The selection of an automotive engine as opposed to a certified aircraft engine left me on my own when it came to the exhaust system. I read a lot of articles about exhaust systems but it seemed that it was much of a black art. I assume that the auto racing guys comprehend the subject but nothing was out there in the literature. What I did glean from the literature was that the goal was to create a negative pressure wave at the face of the exhaust valve for as long as possible while the valve was open. This negative pressure wave (deemed suction wave by the exhaust specialists) assists in drawing the exhaust gasses out of the cylinder and results in depressurizing of the combustion cylinder. The reduction in cylinder pressure enhances the pressure difference between the manifold pressure within the induction system, and the cylinder thus resulting in an improved volumetric efficiency. i.e. more torque.

How does one arrange for this negative pressure wave to arrive at the valve at the correct time?

The positive pressure wave is exhausted from the cylinder when the exhaust valve is opened. This pressure wave moves down the pipe with a velocity of 1200 ft per second. When the pressure wave reaches the end of the pipe it experiences an impedance change in a similar manner to r.f. signals at the end of an incorrectly terminated transmission line. This change in acoustic impedance results in a reflection of the wave back down the exhaust pipe. The significance of this impedance cannot be over emphasized. The new impedance encountered as the pressure wave reaches the end of the pipe is lower than the impedance of the pipe itself thus there is a phase reversal. This phase reversal changes a positive pressure wave into a negative pressure wave (suction wave), and it is this negative pressure wave that is transmitted back to the exhaust valve. If the timing can be arranged such that the negative pressure wave is present when the exhaust valve is open then the pressure difference between the cylinder and the pipe is increased which results in an improved transfer of the exhaust gases and more power is achieved.

If we start using simple numbers: An engine spinning at 6000 rpm emits exhaust pulses at a rate of 50 pulses per second from each pipe. For the pressure wave to be transmitted out to the end of the pipe and be reflected back to the exhaust valve ready for the next opening the pulse must have accomplished a journey of 2 x (pipe length) within 20 milli seconds. At 1200 ft per second, 20 milli seconds covers a round trip distance of 24 ft which requires a pipe length of 12 ft. Clearly 12 ft exhaust pipes are not acceptable on aircraft in a pusher configuration. Reducing the engine speed to 2700 rpm ( certified engines) results in a pipe length requirement of 26 feet to benefit fully from a tuned exhaust length.

There is however, a trick that can be played to fool the engine into thinking that the pipe is really longer than it really is. Consider all of the cylinders connected through long pipes into a common collector the through a single pipe to the outside world. In this instance the pressure wave is launched from the active exhaust valve, along the pipe where it hits the collector. At the collector it experiences the same low impedance that the single pipe system experienced and a negative pressure wave is reflected. This reflected pressure wave is reflected down all of the exhaust header pipes. Now since there are N cylinders operating, the next exhaust valve is opened in a period N times less than the single pipe condition. Going back to the 6000 rpm engine and using six cylinders, the next exhaust valve is open 3.333 milliseconds after the predecessor instead of the 20 milliseconds required for the single pipe condition. Thus the pipe length only need only be 2 feet to accomplish the same effect. Essentially one exhaust pulse is used to augment the extraction of the next exhaust pulse. Since the process is cyclic there is perfect balance.

Now in the real world you never get something for nothing, so what is being lost. When the reflection of the exhaust pulse occurs at the collector the pressure wave is diluted by N (the number of pipes in the collector. In my example N is six so the benefit of the reflected wave is reduced by six but along with this comes an increase in bandwidth so this effect is no restricted to one specific engine speed moreover a band of engine speeds.

The Subaru EG33 engine uses six cylinders and achieves maximum power at 5400 rpm thus the optimum performance is obtained with a tuned pipe length of  26 inches. I have made my pipe lengths 28 inches to optimize the power at a slightly lower engine speed of 5150 rpm but with the wide band experienced by the collection of six pipes the power should be spread across the band necessary to achieve 5400 rpm.

In my design I was unable to collect all six pipes into a single collector so I fabricated two three into one collectors and then coupled the two collectors with a balance pipe. This resulted in a close to optimal condition but more power could be extracted if I had created a six into one configuration.

Using four cylinder engines it is important that either all four cylinders are coupled together or the opposing cylinders are coupled together so that evenly spaced exhaust pulses are entering a collector. Siamese exhaust systems will rob the power and it is necessary to couple both pipes into one to obtain the best condition.

With the Lycoming 0-360 connecting cylinders 1&3 and 2&4 together is a definite NO-NO. The optimal configuration would be to connect all cylinders together. Nat puffer recommends four separate pipes but this is in contrast to the 1&3 and 2&4 condition given above. More performance would be obtained by connection all cylinders together and with the longest distance between the exhaust valves and a collector.

Fabricating the Exhaust
 

Fabrication of the exhaust was relatively easy. Having bought a Henrob oxyacetylene welding torch and a number of 1.75" diameter bent stainless steel tubes from "Aircraft Spruce Specialty Company" and some straight 2" stainless steel tubes, the welding process was relatively straightforward. A stainless flange was made to match the two cylinder heads of the EG33 engine then the various tubes were welded together. Welding used a slightly carburizing flame and was very easy to achieve with only a little practice. The Henrob torch is the key to the simple fabrication using stainless steel tubes. The slip couplings are barley visible on the lower of the two photographs. The couplings are on the diagonal section of the pipes rather than as the pipes exit the cylinder heads. Time will tell whether this is an acceptable means for incorporating the slip couplings.
 

The collectors on this configuration was thought to be too aggressive in taper so a second collector was fabricated this may be seen in the third and last photograph. The balance pipe may be seen connecting the two collectors under the PSRU. Just visible, and attached to the balance pipe, are the two O2 sensors necessary to maintain the high fuel efficiency of the automotive engine. The O2 sensor is located in the balance pipe to reduce the operational temperature and to prevent the atmosphere, which will be briefly sucked back into the exhaust system, from corrupting the O2 readings.

The image also shows a 1/2" pipe (vertical and behind the collector) which is incorporated into the collector. This pipe is configured to provide vacuum for the gyro instruments. There is one on each collector. The second venturi/vacuum source is used to depressurize the crank-case thus preventing piston ring blow-by from resulting in oil leakage.

Last Updated:    Thursday August 31, 2006