Extent of Validation - New machine installed in manufacturing

jkc3usc

Involved In Discussions
We have a new machine installed in manufacturing and paper work has been signed off for installation. If we check 1st piece as well as in process inspection to the part print do I still need to run a capability study? Again the risk for this part is low and critical dimensions are checked in 1st piece. I understand the more data to show capability is great (I also now the machines specs) but the machine is needed to run for parts production.
 
Last edited:

William55401

Quite Involved in Discussions
What is the machine? What is the process? It sounds like you have IQ completed. Without more information, it is uncertain what additional steps (OQ, PQ for example) are expected. Depending on the process, IQ only is often just fine.
 

yodon

Leader
Super Moderator
The regulation / standard requires that any process (i.e., whatever your machine does) whose outputs are not (100%) verified needs to be validated. The means for validation are not prescribed. Some aspects related to validation are, indeed, listed in the standard. If you read through those, have the required documentation demonstrating compliance, and believe you're in a defensible position, then maybe you've fulfilled the requirement. As @William55401 notes, without more details, it would not be possible to give further advice.

Do also note that your internal procedures should be driving this! To have a QMS compliant with 13485, you should have procedures for infrastructure qualification, equipment qualification, and process validation.
 

jkc3usc

Involved In Discussions
Understand. So We plan to run a 30 piece capability study. What about timing? The OQ,PQ can be done later? The machine is installed, employees are trained, Parts are being machined and checked as we do on all other machines. There is no safety risk for employees and parts go through inspection and a final inspection when product is built.
 

Mikey324

Quite Involved in Discussions
I would say 30 piece capability study. If you run any of the same parts on this machine as you do/did on your older ones you could compare results. Just to see if you are producing equivalent quality parts. Just depends on what all this new machine does.
 

yodon

Leader
Super Moderator
Statistical rationale is emphasized in the standard and beibg assessed in reviews. Be sure '30' is statistically based and not just a seemingly good number
 

Ronen E

Problem Solver
Moderator
Statistical rationale is emphasized in the standard and beibg assessed in reviews. Be sure '30' is statistically based and not just a seemingly good number
There is no one-size statistical basis available.
If we are being completely honest, most reviewers lack the background, motivation and time necessary for true statistical rigour. It's just too complicated; too many factors need to be considered for such determinations to be truly rational.
Pseudo-statistics are much more accessible and look just as nice in most cases, though.
 

yodon

Leader
Super Moderator
I don't disagree about the statistical rigor and the ability to pass off pseudo-statistics. Reviewers, at least from my experience are not accepting the seemingly-large-enough sample size (or the rationale that 3 runs is sufficient, etc.). They may not be able to argue whether a statistical approach has legs but the do want to see it stood up, so to speak (some rationale for the decision).
 

Ronen E

Problem Solver
Moderator
Yes, that's exactly what I mean - it has to look "serious", not necessary be truly statistically defendable (which is quite difficult if you apply actual proper statistics and mathematics).
 
Top Bottom