Some exhaust porting rules of thumb. I've picked these up from reading speedtalk.com, emailing with Ake, playing with PipeMax, etc. They'll totally be worth what you paid for them! ;-)

The S4 stock exhaust port is 40mm (1.575") diameter round hole at the exhaust port flange. The area of that hole is 1256 mm^2 (1.95 sqin) With performance cams that have a lot of overlap (not necessarily passing the smog tail pipe tests in some of the **** states), that is quite a large exhaust port that's ideal for hp that almost none of these engines will ever see.

The good velocities for the exhaust port (when tested at 28" inches on the flow bench) are in the 210-310 fps range. With a lot of overlap and you're probably better off at the 310 fps end of the range while with less overlap you're probably better off at the 210 fps end of the range. Better shaped ports can take higher velocities, over 310 fps. The 928 S4 has a very well shaped ports.

According to PipeMax, the recommended cross-sectional areas for the exhaust port for a stock S4 engine are 1.049-1.554 square inches. The stock exhaust port is way outside of that range large. In fact, if the recommended sizes produce 310-210 fps velocities, the stock port will produce 167 fps velocity. Oops.

Another way to ask the same question is what's the crank hp range for normally aspirated engines for which the stock port size is appropriate. According to PipeMax, the least powerful hot rodded 928 S4, baby girl cam engine for which the stock exhaust port is appropriate at the maximum size of the recommended range makes 425 hp. The most powerful engine, with nice high-overlap sports cams, for which the stock size exhaust port is at the small end of the range (with velocity of about 310 fps) makes 660 hp.

To summarize: Unless you are planning to make more than 540 hp normally aspirated (or that multiplied by the boosted absolute pressure ratio), it is highly, highly likely that you shouldn't increase the size of the exhaust port at the flange. The only clear case for making the exhaust port a tiny bit larger is for engines that make more than 660 hp normally aspirated.

Since the 40mm exhaust is too large for most 928 S4 motors, the headers that are usually sold are also too large. In my opinion, the header pipe with 1.75" OD is too large for most 928 engines. The most powerful normally aspirated 928 engine that's making over 700+ hp at the crank sports 1.75" OD full-length headers without any tricky-thin wall thickness. That's estimated 1.625" ID for the full length of the header. And the engine runs like a raped ape well past 7500 rpm self-imposed redline.

For the stock S4, pipemax recommends header primary pipes with 1.362-1.55" ID, which are all smaller than the stock exhaust port opening at 1.575" ID. Usually, one does not want to go smaller than the exhaust port opening with the header pipe ID because that creates an awkward step. In the case of most S4 engine, one probably should step down into a smaller header size and taper the flange and stretch the header pipe at the flange a little bit. Calvin Easton of http://www.elstonheaders.com/ has succesfully fabricated a lot of tapered flange, stretched pipe combos that reduce the header pipe down from the oversized exhaust port, so don't laugh at me and that suggestion! (His exact reducing methods are a business secret...)

So my suggestion is that most people hot rodding the 928 S4 should probably leave the exhaust port the stock size and get the smallest header primary pipes that they can smoothly mate to the exhaust port in the head without a large and awkward reducing step. This is my opinion, others like the step there -- I'd like to have no step there and then step say 10" later in the header pipe.

In terms of what to do with the port upstream, here are some suggestions based on what I've read (but I've never personally tried these):

1. Maximize short side radius. Do whatever necessary / possible to maximize the minimum short-side radius of the exhaust port floor. That is, port the exhaust floor to have a high radius, with the constraint that the short side tangent has to be parallel to the flow at the valve seat. The exhaust port at the flange can be lifted up within the seat ring, much like Porsche did. As long as the header then doesn't make a cheater turn with an angle, lifting the port up should always help. No cheater turns in either end.

2. Smooth out casting imperfections. I would smooth out the casting imperfections in the exhaust port but try not to increase the port cross-sectional areas. These should help by common sense, especially when flow is very fast.

