Is a Split Plenum a good idea? 1

[Gary_321]

I've noticed something glaringly obvious.

My six-cylinder engine is effectively two three-cylinder engines with a common crankshaft.

Cylinders 1-3 have a common exhaust manifold as do cylinders 4-6.

The intake has individual throttle bodies with all six cylinders have a common plenum.

A six-cylinder has a cycle phase of 120° and because the inlet valve has a duration of 293°, there is a 173° overlap.

Graph showing six-cylinder inlet valve interaction

In the linked image, the red line shows the cycle that is following the blue. Positive values show "air" moving from the cylinder into the plenum, negative values show "air" moving from the plenum into the cylinder. You can see that some "air" would move from 3 to 5.

The firing order is 1-5-3-6-2-4. 1 & 5 are four cylinders apart and 5 & 3 are two cylinders apart. This means that 5 will influence 1 less that 3 will influence 5.


To almost eliminate this inlet valve interaction, I propose to split the plenum in two. As the phase is now 240°, the overlap will be only 53° and the flow from consecutive cylinders would be significantly reduced. Here is the inlet valve overlap graph...

Graph showing three-cylinder inlet valve interaction


The proposed modification...

Plenum before...

Image of unmodified plenum
Image of unmodified plenum

Plenum after...

Image of modified plenum

I'll replace the cardboard template with aluminium sheet and secure it with epoxy resin.


Do you think this is a sensible thing to do?

 

[Lou Scannon]

It depends. It would be good to understand the OEM designer's rationale for doing it the way they did, and deciding from there why and how to modify.
Plenum volume has a role to play in volumetric efficiency as well as throttle response, so you should understand what trade-offs were made in the OEM design in these areas before you proceed with modifications.
Keep in mind that volumetric efficiency, not to mention throttle response, is highly rpm dependent, so you need to have a goal in mind for your torque curve, and accept that the torque curve is directly influenced by VE as a function of rpm.

"Schiefgehen wird, was schiefgehen kann" - das Murphygesetz

 

[BigClive]

Individual throttle bodies? What sort of engine is it?

 

[Gary_321]

It's a BMW S50B32. I have an M Roadster. It is the same engine that was fitted to the Euro spec E36 M3.

 

[EdStainless]

It is likely that your mod would improve WOT while hurting part throttle and lower end performance.

= = = = = = = = = = = = = = = = = = = =
P.E. Metallurgy, Plymouth Tube

 

[SwinnyGG]

Intake overlap doesn't cause the behavior your plots say it does.

I don't have your equations to do the math that generated your chart. Where did you come up with this?

If you're plotting flow rate vs time (which you are, using cylinder angle as a proxy for time) than the integral (the area under the curve) of the plot will give you the total flow. Since we don't have access to the math you used to create these, we can only evaluate visually- but from the plots you've created it appears that your model shows Cyl 3 taking in very little total volume over its whole intake cycle. (area under the positive curve minus area under the negative curve looks like a very small number).

This is not how engines work.

Intake reversion is certainly a thing that can exist- and it will continue to exist until someone comes up with a valve actuation system that is continuously variable and universally applied. But the amount of reversion that actually happens is small, and only happens in certain parts of the RPM range of an engine. The true fix for reversion is porting; basically, reversion is caused by restrictive ports which don't allow the intake air to keep velocity through the port throat. Porting the cylinder heads in a way that improves low-valve-lift intake flow fixes this problem.

On top of that... The modification you've pictured takes the plenum opening and cuts it in half. Any cylinder with an open intake valve is now drawing air through an opening that is half as big as it was. This is bad. The point of having a large plenum volume is that it provides dampening for pulses created as the valves open; reduce the plenum volume, and you reduce the damping. That's also bad.

In short, don't modify the plenum. Leave it alone. If you want to make the most power possible, remove the plenum entirely and put a filter on each throttle body. This, in effect, gives you a plenum of infinite volume.

 

[RodRico]

Gary,

Now that you've identified the engine as one of those used in the BMW M-series, I'm not very optimistic about making engine mods without a deep understanding of the current engine's flow over RPM... BMW is no slouch at performance tuning, and this particular S50 engine is the product of more than one phase of evolution over several years.

