Process validation

[processtech]

Hi,
Just interested to know what you do as regards process validation.
At the minute i set upper and lower levels of the following:

Hold pressure
Inj. Speed
Cooling Time
Changeover point
Screw speed

then develop tolerances around the resultants, eg inj. time, plasticizing time, cycle time, melt cushion

What is standard practice on this, im not sure if i need the screw speed, can one just take actual values of these resultants, e.g melt cushion, plasticizing time, and look at the variation over a production run, and base tolerances on this,

Also is there an industry standard practice on hot runner set point tolerances.
Thanks

[Suhas]

ProcessTech,

Good Start.
To begin with tolerance are on inputs and alarms limits are on outputs. Tolerance is what you will let the tech change the value by. Set = 800, Tolerance is 100 - so the tech can go from 700 to 900. Alarm limits - the cushion value is an output and so you will determine it should be for example, +/- 0.05" So without changing anything on the machine, if the cushion values drops by 0.06", the machine will alarm.
Having said that, your list above is for tolerances. Come up with a list for the alarm limits. This is food for thought. Will also post it later.
Suhas

[processtech]

ProcessTech,

Good Start.
To begin with tolerance are on inputs and alarms limits are on outputs. Tolerance is what you will let the tech change the value by. Set = 800, Tolerance is 100 - so the tech can go from 700 to 900. Alarm limits - the cushion value is an output and so you will determine it should be for example, +/- 0.05" So without changing anything on the machine, if the cushion values drops by 0.06", the machine will alarm.
Having said that, your list above is for tolerances. Come up with a list for the alarm limits. This is food for thought. Will also post it later.
Suhas

Hi Suhas,
Thanks for your reply, the following would the list of outputs from those
Fill time
Melt Cushion
Plasticizing Time
Cycle Time
Peak melt pressure

As regards to alarm limits i would base these on the lowest and highest values of each during the run.


What im going to ask is would you create a high and low tolerance say for changeover point, i want to leave a window wide enough that can cover potential changes in the future, without going outside the validated process window.
Can you advise on the temperature tolernaces, or would you say these should be specified with a validation
Thanks

[brentb]

This is where things like DOE come in. Too wide a "window" makes for an invalid validation.

Buy the Suhas Book (Robust Process Development and Scientific Molding) it is worth every penny)

If you mention my name you will get a special( higher) price !

kom

brent