Intake Design Exploration

[vex]

So I've decided I want to produce a custom intake manifold, but I have no idea where to begin. Anyone have any links or words of wisdom on intake design? Plenum design, runner lengths (variable and fixed), pulse tuning, air intake velocity effect on torque and horsepower, etc, etc, etc. I'm looking for it all. I have a rough design in mind, but I would like to refine it and make it better and eventually using CFD to ensure the best possible outcome before producing it in real life.

So... Let 'em rip!

http://books.google.com/books?id=DoY...age&q=&f=false

[need RX7]

http://www.rx7club.com/showthread.php?t=199788

Originally Posted by rotarygod

Well everyone always wants to know the mathematical way to figure out intake runner lengths so they can design their own manifolds. There are so many things to understand and the math seems to go on and on forever. Since there are books dealing with the subject in great depth I'll just get to the simplified math so you can figure out the perfect legth for your port style. Using this formula you will also learn why a halfbridge or full bridgeport engine utilizing the stock intake manifold has no top end but great midrange power. Here it is:

L= ( (1080-EPD) X 650 / (RPM X RV)

L= Legth of the runners. This is your answer.
EPD= Effective Port Duration (how long they are open for)
RPM= The spot where you want peak horsepower to be. (If you still have the stock gear ratio transmission and this is a streetcar there is absolutely no point in making this anything other than 6500-7500 rpm.)
RV= This stands for Reflective Value. The pressure wave reflects back and forth several times inside the pipe. For the intake the second wave is best so use the numeric value of 2. For a carburated car use 3 or 4 since the manifold may be too long. If you are figuring out exhaust length use 1. this will give you the proper length for a short primary collected system. If you want a long primary system, take the short length and multiply it times 4. OK lets plug in some numbers to prove this.

Let's look at just the primary ports in an '86-'88 n/a 6 port engine. The ports open at 32* ATDC (after top dead center) and close at 40* ABDC (after bottom dead center). We use 720* as our base point to start figuring out EPD. Since the port opens after TDC, we subtract 32* from 720* to get 688*. Since the port closes 40* after BDC, add 40* to 688* to get a total EPD of 728*. You now have one number to plug in to the above formula! So far the formula is (1080-728) X 650= 228800. Now we need to know what to divide this by. Since the '86-'88 n/a 6 port engine had a power peak of 6500 rpm this is what we will use for this example. Also since I said the second reflected wave is best to use, use the numver 2 for RV. There are the rest of the numbers for you. Take 6500 X 2 = 13000. Now we have 228800/13000. The answer; 17.6" There is one other thing to consider though. The reflection doesn't take place at the very end of the intake runner pipe but rather at a distance 1/2 the diameter of the pipe out away from the end of it. (what the hell did he just say?) Go back and read that slowly. Since the primary intake runner is 1 1/8" in diameter we must subtract half of this value from the length of the intake runner. .56". According to the calculations, the proper primary intake runner length for a stock port '86-'88 n/a 6 port engine is 17.04" The actual length as published by Mazda is 17.1"!!! Holy crap it works!!! The slight difference can be attributed to several small things. First we did not account for how fast the ports open and close. A peripheral port opens faster than a side port. accounting for this would give us shorter runner lengths. We also didn't account for the distance within the rotor housings that the air has to travel. This would add to our length. Basically these numbers almost cancel each other out so I don't worry about it. If you want to know how to figure out down to the last thousandth of an inch it will take some studying. If you look at the above number though we are within .06" of an inch from actual. Close enough. Your altitude where you live will affect it much worse than that.

If you need port timing specs for different year models I do have them as well as some of the Racing Beat port template specs and peripheral port specs. You will see something very neat when you type in the specs for a streetport but retain the factory intake manifold. Your horsepower peak will get lower! For my streetported GSL-SE I actually need a manifold with runners a little over 1" shorter just to retain the stock 6500 rpm peak number. This gets even worse with a half bridge or semi peripheral port. Many people think that since they ported it, their top end power falls off faster because lack of fuel. That may play a small role but they run rich up there anyways so big deal. The real problem is that the stock manifold they are using is too long. BDC if you are reading this it explains why Tony's half bridgeport car has much more midrange power than the streetport but falls off hard on the top end. Before you spend money on a new turbo, build a new intake manifold! Yours is too long. Yes it still matters on forced induction cars too!!!

