hi i am doing a project aeroacoustic simulation in which i am using Mac Cormack scheme (yes its only 2nd order accurate). i am using linearised euler equations in the primitive form, and forward predictor, backward corrector scheme in 2 Dimensional cartesian grid.and i have a PML domain around my main computaional(from where i get the result). the PML equations i am using r the ones suggested by Mr Fang Q Hu in his paper " on absorbing boundary conditions for linearised euler equations by a perfectly matched layer", in which he splits all the veriables in the main domain(4 variables) into 2 in the pml doamin(thus 8 variables) and adds viscocity terms to the equations. at the PML and main region, i am intrpolating the equations

Now the problem i am facing is that after few (800-900)iterations at the boundary of the main computaional domain on one side the values dip and on the other side they rise... i will like to know if it is because of the reason that i am using such a scheme insteat of some central difference scheme. or there is some other problem.

thanx in advance. vineet

Dear Vineet,

In Hu's paper, a filter process is necessary, where a forth-order central difference scheme was used. MacCormack scheme is 2-order and dispersive in nature therefore an effective filter is also needed. In addition, you may try the staggered grid to suppress the spurious oscillations.

Hope it's helpful

Paul