The EPG (exhaust pressure graph) studies have elicited a useful dialogue between the CAFE Foundation and many Sport Aviation readers with experience in exhaust design. This article will present their views.

6 CYLINDERS From his engineering background and his work on the tuned exhaust system developed for the Questair Venture, Jim Griswold submitted the following comments about exhaust pipe design: "The gross improvement obtainable through the exhaust system is largely a function of how bad the system was to begin with. A more informative portrayal of the potential improvement is to reference from a neutral exhaust system, i.e., short (approx. 8") individual stubs. The potential improvement from a neutral system is 7%. "The measure of a well set-up engine is for all the EGT's to peak simultaneously during a lean-out. This results in more power, less fuel consumption and minimizes vibration. It occurs when the induction air, fuel and exhaust behavior are uniform among all cylinders. An engine without these air/fuel qualities may

"The experiments that I conducted on optimizing exhaust thrust used a normally aspirated engine and resulted in unacceptable BHP/ BSFC's before any thrust effect was detectable. The boosted engine offers the best opportunity for exhaust thrust, i.e., low drag at high altitude with high mass flow through the engine. However, turbocharged engines are generally more sensitive to exit pressure than normally aspirated engines. It is possible that the technique commonly used to accelerate the exhaust is faulty. A lip, insert or converging cone at the exit is typical. This causes an abrupt acceleration of the gas and sends positive pressure waves upstream. Both are inefficient. A better approach might be to use the minimum possible collector diameter, thus minimizing the deceleration upon entering the collector and eliminating the acceleration exiting the collector. The limitation on collector size is avoidance of the riser and collector behaving as a single tube longer than critical length. An exhaust tube at or above critical length causes reverse flow at mid-scavenge stroke, reheating the exhaust valve/cylinder head and interfering with the critical final scavenging. (As a matter of interest, the Venture has a very favorable drag/engine mass flow ratio and it requires an

skew the results of some of the exhaust experi- exhaust velocity increase of 71 fps with no loss ments. The goals of exhaust system design are: of engine power to gain 1 kt. A 10% reduction

1) Achieve near atmospheric cylinder pressure at the beginning of the scavenge stroke, 2) No reverse flow during the scavenge stroke, 3) Maximum negative pressure effective at the valve port while passing through TDC ending the scavenge stroke. "System back pressure, except at TDC, is of little consequence, making the average back pressure meaningless. The maximum torque contribution of maximum negative pressure effective through the entire scavenge stroke is 0.1%. Realistically, 0.01% contribution may be achievable. 80

in exit area is necessary to achieve this, and that materially affected engine performance.) "Pressure waves have no mass, therefore they recognize no obstruction. Anti-reversion is an oxymoron. Fancy collector clusters, pyramid points, etc. are insignificant.

"Departing exhaust gas is subjected to linear acceleration exceeding 30,000 G's. Turning a 100 G corner is insignificant. "Many experiments have been conducted using megaphones. The two most common observations are a lack of effectiveness from the

initial portion of the cone, and an early collapse of the wave expansion. Both are from the same fault. Traversing from a tube to a cone leaves a sharp break that either requires infinite expansion acceleration or separation. The separation makes the initial cone invisible and uses considerable energy that leads to early collapse of the wave. The solution is to make the megaphone by forming a body of revolution using a second degree curve. This provides a flow path that approximates a uniform expansion acceleration rate. One needs only to study a musical instrument to realize the effectiveness of this approach. The part can be easily fabricated by spin forming a welded cone on a mandrel of the desired shape. "Your early work on tuned exhaust demonstrated a significant and sudden loss of power when the engine was operated at lower rpm. Operating above the tuned rpm resulted in a gradual but progressive decay from optimum. This points to the ineffectiveness of early negative pulses and the virtue of having negative pressure present during exhaust valve closing. Tune the system to the bottom of the cruise rpm and accept a small penalty at takeoff rpm, or use a megaphone adequate to cover the range of rpm. "When one compares the timing of the system to the local speed of sound, there is an apparent conflict. When the exhaust system of an airplane engine is tuned, the average speed of the pressure wave is approximately 1200 fps while the speed of sound at the exhaust system temperature is approximately 2200 fps. The pressure wave is formed by the initial exhaust valve opening and travels down an essentially quiescent exhaust system to the exit, where a negative echo wave is formed. Immediately following the initial wave is the rapidly accelerating exhaust bolus. This bolus has an average speed of approximately 1700 fps. This results in the echo wave of having a speed relative to the exhaust tube of only 500 fps until the bolus clears the exhaust system and the system goes quiescent again. Aside from resolving the mystery of the "slow" wave speed, it provides an insight with which to anticipate the effect of tube diameter changes and tube dia./cylinder displacemnt ratio requirements. "The EGT responds noticeably

