Time accuracy of NITA - FRACTIONAL STEP in Fluent

Paolo Lampitella

Hello,

i'm trying to perform some accuracy tests of the NITA - FRACTIONAL STEP time integration scheme employed in Fluent using the Taylor Test Case, which is a double periodic array of vortex whose kinetic energy is decaying in time.

I performed several simulations with several grids, domains, boundary conditions, pressure interpolation schemes and space discretizations but i have misleading results.

Using exactly the same conditions cited in the paper (relative to this scheme used in Fluent):

Sung-Eun Kim Boris Makarov

An Implicit Fractional-Step Method for Efficient Transient Simulation of Incompressible Flows

AIAA 2005-5253

i obtain the same results cited there, a 2nd order accurate in time scheme.

But, just going outside the conditions cited there, just using one more point in the error curve, i always obtain only a first order time accurate scheme or even worst convergence.

So i'd like to know if someone has experience in this or even know if there are some other papers where the scheme has been more rigorously tested and with which compare my results. Obviously, i mean the scheme implemented in Fluent and not just the Fractional Step.

This scheme, in particular, seems to be a fully implicit, approximate factorization, fractional step method with incremental pressure approach. It also seems that the boundary condition for the pressure equation is simply dp/dn = 0 and i don't know if in this particular case it is correct (i think it isn't)

Thanks a lot, any help is appreciated