If we wanted to tune the above engine to have peak power at 7000 instead of 6500 rpm the length would change to 15.75". For 8000 rpm it would be 13.75". Cool huh! Your low end will suffer the higher it is tuned and your power band will get narrower. You may quickly get out of you gear ratio range in which case all your new found power is worthless. If you used a Guru Racing transmission that has a shift point centered at 9500 rpm then you would want to tune for peak power at 9000 which would give a runner length of 12.15". The intake runner diameter may not flow enough air due to size though so this is something else to consider.

Now I suppose you want to figure out proper plenum size. I'll get into that later. I will just say that it gets SMALLER as the rpm limit rises! I'll let you all sleep on that one! If you have any questions about your porting style shoot me a pm and I'll help you out best I can. Have fun!

Cheers! :beerchug:

Fred

[vex]

That works well, but what about the other stuff? Plenum design and so forth? (The RCC is a different beast when it comes to math--we like the nitty gritty... I'm actually going to be looking into a book that deals specifically with this question I posted, though it's out of print)

I'm going to check out this book:
Theory of Engine Manifold Design: Wave Action Methods for IC Engines
ISBN: 0768006562

[vex]

Okay, so that book is okay... but doesn't cover anything specifically with rotaries. I saw maybe 6-7 books total in my schools library concerned about rotaries, and they were all from the 70's... I may have to start looking up more ASE articles for it... sigh, not what I wanted.

[RotaryProphet]

Rotaries aren't that special; treat them as a two stroke two cylinder of stated volume (ie, 1.1 or 1.3 literss), or as a four stroke engine with twice the RPM. Port timing info is available, and with all of that info, you should be able to calculate runner size and length and plenum volume, as well as throttle body size for your given peak torque/horsepower point.

As an interesting note, by using exhaust and intake runners a couple of inches longer on one rotor than the other, each rotor will have a different peak power point about 500 or so rpm apart, which leads to a wide peak power. Helpful if you have a particularly peaky motor like a P-port or a big bridge.

[vex]

Originally Posted by RotaryProphet

Rotaries aren't that special; treat them as a two stroke two cylinder of stated volume (ie, 1.1 or 1.3 literss), or as a four stroke engine with twice the RPM. Port timing info is available, and with all of that info, you should be able to calculate runner size and length and plenum volume, as well as throttle body size for your given peak torque/horsepower point.

So Plenum is only concerned with volume and not geometry?

As an interesting note, by using exhaust and intake runners a couple of inches longer on one rotor than the other, each rotor will have a different peak power point about 500 or so rpm apart, which leads to a wide peak power. Helpful if you have a particularly peaky motor like a P-port or a big bridge.

But would that put a moment on the E-Shaft and cause additional wear or are the forces congruent with doing that negligible?

[RotaryProphet]

Originally Posted by vex

So Plenum is only concerned with volume and not geometry?

In as much as flow patterns are concerned, geometry is important, but as long as it's not shaped so as to -prevent- flow, you should be alright.

In an application where the throttle body is on the side of the plenum (facing forward in a rotary application), it's best to use a plenum four to six inches longer than the distance from the front of the first runner to the back of the last runner, and taper the plenum the entire way. This helps the air "slow down" earlier in the plenum instead of wanting to slam into the back wall, and helps the front runners breath. In a setup where this isn't done, the rear-most runner tends to ingest the most air, and the front-most one (from the throttle body's perspective) tends to ingest the least.

In a side-facing throttle body, the best solution would be a setup that tapered in both directions, with the widest portion in the middle, and some sort of diffuser to help the air with it's right-angle turn into the plenum, going either right or left. However, in a rotary specifically, since the middle ports ingest less air anyway, you can get very good results with a simple tube with a throttle body stuck on it, and let the "bad" shape direct the majority of the air into the big ports, where they're needed.

