When analyzing variation in mass production components, people often fall back to dimensional variation. Often times, dimensional variation falls into known, repeatable patterns allowing statistical models to function properly. Normal bell curve or one sided distributions lend themselves to analysis very well.
Unfortunately, a number of die casting variables and failure modes are not easily measured, such as appearance or porosity. Porosity found after machining or in X-ray in particular is difficult to model or analyze due to its inherent odd shape. Many OEM’s have tried to come up with systems for doing this as an inspection criteria with some success. However, actually utilizing statistical techniques to control porosity is difficult.
With any process control, the inputs into the process are key to maintaining a stable and repeatable output. Die casting presents a large number of variables in inputs such as:
- Gate geometry
- Vent geometry
- Die temperature and cooling profile
- Metal temperature and chemistry
- Injection parameters, such as plunger velocity and pressures
- Lubrication system and set up
It is critical that die casting companies take all of the associated variables into account and control them in order to minimize casting output variation. Even with such, failures are often due to special causes of variation, such as mechanical breakdown of a system instead of normal wear issues that can be caught beforehand by statistical methods.
By minimizing the variation in process inputs while monitoring process output, Basic Aluminum seeks to apply standard statistical techniques to reduce delivered die casting variation to customers.