Technical discussion -- Intake runner

[PrecisionBoost]

Ok, so I've been trying to figure out how to get solid data on tuning the length of an intake runner.


At this point I think my calculations and reasoning are fairly decent...... but I really don't know if it's correct or not.

I'd really like a few "technical" guys to look at this post and comment on where I may have messed up.

I've gone through it twice and I think it looks reasonably accurate.... but it's been many years since I was in Engineering taking stuff like this so I can't be sure it's correct.

All input is welcome.... please make it constructive.




For those who don't know here is how things work with respect to pulse waves generated inside an intake runner.

First off the intake valve is open and it's sucking air into the combustion chamber at a fairly fast speed.

Then the intake valve closes.... but there is still air rushing towards the valve

The air rushing towards the valve builds up pressure at the valve and then is reflected backwards.

Think of it like throwing a rock into a pond near shore.... the ripples go from where you threw the rock towards you and the shore

When the ripples hit the shore they then bounce off and start moving away from you again.

So anyways... the pressure pulse bounces off the valve and starts moving backwards towards the plenum or throttle body

When the wave hits the air built up in the plenum some of it moves into this open area and some of it gets reflected yet again and starts moving back towards the valve.

As this higher pressure wave moves back towards the valve a lower pressure is created behind it.

Now...... if you have just the right length of runner ( tube between the plenum and valve ) the return pressure wave will hit the valve just as it's opening up.

Additionally the low pressure behind the pressure pulse will suck air in from the plenum.

The net result is that you can actually get more than 100% efficency.... sometimes as high as 120%

That is to say that on a 2.0L engine each cylinder pulls in 0.5 L ( 2.0L divided by four cylinders )

At 80% efficency only 0.4L manages to make it into the 0.5L cylinder

At 100% efficency exactly 0.5L manages to make it into the 0.5L cylinder

At 120% efficency you manage to get 0.6L into the 0.5L cylinder.

Obviously your power produced is directly proportional to the amount of air you stuff into the cylinder.

For example.... lets say your intake is junk and it's only 80% efficent.... and you manage to make 100hp

If you had a 100% efficency then you would make 125hp

Guess what..... if you put on an intake that manages to be 120% efficent then you will make approximatly 150hp !!!!

So bad intake = 100hp and good intake = 150hp

Now the numbers won't be quite that pronounced...... and they will typically happen at a particular RPM level.

Manufacturers will make variable geometry intakes to take advantage of intake runner tuning at two different RPM levels.

Typically if you tune the length for high RPM the length of the runner will be short

If you tune the length for low RPM the length of the runner will be long.

If you have a short runner it will make lots of power at high RPM but it will actually hurt the performance at low RPM

If you have a long runner it will make more power at low RPM but it will actually hurt the performance at high RPM.


So that is the basic idea behind tuning the length of intake runners on the intake manifold.

Also keep in mind that it's not just the length of the runner that has to be calculated..... the length of the port inside the cylinder head also has to be accounted for.

For example.... let's say you found the length to tune it at 5000 RPM was 12"...... and you measure the distance from the edge of the cylinder head to the valve as 4"

Then the length of the runner on the intake manifold should be 8" ( since 4" + 8" = 12" )

[PrecisionBoost]

So I've looked around and found a few formulas for calculating the length but many of them are too simple.... they tend to have plug in numbers with estimates.

For example.... one formula I found said to plug in a velocity of between 400 ft/s to 1200 ft/s...... well if you use 400ft/s and calculate the length it's really different than if you use 1200ft/s

As well I calculated the actual air velocity and on most of our intakes its closer to 200 ft/s so I have no idea why they used such a high velocity number

So in the next post I'm going to go through what I think is the correct way to calculate the tube length based upon a number of values.

Generally speaking my calculations seemed fairly close to some online calculators (which seem to leave out a pile of detail.... but they still work somewhat )

[PrecisionBoost]

Ok here we go..... I'm going to use metric mostly because our displacement is in liters and I will add some imperial where I can

First off I'm going to assume that the intake tubing is 35mm = 3.5cm = 1.38" inside diameter

If you calculate the area of that tube via the formula Area = (3.1415)( radius )^2

Area = ( 3.1415 )( 1.75 )( 1.75 ) = 9.62 square centimeters

Now comes the volume of air.... lets pretend we have a 100% efficency..... then that means that each time the valve opens it manages to suck in 0.5L of air

However... keep in mind that the duration of the cam dictates how long the valve is open..... so we have to use that as a correction factor later on.