3. Choose the appropriate exhaust throat diameter. The 928 S4 exhaust port throat is about 28mm. That represents about 85% throat-to-valve diameter ratio. There's nothing wrong that I know with that ratio, although it's in the low end of the range for high-performance heads. Looking at the glass half full, what that gives one is the flexibility to increase the exhaust throat size a little bit, up to say 88% ratio, _IF_ you need the additional area for your super-duper high power car. If you don't need the area and would in fact need a smaller port to keep the velocity up, then somewhat tautologically increasing the throat area will hurt.

4. Keep the port approximately constant CSA from the throat to flange. This is a new rule from Ake. Whatever you set the throat cross-sectional area to be, use the same area at the exhaust port flange.

5. Wide valve margin. Leave the exhaust valve margin as large as you can (within reason if working with custom valves, as large as you can if modifying stock vales). The exhaust valve margin is a part of the funnel into the port.

6. Radius the valve face/margin. Unlike with the intake valve, radius and/or top cut the edge between the valve face and margin. In an intake valve, you want to leave that relatively sharp. On an exhaust valve, you want a generous radius. This is the "bellmouth" flare into the exhaust port.

7. Back cut the valve but preserve the tulip shape. Ake recommends a 30-degree valve back cut after the 45-degree valve seat angle. I would keep that 30-degree valve back cut short to preserve as much of a tulip shape as possible. There's a widespread consensus that tulip shape works the best as an exhaust valve. While the intake valve works better with more of nail than tulip shape (Ake's results are a powerful demonstration of that), the exhaust valves work the best with a fat tulip shape.

8. Wide exhaust valve seat for reliability. The exhaust valve seat width should be at least 1.5mm for our size exhaust valves. It won't really hurt flow or power and it will improve reliability.

9. Radius the seat downstream of the the valve seat. The seat ring should be radiused. Ake recommends the below cutter profile that has a radius downstream in the exhaust port and a sharp top cut in the combustion chamber. My theory is that the sharp top cut may help reversion, as the flow outside thru the exhaust valve doesn't care about the sharp edge and large change in the angle but the (undesired) flow back from the exhaust port into the cylinder may separate at that sharp edge. Just a theory.



10. Ignore the flow bench CFM numbers for the most part. . For the most part, the flow bench CFM numbers don't tell one much about whether the exhaust port works or not. This is the opinion of multiple experienced sources, who do spend a lot of time on the flow bench when porting the intake ports. One can detect some obviously stupid stuff on a flow bench, but especially at low lifts the flow bench isn't replicating the blowdown conditions very well.



Agree/disagree, missed something?

 

I pulled some ported exhaust port photos from the internet with google. I think these ports clearly display different exhaust porting philosophies and the intended power ranges of the engines.

928 Motorsports stock exhaust port and their modified exhaust port right next to it:



The stock port on the right has a 46mm inside diameter of the gasket groove and 40mm circular exhaust port. The exhaust port is offset upwards within the gasket groove such that the margin between the port and the gasket groove is 2mm at the top and 4mm at the bottom. The divider is sharpened. Exhaust valve guides are truncated.

The 928MS exhaust port is taken to the max diameter with just enough margin to contain the gasket ring. I am eyeballing a 44mm+ exhaust port diameter, which corresponds to at least 1520 mm^2 (2.36 sqin) area.

Ake's plans for the exhaust port of his big engine. This engine is designed to be very large and very radical in terms of rpms and camshaft overlap:



Ake's plan is to radically alter the exhaust port shape and sealing method to get both better flow and larger cross-sectional area for a big, high rpm, radical engine. The cross-sectional area is 2.6 sqin which corresponds to 1.82" diameter round pipe. According to PipeMax, this exhaust port size gives a 300 fps very high velocity port for a 850+ hp normally aspirated engine with significant camshaft overlap.


Greg Brown exhaust port posted on the RL:



I am eyeballing this to be about 42mm diameter port with a circular shape, placed concentrically inside the gasket ring. Material has been removed from the bottom of the port to move the short side exit 2mm lower. Cross-sectional area about 1385 mm^2 (2.15 sqin), again based on eyeballing the port. The divider has been sharpened. Valve guides have possibly been shortened somewhat.