when the tuning is correct. Knowing this can sometimes save time running tests. My recollection is that EGT dips down (= 25°F) when tuned on the spot. "I estimated the boundary (layer) to be between 1/16" and 1/8". When taken into account, it shows the effective area diminishes faster than the physical area as tube size is reduced. "Pressure mufflers, regardless of the cleverness of the innards, seem to function at direct relation to the back pressure. This is always a losing proposition. On the other hand, absorption mufflers are heavy, long and have no durability. Given the choice, I would rather deal with the packaging problem of the absorption muffler than the assured poor performance of the pressure type, providing the durability can be addressed. The durability problem of absorption mufflers is attributed to sonic destruction of the glass fiber packing. The continual quest for efficiency by the American auto manufacturers has brought two new products to the market that solve the problem. Stainless steel wool fabric capable of withstanding 2000° F is used adjacent to the gas stream and attenuates sonic intensity. Basalt wool, spun molten lava rock, is much tougher than fiberglass and absorbs better, and is used behind the steel wool. These products can be purchased from American Metal Fibers, 2889 N. Nagel Court, Lake Bluff, IL 60044,312/295-1200. A WORD OF CAUTION — Fuel injected engines sometimes pour raw fuel into the exhaust when windmilled above idle rpm with the throttle at idle. If this collects in an absorption muffler, the explosion that occurs when the throttle is advanced will gut the muffler. "The Continental IO-550G (Venture engine) was certified at 280 hp using 2500 rpm. The engine had tuned induction (top of engine) and neutral exhaust (8" stubs out the bottom of the cylinder). The calibrated test engine delivered 284 hp and .375 bsfc. This

same engine was fitted with the tuned/ muffled exhaust and delivered 302 hp, .365 bsfc, and peaked all EGT's within 1% fuel flow during a lean-out. The system routed the three cylinders on each side to a common collector/muffler. The risers were 1.5"x.035"x34". The collector/muffler was 2.25" x .035x30.4". This 64.5" system was optimum at 2500 rpm."

WAVE GUIDES/MEGS Mike Palmer writes: "Wonder if you could include baro pressure on titles of your performance graphs. I mean, if it's 29.5" one day, and 31.0" another, but you don't tell us, we might conclude that an extra 1" MP engine gain was from the exhaust changes and not the wx. "I did my Masters work at OSU, and one of the things a fellow RA was working on was the effect of microwave horns as efficient antennas. The thought was that a 'smooth' transition (whatever that is) from the tiny confines of the waveguide to the outside, resulted in fewer reflections back to the transmitter. Wonder if there's any duality to that and engine exhaust theory? "Any plans to try to measure back pressure in flight, where the effects of the airstream may change things? (Yes!! Jim Griswold predicts airstream effects will be insignificant.) "Along the same lines, any thoughts on testing augmenters? As an engineer myself (although electrical) I couldn't pass up trying to recover some of that wasted exhaust energy. We modified the exhaust on our Glasair to pump into two augmenter tubes (two tubes about 18" long, that megaphone backwards to the way everyone thinks they should). Haven't got a clue if they're working — I've measured the air pressure in the lower cowl, and it's 100 ft. less than static, with good suction — but seems insensitive to power settings. (Although it's hard to tell.)"

COANDA EFFECT Bob Lockhart writes: ". . . to plant a seed; stuffing all those pipes under a cowling will be difficult, as any manufacturer will attest, and your pictures suggest. In our VW engines which were tuned at 6000 rpm, a length of 60 inches in front of the collector produced the most

power. The airplane application at half that rpm would need considerably longer pipes than that if you make them as long as required, where will you put them? "Secondly, racing cars have noise restrictions, and must run mufflers. Some of the mufflers have produced lower noise and higher power. Super 81

