Compressor Wheel Designs Changing .

[discopotato03]

Hi all , recently I bought a 9RS turbo for the Lancer and I got the chance to look at a wide range of Mitsubishi turbines and compressors many of which were aftermarket .

It was quite interesting to compare OE cast wheels like 15GK2s and 16/18/20G6s and even a few CNCd HTA wheels as well . I would like to have seen a GTX wheel to compare to the HTAs because some of the differences between the cast Mitsy wheels and HTAs look to be the hubs section . The billet wheels have less hub and more blade which may be part of the reason why some of these billet compressor wheels of similar diametres can have greater capacity than earlier cast ones .

I remember reading that part of the compromise with cast compressor wheels is that they have to be easily and reliably removed from tkeir moulds and this could have performance limitations . It was pointed out to me that even GT compressors like the 71/76.2/82mm ones are sort of fat in the hub and this could be part of whats driving Garrett to make their GTX billet CNCd wheels .

I do know that there are other differences like blade counts and that port shrouded compressor housings are more common on these units .

Possibly one of the selling points of the GTX turbos is that they can run port shrouded comp housings and not get the jet turbine sounds which is better IMO . Also wheel development should be faster and cheaper when your residant computer geek can punch the numbers to get the machining center to carve out whatever you like . It would be a similar deal with housing profiles and when you already own extensive test rig equipment its all good .

If only the turbines were as easy , cheers A .

[AngryRB]

looking at the GTX compressor wheel on mine recently, it looks very different to the old garrett that came off, looks soft like aluminium and the very curvy. looks alot more aggressive not that ive seen alot of them..

[GTScotT]

Not sure if you frequent the Hypergear thread Disco but the whole purpose of his CNC billet wheel adventure was to test out the 'more blade less hub' concept.

Seemed to work for him too, there are losses in response evidently yet if manufacturers could hunt for a happy medium we could see great things.

I think one thing that needs mention at the moment is the push to big pressure ratio capable compressors.. There seems to be another tradeoff there which I think could be hindering us 'cheap' folk with older cars and smaller budgets. I would like to see manufacturers go back to low pressure high volume designs for us hobo's.

Otherwise MHI still has our back's.

[Roy]

Yeh, I am getting a HTA wheel on my TD06. Slightly bigger wheel but as you say the hub section is a lot narrower on the machined jobbie so more space for blades, hence the large increase in flow for little change in overall physical dimensions. Will know by end of Jan how it actually works in the field

[hypergear]

Not really bigger the better, the thinner the hub the more inconsistency in flow. Also the shape of hub, blades, wheel dimensions and etc all makes differences in response, and power.

They needs to be very carefully trailed refer to capacity and flow of each engine for the best configuration within the target HP range. We've been making and trailing customized billet wheels for the whole year, and there are only 3 designs made "positive" differences, rest of 10+ went back wards compare to traditional cast wheels, how ever those didn't work provided directions in terms of effects on every little changes made, some of them are considered perfect for SR20dets are not so good for rb25det.

[discopotato03]

Yes I imagine a lot of trial and error goes on which would be expensive and time consuming . Yes and there always seems to be a process of learning about what doesn't want to work .

A .

[scotty nm35]

Im sure Garrett engeneers would be using flow analisis programs for pre-testing not just cutting new wheels and trialing them without some idea of what would work.

[hypergear]

No one would be spending $$$$ in programming, tooling and testing without prove of concepts and directions.

The only way to reach an goal on an engine is to prove the concept behind it on that engine. This also apply to turbine wheel, and housings profiles.

What ever "moved forward" on papers only count when it "moved forward" in real life.

Magic software does not exist.

[GTSBoy]

Im sure Garrett engeneers would be using flow analisis programs for pre-testing not just cutting new wheels and trialing them without some idea of what would work.

If you're thinking "flow analysis" meaning CFD, then I'd be reasonably certain that there's no computational fluid flow modelling software that would do a good job of it. Fluent/CFX/Ansys is pretty good code, representative of the best CFD capability available, and even for relatively simple geometries and flow situations it requires a huge amount of expertise and cross-validation work to generate modelling output that is trustworthy. (Anyone who trusts CFD output without cross-validation does not understand what CFD can and can't do.)

CFD has a hard time getting the simple things right, like a backward facing step in a duct (ie a step change increase in pipe diameter) in steady state flow. If you carefully tune the turbulence models used so that they will handle a backward step, then they completely fail at swirling flows. A compressor wheel is a horribly complicated geometry to model, and it is not stready state, it is time varying, and it is full of swirling components, and it takes the gas velocity and dynamic pressures way up past the sonic limit. All of these things are hell on CFD codes.

You could so some limited modelling, but there is no way that you could satisfactorily CFD model a compressor wheel and get output that you would trust. I would say that these things would still be designed by experience, black art and testwork as much as any other way.

[Roy]

Thats right. Its not just software. You need advanced knowledge of fluid mechanics and turbine design to validate any model.

And then you need to be a boffin in a cardigan to understand the fluids to the point that people are chasing effeciency.

I did a CFD elective years ago at Unni. Hated it. FEA was much better and made more sense to me

Bring on F1 turbos and some rapid developments in the tur ine and compressor design, materials and manufacture

[BLITZ-26]

best thing for turbocharger development will be the new F1 cars

[Rolls]

Im sure Garrett engeneers would be using flow analisis programs for pre-testing not just cutting new wheels and trialing them without some idea of what would work.

It is cheaper to just trial and error model a compressor wheel, I would go to the point of saying I don't think it is even possible to correctly model a turbine wheel in flow dynamics software.

They would use existing models generated from trial and error data to "design" new wheels.

Edited December 24, 2011 by Rolls

[Rolls]

only works at 300rpm lol Edited December 26, 2011 by Rolls