Sponsored By

September 21, 1998

11 Min Read
Tailored data equals fast, accurate analysis

Whether you are trying to perform a moldfilling, structural, thermal, or fluid analysis, you have probably run into the "brick wall" experienced by other plastic part designers. Simply put, plastic materials behave in a nonlinear fashion and are typically subject to large deformations during their use. So the material models generated to help predict their behavior are complicated. First, you need the right material data. Moldfilling and structural programs often require multipoint rather than single values supplied on a datasheet. Also, the data must be reduced into a form recognized by your analysis program.

These problems combined often lead to CAE programs that are under-utilized. Typically, the software and hardware represent a sizable investment. Datapoint Testing Services (Ithaca, NY) has come up with one solution to help designers and analysts make these purchases pay off for their companies. The solution, says president Hubert Lobo, is to provide Datapoint's customers with materials characterization data tailored specifically for the analysis program being used.

A Custom Approach
Let's take a look at the challenges presented during moldfilling analysis. The simulation programs are designed to crunch numbers, but when those numbers are incorrect or approximations, so is the result. Plastics designers need a range of precise properties for these simulations. One such property, illustrated by the p-v-t diagram in Figure 1, p. 53, shows how the material volume changes with temperature and pressure, providing a measure of the material shrinkage. For example, if there is a major volume change when a resin goes from melt to solid, the holding/packing phase of the molding cycle needs to be extended to achieve predictable part size and proper filling. But this data simply doesn't exist for most materials.

Roots of a design dilemma

According to Lobo, material modeling is a critical component of the structural analysis of parts made from plastics, elastomers, and rubber. "While these materials present unique characteristics and are advantageous from the standpoint of cost and processibility," he says, "their complex behavior can present a difficult challenge to the designer and analyst. Linear elasticity, a staple in the modeling of metal behavior, does not provide reasonable results for such materials. This forces the CAE analyst to look at nonlinear material models to adequately describe their behavior."

Historically, there have been two main drawbacks to using non-linear models, according to Lobo: long computation time and complex material models. Affordable, high-speed computers and enhancements to structural analysis codes have all but eliminated the first problem. However, the issue of complex material models has not been easily addressed. "Although analysts now have the computational capability to perform nonlinear analyses, such as structural and moldfilling," he says, "many have been hesitant to step into this arena because of the difficulties related to the generation and use of nonlinear material models."

Several factors complicate the issue. The choice of material model varies according to type of analysis. Structural analysis programs, for example, often provide several material models to describe the same phenomenon, and choosing the appropriate model can be a critical matter. Also, Lobo adds, "With many models, the material property test specification does not clearly outline an unambiguous scheme for generating the property being modeled. This can lead to interpretation difficulties that can compromise the quality of the generated data." Lastly, the data must be reduced into the form needed by the program. Obtaining the right set of coefficients for some models requires a good working knowledge of statistics and nonlinear regression methods.

"This is the driving force behind the creation of TestPaks," Lobo said. Because Datapoint Testing Services is a full-scale materials testing laboratory, it can generate the original material data. More than 40 TestPaks combine all the material properties needed for a specific material model in the format needed for 13 different analysis software programs, including Moldflow, C-Mold, I-deas, Abaqus, and Ansys. In addition, materials are tested according to specifications of each program. Users will spend anywhere between $1000 and $3000 per material for the service, which improves analysis accuracy and considerably slashes the time requirement. In most cases, Lobo noted, data can be shipped on a disk formatted to read directly into the user's software program within five working days. The company also has a 48-hour rush service available.

We spoke with two of the design engineers currently using TestPak data: Gayle Rose, senior engineer at Becton Dickinson (Raleigh, NC )and Brian Zajas of Nypro Design Group (Clinton, MA).

High-Volume Data
A long-term Datapoint customer, Rose explains, "Our group within B-D performs moldfilling and structural analyses for the entire corporation, which utilizes both custom and captive molding operations. We use both Moldflow and C-Mold software for moldfilling analyses, and Abaqus (from Hibbit, Karlsson & Sorensen) for structural FEA (see sidebar, below). Many of our products are injection molded--connectors, caps, test tubes, and well plates for chemical testing." Designers are mainly working with PP and PS, but also ABS, PC, and propionate.