Also it's important to bell-mouth your runner entrance from the plenum, or better yet, use short velocity stacks actually sticking into the plenum.

Originally Posted by vex

But would that put a moment on the E-Shaft and cause additional wear or are the forces congruent with doing that negligible?

Considering that only one rotor is ever firing at a time anyway, you've always got more force (significantly) from one rotor than the other. The difference of maybe 10 horsepower at max between two rotor firings is negligible in comparison. No different than having a slightly low compression rotor and a good one.

It's really much more useful in V8 applications, where each pair of cylinders 360* off from each other is setup with peak power 500 RPM off from each other, creating, say, two cylinders making peak at 4500, two at 5000, two at 5500, and two at 6000, creating a very wide power curve. Obviously you're trading a reasonable amount of peak power (up to maybe 30hp) for this much wider band, but in situations where that's desirable (notably rally and drift racing, and some road racing), this is a good way to help. To really pull it off you need individual cylinder fuel and spark control, however.

[PercentSevenC]

Here are some more relevant posts by rotarygod:
http://www.rx7club.com/showthread.php?t=94362

Originally Posted by rotarygod

I have a custom built aluminum manifold on my 2nd gen. It has a 100 cu in plenum volume. The intake runners are slightly larger than the lowers but there is a taper down to their size. This is done to broaden the torque curve since the air accelerates as the volume decreases. I also have a 75mm Mustang throttlebody on it with my blowoff valve machined about a quarter of an inch in front of the throttle plate. This engine was designed for big power but still be street drivable. I'll have to post some pictures later this week. The mathmatical formulas are hard to apply to a rotary. Keep in mind that on a piston engine the air is stagnant longer in the intake runners than they are in a rotary. On our engines the intake ports are only closed for a very short amount of time. On a piston engine, unless it is a 2 cycle, the air in the ports is not moving until every other time the piston is at the top. Because of this the effect of plenum volume changes in relation to piston vs. rotary. A bigger plenum volume will raise the torque peak. Port runner length is also dependent on port runner size. The effects are due to a volume/velocity relationship. Theoretically peak power (n/a) is achieved at whatever rpm the engine is at when air velocity entering the engine is at .6 mach (60% the speed of sound). This is however offset by runner length as a runner of too much length can kill power. Its almost too complicated for me to get into without writing a huge book (already several out there on this subject) or giving a night school class. I'm afraid something I write will sound contradictory to something else. Lets just say my setup was trial and error and it works damn well! Oh btw a pp engines intake is always open. When the apex seal crosses the port it is briefly open to 2 chambers. Therefore it never closes.

Fred

[vex]

So I've started a preliminary design of the intake. And Suddenly everything I learned in Aero/Hydro dynamics is making sense to me and how I can mathematically apply what I learned. I may have a V2 of the manifold done by the end of the month. This should be interesting...

[vex]

Alright, I think understand what you're saying concerning the Plenum design. Shouldn't be too hard to do a digital mark up in a few. As for the velocity stacks I was already considering doing that.

My question however is for turbo charged applications does wave tuning do that much to begin with? I'm curious because if I'm understanding it correctly the positive pressure comes on (for me anyways--Turbo 6PI) around 2k RPM. If i'm getting positive pressure that quickly, no matter what wave I tune for with the intake runner lengths I'm going to end up with more pressure than the wave could shove in by itself (even when it's not during a compression wave).

If my thinking is correct then I could be able to have rather long runners and be fine power wise. My concern from this however is will throttle response be adversely affected by having abnormally long runners? As it stands right now I'm thinking about keeping the runner length the same as a stock NA (roughly 17" or so), but directed much differently so the air needs to only take one continuous turn once inside the manifold.

I think the best bet for me (and everyone else who follows this thread) right now would be to focus on one stage at a time. For now lets focus on Plenum design and worry about intake runners later:

If I'm understanding you correctly you are telling me that a diverging-converging
(Air-> TB:) nozzle design will work best for when the throttle body is placed on the longitudinal axis of the Plenum. Lets focus on this setup as it seems the easiest to manufacture and produce in ones garage.

