Printer Friendly

'Fuzzy logic' helps sharpen up injection mold designs.

The artificial-intelligence features of a package of injection mold-analysis programs has been enhanced by the addition of a "fuzzy-logic" advisor. This software "expert" examines a mold-filling simulation for potential processing problems and flaws in the mold design, calling attention to potentially ominous "blips" in the data that you might overlook. Then it advises you on ways to rectify the problem.

This is one of several recent additions and enhancements to the TMConcept Expert System for Molding from Plastics & Computer Inc., Montclair, N.J. These include an expert advisor that recommends an initial set of processing conditions from which to run a mold analysis--or an actual injection machine; a new entry-level sprue and runner balancing program; a new pack/hold analysis that facilitates warpage prediction; commercialization of gas-injection simulation; and improvements to existing TMConcept software modules.


The fuzzy-logic expert, which was first introduced over a year ago as part of the faMold 2-D layflat mold-filling program, is now also incorporated into the faBest 3-D finite-difference flow module in the just-released version 9.5 of the software. The program first asks you to designate a quality classification for the part on a three-point scale. From this and the identity of the resin you have chosen, the program recommends allowable limits for certain critical molding variables, which you can accept or alter at will. These include minimum and maximum melt and mold temperatures, maximum pressure and shear stress, and maximum pressure difference between any flow paths--i.e. how unbalanced the mold can be. Then, after you run the analysis, the program puts colored warning flags on any areas of the mold where conditions exceed the set limits during the molding cycle. At the same time, it recommends possible corrective actions and highlights the best of the possible alternatives.

This software advisor both helps less experienced design engineers and saves time for more knowledgeable users. In the latter case, the software can recommend changes to two or three parameters simultaneously, rather than requiring you to fix one problem at a time.

By mid-year, further evolution of this software will enable it to choose between two different mold setup recipes, picking the one that will yield highest quality and shortest cycle. Eventually, the program will even be able to interpolate between the two to find an optimum solution.


Another software expert advisor, which has been available for a little over a year, is called CAMMS-B. It helps you determine the best machine settings to run a mold, or a mold analysis, based on a limited set of inputs. It asks 30-40 questions about the part weight and volume, its maximum and minimum thickness, max. flow length in the runners, the material, overall mold dimensions, mold height, max. daylight, number of cavities, projected area of a single cavity, whether hot or cold runner, type of ejection (manual, gravity or robot), type of gating, and numerical ratings (on a three-level scale) for the degree of precision and quality of appearance required.

From these inputs, CAMMS-B will recommend barrel and nozzle temperatures, mold and water temperatures for cavity and core sides, a 10-step injection profile (calculated to maintain a uniform speed of flow-front advancement), four-step screw-speed profile, suckback position, velocity/pressure changeover position, five-step holding-pressure profile, and a breakdown of times for plasticating, filling, holding, cooling and total cycle. In addition, the program offers comments--for example, that gates may be too small, or that required pressure is near the limit of clamp force.

CAMMS-B is designed to be customized for various injection machines and made available on the machine controller or a central monitoring terminal. It has already been adopted by Sandretto and Arburg, with SHI (Sumitomo) soon to follow. And Plastics & Computer is working with Hunkar Laboratories to enhance CAMMS-B with the addition of recommended SPC limits and install it in Hunkar's CIM monitoring system.


As reported last month, Plastics & Computer also has a new faRun program that's simple to use, fast (1 minute on a 386 PC), and inexpensive ($2000). This stand-alone program helps you balance hot or cold runners for pressure, temperature or shear with minimal cavity description.


The brand-new faHold packing program integrates calculations of part shrinkage, orientation, temperature, and residual stress throughout the holding stage. After the velocity/pressure changover point (specified as a percent of fill), up to 10 steps of pack/hold pressure profile are available. Results from this analysis can be fed into a third-party, general-purpose finite-element stress-analysis program to calculate the degree of warpage (formatting for Abaqus, Ansys and Cosmos is already included). However, TMConcept software developer Giorgio Bertacchi at Plastics & Computer International in Milan, Italy, feels that these programs lack dedicated features to simplify the inputs and evaluation of results. For this reason, he has developed a fully integrated, dedicated package called faShape to determine the final (deformed) shape of the molded part at the equilibrium of all internal stresses, compare it with the original cavity dimensions, and accurately measure the shrinkage of any desired dimension.


