Contemplating an HX82 Turbojet

[monty]

John,

Regarding ABB. There are so many emissions requirements these days....A lot of them based on off design point operation. They may need more width to the map to avoid smoke limits. So many requirements....

Anyway, one thing I noticed, The CC3 numbers fall right on the Noel Penny map. They are similar trims. 1600 ft/s is about 80% design speed on the NP map...- 4.5 PR. That map is a good indicator of what you can do with low trim wheels.

Monty

[monty]

John,

Regarding the CC3/HECC paper....yeah. A lot to digest there. Especially the diffuser stuff. Obviously some opportunity to reduce the size of the engine. But the manufacturing complexity! Probably not something I'm going to be attempting. If I did something like that, I'd probably use aluminum brazing. I have some I'm going to try out in my kiln....just to see if I can do some interesting things with it.

A lot of hair splitting there too.. ;-)

Monty

[racket]

Hi Monty

Undoubtedly ABB have "considerations" , ....LOL, compromises .

The comp maps I have for several Garrett turbos with different wheel trimmings certainly indicate better PRs for the same rpm/tip speed for the low trim variants , but I haven't been able to find one in the 30s , 42 Trim is the smallest turbocharger wheel trim , 48 Trim is popular but ABB are ~52 Trim if I remember correctly .........its a minefield out there :-(

I'll be looking forward to seeing your final choice :-)

Cheers
John

[monty]

John,

Which one I choose depends on whether my inner hot rodder gets the better of me, and I decide to "Go Big" or I decide to be reasonable. I have that 861 wheel....If I put the wheel in the CNC lathe and put a lower trim on it..... I'd have to do something different for the bearings than what I have planned to fit that wheel though. It's a tall one.

I wish I knew the limit load on this turbine... You don't have a map of similar sized one somewhere do you?

I may "Go Big" so I can find the limit. Then back up and figure out what the ideal should be.

Monty

[racket]

Hi Monty

The TV91 map is the F Trim Garrett wheel 129/106mm , same inducer as the G Trim at 129/111mm , so probably ~10% difference in flow area at the exducer .

The GT6041 has 25 mm inducer tip height at 130mm dia and ~119mm exducer





I'll dig out some "thoughts".

Cheers
John

[racket]

Hi Monty

At a 2:1 PR the smaller TV91 has roughly the same flow as the larger Gt6041 , the only difference is the scroll A/R 1.7 vs 1.47 , the GT6041 , if my memory serves me well, was suppose to come out with a 1.7 A/R scroll, but I don't ever remember seeing one on a map .

The TV91 scrolls appear to be running choked as the flow flat lines after 2:1

The GT6041 has a more modern wheel design .

If we're trying for max flow one would feel the exducer has to be running choked so that we are getting as much through as possible , but because density is dropping faster than velocity increase , theres a trade off to be had , less dense gases need more area , but if we clip the wheel the gas deflection and horsepower production drops off , and to compensate we need more "impulse", but again trying to get that balance of power producing deflection and ability to get the gases into the wheel rears its head .

Ideally I guess the equal split of energy production is the way to go , but the wheel inducer width is sorta fixed by what we have , though I have felt its probably worth looking at widening the NGV axially to lower its angle to improve the gas deflection whilst creating enough flow area into the wheel for choked gases from the NGV , the GT6041 at 25mm wide inducer compared to the 19mm of the TV91, a 31% increase , would allow much lower approach angle .

The GT6041 ( Performance Model) is different to the GT6041 (Industrial) turbine wheel , its closer to the TV94/HX82 at 21.6 mm tip height at 130 mm inducer with a 113mm exducer , the Performance GT6041 is for petrol engine whilst the Industrial is for a diesel , gas temps/density undoubtedly playing a part.

LOL.........I've got myself completely lost with this :-)

One of the problems I've had is I'm not certain if the temperature problems I've experienced at higher P2s is a combustion issue or a turbine one , T2 temps indicate reasonable comp efficiency which shouldn't require a lot of turb temp to balance the power requirement .

My combustors are just too short for their widths which is a sure fired recipe for problems .

Cheers
John

[monty]

John,

For an ideal turbine situation, I'd like something that looks more like that Garrett turbine I posted. The inlet area needs to be smaller not bigger, with increased diameter and tip speed. That allows for a choked nozzle while maintaining the triangles such that the inflow is correct (-20 to -40 degrees). I'd prefer higher solidity (more blades) so that ideal number is closer to -20 than -40. That's a juggling act because more blades chokes the exducer sooner. ...but we have what we have.

I took the comp wheels and turbine in to scan with the CMM. I wanted to re-do the turbine to confirm the exit area. I knew more what I wanted in the way of points to get a good model after doing it once. That 8 sqin number is good for the unmolested Holset AFTER assuming BL blockage of 5-10%. The 35deg exit angle is a good number too.

