I have been working on an intake manifold for my LS-Datsun and was playing around with our flow analysis program to run some test trials. I found the flow rate(Lb/min) of my turbo at a given RPM(or pretty close based on compressor maps and calculations) and ran the analysis. We only have the express version so I can only do one inlet and one outlet. I will try it at school(I think we have the full version), to see if I can do all runners at the same time. This was a worst case scenario. Where all of the air was going into one cylinder.

Still working on the plenum volume required(anyone know of any formulas for plenum volume required for a forced induction setup) The runners will be seamless, formed out of one piece of round tubing. I have already machined the dies to be able to form the tubing. I will have done some test runs on some 6061 that I annealed and it formed very nicely.

Lemme know what yall think, any input is appreciated.

 

I have spent several years on this type of thing. When i started a few years ago, I was doing the same thing you were. Just running airflow through a runner. Then airflow through mutliple runners, etc... I got to the point where i began to wonder what I was looking for out of the simulation. This lead to trying to simulate an entire engine cycle - which eventually led me to 1D gas dynamics/engine simulation. Here's my advice for you to save you some time.

Designing an intake manifold with 3D CFD is near impossible (unless you are coulping it with engine simulation code). The results won't really mean much unless you are able to model the head ports, the cylinder with moving piston, and valve motion. This would take a significant amount of modeling in fluent or some other advanced code. Even then, moving boundary condition codes run into convergence issues, and is very sensitive to your modeling. At the end of the day, this probably isn't a very realistic way to do things - there's a better way.

This is not to say CFD is useless however. You just have to use it in the right way (I'll get to that in a second). If you get access to the full flow simulation program, you are not going to want to flow all the runners at the same time. You'd want to flow each runner as it would be in an engine cycle at a given RPM. This leads to the question of boundary conditions. Ideally, you'd use a pressure vs time/crank angle at each of the ports along with your expected boost pressure at the inlet. You would then see where the lb/min fall out, and adjust your boost pressure boundary condition so that you were matching flow and pressure on the compressor map.

However, the problem with all of this is the pressure vs. time trace for each of the ports - how do you determine that? Ideally, this is where you'd go acquire some advanced 1D engine simulation/gas dynamics code. You'd model the geometry and the associated discharge coefficients through the whole system. You'd correlate the model and match either VE or torque to the known engine configuration. You can then examine the pressure at the intake port vs. time and input this as the boundary conditions to your 3D CFD model. This is the route I'm going, but between building the models, doing some coding in excel, correlating to a stock LS2, I probably have 300 hours in, and it's been very frustrating.

Ultimately though, 1D simulation is how you study the effects of plenum volume, runner lengths/diameter, runner taper, cam selection, etc....

Where the CFD becomes useful is for what I'll call detailed design. It's useful doing things that 1D code can't do. For example, I use CFD to predict dishcarge coefficients where you have pipes dumping into a large volume. Or another good example is using CFD to optimize your runner entry (i.e. reducing the losses). Too large or too narrow an inlet radius and you give up aerodynamic performance of the runner. For advanced manifolds, CFD is also useful to look at an entire engine cycle and determine where the air is separating from flow surfaces.

Since the full 1D simulation is probably more than you are looking for at this point (cost to buy a good program, or the time/knowledge required to code your own), I'd suggest downloading Lotus's free 1 cylinder engine simulation model. With some basic inputs, you should be able to simulate 1 port and use it to find a sample pressure vs. time trace to use as your input to your 3D CFD model.

This subject is a bit more complex than can be covered in 1 post on the internet, but I hope this helped some.

 

On a note related to your pictures.

It looks like your peak velocity is around 1200 ft/s at the throat of the runner. Depending on the temperature and pressure you used, this is very near Mach 1, ( choked flow). This is much too fast. You definitely need a better set boundry conditions.

 

Runner length looks short, plenum volume looks large. Should work good for drag racing. Conversely, others with manifolds of similar characteristics actually went faster by switching back to factory-type manifolds (LS6/ FAST/ etc).

What are the goals of this intake manifold design? Are You redesigning for packaging/ clearance? Better performance? Strip/ Roadcourse/ Street?

You may get more useful info from the CFD by having all cylinders getting air at the same time. Ideally You would have the cylinders drawing air in the firing order (at specific rpm) but You'll need to rerun the analysis for each rpm (at minimum 3 runs).

Challenge Yourself on the setup conditions to see if Your assumptions make sense. (Easy mistake is boost preesure in the manifold at idle). Not saying You made a mistake, but be sure to get a second set of eyes on the setup/ constraints - even if those 2nd set of eyes are Your own.

Collect some data on the most common manifolds that work (LS6, LS3, FAST) and compare them to the aftermarket offerings (esp the dyno sheets). There are a couple "aha!" features in there that would probably change Your design significantly.


Regards,
Kurt Betton

 

Wow, awesome info. Like I said my program only allows one entrance and one exit, and is very limited in is boundary conditions. Setup is a T76 on a built bottom end LS1. I am mainly building it just to build it. I am finishing up my ME degree and also do quite a bit of fabrication work. Just wanted to play around with some designs, and build a clean functional manifold. I have an LS6 manifold for it right now. I am only looking for 650 wheel . I also realize that the shorter runners will sacrifice the bottom end a little, which is fine since its going in my 280 datsun. I really wish there were more schools that taught flow science and design in reference to automotive. I know there is a large amount of science in flow characteristics of a motor, NA or FI.

 

I will re-run the program with 1/8 of the flow to that one cylinder. Velocity numbers should be much better. Car is also a fun street car, maybe take it to the drag strip or a track day. but mainly just to mess with all the Lambo's and ferrari's around me

 

I'm no engineer but I've done some engine dyno testing and track testing with FAST vs CARB vs Sheetmetal and for your application, the sheet metal will suffer everywhere below 6600 RPM. If you are wanting to build one anyway, I'd keep the plenum volume as close to the sum of the runner volume as possible and even error on the smaller side. The plenum pictured with the short runners will hurt your throttle response and 60' ET horribly vs say an LS6 intake.

 

^ Thats kind of what I am looking for, people who have done the dyno testing, also have been hearing from other friends as well. Keep everything small to keep the throttle response up. Keep everything smooth and rounded to keep turbulence down. I am working on another design with longer runners as well.