Now.... the RPM is the main thing that allows you to calculate how many liters per second ( cubic cm per second, cc's per second, CFM ) the engine uses

So lets pic a nice number like 5500 RPM to tune our port length.... since this is typically the area where maximum power is made and where the tuned length will make the biggest difference.

So.... at 5500 RPM the crankshaft turns 360 degrees 91.66 times per second.

That is to say that the crankshaft takes 0.011 seconds to make one 360 degree revolution.

Now.... we all know that the intake valve only opens every second rotation of the crankshaft (since it's a 4 stroke motor) but at this point it doesn't matter.

Basicly we want to look at a single intake stroke.

During that 360 degrees of intake stroke the intake valve is usually only open for roughly 275 degrees.... but we are going to use 240 degrees for our number because at the very start of the lift of the valve it's not letting much air in and at the end of it's cycle it's practically closed.

What we need to figure out is the number of degrees that air is actually flowing at a resonable level..... and I've estimated that 1mm of lift at the front end to 1mm at the back end is about 240 degrees.

So...... during that 0.011 second rotation the valve is only open for 240 degrees out of 360..... so that means that we need to do a correction on the speed to reflect that it has to go in much quicker.

So..... 360/240 = correction factor = 1.5

Next we take that number of 0.011 seconds and divide it by the correction factor = 0.007333 seconds.

So at this point we can say that at 5500 RPM it takes exactly 0.007333 seconds for that 0.5L volume to go into the cylinder...... so now we can calculate the speed of movement of that volume.

Now.... the smaller the intake runner tube.... the quicker the speed of the air..... so we use the cross sectional area of the tube, volume and time to calculate the speed at which each air molecule is travelling.

0.5L = 500CC = 500 cubic centimeters

So next we use this information to calculate the velocity...... ( 500cm^3/ 9.62 cm^2 ) / 0.007333 s = 7088 cm/s

Velocity of a single air molecule traveling down the tube = 7088cm/s = 70.88m/s = 232.5 ft/s

Now..... here is where the tube length becomes a major issue.

When the intake valve closes the very last molecule of air moving into the end of the intake runner from the plenum is traveling at 232.5 ft/s

If the length of the runner was say 8" and the length of the port in the cylinder head was 4" then we can say that that last molecule has to travel 12" or 1 foot before it hits a dead end at the closed valve (basicly like hitting a brick wall )

Then that molecule bounces off the valve and starts travelling back towards the intake.... so it travels another foot to get to the plenum.

Lastly it hit's the plenum volume (which doesn't act like a brick wall.... it's more like a spring ) and that molecule then bounces back and travels towards the valve yet again.

So all in all it has travelled 3 feet to get back to the valve....... this is the 3rd harmonic..... which is the ideal tuning frequency.

Now.... what we want is for the valve to open just as that molecule is coming up the valve.... because it's allready travelling at high speed it shoots into the cylinder as if it were under pressure.

So this is what causes a 0.5L cylinder to fill up with 0.6L of air..... the pressure wave pulls along air and pushes into the cylinder at just the right time..... it's kind of like a turbo or supercharger effect.

Now..... 12" might not be the right lenght.... it's possible that the pressure wave isn't at the valve.... it may be just before the valve or it may have allready hit the valve and is now travelling backwards towards the plenum again.

So..... how do we figure out the right length...... well it's actually fairly simple.... we know how long it takes to travel three lengths of the intake runner..... now we adjust that length until it is exactly equal to when the valve opens.

Now.... here is where the 4 stroke stuff comes in..... the intake valve only opens every second rotation of the crankshaft.

What we want to know is the time from when the intake valve closed to when the intake valve opens up again

The intake valve was open for 240 degrees out of a 720 degree rotation ( two 360 degree rotations of the crankshaft )

Now.... if we take 720 deg and subtract 240 that will give us the number of degrees between when the valve closed to when it opens again.