The faBest 3-D flow analysis offers several new features. One is a calculation of part projected area and of mold-opening forces in any direction. In addition, faBest now displays the flow velocity vectors for every filled element at any moment in the cycle (before, only the velocity vectors at the flow front were shown). This visualization of velocity vectors can be very important for judging fiber orientation and potential causes of warpage, particularly for materials with anisotropic shrinkage.

Also, the large increase of computing power in workstations and PCs has permitted increasing the density of the finite-difference mesh in the thickness direction. The number of nodes has been made geometrically proportional to the rate of cooling. Field evaluations have demonstrated greater accuracy with this update in cases with very thin walls and/or materials with very high cooling rates.

The 2-D faMold filling program also adds a new feature that considers the limited surface area available for heat removal in the area of small or tunnel-type gates. Bertacchi has introduced an experimentally derived coefficient to correct what he says has been the tendency of all mold-filling programs to predict too short a cooling time to reach the no-flow temperature in the gate area. This correction is very important in predicting packing of crystalline materials, he says.


Although initially released in 1991 (PT, Aug. '91, p. 76) for use on a consulting basis, faGaim gas-injection molding simulation is now fully commercial and has been verified in several applications. Some challenging examples were molded at the K'92 show in Dusseldorf, where faGaim analysis was demonstrated concurrently. At the Sandretto booth, for example, a gas-injected part was molded with a hot runner. And at Battenfeld, a photocopier cover was molded with four gates and two gas-entry points.

"You define what skin thickness you want," says Bertacchi, "and the program tells you what percent of total part volume will be plastic." The program also calculates how much of the part should be filled with melt before gas is first introduced. The user enters the maximum allowable gas pressure and/or flow rate, and the program calculates the flow rate and pressure of gas at every instant during filling. The program automatically profiles the gas pressure/flow rate so as not to disturb the uniform advancement of the melt flow front, which can result in surface flaws. Anne Bernhardt, president of Plastics & Computer Inc., says it can be quite difficult to set gas pressure or flow rates properly without the aid of computer simulation. "When the gas inlet is in the cavity," she explains, "the pressure of gas must be the same as the pressure of the melt at that location in order not to accelerate the melt-front advancement. If you try setting the machine without first doing the calculations, you have no idea what that pressure value will be."

Bertacchi notes some special problems in accurately modeling gas injection. First, he says, it must be recognized that the gas displaces only the most fluid melt--not all the plastic up to the frozen skin layer. His model takes into account what he calls the "stationary layer," consisting of both the frozen skin and cooler, viscous melt adjacent to the skin (see diagram). Second, he says the model must use different coefficients of heat transfer, because the gas pushes the plastic against the cold mold, resulting in faster cooling.
COPYRIGHT 1993 Gardner Publications, Inc.
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 1993, Gale Group. All rights reserved. Gale Group is a Thomson Corporation Company.

Article Details
Printer friendly Cite/link Email Feedback
Title Annotation:Software Review
Author:Naitove, Matthew H.
Publication:Plastics Technology
Article Type:Evaluation
Date:Feb 1, 1993
Previous Article:Screenchanger valve stops unwanted flow.
Next Article:K'92 news in RIM & urethanes.

Related Articles
New dimensions in mold analysis.
'Artificial intelligence' already taking many forms in plastics processing.
The AI factory; how artificial intelligence will create 'smart plants.' (Cover Story)
700 readers say: mold analysis makes the grade.
Computer software will 'automate' concurrent engineering of plastic parts.
Ohio State University's.
CIM & controls.
Fast & easy flow simulation runs inside 3-D CAD.
'Fuzzy logic' makes controls think like you do.
Injection molding research targets micro-molding, sinks & foams.

Terms of use | Privacy policy | Copyright © 2021 Farlex, Inc. | Feedback | For webmasters |