It's hard to say about the TV91, but it looks to me like the exducer chokes at PR 2. You can see from the turbine map, why I am trying to drive the inlet triangles more towards large aspect ratio equivalent. (away from impulse) That is the way to more mass flow because of reduced flow loss into the impeller. Increasing impulse can allow the turbine to operate at higher PR, but with lower mass flow and increased loss. The GT6041 has a much larger exducer. So you can push to higher PR before it chokes. The larger inducer of the GT6041 is in keeping with non-choked inducer design practice. The turbocharger designers seem not to want that, based on the area ratios they are using. I think without an NGV, just using a scroll the shock losses would be too high. Knowing what PR and RPM the turbine exducer chokes allows choosing the comp diameter. The TIT and RPM set the choke mass flow and allows choosing the comp trim.

The numbers I'm getting for desired comp trims are in the 30s. PR is our friend, if the turbine can take it. Somewhere between 5 and 6 PR gives the best result. But the turbine PR is going to be ~2.5 or so. I decided to take a look at trimming the larger diameter comp wheels I have. The tip speeds are in the 1700 ft/s range. With the right trim they would be good for somewhere in the 5-6 PR range. The problem is, the inducer vane angles are all too shallow for the mass flow the higher PR makes possible. If you size for the mass flow, the PR goes down because of the increasing trim value.

I wanted to do a better job than the cardboard cutout method for finding the inducer throat area. I also realized I need to re-visit the math on the comp/diffuser spreadsheet I have. It's all based on a measured throat area. I need to be able to figure out what throat area I need choke to occur at instead.

Unfortunately I think the conclusion is: I really need a custom wheel to max out the turbine...but first I need to know what that PR is.

Monty

[finiteparts]

Monty,

I have been following the discussion here and I am trying to understand what you are proposing is the driving mechanism for shallower inducer blade angles ( I follow the industry standard of defining the angles relative to the primary flow direction...so on the inducer, the primary flow direction is axially inward to the inducer face and thus the beta angle of the inducer blade is shallower on some of the aviation impellers).   I think you might be confounding things by 'mixing' very dated subsonic designs with more modern designs that are capable of flowing transonic across the inducer span.   Generally, the steeper the inducer angle, the lower the inlet relative Mach number...generally.

Here is a aviation compressor that doesn't follow the argument.  It is a Honeywell design from the 80's that can handle higher tip relative Mach numbers.







An ASME paper that discusses the design point conditions for this application can be found here:

appliedmechanics.asmedigitalcollection.asme.org/GT/proceedings/GT1986/79290/V002T04A011/234756

My opinion on what you are seeing on older centrifugals used in aerospace applications is the need to keep the tip relative Mach number subsonic due to shock losses in the inducer and there lack of designing for supersonic flows in such complex flowfields (pre-CFD).   The design space really opened up in the 1980's and 90's as manufacturing and design capabilities began to open up the design space of transonic impellers.

Lower trims are definitely goodness for our applications, but if you are designing for peak pressure conditions (i.e. at the top of the map), then this region is usually dominated by inducer stall, but a close second to the vaned diffuser stall, which means more than anything that you need to have the flow capacity of the compressor, the diffuser and the turbine stage properly matched.   You are juggling a lot of different design parameters and I am not sure if you are prioritizing pressure ratio over mass flow or what it is that the aerospace-like designs prove advantageously.  If it is the stage efficiency, then I think you might be confounding things by trying to compared stages with vaned diffusers and stages with vaneless which will have drastically different stage efficiencies.

Lower trim stages that are driving to high pressure ratios will by design need to have low diffusion ratios in the impeller (thus the shallow exducer heights) and thus will have very high discharge relative Mach numbers, leading to the need to increase the backsweep.  But the backsweep reduces the work in the stage driving a need to increase the tip speed, either by rpm or exducer diameter increases.  The shallow passages mean that the clearance losses become more dominant and thus you need to increase the number of vanes to minimize the blade loading which drives the clearance leakages.  These long and skinny passages also lead to larger passage frictional losses and thus there is a balance of how far to go on the trim and backsweep. Diffuser performance is adversely impacted by high entry Mach numbers and once you get to a difuser entry Mn ~ 1.0, you are in the realm of >10% dp/P and going up fast.   So what is my point here?  Trying to optimize on a single parameter requires thinking about all the other "systems" variables that will be impacted and that requires a lot of design work.

So I did a few quick calcs on three centrifugal compressors, the 85mm Gtx5533, the Noel Penny one and the GTCP36-300 impeller that I show above.
I just wanted to look at the inlet conditions, since this is where the inducer angle plays a role.