720 - 240 = 480 degrees.

Now.... back to the tuning RPM..... which we decided would be 5500RPM

We allready figured out that at 5500RPM the crankshaft rotates around 360 degrees in exactly 0.011 seconds.

If it takes 0.011 seconds to do 360 degrees then we can calculate that 480 degrees takes 0.0147 seconds

So now we know how fast the air molecule/pressure wave is traveling and we know the time we want for it to hit the valve as it opens up.

Time of travel = (3)(length of tube) / velocity of air molecule

0.0147 s = (3)(length of tube) / 7088cm/s

Length of tube = (0.0147s)(7088cm/s) / 3

Length of tube = 34.65cm = 13.64 inches

So there we have it..... if the length of the tube is 13.64 inches (intake runner + port length) the pressure wave will hit the valve just as it opens making for more than 100% efficency

If we were to guess that the length from the valve to the manifold (outside edge of cylinder head) was 4" then that means that the runners should be 9.64" long

[PrecisionBoost]

Now for other questions.....

What effect does the diameter of the runner pipe have on the tuned length of the runner???

If the diameter of the runner pipe is smaller then the velocity is higher (smaller cross sectional area).... meaning it's traveling quicker..... so the tuned length of the port will have to be longer

If the diameter of the runner pipe is larger then the velocity is slower (larger cross sectional area).... meaning it's traveling slower... so the tuned length of the port will have to be shorter.


Why are turbo manifolds shorter than N/A manifolds?

Well that is a complex answer and I have to think it's related to a number of issues.

First of all if a turbo is pressurizing the air to 15psi of boost then it will be trying to force 1.0L into the 0.5L cylinder

A larger volume of air with the same intake manifold runner diameter will result in a pulse wave velocity that is double that of the first example.... which is to say it's travelling much faster and requires a longer runner.

If all of a sudden we needed to increase that runner length from 12" to 24" it would make it very hard to fit the manifold inside the engine bay.

Additionally a really long runner will create issues with respect to throttle response and boost pressure.

The shorter the runner length the closer the cylinder pressure will be to that of the plenum.

You don't want to run into a situation where your shifting and the runners are still at 15psi and your blow off valve has dropped the pressure in the plenum to 2psi via the blow off valve

If the runners are short then odds are the pressure in the runners will be very similar to that of the plenum.

Also consider the fact that the boost is measured in the Plenum..... if the runners are long and pressure in there is different than the plenum then the ECU will inject the incorrect amount of fuel into the system.

So.... how do they manage to tune the runner length on a turbo?

Simple.... they use a different odd number harmonic.

Instead of the molecule hitting the valve then bouncing of the plenum volume and then back to the valve as it's opening they allow it to do one more bounce.

So it hits the valve, bounces off plenum volume, hits the valve a second time, bounces off the plenum volume again, then it hits the valve just as it's opening up.

That way it travels 5 times the length of the runner instead of 3.

You could also take it a step further an design it to do the 7 times the runner length which would decrease the runner length even further.


Does the pulse wave decrease in amplitude as it bounces back and fourth ?

Yes.... ideally you want to do it on third bounce.... when it hits the plenum volume the intensity of the pulse wave decreases by a significant factor.

If it hits the plenum twice that means that the pressure wave will be significantly lower so you will end up with a lower efficency.

In a turbo system that is really not a big deal because you can just add a tiny bit more boost to accomidate the lower intake efficency.

With a naturally aspirated engine you don't have the ability to add more air so you want to make the best use of the intake and tune the length for just once bounce off the plenum volume.

[Stefan]

Well, you have picked one of my favourite subjects.

First off, if you want an in depth view of the intake process I'd recommend you look at the four stroke book by Prof G. P. Blair.

