an:01571128
Zbl 0995.65086
Akhtar, J.
On the optimisation of discretising steps in the space and time domains along with over-relaxation parameter in the finite difference solution of the transient heat-flow equation
EN
COMPEL 19, No. 4, 956-973 (2000).
00069956
2000
j
65M06 35K05 65M50 65M15 80A20
grid optimization; error bound; transient heat flow equation; successive-overrelaxation method; finite difference; optimal overrelaxation parameter
Optimisation of the discretising steps in the space and time domains is studied for the evaluation of the corresponding optimum value of the overrelaxation parameter in the numerical solution of the transient heat flow equation using the successive-overrelaxation method in the finite difference code. No closed form solutions are available for the optimisation of a complete set of the involved parameters in such problems.
The present work deals quantitativly with the need for a more generalised closed form relation involving discretising steps in the space and the time domains for an optimal overrelaxation parameter. The maximum finite difference error and the number of iterations required to achieve a reasonable error tolerance in functional value are the two criteria used to obtain an optimised set of parameters. The effect of deviation from the optimised values of any of the involved parameters is shown by means of a model problem of a one dimensional diamond-IIa medium of 100 micrometer length and for a time duration of 1.24 micro-seconds.
Peter Matus (Minsk)