Some people have, of course, modified their M-series cars for even more performance, but they typically use off-the-shelf parts designed by engineers who have comprehensive design tools (FAE, CFD, etc.) and/or good test facilities (flow bench, dyno, etc.). If partitioning the intake were effective, I imagine someone would have done so an after market performance manifold.

I found a website that calls out some off-the-shelf mods for the S50B32. I certainly can't vouch for its credibility, but the website at

https://en.wikipedia.org/wiki/BMW_M3

says,"[The S50B32] motor showed 321 HP at 7,400 rpm, and a torque 350 Nm at 3,250 rpm. The best mods for your S50B32 are following: a cold air intake, Schrick 284 camshafts, with valve springs, headers, performance exhaust system (or straight pipe). Then you have to adjust ECU for new performance parts. As a result you get about 360+ horsepower. For even more power, you need to remove VANOS. Then buy 316/308 Schrick camshafts, tappets and springs. You will also need lightweight rods and high compression pistons (CR~13), large valves and port and polish cylinder head, airbox, fuel injectors 500 cc/min. These mods will increase the maximum revs to 9,000 rpm and develop to 400 horsepower."

If you really must do it all yourself, maybe you can copy some of the after-market components made for your engine.

Rod

P.S. A bit more research...
No plenums listed as E36 performance parts on Dinan's site:

https://www.dinancars.com/products/?keywords=E36

No plenums listed in this full tweak of an S50 engine:

http://www.eatsleeptinker.com/2015/01/10/fs-bulletproof-s50-arp-s54-pan-mls/

(the guy who did the mods works at

https://www.moreheadspeedworks.com/product-category/engine-drivetrain/#...

no plenums)

 

[Gary_321]

I've only shown the intake overlap period as that is the period when the cylinders interact.

The maths that I used was the in-cylinder pressures from http://performancetrends.com/Definitions/Images/Cylinder-Pressure-Lrg.gif]here[/url] and the minimum between the inlet valve curtain area or the most restrictive area in the inlet valve throat.

Inlet valve curtain = pi * 35 * valve lift * 2 (35mm valve head and 2 valves)
Minimum inlet throat area is around the valve stem. By measurement, I estimate this as 333 sq mm in each intake valve throat, so 666 sq mm because there are two valves.

By estimated measurement with a tape-measure, the plenum is 10,000cc and the "scoop" to the MAF is 1800cc. The "scoop" will remain standard. The engine is 3201cc (each cylinder is 533.5cc) and max rpm is 7650rpm.

I understand (and agree) with the dampening comment, but the way I have looked at it is...

Firing order is 1-5-3-6-2-4. If you split the plenum into Left and Right plenums...

1-LEFT
5-RIGHT
3-LEFT
6-RIGHT
2-LEFT
4-RIGHT

Rather than using a 10 litre plenum to feed six cylinders, I would be using 5 litres to feed 3 cylinders. In theory, the new plenum is 9 times bigger than the cylinder.

 

[SwinnyGG]

So you produced that plot by hand-calculating valve flow?

Using what pressures?

You're basing a lot of things on a port flow chart that is not accurate.

Your chart indicates that air is flowing OUT of the intake port for nearly 90 degrees of crank rotation. It also indicates that the peak flow rate of air being pumped back into the plenum is higher, by 50%, than the peak flow rate of air flowing into the valve later on.

This chart is nowhere near realistic. You should throw it away and start over, instead of using it to make decisions.

 

[Gary_321]

jgKRI - I've expanded the graph to show all 720° and the net intake value is obviously wrong .

I need to have another think about this.

I agree with your suggestion about no plenum, but I would need six new MAFs and an ECU to control the fuel.

 

[SwinnyGG]

You could move to an Alpha-N control scheme, although with your stated goal this doesn't make any sense.

I was just using that as an example of the other end of the scale with regard to plenum size to try to illustrate why you don't really want to modify the plenum that you have.

I'm still trying to figure out exactly what part of your model makes you think that intake ports open at the same time causes air to flow between cylinders. This simply does not happen.

There's a couple of things I think you may be neglecting when you think about this:

1) Air is compressible. The pressure inside the cylinder is lowered as the intake valve opens and the piston starts to move down during the intake stroke- but the transfer of negative pressure happens initially as a wave up through the port into the manifold. In other words, pressure takes time to build. It isn't instantaneous. The time is microscopically short, BUT at high RPMs processes that take a single millisecond become significant.

