A Series simulation thread.
Turbo Mini Forums

valvetrain really has less losses than the crank? even taking into account the valve-train operates at half speed seems counter intuitive! v.interesting.

turbo 16v k-series 11.9@118.9 :)

Denis O'Brien.

The model is based on Patton et al. (1989) and the improved version of Fmep by Sandoval and Heywood (2003)

SAE papers are quite expensive when you don't have access via a library but the relevant content is described here as well for a improved model:

http://umpir.ump.edu.my/3984/1/P176.pdf

Edited by Sir Yun on 30th Oct, 2014.

That sir, is not rust, it is the progressive mass reduction system

http://aseriesmodifications.wordpress.com/

doesn't that paper put train loss at 16% and crank at 9% joost ?

Medusa + injection = too much torque for the dyno ..https://youtu.be/qg5o0_tJxYM

It does , but does not mention for which type of valve train. and is the newer friction model. That said I shall and see if I made a mistake in the Fmep inputs

Edited by Sir Yun on 30th Oct, 2014.

That sir, is not rust, it is the progressive mass reduction system

http://aseriesmodifications.wordpress.com/

This is the Sandoval & Heywood data for a 998:



This does not include pumping losses as they are calculated within the simulation.

Other optional methods of calculating friction are available in my software and the highest values are about the same as yours but they are for a diesel engine.

Edited by Paul S on 31st Oct, 2014.

Saul Bellow - "A great deal of intelligence can be invested in ignorance when the need for illusion is deep."
Stephen Hawking - "The greatest enemy of knowledge is not ignorance, it is the illusion of knowledge."

this is the data as i use it.

Attachments:

29.29kb

That sir, is not rust, it is the progressive mass reduction system

http://aseriesmodifications.wordpress.com/

This is after working on the burn model (it now uses a turbulent entrainment model to get the values for the Vibe/Wiebe parameters once as otherwise the TE model just takes too long (>7 hours) to do a run . This is about a good as I'm going to get it for now. I did not tweak the model further to force it into shape as that is probably counter productive for the predictive capabilities.


I don't have dyno data for lower rpm's so at this point I can't really tell if the data lower down will correlate as well.

There are a few issues left.

The timing figures are based on my best guess of what you need for such a build.

This model assumes a constant AF ratio which with a Weber carb (well any carb) is unlikely at best.


I tried to model a reverse7 port head, so two siamese ports in but four proper exhausts out. It looks like the exhaust are quite material to the whole robbing thing, more so than the intakes ports being shared.

Must verify the results a bit though

Attachments:

16.41kb

Edited by Sir Yun on 12th May, 2015.

That sir, is not rust, it is the progressive mass reduction system

http://aseriesmodifications.wordpress.com/

well done joost.

Medusa + injection = too much torque for the dyno ..https://youtu.be/qg5o0_tJxYM

Saul Bellow - "A great deal of intelligence can be invested in ignorance when the need for illusion is deep."
Stephen Hawking - "The greatest enemy of knowledge is not ignorance, it is the illusion of knowledge."

it only seems to do that in very specific situations. and I initially made an error in the direction of the flow so the CSA increased the wrong way round . That is all fixed.It still does it a bit though.

The volume of the siamese sections seem to have something to do with it

That sir, is not rust, it is the progressive mass reduction system

http://aseriesmodifications.wordpress.com/

I also noticed from your blog that you are treating the junction of the siamese ports within the head as a "plenum".

I don't think that is correct as it is clearly a well radiused joining of the ports. Treating it as a low volume plenum could cause changes to the pulses.

Saul Bellow - "A great deal of intelligence can be invested in ignorance when the need for illusion is deep."
Stephen Hawking - "The greatest enemy of knowledge is not ignorance, it is the illusion of knowledge."

No/Yes it is treated as a three way joint just like in a collector. however The CSA's and angles are taken from casts I made from a head so it does have the xtra volume as well

if you look at the cast the only circular proper ''port runner '' is the 30mm just before the valve.

plenums do change the pressure /wave behaviour . I discussed this with Neels but i'm unable to find in it the chaos that is my mailbox.

That sir, is not rust, it is the progressive mass reduction system

http://aseriesmodifications.wordpress.com/

Apparently this paper is one of the best references on the subject of junctions:

http://papers.sae.org/1999-01-1186/

Saul Bellow - "A great deal of intelligence can be invested in ignorance when the need for illusion is deep."
Stephen Hawking - "The greatest enemy of knowledge is not ignorance, it is the illusion of knowledge."

I tried the size you mentioned for the manifold. so slightly smaller than what I had. Seems to work better.

Attachments:

29.54kb

That sir, is not rust, it is the progressive mass reduction system

http://aseriesmodifications.wordpress.com/

The LCB test is a very good indication that the model is working correctly.