As air inters the divergence portion of the plenum the air will slow down according to thermodynamics:

A/A*=1/M[(2/(k+1))(1+(k-1)/2*m^2)]^((k+1)/(2*(k-1)))

Since it's air, k=1.4 and M<=0.6 the formula will give you A/A* for the divergence. (Note: A- Area when gas inters, A* When gas is at M speed)

If the pressure drop is significant enough the temperature will drop, but flow speed will suffer as it drops down in mach number. Knowing what the temperature will drop to we can solve for velocity using

Ve=sqrt(k*R*Te)

Note: There is a conversion factor in here (and this will end up in metric units)

Now you mentioned something about a baffle or is that not needed in a rotary application? Do we even need a Divergence-Convergence Plenum, or is it just a simple matter of getting a big enough pipe and sticking a throttle body on the end of it?

Aside: Can anyone scan a lower intake gasket for a 6PI and take a single measurement for me? I wish to be accurate and start a digital construction of the intake system so when I have access to CFD I'll be able to accurately see where potential flow issues are. Thanks

[tervo rodriguez]

here is a look at my custom intake that positive pressure comes in at 2200k w/webbers dcoe 45 on 13psi boost at 3200k and puts out 322 to the wheels 360hp on my raceported 12A

Originally Posted by vex

Alright, I think understand what you're saying concerning the Plenum design. Shouldn't be too hard to do a digital mark up in a few. As for the velocity stacks I was already considering doing that.

My question however is for turbo charged applications does wave tuning do that much to begin with? I'm curious because if I'm understanding it correctly the positive pressure comes on (for me anyways--Turbo 6PI) around 2k RPM. If i'm getting positive pressure that quickly, no matter what wave I tune for with the intake runner lengths I'm going to end up with more pressure than the wave could shove in by itself (even when it's not during a compression wave).

If my thinking is correct then I could be able to have rather long runners and be fine power wise. My concern from this however is will throttle response be adversely affected by having abnormally long runners? As it stands right now I'm thinking about keeping the runner length the same as a stock NA (roughly 17" or so), but directed much differently so the air needs to only take one continuous turn once inside the manifold.

I think the best bet for me (and everyone else who follows this thread) right now would be to focus on one stage at a time. For now lets focus on Plenum design and worry about intake runners later:

If I'm understanding you correctly you are telling me that a diverging-converging
(Air-> TB:) nozzle design will work best for when the throttle body is placed on the longitudinal axis of the Plenum. Lets focus on this setup as it seems the easiest to manufacture and produce in ones garage.

As air inters the divergence portion of the plenum the air will slow down according to thermodynamics:

A/A*=1/M[(2/(k+1))(1+(k-1)/2*m^2)]^((k+1)/(2*(k-1)))

Since it's air, k=1.4 and M<=0.6 the formula will give you A/A* for the divergence. (Note: A- Area when gas inters, A* When gas is at M speed)

If the pressure drop is significant enough the temperature will drop, but flow speed will suffer as it drops down in mach number. Knowing what the temperature will drop to we can solve for velocity using

Ve=sqrt(k*R*Te)

Note: There is a conversion factor in here (and this will end up in metric units)

Now you mentioned something about a baffle or is that not needed in a rotary application? Do we even need a Divergence-Convergence Plenum, or is it just a simple matter of getting a big enough pipe and sticking a throttle body on the end of it?

Aside: Can anyone scan a lower intake gasket for a 6PI and take a single measurement for me? I wish to be accurate and start a digital construction of the intake system so when I have access to CFD I'll be able to accurately see where potential flow issues are. Thanks

[vex]

Originally Posted by tervo rodriguez

here is a look at my custom intake that positive pressure comes in at 2200k w/webbers dcoe 45 on 13psi boost at 3200k and puts out 322 to the wheels 360hp on my raceported 12A

Your plenum and intake are simple enough. Though I don't think your runners can readily be seen in the pictures you provided (since you're running a carb the runners would be underneath the carb).

Though I must say, I really like the way your plenum is set up. It looks somewhat similar to my first attempt to digitally model the intake, though I didn't place the blow off valve on the plenum. I like the thought of having the vac/boost lines coming off the plenum. Do you run into any issue with pressure sensing with a turbulent flow? I suppose you really wouldn't because stock does the same thing.


