N54 Manifold simulations

[digger]

i am not really convinced that having all runners drawing from the plenum at the same time gives a realistic picture. if you look at the firing order and cam profile for each cylinder and plot it out over a 720* there is really only a few cylinders drawing at the same time and it is mainly two at any point in time that are drawing hard on the intake system.

Look at the M50, M52 and S52 manifold, looking at the OP plots the distribution would be so bad that one cylinder would be 10:1 AFR and another 15:1 if you tried to draw through all runners simultaneously

there is of course pressure gradients in the system that can affect distribution but you need to do a transient analysis

[digger]

Hey guys I have been really busy so I didn't have much extra time to spend on the simulations. I did begin putting together a new spreadsheet that captures valve events so I can use these boundary conditions for my transient flow simulation. I didn't forget about it.


Attachment 18549

maybe no one is active in this thread but i dont think this graph is correct. cylinder 1 is at TDC on the induction stroke every 720* so on a 6cyl the offset between each cylinder should be 120* ? not 70* as shown?

[Sticky]

maybe no one is active in this thread but i dont think this graph is correct. cylinder 1 is at TDC on the induction stroke every 720* so on a 6cyl the offset between each cylinder should be 120* ? not 70* as shown?

@DBFIU

[mjmarovi]

Ya, I'm still here. I made the decision to move over platforms to the Ecoboost, but I am working with Vargas on his projects now, and we both feel that it makes more sense to put our resources into a the twin turbo kit. The intake manifold will make more sense when we need a aux fuel rail for port injection when the hp numbers get to that point. Naffo and I got together about a month ago and tried to fit the wood model I made, and it will not fit as it is with stock rear air snorkel, and I have to go back and edit a few things. Also, I have had a crap ton of personal stuff happen to me this last month that kinda made car parts irrelevant. Sorry, $h!t happens, and sometimes you got to crap back at it. I am working on N54 stuff, I just do not post about it on the internet.

That fitment issue on the rear snorkel is no issue for us single turbo guys who would be the biggest customer

[Payam@BMS]

@DBFIU

He hasn't been active in years LOL

[Sticky]

He hasn't been active in years LOL

You're wrong.

[mjmarovi]

You're wrong.

Yeah, I show he posted on 5/20/13 that's less than a year ago Payam...

[Sticky]

Yeah, I show he posted on 5/20/13 that's less than a year ago Payam...

[DBFIU]

i am not really convinced that having all runners drawing from the plenum at the same time gives a realistic picture. if you look at the firing order and cam profile for each cylinder and plot it out over a 720* there is really only a few cylinders drawing at the same time and it is mainly two at any point in time that are drawing hard on the intake system.

Look at the M50, M52 and S52 manifold, looking at the OP plots the distribution would be so bad that one cylinder would be 10:1 AFR and another 15:1 if you tried to draw through all runners simultaneously

there is of course pressure gradients in the system that can affect distribution but you need to do a transient analysis

digger you are 100% correct, but for a good baseline design we first analyze this in a steady state simulation. If you see any gross mistakes such as very large deviations in flow from runner to runner, you can correct and re-run the analysis very quickly. Generally in industry simulations can take hours or days, and that costs a lot of money since you need a powerful computer to run these simulations. So for starters I ran a steady state simulation because I am accustomed to doing things this way (it saves more time in the long run).

Transient simulations take 10 times or even up to 100 times more allotted computer time to complete and then they need to be post processed and assessed with takes a very long time, that I will do once we finalized a generally good intake design, and minor tweaks after the transient are not uncommon.

Good catch, do you run CFD?

[Payam@BMS]

Yeah, I show he posted on 5/20/13 that's less than a year ago Payam...

http://www.JaguarBoost.com/images/im...74169662-1.png

LOL, close enough. I had to check his profile too so he was here about two weeks ago just snooping around.

[mjmarovi]

LOL, close enough. I had to check his profile too so he was here about two weeks ago just snooping around.

Actually he was on earlier posting, see above? lol

[digger]

digger you are 100% correct, but for a good baseline design we first analyze this in a steady state simulation. If you see any gross mistakes such as very large deviations in flow from runner to runner, you can correct and re-run the analysis very quickly. Generally in industry simulations can take hours or days, and that costs a lot of money since you need a powerful computer to run these simulations. So for starters I ran a steady state simulation because I am accustomed to doing things this way (it saves more time in the long run).

