Eliminating Missing Steps In Manual Operations


Starting to get Involved
I know this may be a difficult issue and probably has been from the beginning of time but here it goes.

In our case, not often but once in a while our fabricators miss a weld. I would guess human error is to be expected but our customer simply has no tolerance for it. One solution is 100% inspection, but that's stop gap and isn't addressing the core issue.

Aside from converting to robots (not an option) what suggestions or methods are out there that may be helpful?
Re: Missing Steps In Manual Operations

Something to consider: Add accountability, with more granulation in the flow inspection points. Have two lines for sign-off in the work order line: one for the operator, one for the verifier (usually a charge-hand or QC inspector). This will identify where the failure is occurring and the methods that aren't working.

You didn't mention if the part goes from Weld to another area for further processing, or if it goes directly to the customer. If it's the latter, then the final inspection step isn't effective or doesn't exist. If it's the former, the group receiving the part should always be verifying the previous step. If it hasn't been done, the part is simply returned to the previous cell for completion.


So was wondering if there was any other scientific approach to identify needs based on which a communication model can be developed which clearly identifies 5W 1H of the communication to be done.



Inactive Registered Visitor
There are weld monitoring devices capable of counting welds during the process.

Arc Agent is a name that comes to mind. Sure there are others just one that I have experience with. No affiliation just a past customer.


Involved In Discussions
I'd like to share my experience with the same issue some years back when I was with car manufacturing plant. The area was where we manufacture steel made re-usable pallets that are used for stacking/shipping automotive transmissions to Asia. During training there was a piant-pen that we use to mark the weld location after we had finished welding that particular portion. That part was to train the welder to inspect what he had welded. It worked for a time until the welders marked the welded part even when they did not actually checked it. Worse was after training the paint marker was removed and as the welders became accustomed to the work the became complacent and it paved the way for non-checking.

From there we would have missed welds or in most cases welds that are not good/proper. The end users have pointed the defective parts to use when the pallet arrived at Malaysia/Thailand with supports broken because one end seemed not to support the whole load. The broken pallets was shipped back to us with the photos identified as weakpoint etc. We saw it all. We have traceability codes to determine what shift and who was/were the welder/s but that is a different issue( retraining and suspension etc). This is where it is tricky because even if the guilty welder says he had inspected/checked the welded portions properly there was in no way that we can be sure prior shipment so we had added a station just for weld final inspection. So there was an assigned inspector just to inspect the welding 100%. Complete with a map for checking/inspection guides plus tools to check if the weld is stable. That weld map checklist was also crucial when we had to evaluate who was prone to poor welding on what weld location on what station. We even plotted it in a graph for every welder to see so they know where they stand. That is also a key factor why on that particular weld location this welder welds poorly, at what time/shift.

This was also included in a Quality Circle project which was to eliminate weld misses/poor welds. A lot of improvements came from it. The lighting, space, improved visual aids/keypoints at every station, rotating welders so that they may not be complacent in one location/station...environment improvement..etc. But as mentioned earlier the best solution that we had at that time was an additional inspector outside those welders. An additional inspection procedure aside from the welders inspection procedure. There was also a strike out penalty so if a welder has found to be re-miss on his welding quality there was an appropriate penalty. It still is the same until today except that they are now shifting to robotics slowly.
There are weld monitoring devices capable of counting welds during the process.

Arc Agent is a name that comes to mind. Sure there are others just one that I have experience with. No affiliation just a past customer.
This won't help with missing welds - it's used to identify welds made that are outside specified parameters. The system can't flag something that never existed in the first place.


Starting to get Involved
Hi folks,

(Apologies for misbehaving by jumping straight into a forum without introducing myself properly ;) ).

Just to add something to the power of 100% overchecks/inspection

I have just carried out some Human Factors training with our shop-floor, and showed them the following simple bit of 6-sig analysis:

After a bit of discussion with the operators, we agreed it was reasonable to assume 1) that humans were 99% accurate and 2) our typical routing gave a person roughly 50 opportunities to make a human-based error (in actual fact our routings give probably closer to 200, but that is my problem, not theirs!)

Therefore, only 0.99^50 = 60.5% of routings reach dispatch with no errors.

The simplest way of improving this unimpressive figure is through a 100% overcheck. Assuming your inspector is also 99%, then the maths is now 0.9999^50 = 99.5% and through adding an extra op we've suddenly turned the figure into something acceptable.

I have also used this to great effect with top management, who think that Quality Inspection is a waste of resource.