Numerical diffusion and the tracking of solute fields in system codes Part II. Multi-dimensional flows

In Part I of this series, it was shown that high order numerical methods can significantly reduce numerical diffusion during the simulation of 1-dimensional solute transport by system codes. The current paper extends these conclusions to multi-dimensional flows by showing that the use of such methods can greatly improve predictions of the solute transport in a vessel. A high order method, Smolarckievicz's flux corrected method, together with an improved numerical flux limiter, has been implemented in a system code (TRAC-PF1/MOD2) in order to perform the simulation of an experimental dilution transient. The analysis of the results shows that the method is able to predict the main local characteristics of the core inlet and solute concentration field whilst introducing very low numerical diffusion, which is smaller than the experimentally observed turbulent diffusion of the flow. This substantial reduction in numerical diffusion even permits us to see the effect on the core solute distribution by accurate prediction of the multi-dimensional flow field transporting the solute.