Published on: 

All centrifugal compressors have at least an impeller and a diffuser. But why include a diffuser when it is the impeller that does all the work and gives all the total pressure and total temperature rise? Furthermore, the diffuser always adds some total pressure loss, which always reduces the total efficiency. The reasons for the diffuser are several:

• Diffusers recover kinetic energy, which is excessive at impeller exit

• Diffusers regulate flow (a good impeller may operate at low flow by matching to a low-flow diffuser)

• Diffusers can be a significant structural element

• Diffusers can match the flow to the next element (set the velocity triangles) • Diffusers may “clean up” the flow field (more steady and uniform)

• Diffusers can reduce radial side loads

• Diffusers may potentially reduce noise and vibration

The reader might also wish to review the author’s article “Focus on Diffusion” (Turbomachinery International, July/August 2008, p. 26).


The first item above is the key point: Compressor impellers produce too much kinetic energy and not enough static pressure. We must therefore convert some of the kinetic energy by diffusion over to static pressure by allowing a greater flow area, hence lower velocities and higher static pressure.

Recall that the work input is given by Euler’s turbomachinery equation: Wx = U2CƟ2 Therefore, the values of CƟ2 will be quite high and the value of C2 2 entering the diffuser will thus be much higher still. This must be diffused down to an acceptable level. Diffusion is achieved by conservation of linear or angular momentum, the latter being the case for a vaneless diffuser.

Two examples of industrial diffusers are shown in Figures 1 and 2. Figure 1 is a channel or straight centerline diffuser, and it recovers kinetic energy and transforms it into static pressure rise by using a simple area increase to reduce velocity, and hence, static pressure is recovered by the Bernoulli principle.

Figure 2 illustrates a three-row airfoil diffuser which works with the same basic principle, but now the area increase is also brought about by changing the mean flow angle as well. Each diffuser can work well, and it is not uncommon to recover more than 70% of the kinetic energy as a static pressure rise.

The graph shows the effect of good diffuser matching to achieve low-flow operation; all cases in the graph are with the same impeller. This case is for a lowvolume, gear-driven, centrifugal compressor using a pressure ratio of 3.5 for the last stage; the same impeller can be used quite well from a flow rate of about 0.5 down to about 0.2 and perhaps even further with additional development. This represents a flow reduction of about 60%.

Seven major attributes were listed at the top of this article; several of these have been illustrated with the chosen examples. Further details are available in the author’s textbook on centrifugal compressors (Japikse, D., Centrifugal Compressor Design and Performance, Concepts ETI, Inc., Wilder, VT, 1996). In the next article for this series we will look at the optional elements and learn how they also leave their mark on stage performance.


Dr. Japikse will be teaching several weeklong seminars in 2012. “Experimental Techniques for T u r b o m a c h i n e r y Development”, is the first, on March 19-22, 2012 in White River Junction, VT. He will also be teaching some courses in SE Asia as well. Please visit