Gentlemen I run a completely stock TB and intake manifold no port nothing and I lost I believe 5hp over the Wilson and it spooled quicker also. After trying every manifold and there mother on my car Im not even going with a ported mani or TB. Just stock. best luck no issues and car spools mighty quick. the 10hp is not worth trying and hoping and spending all that money. you want 10hp turn the boost up lol. My IMO


S

 

I have updated the thread on our site again this morning with the latest testing.

 

I hope to clear things up, because I'm not really anti-buschur like most of my posts come across. I just like to think there is a large amount of uncertainty in this world and any time somebody comes across as being certain about something, I like to argue. You are almost the opposite of that, you seem to like the idea of certainty, the words "It is what it is" (or whatever it is I've seen you quoted as saying) is very strong in that idea of certainty. Maybe I do it to pump up my ego by trying to throw question into somebody else judgement? Maybe I do it because I do want to pursue a higher level of understand? I don't know...

On topic
I've seen those same lines and screen captures. Go read my posts in regards to them that have been on this site. Let's just say my comments didn't make one particular shop too happy.

David, you have seen what CORRECT computer simulation comes up with, because I would place a pretty large junk of money on a bet that Mitsubishi designed the OEM manifold heavily with computer simulation.

The reason you don't see aftermarket companies with computer simulations that match performance is because the computational power needed to do it correctly is mind boggling. I have a solid model of a manifold that I designed. I came up with the design from 3D and simplified 2D steady state computer simulations. I have all those screen shots that these high end shops have.

But then I tried to do a 3D transient state flow analysis to verify my design with inlet conditions that would simulate a true engine environment. Now, I'll say this, my computer isn't anything amazing, but for a home PC, it's quite robust. Where it could run these 2D simulations in about 5 minutes and 3D simulations in about an hour, my machine ran 24 hours a day for 2 days solid and still was not able to finish the calculations in that time, not even close actually. I think it got to like 50% of the error targets. By then, I realized that it was a waste of time because there was no way in hell I could do several revisions of this manifold and test the changes. And that was just for one RPM point. Just to develop the flow characteristics of one design at one particular operation state across a usable RPM range would probably take months.

The only way a company is doing 3D transient response computer simulation is with something like a large scale super computer like you would see at MIT or similar engineering school or in the Mitsubishi engine development program. A $10,000 computer in a shop isn't even going to come close to finishing up a project like that in under a year.

That's just my opinion though. I'm not a flow analysis expert, but I do know enough about Finite Element Analysis to know that doing a computer simulation that would mimic a motor’s operational characteristics is not an easy task.

Now, there are some programs out there that can take flowbench numbers and produce simulated transient responses in reasonable times. Lotus is one such program that I have messed around with and just didn't have the time to get valid numbers into it (garbage in->garbage out).

 

F1 teams have serious super computer setups. I think right now one of the F1 teams has the largest super computer in Europe

Given that... perhaps Dave is right in that the computers tuners in the aftermarket world have access to are not powerful enough to really give results that translate to real world performance.

 

Good info and good point.

 

There are lots of CFD modeling software that can be ran on pretty basic computers that can handle things more complex than some of our vendors can even fabricate.

Are you sure he believes that personal computers are not powerful enough to give results that translate to real world performance? I feel like there is a TON that can be learned from simple modeling techniques, lots of apps out there.

Scorke

 

Sorry I did not meant to put words in Dave's mouth. Dave has said recently he thinks that perhaps computer modelling is not generating designs that deliver in the real world for aftermarket parts. Then 03whitegsr seemed to confirm that you basically need super-computing power to really do it right.

I mean can a normal, single Intel PC simulate air flow at 40 psi while pushing through intake manifold, head ports, cylinder, exhaust ports and simulate the valves moving at 8000 rpm given a camshaft profile?

 

Depends on the Intel PC and the software being used?

