The EJ20 Turbo Choice thread - Comp maps, CFM flow ratings.

[clutchcargo]

I've been looking at my turbo options for my next phase, and have spent a bit of time looking at the flow characteristics of the EJ207, as well as a range of different turbo compressor maps. This is a relatively basic analysis, but I thought I'd share my findings anyway (and by all means let me know if something doesn't sit right or you want me to add a map).

Currently I'm running an AUDM 2005 STi, maxing out the VF35 (205kw/275hp@19psi at the wheels on the STi dyno here in Melbourne), but am looking for about 330hp at the wheels.

These results really helped me understand which turbo I wanted to go with. Some turbo vendors provide a flow rating, or CFM (Cubic Feet per minute) which provides you with a bit of a guide as to how much a turbo can push. However usually this rating is at the peak efficiency of the turbo, so it doesn't translate in any real way to how a turbo will respond to your setup.

So I decided to map out the EJ20 CFM at varying boost amounts and RPM, so I could plot this flow right on top of the turbo maps and get a sense of which turbos would be best suited to me, and there were some interesting results!

To start with, the CFM Engine flow rate is (cid x rpm x 0.5 x VE) / 1728
CID = Cubic inches of displacement (In a 2ltr this is 122.05)
RPM = Obvious really
0.5 = This is a four stroke so we need to halve
VE = Volumetric Efficiency. The EJ20 is regarded at about 85***37; however when you flow heads etc this can be as much as 90%
1728 = Number of cubic inches in 1 cubic foot

Once I plugged these numbers into a spreadsheet, I ended up with the following table, this shows at various RPM and boost settings what CFM the 2ltr outputs. Note that Pressure ratio 1 is atmospheric pressure or 0psi, 1 bar is actually PR2 (or two atmospheres of pressure, aka 14.7psi).



Next I set up a graph showing the CFM range for each boost setting and RPM range. So assuming I can boost to PR2.0 (14.7PSI) at 3000 rpm, the EJ20 will output 180CFM@3000RPM and 480CFM@8000RPM, but at 20PSI the EJ20 will output 288CFM@4000RPM and 576CFM@8000RPM! You can start to see why guys run strengthened internals to run bigger boost and higher redlines, and do head flow mods to increase their VE rating!



So how does that make my life easier choosing a turbo? Simple! I first decide on a prefered boost level and rev limit (19-20psi and 8000rpm). Then I plotted my RPMs/boost on the flow map. I also plotted 14.7psi for comparison. The numbers here are RPM values. So at 20psi, at 5500RPM, there is 400CFM of flow.




Next I got a series of turbo compressor maps and overlaid them on a common scale. Where some turbo providers give measurements in CFM, others do it in lb/min, so to convert back (assuming you're at sea-level) you just divide by 0.8.
e.g. 45lb/m / .08 is 562 CFM

So, to choose my turbo, all I had to do was consider where in the rev range I'd be doing most of my driving, and find the turbo compressor map that matched my 20psi plotline! The answer, luckily for me given I'd already bought one, was the Garret GT2871R, otherwise known as the AVO400L which I picked up for a great price (about the same as a TD05).

What do flow mods do here?? Well they allow the turbo to spool quicker, and for the exhaust to rid the system of back pressure. Therefore again if I have a more restricted flow my boost plot would come on later, and in the top end it would fall away faster.

You want to avoid choosing a turbo where your boost plot line is going to run to the left of the surge line, in this case (i.e. assuming a high-flow set-up) you're going to see compressor surge (which I currently suffer with in higher gears with the VF35). Note that my assumptions on boost plot are based on my VF35 spool characteristics, the GT40 wouldn't likely see 20psi till much later, so conversely you want to avoid a turbo that's too large for your CFM output.

I've included that plot below, as well as a TD04 (its out of its peak efficiency even at 14.7PS, but is still at 76% at 20psi!), 400s, TD05-16G/GTPS02 (great wrx upgrade IMO), PE1820, TD06-20G, and for a laugh a GT40. More comparisons below:















Next up I'd like to find a VF30/35/34/39 which are a PITA to find, and show them up against the TD/GT variants.
Ben.

* Registered users of the site do not see these ads.

[clutchcargo]

New Plots:

APS SR40


APS SR50


TD05-18G
Poster, please send me original map.

VF23


VF22



Comparisons:
Lots to be learned about the options even from just looking at these charts, regardless of parts fitted, dyno/road tuned, even 1/4 mile times. It explains graphically why some turbos don't really come alive until 18-20psi (GTPS02/TD05), why some turbos need/love high flowing mods (GT/400L), why VF turbos experience compressor surge, why its stupid to think TD04s are good for 180kws or GT40s are suitable for 2Ltr applications, etc etc.


400L vs 400s (400L in pink)


400L vs PE1820

[clutchcargo]

Saved for more comparisons

[slowmike]

Beautiful presentation of the data. I would also like to see a VF34 map as mine is starting to show some shaft play and will need to be replaced soon. I would like to compare the VF to a small/big 16g.

[clutchcargo]

Yeah that 16G plot looks almost perfect for a high-flowing 18psi setup, you get quick spool and between 4000-6000 it is in 75***37; efficiency or higher and should be around 65% at redline (so it might drop off right at redline, could someone confirm?)

However, if ANYONE can find me ANY VF comp map I'd be eternally greatful, they must be closely guarded secrets. I think based on my VF35 experience, I can guess what a plot would look like compared to the TD05-16g small.

