Finite Element Modeling of Thermoplastics at Different Temperatures

Transcription

1 1 Finite Element Modeling of Thermoplastics at Different Temperatures J.S. Bergström, Ph.D. Veryst Engineering, LLC

2 2 Outline of Presentation Experimental Data for Polycarbonate (Lexan) Calibration of Plasticity and Creep Models Parallel Network Model: Theory Calibration FE Simulation

3 3 Problem Statement Limited experimental data available from the raw material manufacturer (non-linear viscoplastic material) What material model to select? Goals: Demonstrate how to effectively use both creep and uniaxial tension data Demonstrate the material model calibration procedure Demonstrate the use of an advanced user-material model in Abaqus

4 4 Experimental Data Uniaxial tension at different temperatures

5 5 Experimental Data 3000 Young's modulus as a function of temperature 2500 Young's Modulus (MPa) f(x) = -3.39x R² = Temperature (deg C)

6 6 Experimental Data Uniaxial creep experiments were performed at stresses between 15 MPa and 40 MPa Region for creep experiments

7 7 Experimental Data Creep strain as a function of time for different stress values 27 days

8 8 Experimental Data Creep compliance as a function of time and stress The material is NOT linear viscoelastic 27 days

9 9 Experimental Data Uniaxial fatigue data

10 10 Material Model Selection Linear viscoelasticity not-sufficient since the material is non-linear viscoelastic Elastic-Plastic with rate-dependence Elastic-Plastic with creep Viscoplastic user-material model Parallel Network Model from the Veryst Engineering PolyUMod library What material model works best?

11 11 Metal Plasticity Model Calibration *Elastic *Plastic *Rate Dependent, type=power law

12 12 Metal Plasticity Model The rate-dependent plasticity model does not capture the non-linear creep response very well

13 13 Metal Creep Model Calibration *Elastic *Plastic *Creep, law=strain The material model calibration was performed using the MCalibration software from Veryst Engineering. This software can calibrate any material model in Abaqus.

14 14 Metal Creep Model The plasticity-based creep model does not capture the non-linear creep response at high stresses

15 15 Parallel Network Model (PNM)* Neo-Hookean with piecewise linear temperature dependence Neo-Hookean with piecewise linear temperature dependence Power-law flow with exponential yield evolution and piecewise linear temperature dependence *The PNM is commercially available for Abaqus/Standard and Abaqus/Explicit from Veryst Engineering

16 16 Viscoplastic Flow Behavior Power-law flow rate: γp = viscoplastic flow rate τ = driving shear stress τhat = flow resistance m = flow activation exponential fp = 1 fe = yield evolution factor fθ = piecewise linear temp factor Exponential yield evolution: ff = final yield evolution factor εpmax = max plastic strain εhat = characteristic transition strain

17 17 Damage Model Rate of damage accumulation: No damage (D=0) at time t=0 σe = Mises stress [t0, σref, m] = material parameters Element failure when D > 1 This damage model enables fatigue predictions

18 18 Material Model Calibration

19 19 Material Model Parameters Network A Neo-Hookean hyperelastic: [µ,κ] [174 MPa, 2000 MPa] Network A temperature dependence: 5 pairs of [T, f] values [(243 K, 0.64), (273 K, 0.53), (296 K, 0.76), (333 K, 0.67), (363 K, 0.60)] Network B: Neo-Hookean hyperelastic: [µ,κ] [601 MPa, 2000 MPa] Network B hyperelastic temperature dependence: 5 pairs of [T, f] values [(243 K, 1.11), (273 K, 1.24), (296 K, 1.28), (333 K, 1.12), (363 K, 1.30)] Network B flow: [tauhat, m] [14.2 MPa, 33] Network B flow evolution: [ff, epshat] [3.2, ] Network B flow temperature dependence: 5 pairs of [T,f] values [(243 K, 0.006), (273 K, 0.023), (296 K, 18.1), (333 K, 4243), (363 K, 12784)] Damage accumulation model: [t0, σref, m] [5e-6 s, 50 MPa, 1]

20 20 Model Predictions The PNM captures the strain-strain response in monotonic loading at different temperatures.

21 21 Creep Predictions The PNM captures the non-linear creep response.

22 22 Creep Predictions Comparison between experimental and predicted creep behavior

23 23 Damage Model Predictions Predicted damage accumulation during a monotonic uniaxial tension simulation

24 24 Damage Model Predictions Predicted fatigue behavior from the damage model

25 25 Exemplar FE Study Fixed end PC hook loaded with an increasing force The top of the hook was fixed Abaqus/Explicit with 27k C3D8R elements Failure at a critical damage level Room temperature Force

26 26 Exemplar FE Study Large pre-existing notch

27 27 Damage Model Predictions Configuration just before final failure

28 28 Summary Thermoplastic materials are non-linear materials: viscoplasticity, creep, failure Abaqus built-in plasticity models with strain rate dependence and creep can capture some of the experimental data Advanced user-material models (UMAT/VUMAT) can be used to accurately predict almost any aspect of isotropic and anisotropic polymers