Transient simulations take 10 times or even up to 100 times more allotted computer time to complete and then they need to be post processed and assessed with takes a very long time, that I will do once we finalized a generally good intake design, and minor tweaks after the transient are not uncommon.

Good catch, do you run CFD?

im an amateur CFD person but have been doing professional FEA for over a decade now. since starting using ansys CFD became available and i just learn from a peronal point of view

ive never done transient myself for that reason. i see alot of people online CFD that draw from all runners to look at distribution but when you do this is too "harsh" a test for the manifold

this is a naturally aspirated simulation whereby m50 all runners drawing on the manifold (flow is 3.2L 100% VE 8000rpm) see how crap it is but obviously it cant be this bad otherwise some cylinders would be hardly run





angled entry all runners drawing (flow is 3.2L 100% VE 8000rpm) is alot better but still not great





Since all runners never are drawing at the same time only 2 to 2.5 runners are drawing depending on cam duration

i looked at the above "all" runner sim and firing order to work out which two cylinders will produce the worst distrubution


m50 runners 1&4 drawing (flow is 3.2L 100% VE 8000rpm) 11% variance which is not great but not exactly bad when the flow and inlet tube velocity is probably double what factory intended





angled entry runners 1&4 drawing (flow is 3.2L 100% VE 8000rpm)
2.5% variance





this is a steady state flow with known mass flow into the system.

[Sticky]

LOL, close enough. I had to check his profile too so he was here about two weeks ago just snooping around.

Or chatting with some of us

[DBFIU]

digger I have a lot of experience with ANSYS CFX and ANSYS Fluent FEA. If you want to run a simulation using those programs let me know. Also CFD meshing is important if you want to get the correct pressure drop due to boundary layer viscous forces. You know how to calculate Y+ and how to set up boundary layer mesh? Very important! Also your Courant number is very important for transient unsteady CFD!!

[digger]

digger I have a lot of experience with ANSYS CFX and ANSYS Fluent FEA. If you want to run a simulation using those programs let me know. Also CFD meshing is important if you want to get the correct pressure drop due to boundary layer viscous forces. You know how to calculate Y+ and how to set up boundary layer mesh? Very important! Also your Courant number is very important for transient unsteady CFD!!

i always inflate the mesh at the surface and check mesh quality, i did a 5 day training course with ANSYS fluent (not CFX) where that was convered but have not done much hands on since so have forgotten alot.

i dont know how to do transient in fluent. im not even sure how you would do this from a strategy perspective let alone specific details in the GUI with regard to boundary condition without resorting to moving boundaries and remeshing. i guess you would specify a time varying pressure at the outlet? you could get an idea on the depression using EA Pro or dynomation

[DBFIU]

You would have to set a varying pressure at the outlet. This is still not very accurate because the outlet pressure depends on the inlet mass flow due to the fluid intertance (helmholtz resonance). So how do you predict the outlet pressure of this model? They are coupled and the BC's for the outlet are tricky because they depend on the mass flow. See where this can get complicated?

[digger]

You would have to set a varying pressure at the outlet. This is still not very accurate because the outlet pressure depends on the inlet mass flow due to the fluid intertance (helmholtz resonance). So how do you predict the outlet pressure of this model? They are coupled and the BC's for the outlet are tricky because they depend on the mass flow. See where this can get complicated?

or in otherwords without a highend 1D sim that acoounts for these phenomenon to tell you pressure history and feed it into a 3D analysis it is still more accurate to build a prototype and and test it. its the same with FEA unless you know the loads and BC mostly you generate pretty pictures. without enough experience you can do approximations that give 95% of what you are after though.

luckily i am mainly interested in NA engines with individual throttles and large volume air boxes where resonances are not as important or tune-able (usable).

im surprised no one has looked at making some aftermarket manifolds with two plenums each feeding 3 runners to maximise the helmholtz effect. it might be possible to boost performance when turbo hasnt kicked in

[DBFIU]

or in otherwords without a highend 1D sim that acoounts for these phenomenon to tell you pressure history and feed it into a 3D analysis it is still more accurate to build a prototype and and test it. its the same with FEA unless you know the loads and BC mostly you generate pretty pictures. without enough experience you can do approximations that give 95% of what you are after though.

luckily i am mainly interested in NA engines with individual throttles and large volume air boxes where resonances are not as important or tune-able (usable).

im surprised no one has looked at making some aftermarket manifolds with two plenums each feeding 3 runners to maximise the helmholtz effect. it might be possible to boost performance when turbo hasnt kicked in

I would say in this case you're right, high end FEA analysis like this requires thousands of dollars in computing equipment and software. For that money, you can buy a BMW, cast a new manifold and test it at a dyno. For the gas turbine industry where a new engine program costs 100 million, it pays off though. Still, you can use some basic CFD techniques to get you in the ballpark which is what I am trying to do. I won't get it optimized to a nats ass, but it will be close to optimal, at least.

