The p and hp finite element modeling
of thin structures
Manil Suri
Dept. of Mathematics and Statistics, UMBC
Tuesday, October 30, 2001 13:00-14:00,
Room 145, NIST North (820)
Gaithersburg
Tuesday, October 30, 2001 11:00-12:00,
Room 4550
Boulder
Abstract:
The classical h version finite element method uses mesh
refinement with a fixed polynomial degree (usually 1 or 2) to increase
accuracy. In contrast, the p and hp versions allow the
user to change the polynomial degree for increasing accuracy. We
survey these methods and show how this flexibility can be exploited
to counter four difficulties that occur in the approximation of
problems over thin domains (such as plates and shells) which arise
often in engineering practice. These difficulties are:
(1) control of modeling error,
(2) approximation of corner singularities,
(3) resolution of boundary layers, and
(4) control of locking.
Our guidelines enable the efficient resolution of these
difficulties when a p/hp code is available.
Contact: A. J. KearsleyNote: Visitors from outside NIST must contact
Robin Bickel; (301) 975-3668;
at least 24 hours in advance.
|
