# Courant Condition Stability

Radioss uses elements with a lumped mass approach. This reduces computational time considerably as no matrix inversion is necessary to compute accelerations.

The integration scheme used by Radioss is based on the central difference integration scheme which is conditionally stable, that is, the time step must be small enough to assure that the solution does not grow boundlessly.

The stability condition is given in the last sections. For a system without damping, it can be written in a closed form:(1)
$\text{Δ}t\le \frac{2}{{\omega }_{\mathrm{max}}}$
Where, ${\omega }_{\mathrm{max}}$ is the highest angular frequency in the system:(2)
$\mathrm{det}\left(K-{\omega }^{2}M\right)=0$

Where, $K$ and $M$ are respectively the stiffness and the mass matrices of the system.

The time step restriction given by 式 1 was derived considering a linear system (Explicit Scheme Stability), but the result is also applicable to nonlinear analysis since on a given step the resolution is linear. However, in nonlinear analysis the stiffness properties change during the response calculation. These changes in the material and the geometry influence the value of ${\omega }_{\mathrm{max}}$ and in this way the critical value of the time step.

The above point can be easily pointed out by using a nonlinear spring with increasing stiffness in Body Drop Example. It can be shown that the critical time step decreases when the spring becomes stiffer. Therefore, if a constant time step close to the initial critical value is considered, a significant solution error is accumulated over steps when the explicit central difference method is used.

Another consideration in the time integration stability concerns the type of problem which is analyzed. For example in the analysis of wave propagation, a large number of frequencies are excited in the system. That is not always the case of structural dynamic problems. In a wave shock propagation problem, the time step must be small enough in order to excite all frequencies in the finite element mesh. This requires short time step so that the shock wave does not miss any node when traveling through the mesh. It follows that the time step should be limited by the following relation:(3)
$\text{Δ}t\le \frac{{l}_{c}}{c}$
Where,
${l}_{c}$
Characteristic element length, representing the shortest road for a wave arriving on a node to cross the element
$c$
Speed of sound in the material
$\text{Δ}t$
Time step
The condition 式 3 gives a severe time step restriction with respect to stability time step, that is, $\text{Δ}t<\frac{2}{\omega }$ . It can easily be shown that for a simple case of a bar element, the two expressions 式 3 and 式 1 are equivalent.
If a uniform linear-displacement bar element is considered, (図 2), and a lumped mass formulation at the nodes is used, the highest frequency of this element can be obtained by a resolution of an eigen value problem:(4)
$\left[K\right]\left\{X\right\}={\omega }^{2}\left[M\right]\left\{X\right\}$
(5)
$\mathrm{det}\left(\left[K\right]-{\omega }^{2}\left[M\right]\right)=0$
For a lumped mass bar, you have:(6)
$\left[K\right]=k\left[\begin{array}{cc}1& -1\\ -1& 1\end{array}\right]$
(7)
$\left[M\right]=m\left[\begin{array}{cc}1& 0\\ 0& 1\end{array}\right]$
Where, $m$ and $k$ are respectively the nodal mass and stiffness of the bar:(8)
$k=\frac{EA}{l}\text{ };\text{ }m=\frac{A\rho l}{2}$
Equation 式 5 yields:(9)
${\left(k-m{\omega }^{2}\right)}^{2}-{k}^{2}=0$
then:(10)
$\omega =\sqrt{\frac{2k}{m}}$
which can be simplified with 式 8 to obtain:(11)
$\omega =\frac{2}{l}\sqrt{\frac{E}{\rho }}=\frac{2c}{l}$
Where, $c$ is the speed of sound in the material and its expression is given as:(12)
$c=\sqrt{\frac{E}{\rho }}$
with $\rho$ the material density and the Young’s modulus. Combining 式 11 and 式 1, you obtain:(13)
$\text{Δ}t\le \frac{l}{c}$

This relation is that of 式 3 and shows that the critical time step value in the explicit time integration of dynamic equation of motion can be carried out by the interpretation of shock wave propagation in the material. This is shown for the first time by Courant. 1 In spite of their works are limited to simple cases, the same procedure can be used for different kinds of finite elements. The characteristic lengths of the elements are found and 式 3 is written for all elements to find the most critical time step over a mesh. Regarding to the type (shape) of element, the expression of characteristic length is different. 図 2 shows some typical cases for elements with one integration point.

1 Courant R., Friedrichs K.O., and Levy H., 「About the partial Differenzensleichungen Bogdanova of Physics」, Math.A nn., Vol. 100, 32, 1928.