As for the code generation for calculating intake runner length and diameter I have some good news to report:


I've also included the code below for further investigation for users who wish to run it themselves and maybe even critique my math. I wrote it in matlab which allows easy 3d surface rendering accurately.

function []=runnerlength()
gamma=1.4;
R=287;%J/kgK
T=293.15;%K

a=sqrt(gamma*R*T);%M/S

redline=8000;%rpm redline
v=0.0013;%m^3
r=linspace(0.001,0.015,100);%m
A=@(x)pi().*x.^2;%m^2

rpm=linspace(3000,10000,100);
f=@(x,y)a.^2.*A(y)./(4.*(x./60).^2.*pi().^2.*v);
z=zeros(length(rpm),length(r));
for i=1:1:length(rpm)
    for j=1:1:length(r)
        z(i,j)=f(rpm(i),r(j));
    end
end

surfc(r,rpm,z)
colorbar
xlabel('Radius, r (m)')
ylabel('RPM')
zlabel('Length (m)')

[NoDOHC]

Be careful applying Speed-of sound math here... If you do a good job on the plenum, you should be able to mostly ignore the effects of the compressibility of the air. Most of your resonance tuning is due to the Helmholtz effect (AKA: organ pipe, more of a dynamic systems model than anything to do the compressibility). The air in the plenum should not be moving anywhere near the speed of sound.

Don't think about this too hard man, intake manifold are simpler than they would initially seem. Don't ever try a dynamic model on a manifold unless you are a glutton for punishment (I have tried it, it is not easy).

Basically, your air velocity will follow the offset-sinusoidal waveform typical of an infinite-length-connecting-rod reciprocating engine (or a rotary, which has similar characteristics). The pressure drop at each transition is easily determined by using the lookup tables in the back of your fluids book, no difficult math required. Basically, you can get easy cross-sectional area requirements by taking the peak flow into the chamber and dividing it by the desired velocity (no rocket science there).

With the plenum, everyone has their own idea as to how bast power is obtained. I won't take to time required to explain my opinion on that. As I said before, you can easily find the flow through any given portion of the manifold at any given time with reasonable accuracy.

I will venture to say that I have seen tapered plenums, log plenums, cross rams, tunnel rams, inboard velocity stacks, tapered tubes, straight tubes, etc. in operation and I have not seen the simple log with beveled, constant cross-sectional area runners beat yet.

I hope this helps some. I know that math is awesome, but don't let it bog you down. Seriously, I found that going by my intuition and what feels right is often better than trying to crunch crazy numbers, there are too many x-factors to make any good simulations given the typical person's toolbox.

Edit: I hunted high and low for an intake manifold gasket and only succeeded in concluding that it is high time to clean out the garage (I know I have two brand-new ones, somewhere).

Can I take a scan of a LIM for you? (I can find that...) What measurement do you need?

[vex]

Originally Posted by NoDOHC

Be careful applying Speed-of sound math here... If you do a good job on the plenum, you should be able to mostly ignore the effects of the compressibility of the air. Most of your resonance tuning is due to the Helmholtz effect (AKA: organ pipe, more of a dynamic systems model than anything to do the compressibility). The air in the plenum should not be moving anywhere near the speed of sound.

Understood. I doubt very much that the velocity of the intake stream will be higher than .6M which minimizes compressibility of the air. If I maximize intake velocity then wouldn't that mean that I have the ability to pull higher velocity air into the intake stroke for a higher torque curve? As for the Helmholtz effect, would actually tuning for that even though I'm turbo'd be beneficial? I suppose for cruise when I'm not in positive pressure it may be beneficial though I'm having a hard time reconciling the previous posts mathematics with my engineering brain (units don't add up). Is there a full formula somewhere that I may be able to tinker with?

Don't think about this too hard man, intake manifold are simpler than they would initially seem. Don't ever try a dynamic model on a manifold unless you are a glutton for punishment (I have tried it, it is not easy).

