Eliminating a Recurring Problem of Missing Weld Nuts/Fasteners

[AgnieszkaSz]

I am having a very similar problem, only that the missing components are marking sleeves and therefore cannot be counted by weighing. We started with brainstorming sessions with operators - they named several root cause factors and proposed several ways of preventing them. Now we decided for batching AND one-piece-flow; it means, if you have, say, a production order for 50 products, you prepare 50 sets of components (NOT boxes with 50 pieces of each component); in this way, the possible mistakes are detected after first produced piece. We will see how it works. We also designated one persons as "fast agent", resolving problems of missing or illegible markers in the perpared sets. FAI (First Article Inspection) is a routine.
By the way, check the work instructions - whether people read them, understand and are willing to refer to them at all; instructions are often treated as kind of decoration, particularly when the pay system is per piece and people want to produce as fast as possible in order to earn as much as possible. I estimate that the latter factor causes ca 60% of all mistakes we make.

 

[DrM2u]

Re: Missing Weld Nuts/Fasteners

We are a contract manufacturer, we weld on several different weld nuts and fasteners per customer specs, however, we seem to have a re-occuring issue with missing weld nuts/fasteners. Some parts have several weld nuts/studs on them, some only one, obviously the parts with only one feature needing welding are never missed, however sometimes with multiple weld nuts/fasteners we struggle. Any additional advice would be appreciated.

Thank you.

You have to look at the corrective actions from both prevention and detection perspectives. The first question to answer is whether is acceptable to have missing nuts within the plant but not to reach the customer or not acceptable to have missing nuts alltogether within the process. The first option could be easily address by looking at 100% inspection of the completed part (final inspection) using a detection device, something as simple as a fixture with GO/NOGO features or similar. Some prevention steps couls include counting the number of nuts, re-organizing the work centters, etc. The second option implies minimizing or reducing the human factor from the process, which you already said it was not feasible.

Again, these are just some potential solutions. What you need to do though is a root cause analysis to identify where, how and why you are losing nuts. Focus on the system and look at the human factor as part of the system. 'Lack of training' and 'employee error' are realy lame causes and lead to ineffective solutions (a.k.a 'band aids'). Only after you identify the root cause you'll be able to select the proper and effective action(s).

 

[DrM2u]

I am having a very similar problem, only that the missing components are marking sleeves and therefore cannot be counted by weighing. We started with brainstorming sessions with operators - they named several root cause factors and proposed several ways of preventing them. Now we decided for batching AND one-piece-flow; it means, if you have, say, a production order for 50 products, you prepare 50 sets of components (NOT boxes with 50 pieces of each component); in this way, the possible mistakes are detected after first produced piece. We will see how it works. We also designated one persons as "fast agent", resolving problems of missing or illegible markers in the perpared sets. FAI (First Article Inspection) is a routine.
By the way, check the work instructions - whether people read them, understand and are willing to refer to them at all; instructions are often treated as kind of decoration, particularly when the pay system is per piece and people want to produce as fast as possible in order to earn as much as possible. I estimate that the latter factor causes ca 60% of all mistakes we make.

I'd suggest that you look at other processes and see if you have similar problems. If you do, then your root cause is much deeper that just the process you are refering to. I came across organizations where there was a general lack of process planning activities caused by the executive management's failure to recognize its improtance. Remember what our good ol' friend Deming used to say, that somewhere around 90% of the problems are traceable to the management?!?

 

[AgnieszkaSz]

Thank you - unfortunately, I know that this is a gap in process design. It is always subject to a heated discussions whether preventive solutions are economically sound in case if niche production (short series, many products).

 

[Jim Wynne]

Thank you - unfortunately, I know that this is a gap in process design. It is always subject to a heated discussions whether preventive solutions are economically sound in case if niche production (short series, many products).

There's nothing wrong with discussions about feasibility. It's helpful to remember that when considering preventive measures you shouldn't focus strictly on preventing defective products. It's possible to wind up in the poorhouse surrounded by a lot of defect-free products. You should focus on saving money, and reducing the level of defects is just one way of doing it.

 

[Umang Vidyarthi]

Thank you - unfortunately, I know that this is a gap in process design. It is always subject to a heated discussions whether preventive solutions are economically sound in case if niche production (short series, many products).

'Hot or cold' is immaterial, as long as the discussion is healthy, since that is the basic purpose of this forum. If you take a macroscopic view, then the cost of the preventive measures is negligible comparitive to the cost of rejections and loss of production.

Look at the mega picture like this- you can't expect your people to 'over look' the preventive measures for small lots and take the measures for large lots. It is human to form a habit, good or bad, which just can not be switched 'off' & 'on' at will. Beware, you are dealing with humans and not robots! Therefore, neglecting the preventive measures for small lots is also going to take its toll from the larger lots as well. Now compare the cost of poor quality vis-a-vis the cost of preventive measures!

The Japanese take great measures in this regard. They prefer to work with 'single cell system'. The whole line stops when problem occurs at any work station, and will resume only after taking full corrective and preventive measures. Prima-facie it appears to be a loss of production, but contrarily this has controlled the rejections and increased the production in the long run.

Umang

 

[basselope]

Old topic, but what the hey!

Based on the OP's original description, the error proofing needs to be flexible enough to cover multiple designs, inexpensive enough to be justifiable and robust enough to be effective.

Perhaps a solution could be:

1) Work order or check sheet to follow each assembly.
2) Work order or check sheet to include something like:
M8 studs, 12:___
M10 nuts, 7:____

3) Operator will physically number each stud 1 thru 12 (china marker, ink marker, etc)
4) Operator will write the piece count on the work order/check sheet signifying that each component has been numbered and count is correct.

Not bullet-proof, I admit, but I have seen this be effective and inexpensive. When the last number you write on a part is 11 but there is a 12 next to where you fill in the blank, folks tend to notice.
This also works well when the type of product is similar in design but with slight variation, i.e. 7 stud design versus 12 stud design running back-to-back. We used to have to do this with robotic weld stations and automotive dash board frames. The robots were not trust-worthy enough to allow them to run unsupervised. This also helped ensure that the correct weld program was associated with a particular dash frame, in this case it was applying M8 weld studs.

 

[Helmut Jilling]

Cameras or light sensors generally work very well. The costs have come down significantly. A couple of customer complaints for one missing nut can pay for the systems. Done correctly, it is virtually error-proof.