Piechna's book has a short chapter on train aerodynamics, so I though I'd add that info to this discussion.
Here is a simple comparison of how nose and tail cones effect the drag coefficient Cx.
This data comes from Wysocki's testing of train models ([1]), in this case a 250 km/h train.
Next, a more detailed comparison of different train sets.
Example
a is from Wysocki ([1]),
b and
c are from Gackenholz ([2]).
Here is the effect of each element of a loco (Wysocki [1]).
And the same for wagons/cars (Wysocki [1]).
Notice the effect of covering the underside.
All models were done with Reynolds number above the critical value (which most researchers assume to be greater than 10^5), based on the height of the train. Gackenholz ([2]) used Re > 10^5, Neppert and Sanderson ([3]) used Re = 4x10^5, while Wysocki ([1]) noticed that the drag coefficient does not change above Re = 5x10^5.
Sources:
[1] Wysocki Z.,
Badania aerodynamiczne zespołu trakcyjnego 4WE, Spr. Inst. Lot. nr 29/BA/76, 1976.
[2] Gackenholz L.,
Ergebnisse neuerer Untersuchungen zum Luftwiderstand von Fahrzeugen in Zugverband, Elektrische Bachnen, Heft 12/42, 1971.
[3] Neppert H., Sanderson R.,
Untersuchen zur Schnellbahn-Aerodynamik, Z. Flugwiss 22 (1974), p.347.