In a nutshell - A negative wave is generated during the intake stroke at peak piston speed (~ 75-80 deg ATDC) and travels up the intake duct. At an area expansion, be it at the atmosphere or plenum, a positive rarefraction wave is generated that travels back to the valve and should arrive there between BDC and IVC. This wave when going outwards travels at less than the speed of sound and on the way back travels faster than the speed of sound - this is because it is not a sound wave. A IVC the wave is reflected again back towards the area expansion where it is then reflected back again. Each time changing signs. During this period the wave travels at approximately the speed of sound. From IVC to IVO it will travel backwards and forwards a number of times as determined by the intake duct length hopefully arriving at IVO to help purge the cylinder of residual gases.
A typical intake pressure trace is below (from TFX Engine Technologies site)



Be aware that the pressure peak near IVC comprises of the wave action and inertia ramming which are seperate functions.

I've done a lot of work with intakes over the last few years, tried lots of different formulas, various programs, etc. I have formulas that I now use for initial manifold dimensions that work very well.

These formulas I've put into a spreadsheet that does the hard work for me. Here is a screenshot.



When I have some more time I will post some more information for you. In the mean time you should read these.

http://hdl.handle.net/2060/19930084998

http://www.profblairandassociates.com/p ... basics.pdf

http://www.profblairandassociates.com/p ... h_Sept.pdf

Stef

[PrecisionBoost]

In a nutshell - A negative wave is generated during the intake stroke at peak piston speed (~ 75-80 deg ATDC) and travels up the intake duct. At an area expansion, be it at the atmosphere or plenum, a positive rarefraction wave is generated that travels back to the valve and should arrive there between BDC and IVC. This wave when going outwards travels at less than the speed of sound and on the way back travels faster than the speed of sound - this is because it is not a sound wave. A IVC the wave is reflected again back towards the area expansion where it is then reflected back again. Each time changing signs. During this period the wave travels at approximately the speed of sound. From IVC to IVO it will travel backwards and forwards a number of times as determined by the intake duct length hopefully arriving at IVO to help purge the cylinder of residual gases.
A typical intake pressure trace is below (from TFX Engine Technologies site)

Stef

Thanks for the input and links, I've found very few sources on intake runner lengths that seem to explain what is going on within the runner...... so I just started thinking about it and came up with some theory in an attempt to figure out what was going on.

I haven't read the links yet.... but I will probably tonight..... I've just been looking at this as if the pressure wave were an ocilating spring with an inertial mass.

Question for you.... the wave speed going towards the valve is a function of mass air flow versus cross sectional area of the tube right?

So why does the return reflected wave wave move faster than the speed of sound?

I would have thought the pressure wave would move at similar velocity to that of the initial pressure wave velocity (that is to say if the pressure wave was traveling at 750ft/s wouldn't it also rebound at slightly less than 750ft/s instead of at the speed of sound ( 1130 ft/s ) ??

Where I get even more confused is when we add pressure from a turbocharger into the equation.

I've been seriously contemplating making myself a test bench and simply fool with the intake at various mass air flow numbers.

After thinking about it I realized a testbench of this sort would be very expensive ( partially due to the fact that I would like it to create pressure on the intake side to simulate a turbo and vacuum on the other side to simulate the pistons )

So in the end I think it might be easiest to simply take a 2.0L block, connect my spare cylinder head and spool up a turbo with an external air source while rotating the crankshaft via external gasoline engine.

I was thinking a 7.5hp lawnmower engine should be enough power to get it up to speed..... but I'm not really sure if that's a reasonable assumption.

I was thinking about removing the spark plugs and putting in a one way valve so that the air inside the cylinder simply goes out the valve during the compression phase (to decrease horsepower requirements)

I'm only really interested in the intake stroke.... so I figure that it shouldn't matter if I skip the compression stroke.

Then I was thinking I would have one MAF on the intake side and four MAF's on the exhaust side to measure the output of each individual cylinder.

I would plumb the one way valve into the individual exhaust pipes prior to the exhaust side MAF's to ensure that all airflow it accounted for.

I know this isn't a "perfect" replication of a running engine..... but it should be pretty close right???


This way I can try different designs and see if I get even flow numbers from each cylinder.... and I'd also like to plumb in four MAP sensors to measure pressure in each runner ( figured I could simply use the injector holes )

So what do you think??? Am I crazy??? I allready have all the parts I need to make this work.... so it's definitly tempting to try it out.

[PrecisionBoost]

Here is a general idea of the test bench I was thinking about building using a 2.0L engine.....




