Intake Design Exploration

[NoDOHC]

Each Rotor has one intake stroke per 360 degrees of rotation.

The rotor completes 1 full revolution every 1080 degrees of eccentric shaft rotation. This is 3 intake strokes.

I think this will change the dynamics.

[vex]

Him... I don't know then. I may have to re-evaluate the length formula, though from my understanding and checking it against the stock length on the NA version of the FC it seems to be accurate.

[NoDOHC]

The nice weather will soon be over. This sounds like a winter project with a notepad and a calculator. I think I loaned engine dynamic analysis book to a friend, as I can't find it anywhere. If I can find it, I will try to derive the equations for a rotary engine (from the piston engine equation derivation explanations in the book). It has been too long since I studied any of this for me to trust my memory.

I know that the rotary engine has a 298 degree intake port duration each revolution, making the airflow into the two rotors overlap (one is still filling when the other starts). This is also why the engine is so smooth, as the power strokes of the two rotors overlap. The actual maximum to minimum volume cycle of the engine occurs in 270 degrees of eccentric shaft rotation.

Do you have an engine apart so that you can get a feel for what it does as it turns?

I can try to get pictures, but a photographer I am not (as I think is evidenced by many of my pictures on this site).

For what it's worth I think this development is very helpful to the community and I appreciate your enthusiasm and effort on the project so far.

Thanks!

[vex]

Originally Posted by NoDOHC

The nice weather will soon be over. This sounds like a winter project with a notepad and a calculator. I think I loaned engine dynamic analysis book to a friend, as I can't find it anywhere. If I can find it, I will try to derive the equations for a rotary engine (from the piston engine equation derivation explanations in the book). It has been too long since I studied any of this for me to trust my memory.

Yeah, sounds good... though with the colder weather comes more homework assignments for school for me.

I know that the rotary engine has a 298 degree intake port duration each revolution, making the airflow into the two rotors overlap (one is still filling when the other starts). This is also why the engine is so smooth, as the power strokes of the two rotors overlap. The actual maximum to minimum volume cycle of the engine occurs in 270 degrees of eccentric shaft rotation.

That I understand, but for sizing the actual intake runner length you need only concern ourselves with one intake volume as the pulse from the closing of the intake port results in the wave propagation to achieve higher VE of the engine (in other words we don't really need to worry about the intake overlap charge). The wave travels the length of the runner at a specific RPM and results in a compression wave of air being sucked into the next rotation of the rotor.

Do you have an engine apart so that you can get a feel for what it does as it turns?

No... but I do have a nice program to validate different cycles of various engines to ensure synonymous application of the formulas considered.

I can try to get pictures, but a photographer I am not (as I think is evidenced by many of my pictures on this site).

There's really no need unless you really desire to.

For what it's worth I think this development is very helpful to the community and I appreciate your enthusiasm and effort on the project so far.

Thanks!

As always I appreciate your input and help with my engineering cog turning exercises. Hopefully the final iteration of the formula will facilitate the proper lengths and diameters of the intake runners.

In the meantime however I need to do my boundary layer theory report.

[vex]

Of interest and the reason for using 0.0013 cubic meters:

Originally Posted by wikipedia

There are various methods of calculating the engine displacement of a Wankel. The Japanese regulations for calculating displacements for engine ratings use the volume displacement of one rotor face only, and the auto industry commonly accepts this method as the standard for calculating the displacement of a rotary. However, when compared on the basis of specific output, the convention results in large imbalances in favor of the Wankel motor.
For comparison purposes between a Wankel Rotary engine and a piston engine, displacement and corresponding power output can more accurately be compared on the basis of displacement per revolution of the eccentric shaft. A calculation of this form dictates that a two rotor Wankel displacing 654 cc per face will have a displacement of 1.3 liters per every rotation of the eccentric shaft (only two total faces, one face per rotor going through a full power stroke) and 2.6 liters after two revolutions (four total faces, two faces per rotor going through a full power stroke). The results are directly comparable to a 2.6-liter piston engine with an even number of cylinders in a conventional firing order, which will likewise displace 1.3 liters through its power stroke after one revolution of the crankshaft, and 2.6 liters through its power strokes after two revolutions of the crankshaft. A Wankel Rotary engine is still a 4-stroke engine and pumping losses from non-power strokes still apply, but the absence of throttling valves and a 50% longer stroke duration result in a significantly lower pumping loss compared against a four-stroke reciprocating piston engine. Measuring a Wankel rotary engine in this way more accurately explains its specific output, as the volume of its air fuel mixture put through a complete power stroke per revolution is directly responsible for torque and thus power produced.

