It ain’t just pipes, it’s science!
I’d like to try to explain some basic exhaust theory and clear up some issues that may not be completely clear.
Everyone knows the purpose of an exhaust system is to provide a means for the exhaust gases to be removed from the cylinder. You might wonder why have an exhaust system at all. Other than the frequent need to muffle the noise of rapid combustion, why not simply open the exhaust port to the atmosphere, thereby saving both weight and expense?
Some time back in early internal combustion engine history, it was discovered that attaching a length of pipe to the exhaust port (probably to direct the noxious exhaust fumes away from a passenger compartment or out of a room where a stationary engine was housed) often had an effect on the performance of that engine. Depending on parameters such as pipe diameter and length, the performance could be adversely or positively impacted.
I expect it was clear from the very beginning that exhaust gases have momentum. What may not have been known at the outset is that they also exhibit wave properties, specifically those of sound. Both those properties can be utilized to evacuate the exhaust gases more quickly and completely. The usual term for this removal process is “scavenging.”
There are two types of scavenging: inertial and wave. Inertial
scavenging works like an aspirator whereby some of the kinetic energy of a
moving fluid stream (air, water, etc, generally in a pipe) is transferred
to the fluid in an adjacent pipe. You may remember from high school
chemistry lab class where you used water traveling through the top of a
“T” fixture to draw a quite powerful vacuum in an attached vessel.
The “T” can be likened to a merge collector as used in virtually all successful racing cars (although often not in dragsters). The most effective merge collectors minimize the volume increase at the juncture of the pipes. If this volume is too large, gas speed is diminished and less kinetic energy is transferred to the gases in an adjacent pipe. Thus, the scavenging is less complete. Well, so what if there is a little gas left in the pipes? Consider the engine cylinder as an extension of the exhaust pipe. A cylinder with residual exhaust gases has less room available to accommodate the incoming charge of gas and oxygen. Obviously, the more gas and air you can get into a cylinder, the more power is developed; that is why superchargers are so effective.
Not only can scavenging be utilized to empty the cylinders, it also can help to draw in the new charge, by producing a negative pressure in the cylinder. This gets tricky because there has to be adequate time in which both the intake and exhaust valves are open, and there is the potential problem of the new charge passing right through the cylinder into the exhaust pipe! Gas is wasted and power is lost. Maybe you can design your cam such that it closes at just the right time to prevent this from occurring. Or maybe you can make the exhaust pipes just the right length so that the reflected sound waves (at a particular engine speed) prevent the incoming fuel and air from spilling out of the cylinder. More on this later.
A stock S4 engine has very little valve overlap (some at small valve openings) and therefore there is only a short time during which scavenging of the cylinder can be accomplished. Even still, there is opportunity for significant performance gains with effective scavenging of the primary exhaust pipes (the first pipes that emanate from the ports) where it’s possible to produce a negative pressure so that when the exhaust valve opens, exit speed is increased. The result is increased momentum and possibly improved cylinder evacuation.
On to wave scavenging. An analogy would be tuned organ pipes in which
their length is adjusted such that a standing wave of a particular desired
length (and frequency) is established. This means that some whole number
of waves will fit exactly within the length of the particular pipe. When the point of maximum amplitude of a wave comes to the end of the pipe or a change in diameter, the wave is reflected back up the pipe, but as its mirror image. Thus a positive pressure wave is reflected as a negative pressure, or rarefaction, wave which, in turn, helps to draw spent gases from the pipe/cylinder. Wave scavenging is most effective over a narrow speed range that can be adjusted by changing the primary pipe length.
Thus a torque or power peak can be designed to occur at a particular engine speed to suit the application whether it is racing or everyday driving.
What are crossover headers? There are numerous types of headers, tri-Y, equal length, stepped, unequal length, crossover, etc. Unequal length headers are by definition not tuned at a specific rpm; rather each pipe is tuned for a different speed. They tend to perform better than the stock manifold and may increase performance over a broad speed range. Because of their unequal length, each pipe will utilize wave scavenging at a different speed, thus reducing the effect at any single or narrow band of speeds. They often have sub-optimal merge collectors and so, do not make the best use of inertial scavenging. Equal length headers can be excellent wave scavengers, but often have inferior collectors, so inertial scavenging is not optimized. The tri-Y design is especially good on 4-cylinder engines and is now being used almost exclusively on NASCAR engines with 8 cylinders. Stepped headers gradually increase the pipe diameter going away from the port. I believe at least one of the purposes is to inexpensively approximate a megaphone which is the most efficient device for returning the pressurized gases back to the surrounding atmosphere.
