probability of producing a good part with the “pre-qualified” material and the cost to validate the material prior to manufacturing.
The probability of producing a good part would depend on the accuracy of the qualification data (measurement accuracy, number of samples, etc.) and the actual property requirements to produce a good part. (ie is there a lot of margin?) Costs would include holding material in stock, validation testing, labor, material lost during test process and the cost of rejects. Depending on the material and property degradation while being held in stock could also come into play.
The answer, of course, is it depends. It depends on what industry you're in first and foremost. Beyond that, there are a number of questions that should be incorporated into a risk assessment, such as:
- Has the vendor validated their own process?
- What kind of QMS does the vendor have?
- Is their process established and stable?
- Do they have deep technical literature available?
- What quality certs do they have?
- Have you done an on-site audit?
- What's their reputation in the industry?
- What claims do they make about their products and how well are they documented?
All we need to see Risk associated with the process, in other words how the far raw material qualification process is critical to Quality (CTQ)
If the process associated with low risk you can proceed with large scale production process
otherwise its wise to validate the raw material to avoid the sunk cost .