Small questions about injection moulding practice

How do operators today determine good injection moulding conditions?

Is there software available which allows the user to determine good injection moulding conditions (melt temperature, mould temperature, back pressure, etc.) based on empirical models?

Would it not make sense to correlate product characteristics with injection moulding conditions? I don't find journal papers mentioning anything of this sort. I understand that this would be mould-dependent and machine-dependent, but if the dependency is significant, the operators too would have to deal with that and find good conditions for each mould, each machine.

With kind regards,

A. Bulsari

I looked through and it looks like their software is not based on empirical models. The focus is more on the mould and fluid dynamics than on finding suitable moulding conditions. Please correct me if you know that they also have solutions based on empirical models.

A. Bulsari

In the medical device field, the FDA consider plastics processing as a "special process", where the final parts are not 100% verifiable.

In this industry, there is a huge amount of up front work done in both molding and extrusion in process characterization just for what you are describing. The process needs to be pushed to failure during characterization, where that failure is either in the process's ability to make the part, or until the part itself fails some specific predefined test or tests. It is best to do this work with multiple lots of the resin.

After completion of the characterization, a validation protocol is drafted with IQ for the equipment (installation qualification) if it is new, OQ (operational qualification) for the specific part on that specific equipment where an operational "range" is established as outlined in the protocol, and a PQ (performance qualification) where the nominal conditions are tested, most likely numerous times. All of the data is then looked at statistically and matched up with the predefined criteria established in the validation protocol. All associated tools such as molds, dies, tips, etc. are given drawings and part numbers.

Once the protocol is completed, from the validation, a process specification is authored. Once completed, the parts have to be manufactured within the constraints of the process specification, or else they cannot be used unless they are determined by cross functional team of QAE, Manufacturing Engineering, Regulatory, and other functions as usable (usually called a Material Review Board, or MRB). For the most part, it would include a lot of testing of the component and the product.

In the past, in extrusion, I have seen many engineers institute validation protocols that lead to many problems in manufacturing. Such things as specifying a screw RPM range and a FPM range. A slight shift in the SG or bulk density, and you cannot produce parts within the specified range, and you need to convince a panel of other individuals that know very little, if anything, that a change needs to be made (DCO), which may push the part back into the validation cycle.

In a well run organization, this detailed approach can work wonders. With the proper controls in place, the decisions that can be made by the operators are (theoretically) proven to make functional products. ____________________ Lawrence Alpert Manager, Extrusion Boston Scientific Fremont, CA

Look up articles on scientific injection molding by John Bozzelli (SP?). He has dedicated much of his work to scientifically determining the proper molding parameters for parts. He has written many articles about this as well as the concept of a universal set-up sheet. Please let me know if this is what you are looking for.

Bozzelli's ideas are also interesting, a bit different from what I am thinking of. Articles/documents I found describe procedures without much theory, or an explanation of why his way. Calling it scientific did not sound convincing. I agree that injection moulding work is going more to cheaper countries in Asia, because the developed countries are still not offering anything which is of much superior quality.

I think we have a good solution to determining suitable injection moulding variables (and looks like there are no competitors either). (1) The user can decide the product characteristics he wants to take into account. (2) The software system plans out a set of experiments. (3) The user carries out the experiments and feeds the data to the system. (4) The system tunes models (not plain linear regression like SPC) based on that data for each product characteristic. (5) The user can specify his requirements, and the system calculates suitable variables (injection moulding conditions).

Now we should slowly start looking for representatives willing to distribute this software system. Our (Nonlinear Solutions Oy) modest Internet site does not tell very much, and there is still no description of the software system there, but I will be glad to communicate with interested people first. (My e-mail address is snipped-for-privacy@abo.fi, or write from the contact page of the company's www site.)

With kind regards,

A. Bulsari