Trapp comes to mind. People at Sears Point could tell you what works, and give you their thoughts. "Thirdly, I made a system with a 4 in 1 Cyclone collector that had a sharp (3 inch radius) 90° bend at the collection point. It had two pipe inlets stacked on a side with the collector in the middle — an 'H' with a circle in the middle. It performed as well on the dyno as the best of the 4 in 1 systems and was very compact. "Fourthly, an article in a really old Hot Rod or maybe Motor Trend told of a megaphone Henri Coanda made in the twenties. It said that the megaphone produced 20% more power on the buses of Montreal. It had a cone of individual pipe diameter tapering to twice the original diameter at a ratio of— I think — 18 to 1. Inside this cone was a second cone which tapered from a point to the original diameter. The inside cone was capped with a hemi- sphere. So the flow cross sectional area was a constant throughout the megaphone, but the low pressure area behind the ball sucked exhaust out. I've always wanted to try it! If you do, please let me know. "Lastly, harvesting the original seed: couldn't something akin to the Lotus noise-canceling technology be used to generate a negative pressure pulse at the exhaust valve? And also cancel exhaust noise at the pipes' end? Wouldn't an electro/mechanical system make it possible to enhance power and minimize fuel needs? It seems you're enjoying your experiments a lot — I certainly am." Ed Vetter replies: "I had a similar thought with regard to noise-cancelling mufflers. My background is in Tele- communications where echo cancellation is standard technology. The exhaust system creates some

James Lenahan, EAA 327269, writes: ". . . .1 did at one time gain tuning on an opposed engine using short straight head piping by placing a crossover between two opposed pairs of pipes at 4 inches from the pipe ends. The critical thing here was to make the crossover joints at 90 degrees to the exhaust pipes. I was able to achieve resonant length from one exhaust valve down the pipe and around back up to the back of the other exhaust valve, achieving what functioned like exhaust pipe length without actually having it (so to speak!). "The analysis of resonance in exhaust systems is a combination of the gas flow dynamics and acoustics. This can be a complex situation where dual engineering disciplines are not imbedded in a same individual. To make matters more complex but somewhat simpler to analyze, a person also practicing electrical engineering in addition to gas dynamics and acoustics can set up analogies of the exhaust using electronics circuits. I did this in designing my mufflers. I started with an electronics power supply filter and looked for features that could react like inductance and capacitance within the muffler's gas flow system. "It was mentioned that a megaphone on the exhaust helped. To better understand this we need to look to acoustics. A continuous diameter bore exiting to the atmosphere with pressure waves (notes) presents an acoustic impedance mismatch and therefore a resistance to the exiting of the notes. An exponentially curved bellmouth (bugle or other wind instrument end) provides an excellent impedance match when properly designed. To find the 'perfect' length on a race car, 1 used to

Paul Schalk writes: "I'm glad to see some work being done on the exhaust systems of airplanes. The short P-51/P-38 stacks may look and sound impressive but there is a lot of power given away in doing so. "In making a brief review of some of our files, the following items may be of some help to you: SAE Papers # 930621 and 930625. Ed Henneman-Headers by Ed 2710 16th Ave. South Minneapolis, MN. 55407 612/729-2802. "His catalog is a tech manual; this guy knows his stuff. Jim McFarland-Auto Com Austin, TX 512/833-8813. "Jim is a personal friend and has been around the automotive industry since the '60s. Borla Performance Oxnard, CA 805/988-0968. "They have some very interesting Hi-performance (racing) mufflers. Smokey Yunick Daytona Beach, FL. "Keep the firing order in mind when designing an exhaust system and remember that a single outlet sys-

does one find a speaker that can put out 90 Hz at 160 dB while standing up to the corrosive environment and high temperatures of the exhaust gases. It may be possible to create a noise cancelling muffler that could provide either minimum noise or optimum pressure during valve overlap. During takeoff you would use the muffler mode to minimize noise and in cruise you would use the performance mode to get maximum power."

output then heat the exhaust pipe (to almost white hot) with a torch. Violent pressure waves from a high power large cubic inch engine exhausting to a tuned system would cause the red hot exhaust pipe to deform and expand at the peak pressure wave point. It was here (at the peak of the bulge) we would cut the pipe off and bellmouth the pipe. "To further enhance the impedance match to the atmosphere of an exhaust gas system, one ought to consider mod-

outlet one, expecially on a low speed engine." Robert Addoms writes: "It appears that you have already invested a substantial amount of effort and money in test facilities and instrumentation. You might find it helpful if you had some computerized engine simulation capability to complement your testing facility. With such software very informative numerical experiments can be run at low cost. . . I have used several of these engine

new problems for us, such as, where

82

CROSSOVER SYSTEMS

place the engine at maximum power

ifying it to gain the advantage of using

the Coanda effect which induces vortices to lower the exit pressure more

without destroying the potential of the tuned system. I emphasize tuned because the third harmonic is often possible to play with where the piping would be too long at the fundamental frequency. Another thing to consider is reducing the 'Q' of the tuned system to enough of an extent to have tuning useful over a practical rpm range (i.e.: 2300 to 2650 rpm)."