My "head case" exhaust port:



Almost exactly the stock size, maybe a tiny bit off from the top moving the port a tiny bit up. The valve guides have been shortened a little bit (they are aftermarket guides anyway, so speaking relative tot he originals) and the divider has been sharpened somewhat.

Mark Lowe exhaust port from extensively modified heads:



Almost exactly stock size, maybe a tiny bit off the top. The diameter measures maybe 0.5mm (0.02") over the stock exhaust port at the manifold flange. The exhaust port volume is 7.8% greater than the stock. In the divided part of the port, the exhaust port diameter is increased 2-3% depending on the exact location. The valve guides are left relatively long, but reshaped somewhat. The port divider is not ported to be particularly sharp.

Above are all just eyeball measurements, with some exceptions for the stock and the last head. I do have more exact numbers of some of the above ports in a spreadsheet somewhere, but the exact numbers aren't that important for the topic at hand.

One practical consideration for the normally aspirated folks is that the most common and commonly available header pipes are 1 5/8 and 1 3/4 OD 16 gauge size. That means either 38.0mm (1.495") ID or 41.1mm (1.62") ID after the flange. Header builders seem to want the head porters to match the exhaust port sizes with standard tube sizes or they get very mad. In our case, we'd either have to open up the (already too large for most purposes) exhaust port by 1.1mm or fabricate a reducer flange that smoothly transitions from 40mm into the 38mm ID 1 5/8 header tube. If opening up the exhaust port, I would open it up on the sides by about 1mm per side into an oval or elliptical shape, match that shape at the flange, and then press flatten the 1 3/4 OD header tube flange end into that oval or elliptical shape. If reducing the header, I would get a conical steel die and expand the 1 5/8 header tube into 40mm ID size at the flange. Neither is perfect, but then again perfection is not a good standard when trying to get something actually done.

Any thoughts on these different exhaust approaches?

 

I spoke to and read some writing done by a retired nasa rocket scientist who told me about the exhaust charge going supersonic and producing a sonic scream. He has a few records in NHRA, and had designed heads for ford racing. The effects are on much more than just the valve shape, and extend all the way to the headers..

You measure intake on the flow bench, and exhaust on CFD model using heat and other factors from the explosion in the cylinder. Yes, math is a tyrant, but because of pressure and temperature, you're not going to be able to do this on a flow bench..

Based on his inputs, I made a comment a while ago that all 928 exhaust headers I have seen have not been designed with the scientific principles he mentioned to me..

Imagine calculating the values of heat and pressure based on the fuel and air mass in the cylinder and then adjusting for unburned fuel after the ignition of the charge in the cylinder.. Sonic scream is not a myth in exhaust port and header design...

Even misinformation has value, depending on what you intend to use it for..

Having the knowledge to tell the difference between information and misinformation is the key, no?

Sent from my iPhone using Rennlist

 

There's definitely misunderstandings, misinformation, and flat-out deliberate lies being spread on RL about this porting subject. I am too old for those kinds of games, so what I've posted in this thread over the last two weeks is what I actually believe is true. That doesn't mean it's necessarily true, it just means that I am acting in good faith.

I agree with you on a lot of what you wrote.

However, I feel that CFD is too often used as a bail-out from having to do some old fashioned analysis. Yes, it would be nice to be have access to CFD software and models of the engine gas exchange. At this point at least I don't have that access, so is there nothing that can be done without them?

I think one should be able to do a lot of old-fashioned engineering analysis that predates widespread availability of CFD. A lot of very powerful engines were designed in the 1970's and 1980's, with minimal help from computers. We should be able to do the same today, no?

The case in point is the sonic velocity of the exhaust port. If we know the valve seat geometry, valve diameter, etc. we can compute the valve curtain area at each point. We know the camshaft profile. We can also use cheap simulation programs to estimate the cylinder pressure and temperature at each crankshaft degree, ballpark. (Or better yet, measure them, but that is a big project even after shelling out the bucks for the measurement equipment.) This should allow us to use 1950's rocket scientist formulas to compute at what lift and crankshaft degree the exhaust port goes from supersonic flow to subsonic flow.

Below that lift when flow is supersonic, we can use the analytical tools for supersonic flow to guess what shapes work well in accelerating flow from a supersonic nozzle without the shock waves converging on each other and chocking the flow. Just google a hobbyist model rocket builder's site and they have the formulas.

