Smart-labels that automatically monitor lapsed time - Sample and Testing Plans, Help!

[shushd]

Hi everybody, I'm new to this website, and must say I am learning alot. I started working in a new factory (2-3 years old), which manufactures smart-labels that automatically monitor lapsed time, ranging from under 1 day to 12 months in variable conditions, the label allowing a tinted liquid to migrate through a micro-porous material at a consistent rate. It actually replaces the Effective Date, Due date and so on.
Any way, the company is small but growing rapidly, I started working about a month ago, and my job is to implement QMS (from scratch...). right now we are trying to figure out how many samples do we need to take for inspection, the problem is that the inspection and analyzing consumes a lot of time, we are affixing a number of labels (from the beginning, middle and end of production process) onto a paper, and scan. then, depending on the product we need to scan again and again, in order to see the if the liquid behaves like it should, and compare the results from the different labels. So, you can imagine how much work needs to go in to this. We have a lot of production problems because the whole process is not stable yet, and we need to check the products all the time. I would appriciete any kind of help, new methods for testing, sampling plans, any information relevant will be much appricieted.

 

[True Position]

I do not know how to help you, generally acceptable quantities for sampling is something agreed upon with your customers, I'm sure the more knowledgable members of this forum will offer what guidance they can.

Now to my real comment, how many people reading this post thought it was a sales pitch until he got to: 'Any way, the company is small but growing rapidly, I started working about a month ago, and my job is to implement QMS '.?

 

[shushd]

I am sorry if it came out like that, that was not my intention at all. I just wanted to give a little background, so you guys understand more about what I am talking about, because each company is different, (and what we do is quite unique), and a lot of our actions depend on the type of product we manufacture, don't you think?

 

[True Position]

No no, at the halfway point it becomes perfectly clear, but my brain was looking for the report button for a second. Just teasing a little bit.

 

[Tim Folkerts]

It sounds like you have several challenges to face; several different jobs to do.

• Research & Development: the process isn't stable, so research is needed to determine what factors in the manufacturing process might improve reliability and consistency of the final product.
• Metrology: you need to accurately and consistently measure slow movements of an indistinct boundary of colored liquid.
• Control: You need to be able to spot and remove bad products before they are shipped.
• Training: once you know the ways you want things done, you need to train people in the proper techniques.
• Management: you need to get these systems in place and formalized so that they are done consistenly.

These are all jobs that are within the scope of "quality". Where to place your emphasis must depend on your understanding of the severity of the problems and your own strengths. You will almost certainly need support from a variety of people to accomplish all these goals.

For the specific problem of how many samples to take, that will depend on the ability of your tests to measure the motion (Gage R&R). If there is poor resolution in the measurements, you might need to take extra samples and average in order to reduce random variations. Are you using image-processing software to analyze the scans? There are various software applications that could semi-automatically look for boundaries, measure areas, etc.

The number of samples would also depend on just how good or bad the current production is. If you are trying to spot whether a batch has 0.1% defective vs. 100% defective, then just a few samples from each batch should spot the problem. If you are trying to spot whether a batch has
1% defective vs 5% defective, then you will need a larger sample.

Hope that gives you something useful to think about.


Tim F

 

[shushd]

Thank you Tim. and yes we do have a software that analyzes the scans. I just don't know how many samples I need in order to check a batch in %. the problem is that we scan the same samples several times, depending on the length of time the label should measure, for example: a hour label, should be measured at least 4 times during those hours from activation point, that means 4Xscanning, a 6 months label requires more testing. We also take samples from 3 points of production, that is in the beginning, middle and end. so you see, it takes a lot of time. Do you think it is necessary? Can you explain some more about how to determine the % of samples or maybe refer to a reading material on the net that might help. I realize I have loads of work implementing QA in the organization, but it is a huge challenge for me, and I though the production will be a good place to start, because it is the most important factor, and I don’t want to release any defect products. It is the name of the company on the line, and now it is my name as QA manager.

 

[Wes Bucey]

Thank you Tim. and yes we do have a software that analyzes the scans. I just don't know how many samples I need in order to check a batch in %. the problem is that we scan the same samples several times, depending on the length of time the label should measure, for example: a hour label, should be measured at least 4 times during those hours from activation point, that means 4Xscanning, a 6 months label requires more testing. We also take samples from 3 points of production, that is in the beginning, middle and end. so you see, it takes a lot of time. Do you think it is necessary? Can you explain some more about how to determine the % of samples or maybe refer to a reading material on the net that might help. I realize I have loads of work implementing QA in the organization, but it is a huge challenge for me, and I though the production will be a good place to start, because it is the most important factor, and I don’t want to release any defect products. It is the name of the company on the line, and now it is my name as QA manager.

The product is interesting. What triggers the "start countdown" point? When it is applied to substrate?

If the label doesn't count down until activation, there are several methodologies you can use to generate sample pools for analysis. I'll include some links at the end of this post.

If the act of production starts the countdown (unlikely, I'm sure), you have a tough row to hoe.

I am referring you to a free manual for statistics provided by the United States National Institute of Standards & Technology (NIST)
http://www.itl.nist.gov/div898/handbook/index.htm
Specifically look at Chapter 6 "Monitor" here http://www.itl.nist.gov/div898/handbook/pmc/pmc.htm
which discusses sampling plans and process control techniques, including

• Histograms
• Check Sheets
• Pareto Charts
• Cause and Effect Diagrams
• Defect Concentration Diagrams
• Scatter Diagrams
• Control Charts

 

[favqty]

SHUSHD,
You are when you think there is a lot of work to do. You are a new guy in town dealing with a "unique" sort of product. I whish you have patient wnough and the courage and ability to deal with people and process.

The manual that Mr. Bucey link to you will be such a great tool. It will give you the basis to back up the methodology you must to define. You also right when you say you must focus on production first and then there will be other priorities. I want to give you a recommendation to start.
1.- Identify the variables that affect the product performance (Raw materials, enviorement, machine, people, methods). It is pretty much an Ichicagua exercise. It will help you to understand which is more critical.
2.- Run PFMEA to figure out failure modes and action to prevent and detect. Because your process is already running it can help you to control the current modes which are the most critical now. Whenever you controol it there will appear some news.
3.- Define the machine variables, people rutine that are related with the current failure modes.
4.- Define the actions to keep this variables into control and make/deploy the control plan (Train people).
5.- Once you have this done will be easier to keep the process a little bit more consistent and you will have the time to see the machine running and try some DOE's and even better you will have operator's support to make proper test runnings and get back to your control settings (Firts reference).
6.- Doing this you can make your first process assestment and get your process capability plot to get into the process control stage.
7.- Now you can take a product sample (1000 consecutive samples for example) and make product evaluation to understand your current defective level. You can make the PARETO to see which are the most critical defects (Failure modes) and set actions to control it and update your PFMEA.
8.- Everytime you control one failure mode you must to deploy the process key control characteristics through you operational crew. Education is the most important task when you want to control production processes.
9.- Whenever you control your three most critical failure modes you can apply any method to set you sampling plan to keep and eye in your product and evaluate the effectiveness of your current process control plan to keep the variation within specs.

I do not want to give you a solid rock formula to control your process. I just want to share my thoughts with you hoping that any idea given works for you. Enjoy your challenge and have a nice end of it.

P.D. Excuse my English.

 

[shushd]

Thank you both very much, the information is very helpful, I have a lot of reading to do, if I have anymore questions, I'll let u know.