This page is a draft intended to collect thoughts on the topic of the advantages and disadvantages of 100% screening inspection.
What is the goal?
To inspect and screen out nonconforming product?
To monitor the process?
To study and improve the process?
Is the process not validated?
High risk to customer?
Test is non-destructive?
Test is not too costly or does not consume too much time?
100% testing is possible from the process?
Then 100% screening might be appropriate.
Does the output of the process allow verification, to include either sampling or 100% screening?
Classic example: processes resulting in sterile product (the sterilization process itself, or the packaging process resulting in a sterile seal).
Characterisitics of rolls of film - not feasible to measure the characteristics of the entire roll once the roll of film (laminate, etc.) is formed. Would need to do 100% screening/monitoring during the process.
One recent presentation suggested that the FDA defines (ongoing) process verification as only meaning 100% screening, and does not include lot-based sampling acceptance plans.
This position warrants further exploration.
Does the process result in discrete components that can each be individually inspected? Or is the process output continuous, such as extrusion?
Which products and characteristics would be submitted for 100% inspection? What is the decision criteria?
Juran, 5th Edition, page 23.13
"It is generally conceded that no screening operation is 100% effective and, in the real world 100% inspection of a large number of units may be only about 80% effective."
Therefore, a lot-based sampling plan with a rectification scheme can provide more protection than does 100% screening.
Ott, Schilling, Neubauer, page 153
"[...] 100 percent screening inspection of large lots does not ensure 100 percent accuracy. The inspection may fail to reject some nonconforming units and/or reject some conforming units. Fatigue, boredom, distraction, inadequate lighting, test equipment variation, and many other factors introduce substantial errors into the screening inspection."
If we need to rely on a plan based on 100% screening, and that does not use a validated system for automated detection, then the Interrater Agreement method can be used to evaluate the effectiveness of a human-based, subjective method of screening.
A lot-based sampling scheme can be used as a check on the 100% screening inspection.
A sampling plan can be used to monitor a non-automated 100% screening inspection process.
100% screening might be more effective when the screening is done by an automated (and validated) system, such as a vision system. However, a sampling plan can still be used to monitor the automated 100% screening system.
If the reason for 100% inspection was to study and improve the process, then there may be some benefit.
If we want to 'set a baseline' for a measured characteristic, then SPC methods are probably more efficient.
If we want to establish a baseline for percent nonconforming, then attribute sampling plans (with p-charting on the results) might be a more attractive option.
Time-frame for the use of a 100% screening inspection method
Will it be used only as a temporary measure until the process can be improved and brought into a state of control?
Will it be used on a long-term basis?
What is driving the need for a 100% screening inspection plan?
Is it a legal or regulatory requirement?
Is it meant to mitigate a high level of potential risk to the customer or the company?
"The purpose for inspection is to [improve the process and] eliminate the need for inspection."
~Deming (taken from Stephens, p. 20)