Exhaust Science Demystified

The fact is most cars are leaving horsepower on the table. We show you how to get it back.

David Vizard May 16, 2005 0 Comment(s)

Muffler Flow -- How Much is Needed?
The first point to appreciate here is that optimally-sized collectors/secondary pipes are not sized so as to meet the engine's flow requirement, but more by the need to produce the desired pressure wave characteristics. For instance, a 700hp engine may have a dyno-optimized 3.75-inch diameter collector. This diameter, in conjunction with the length used, results in the system "tuning in" at the desired rpm. But from the standpoint of flow, a 3-inch pipe from each bank would be capable of handling all of such an engine's flow requirements.

Without data to the contrary, it seems safe to assume that the more a muffler flows, the better. This, fortunately, is not so and here's why. Increasing muffler flow unlocks potential engine power. Once all the potential power is unlocked, further increases in exhaust system flow will not produce any further benefits in terms of power. But what may be good for power may not be good for noise as any excess flow capability can lead to a noisier system. From this we can conclude that too much muffler flow serves no useful purpose and possibly costs more money than was really necessary. The trick here is to use just the right amount of muffler, no more and certainly no less. This allows the full power potential of the engine to be realized at the lowest cost without undue compromise in terms of noise. Now the question is, how much flow is enough?

Some years ago, in anticipation of the fact that eventually almost all race cars would need to be equipped with mufflers, I embarked on a series of tests to establish what a race engine's minimum flow threshold was. Initially, such tests looked easy but, to get meaningful results, it was necessary, as far as possible, to isolate the effects of flow from the effects of pressure wave tuning. This can be done with a pressure wave termination chamber more commonly known as a resonator box. Knowing when and how to use a resonator box can be a very important part of building a high-performance system and we will look at these shortly to see the role they play. For now, let us look at some flow-oriented test results.

In Fig. 7 you will see the results of tests run on a number of engines of various types. The only common element of significance between these engines was the use of a cam with 290 degrees or more of seat (advertised) duration. As you can see, the trend is that as flow is added to an initially flow-restricted engine, power increases rapidly at first then gains tail off. Once the available flow exceeds about 2.2 cfm per hp, the gains possible by increasing muffler capacity drop to less than 1 percent.

Knowing that 2.2 cfm per open-pipe hp means zero loss from backpressure allows us to determine how much muffler flow your engine needs. Just make a reasonable estimate of its open exhaust power potential and multiply by 2.2. For instance, a V-8 making 500 horsepower on open exhaust will require 500 x 2.2 = 1100 cfm. Two 550-cfm mufflers will get the job done and contain the backpressure-induced power loss to 5 horsepower or less. With mufflers rated in cfm, see how easy making an appropriate choice gets?

Pressure Waves
With muffler flow requirements out of the way we can move on to methods of applying suitable capacity mufflers to the "system" without needless disruption of length-induced pressure wave tuning. Probably the best way to ease into this somewhat complex subject is to consider some of the published muffler test results done in recent years. These tests appeared to have shown that, sometimes, lower flow mufflers inducing at least some backpressure were required to make best power. In all such tests that I have studied, the conclusions (as apposed to the tests) were invalid. There turns out to be several reasons for this and all are relevant to building a near zero-loss exhaust system.

The first point canceling the supposed validity of back-to-back test results is due to the varied internal designs seen amongst the test pieces Fig. 8. Many mufflers are made up of a number of inter-connected chambers having varying degrees of access ease by the exhaust. Others are of the "glass pack" variety. These types represent opposite ends of a spectrum and have a substantially differing response to arriving pressure waves.

When we dealt with collector length it was emphasized that it was, in most cases, more critical than the primary pipe lengths. Adding a muffler (even one with zero backpressure) to a system with already optimized lengths can alter the pressure wave response such that the tuning is now out of phase with what is required and as a result, power drops. The trick here is to install mufflers such that they don't alter the tuned lengths of the system. Let us assume that the test muffler is attached directly to the end of the collector. A pressure wave is reflected either at the end of the exhaust pipe or when a sizable increase in cross-sectional area occurs. Open chambered mufflers such as Flowmasters often appear to the pressure wave much the same as the end of the pipe. This means the pressure waves see no change in length and reflection occurs largely as it did prior to the fitment of the muffler.