The crankshaft in the 928 is a 90-degree dual-plane design that dictates a
particular firing order (1-3-7-2-6-5-4-8 where #1 is at the right front
and #5 at the left front). I think that’s right—my engine is usually
upside down when I’m looking at it! A complete cycle on a four-stroke
engine requires two revolutions, 720 degrees. Divide that by 8 cylinders
and you get 90 degrees. An exhaust valve opens on a particular cylinder
every 90 degrees. Let’s look at the arrangement of exhaust pulses in our
engines. Cylinder #1 fires, 90 deg later #3 fires, and then another 90
deg and #7 fires. Note that cylinder 7 is on the opposite bank from 1 and
3. After another 90 deg, #2 fires, another 90 deg for #5 and another 90
deg for #4.
Boring, I agree, but let’s look at what this means for the arrangement of pulses on a single bank of cylinders: 90 degrees between 1 & 3, 180 degrees between 3 & 2, and 270 between 2 & 4. Graphically it looks like this: 1---3----2------4 90 degrees to get to #8 and for 5 - 8 where we have 270, 180, 90 or 8------7----5--6. Obviously the pulses are not evenly spaced and this is what accounts for that V-8 sound we all love. When it comes to effective scavenging and ultimate performance, this unequal spacing has dire consequences. Virtually all headers for V-8 engines have 4 primary tubes attached to one merge collector for each cylinder bank.
The scavenging between cylinders 1 & 3 and 5 & 6 is about as good as it gets since the exhaust pulses are spaced 90 degrees apart. The pulses for the other cylinders, however are all 180 degrees apart! This means that kinetic energy is imparted to the other cylinders only half as often as 1 & 3. Ok, this isn’t exactly true since there are harmonics involved and the second and other harmonics may provide some lesser degree of scavenging. In any case, inertial scavenging is substantially reduced.
The crossover group (to which my Scavenger system belongs) routes the pipes from cylinders 2 & 3 over to the other side to connect to the collector for 5 & 8. Similarly for 6 & 7 and the collector for 1 & 4. This spaces the pulses evenly at 90-degree intervals and optimizes the inertial scavenging effect. One of the results is a V-8 that sounds smooth and melodic like a V-10 or 12. It is simply not possible to do better than this for an engine of the 928 design. And this is not coming from just me, but also from prestigious organizations such as Burns Stainless, with whom I collaborate. There is simply no better design for engines like ours than the crossover system. It’s an engineering fact, not idle conjecture. Just as an interesting aside, Burns Stainless assisted Kelly Moss Racing with the design and construction of the tri-Y system for the 928/944 car.
Given the same engine and assuming that appropriate crossover primaries can be fitted, the crossover system will usually produce more power than other designs. Now there are cases such as with stock engines with little valve overlap (such as the 928) where the difference may be small, but as the valve opening is increased and the duration lengthened, the superiority of the crossover design will be clearly evident. So why doesn’t everyone use crossovers if they’re so great? They simply cannot be implemented on many engines and even if it is possible, usually not without a great deal of effort and expense. Most V-8s have the transmission located at the rear end of the engine. This means that in order to provide adequate ground clearance, the primary pipes would have to be so long that they wouldn’t provide any scavenging except maybe at idle!
I first became aware of the use of crossovers in the ‘60s (I think) when
the Ford GT40s had them. It was a devastating design. Other teams tried
to implement the same design, but because of the ground clearance problem,
had to resort to crazy and expensive schemes like turning the heads around
so the exhaust ports were on top! Ultimately, crossover headers were
deemed an unfair advantage and so expensive as to limit competition and
they were declared illegal in many forms of racing. It just happens that
the placement of the 928 transaxle at the rear creates a perfect location
for two merge collectors and crossing over of primary pipes.
Well why does the Scavenger cost so darned much? The fact is that the
materials in a complete stainless steel system cost about $3000. It is
very difficult to construct. I may spend an entire day just fitting a
single pipe! There are no voids in the pipe joints filled with ugly welds
as is often the case with less careful constructors. The merge collectors
are provided by Burns Stainless, and are some of the very best available
anywhere. They can be found on some of the most successful race engines
in nearly all forms of racing. I spent the weekend at Sears Point where I
talked with a fellow from S-Car-Go Racing who was carrying a set of
stainless 911 racing headers—6 pieces of slightly curved pipe and 2
megaphones. “How much,” I asked. “6500 bucks,” he replied and also said they cost about $17,000 from the Porsche factory. I’m thinking 5 grand for a complete state of the art stainless exhaust system is cheap. By the way, Burns would charge about $5k for just the headers alone.
I hope this was at least somewhat informative and not too boring.
Some additional interesting reading: http://www.bolly.com.au/book/Book.asp?Chapter=6&Section=2