A friend tells me: "Ben, the VF34/35 was designed between Subaru and IHI as the Homologated turbo for the group N/PWRC cars, and has been designed to run optimally with a 34mm restrictor. This results in the turbo running out of flow over about 5500/6000rpm (about where the restricted engines rev to) and going into a surge condition."

So, I would say the VF3x turbo would look like the 16G plot (below), but with less tolerance for surge limit (see 1820 vs AVO400L above), and thinner efficiency islands. As a result the 16G should spool similarly (if not quicker) and have a similar top end, but minus any comp surge issues.

What are the MPH/dyno results between the two? VFx pull about 280hp, 16g about the same??

See below:

[fastwrx25]

wow that is very thorough and packed with info. great job man.

I have seen compressor maps for the vf23 turbo. would that be of use to you?

[clutchcargo]

Thanks fastwrx25, yeah shoot it over to me, it would probably be between the TD04 and VF3x in terms of flow, but might give some clues as to the look of VF maps.

[prometheum]

you divide by 1728 because thats how many cubic inches are in 1 cubic foot

[ride5000]

http://ken-gilbert.com/wrx/vf23.gif



ken

[clutchcargo]

Nice, thanks Ken, I've added that at the bottom of the plots so you can take that image down if you like. And thank you prometheum, damn you progressive education and your metric system!!

[Redline8k]

can you do something like that for a gt32twinscroll or a gt3076 for a 207. I know both are garrett turbos.
this is a great way to look at the data!
thanks
-nick

[ride5000]

juding by the high PRs indicated, i suspect that pe1820 comp chart is based off the "old style" compressor cover.

[nhluhr]

Added to the Threads Of Note sticky. Great job

[eqlized_aero]

TD05H-18g looks pretty good as well

[clutchcargo]

Thanks nhlurhr.

redline8k: can you shoot me URLs of the twinscroll maps on the turbosbygarrett site and I'll put up a comp map. I suspect they have really wide efficiency islands.

Ride5000: you know I thought something might have been up with that map, I wasn't aware the 1820 had comp surge issues or that it ran out of puff at redline, so if anyone has a newer map I'll happily put it up.

Eqlized, that was my original choice or a TD05-20G. Do you have a url for the 18g comp map above?

[clutchcargo]

SR40 and SR50 maps added.

[subenerd]

Get an EVO.

[NSFW]

Originally Posted by eqlized_aero

TD05H-18g looks pretty good as well

I don't think that chart is scaled correctly - it shows the 18G putting out over 50 lb/min, but IIRC it's actually maxed out at 40 lb/min.

Other charts have the 16G putting out over 40 lb/min (more than an 18G should), 20G peaking at ~55 lb/min (should be 44 according to Blouch), and so on.

Are the charts wrong, or am I reading them wrong?

[clutchcargo]

Originally Posted by NSFW

I don't think that chart is scaled correctly - it shows the 18G putting out over 50 lb/min, but IIRC it's actually maxed out at 40 lb/min.

Other charts have the 16G putting out over 40 lb/min (more than an 18G should), 20G peaking at ~55 lb/min (should be 44 according to Blouch), and so on.

Are the charts wrong, or am I reading them wrong?

That chart isn't mine, if the poster can provide the original map I'll scale it correctly.. This is one thing I've been careful with

[eqlized_aero]

The top of the compressor map says how it was created.

I believe I've seen deadbolt post the same map however. Correct the image if I got the scaling wrong.

(pretty sure they don't allow hot-linking, scroll down to td05h-18g)
http://www.stealth316.com/2-3s-compflowmaps.htm

http://www.stealth316.com/2-3s-cfm-maps.htm

[SublimeGTP]

Great work on this... one thing I would suggest from some extra accuracy.

VE is not constant across engine rpms. The accurate way to figure out the VE would be to disconnect the turbo and get the MAF readings. Since torque is a measurement of cylinder pressure and cylinder pressure is mostly related to the volume of air combusted, the easiest way to estimate is to look at the torque curve of a NA 2.0L Impreza (if it used the same heads). It will be slightly off because of the compression ratio, spark advance, etc but it would be close enough for government work

[clutchcargo]

Has anyone done this, or would anyone like to suggest a correction factor? I could work in a notional one to show what difference it really makes.

For example, a VE of .85 @ 20psi/8000RPM is 576CFM, whereas assuming VE diminish to .8 then @ 20psi/7000RPM flow is reduced to 542. On the graph this would mean the 8000RPM line moves LEFT about 1/4".

It doesn't make a huge difference in most cases, but yes it is important to bear in mind that there will be slight left/right shift in the maps depending on your engine's VE, but for my purposes this was great in determining that the 400L was ideal, and that the 16G would drop away at approx 7000 in my application.

Unfortunately what this can't do is factor in one's propensity to require more power over time, nor provide an accurate indication of disposable income vs cost of parts to acquire said power.

[banzai]

OP, outstanding work! I am very impressed!

[slowmike]

clutch, I'm afraid the the formula for calculating one's propensity to require more power over time would be rather complicated and lengthy.

[Uncle Scotty]

...something else to point out is that the EJ207 that the USDM never got, is a much different(and far better) 2.0 than the EJ205 that we DID get in the US.....and what works well on a 207 may not be(almost certainly WILL NOT be) as good on a 205