If I'm reading this correctly a proper displacement of 0.0013 cubic meters is achieved with 1 revolution of the eccentric shaft. Validating the Helmholtz equation used.

[vex]

So... I figured I should update this thread.

I'm currently going through design iterations of a new intake. As it stands I have balanced the runners with plenum geometry.

Some pointers to keep in mind when doing this yourself:

• Plenum Geometry is very important with flow distribution to the runners
• Hemholtz equation is useful for understanding the ideal length for a specific RPM
• Plenum volume is important under natural aspiration criteria and less so in forced induction.
• Slowing the intake charge once into the plenum will result in more uniform flow distribution of the runners.

[NoDOHC]

I should clarify a previous post. I re-read it and got confused myself.

Originally Posted by NoDOHC

I know that the rotary engine has a 298 degree intake port duration each revolution, making the airflow into the two rotors overlap (one is still filling when the other starts). This is also why the engine is so smooth, as the power strokes of the two rotors overlap. The actual maximum to minimum volume cycle of the engine occurs in 270 degrees of eccentric shaft rotation.

It is true that Rotor 1 (front) is taking in air during the same time that rotor 2 (rear) is taking in air (port active 288/360) = 80% of the time - therefore 60% of the time, both rotors are on an intake stroke. Unfortunately, I improperly computed the stock port open time for a 4-port (It is actually 288 degrees - 32 degrees ATDC and 50 degrees ABDC = 270-32+50 = 288). The 6-port is 268 degrees with ports closed (270-32+30) and 318 degrees with ports open (270-32+80).

My apologies on the misinformation.

The cool thing about the rotary engine is that although the port open time is overlap-equivalent to 2/3 the timing for a piston engine - 192 degrees for a four port (talk about a tame cam), the air has the same amount of time to enter the rotor as it would with 288 degrees of intake duration on a piston engine. Bear in mind too that well-ported ports on a rotary flow much better than an equivalent-sized valve (over 3 square inches of area between the ports would require one 2+" valve or 2 1.4+" valves). This is why the intake manifold is your biggest friend (or enemy).

Thanks for all your research in this vex!

[vex]

Time for another update (this will be used in conjunction with the NA to Turbo sticky found in the 2nd Gen forum):

To further understand the flow distribution to each runner in a manifold one needs to consider the geometry one has chosen. Flow distributions are going to be significantly different if one chooses runners 'inline' with the throttle body as compared to a 'non-inline' manifold.

A picture is worth a thousand words:
for a 'non-inline' type of manifold think of this:


for a 'inline' type of manifold think of something similar to this:


Both have their pro's and con's. For most people an 'inline' style manifold is going to be just fine for what they want. However there are some things to be made aware of if you're going to be making your own.

First and foremost, any changes to the stock manifold should be qualified. That is; use a CFD program to calculate how your flow is going to be affected. I'm not going to get into the nitty gritty of CFD programs or even how to run them, but suffice it to say that it's better to use one than to not (you can look up OpenFOAM for a free CFD program with a large support community--I however do not use it, so don't ask me). Once you have qualified your base design to ensure uniform airflow distribution between both the front and rear rotors look at the size of the throttle body. If you can't get the flow just right, take a close look at your intake velocity .

When using a CFD program, you may notice that some of the issues you encounter come from too fast of an airflow into the manifold (if you're doing boosted applications you'll need to calculate the volumetric flow rate coming through the throttle body). There are a few ways to overcome too fast of an airflow into the manifold. First and foremost increasing the size of the throttle body will inherently slow down the flow for a better and more uniform distribution. However, just upping the throttle body may not yield desired results. One must further look at how the air is expanding into the plenum. Without getting into a lot of engineering speak and mathematics; a gradual enlarging of the effective area of the flow will slow down the flow. Such an object is called a diffuser (no we're not talking about the thing that goes under the rear bumper--though they do the same feat).

Diffuser geometry is very tricky to get right. If, for instance, you enlarge the diffuser too quickly you will cause the air to separate (engineering speak: The boundary layer separates) from the wall of the diffuser causing turbulence and circulation (something we're trying to avoid at this point of the flow). If you carry on the diffuser for too long the boundary layer will begin to enlarge and separate; eventually causing circulation. This is, of course, based on the intake velocity. Slower traveling intake charge will be easier to expand than a faster moving charge.

For understanding the above here are some images of 75mm 'non-inline' manifold compared to a 90mm 'non-inline' manifold.

(reserved)

[blaen99]

Here's a very early beta of Vex's proposed program written in C# for ease of development.

It only accepts Vex's proposed metric values for now, but I need to verify if the values and math work correctly or not to go forward with further development.

