RomRaider

ok guys, a very useful comparison can be made by plotting engine load vs MRP, and determining a best fit line.

basically the slope of this line shows you relative VE. you want a nice steep slope since that means you get high load for low MRP.

now, on my car, with a vf23, the average slope has worked out to be 0.15. i've taken a lot of logs over the last 2-3 seasons and it is remarkably stable at 0.15. some days it's 0.145, and others it's 0.155, but like i said it's a shockingly tight fit.

so what i'm interested in is the slope and intercepts for these data items from various setups.. ie different turbos and intake/exhaust mods.

you can get this very quickly in excel by using the following equations:

slope:

y-intercept:


replace the "load" and "mrp" strings with your data range... ie "c2:c3765" (see note #1)

it is also useful to put this equation in the context of an IAT, so if you have that, post up the average IAT:



if you don't have IAT, don't worry about. i'd still like to see the equations.

i am in the process of compiling the data from my car and will be posting it shortly.

ken

note #1:

for those who aren't aware, there is also an easier way to work with excel data--especially ranges of data--and that is to use "names." it's very simple and quick to set up and ends up saving lots of time and confusion. a few screen caps of the process follow:







there are also some other tricks you can do to define names but this method is probably the quickest. you can obviously perform the same name definition steps to iatc, rpm, etc etc.

HTH!

For those submitting data: For this data, you need to be sure to plot and include only relevant data. That is, the first 100% throttle data point and possibly the last, may not be relevant. If you look at a plot, you will see the slop is constant in the middle, but may trail off at the first and last data point.

fxtiam, when i perform this test i am interested in the ENTIRE datalog, and use the whole set of points. i don't mind the outliers because they are few and far between. i consider them all relevant!

the only thing that i have seen shift the slope ever so slightly is the opening of the wastegate, making it steeper.

for example, some old plots:

Seems like you should be using MAP, not boost. The boost measurements at sea level and at 6000 ft could be exactly the same, but the MAP measurements (and, therefore, air density) would be very different.

Ok, now I am confused. I briefly looked at it earlier, and apparently had some bad data points at where the throttle was snapped open.

But, why does the throttle plate not mess this up? Seems like you would have energy loss across it, which would vary significantly with position, and hence VE.

And here is one of my plots - '06 FXT with VF39.

Ok, here are some results.


M - Slope
b - Intercept
IAT - IAT F
N - Number of data points
R^2 - Closer to 1, better the data fit - Thats all I remember about it.

that's a good point, jon. sometimes i get lazy and trapped in a "sea level" mindset. it would also eliminate the y intercept as load must be zero at 0 psia.

unfortunately my historical data doesn't include MAP, only MRP corrected.

perhaps i should also compile approximate altitide (if not close to sea level).

Load = 0.185 * MRP + 2.369

~4,600 data points. Outside temp would have been about 65-70F and barometric pressure 28.87-28.85 inHg.

Wonder what that downward spike is at the high MRP? I'll have to start keeping track of this out of curiousity.

I may be wrong, but let me think about this...

Load = air in grams / revolution
Manifold Pressure = Pressure in the intake manifold
VE = Load / ManifoldPressure

So basically, volumetric efficiency is higher when we get more air into the cylinder (more load) for a given manifold pressure. Since manifold pressure should tell us how much pressure is actually in the manifold right now, it shouldn't be greatly affected by throttle position. Less manifold pressure or less VE, less air into cylinders - more manifold pressure or more VE, more air into cylinders.

Quick throttle changes would probably affect your data mostly because of poor sampling rate. If you open or close the throttle plate quickly enough your two sample points for load and manifold pressure could be on either side of the throttle plate transition. This would skew those particular data points.

Engine VE is also be dependent on RPM, and I think that is probably a lot of the "width" that I see to the data point graphs.

Also, looking at your graphs & numbers it appears that you are reading load in g/rev rather than 2*g/rev, so in reality your VE (slope) should be twice what they are to convert into 2*g/rev. So more like .18 rather than .09.

Eqtrian, this "spike" corresponds to the RPM range starting from where you reach your max boost and the red line (MRP stays high, but load decreases). It's more effective on my 2D log here (MRP in bars) taken from 2000 to 7000 RPM (max boost around 3400 RPM):
[img][img]http://wcssp.webclub-subaru.com/XNWRX/MRPvsLoad.jpg[/img][/img]

It demonstrates that our VE becomes catastrophics at high RPM !
Another comment, relevant ?, my downward spike starts when my STi cams start to close (35° before, 25° @ 3600, 0° @ 6000) ?

Makes sense, that was going through my head, but what I was too tired to put together was that the upwards part of the spike is due to the increased VE around torque peak. Then for sure the downwards part of the spike is while nearing redline.

I need to bring a few pulls up in an ecuedit 3d graph of MRP, Load, and RPM to figure out exactly what the VE curve looks like across an entire pull. Interestingly I would think that that VE vs. RPM would give a really good idea of what the torque curve should look like (then again so will a road dyno graph )

So here's a question to ponder....

Could one evaluate intake/exhaust changes to their car based on this data?

Perhaps only use data at WOT? Normalize the load values by dividing them by MAP to get VE. Then plot VE vs. RPM for a few nice high gear WOT pulls. Throw out the outliers and average the rest to form a new VE vs. RPM plot.

Then, when you do a modification you could do another set of pulls with the same analysis and compare the VE vs. RPM plots.

Think anything useful would come of this? Would be interesting to see if it shows different information than a road dyno would. Road dyno is going to be highly dependent on the timing you are running, while I would think VE would be relatively consistent at different amounts of timing. Then again maybe I am dead wrong on that one?

Cam timing would obviously have a big effect I would think, but that would be a good thing. It would probably allow you to see the effects (depending on how large of an effect the intake cam timing has).

Seems like some real experimentation is due!

Relevant to the discussion:

http://www.kickflop.net/wp/?p=17

'06 WRX with TBE and SPT intake. IAT ~55F

Appart from the effective VE number, I think it could be interresting to see the effect of tuning the engine on VE (particulary turbo efficiency).

I took logs from a EUDM WRX02 that I've remapped few weeks ago. This WRX is stock except 2.5" turbo back exhaust. (I took a WRX because it has no VVT).

Here is the MRP vs. load graph at the beginning of mapping (just few adjustments done to turbo pressure).


As you can see, there is a small area where the downward curve is lower than the upward curve during descent, this occurs between 4700 and 5500 RPM, then, after 5500 RPM both upward and downward become identical. I would suggest that this is due to the efficiency loss of the TD04L in the 4700 - 5500 area whilst producing about 1 bar.

Now the same car at the end of the remapping session. The turbo pushes 1.25 bar tappering down to 0.95 bar @ 6000 RPM :


This time it is clear that the downward curve is lower than the upward one. The turbo is out of its efficiency area and pushes hot air leading to a lower VE. No doubts than a FMIC for example would recover that.

Back to my STi curve, it is also clear that my 20G is pushing hot air. I reached the choke area !!! I'll swapp my 20G for a SZ49 in the following weeks and will make the same log...