[Flinchy]

Just wondering if there's actually anyone anywhere in the world anyone knows about.. who is actually looking seriously into providing an improved N54 manifold?

power numbers are starting to get to a level where the gains in an optimized manifold aren't looking as expensive as they used to be lol... I'm being tutored in inventor/CFD by an engineer friend who is interested in this stuff, but nothing is likely to come of it, at least not for anyone else product wise haha.

[Payam@BMS]

Just wondering if there's actually anyone anywhere in the world anyone knows about.. who is actually looking seriously into providing an improved N54 manifold?

power numbers are starting to get to a level where the gains in an optimized manifold aren't looking as expensive as they used to be lol... I'm being tutored in inventor/CFD by an engineer friend who is interested in this stuff, but nothing is likely to come of it, at least not for anyone else product wise haha.

My fabricator wants to build me one, with injectors etc. Him and Terry were talking about it in person but I'm not sure if we're going to go with it. For now the stock one is doing okay, and it does great with heat compared to aluminum.

[klipseracer]

I would say in this case you're right, high end FEA analysis like this requires thousands of dollars in computing equipment and software. For that money, you can buy a BMW, cast a new manifold and test it at a dyno. For the gas turbine industry where a new engine program costs 100 million, it pays off though. Still, you can use some basic CFD techniques to get you in the ballpark which is what I am trying to do. I won't get it optimized to a nats ass, but it will be close to optimal, at least.

Glad to see you post, its been a while. Where did you run off to?

[Flinchy]

My fabricator wants to build me one, with injectors etc. Him and Terry were talking about it in person but I'm not sure if we're going to go with it. For now the stock one is doing okay, and it does great with heat compared to aluminum.

not hard or expensive to coat a part in gold tape or get it ceramic coated or something (or both.. etc.) no?

i'm having a ball.. then getting frustrated.. then having a ball again.. learning inventor haha...

I'm sure it'd be a lot faster to just jump to creating a part then trial and error-ing until you get the gains you want.. but always useful to add other skills imo... the geometry on these manifolds isn't exactly complex to put together either. just then trial and error-ing in software is a lot more complex to learn... already a hell of a lot of progress by @DBFIU though.. did just about everything up to making a physical part and seeing if it works, it seems :/

[Tamut]

wow, i would have happily pay for the #2 style manifold!

im an amateur CFD person but have been doing professional FEA for over a decade now. since starting using ansys CFD became available and i just learn from a peronal point of view

ive never done transient myself for that reason. i see alot of people online CFD that draw from all runners to look at distribution but when you do this is too "harsh" a test for the manifold

this is a naturally aspirated simulation whereby m50 all runners drawing on the manifold (flow is 3.2L 100% VE 8000rpm) see how crap it is but obviously it cant be this bad otherwise some cylinders would be hardly run


http://www.60-130.com/images/im...5c183ac6-1.jpg


angled entry all runners drawing (flow is 3.2L 100% VE 8000rpm) is alot better but still not great


http://www.60-130.com/images/im...8671aa5f-1.jpg


Since all runners never are drawing at the same time only 2 to 2.5 runners are drawing depending on cam duration

i looked at the above "all" runner sim and firing order to work out which two cylinders will produce the worst distrubution


m50 runners 1&4 drawing (flow is 3.2L 100% VE 8000rpm) 11% variance which is not great but not exactly bad when the flow and inlet tube velocity is probably double what factory intended


http://www.60-130.com/images/im...afbcfbd6-1.jpg


angled entry runners 1&4 drawing (flow is 3.2L 100% VE 8000rpm)
2.5% variance


http://www.60-130.com/images/im...e1a65b6e-1.jpg


this is a steady state flow with known mass flow into the system.

[654]

Anyone fabricating one?