The basic idea is that I could use a feedback circuit on the intake MAP to control a drive by wire throttle in order to control the flow of air to the turbine side of the turbo.

That way I can pretty much set the desired boost electronicly.

I would probably use a large dust collector fan running directly off the small gas engine to produce the pressure required to spool up the turbine.

I'd probably also add in some IAT sensors to monitor temperature within the system ( which obviously affects the speed of sound quite significantly )

Still not sure what size of engine I'd need to rotate the engine up to 6000 RPM and running the turbine air source.... I guess I also have to account for the energy required to run the oil pump system as well.

In any case.... I'm sure I could pick up a suitably sized motor.... or if need be.... I will build a frame for one of my 1.6L motors and use it to run the 2.0L engine.


So.... do you see any major flaws in my idea for a test bench???

[PrecisionBoost]

Hmmmm..... I was thinking it might be a bad idea to use those one way valves in the spark plug holes because it will make data aquisition a bit harder.

If I let some air out the valve and some out the exhaust valve there will be a complete rotation between events.... I will then have to sum the voltage from the two events to get the total airflow during the pulse.

If I let it go through compression it will make it easier to capture a single MAF event..... even if it means I have to use a much larger motor to run it.... I think it will be easier in the long run.

[Stefan]

Question for you.... the wave speed going towards the valve is a function of mass air flow versus cross sectional area of the tube right?

So why does the return reflected wave wave move faster than the speed of sound?

I would have thought the pressure wave would move at similar velocity to that of the initial pressure wave velocity (that is to say if the pressure wave was traveling at 750ft/s wouldn't it also rebound at slightly less than 750ft/s instead of at the speed of sound ( 1130 ft/s ) ??

The wave speed is dependent on the cross-sectional area of the duct and the mass flow within it. On it's way out from the valve it is slowed down by the massflow going into the engine. On the way to the valve it's speeded up by the massflow going into the engine. *edit* The wave also imparts motion to and from the massflow.
I know this seems really confusing and I'm probably not doing a good job of describing it. Prof Blair does a much better job at this!
An explanation of events is in the following paper(s)...

http://www.lesoft.co.uk/files/Bratislava2001.pdf
http://www.lesoft.co.uk/files/Detroit2003_0370.pdf
http://www.lesoft.co.uk/files/ASME_ICES ... 44_FIT.pdf

The last two papers deal specifically with exhaust ducting but the wave action process in the intake is similar.

Where I get even more confused is when we add pressure from a turbocharger into the equation.

It's not confusing. A turbo or supercharger just provides a different atmosphere from an engine dynamics point of view. The wave profile is just superimposed on top of the nominal intake pressure, be that 14.7psia or 29psia. Wave speed changes due to the higher intake temps and inertia ramming changes a little due to the higher mass flow.

I've been seriously contemplating making myself a test bench and simply fool with the intake at various mass air flow numbers.

After thinking about it I realized a testbench of this sort would be very expensive ( partially due to the fact that I would like it to create pressure on the intake side to simulate a turbo and vacuum on the other side to simulate the pistons )

So in the end I think it might be easiest to simply take a 2.0L block, connect my spare cylinder head and spool up a turbo with an external air source while rotating the crankshaft via external gasoline engine.

I was thinking a 7.5hp lawnmower engine should be enough power to get it up to speed..... but I'm not really sure if that's a reasonable assumption.

I was thinking about removing the spark plugs and putting in a one way valve so that the air inside the cylinder simply goes out the valve during the compression phase (to decrease horsepower requirements)

I'm only really interested in the intake stroke.... so I figure that it shouldn't matter if I skip the compression stroke.

Then I was thinking I would have one MAF on the intake side and four MAF's on the exhaust side to measure the output of each individual cylinder.

I would plumb the one way valve into the individual exhaust pipes prior to the exhaust side MAF's to ensure that all airflow it accounted for.

I know this isn't a "perfect" replication of a running engine..... but it should be pretty close right???


This way I can try different designs and see if I get even flow numbers from each cylinder.... and I'd also like to plumb in four MAP sensors to measure pressure in each runner ( figured I could simply use the injector holes )

So what do you think??? Am I crazy??? I allready have all the parts I need to make this work.... so it's definitly tempting to try it out.

