radial diffusers

[ripp]

Hi,

who can say something about the current radial diffusers?

Current RC turbines have a very small annular space.

What does the theory say about this?

My Xicoy X45 has none at all, but it is very powerful.

[

www.youtube.com/watch?v=sUoqDp-N5C8

I hope that Chris and others can clarify this.

Thanks in advance

Ralph

[finiteparts]

Hi Ralph,

I am assuming that you are referring to the distance between the impeller exducer diameter and the leading edge of the diffuser vane (Vaneless space diameter ratio)?

I did some digging and there is very little guidance given on sizing the vaneless space between the impeller discharge and the diffuser vanes.  Many of the sources point to a paper that Colin Rodgers put out that can be found here:

asmedigitalcollection.asme.org/GT/proceedings/GT1982/79566/V001T01A003/234803

As you can see in figure 13,



that he achieved the highest performance with a vaneless space diameter ratio of 1.125, but as stated in Cumpsty's "Compressor Aerodynamics" book, the universality of this result is uncertain.  I think his comment at the  end of the section discussing the vaneless space effect is very interesting;

    "These specific test data indicate that diffuser

    flow range was not materially influenced by vaneless

    space diameter ratio, but diffuser losses apparently

    increased when the diameter ratio departed from the

    optimum value of 1.125."

Dr. Japiske suggested in a presentation to use a vaneless space diameter ratio of 1.03 to 1.07 for lower discharge Mach numbers, and 1.10 to 1.12 if the impeller discharge Mach number is large...but it wasn't clear where the split between lower and higher Mach numbers were to be defined.

Rene Van den Braembussche states in his book ("Design and analysis of Centrifugal Compressors"), "No clear criterion is available to define the optimal diffuser leading edge radius ratio.   A value of R3/R2 between 1.05 and 1.12 is generally accepted."  He then discusses some of the noise and blockage concerns that occur due to a very close spacing of the impeller and diffuser.

The vaneless space gives time for the impeller discharge flow to mix and diffuse prior to encountering the blockage of the diffuser vanes.   The mixing is due to "wake-Jet" structure of the flow from each impeller passage.   Due to secondary flow effects, the discharge flow is far from uniform and one the forward portion of the passage, there is a region of low energy (usually separated) flow, which is termed the "wake".   On the back side of the passage (or the pressure side), the flow is more well behaved and uniform, thus the term "jet".   The outflow of the impeller is seen as a periodic pulsing flow when viewed from the diffuser vane.  This pulsing flow swings the incident angle due to significant changes in the radial velocity component.   It is thought that if there is some small amount of time and diffusion that can allow this to "mix" into a more uniform flow structure before encountering the diffuser vanes, the more efficiently the diffuser will operate.

The other thought is that if the vaneless and the semi-vaneless space (area between one vane leading edge to the next vane leading edge) can diffuse the discharge flow down to a lower Mach number before entering the vane passages, then the diffuser losses should be lower.  In the work done by Runstadler back in the 1960s,

    apps.dtic.mil/sti/citations/AD0865300

the impact of the inlet Mach number on the diffuser pressure recovery shows a general trend to reduce diffuser pressure recovery with Mn increases.  The blockage ratio is also shown to exert a very strong impact on pressure recovery.  The complexity of the diffuser inlet flowfield makes this a challenging argument.



In the above plots, we can see the impact of Mach number on the diffuser performance curves for a fixed geometry.   In the furthest plot to the right, you can see the stronger influence of the blockage verses the Mach number.   Since it is very hard to say if the flow entering the diffuser will be more disrupted by the boundary layer growth in the relatively longer flowpath due to a larger vaneless space, or the off-incidence flow losses due to the poorly mixed out flow for a diffuser with no real vaneless space, it is tough to make a recommendation.

For my designs, I tend to default to the 1.125 that performed well for Colin Rodgers...in my mind a single data curve is better than guessing.   But I admit that it may be not actually be optimized for my particular designs.  Now, as stated above, there is little guidance or evidence of this, so it is sort of an open question as to the vaneless space radius ratio impacts the overall performance of the diffuser stage.

I hope that helps,

Chris

[finiteparts]

I guess I hinted to the noise issue in the previous post, but I wanted to expound on that a touch. There is quite a bit of discussion in the technical literature on the diffuser vane proximity to the impeller causing vibration and noise. There are cases of Garrett / Airesearch impellers fatigue fracturing the trailing edge sections of the impeller blades due to excitation forces driven by the passing frequency of the diffuser vane (i.e., increased local static pressure). You can think of it as tapping the impeller blade a little bit every time it passes a diffuser vane. Even though they are small forces, when they hit a resonant frequency, they can excite flex modes on the impeller that will cause rapid high cycle fatigue.

