Possibly one of the difficult task to use turbulence model is to specify the appropriate BCs. For specific case like flow through pipe, turbulence intensity (%) can be specified as I = 0.16Re^(-1/8) while the corresponding length scale can be specified by hydraulic diameter of pipe. When this is not the case, an ad-hoc turbulence intensity in the range of 4-20% (low to high) as in most industrial applications, turbulence intensity falls in this range.

Possibly one of the difficult task to use turbulence model is to specify the appropriate BCs. For specific case like flow through pipe, turbulence intensity (%) can be specified as I = 0.16Re^(-1/8) while the corresponding length scale can be specified by hydraulic diameter of pipe. When this is not the case, an ad-hoc turbulence intensity in the range of 4-20% (low to high) as in most industrial applications, turbulence intensity falls in this range.

I think that David Harland's answer is more accurate, because the expression I_tur = 0.16Re^(-1/8) is valid for developed flow inside a cirular section pipe. This formula is an expresssion obtained from classical solutions for developed flow inside of these kind of pipes (see e.g. books as Turbulence from Hinze or Boundary-Layer Theory: H. Schlichting).

Regarding the turbulence intensity level in pipe flow, I have written a paper on this with a colleague: 

https://www.researchgate.net/publication/308938793_Scaling_of_turbulence_intensity_for_low-speed_flow_in_smooth_pipes

Article Scaling of turbulence intensity for low-speed flow in smooth pipes

I revisited your question and I think that if you must to model transition from laminar to turbulent flow, you have big trouble if you use the common RANS models availble in CFD codes. None of them take into account the transition in an accurate way. Only SST model (only two equation not four ones, I think you have a mistake in your question) takes into account transition but in a very rudimentary way (an intermittence coeff. plus an amortiguation (damping) function only). Instead, if you need only to model the turbulent flow with swirling, the BCs at the inlet suggested by David could be a good option.

I have added the results from the paper to CFD-Wiki:

https://www.cfd-online.com/Wiki/Turbulence_intensity