You are on page 1of 2

MEMA/CSES 648 Spring 2002

Nonlinear Finite Element Analysis

Date: April 4, 2002 Due: April 11, 2002

ASSIGNMENT No. 8
Formulation of Axisymmetric Problems
and Stability of Temporal Schemes

Problem 1. Symmetries present in a problem can be exploited both in formulating and solving
the problem. Axisymmetric problems have cylindrical geometries (see Figure below) and exhibit
symmetry (in addition to the geometry) in material composition as well as loading about the
axis of symmetry, which is the centerline of the geometry. Such problems use the cylindrical
coordinate system for analytical description of the governing equations. For an axisymmetric
problem, all quantities (hence, the solution) are independent of the circumferential coordinate θ.
This exercise is aimed at developing the weak forms and finite element model of the equations
governing flows of viscous, incompressible fluids in axisymmetric geometries. For axisymmetric
viscous incompressible flows, the Navier-Stokes equations take the form
1 ∂ ∂vz
(rvr ) + =0 (1.1)
r ∂r ∂z

Dvr 1 ∂(rσrr ) ∂(rσzr ) σθθ


´ µ
ρ0 = ρ0 fr + + −
Dt r ∂r ∂z r
(1.2)
Dvz 1 ∂(rσzr ) ∂(rσzz )
´ µ
ρ0 = ρ0 fz + + (1.3)
Dt r ∂r ∂z
where

σrr = −P + 2µDrr , σθθ = −P + 2µDθθ ; σzz = −P + 2µDzz , σzr = 2µDzr (1.4)


∂vr vr ∂vz ∂vr ∂vz
Drr = ; Dθθ = , Dzz = , 2Dzr = + (1.5)
∂r r ∂z ∂z ∂r
Develop the semidiscrete penalty finite element model of the equations. Define all elements
of the finite eleemnt model.
2

Problem 2. A study of the numerical stability of time approximation schemes is the topic of
this exercise. Recall that all time approximation schemes ultimately reduce a set of ordinary
differential equations, in matrix form,

[M ]{ü} + [C]{u̇} + [K]{u} = {F } (2.1)

to a set of algebraic equations of the form (for homogeneous case of {F } = {0})

{u}n+1 = [T ]{u}n (2.2)

Here the subscipts refer to time step number. The stability requirement (i.e., in order that the
error introduced by the time-approximation scheme does not grow unboundedly as we solve (2.2)
repeatedly, marching from one time step to the next) is that the maximum eigenvalue of the
matrix [T ] be less than unity. Establish the stability criterion (i.e., determine the critical time
step in terms of the maximum eigenvalue and parameters of the time-approximation scheme)
for (a) the α−family and (b) Newmark family of approximations (see the notes for the criteria).
It is sufficient to work with the one-dmensional forms of Eq. (2.1):

C u̇ + Ku = 0 and M ü + Ku = 0 (2.3)