A glass pack muffler can act significantly different. It does not appear as a pipe end but as a substantial increase in collector length. Result: a reduction of power even though there is no measurable backpressure involved. From this we can see that many comparative muffler tests were in fact "pseudo pipe-length" tests. Although many invalid conclusions were drawn, these tests still demonstrated some important facts. The most important is that the engine's needs in terms of flow and pressure wave length tuning must be isolated, one from the other. This is easy to do by means of the pressure wave termination box (resonator box) mentioned earlier. Incorporating a resonator box into a system produces a layout along the lines seen in Fig. 9. With enough volume, the resonator box makes everything down stream appear invisible to the header's primary- and secondary-tuned lengths. With a flow capability of 2.2 cfm or more, the muffler appears virtually invisible from a flow standpoint. As a result, we have a muffled system that produces virtually the same power as an open exhaust.

Cross Overs and Balance Pipes
The object of the entire muffler tech so far discussed is to end up with an acceptably quiet system; otherwise the point of the exercise is lost.

By using no more muffler flow than needed we are giving whatever mufflers are selected the best chance of doing the job. Unfortunately, mufflers can be a little inconsistent and unpredictable in terms of noise suppression from one engine type to another. Situations involving high compression ratios and long-period cams are usually more demanding in terms of noise reduction. Big cubic inches, shorter cams, and lower compression ratios are easier to muffle. The biggest problem in this area is knowing whether or not a possible combination is quiet enough. If you hit the Dynomax web sight you can hear chassis dyno tests of a wide variety of mufflers (including stock) on an extensive range of vehicles.

Be aware that how the system is installed can also affect the sound level, especially in the vehicle's interior. Do not have the tail pipe ending under the car, as the bodywork will act as a sound box in much the same way as a guitar body. Either have them go all the way to the rear, with down turned exit pipes angled slightly in towards each other, or have side exits aimed 45 degrees to the ground.

As far as power is concerned, tail pipe length after the mufflers has no measurable effect on the power if a large change in cross section is present up stream (toward the motor) of the tail pipe. An open-type muffler, or a resonator box, provides this cross-sectional change. The tail pipe length exiting most glass pack installations is also of little consequence if a resonator box is used, but is of significant influence if not.

Virtually all V-8 exhaust systems can be refined by the addition of a balance or X-pipe. These have two potential attributes: increased power and reduced noise. Extensive dyno testing on both of these factors has indicated balance and X-pipes are 100 percent successful at reducing noise. The reductions amount to a minimum of 1 dB to a maximum of 3 dB with 2 dB being common. As far as power is concerned, things are a little less certain. With engines between about 325 to 550 hp, experience indicates that in about 60 percent of the cases (mostly with balance pipes), the engine can deliver as much as 12 additional hp, with 5-8 being the most common. The other remaining 40 percent tested showed virtually no change in output either up or down. Based on such results, we can conclude that a balance or X-pipe is always a positive asset and never a negative.

Balance pipe sizing seems not to be overly critical. The only really influential dimension is the pipe diameter. This needs to have an area at least equal to that of a 2.25-inch diameter pipe (4 square inches) with 2.5 to 2.75 inches being preferable. Though limited to tests on engines up to a little fewer than 600 hp, there seems to be no measurable benefits to using a crossover pipe bigger than 2.75 inches in diameter. As for the crossover length, dyno results indicate that 18 inches responds in virtually the same manner as 72 inches long.

The Final System
Take a look at Fig. 10. This is a system I designed for a 700hp normally aspirated non-nitrous street/strip small-block Chevy that was installed in a 1986 Corvette. It produced acceptable street noise levels without any measurable drop in power. Although you may have to adopt some slightly different steps toward getting an acceptable installation, keeping sight of the principles involved will deliver similar results. Step outside the guidelines and you are on your own!


Fig 7 The line rising from left to right shows muffler flow versus the percent of maximum power retained compared with open-pipe power. Once the flow reaches 2.2 cfm per hp, the output seen is as per open pipe output. The line descending left to right shows the typical backpressure seen. At 2.2 cfm per hp, the backpressure should be down to as little as 0.2 psi (a little less than 0.5 of an inch of mercury).

Fig 8 Understanding the concept outlined here is vital to understanding how different styles of mufflers affect the apparent tuned length.

This is classic race Flowmaster. The open internal design allows the pressure waves to react as if they had pretty much reached the end of an open pipe. This means whatever pressure wave tuning existed before the muffler was attached, is largely unaffected.

Fig 9 Follow the system construction guidelines shown here and you will be pretty much assured of a zero-loss exhaust system.

This 4-inch Borla muffler topped out the UNCC flow bench. Readings were taken at 6 inches of depression and corrected to 20.3 inches (10.5 inches Mercury). The result: 1100 cfm!

Here's one of the "new generation" flow-bench developed three-pass Dynomax mufflers. Our flow bench and dyno tests show this unit delivers really good results for minimal money.

