If you take computer simulations of thermoplastic elastomers with a grain of salt, you’ve probably done the right thing.
Finite element analysis of elastomer parts hasn’t truly measured up to reality in semi-dynamic applications — those involving cyclic loading at no particular frequency. “The error in these cases typically ranges from 10 to 30 percent and varies from elastomer to elastomer,” says Thierry Burton, worldwide design supervisor for ExxonMobil’s Santoprene division, a leading elastomer supplier.
This error has its roots in a physical phenomenon that affects thermoplastic elastomers in general, not just those supplied by Santoprene. “Elastomers experience a cyclic softening. So what you see on the first cycle is not what you see on the fifth cycle,” explains Ward Nahri, a senior design engineer for Santoprene.
In the past, Nahri continues, the mechanical testing that underpins FEA material models didn’t take cyclic loading into account. So engineers instead relied on models based on single-cycle tests of each elastomer grade in tension, compression and shear.
While useful, this data only mimics static applications. “Things like a pipe seal that you would install once and forget about,” says Nahri. When applying this static data to semi-dynamic applications, FEA would tend to over-predict stiffness, Nahri explains. And because engineers too often design for initial stiffness, the over-prediction holds true even though the softening effect tends to level off around the fifth loading cycle, he adds.
Santoprene has now addressed this prediction problem by creating a database of semi-dynamic material properties. So far, the company has about 10 of its elastomer grades tested, the ones most likely to see use in semi-dynamic applications such as window or appliance seals. “We’re just at the starting point,” says Burton. “But the intention is to develop semi-dynamic data for most of our materials.”
That will be a big job. Whereas the static data is based on just three mechanical tests — tension, compression and shear — the semi-dynamic data requires the same three tests over five cycles and at multiple strains. “Characterizing each material from a semi-dynamic perspective requires the equivalent of 90 tests rather than just three,” Nahri says.
Santoprene’s push to add more useful FEA data won’t end with the semi-dynamic information. “We’re always working to give our customers more robust modeling capabilities,” says Nahri. Next up will be viscoelastic data. According to Nahri, this FEA data will primarily help engineers predict stress relaxation, an important consideration in elastomer seal design. The same data can also be used to predict creep, he adds. Nahri expects the viscoelastic data to be available sometime in the first quarter of 2007.
In the meantime, an example of the new semi-dynamic data can be found on the company’s FEA datasheet page, along with reams of raw static data and material models. Registration is required, and you may need to interact with the company’s online Answer Person to get the data you need.