2) Air has inertia. Air flowing into the port is stopped by the back of the intake valve after it closes, and the inertia of that air doesn't just disappear- it is transformed into local pressure in the port, acting against the back of the valve. This means that at certain RPM points, it is possible for there to be positive pressure, above plenum pressure, on the back of the valve as it starts to open. This helps prevent reversion and is very significant.

3) The plenum is unthrottled. This means that it is always full of ambient pressure air, and acts as a near-infinite reservoir (the only pressure drop is due to the maf/filter/plumbing upstream). The dampening you get from the plenum is due to wave propogation, but if you neglect that, you can model the plenum as basically an infinite reservoir with a static pressure slightly below ambient. This pressure is very, very, very close to ambient at idle and moves away from ambient as RPM increases, due to increasing flow over a filter element and plumbing which have not changed between RPM points.

Point is, in order for air to flow OUT of an intake port and INTO another cylinder, the pressures inside each cylinder mean nothing- for air to flow OUT of an intake port, the pressure in the cylinder needs to be higher than pressure in the plenum (which only happens right at the end of the intake stroke as the piston moves up from BDC) AND the column of air moving through the port needs to have had enough time working against that cylinder pressure to stop and change direction.

The combined effect of pressure 'stacking' on the back of the valve and the inertia of the column of air in the port is significant. There are certain highly optimized engines (think F1, motorcycle engines, etc) which are capable of producing VE greater than 100% in certain parts of their RPM range precisely because of these effects working in combination.

 

[Gary_321]

I think that I need to do a short video to explain what I am thinking. It helps me to get my head around it, as well as being able to seek advice.

At times like this I wish I didn't have variable valve timing! It would make the "problem" much easier to understand.

At the moment, I'm charting all the cylinders and their valves and there are definitely times when multiple inlet valves are open AND there is an exhaust valve in overlap.

For instance,

At a certain at one point (angles are +/- 5°)...

Inlet#1 is just opening
Inlet#2 is just closing (45° to close)
Inlet#4 is open
Exhaust#1 is closing (85° to close)
Exhaust#5 is open

I can ignore E5 as its "air" will go out the exhaust pipe, since I5 is closed.

E1 and I1 have a 91° overlap, so there is a mixture of E1, I1, I2 and I4 for at least 45°.

All that "air" is able to flow either into or out of the plenum. Any uncontrolled "air" movement is undesirable.


I agree with you that any "air" coming out of a cylinder is unlikely to flow into another, but I think it is worth investigating.

I think that I may have fallen into the trap of thinking of air as a particle that moves from A to B. Instead, I should think of it as a particle that bumps into the particle next to it - and it's the pressure wave that moves, not the air molecule itself.

 

[Gary_321]

I've done the analysis of the valve timings. The timings are different for each variation of the VANOS* angles. These timings are for a VANOS position during a 0-60mph test

At least 2 inlet valves are open 100% (720°) of the time.
At least 2 Inlets PLUS an overlapping exhaust valve are open 73% (528°/720°)
Only 3 inlet valves PLUS an overlapping exhaust valve are open 44% (318°/720°)

Cylinders 1, 2 & 3 in isolation...
2 inlets PLUS an overlapping exhaust valve are open 22% (159°/720°)
Cylinders 4, 5 & 6 will be the same as 1, 2 & 3.

*VANOS is the variable cam system

Valve Chart
Each line represents the period the valve is open.
Inlet#1 is the bottom row, Exhaust#6 is the top row
The left end of each line is when the valve opens, the right end of each line is when the valve loses.

 

[SwinnyGG]

I think I understand now how you're getting to that conclusion... maybe.

Like I said in my earlier post- the fact that air has inertia is the reason that what you're saying happens, doesn't happen.

The air in the plenum is not static- is is moving, constantly, and it is under higher pressure than the air inside the cylinders (while the intake valves are open) almost all the time. During the (short) times when the intake valve is open and cylinder pressure is higher than plenum pressure, inertia of the air in the ports takes over and the cylinders continue to fill.

You can look at the air as individual particles- because it is. But the particles aren't static, they are in motion and under varying pressures all of the time.