This is what Hooker's Aero Chamber muffler looks like inside. Our tests showed these well-made 2.5-inch units to be good for no loss on a 375hp engine while delivering a conservatively sporty, yet authoritative, exhaust note.

Although there are many performance enhancing "tweaks," most straight-throughs, such as great-looking stainless Magnaflows, are a variation on a common theme.

Many of the smaller-bore Flowmaster installations can be fine-tuned to make a little more torque everywhere in the rpm range by selecting a muffler with about an inch larger in/exit diameter. Bell-mouth adaptors (available from commercial truck supply stores) are then used as shown here. It is important to make the inlet as per this drawing, otherwise the low-speed gains will not be realized.

With high-flow cats and an X-pipe, this 5.0L Magnaflow system has all the ingredients for strong street performance and moderate noise levels.

Fig 10 Here is what a complete zero-loss header/muffler system looks like in finished form. A lot of work went into this but the results were worth the effort involved.


John Kaase: "I used a straight-through glass pack muffler design specifically because of the high-flow they can deliver. My dyno testing left no doubt as to how important collector length was and that a straight-through glass pack contributes to that length. By getting the collector/muffler length right, which in our case was about 40 inches, the torque at 3,500 was increased substantially. That gain is probably what won the Engine Masters deal for me the first time. I have seen an incorrect length along with less than the critical minimum flow cost 40 hp. Short change efforts on the collector/secondary and it will short change you."
Joe Sherman: "If you are building a serious performance system, then assuming you have a near-optimal header set-up, the place that is most critical when it comes to avoiding power loss is from the collector back. Also, don't be fooled into thinking that big tailpipes contribute to power. In all my years of dyno testing, I never have seen that work. For me, the straight-through Magnaflows when used as part of the collector length, show only very small losses in power over an open pipe. It's all about the right length and sufficient flow. I have seen mistakes in this area cost 85 horsepower."

0505phr_exh_25_z 13/25


If, to stay legal your exhaust system must run catalytic converters, then the possibility of loosing power goes up dramatically, but it certainly does not mean the game is lost. The first rule of thumb here is if the cats must be in the original position, use the highest-flow components that can be physically installed. For high-flow, high-performance cats, one of the first places I would try would be Random Technologies. Some of this company's key employees drag race late model-street legal machines and are serious about performance. Also in the business of marketing genuine hi-flow cats and cat systems are Walker (Dynomax), Magnaflow, Dynatech and, for a number of specialized truck installations, Gale Banks. These are not the only ones, but they are all the companies of which I have experienced the no-nonsense functionality of their products.If the position of the cats can be moved to such an extent that the length going into the cats represent the secondary tuned length, then we find that to an extent, the cat, if large enough, can, in part, act as a resonator box. Moving the cats to a more favorable position then is rule number 2 when cats must be used.Rule number 3 is that if there is room to put a crossover or an X-pipe before the cats, then that's almost always the best place. Anything after the cats will drop the sound level but is unlikely to increase power unless the flow of the mufflers you chose was significantly short of what was needed.

0505phr_exh_23_z 14/25

High-Tech Collectors
Technology to make a good header has been around for 30-plus years. These days, making a top-notch header is very much a question of refinements to eek out whatever potential may be remaining. One area of research that has paid dividends in the past decade is in the collector design. Example number 1 on a system built by Kook's Headers is a 4-into-1 merge collector (arrowed). Dyno testing this type of collector, versus a regular parallel one, shows that the merge collector tends to pull up torque from the lower speed range with ever decreasing amounts, thus delivering a fatter torque curve but not necessarily any more peak hp.

0505phr_exh_24_z 15/25

Another header/collector worthy of note (our example is again from Kook's) is the type shown in photo number 2. This is much favored by Busch and Nextel Cup engine builders. Essentially it is a long 4-into-a-short-2-into-1 system. The parts that go to make up the system between points A and B are shown in the top right hand corner of number 2. About 10 years ago, Flowmaster introduced a collector that converted a regular 4-into-1 system into the system seen here. This was my introduction to testing this configuration of collector. The dyno indicated only marginal gains in peak power. Like the merge collector, this collector style fattened up the torque curve, but usually to a greater extent.


Flowmaster Inc.
Santa Rosa, CA 95401
Gale Banks Engineering
Azusa, CA 91702
MagnaFlow Performance Exhaust
Rancho Santa Margarita, CA 92688
Hedman Hedders
Whittier, CA 90602
Schoenfeld Headers
Borla Performance Industries
Oxnard, CA 93033
Stahl Headers
Kook's Custom Headers
North Bayshore, NY 11706


Connect With Us

Get Latest News and Articles. Newsletter Sign Up

subscribe to the magazine

get digital get print