Edited by Leslie Gordon
When it comes to rubber products, designs, materials, and manufacturing processes are still largely trial and error. CFD or FEA usually comes into play only after problems arise, which is unfortunate because these tools can, for instance, predict mold filling, material scorching, curing, cooling, and distortion. CFD results can even be used as FEA inputs, letting users see how molded, extruded, or cast parts will perform in service.
Rubber manufacturers continually collect quantitative data regarding material-processing temperatures, pressures, and efficiencies. Yet product-development engineers usually check a rubber sample just to determine whether it meets one of two possible outcomes, such as passing or failing a test. Information is thereby discontinuous and of little use in understanding and optimizing rubber products. Correlating physical tests with simulations helps bridge this gap.
On the simulation side, numerical analysis of rubber products requires nonlinear viscoelastic models for component shaping. Nonlinearity arises because rubber goods can involve large rotations and translations as well as complex interactions with connectors. For instance, consider the excessive deformation of a rotating axle CV boot.
For physical verifications, lab tests are needed to quantify shrinkage and warpage and determine exactly where knit-lines, flash, and air traps crop up. Also critical tests pinpoint how time and temperature domains affect rubber. Simulation codes such as Moldex3D and the Virtual Extrusion Laboratory (VEL) can do the same, based on lab data on noncured rubber as measured or after external fits to models subjected to compressed laboratory probing.
Lab results are used to determine the parameters (constants and functions) of the equations involved in simulating rubber. For linear materials such as steel, a familiar example of a material parameter is Young’s modulus. Rubber, though, deforms nonlinearly from the start of loading, and it can withstand extremely large reversible strains. Hence, many more parameters are needed to characterize a rubber compound than a linear material. Further complicating matters, attributes in any given deterministic equation are generally unique. Thus, tests must be repeated for every compound and condition (temperature, media, and time period).
How rubber deforms
Rubber can deform in uniaxial, equibiaxial, planar, or volumetric modes. In turn, each mode can be tensile or compressive, for a total of eight tests to characterize rubberlike materials when loading begins. To date, only the uniaxial deformation mode falls under ASTM Standards. The other modes are still in R&D.
The most important rheological property of rubber is viscosity. Factors affecting viscosity include polymer type, concentration, added particles size and distribution, emulsifier, temperature, and shear rate. Other rheology parameters include specific heat capacity, specific volume versus pressure and temperature (PvT), thermal conductivity, melt density, ejection temperature, no-flow temperature (“freeze temperature”), and curing.
Measurements performed at polymers rheology laboratories such as Rapra Technology in England or DataPointLabs in Ithaca, N.Y., use special instruments such as precision capillary rheometers. The labs produce data isothermally at three temperatures representing the process range of the test rubber. The labs run tests using two capillary dies of the same diameter but differing lengths (L/D ratios). The data feeds Bagley’s end-correction equation to eliminate entrance and exit pressure effects.
Results show viscosity as it relates to the steady shear flow in the test capillary. These results are entered directly into Moldex3D or VEL, for example. DataPointLabs requires 2 lb of pellets to characterize plastics for flow analysis while Rapra works with 2 lb of rubber without curatives to generate viscosity data. Costs and lead time for rubber characterizations are worthwhile because the same data can be used in future rubber-product-development projects.
Developing a rubber product requires understanding manufacturing limits such as the compression needed to make a rubber gasket seal, or the internal pressure at which a rubber air duct would collapse. Design parameters depend on the application at hand.
When sealing is important, the least material condition (LMC) must address contact pressure over the life of an assembly under light compression. Additional tests should record rubber-compressive stress-relaxation (CSR) effects. These factor time into the prediction of sealing and show, in particular, how long the seal will last. Labs such as ours house CSR equipment to account for polymers viscoelastic effects.
Another design parameter could be the force needed to assemble or actuate a system involving a rubber component. Examples include closing an electrical connector or mounting rubber hoses on nipples of varying materials in medical setups.
Breaking the contact takes some load that varies from one elastomer to another. Besides, loads fluctuate, depending on speed, contact area, and magnitude of normal weights. One solution is to use an average as a friction coefficient or program a “reaction to drag with amounts of movement” in general-purpose FEA software such as Ansys.
Modeling the performance of a rubber product using FEA requires several elements, such as the geometry in a nondeformed state, a model representing the compound, boundary conditions such as contacts or restraints, and loading conditions. Performing FEA on a rubber product must be incremental and iterative, gradually applying displacements and loads to simulate, for instance, placing a rubber hose on a plastic nipple and then pressurizing the hose after assembly.
CFD requires the geometry of the rubber part and tools. It also needs the rheology of the rubber and various process conditions. An example model might show the tooling cavity or the cuts in the die to form round rubber extrusions into a needed profile. Mold runners, which move the polymer from the production machine to finished parts, are commonly modeled by tubular beam elements.
Most commercial or research CFD software for polymers includes material libraries. For more-precise process modeling, special one-time tests should be conducted on the specific material. These tests often fall beyond the scope of support services that elastomer suppliers provide.
Our lab recently helped a U. S. custom molder that had built tooling for rubber overshoes for medical clean rooms. It was necessary to simulate XL and XXL sizes in Moldex3D to reproduce a defect above the heel. The finest-meshed model had 451,973 elements connected at 641,403 nodes. The coarsest model ran in a few hours. The software showed the advancement of the rubber front and indicated an air trap above the heel, confirming earlier “shortshots” at the molder. The molder used this data to redesign the gating and eliminate the defect.