                                                Gtx5533                            Noel Penny                            GTCP36-300

Inlet face Mach Number:                0.122                                0.116                                    0.085   (Mn of mass flow through axial projected area of shroud - hub dia.)
Tip Mach number (U2/a01)             1.738                                1.811                                    1.817   (Industry correlation of exducer tip speed referenced to inlet a)
Relative Inlet Mn (Radius_rms)       0.809                                0.743                                    0.957   (Mn at rms radius location)
Relative Inlet Mn (Radius_tip)         1.117                                0.967                                    1.248   (Mn at impeller inducer tip)
Air relative inlet angle (deg)          -83.8                                 -83.1                                     -86.1    (these are at the inducer tip location)
Exducer tip speed (ft/s)                  1937                                2019                                      2026
Pressure ratio                                5.1 (t-t)                            9.4 (t-t)                                 6.08(?)
Corrected Mass flow (lbm/s)           2.42                                  7.1                                        3.67
Mass flow per area (lbm/s/in^2)     0.0714                              0.0683                                   0.0499   (assumes geometric area
Diffuser tech.                                 Vaneless                           Vaned*                                   Vaned  (* - the shape of the map suggests a vaned diffuser, but not sure)
Stage Eff (isentropic)                      70% (t-t)                          76% (t-s)                               82% (t-s)


As you can see, they aren't that different from each other relative to the inducer properties.   

Since it is commonly stated that roughly 50% of the kinetic energy introduced into the flow by the compressor is still in the flow as it exits the exducer, you can see that one of the largest challenges to produce the stage pressure rise efficiently is to recover the discharge kinetic energy.  I would suggest that the largest efficiency that a home-builder can achieve is to really scrub the diffuser design. 

If you look at the performance of the Gtx5533 with a small vaneless diffuser (radius ratio ~ 1.25), it is only recovering at "ideal" performance 40% of the discharge KE into static pressure rise.   If you do well with a vaned diffuser, you can get up to 65-70% of the KE recovered...for a high speed point, this could be the difference between a PRt-t of 4.5 verses 5.25.  But this is also one of the highest flow velocity regions in the engine, which means small flow issues can become very large losses in stage efficiency.

The best references that I can provide are Michael Casey and Chris Robinson's book, "Radial Flow Turbocompressors, Design, Analysis and Applications" and also this paper on the design space boundaries :   

Daniel Rusch and Micael Casey, "The Design Space Boundaries for High Flow Capacity Centrifugal Compressors"
asmedigitalcollection.asme.org/GT/proceedings/GT2012/44748/543/365301?searchresult=1

Lastly, choking in the exducer first is a wonderful means to make a super inefficient, low power density-turbine stage.  I understand the desire for higher mass flows and since the exducer has a larger physical area than a properly designed NGV stage, it is an easy way to try to increase the system mass flow, but it is also a way to increase the challenges of component matching.  I will save further comments for a future thread that I am working on.

- Chris

[racket]

Hi Monty

With the TV91 map , if the exducer was choking , wouldn't both Corrected flows be the same irrespective of the scroll A/R change ??

Cheers
John

[finiteparts]

John, I agree with you.

Also, if the exducer was choked, the corrected mass flow verses PR would have multiple curves since the critical flow in the relative frame is speed dependent, and as the rotor speed increased, the corrected mass flow would fall due primarily to the centrifugal force opposing the inward radial flow. The flat line after choking is when you have a critical flow in the absolute frame of reference, i.e. a fixed orifice...thus choking at the tongue in the scroll.

- Chris

[monty]

Chris and John,

I'll answer the things I can at the moment. Some others will take some thinking.

To your point Chris-Yes, I am using outdated design. I have dabbled enough with CFD to know the pitfalls. I don't have the data base. I don't know the right turbulence and BL models to use. I don't have the software tools.The compressors I'm using have big fat "hopefully" elliptical leading edges, they certainly aren't transonic airfoil designs. So I'm using old low tech stuff. Probably not even cutting edge 80's tech....It's what I got. But you may have answered John's earlier question about ABB's comp trims. I didn't even think about that. Of course they are going to be using CFD and transonic design. So what they come up with won't look anything like what I'm doing, or previous practice.

Regarding the choked turbine exducer....My current "understanding": Choked isn't limit load. First the throat in the turbine will choke. As the PR increases the flow goes supersonic exiting the turbine. The process of turning to axial through a shock system, rapid expansion and rotating reference frame is.....complicated. So *I* can't do a good job figuring all that out. I need a turbine map. I don't have one. At some point the axial annulus flow across the turbine exducer face goes sonic. That is the end of increasing mass flow through the turbine. More PR or not. When this choke happens is a function of how hard you are working the exducer. How much delta P in the nozzle vs exducer to balance the work. It's a trade-off. If you do more delta P in the nozzle, it gives you more room till choke in the exducer. With more loss, and less mass flow, but more specific work (turboshaft). In a pure jet, you want most mass flow with less specific work. Until you reach the limit. I'm trying to balance a bunch of bad things off one another. I'm not focused on one thing! I'm trying to find an optimum combination of bad things...

