Hi All:

Anyone using this software in a pretty high piece rate mfg environment?

Screw machines, cold headers 20k to 40k pcs a day.

Seems cumbersome, not specifically this product but how do we do SPC without cutting into production?

There is resistance, since experienced operators know almost by intuition if tolerances are drifting, they don't need SPC to tell them the obvious.

How do you answer such very legitimate questions?

I mean paper charting seems like it would be faster.

Been there in a previous life. We got a lot of grumbling at the beginning of the project, but once operators gained proficiency they told us it was easier to input values than to chart by hand. It will definitely save you time in calculating capabilities, re-calculating control limits, etc. There are remote sending/receiving units out there to capture your readings and eliminate some of the manual data entry.

Going to a computer system from paper will seem like a disaster for awhile until you reach the top of the learning curve. Good luck on your project, you'll be glad you did it.

Hmmmmm --- need more information to give you a "good answer"

Whats the current sample size?

How much of a "Cut into production are you seeing"?

> can't say I've heard anyone that would agree with the statement SPC by Intuition is an acceptable method....

it may be useful to your operation - maybe not - smallest realistic sample sizes could help but I'd explain to the operators that you need some data to prove that it isn't needed by using it for a short term anyway http://www.16949.com/ubb/smile.gif

Regards
Jim

We are testing a package from CimWorks that looks promising. It is flexible enough that we can network it to multiple operations and processes.

Defining which software works for our application has been a long road. We feel that the investigation process has also been a good learning process, as the operators that will use the system are part of the group evaluating the product.

I'll keep you posted as we continue...

ASD...

"Good luck on your project, you'll be glad you did it."

Thank you and I truly hope so.

Our SPC is a "flowed down requirement" and its not really clear that quality has improved since going to SPC. The parts function as they always have only now we have the new SPC requirements.

We could get a whole lot more "buy in" if we could get some easily observable quality, cost, or thru-put improvements with the new SPC.

Thank you

"Whats the current sample size?"

Our sample size is pretty modest, our people check and verify dimensions probably 2 or 3 times more frequently.

Its just the "overhead" of getting one's hands ready for computer work, fishing the right chart for the machine out of the computer, when one has been on a screw machine or header. Also we share computers, so people tend to "customize" the setting for themselves.

I'm trying to get our people their own computers, this maybe doable since computers are getting so cheap.

How much of a "Cut into production are you seeing"?

Significant enough that operations is aware, if you know what I mean.

"that you need some data to prove that it isn't needed by using it for a short term anyway"

I like this approach, maybe I can minimize it altogether :^o

BTW, many of our people still remember very fondly old DOS based SPC software since 1) the font was huge, small font and oil smudges on monitors are a real nusance. 2) DOS based apps were much more simple, no Windows overhead, and actually was faster.

"Experienced operators know almost by intuition if tolerances are drifting".... that sounds like they are over controlling the process. SPC will find that out for you, and lets not forget over controlling cost money and downtime.

We used to have operators that kept away from SPC, but now several years later, they REFUSE TO RUN without SPC charts.

Geoff

(Tried computerized SPC and found it too easy to fiddle the data. Our guys prefer paper charts as they use them to make notes on about how the process has run etc. The notes are really usefull when you have to try to remeber what happened months ago)

We are also thinking about using Visual SPC for a high speed manufacturing environment. We are currently using an old DOS based system and I would welcome any thoughts from anyone who has experience of using the Visual SPC system. (ie benefits, problems etc) My impression is that it is not easy to set up in the first place, but once it is set up the operators find it easier to work with.
Thanks

Hi All:

Anyone using this software in a pretty high piece rate mfg environment?

Screw machines, cold headers 20k to 40k pcs a day.

Seems cumbersome, not specifically this product but how do we do SPC without cutting into production?

There is resistance, since experienced operators know almost by intuition if tolerances are drifting, they don't need SPC to tell them the obvious.

How do you answer such very legitimate questions?

I mean paper charting seems like it would be faster.

These problems are real, and it is because if you are using X-bar-R charts, they are not the correct SPC technique for precision machining. Unfortunately, no software provides the correct technique. Some have said that machining is too specialized of an area for an addition to their software. But, you can read about the correct technique here:

Statistical process control for precision machining (http://elsmar.com/Forums/blog.php?b=79)

If you implement this technique, your operators will better understand the data that is generated, and decisions of when to make adjustments, tool changes, etc. will be far easier to make. It also explains how to make sane SPC sampling frequencies.