 

[Gary_321]

During the (short) times when the intake valve is open and cylinder pressure is higher than plenum pressure, inertia of the air in the ports takes over and the cylinders continue to fill.

I agree with this ATDC, but from approx 70° BTDC, the inlet valve is open and the piston is rising. From around 70° BTDC, the exhaust is closing, the inlet is opening and the piston is rising.

I think there is the possibility that "air" at 23psi in the cylinder could choose the 15psi plenum over the exhaust as the easiest exit.

This "air" could have the positive effect of pressurising the plenum, or the negative effect of disturbing the flow of air into the cylinder that is 120° further through its phase.



Valve Lift Chart showing opening/closing times (720° = TDC)

 

[SwinnyGG]

Once again- you're neglecting velocity and the inertia of the air in the port. The system is not static- if you try to analyze it like it is static, all of your conclusions will be incorrect.

You seem to be of the mindset that 'it's just air, how fast could it be moving'. This is a common line of thought in the minds of people who have never done a deep dive into cylinder head design.

Well.....

Your car's S50B32 makes approximately 320 brake horsepower.

From that number we can directly approximate the amount of air the engine consumes- this works out to about 220 cubic feet per minute for 320 bhp.

220 CFM x 28316.8 cc/CF = 6,229,696 cc per minute. That's roughly a cubic meter every 10 seconds, or 100 liters per second. We're talking about a LOT of flow here.

6,229,696 cc/min / 60 s per min / 12 ports = 8652.36 cc per second per port.

Given your flow area of 333mm², this gives an average port velocity of 26 m/s.

But that's just an average- the intake port is actually only open about 40% of the time (290/720).

Average port velocity for just the time the port is open is 65 m/s. That's almost 150 mph, and that's just the average for the time the valve is open!!!!!!! That velocity takes time to build, meaning that the true peak velocity in the port is much higher.

The cylinder pressures are not high enough to slow down this high speed jet of air, stop it, and accelerate it in the other direction.

It is not just about pressure- you cannot neglect the inertia of the air in the port. It is massive.

The air in the cylinder is also a compressible fluid. When the piston starts back up, it is accelerating. These components are moving VERY fast- so fast that the air pressure exerted on the face of the piston is not the same as the local air pressure in the flow area of the valve.

This is a high speed machine. You can't evaluate it by disregarding the kinetic energy of the moving parts- and that includes the air being pumped.

 

[Gary_321]

I wish that I had done the calculations earlier...

I reckon that only 35cc would be "pushed" into the inlet tract. The volume of the inlet tract from trumpet-to-valve is 700cc.

 

[SwinnyGG]

Don't worry about it- this is why this forum exists.

I hope my posts have come across with a tone of wanting to help, and nothing else.

As a guy whose weekend toy is an "antique" BMW ('95 M3, heavily modified) I wish you the best of luck.

 

[Gary_321]

As a guy whose weekend toy is an "antique" BMW ('95 M3, heavily modified) I wish you the best of luck.

No problem with being advised that I shouldn't go down a certain path. I've seen enough posts on the internet to know that not being face-to-face can cause the wrong thing to be inferred. If I thing that I've got a problem, I'll ask for confirmation before assuming the worst .

How did you decide on your modifications?

As you can tell, I'm trying to confirm that a mod has merit before parting with any cash. I'm a bit more busy putting the car back together now, as the weather is dry.

A question from something that I read last night (if I need to start a new thread, just say)...

The text on this website suggests that the pulse is generated by the piston changing direction at TDC, but the calculation suggests the pulse is generated by the inlet valve closing. The text also suggests that you want the pulse to arrive as the inlet valve is close to closing, but the calculation suggests you want the pulse to arrive 30° after opening.

[edit] I've just re-read the website and the calculation agrees with "The third and most complex cause of pressure waves".

 

[SwinnyGG]

The answer to that last question is that both events cause a pulse. Any component which

-causes a volume change in an airflow area
-causes a cross-section change in an airflow area

will cause a pressure pulse (either positive or negative).

It is absolutely correct that the interactions of all of these discrete pulses are complicated. Because they travel as waves with fixed speed, they don't affect the timing of other pulses very much, but they do cause scalar changes in magnitude when they interfere.