Method Development & Analysis of Seals using FEM

Abstract: Hyperelasticity is a significant property of rubber, taken advantage of in engineering applications. A common application is the use of seals to prevent fluid transfer (liquid or gas) between solid regions. Volvo CE is often depending on external supplier when developing seals. However, it could be beneficial to be able to do design and analysis in-house. Thus, they want with this master thesis to increase their knowledge about rubber and FEM simulations of seals in ANSYS. The aim with this work is to develop a method and guidelines for analysis and simulation of seals of hyperelastic materials. Components analyzed in this thesis work are two static seals, an O-ring andan in-house modified X-ring design. Selected materials, HNBR and FKM, are commonly used elastomers at Volvo CE. Material tests performed at RISE are for three different load cases:uniaxial tension test, planar tension test and biaxial tension test. Quasi-static analyses are performed in ANSYS. Hyperelastic materials need different constitutive models, hyperelastic material models, to describe their material behavior and these are defined in terms of a strain energy density function.However, the challenge is to determine the material constants in the equation, to characterize the material properties, by processing test data. Research questions answered are ‘’What material tests are needed for hyperelastic materials?’’, ‘’How is the test data converted to work as input to ANSYS and obtain material constants?’’ and ‘’How is an appropriate material model selected for simulation in ANSYS?’’. The study shows the importance of that material test represents the condition the application will experience. It should capture material behavior at the specific frequency, strain amplitude and temperature range for the application. The expected strain range and deformation modes that will play a functional role in the application should be considered in the material testing. Material constants can be determined from test data separately or simultaneously. Test data from at least one deformation mode is required, but one can't accurately predict full deviatoric behavior of hyperelastic material models by using one mode. If data only is used for one deformation mode, simulations in other deformation modes can yield erroneous results. It is therefore recommended to use several deformation modes. For applications with more complex load cases more deformation modes are needed. Generally, recommended tests are uniaxial tension test, planar tension test and biaxial tension test due to homogenous deformation is achieved. It is important to verify the material model before analysis. Using test data from one deformation mode can still provide a good fit. In the cases investigated verifications of the material model Yeoh 3rd order show that the fit obtained by only using uniaxial tension test data and using test data from three tests doesn’t seem to differ. Both uniaxial tension test data and test data from three tests give agood fit when simulating the tests with this material model. The benefit of using test data from three tests is questionable due to costs. It seems that only uniaxial tension test data could have been used as it provided a good fit. Moreover, test data must be processed to work as input to ANSYS. ANSYS requires engineering stress-strain test data for hyperelastic materials besides from the volumetric test, where true stress strain is required. The biaxial tension state which is realized with so called Bulge test thus needs to be converted to engineering stress. Then, test data needs to be adjusted to account for effects such as hysteresis and Mullin’s effect, where choice of curve and a process zero-shift must be done. Hyperelastic material models have different validity for different strain ranges. The selected material model was Yeoh 3rd order, which showed be a good fit for both the materials, HNBR and FKM, in strain range 30 %. The curve fit is based on three tests. The selection was based on the material model with lowest relative error with stability. Material constants were obtained for that material model, and these were used in simulations. Material models tends to be unstable for strains outside the test data. Simulations of seals with fluid pressure were performed for different pressure and stretch of the seal. If the contact pressure is larger than fluid pressure in the seals no leakage will occur.