Basic Steps in FEA


Notation

Explanation

U = N.d

  • U: column matrix of unknown displacements within discrete elements (translations & rotations)
  • d : column matrix of known displacements at nodes
  • N: matrix of shape functions relating known d to unknown U
  • N: derived by Pascal's triangle depending on degree of polynomial
  • N: =1 at node I; =0 at others

e = B.d

  • e : column matrix of strains
  • d : column matrix of known displacements at nodes
  • B: strain-displacement relation; derivative of N w.r.t. d (x,y,z,xy,yz,zx)
  • B: =derivative

s = D.e

  • Stress-strain relation
  • s : column stress matrix
  • e : column strain matrix
  • D: stress-strain relation matrix
  • D: depends on material - linearly elastic; non-linearly elastic; inelastic
  • D: derivation is assumed elastic or discretised into linear regions for non-elastic
  • Plane strain: e z=0
  • Plane stress: s z=0
  • Axisymmetric: s q =0

F = A.s

= K.d

  • F: force excitation column matrix
  • A: area matrix
  • s : column stress matrix
  • d : column matrix of known displacements at nodes
  • K: stiffness matrix, load-displacement relation
  • K: 2 types - global stiffness matrix for all nodes of structure; local or element stiffness matrix for element nodes only
  • K: thus transformation required
  • K = ò BT.D.B dV (or dA, dl)

Solution

  • Derivation of K matrices
  • Apply excitation loads
  • Apply B.C.
  • Using F=K.d , solve for d (degrees of freedom at nodes)
  • Backward solution process:
  • Using e = B.d , solve for e
  • Using s = D.e , solve for s
  • For both dynamic & static conditions, & total effect

Solution techniques

  • Since all variables are conveniently formulated into matrices, use matrix operations
  • Inverse matrix: direct solution, but high computation needed
  • Cramer's rule: limitations
  • Gaussian elimination: frequently used, less computation
  • Numerical techniques: finite difference method, Jacobi, Gauss-seidel, Runge-Kutta, etc.

Hosted by www.Geocities.ws

1