Learning Curves - Biggest Process "VARIABLE" is mobile workforce

[Chris May]

I have a problem that maybe someone else has come across........

I will just set the scene first.

We are setup to run Cell manufacture of PCB's for our own designed products.

Our operators are all trained at the training school for 3 days at a time and refresher course every 6 months (ish).
They are trained in general soldering skills and rework and this is all based upon the IPC-A-610 acceptance criteria. All records are updated, skills matrices etc; It is not the documentation at issue here.

So, they are all very good at generating decent soldering and have the skills for any necessary rework. Have excellent component recognition and are not colour blind.

The "textbook" method is that we can move operators around to any cell and they can produce conforming product. They can........after a while.

Each cell has a core team and now and again due to workloads varying (shortages in one area, change of quantity so fast response required), people from another cell are drafted in to help.

Now we only have about 12 PCB types and most of the people have worked on all of them at one time or another. But, when they revisit a product, they do make elementary mistakes until they "get back into the swing of it".

This is the biggest "VARIABLE" in my process at present........mobile workforce.

Each cell completes a simple tally chart which shows an array of defects during a day. I collect these and correlate them into weekly charts to spot trends, possible design improvements etc.

When I get these, I can immediately go to Carol or Marilyn and ask "Who was drafted in on Wednesday to help??
because there is a big spike for Wed.

I know that while we have manual assemblers every product has the opportunity to be unique.....it is by no means repeatable from the start, but after a while it is pretty **** good.

I dont go and beat people up over this at all, but it does affect the figures...........and I get beaten up.

Apart from automating (costly), I am thinking of some sort of Poke-Yoke arrangement...but I am not sure yet.

Hope this makes sense to someone.

Best Regards,

Chris

 

[Al Dyer]

Maybe instead of waiting for a "hole" in the cell then bringing in someone, you could structure moving cells on a determined time basis. Maybe every month you rotate entire cells to help reduce the "learning curve".

Just a thought

 

[CarolX]

cross train

I have to agree with Al. Perhaps as standard procedure people rotate every day. Then in a 2.5 week time period, everyone has rotated through all the cells.

Just my 2 cents worth.

Carol

 

[Mike S.]

Chris,

Use-it-or-lose-it as far as skills and recall is not unusual. I agree with the others -- rotating the folks regularly enough that they are always "up to speed" on every PCB type or cell. If this is too much for a person to keep up with, maybe you could have something like "teams" where team A are all regularly working on PCB types 1-4, team "B" all regularly work on PCB types 5-8, etc.

Aside from that concept, the only other possibility I can think of is to determine if the errors made by the "new draftee's" all fall into a few narrow categories regardless of the cell they are in. In that case perhaps some targeted training can be done. Or, maybe "cheat sheets" can be made-up for each cell -- something highlighting the errors new draftee's often make in that cell so the draftees can read it and refresh their memories before stepping in to help.

 

[rrramirez]

Some response to CHRIS

So, they are all very good at generating decent soldering and have the skills for any necessary rework. Have excellent component recognition and are not colour blind.

IMO you must define (operational definition) what is "a decent soldering" in order to have the right way to measure the expected quality for each of the 12 product you manufacture. This will help you to train better yoour workforce. The necessary rework is not a quality oriented approach but may be required by the need of the processes of your company.

we can move operators around to any cell and they can produce conforming product. They can........after a while.

This one of the Dr. Deming´s Funnel Experiment: Move the target each time you are out of it. You are introducing special cause of variation. Try to mantain consistent teams for each kind of product (May be you could divide them in 3 or 4 teams; but don´t move one team to other team for different kind of product)

Each cell has a core team and now and again due to workloads varying(shortages in one area, change of quantity so fast response required), people

See my point before.

Now we only have about 12 PCB types and most of the people have worked on all

See Dr. Deming´s read beads experiment.

This is the biggest "VARIABLE" in my process at present........mobile workforce

This the your principal source of variation; before go to implement a poka-yoke system you must organize your workforce in a group of consistent teams by 3 or 4 types of PBC.

You have too much variation in what you are doing right now.

 

[E Wall]

Rotation on schedule.

Chris, I have to say that Al's suggestion is probably the most efficient way to handle the problem. We do the same with departmental certification. Several departments use the 'team' approach and rotation of jobs is key to keeping skills honed so if/when a change is necessary, all team members can fill in without product quality taking a hit.

NOTE: I HAVE DELETED SMACKER'S OFFENSIVE COMMENT AND AL'S REPLY TO IT - LET'S ALL TRY AND KEEP ON SUBJECT IN A PROFESSIONAL MANNER AND REMEMBER YOUR POSTS CAN BE SEEN BY MANY AND THIS SITE JUDGED BY IT.

YOUR COOPERATION WILL BE APPRECIATED.

Eileen

 

[JodiB]

Offensive??? I thought he/she was sarcastically making the point that blaming/punishing employees for mistakes would be ridiculous. Chris never indicated that sort of mentality exists in his company, but it does happen in some companies and is not the most fruitful approach. Seemed a valid opinion and posting IMHO.

Perhaps Smacker did mean it seriously or frivolously - I just didn't read it that way.

 

[RosieA]

Hi Chris,
I have the same issue in my manufacturing area. We aren't making PCBs, but assembling printers. We routinely shuffle trained assemblers from one printer type to another with the same issues with variation that you've encountered.

Rotating people through all cells would not be practical here, so what we came up with was a re-training period. If the Associate had not been in a cell in the last month, then we required a one day training refresher with a certified cell trainer. The trainer would go through any process changes or quality alerts since the Associate had last worked in the cell, and work side-by-side with the Associate for a day at minimum, and more if required.

That helped drop the error rates, while allowing manufacturing management the flexability they needed.

 

[barb butrym]

what about a buddy system for "draftees ", with a check for the typical curve type errors, then release them from the buddy when all is well with the world again.

 

[barb butrym]

oops i didn't see rosie's reply...mine is a bit simplier than a training day, cause to me the stuff you are talking about is not trained, its technique and unique...if i read it correctly