I have access to CFD software and already have the wherewithal to create digital representation of the manifold in a few hours time. From my understanding once I have that all done and taken care of it shouldn't be more than a few more clicks and having it run to numerically solve the flow potential.

Basically, your air velocity will follow the offset-sinusoidal waveform typical of an infinite-length-connecting-rod reciprocating engine (or a rotary, which has similar characteristics). The pressure drop at each transition is easily determined by using the lookup tables in the back of your fluids book, no difficult math required. Basically, you can get easy cross-sectional area requirements by taking the peak flow into the chamber and dividing it by the desired velocity (no rocket science there).

Don't have a fluids book I have an aerodynamics book, a couple ocean engineering books, and thermo book. I'm currently taking a compressible aero course but no fluid tables for different offsets.

With the plenum, everyone has their own idea as to how bast power is obtained. I won't take to time required to explain my opinion on that. As I said before, you can easily find the flow through any given portion of the manifold at any given time with reasonable accuracy.

I will venture to say that I have seen tapered plenums, log plenums, cross rams, tunnel rams, inboard velocity stacks, tapered tubes, straight tubes, etc. in operation and I have not seen the simple log with beveled, constant cross-sectional area runners beat yet.

I imagine that the velocity increase from velocity stacks, tapered tubes, etc wouldn't yield a high enough velocity increase to matter.

I hope this helps some. I know that math is awesome, but don't let it bog you down. Seriously, I found that going by my intuition and what feels right is often better than trying to crunch crazy numbers, there are too many x-factors to make any good simulations given the typical person's toolbox.

I'd hate to be argumenitive (and I don't want to come off like that) but I do not have the typical person's toolbox: CFD, Numerical analysis, and a hand full of individuals that have 800+hp cars. I've bounced my thoughts and what I am thinking off of them and they seem to be of the mindset to use the CFD software to ensure proper air distribution to all runners is key in plenum design. I appreciate the information that you're giving to me and by no means am discounting it off hand, i'm just trying to learn as much as I can in as little time as possible.

Edit: I hunted high and low for an intake manifold gasket and only succeeded in concluding that it is high time to clean out the garage (I know I have two brand-new ones, somewhere).

Can I take a scan of a LIM for you? (I can find that...) What measurement do you need?

It's okay, I don't want you to break you scanner. I may actually be getting a LIM in the not too distant future to take the measurements off of myself. But if you're wanting to do it you'd just need to scan the profile of the mating flange, and measure say a mounting hole diameter. This sets a scale for a digital copy and will allow me to pull the measurements off the copy with out worrying if they're right.

Basically Dreal/Dscale=Dreal/Dscale: 10.5mm/1.25mm=Dreal/8.9mm; Dreal=74.76

[NoDOHC]

CFD software is outside my experience, but if it is capable of simulating the flow in the manifold accurately (assuming that it is given good data) you have an excellent opportunity to maximize your learning as you go, while using the resources that you have to their fullest potential.

As you said, you do not have the typical toolbox.

Helmholtz tuning works for any level of boost (it is basically the natural frequency of resonance of the fluid system). The speed of sound varies with density of the charge, so you will have to that into account on the Helmholtz equation.

You can write the Helmholtz equation most easily in terms of the resonance frequency (First equation) Solved for L gives the second equation. This expects a uniform cross-sectional area for the runner, as any changes in velocity will create additional and possibly conflicting pressure waves.

(a = speed of sound, V = velocity in the runner at time of wave excitation, L = Runner length from source to plenum, A = cross sectional area of runner, f = frequency of resonance (which is related to engine speed, obviously).
I hope this helps some.

With the tools at your disposal, this should be one awesome manifold.

The offset sinusoid expresses the chamber volume as a function of E-shaft rotation (the period is 270 degrees, the amplitude is 20 in3 (327 cc) (654 cc peak to peak). Taking the derivative with respect to time requires that the x axis be in time units (pick an engine rpm). This will give you the rate of change of chamber volume with respect to time, which should give you a good velocity characteristic (given port cross-sectional area). Hopefully this will be good enough input data to get a reasonable approximation of how the manifold will flow.

Keep up the good work!