I'm trying to balance the mass flow and pressure ratio of the comp wheel to coincide with the choke flow of the turbine stage, using existing low tech, with minimal IGV loss. It's not what one would do with a clean sheet design.

Regarding the diffuser. I've been re-reading Casey and Rusch "The Matching of a Vaned Diffuser With Radial Compressor Impeller and its Effect on the Stage Performance" One paragraph stood out like a sore thumb:

"Finally, it would appear that fine details of secondary effects such as the complex unsteady flow in open impellers, possible leakage flow paths in the impeller back-plates or shrouds, and design features such as leading edge incidence, inducer shroud bleed and pre-throat curvature not crucial in the selection of the design throat area. They are of course, crucial in determining the loss mechanisms and operating range but much less important than selecting the right size of the hole between the diffuser vanes to pass the flow"

If you match the diffuser to the wheel, and the wheel/diffuser to the turbine and NGV you get max flow at the PR the turbine I can have delivered to my house will support.

That said, I'm currently crawling through the comp tip ht calculation proper. Taking into account the density increase in the wheel. I will attempt to maximize diffusion without seperation....it's a juggling act. The diffuser becomes a nightmare with low tip hts. No doubt. I'm not done with that trade study yet.....

Is it the best....No. Merely good enough for the girls that will run with me....or something. ;-)

Monty

 

[monty]

Chris,

Looking at your comparison numbers. Yes, they are roughly the same. They are all aiming for the same thing. Max flow at the PR they are after. That sets the inducer size and angle of attack. Not considering transonics. Which area is the mass flow per area for? Throat or annulus? minus hub and blades? Is there a zero missing in the GTX5553 number? Just trying to understand, not poke holes.

Smaller trim means greater mass flow per area in the inducer annulus, but the relative Mach #s stay about the same.

Define air relative inlet angle. I'm not clear on that one.

Monty

[monty]

Chris,

That Honeywell comp inducer appears to have around 40ish deg angle of attack....which is what I'm after. The limit is about 50. 45 would be ideal.

The turbocharger wheels are mostly all much shallower. 20-30 deg. From tangential. So 70/60 in aero speak. Everyone is looking for max flow and PR at the point they need it. The turbo world has surge slots, blow off valves, and waste gates. They don't care about matching to the turbine, so much as matching to the engine/turbo system.

I'm trying to pick parts from the bin that will work best together. The problem I'm running into is the PR/mass flow rpm or specifc speed trade the turbo world uses isn't the best for us. I'm now looking at trimming wheels to get closer, but the inlet angles are not quite right. They are for lower axial mass flow/area.

If designed properly the relative mach numbers are going to be the same for low and high trim wheels but the incidence angles will change. So the "leading edge twist" will be more for lower trim values. Lower trim increases delta H in the wheel at the same tip speed....flow coef. changes though.

Monty

[monty]

Chris,

I think I understand our confusion. You are talking angles relative to "flow". I am talking static geometric angles. I "think" this is the issue.

Angles relative to flow will be the same for different trims, but static geometry will be vastly different.

Monty

[finiteparts]

Dec 22, 2023 0:47:56 GMT -5 monty said:

Chris,

Looking at your comparison numbers. Yes, they are roughly the same. They are all aiming for the same thing. Max flow at the PR they are after. That sets the inducer size and angle of attack. Not considering transonics. Which area is the mass flow per area for? Throat or annulus? minus hub and blades? Is there a zero missing in the GTX5553 number? Just trying to understand, not poke holes.

Smaller trim means greater mass flow per area in the inducer annulus, but the relative Mach #s stay about the same.

Define air relative inlet angle. I'm not clear on that one.

Monty

Yep...I was missing a zero...fat fingered that one.   Thanks for catching that.

It is the inlet face annulus area....(pi/4)*(D_inducer shroud^2 - D_inducer hub^2)...the mass flow just at the forward plane of the compressor inducer face.  This is a typical comparison technique used to do competitive analysis of things such as fan technologies, compressors, etc...

I was trying to illustrate that the existing turbocharger compressors are actually pretty good and are a really good option for the majority of cases.   Coupled with a well matched diffuser I think they will really do well.   Just trying to help you reduce your scope of work.   I think trimming an existing compressor may be an interesting way to go, but I also think it is going to be quite a challenge to get it done well without significant engineering work.   The shroud profile is a strong driver of the rate of diffusion and flow migration in the passage, which opens up quite a box of worms.   But I am not trying to disparage your idea, I am just highlighting some challenges that you might research.   I think I have a few papers somewhere that discuss the impact of shroud profiles; if I can find them I will send them your way.

- Chris