Langford Analytic · Knowledge Base

How to Perform Creep Analysis in FEA

The complete workflow for performing a creep analysis in finite element software — from thermal and structural model setup through creep law definition, time stepping and result extraction.

Creep & High-Temperature12 min read
how-tocreepFEAtime steppingnonlinearresults

When to use this guide

Use this guide when setting up and running a creep analysis in commercial FEA software (e.g. Abaqus, ANSYS, Nastran).

Step-by-step method

  1. Run a thermal analysis to determine the temperature field (if temperature is not uniform)
  2. Define temperature-dependent material properties: elastic modulus, yield strength, thermal expansion
  3. Define the creep constitutive model with calibrated parameters
  4. Set up the structural model with the correct boundary conditions and loads
  5. Apply the thermal field as a temperature load on the structural model
  6. Run an initial static step (no creep) to establish the initial stress state
  7. Activate the creep model and set the time stepping: total time, initial increment, maximum increment
  8. Run the creep analysis — monitor convergence at each time step
  9. Extract results: creep strain, total strain, stress, deformation at key locations
  10. Check energy balance and convergence indicators
  11. Assess creep rupture life using the stress and temperature results

Checks

  • The initial static step gives the correct stress distribution before creep
  • Time stepping is adequate — the creep strain per increment is not excessive
  • Convergence is achieved at every time step
  • Creep strain accumulates monotonically (no unphysical negative creep)
  • Stress relaxation occurs in highly stressed regions as expected

Common mistakes

  • Using too-large time increments, causing inaccurate creep strain accumulation
  • Not running an initial static step, so creep starts from zero stress
  • Using a creep model outside its calibrated temperature-stress range
  • Ignoring thermal-structural coupling when temperature gradients are significant
  • Not checking convergence — accepting non-converged results

Related Knowledge