I note that the paper concluded that a 3-vane setup with "streamlined" input (called a "bellmouth" in this forum) and a "central body" ("Type E1" in its Figure 3 
) was superior to 3-vane without central body (Type C1) because it completely killed the "swirl velocity component". The "central body" is a cylinder, small in diameter compared to the output tube, axially centered in the output tube, with a rounded leading/entry/downward-pointing end and long-tapered trailing/exit/upward-pointing end -- similar to the nacelle of modern wind turbines. BTW, I was able to find the paper/PDF with a web search for its title, "Application Of Deswirl Device In Cyclone Dust Separator".
) was superior to 3-vane without central body (Type C1) because it completely killed the "swirl velocity component". The "central body" is a cylinder, small in diameter compared to the output tube, axially centered in the output tube, with a rounded leading/entry/downward-pointing end and long-tapered trailing/exit/upward-pointing end -- similar to the nacelle of modern wind turbines. BTW, I was able to find the paper/PDF with a web search for its title, "Application Of Deswirl Device In Cyclone Dust Separator".