OTHER EXPERTS

tem will generally out-perform a dual

simulation packages. I have written enhancements to them. "If you are primarily interested in studying unsteady flow in exhaust manifolds and its effect on engine performance, one of the best simulation packages available is the University of Manchester Institute of Science and Technology (UMIST) simulator. The last time 1 checked, the retail price for the package was about $30,000. There are cheaper packages available, but not all of them are adequate to study wave action in exhaust manifolds. Whether UMIST would be willing to give you a special price, or would be willing to exercise their program in your behalf, is something 1 don't know."

AN UNSOUND THEORY The major pressure energy in the

exhaust is in the main pulse leaping out of each valve 22.5 times per second at 2700 rpm. That gives a fre-

quency of 22.5 Hz seen as the EPG's primary or P wave. When all 4 cylinders are ganged with equal length headers into the collector pipe, the collector shows regularly repeating waves of 4 x 22.5 Hz = 90 Hz. Aside from these two frequencies, our pressure transducer, rated with a typical minimum response time of 0.3 msec, reveals no other significant 'sounds' in the exhaust. Furthermore, significant 'harmonics' are not identified after Fourier analysis of the collector's EPG waveform. Are aircraft engines, because of their low rpm , peculiarly inharmonious or 'non-musical'?? As a musical instrument, an aircraft exhaust system probably does emit a chaotic mix of overtones of varying higher frequencies dependent upon cam ramp shape, valve lip and valve seat shape, compression ratio, mixture, rpm, etc. The chance that any of these higher frequencies are of sufficient amplitude to be useful forces in scavenging would seem very remote, as shown by the following computations: If the overlap stroke in the Lycoming IO-360 engine is effectively about 30° of crank rotation then it occupies only 1/12th of the time required for a full crank rotation. At 2700 rpm, a full crank rotation takes 22.22 milliseconds, so the overlap stroke duration is 22.22/12 = 1.85 msec. For a negative sound wave to have a long

enough wavelength to remain negative for the entire 1.85 milliseconds, it must have a full wave duration of at least 3.7 milliseconds. This would equate to a frequency of 1000/3.7 = 270 Hz or lower for such a sound wave. The third harmonic of the collector's 90 H/ wave would have a frequency of 270 Hz. Higher frequencies, even if perfectly resonant, could not scavenge the entire duration of the overlap stroke. Since sound waves in exhaust pipes are thought to travel at about 2000 ft/sec, the wavelength of the 270 Hz sound would be about 2000/270 = 7.4 feet. Half of that wavelength is positive and the other half is negative, and the negative pressure part would occupy a pipe length of 7.4/2 = 3.7 feet or 44.4 inches. Frequencies lower than 270 Hz would give even longer segments of low pressure in the pipe. For example, 22.5 Hz would give a negative half wavelength of 533 inches! This may explain why the small changes in header lengths we tested by EPG showed little influence on the backpressure. Is the real exhaust tuning story simply told by the negative portion of the EPG's primary wave or are there 'hidden' sonic influences lurking undetected? Is the idea of sonic tuning unsound? Stayed tuned for EPG IV!!" Brien Seeley

Bibliography 1. Jim Griswold. Work phone 816/ 426-6941

2. Mike Palmer, email: MPPalmer @aol.com. 3. Bob Lockhart, 370 Altair Way

#188, Sunnyvale, CA 94086, 408/733-

3339.

4. James Lenahan, Custom Railway Supply, 118 N. 7th Street, Colorado Springs, CO 80905, 719/471-0110. 5. Paul Schalk, personal communication. Ricardo North America. Engineer experienced in OEM automotive exhaust design by flow bench. Phone

313/397-6666.

••

.

6. Robert Addoms, 2719 Manhattan

Beach Blvd., Gardena, CA 902494531,310/329-8104. 83