[vex]

it looks like it's outputting the same values. Here's some tests for verification:

inputs:
r=0.019
rpm=6500
vol=0.0013
temp=293.15

Your output: 0.22178 m
My output: 0.2218 m (Due to format truncation).

r=0.019
rpm=3200
vol=0.0013
temp=293.15

yours: 0.91507
mine: 0.9151

Looks like your math is fine to me. Is there a specific error you're concerned with?

[blaen99]

Originally Posted by vex

it looks like it's outputting the same values. Here's some tests for verification:

inputs:
r=0.019
rpm=6500
vol=0.0013
temp=293.15

Your output: 0.22178 m
My output: 0.2218 m (Due to format truncation).

r=0.019
rpm=3200
vol=0.0013
temp=293.15

yours: 0.91507
mine: 0.9151

Looks like your math is fine to me. Is there a specific error you're concerned with?

If it's good, it's good. I'm working on the PHP end, which is why I'm releasing a super-simple and shitty C# version while I get that done. The C# version can be used to find any bugs or problems in the PHP version before the PHP is complete.

[vex]

There's been progress on this topic; I just have not been able to get back to it in a very timely manner. As a teaser: Here is a quick and dirty of a prototype intake manifold under CFD testing. I have forgone the use of a diffuser for this run; just to ensure that if I can get a uniform flow at the entrance of the manifold I get a good flow at the runners:

Here's the initial design:

notice the poor flow distribution to the runners. This required me to alter the geometry of the plenum.
Here's the final prototype:

[RICE RACING]

When I was helping a student do his Thesis (I was operating the engine dyno and doing the mapping) for SAE intake manifold runner length experiments he was tossing up to use 1st or 2nd order waves for packaging reasons.

It would be nice to use (at lesser effect) the later orders to allow a shorter more direct manifold. Have you looked into that?

It's a bit like the header discussions from old people who used to race 13B P.P.'s fitted to a prototype mid engine set up. No way could you use the long header but the later order waves allowed a much shorter version and the power differences were in favor of the short through less system length and pipe friction losses.

Keep up the good work.

[vex]

Originally Posted by RICE RACING

When I was helping a student do his Thesis (I was operating the engine dyno and doing the mapping) for SAE intake manifold runner length experiments he was tossing up to use 1st or 2nd order waves for packaging reasons.

It would be nice to use (at lesser effect) the later orders to allow a shorter more direct manifold. Have you looked into that?

It's a bit like the header discussions from old people who used to race 13B P.P.'s fitted to a prototype mid engine set up. No way could you use the long header but the later order waves allowed a much shorter version and the power differences were in favor of the short through less system length and pipe friction losses.

Keep up the good work.

When I saw you posted in this thread, I was a little nervous that I made a mistake somewhere.

I have actually avoided working on the intake runner length problem for some time as I didn't have a good grasp on the physics or math of what was going on (not to mention the time I didn't really have to devote to it). I may revisit them today and better take into account the various order waves/harmonics you reference--and if I'm feeling adventurous I'll even put in pipe/friction losses. As always I'm glad to have input on matters that I don't have any experience in.

As an aside I no longer have access to CFD to verify diffuser geometry or even flow distribution (as the latest update to Autodesk's suite of CFD solvers requires more resources than my little laptop can handle). I hope to remedy this situation in the coming months; but we'll see how it all plays out.

Edit: Looking back over some of the Helmholtz equations for a tuned resonator I can't seem to locate any 'simple' references to what wave number enters into the calculations. Thinking over the physical process of how the intake runners work and tune I'm debating on if I need to even progress down this road. Rotarygod already did a majority of the simplification for the combined area of the stock intake runner port size. Ideally I would like to just adjust the posted forumula to account for a different runner area (combined primary and secondary ports), but if I do any more detail than that I think I would just be wasting time. Maybe I'll put this all on the back burner for awhile...

[j9fd3s]

neat thread! i've not much to contribute except that i've read a lot of SAE papers and Mazda has published a lot of rotary specific info.

particularly 831010 which is the 6 port intake system. the 90032 rotary P port modeling paper. also the S2000 paper is good, although isn't rotary specific. there is an S5 paper too.

the crib notes, we know about the lengths, i think, shorter = higher rpm peak.

they have a graph of plenum size vs VE, and due to the 270 degree cycles the rotary has what they call a tournament effect, when one rotor closes it uses the reflection to charge the other rotor. Mazda found a smaller plenum increases this effect. honda actually found the same, which is why the S2000 has a teeny plenum.

with the Rx8, all the runners (primary, secondary, thirdinary, corollary, arbitrary, and fred), have tuned lengths AND diameters.

from the info provided it really looks like Mazda made something that they could easily change the lengths on, and then they just tried a bunch, easier to do on a P port.

mike