I can simulate fluid flows with existing opensource softwares such as openFOAM, gerris, basilisk and so on based on my discretion as some of these are better suited for certain classes of fluid flows. For example, I would use openFOAM if you have a complex geometry and it requires unstructured grid. I would use Gerris if the flow problem involves multiple fluid phases in relatively simpler geometry. Flow variables like velocity, temperature, pressure can be easily observed in the form of vectors, contours and plots at the end of these simulations.
I would start with making the geometry in Blender. If the geometry is simple it can be done using the native geometry within the flow solver. I would then import the geometry into the software, say in Gerris, and mesh it there with sufficiently fine mesh. Thereafter I would write a script which defines the initial conditions, models for viscous stress, gravity, boundary conditions on the domain and on the geometry boundaries. Finally, the calculation would proceed on 8 parallel cores with 16GB RAM (3.2 GHz per core). An adaptive mesh refinement technique is used so that the computing time is reduced by as much as 4 times.
Afterwards, a postprocessing of the results would be done which will include the plots of the variables of interest and a report. Depending upon the complexity of the geometry of the problem, the preprocessing time may vary from no time to 1 day. My experience shows that the calculation usually finishes in about 2 days including the postprocessing for average sized problems. See this example for an estimation of time (it takes 2 days on 4 cores): http://gfs.sourceforge.net/examples/examples/atomisation.html
Open-source codes are famous in the cfd industry/academic circles because of the accuracy they offer just because they are custom made and the source code is available for all to see. Therefore, the chances of bad CFD practices like smoothing the variables unnecessarily is strictly avoid