Above that lift when flow is subsonic, we can use experimental tools like flow bench to develop and exhaust port that works well with subsonic flow while taking our supersonic flow considerations as a constraint.

Do you agree that we should be able to do that with our modest computational tools?

 

Tuomo,

Yes, I fully agree that you can use math and engineering tools to calculate most, if not all the variables about the engine. The hard part is knowing what tool and or model to use for what, and how to relate it all together given that an internal combustion engine is a dynamic model.. The rocket science I mentioned earlier was done pre CFD software, and is from the 60's and 70's

Yes, IMHO, and what I have learned from the scientific community, in the example you described about the exhaust valve and sub/supersonic, you can do all of this with engineering calculations.

I appreciate that your presentation of data and opinions are truthful and you clearly state your opinions.

Many on RL do not do this.


Cheers
Sent from my iPhone using Rennlist

 

While trying to figure out how to do even the basic calculations, I got bored and started surfing the web. I came across this thread on exhaust ports. I think that it is highly informative, if of nothing else just about the fact that hot rodders don't know much about what's going on inside the exhaust port or even how to modify exhaust ports! I enjoyed reading it, here it is: http://speedtalk.com/forum/viewtopic.php?f=15&t=2598

 

At least you surf the web, I usually just fall asleep with the reading materials on my head!

I find it even harder to figure out what happens in the dynamic vs static situation. That just leaves me mumbling to myself like a crazy person when I try to visualize it.

Sent from my iPhone using Rennlist

 

Here's some more web surfing results, in case you are interested in turbocharged engine exhaust port and runner size. Let's take Audi 2.7T engine. It has 450cc cylinders, and the 928 S4 thus has about 40% larger cylinders. However, each cylinder makes about exactly as much power in the two engines, with the six cylinder 2.7T being rated at 250hp.

The 928 S4 has 40 mm exhaust ports, giving 1257 mm^2 cross-sectional area. Guess what the exhaust runner diameters are for the 2.7T? 27 mm, giving a 573 mm^2 cross-sectional areas. The 928 S4 has 2.2x the cross-sectional area for the exhaust port / runner as Audi 2.7T putting out equivalent power! I hope this causes people to at least pause a bit before taking the grinder to the 928 S4 exhaust ports under the theory that forced induction means the exhaust ports must be made larger. "Sir, please step away from the grinder!"

If that didn't blow your mind, then consider this. When the 2.7T turbo exhaust manifold merges the three 27 mm diameter primary runners into a single secondary runner, guess how they sized the secondary runner that combines the flow from all three primaries? Drum roll... 27 mm! That is, the three cylinders each making as much power as the 928 S4 cylinder breath together in aggregate through a single 27 mm pipe.

An aftermarket cast exhaust manifold for that engine by Wagner continues to neck down the exhaust runners slightly from the ports:



These go on high powered cars with bigger, free flowing turbines -- but not massive turbine inlets!

On the exhaust side, the velocity created by small ports and header pipes is your friend if you want to empty the combustion chamber from exhaust gasses.

 

I noticed a couple things..

1. This is a 6 cylinder, so the exhaust pulses are every 60 degrees vs 45 degrees for a v8, and collectors can be different due to the pulse not colliding in the header junction

2. The velocity is important for a turbo to get going initially, but total heat energy in the exhaust charge also plays a very large factor, as the heat energy drives the turbines not necessarily the velocity per se..

3. I have not seen the port on the Audi, but if similar to the 928 port with sealing rings, then the header tube could be allowing the seal ring to seat properly cause less disturbance in the exhaust flow at the header to head face transition.

4. Unless it is on an unlimited class open wheel race car, the headers will always be compromised by packaging constraints in the engine compartment. (Even open wheel racers have packaging constraints)

I'm sure there are more things to spot if this is mulled over a bit more, but I think you get my initial observations.

However, I do agree that bigger ports are not always needed. The engine design should always be optimized based on intended use.

Just like big fat turbochargers, unless you have the heat energy in the exhaust to spin the turbines, you will have a pig-car that is unresponsive at best....

