Process capability

Sabertooth · May 1, 2024

Our pipe sizing operation (Tube end forming using I/O MACHINE) and then flange and another pipe gets welded to this sized pipe. After all this operations, pipe OD at inlet should maintain a tolerance range of 53.7mm to 54.3mm. Following a capability study, we determined that our Cp value stands at 1.03, with a sample mean of 53.9921mm. The customer has requested shifting the sample mean to 53.85mm. Considering the process variation and capability, we need to clarify the feasibility of this change to the customer. How can we effectively communicate this to them?

ScottK · May 1, 2024

Sabertooth said:
Our pipe sizing operation (Tube end forming using I/O MACHINE) and then flange and another pipe gets welded to this sized pipe. After all this operations, pipe OD at inlet should maintain a tolerance range of 53.7mm to 54.3mm. Following a capability study, we determined that our Cp value stands at 1.03, with a sample mean of 53.9921mm. The customer has requested shifting the sample mean to 53.85mm. Considering the process variation and capability, we need to clarify the feasibility of this change to the customer. How can we effectively communicate this to them?

Are they looking for a spec change to re-center the target at 53.85mm where it would currently be 54.0? Or are they asking that you just run the process to get what they want without changing the specification?
If it's the latter I would push back and and tell them they need to update the specification because right now you're meeting the current spec very well. And changing the spec would require another capability study with the cost associated.
I've run in to too much trouble in the past when asked to run to one side of the tolerance or another and the customer complaining that we were off even though we are in spec to the drawing. Targets and tolerances need to be black & white and indisputable.

Miner · May 1, 2024

One way you can visually demonstrate the impact to your customer is to do a Monte Carlo simulation of the process when centered at 53.85mm. You can show the tail area below the LSL and the estimated percent out of spec, or PPM.

John Predmore · May 1, 2024

I am curious whether the risk of occurrence and deleterious effect of oversized diameter is equal to the risk of undersized diameter, and that could be one reason for a process target not centered in the tolerance range. @Sabertooth has a pipe end sizing total tolerance of 0.6 mm on a diameter, so the pipe end contributes a tolerance of 0.3 mm to clearance on 2 sides. I am not a welding expert, but a tolerance of 0.6 mm on a tube in a weld seam seems rather tight. The OP didn't indicate which type of welding or whether material is added to the joint during welding. All metal welding involves substantial heat (even laser welding on a localized level), and I would expect to see distortion to diameter and roundness of the pipe during weld, regardless of what the print says. Maybe it is not your place to push back, but the design authority should talk to the parts and the process for guidance, rather than seek some statistical ideal.

Golfman25 · May 1, 2024

ScottK said:
Are they looking for a spec change to re-center the target at 53.85mm where it would currently be 54.0? Or are they asking that you just run the process to get what they want without changing the specification?
If it's the latter I would push back and and tell them they need to update the specification because right now you're meeting the current spec very well. And changing the spec would require another capability study with the cost associated.
I've run in to too much trouble in the past when asked to run to one side of the tolerance or another and the customer complaining that we were off even though we are in spec to the drawing. Targets and tolerances need to be black & white and indisputable.

Your dead nuts right now, and probably have "too much variation" causing your CPk issues. So the question is why are they asking for the re-center to 53.85. You would need to change your process, wouldn't you, to bring the dia down a bit? You'll still have the same variation, unless you figure that part out as well.

Sabertooth · May 1, 2024

ScottK said:
Are they looking for a spec change to re-center the target at 53.85mm where it would currently be 54.0? Or are they asking that you just run the process to get what they want without changing the specification?
If it's the latter I would push back and and tell them they need to update the specification because right now you're meeting the current spec very well. And changing the spec would require another capability study with the cost associated.
I've run in to too much trouble in the past when asked to run to one side of the tolerance or another and the customer complaining that we were off even though we are in spec to the drawing. Targets and tolerances need to be black & white and indisputable.

They want it without drawing change. Problem is this is a customer that I can't piss off as I have multiple quality issues and they have been lenient in issuing complaints for those

Sabertooth · May 1, 2024

John Predmore said:
I am curious whether the risk of occurrence and deleterious effect of oversized diameter is equal to the risk of undersized diameter, and that could be one reason for a process target not centered in the tolerance range. @Sabertooth has a pipe end sizing total tolerance of 0.6 mm on a diameter, so the pipe end contributes a tolerance of 0.3 mm to clearance on 2 sides. I am not a welding expert, but a tolerance of 0.6 mm on a tube in a weld seam seems rather tight. The OP didn't indicate which type of welding or whether material is added to the joint during welding. All metal welding involves substantial heat (even laser welding on a localized level), and I would expect to see distortion to diameter and roundness of the pipe during weld, regardless of what the print says. Maybe it is not your place to push back, but the design authority should talk to the parts and the process for guidance, rather than seek some statistical ideal.

This is a Mig welding. Weld distortion also came to my mind

Sabertooth · May 1, 2024

Golfman25 said:
Your dead nuts right now, and probably have "too much variation" causing your CPk issues. So the question is why are they asking for the re-center to 53.85. You would need to change your process, wouldn't you, to bring the dia down a bit? You'll still have the same variation, unless you figure that part out as well.

At customer location, gasket goes on top of this and gasket is also within spec and its creating too much tightness recently. I requested if temporarily/ informally if they can allow us to make it little bit under spec, keeping same tolerance width

Sabertooth · May 1, 2024

Miner said:
One way you can visually demonstrate the impact to your customer is to do a Monte Carlo simulation of the process when centered at 53.85mm. You can show the tail area below the LSL and the estimated percent out of spec, or PPM.
View attachment 30200
View attachment 30201

% of parts having diameter above 54.10	18%
% of parts having diameter above 53.9 and less than 54.1	58%
% of parts having diameter less than 53.9	23%

Bev D · May 1, 2024

An actual histogram is more informative. Any time you provide a table of numbers people try to draw the graph in their head…

If you are all in spec yet there is ‘too much tightness’ of the gasket then you have a stack up problem. (Even when all parts are “in spec”). While moving parts distributions around will help now, the gasket will move eventually and you’ll have to move your process again. This is the initial reason that the Japanese decided to quantify variation vs. the specification with Cpk (later known as Ppk).

Process capability

Sabertooth

Registered

ScottK

Not out of the crisis

Miner

Forum Moderator

John Predmore

Golfman25

Sabertooth

Registered

Sabertooth

Registered

Sabertooth

Registered

Sabertooth

Registered

Bev D

Heretical Statistician

Similar threads