Why does B-D invest in materials characterization data? "You can't get the information you need for structural analysis or advanced moldfilling simulation from a data sheet," she says. "In the beginning, one of our engineers would get stress-strain curves and reduce them by hand into the format required for Abaqus. But that is time consuming. When it comes to some of the grades, at times we can't even get basic data from the material supplier."

Standardizing analysis software interfaces

To solve the apples-and-oranges comparison problem for plastics designers around the globe, major resin suppliers joined together to develop a standardized database called Campus.Now 10 years old, more than 150,000 copies of the materials database are in circulation. Campus itself contains data on over 5000 plastics from 50 materials producers worldwide. Recent U.S. additions to the consortium--GE Plastics and Amoco--will soon be adding their materials. Now that the database is standardized, consortium members have been seeking to do the same for analysis software interfaces. The first three are now on the market. One is an interface to Cadmould (from Simcon GmbH) and the other is an interface for transferring stress-strain diagrams to Abaqus (from Hibbit, Karlsson & Sorensen). Formerly, Abaqus users had to manually convert stress-strain curves from Campus into the format required by the material model of the system. The interface will accurately perform the conversion in seconds. Another Campus software program, MCBase (from M-Base Engineering), merges data from different manufacturers, allows users to enter their own data, and converts Campus material data into various simulation formats. Look for Campus at K'98 next month. In its newest incarnation, version 4.1, the database will use international standardized abbreviations for basic polymers, listed below. In addition, 4.1 will offer information for calculating density, heat conductivity, and freeze temperatures, along with p-v-t data. Also look for DSC diagrams, which present specific heat as a function of temperature and supply data for calculating cooling times.

In certain cases, Rose's group opts for TestPak data, including the long-term creep data needed by Abaqus and specific data moldfilling programs--basic rheology information such as viscosity, thermal properties, and p-v-t data. "If we have a material that will be used in a lot of different products, it's easier to decide to invest in the characterization," Rose says. "If it's a material that will only be used once, we might spend some time talking to the materials supplier to find a similar material in the Moldflow or C-Mold database that would approximate the grade we are using for analysis purposes."

There is another reason B-D turns to outside testing: Many material vendors don't have data on all of their grades. "Even GE Plastics," notes Rose, "whose database is very good, doesn't always have the grades we are using listed."

Aiming at Accuracy
Nypro's Zajas, another TestPak user, works in the Engineering Services Group to perform moldfilling analysis for many of the company's divisions. This central design support function has also been providing product design and development work to customers for the past 15 years.

Prior to using TestPaks, Zajas recalls, Nypro designers were unsure about the validity of material information. At times, there were no data sheets for specific grades. "We'd have to find a similar material in the Moldflow database and hope it had the same characteristics by matching melt flow rate and other properties," he adds. "Still, it was a substitute for the real thing."

Zajas now receives analysis-ready rheology data from Datapoint as specific sequences of viscosity points at certain temperatures and shear rates. "These are the only data accepted in moldfilling analyses," he says. Although resin suppliers do provide rheology data, it is often not in a format that's ready to load into a CAE program. And p-v-t data remains elusive, so Engineering Services outsources the materials characterization.

"After simulating how the mold fills, we naturally turn to the next phase in the injection molding cycle, which is packing. In the packing phase, you are compressing more plastic into the mold," reports Zajas. "Because plastic is compressible, its specific volume changes in response to temperature and pressure. Without p-v-t data, we cannot predict the material's shrinkage response." Nypro's engineering group then uses this data to determine gate seal-off. "Also, what happens in the packing phase determines part size, so it is a critical phase," he points out.

Time savings also justify the cost of the service, adds Zajas. "All I need to do is read an ASCII file into the database--it's amazingly simple." The alternative was for Zajas to manually enter all data points, which could take hours.

International standardized abbreviations






























































Contact information
Simcon Kunstofftechnische
Software GmbH
Herzogenrath, Germany
Phone: +49 (2407) 50 88
Fax: +49 (2407) 59 45 3

Hibbit, Karlsson & Sorensen
Pawtucket, RI
Phone: (401) 727-4200
Fax: (401) 727-4208

M-Base Engineering & Software
c/o The Madison Group
Madison, WI
Paul J. Gramann
Phone: (608) 231-1907
Fax: (608) 231-2694
Website: www.madisongroup.com

Sign up for the PlasticsToday NewsFeed newsletter.

You May Also Like