With complexity of today’s supply chains and complicated
product & manufacturing technology,
error is inevitable in final product.
Therefore inspection is required to verify the product along the
manufacturing processes before the
product reach the end consumer.
Inspection is the fastest way to
ensure only quality products which fulfill the customer requirement are
delivered to end customer.
Product verification through test and inspection process
are non value added from lean perspective as they do not transform the
product. The product will look the same
before and after test or inspection!!.
If possible we would like to minimize product verification process as
much as possible. Although quality
engineering is NOT about inspection (per my blog in Aug 2017), however establishing an effective inspection
process is an important undertaking in quality engineering.
Most of the time human judgement is still used even
though there are advancement in automated inspection. No matter how
sophisticated an automated inspection there is always some limitation and human
inspection is the most versatile although it is NOT the most reliable inspection
method. Therefore attribute gage
repeatability and reproducibility study conducted on human judgement visual
inspection tends to have a lower
repeatability and reproducibility value.
In most cases, after machine
inspection, a human verification is still used to verify the reject call by
automated equipment.
Before establishing any inspection process, we need to understand the following :-
- What do you want the inspection process to detect in your product. Normally items which is NOT detected by automated test or inspection should be targeted for visual inspection.
- How do you want to inspect your product (human judgement such as visual inspection or automated inspection equipment)
Once we are figure out the above 2 points, there we will need to follow a proper
protocol to set up an effective
inspection process. Effective
inspection process means that the process is able to make correct judgement
call base on operational definition created from customer feedback.
We must follow a proper protocol to set up an inspection
process which is effective in detecting any potential reject for both human and
automated inspection process.
Guideline to set up a proper manual/human visual
inspection process
- Study the process flow to determine where should we place inspection process in the production flow. We should work base on the principal that we would like to detect the reject as early as possible. If there is a complicated process which is prone to error, then we will put inspection process right after the process and not wait until outgoing inspection to screen the reject.
- Establish clear operational definition inspection criteria. We must clearly specify as detail as possible what should be rejected and accepted. If possible try to establish limit sample with marginal reject and accept criteria.
- Document operational definition and establish inspection sequence of the part inspection by area or by surface. Inspection can be done according to predetermine sequence per example figure below.
- Inspector must go through proper visual test to ensure their vision is good for the intended inspection.
- Train and certify operator according to the documented procedure to inspect the part.
- Conduct attribute GR&R study on operator to check for their repeatability and reproducibility (refer to my article on Importance of Performing Gage Repeatability and Reproducibility (GR&R) before actual quality data collection published on Sep 2017. Results Attribute GR&R will enable you to know which operator is not suitable for the job or need more traing or if inspection criteria or method is not clear.
- Ensure the lighting is bright enough at appropriate angle, above 100 Lux is recommended.
Do refer to my course for more information on setting up effective inspection system: Click on image to get link
In the second part of this article, I shall write about
the guideline to set up the automated inspection process.