As the vane is moved further away from the impeller, the local relative static pressure that the impeller vane "sees" is reduced and thus the driving force of the high cycle fatigue is attenuated.

The passing of the impeller vanes through these higher static pressure regions not only creates these local forces, but they are also noise creators, since noise is just pressure fluctuations. Noise and HCF forces are the same thing, just to different degrees.

There are equivalent design rules for noise or HCF driving forces to set the distance between axial compressor blades and their stators, fans and their stators, axial turbine blades and their NGVS, etc... passing frequencies are a big issue in turbomachinery and generally, if the distance between the rotating and stationary components are very low, then the risk of coupling and HCF fatigue is high.

- Chris

[racket]

Hi Ralph

The RC micro turbines all want to have the smallest diameter possible for the thrust produced , so they trade off things like radial spaces .

As Chris has mentioned , theres a lot of variation in recommendations .

I like to use 15% as I use more generously sized engine casings , so for my 175mm dia comp , 87.5mm radius X 0.15 = 13.1mm gap between wheel and diffuser tip for the vaneless space , the semi vaneless space before entering the diffuser throats adds even more distance , and importantly , time , before the air has to adjust to the vanes completely, theres a lot of diffusion done across "the space" .

Then with a larger casing most of the rest of diffusion is done radially before theres any disruption to flow going "around the corner".

The small RC casing will require more complex machining of diffusers to attain good efficiency , your engine has a pretty small comp wheel so radial spaces will also be very small , a 40 mm dia wheel even with my generous 15% gap will only have a 3mm gap , at 5% its only 1mm , unfortunately the airspeeds are the same for your wheel as a big one , but the time frame for airflows to settle are greatly reduced .

Probably because of the limited life spans of a RC engine they aren't as concerned about HCF .

Cheers
John

[ripp]

Hi Chris and John,

Thanks for your posts!

Jan 6, 2024 17:47:31 GMT -5 finiteparts said:

I guess I hinted to the noise issue in the previous post, but I wanted to expound on that a touch. There is quite a bit of discussion in the technical literature on the diffuser vane proximity to the impeller causing vibration and noise. There are cases of Garrett / Airesearch impellers fatigue fracturing the trailing edge sections of the impeller blades due to excitation forces driven by the passing frequency of the diffuser vane (i.e., increased local static pressure). You can think of it as tapping the impeller blade a little bit every time it passes a diffuser vane. Even though they are small forces, when they hit a resonant frequency, they can excite flex modes on the impeller that will cause rapid high cycle fatigue.

As the vane is moved further away from the impeller, the local relative static pressure that the impeller vane "sees" is reduced and thus the driving force of the high cycle fatigue is attenuated.

The passing of the impeller vanes through these higher static pressure regions not only creates these local forces, but they are also noise creators, since noise is just pressure fluctuations. Noise and HCF forces are the same thing, just to different degrees.

There are equivalent design rules for noise or HCF driving forces to set the distance between axial compressor blades and their stators, fans and their stators, axial turbine blades and their NGVS, etc... passing frequencies are a big issue in turbomachinery and generally, if the distance between the rotating and stationary components are very low, then the risk of coupling and HCF fatigue is high.

- Chris

I wasn't aware of the possibility of HCF.

Ralph

[monty]

Hi Ralph,

Have you ever taken the front cover off? Is there a radial diffuser stage? With a low PR the discharge Mach number is low enough you can turn the flow without huge loss, then just use a single axial diffusion stage.

Monty

[ripp]

Hi Monty,

I suspect the Diffuser looks very similar to the Kolibri Turbine ( Ø 55mm )



www.youtube.com/watch?v=aIBtZZOZ_54&t=1518s


X45 Turbojet



www.youtube.com/watch?v=sUoqDp-N5C8

Technical data: Ø 60mm, PR 3.2 par, max 225,000 rpm, compressor wheel Ø 40-41mm (471m/s)

Ralph

[monty]

Ralph,

I know some of these engine manufacturers have moved to an axial design to get smaller diameters for the engine. No matter the diffuser design, the performance of these little engines has come a long way!

Monty

[finiteparts]

Ralph,

Those impeller vanes are definitely going to feel the presence of those diffuser vanes. I have no data, but my gut feel is that the vanes will be excited by the passing frequency of those diffuser vanes, and hopefully, there is no point in the operating envelope that overlaps a flex mode on the impeller.

That is too close for comfort for me....doesn't seem to pass the eyeball test. Just an opinion though. Turbomachinery issues often occur after some amount of operational time. Perhaps the saving grace for these small engines is there not getting 100's of hours.

- Chris

[ripp]

Hi Chris,

yes the full throttle part when operating such a turbine in an RC airplane is very low (the start and 2-3 full throttle overflights for a few seconds).

thanks Ralph