Do I need to approach galvanized steel sheet flatness issue with DOE?

Nath2892

Starting to get Involved
#1
Hi there! So we have a galvanized steel sheet coming from a punching machine and then going to flattner. Even after going through flattner we still have flatness issues and its not meeting customer specifications. I was planning to do DOE using rolling speed, no of times a sheet is rolled, roller pressure as factors and with 2-3 levels. Do you think this is a better idea in figuring out optimal combination of factors or do i need to accept that there's a inherent fault in machine and it's simply not capable to produce good parts anymore? I really need some guidance in solving this issue. Please help!

Thank you!
 

Miner

Forum Moderator
Staff member
Super Moderator
#3
Never give up. Consider following two paths. If, as you say, the process formerly produced good product, follow a problem solving strategy (as suggested) to determine what changed.

You can also follow an exploratory strategy using DOE to determine which control factors are important. However, I recommend starting with a screening DOE at 2 levels with a couple of center points. After you narrow the field, you can go into modeling and optimization.
 

Nath2892

Starting to get Involved
#4
Good parts anymore?

If that is the case, what has changed?

I would start with the spec itself. How flat?
Thanks for the reply. We are trying to figure out what has changed and I need some directions on how to investigate what has changed. Overall flatness spec should be within half inch. Currently we running over an inch flatness requirement.
 

Nath2892

Starting to get Involved
#5
Never give up. Consider following two paths. If, as you say, the process formerly produced good product, follow a problem solving strategy (as suggested) to determine what changed.

You can also follow an exploratory strategy using DOE to determine which control factors are important. However, I recommend starting with a screening DOE at 2 levels with a couple of center points. After you narrow the field, you can go into modeling and optimization.
Thank you very much for your advice and encouragement. That was very informative.
 
#6
Thanks for the reply. We are trying to figure out what has changed and I need some directions on how to investigate what has changed. Overall flatness spec should be within half inch. Currently we running over an inch flatness requirement.
With that big of change, I'm not sure I would do DOE. I'd stick with a basic root cause analysis -- method, machine, people, etc.

Just a hunch, but it sounds like something is broken. Take a look at the equipment.
 

kzachawk

Involved In Discussions
#7
Holy Cow.... a DOE on a machine which cannot hold tolerance? I would not think so.....
Think Simple....
All failure is the result of variation occurring related to requirements and expectations.

Therefore if your machine was ever able to produce conforming parts after setup, then at what point do those parts start to vary from setup?

Most manufacturing sets machines 3 sigma inside the tolerance range and then uses a simple tool such as SPC to maintain machine setup, by observing machine drift during operation.

Your problem is really that simple....

Is the machine setup to achieve tolerance 3 sigma within the bell curve of the specification?

At what point using SPC does it drift away from setup?

The obvious answer here relates to three things
1. Your initial setup is too close to spec
2. Your machine is moving around during operation (Variation from setting)
3. Your machine operators/managers/supervisors are monkeying around with the controls during operation. (again causing variation)

Set an SPC control on the machine and require operators, managers, supervisors to log all changes .... then you will observe where the variation is occurring. Sometimes employees must be restricted from changing settings (locket out) The SPC chart will instruct that change is occurring and therefore a reset to setup is necessary.

A DOE is not warranted unless you have data which indicates the machine is incapable of producing any conforming components for the customer. If its ever produced one conforming part it can produce two ... its just a machine it only does what humans tell it to do.
 

Ninja

Looking for Reality
Trusted
#8
With that big of change, I'm not sure I would do DOE. I'd stick with a basic root cause analysis -- method, machine, people, etc.

Just a hunch, but it sounds like something is broken. Take a look at the equipment.
Agreed. That's a big change.

I would start with the machine operators...not in "What are you doing?", but in "What changed with the equipment? What do you see different now that didn't happen before?"
If it changed that much, someone likely saw it happen.

I would also look at the feedstock to make sure that didn't change in some way.
(Wound tighter, smaller ID core, thicker, stiffer, etc.)
 

kzachawk

Involved In Discussions
#9
For posterity ..... Deming's problem solving technique is called Plan Do Study Act... Should a machine no longer produce conforming product... then study needs to be undertaken as to when was the last time it produced conforming product and what has changed since, to cause it to no longer produce conforming product.... Its not rocket science folks .... keep your eye on common sense....
 

optomist1

A Sea of Statistics
Trusted
#10
or stated another way if after you analyze this to death DOE, ANOVA, etc....the common cause variation is "not acceptable"....you may have to redesign the process/apparatus etc....
 

Top Bottom