Making a muffler

[ptuomov]

And here's something little from the Handbook of Acoustics by Malcolm J. Crooker:

How much does a turbocharger reduce the exhaust noise at each frequency? Here are some estimates:
Octave Frequency Band (Hz), Unmuffled Exhaust Attentuation (dB)
31, 5
63, 9
125, 3
250, 7
500, 15
1000, 19
2000, 25
4000, 35
8000, 43
A-weighted, dB(A): 12

So what's the moral of the story here? The turbocharger muffles the high frequencies very effectively. It doesn't do much for low frequencies. Therefore, one should spend the available space to muffle low frequencies in a turbo car. Pure absorptive mufflers, such as small case diameter glass packs, are a waste of space on a turbo car.

[Tuner]

If someone knows of an affordable exhaust design software that can be used to predict sound attenuation, I'd love to get some leads. I am thinking about a software that would take pipe lengths and diameters and basic muffler configurations as given and then predict the whole system transfer loss.

A section in this software http://www.profblairandassociates.com/G ... _Main.html calculates dB reduction obtained using side-resonant and diffusion means. You enter cavity size, hole size, material thickness, lengths, diameters, etc, and the calculation results in a graph showing predicted dB reduction vs. frequency. This software is best used in companion with Prof. Blair's book http://books.sae.org/r-186/ as the text and pictures in the book will make it much easier to visualize the design goal. If you have an audio recording of the sound you want to attenuate, you can use an audio program such as "Audacity" https://sourceforge.net/projects/audacity/ to determine what specific frequencies you want to target.

[hoffman900]

EngMod4T has a noise spectrum processing component.

[ptuomov]

If someone knows of an affordable exhaust design software that can be used to predict sound attenuation, I'd love to get some leads. I am thinking about a software that would take pipe lengths and diameters and basic muffler configurations as given and then predict the whole system transfer loss.

A section in this software http://www.profblairandassociates.com/G ... _Main.html calculates dB reduction obtained using side-resonant and diffusion means. You enter cavity size, hole size, material thickness, lengths, diameters, etc, and the calculation results in a graph showing predicted dB reduction vs. frequency. This software is best used in companion with Prof. Blair's book http://books.sae.org/r-186/ as the text and pictures in the book will make it much easier to visualize the design goal. If you have an audio recording of the sound you want to attenuate, you can use an audio program such as "Audacity" https://sourceforge.net/projects/audacity/ to determine what specific frequencies you want to target.

Thanks, will take a look.

My current guess is that these will be the dominant frequencies:

rpm/EO 4 2.5 2 1.5
1000 66.7 ----- ----- -----
1500 100.0 ----- 50.0 -----
2000 133.3 ----- ----- -----
2500 166.7 ----- 83.3 62.5
3000 200.0 ----- 100.0 75.0
3500 233.3 ----- 116.7 -----
4000 266.7 ----- 133.3 -----
4500 300.0 ----- 150.0 -----
5000 333.3 ----- 166.7 -----
5500 366.7 ----- 183.3 -----
6000 ----- ----- 200.0 -----

I think I'm well muffled above 400 Hz no matter what since I've got the turbos, at least when the wastegates are closed. Also, I don't particularly care whether it's loud above 3000 rpm. Frequencies at 200 Hz and below that occur at 2500 rpm and below are the once that most concern me.

[vannik]

EngMod4T has a noise spectrum processing component.

I think neither EngMod4T nor Prof Blair's 4stSOFT software has the capability to do what Tuomo is attempting, for that he needs GT/Power or Ricardo's Wave in conjunction with 3D CFD.

[ptuomov]

EngMod4T has a noise spectrum processing component.

I think neither EngMod4T nor Prof Blair's 4stSOFT software has the capability to do what Tuomo is attempting, for that he needs GT/Power or Ricardo's Wave in conjunction with 3D CFD.

I don't think I'll need all that to make a more educated guess!

Unfortunately from the result perspective but fortunately from the optimization difficulty perspective, we are highly constrained by the space available. We're going to use the space available and our existing rules of thumb to build the next system, parts have already been ordered. The system will have all the tricks we think we can put in it based on simple spreadsheets and logic, short of active valves. The center mufflers will be dual 3.5" Borla XR-1 Sportsman 12" ovals. There will be two cross-overs, including a large oval center pipe and a trombone resonator in the the end. The tailpipe muffler will be the largest custom fabricated can that fits there while still looking right. It'll be filled with absorption material and have a 4.5" round outlet pipe.

The probable test plan with that system is to measure the power and noise levels on chassis dyno, while inserting different size restrictors to the 4.5" tailpipe. Hopefully, based on those tests we can decide the final tail-pipe size. After the tail pipe size has been optimized, we'll document the system and record the sound in different operating modes both inside the cabin and outside the car. At that point, we may or may not choose to custom fabricate larger center mufflers to replace the dual 3.5" XR-1 to take out more low-frequency sounds. If this turns out to be necessary/desired, I'll come back for more advice.

