Spring 2010, MET 415 - FEA Applications I

Prof. Dave Johnson, psuprofdj@psu.edu
Penn State - Erie, The Behrend College

Homework Assignment 9B


Typical MET 320 Test Problem for FEA-1: 

You decide: 

Two sections of metal strap are 2 mm thick and are welded together along the 25o angle joint.

2D planar or 3D Solid or 3D shell ? Surface Area/thickness = (125mm)(50mm)/2mm = 3215 mm

The centered force P is 2000 N.
Material properties: structural steel 
Evaluate the normal and shear stress along the weld line.
Model with at least two bodies to include the weld joint.
Create a Simulation Coordinate System, oriented along the weld line to permit presentation of stress parallel to and perpendicular to the weld line.

Placement: origin on vertex, orient with the “weld line” edge. Reference for results: Plot SXY and SY using your new local coordinate system

Use “scoping” or "selecting" to see the real stress in each body separately.

Are these results accurate considering finite element mesh error ?

Make TWO models:


Turn in:

  1. Document your modeling decisions:
  1. Symmetry ?
  2. Linear or Non-linear material properties ? 
    • Small or Large deformation effects ?
    • Are load steps/substeps/gradual loading necessary ?
  3. How to apply the given load ?
  4. How to constrain the part ?
  5. How to evaluate the mesh error ?

For each model (the 2D plane stress and the 3D shell - label plots clearly):

  1. A plot showing the mesh 
  2. An "environment" plot showing all the loads and constraints on the model
  3. A plot CLEARLY showing the distribution of the shear stress parallel to the weld, along the weld line.
  4. A plot CLEARLY showing the distribution of the normal stress perpendicular to the weld, along the weld line.
  5. A plot showing the maximum deformation that occurs on the model.
  6. A plot (or plots) evaluating the mesh error

Summary:

  1. List all singularities which may exist on this model.
  2. Any hand calculations you used to validate and verify your model.
  3. Compare the two models (results quality, modeling difficulty, solution time).  Conclude - which model is the more appropriate ?