So, I repeat, design and build for intended use..

Just two cents.. Maybe even euro cents that will be worthless in a few months!

Cheers

Sent from my iPhone using Rennlist

 

Have you zeroed in your internet research into your e63? Or maybe do the 3 valve e55? Very little scientific discussion on the w211 boards - it's very non technical in fact.

 

The V6 has crankshaft 240 degree pulse separation per bank. The 1-3-7-2-6-5-4-8 has 90-180-270-180 degree separation. This is obviously relevant.

My view is that for a V8 such as ours, the secondary (or tertiary in 4-2-1) pipe that combines the 1-3 (and 5-6 on the other side) should be a little bit larger than the primary pipe and a little bit larger than the secondary pipe in a V6 bank. This is because the 90-degree pulses land largely on top of each other.

In contrast, I would make the argument that the primary pipes of the V8 should be even smaller than those in a V6. We've got one weapon to fight the reversal and blowdown interference, which is high velocity in the primary pipes. The main way of getting high velocity into those pipes is to keep the primary diameter small! (This is a theory, not a fact.)

All three drive it, heat, pressure, velocity.

Stay with the trend as the trend is your friend, until the bend right at the end... too late to short on news IMO.

Haven't really looked into it that much. What I do know about the engine in that car is that if you want to make more power with it, you need a new larger compressor wheel but you shouldn't increase the exhaust port size or the exhaust primary runner size. The turbine need not be increased either, at least when going up the compressor one size.

 

Tuomo, when Audi designed the 2.7T engine I strongly believe one of the top priorities was boost at low rpm and good turbo response. One of the tricks to achieve that goal is high exhaust gas velocity at low rpm. In order to get high exhaust gas velocity at low rpm the primary pipe diameter must be kept small. The trade off is less top end power. It is always a question of what kind of engine and what kind of power curve you like to have.
Åke

 

Point taken, of course you're right.

A counterpoint: On that particular engine, they can make a lot more power with those stock exhaust manifolds that have 27mm diameter primaries and also 27mm secondary.

For even higher power 2.7T motors, there are aftermarket exhaust manifolds. The best aftermarket manifolds, modeled and cast, also neck down the exhaust after the port to get the velocity up, as the picture in my earlier post shows.

In the 928 S4 world, another data point is Todd Tremel and his turbo car that runs the stock S4 exhaust manifolds with a flange for the turbo welded to the end. A hair shy of 1000 rwhp is going thru those slightly over 40mm ports and primaries very well.

I don't know about normally aspirated engines, but without any experience and just theory I'd say that high overlap engines should like small, fast, and straight exhaust ports and primaries. You'd know the answer that better, of course.

 

Tuomo, you and John combined have collected a tremendous amount of knowledge when it comes to turbo-charging. I have been mostly working on NA engines. I cannot disagree with you, common sense tells me high exhaust gas velocity is the way to go when turbo-charging. For a high performance, high revving NA engine running wild cams with much overlap, you need fairly big header pipes. The diameter of the header pipe very much determine at what rpm peak torque will occur. The length more or less shift the power curve around that point. The rule of thumb for a very high performance 427CI V8 engine having peak torque at about 6000 to 6500 rpm and putting out well above 700 HP, tells me the optimum header size is somewhere in the range 2 1/8" to 2 1/4" OD. Here is a test performed on a less (600 hp) powerful 427 CI engine. Best size is 2.0". http://www.hotrod.com/how-to/engine/...ers-dyno-test/
I must say I was surprised to hear Mike Simard only had 1 3/4" headers on his very powerful engine but I think I recall he once said he felt the headers were not the optimum design. For an all stock 928S4 I can see 1 3/4" headers work very well and for a modified engine putting out a hundred or so extra horsepower 1 7/8". As said before it is a matter of what kind of engine you like. For good low and mid range torque go for smaller headers.
Åke

 

I just did put the valve seat rings in for 42mm intake valves. Ring OD 44mm, ID 37,5mm, interference fit 0,15mm. The head was heated to 250C or 482F in order to expand sufficiently. This will be a test port in order to find out how high flow numbers can be found for an all out race engine.
Åke