Not sure why you would want to see the flow through the exhaust ports, sure it ends up effecting the flow out of the cylinder which the intake manifold has to then fill but unless your designing an engine you just need to know the limits/characteristics of the air coming into/going through the intake ports, supply going into the manifold, to do a pretty decent job of modeling the intake manifold.

Modeling head ports, exhaust ports, and valve mechanics along with a moving piston is going to take a considerably more complicated program than one to produce an intake manifold.

Scorke

 

ugh, I hate when threads dissolve into theory...keep up the real world testing David.

 

But doesnt the valve opening and closing create pulses back into the manifold and therefore you need to model this behaviour to design a good intake manifold?

 

Mellon, quite to the contrary, read my last post and I think you will see that I feel the idea of FULLY simulating the performance of an intake manifold isn't going to happen at any normal shop. I agree, David's method is likely going to be the fastest and most proficient method of developing parts that work. However, some computer simulation can be beneficial like evaluating runner inlet profiles and runner airflow distribution if using non-symmetrical runners.

That is why it shouldn't keep us from discussing ways to improve on the process.

Scorke, I'm very interested in how you purpose to do a 3D transient state analysis in a quick manner? I'll shoot you a PM in how I was going about it and would love to hear back in here of how you would do it.

Every screen shot I have seen on here has been of a 3D steady state analysis with either a pressure differential between inlets and outlets or a mass flow rate input and a pressure on the outlet. I can whip one of those out in an hour no problem.

As for the question of can a standard computer can run this stuff, here is my PC setup. It's nothing amazing, but it's very stable, which is worth a lot in FEA. Nothing more frustrating than running an analysis for a day just to have it crash at 90% complete...

AMD Phenom 9850 BE (2.5GHz Quad core)
AMD 790FX/SB600 chipset (I need to go over to something with the 790FX/SB750 but haven't cared enough lately)
8GB DDR2 800 RAM (790FX/SB600 chipset has problems with 1066 with anything over 2G and with FEA, lots of RAM reduces the amount of times it has to go to the harddrive while processing large amounts of data)
Western Digital Velociraptor 10,000 RPM Sata 3.0 harddrive (essential because of the amount of data used in a 3D analysis)
4GB Ultra low latency flash drive (used for page file to reduce harddrive usage)
All factory clocking to assure stability
Vista 64 Ultimate

Now, most of the analysis I have done has been on Floworks because it's easy to work with and on the steady state stuff works well enough. Floworks is not written to multi-thread all that well so it only runs three cores at like 50% and then 1 core at about 50%. I'm sure Ansys would speed things up a lot as it's written to be ran on multi-processor servers, not to mention it would be a lot more stable. Floworks has some issues once you start doing 3D and really hates unsteady flow.

 

If you are going to that extent your using something like Recardo and you already have the engine fully accounted for.....

The software that Borg Warner turbo engineers use to design turbine wheels to computate CFD at extremely precise and detailed levels DOES take SUPERCOMPUTERS to model. A buddy of mine along with my past roomate interned there, supposedly the amount of individual things that must be accounted for, and the accuracy with which is needed to get things right inside the compressor housings is insane.

Scorke

 

03white, glad you and I have come to an understanding of each other. Notice I did say "aftermarket computer simulations" I know Mitsubishi got it right with their software. haha That stock intake is hard to beat.

You guys are talking over my head with the computer programs you use and the computers you use. Maybe if I had the ability/knowledge to do some of that work I would, I don't know. I am grateful to be filled with common sense beyond what most normal people seem to have and the drive to keep at something until it is finished (atleast finished in my mind).

I don't have much more to add. I'll just keep plugging along at this and see where it ends up. Next week we should have our 2nd and 3rd intake built. They will be as nice a piece as I can build and the last two ideas I have right now.

Thank you for the good conversation.

 

david can you post pics of the one you build and tested already?

 

Did I hear about a 3rd intake manifold!

Nice Dave keep pushing we all support your efforts.