Direct numerical simulation (DNS) allows for the studying of all the scales of turbulence, without assuming a model. However, even at low Reynolds, and even using nondimensional models, such as those used in this paper, computation can be expensive. Thus, not many people attempt to implement DNS, hence the importance of reading this paper. The results are quite convincing, reproducing several vortex patterns observed in experiments.
I recommend this paper to physicists and engineers who work on composed chimneys or other devices describable with similar physics. Advanced students and researchers may also profit from the description of the implementation of the numerical techniques and the cited literature.