The real problem are special causes that create a lot of scrap fast - such as broken tools. If you do not check your parts frequently enough to catch them, you will have real problems.

Good luck!

We tried hooking Computerised Control charts to an Extruder machine - the operators were like almost 20+ yrs of exp. There was initial resistance nonetheless they started appreciating the inputs from the TV (is what the screen was called by them) which did away with human scope of misjudgement and errors of putting the values on paper.

Regards - Amit Sharma

We tried hooking Computerised Control charts to an Extruder machine - the operators were like almost 20+ yrs of exp. There was initial resistance nonetheless they started appreciating the inputs from the TV (is what the screen was called by them) which did away with human scope of misjudgement and errors of putting the values on paper.


I agree, use of SPC provides a valuable feedback to the operators, and - when performed correctly - minimizes operator intervention (overcontrol).

Unfortunately, most of the SPC packages do not have the correct statistics for machining. Attempting to use X-bar R charts (which should be adequate for extrusion) in machining encourages overcontrol by its incorrect statistics for the distribution that should occur.

I used visual spc with a bunch of toolmakers, it did take a whiel for them to buy into it. I used modified limits after i gathered enough data. I also worked on automating the gauges so operators didnt make so many mistakes and it took less time . Ended up with touch laptop computers abotu 18 of them.
Linda

I used visual spc with a bunch of toolmakers, it did take a while for them to buy into it. I used modified limits after i gathered enough data. I also worked on automating the gauges so operators didn't make so many mistakes and it took less time . Ended up with touch laptop computers about 18 of them.

The use of automated gaging is a great idea. For machining, use of modified limits is also a good idea when trying to force the software to work - I recommend 75% of the tolerance. However, Xbar-R is the worst possible charting methodology for machining. It encourages adjusting to the mean, which is overcontrol - and makes the operator the process instead of the process itself. It looks like it is working, but it really is not. I-MR is likely your best bet, but it is not as powerful or meaningful as X hi/lo-R for machining (http://elsmar.com/Forums/blog.php?b=79)- which is not available on the software.

So, automated SPC is nice, but using the correct charting methodology would be nicer.

I set the control limits manualy in visual spc where as my x bar became my nominal. This was done because the nominal is not always the way we want to go with tool wear and such. The range chart becomes the most improtant signal to assignable causes after you do a study. After the study you could set the x bar limits 3 to 4 std dev away from the max and min spec. Improvement is when you start tightening control limts. But I guess it depends on the goal to kep in tolerance to make each part simialr from batch to batch
Automatic gaging can be as simple as pressing a switch on an indicator
Linda

I set the control limits manually in visual SPC where as my x bar became my nominal. This was done because the nominal is not always the way we want to go with tool wear and such.

Exactly, tool wear is the issue. Correctly controlling a process with tool wear would be (for an OD example) setting up to the lower control limit, then allowing th tool to wear to the upper control limit, then repeating.

The range chart becomes the most important signal to assignable causes after you do a study. After the study you could set the x bar limits 3 to 4 std dev away from the max and min spec. Improvement is when you start tightening control limits. But I guess it depends on the goal to keep in tolerance to make each part similar from batch to batch
Automatic gaging can be as simple as pressing a switch on an indicator
Linda

Range is key, except in tool wear, the range in X-bar is the difference between consecutive part - which should be minimal. Generally what you get is measurement error represented in that range. The link explains that issue in detail.

After the study you could set the x bar limits 3 to 4 std dev away from the max and min spec.

Those calculations assume a normal distribution. If you are getting a normal distribution, you are likely charting measurement error and overcontrol by the operators. It is also explained by the link. You should set the limits to 75% of the tolerance, and let the tool wear.

Improvement is when you start tightening control limits.

That is incorrect in machining. That generates overcontrol. Improvement is reduction in the slope in tool wear, which reduces need for operator intervention and provides a more consistent process.

But I guess it depends on the goal to keep in tolerance to make each part similar from batch to batch Automatic gaging can be as simple as pressing a switch on an indicator
Linda

True!