Thanks for everyone's help. I'll update this based on results in about a month.

[Sir Yun]

I looked it there was something that would help within the OpenFoam CFD framework but it's all hideously complicated

[Tuner]

EngMod4T has a noise spectrum processing component.

I think neither EngMod4T nor Prof Blair's 4stSOFT software has the capability to do what Tuomo is attempting, for that he needs GT/Power or Ricardo's Wave in conjunction with 3D CFD.

It must be you say this because you haven't used Blair's....?

[ptuomov]

EngMod4T has a noise spectrum processing component.

I think neither EngMod4T nor Prof Blair's 4stSOFT software has the capability to do what Tuomo is attempting, for that he needs GT/Power or Ricardo's Wave in conjunction with 3D CFD.

It must be you say this because you haven't used Blair's....?

I tried looking up the software on the page, but I couldn't find anything on muffler design or even exhaust design using standard and idealized muffler components. Can you point me to the exact location to read about those features of the software? I can get the waves coming out the tail pipe from Vannik's software, and probably add expansion chambers etc. simple structures to it, but I don't know how close to the reality that would get.

Right now, the plan is to put together the exhaust in away that flows and uses the obvious noise suppression and ducting tricks first. Then, measure where we are and identify any areas that we might be able to change with positive consequences.

[Tuner]

Zoom in on the numerical list here. # 16 and 17 http://www.profblairandassociates.com/G ... _Main.html

Perhaps your local library or bookstore has a copy of Blair's book, "Design and Simulation of Four-Stroke Engines".

[ptuomov]

Zoom in on the numerical list here. # 16 and 17 http://www.profblairandassociates.com/G ... _Main.html
Perhaps your local library or bookstore has a copy of Blair's book, "Design and Simulation of Four-Stroke Engines".

I've done the expansion chamber, side branch resonator, and Helmholtz resonator math separately and in series in .xls already. Vannik's software can do the waves with the two unequal paths to the rear muffler, but I haven't programmed that yet. Can this software do that? Or is it limited to having the elements in series?

I own the relevant chapter from Blair's book.

Thanks for your help, by the way.

[Tuner]

The software calculates the math in Blair's book.

[vannik]

It must be you say this because you haven't used Blair's....?

4stSOFT can do a basic diffusing element and a basic side resonance element.

I have both 4stSOFT and 4stHEAD, maybe you are thinking of Virtual Four Stroke?

[ptuomov]

Ok so this software by Blair is 1D and I've been programming simple acoustic wave formulas to a spreadsheet. Different.

[ijames]

I found this old document very useful in explaining how these different muffler/resonator combinations work:

https://crgis.ndc.nasa.gov/crgis/images ... fflers.pdf

Thank you for this link. I know you already have a plan in hand for your project but if you don't mind I'd like your opinion on an extension of my tubes within tube idea. In the naca report they characterize expansion chamber mufflers based on the ratio between chamber diameter and inlet tube diameter, with bigger ratios yielding more attenuation. Any idea if this would still hold for multiple inlet tubes in parallel? My idea is to fill a 3" inlet tube with 1/2" ID tubes, with these tubes 3 to 5 diameters long so 1.5-2.5" long, with the ends of the tubes in the plane of the flat endplate that begins the expansion chamber. If that chamber ID is 6" this gives a ratio of 2 without the tubes and 12 with the tubes. Without the tubes the attenuation would be minimal, much less than 10 dB, but with the tubes the attenuation should approach 20 dB. Keeping the expansion chamber short extends the upper cutoff frequency so I think 6" would be best so start the divider wall to the next chamber 6" from the endplate, and make this divider wall from another bundle of 2.5" long 1/2" ID tubes, then follow that with another 6" long chamber to the flat end wall with 3" tubing in the center as the exit. This two chamber muffler should give 30+ dB of attenuation and I think should have minimal backpressure. There will be flow through the 1/2" ID tubes, and also some flow in the gaps between the tubes which will act like smaller diameter tubes for even more attenuation. The divider wall should have minimal restriction since it has 4x the open area, so the primary restriction will be the entrance bundle of tubes. If this proved too much a larger ID tube could be used here (3/4"?), and maybe a third chamber added. This could also be done using an oval shaped chamber which would probably give more attenuation. I know a drawing would be good, but think of this as a 2 or 3 chamber muffler 14 in Fig 11 with the chamber length as in muffler 5 in Fig 9. I admit this is basically intellectual curiosity on my part, but one thing they stress in the paper is that resonators don't work well with high flows and sound pressure levels so my idea might be a possibility to replace your Borla first stage mufflers before your custom resonator. Thoughts, boos, catcalls ?