I built an NA 998 model in the first instance, before adding a turbo etc. I built an LCB into the model using the dimensions given by Vizard in the yellow bible.

I then used the parametric optimiser feature of the software to test different lengths and diameters of the LCB. The dimensions given by Vizard could not be bettered in terms of area under the torque curve. I could send peak torque up and down the rev range a little, but not without reducing the peak value and peak Hp.

Given that Vizard probably took years to get those dimensions just goes to show how useful and valid these software tools are.

Saul Bellow - "A great deal of intelligence can be invested in ignorance when the need for illusion is deep."
Stephen Hawking - "The greatest enemy of knowledge is not ignorance, it is the illusion of knowledge."

I find it interesting that it mostly affects the inner cylinders. without software you have to to to in cylinder pressure sensors.

It used to be ridiculously expensive ( spend 100k no problem) but now you can get a setup you can just stick on you engine in car or on a dyno for 3-5 K from TFX.

still can't afford it but it is no longer into the pure fantasy range anymore. I'd wager a bet your program cost a lot more :)

That sir, is not rust, it is the progressive mass reduction system

http://aseriesmodifications.wordpress.com/

The software and pressure sensors serve two different purposes.

The software assists the designer in optimising the performance. The sensors verify the design.

You have to build the engine to use the sensors whereas with the software, you can build and test umpteen engines without cutting any metal.

Saul Bellow - "A great deal of intelligence can be invested in ignorance when the need for illusion is deep."
Stephen Hawking - "The greatest enemy of knowledge is not ignorance, it is the illusion of knowledge."

Edited by adcyork on 12th Nov, 2014.

Saul Bellow - "A great deal of intelligence can be invested in ignorance when the need for illusion is deep."
Stephen Hawking - "The greatest enemy of knowledge is not ignorance, it is the illusion of knowledge."

You have a point if you wish to apply OEM standards. The problem with model calibration is that it is disproportionately expensive to do and in the majority of real world cases, impractical.

I first posted this video about 4 years ago. Very much of what they say explains my approach:

http://www.worlddragracing.com/downloads/U...TBHP_F_800.html

Edited by Paul S on 13th Nov, 2014.

Saul Bellow - "A great deal of intelligence can be invested in ignorance when the need for illusion is deep."
Stephen Hawking - "The greatest enemy of knowledge is not ignorance, it is the illusion of knowledge."

I agree model calibration is expensive and time consuming, when OEM standards of accuracy are required.

However in this case even a small amount of well captured data would suffice to provide a footing into the behaviour of these siamese ports to validate the modelling philosophy of the ports.

The moddelling methodology outlined in that video is inline with the typical initial design practice carried out in industry, which is as I understood you would be operating, with a view to reducing the required physical testing iterations.

But when the task of modelling the siamese ports is presented, then in my opinion no number of simulation iterations is going to give an accurate enough answer due to the lack of published theory and data on the matter.

Like you said, the last work conducted on an engine of this nature is now a long time ago, and almost certainly not simulation based so there is little to go on without capturing some data known to be reliable in my opinion.

Whilst it would be ideal to have some data to calibrate the models, particularly in respect of the siamese ports, we do have the option of using sensitivity testing to establish if the engine model responds to known improvements as would be expected. This then gives us the confidence that the model is sufficiently reliable to our needs.

After sectioning a head to establish the port dimensions, I'm more than happy to model the siamese ports as a simple junction.

I've already detailed the verification of the LCB. I've also ran extensive modelling of scatter cams that replicates Vizard's findings.

The most convincing model is the table top shape torque curve from the rolling road data of my original 998 turbo. This is due to siamese port specific issues that can be readily explained by examination of the pulse patterns on the centre exhaust port.

Anyway, I'll be putting my latest engines on a rolling road sometime next year. Let the numbers do the talking.

Saul Bellow - "A great deal of intelligence can be invested in ignorance when the need for illusion is deep."
Stephen Hawking - "The greatest enemy of knowledge is not ignorance, it is the illusion of knowledge."

Paul S and Joost, keep the dialogue going. most of it goes well above my head, but the modelling/calibration against known to predict future change trends is seriously good. Paul S , without giving away secrets is the cam in your next build going to be single or scatter? And finally for scatter cams, i gather the more scatter the timing events/valve piston clearance becomes messy, will these models black flag options that cause metal against metal, .. eg scatter of 3 degrees all ok, scatter 5 degrees ERROR reduce lift or timing etc)

Saul Bellow - "A great deal of intelligence can be invested in ignorance when the need for illusion is deep."
Stephen Hawking - "The greatest enemy of knowledge is not ignorance, it is the illusion of knowledge."