A test rig is a good idea and has been used by many researchers in the past for the study of inlet dynamics. You would need to use some Kistler pressure sensors as the map ones don't have enough resolution or are fast enough.

http://www.kistler.com/medias/7ayeO427s ... BEd-30.pdf

Also worth reading this journal http://www.fs.uni-lj.si/sv/English/2004/1/sv-01-an.pdf from page 56.

Stef

[PrecisionBoost]

Thanks for the other links..... what I meant about the turbocharger adding difficulty is more or less that it's not a static pressure.... that is to say that the boost will be rising during the mid band.... and I really don't know how to calculate that pressure without taking physical measurements of the boost as the turbine spools up.

That is strange.... here at work my drawing of the test bench doesn't show up.... weird.... I will have to check it when I get home.

[Stefan]

Thanks for the other links..... what I meant about the turbocharger adding difficulty is more or less that it's not a static pressure.... that is to say that the boost will be rising during the mid band.... and I really don't know how to calculate that pressure without taking physical measurements of the boost as the turbine spools up.

That is strange.... here at work my drawing of the test bench doesn't show up.... weird.... I will have to check it when I get home.

It doesn't really mater that it's not a static pressure - to a lesser degree it's not static in an N/A application either. For basic determination of intake lengths and volumes in a turbo application you only need to make a good estimation of the air temperature inside the intake. Despite the apparent lack of variables involved in the simpler calculations the use of the correct formulas give surprisingly good results.
Further calculation of intake pressures and mean velocity in the intake is light years beyond what I could post here. There is a reason why the good intake wave action programs are so expensive and that's because of the difficulty in accurately modelling this phenomena.
I've given links to some of the better free information on the web. For better info you will have to pay for it. i can give you a list of SAE papers that you would find very useful. I personally, over the last two years have collected a 3" stack of SAE papers on intake tuning alone. Out of all of that I still go back to a select few simple formulas because they work extremely well for initial sizing.

Regarding a test setup, you'll need more than 7.5hp to spin an engine. If you take the average 2L engine producing 150hp at 6000rpm it'll have approx 80% mechanical efficiency, possibly less. Thats 35+hp lost to friction etc. You would need a minimum of a 20hp motor to spin an engine up to a sensible speed for data collection.

[PrecisionBoost]

I was under the impression that many of the mechanical losses are more or less related to things like power steering, alternator, water pump, oil pump ( of which the oil pump is the only thing I have to worry about )

The way I see it is that I can turn the engine over with the spark plugs with very little torque.... perhaps 2 ft-lb of force is required.

If horsepower = torque X RPM / 5252 then the horsepower requirement at 6000 RPM is roughly 2.5hp.

I was guessing 2 or 3 horsepower for the oil pump.... but I really have no idea.

I could be completely wrong, but I don't see friction being a really huge issue when the engine is not actually running and producing power.

To me.... the friction arrises when loads are placed on the engine.... perhaps there will still be some from the acceleration/deceleration of the pistons but I can't imagine that it really adds up to much.

[Stefan]

There is some hp lost to accessories but there's is still a decent amount of hp consumed by friction etc. These losses go up by a power law with rpm so simply taking a torque reading at practically zero rpm won't work. The losses can be roughly broken down,

1.Valvetrain friction
2.Cranktrain friction
3.Piston and ringpack friction
4.Windage losses

Although 2 and 3 won't have any combustion loads they will still have inertia loads which will increase with the square of the engine speed.

The slower you run the engine the less hp you will need to drive it, but to see decent pressure traces in the intake you will need an induction length suitable for that rpm which would be quite long.

To get data with the use of a standard type of intake you will need to run it at 3-4000rpm minimum. You could maybe do this with a small 4cyl engine (maybe a motorcycle engine?) with your 7.5hp. A bigger engine may need 10+hp. The smallest size of Spintron available is a 20hp drive.

Lots of research engines are single cylinder and testing on a small multicylinder engine would be a better way to go for your research perhaps.

Stef

[PrecisionBoost]

Fair enough.... I have several other engines kicking around that I could use to power the "test bench".