The classical $k-\varepsilon$ model and a new implementation of wall-laws have been evaluated for unsteady turbulent flows past square and circular cylinders. We show that good numerics can lead to a serious improvement of the results. In the case of the flow past a square cylinder (Lyn's test case), results agree surprisingly well with the experimental data. In the analysis of the vortex shedding flow past a circular cylinder also the agreement is quite good for the sub-critical regime but not entirely satisfactory after the critical Reynolds number (i.e. the drag crisis phenomenon