Finite Element Modeling of a Thermoplastic Seal at High Temperature and Pressure Jorgen Bergstrom 1, Ph.D. Brun Hilbert 2, Ph.D., P.E. Email: jorgen@polymerfem.com 1 Veryst Engineering, LLC Needham, MA 2 Exponent Inc. Natick, MA
Outline of Presentation Description of the Problem Mechanical behavior of Teflon New UMAT for Teflon Calibration and validation of the UMAT FE simulation of Teflon gasket Conclusions
Threaded Connection Simulation Steel Pipe Teflon Seal Steel Coupling
Threaded Connection Simulation The two steel pipes are threaded together The assembled pipe transports gas at high temperature and pressure The Teflon seal acts as a secondary seal How much pressure can the Teflon seal take before leaking?
Threaded Connection Simulation What is the pressure between the Teflon seal and the steel pipes at different temperature and times?
Finite Element Modeling Geometry and BC Loading Specification Material Representation Finite Element Modeling The results from FEA are only as accurate as the input values The most difficult part is typically the material representation
Geometry and Boundary Conditions Axisymmetric representation The model contains 3 parts: Lower steel pipe Upper steel pipe Teflon seal
Geometry and Boundary Conditions The analysis is performed using ABAQUS/Explicit The 3 parts are initially overlapping No contact activated
Geometry and Boundary Conditions *Expansion, type=ortho,, zero=<t0>, dependencies=1 ** alpha11, alpha22, alpha33, T, field 6.000e-4, 0.000e-5, 6.000e-4, <T0>, 1 1.242e-5, 1.242e-5, 1.242e-5, <T0>, 2 *Initial conditions, type=temperature alln,, <T0> *Initial conditions, type=field, var=1 alln,, 1 FIRST STEP: *Temperature intpipe.alln alln,, <Tcool< Tcool> *Field, op=mod, var=1 alln,, 1 SECOND STEP: *Temperature intpipe.alln alln,, <T0>
Experimental Data for Teflon Uniaxial Tension T=20 C
Experimental Data for Teflon Stress Relaxation Triaxial Compression
Mechanical Behavior of Teflon The Response is Characterized by: Creep Stress relaxation Temperature dependence Yielding Large deformations How can the Teflon material be modeled using ABAQUS?
Constitutive Model Description *Material, name=teflon *User material, constants=17 100, 1, 0, 1.11e-4, <T0>, 6.0, 3.5, 600 100, 200, 3.5, 600, 0.0, 1.35, 3.0, 165.0 0.01 *Depvar 18 *Density 2200e-12
Constitutive Modeling Viscoplastic Flow Time-dependent response Chain slippage driven by a stress driven representation Equilibrium response 8-chain model Viscoelastic Flow: 8-chain model Modeled with a reptation based energy activation representation
Response of the Equilibrium Network ( ) $ 1 ve* lock ( ) L! /! 0 µ A " ve* ve TA = dev % & + # % J $ 1& ve * $ 1 J! L ' B ( ' ( 1 ( lock 1/! ) 8-chain model L -1 (x)) is the inverse Langevin function Hyperelastic representation Micromechanism inspired Accurately predicts large strain multiaxial deformations 0 "! 0 $! # µ A (! ) = µ A exp % & '! base ( ve ve J = det! # F " $ ve* ve! 2/3 ve ve ( J ) ( ) B = F F ( ve ) ve* *! = tr B / 3 T
Constitutive Modeling The details of the material model are available in: A A Constitutive Model for Predicting the Large Deformation Thermomechanical Behavior of Fluoropolymers,, J.S. Bergstrom, L.B. Hilbert, Mechanics of Materials, vol 37, pp. 899-913, 2005.
Constitutive Modeling Available for both ABAQUS standard and explicit Physically motivated Incorporates: Rate effects Viscoelasticity Viscoplasticity Permanent deformation Temperature effects Volumetric creep
Determination of Material Parameters Calibration Verification Evaluation 1) Calibrate model to available uniaxial data (different strain rates, temperatures, and strain histories) 2) Simulate multiaxial tests using the calibrated model 3) Evaluate performance of the model
Material Parameters for PTFE µ! Network A " # A base lock = = s = B 8.52 MPa = = 71.2 C 5.0 o 500 MPa Network B 12.97 Viscoelastic flow C = # 1 m = 9.11 n = 28.9! " base vol 0 = 19.0 MPa = 152 GPa Plastic flow a = 0.046 b = 1.0! = 19.0 MPa
Glass Fiber Filled PTFE
PTFE
PTFE Triaxial Compression (T=20 C)
Limit of application Deformation The model works for arbitrary multiaxial deformation states The model has been tested for deformation rates between 10-5 /s to 1/s Temperature ranges The model has been tested for temperatures between 20 C C and 200 C Software implementations Implemented and tested for ABAQUS (both Explicit and Implicit)
The Need for Multiaxial Testing Uniaxial testing only probes one aspect of the material models Many models can predict uniaxial deformation, only a few can predict multiaxial loading In many important applications the applied load is multiaxial
Verification: Punch testing Specimen geometry: Diameter=6.4 mm Thickness=0.5 mm
Experimental Data
Model Predictions
Threaded Connection Simulations
Threaded Connection Simulation
Threaded Connection Simulation
Threaded Connection Simulation
Threaded Connection Simulation
Conclusions FE analysis generally requires 3 parts: Geometry specification Load and boundary conditions Material models The specification of the material model is often the most difficult part
Conclusions Accurate FE analysis of polymers requires: Careful experimental testing Material model calibration Material model validation Specialized user material models (UMATs)) can provide accurate predictions for many tough problems
Exponent UMAT models for Elastomers Filled or unfilled Semi-crystalline glassy polymers Polyethylene Fluoropolymers Foams Silastic foam
Special Thanks: Shell Exploration and Production Company Sina Ebnesajjad at DuPont Fluoroproducts Pradip Kaladkhar at DuPont Fluoroproducts