Turbine Housing Scroll Design 2

[L0s7]

Hi Guys!

Hoping to get someone with more knowledge on the subject than I. I'm reverse engineering a Toyota CT26 turbine housing for an MR2. I have the exterior locations accurately modeled and now I'm moving to incorporating the scroll, while adapting it to mate up with a Garrett G series CHRA. In this way a G series turbo is completely bolt on and stock looking. My end goal is to get the A/R of the turbine equivalent to the Garrett's 0.72A/R housings, seems to be a fair compromise.

I have a few questions;

Aspect Ratio:
From most sources is defined as the cross sectional area of the volute at any given azimuth described as a percentage of its radius from (what I understood to be) the center of the turbine. However, roughing out some real life measurements of a sectioned housing produces areas that are way out from what was expected based on the definition.

However, if I measure the radius from the point in the casting that flow exits the volute, and enters the turbine wheel itself... I'm getting A/R numbers that are very reasonable and close to numbers I've found online (of an unknown source quality) generally cited as 0.44-0.48 A/R (I'm getting 0.48-0.50 averaged between the two divisions)

I notice that one side of the scroll has a lower aspect ratio than the other. Does anyone know more about the logic behind this? Hypothetically all 4 cylinders would have equivalent mass flow, then are split in to equal pairs by the exhaust manifold... But then one pair gets a 'restricted' scroll? Purely for low RPM response? Wouldn't designing a divided housing be the largest benefit in the first place? Perhaps it is traditionally fed by outer/further cylinders whose exhaust gasses may have less energy by the time they're entering the scroll?

If anyone has experience in this field, pointers are very much appreciated!

 

[Lou Scannon]

Surely there is a wastegate port somewhere. Please describe its location.

"Schiefgehen wird, was schiefgehen kann" - das Murphygesetz

 

[L0s7]

Each division of the scroll gets a 22mm wastegate port that dump adjacent to the turbine exit, see:

http://www.mr2turbo.info/ct26/

 

[Lou Scannon]

OK. I was looking for an explanation for the unequal aspect ratios, and that's not it. But as you say, it could be a devious effort to optimize spooling. I certainly have no MR2 specific knowledge.

"Schiefgehen wird, was schiefgehen kann" - das Murphygesetz

 

[EdStainless]

It is an attempt at getting broader performance. This is for drivability.
If you are going to race then it doesn't matter since you will be >80% or redline all of the time.

= = = = = = = = = = = = = = = = = = = =
P.E. Metallurgy, consulting work welcomed

 

[L0s7]

One thing I am interested in is exactly how the CT and G housings compare in terms of scroll cross section. By definition, since the G housing is more compact it is modeled with a smaller spiral pitch- so a given G A/R is 'equivalent' to a smaller CT A/R?

 

[Kim_W]

Like Edstainless says, it is to give the car a better and broader torque distribution. The small and big volute generates two different air speeds which will spool up the turbine wheel faster at low rpm. If you do racing, there is no need for twinscroll. Then a single scroll might even be better in some cases.

Sidenote: i also own a 3SGTE, but i forged the engine, made my own runner from SS 321 (cnc merged collector) coupled with a BorgWarner EFR7670, build is still ongoing And yes this is also a twinscroll turbo indeed but not really necessary for a racecar (F1 racecars might be another story).

Best regards,
Kim W
Formal Mit***ishi Turbocharger engineer

 

[EdStainless]

It is analogous to cross-plane intake manifolds.

= = = = = = = = = = = = = = = = = = = =
P.E. Metallurgy, consulting work welcomed

 

[turbomotor]

LOs7,

Addressing your original post, I think the A/R for Garrett turbocharger turbine housings is defined only at the section of the turbine housing casting called the T-T section, which is the last fully enclosed flow section in the housing. Theoretically, this section determines the tangential flow velocity at a specific radius, which is defined as the radius with half the nozzle tangential flow area outboard and half the nozzle tangential flow area inboard. The tangential velocity of the gas is then accelerated by conservation of angular momentum as the tangential flow component moves to a smaller radius and enters the wheel.

Bottom line - if you are measuring the A/R at the T-T section (which it appears your are doing), then I think you have it right.

Regarding the difference in the area of the two scrolls, I can only guess the same guess as others have already done. It may have been done to enhance transient response.

