C
C. Oertli
this is my first time in a forum:
The purpose of this post is really to determine if anybody is intersted to discuss the particularities of the chemical manufacturing industry concerning application of MSA and if this post is conidered OT, to obtain some guidance or redirection.
We are a chemical bach processing business supplying high tech resins to various industries, including automotive. Through this relationship we have been introduced to MSA by customer demand. The merits this system brings to the manufacturing industry are obvious consequentially its application to a processing environment is tempting for guidance in the design of process improvement projects. For about a year I have been trying to adapt MSA principles to our industry. While some sucesses have already been shown, some significant differences in the application of statistical principles between the two industry branches remain unresolved. In the interest of brevity I will be synoptic.
Terminology: The translation of manufacturing terms into processing terms is ambiguous but necessary to design experiments. I found little guidance in the MSA literature. (For example I declared a batch as a 'part' after establishing that the batch is homogenous. The advantage is that destructive tests can be replicated by repetitive sampling).
Process Stability/statistical control: is much lower than manufacturing. Batch operations with short production campaigns (as low as three batches/campaign) where start-up transients do not provide a statisitcal stability comparable to the manunfacturing industry. Economy and ecology prohibits to discard the start-up 'parts' (up to $60,000 per batch) until process stability is reached. Process capabilities are much lower than customary in the maufacturing industry and hence the unquestioned adoption of such criteria may not be prudent. Consequentially it is customary to test all batches (parts) rather than a sample. This is effective for Quality Control to prenvent shipment of non-conforming material but process stability as required to predict process behaviors is limited.
Cost of chemical tetsting is high and significantly more complex than physical testing. Many tests are applications tests rather than by chemical analysis. Gage variation is comparably high.
The above is only a small sample of the differences I see with consequences in the interpretation of the output from a MSA/GageR&R study. As a somewhat critical scientist, I am reluctant to purchase a GageR&R stat pack, 'plug in the numbers' and use the output unquestioned in the justificaiotn of six figure improvement projects. For the time being I enjoy the support of management in the introduction of MSA principles in our operation, but is appears that a rational review of the statistical approach taken is in order.
Comments welcome and thanks in advance!
CO
The purpose of this post is really to determine if anybody is intersted to discuss the particularities of the chemical manufacturing industry concerning application of MSA and if this post is conidered OT, to obtain some guidance or redirection.
We are a chemical bach processing business supplying high tech resins to various industries, including automotive. Through this relationship we have been introduced to MSA by customer demand. The merits this system brings to the manufacturing industry are obvious consequentially its application to a processing environment is tempting for guidance in the design of process improvement projects. For about a year I have been trying to adapt MSA principles to our industry. While some sucesses have already been shown, some significant differences in the application of statistical principles between the two industry branches remain unresolved. In the interest of brevity I will be synoptic.
Terminology: The translation of manufacturing terms into processing terms is ambiguous but necessary to design experiments. I found little guidance in the MSA literature. (For example I declared a batch as a 'part' after establishing that the batch is homogenous. The advantage is that destructive tests can be replicated by repetitive sampling).
Process Stability/statistical control: is much lower than manufacturing. Batch operations with short production campaigns (as low as three batches/campaign) where start-up transients do not provide a statisitcal stability comparable to the manunfacturing industry. Economy and ecology prohibits to discard the start-up 'parts' (up to $60,000 per batch) until process stability is reached. Process capabilities are much lower than customary in the maufacturing industry and hence the unquestioned adoption of such criteria may not be prudent. Consequentially it is customary to test all batches (parts) rather than a sample. This is effective for Quality Control to prenvent shipment of non-conforming material but process stability as required to predict process behaviors is limited.
Cost of chemical tetsting is high and significantly more complex than physical testing. Many tests are applications tests rather than by chemical analysis. Gage variation is comparably high.
The above is only a small sample of the differences I see with consequences in the interpretation of the output from a MSA/GageR&R study. As a somewhat critical scientist, I am reluctant to purchase a GageR&R stat pack, 'plug in the numbers' and use the output unquestioned in the justificaiotn of six figure improvement projects. For the time being I enjoy the support of management in the introduction of MSA principles in our operation, but is appears that a rational review of the statistical approach taken is in order.
Comments welcome and thanks in advance!
CO