For maximum system power and efficiency (and I have done this with 400 mph race aircraft at the Reno Air Races), get rid of the wastegate completely and carefully size the turbo expander nozzle (or A/R for a "nozzleless" volute) to achieve the correct compressor and expander work balance at the design peak power operating point. Design point efficiency will be optimized, but transient response will be slow.

 

[L0s7]

Thanks for the input everyone.

After quartering the Toyota housing I attempted to get some rough cross sectional area figures, it seems at the throat of the housing the Toyota housing has ~7.5cm^2 of area, not small for a stock housing it seems. But the rotor design is very poor by todays' standards. IIRC it's a 60/50mm wheel, which by comparison the Garrett GT28 turbine which is ~20 years old at this point likely has comparable mass flow at 54/48mm. The G25 series then flows 15% more than that.

Read a very informative paper on testing different aspect ratios (maintaining equivalent cross sectional area) the gist I got is that A/R is relatively unimportant as far as how much mass can flow, but will affect air speeds which will affect response. Even still the paper noted that even the housing optimization that can be done is largely dwarfed by the characteristics of the rotor. In essence the housing provides a 'pool' for air to flow through/mass in and can be considered a balance between how quickly the rotor empties the pool and how quickly the engine fills it. As long as your design is reasonable, you avoid any glaring aerodynamic problems, performance should be OK. Not to dismiss optimization... The paper did discuss (somewhat) friction along the 'roof' of the scroll being a primary reason why higher A/Rs have lower airspeeds...

Am I correct to take from that, that: a higher flowing wheel can be put in a larger housing and maintain comparable response (as seen by the driver) to a lower flowing wheel with a smaller housing? Generally.

Used some clay to make a mold of the throat of the Garret G housing and it uses ~9cm^2 undivided throat area and dyno sheets I've dug up on the web have reasonably comparable spool (if a tiny bit slower) to the old GT2860rs. Trying to avoid chopping up the Garrett housing... So I've sketched out a couple potential shapes that would begin at ~8.6-8.7cm^2 with a divided design. The goal is as close as OEM Toyota appearance as possible while bringing in a 30 year newer turbo design... Maximize area under the curve.

@Kim_W,

Cool I have a forged Rev3 motor, I've drilled it for EGR, runs a UK Alltrac ECU that operates it so it passes emissions testing in my area. I've designed a crank pulley with 12T trigger wheel that replaces the NE crank signal in to the ECU... Requires a Caldina oil sump that provisions a mount for a reluctor sensor. Timing is much more stable than with a distributor. I also want to draw up a replacement for the distributor, essentially a cam driven shaft with a single tooth wheel and a plate at the end that can mount a stock Toyota VR sensor that cost $20-50. Having 3x VR sensors in close quarters in the distributor works OK but have read reports of noisy signals from it. I'm also incorporating an ignition box that a Swedish owner designed that splits the stock IGt signal out to 4 different pins to operate COPs with the OE ECU. My ECU also has daughter boards that allow me to modify the tune/code.

 

[Lou Scannon]

The scroll serves as a nozzle, the effective area of which strongly controls turbine performance. More so than the physical size of the turbine wheel and its design, which certainly play their part as well. To put it more simply, the engine speed at which the turbo wakes up is more a function of the nozzle area than the turbine wheel itself, assuming no gross mismatch of the turbocharger components in the first place.

"Schiefgehen wird, was schiefgehen kann" - das Murphygesetz

 

[turbomotor]

For whatever it is worth, I fully agree with Hemi. First principles of momentum transfer (as defined by the Euler Equation (Leonhard Euler)) look specifically at the tangential component of the gas velocity as it enters a radial inflow reaction turbine. This tangential component of the gas velocity is set by the gas conditions (temp, press, mass flow) and the nozzle size (A/R, which is shorthand for Area/Radius, and not necessarily Aspect Ratio).

The design of the wheel usually dictates the amount of diffusion in the wheel and the loss associated with the kinetic energy dump at the turbine exit (often times in the form of exit swirl). Other loss factors in the wheel design can be abrupt changes or reversal in the blade/gas relative velocity.

Regarding turbocharger transient response:

Generally, smaller A/R provide better transient response, while larger A/R provide better top end performance. I think (and this comment for sure is open to debate) that the biggest impact of the wheel design on transient response is the polar moment of inertia of the wheel.