#1951: Ssshuperposition

According to my back-of-the-envelope calculation, the flow velocity within an exhaust pipe of a standard sized car is a small fraction of the speed of sound in carbon dioxide.

This allows me to treat the exhaust pipe, for acoustic purposes, as if it were filled with static fluid.

Today’s invention is therefore a suppressor for combustion engine exhaust noise.

Each pulsewave emitted by the opening or closing of the exhaust valve is allowed to pass into the clockwise arm of the looped pipe at 1 o’clock * (using a valve linked to the exhaust valve itself -not shown).

Once the pulse travels around the loop, it is allowed to leave using another linked valve at the 11 o’clock position.

By tuning the length of the loop to be (n+ 1/2)*the exhaust note wavelength, destructive interference can occur between the exhaust and loop waves, greatly reducing the engine noise emitted.

*I had to rethink this whole thing to avoid using the valves and fixed-length loop (thanks Andy, see below). Instead, a toroidal loop would be inserted in the exhaust as shown. A port would allow waves to enter and propagate around the loop. The loop itself would be spun, much faster than the wave speed, so as to position its port against the exhaust pipe. It would do this in such a way that if a compression were moving down the exhaust, the loop would inject a rarifaction and vice versa. In this way, pressure fluctuations could be eliminated within the exhaust, without introducing extra ones by the operation of conventional valves.


  1. If the loop was made of a flexible material then it could be stretched to tune the length dynamically? You might find though that the engine noise is made up of multiple harmonics so perhaps multiple loops could be utilised?

    Ref: http://www.f2d.dk/noise/noisefreq.htm

  2. A related thought was that you could simply have a series of tubes at right angles to the exhaust then like blowing over the top of a milk bottle the pipes would resonate at a particular frequency.

    • Interesting idea. My view of the exhaust flow is that it’s much much slower than the speed with which waves, caused by valves snapping open and shut, can pass through the moving gas. The gas can be treated as if it were stationary (once the waves fill the rotating section, in the above invention, they can be spun around ultra-quickly, with the fluid acting effectively as a solid. so that nodes get dumped onto antinodes).

      If engine noise were just caused by blowing through a tube using a low-speed, continuous flow, (like eg a central heating boiler), then a different solution might work. If a standing wave pattern were to be set up in a tube at right angles to the exhaust pipe (with antinode at the entrance and node at the closed end of the tube) then a quick sum shows that the tube length would need to be around 1m, at maximum (which seems feasible). I guess the trouble is that, like blowing over a bottle, the whole thing might actually make more noise rather than less. There needs to be a way to superimpose nodes on antinodes, so that the energy is dissipated before it gets transmitted as acoustic waves.

      So, your milk bottle approach might work in the following way. This suggests having two milk bottles being blown over by the continuous, low-speed exhaust. One is closed at the end, one is open. Whenever a stable wave pattern gets established, with a node at the closed tube end and an antinode at the open tube end, a valve in the closed tube end is quickly opened, causing adjacent node and antinode to interfere with each other.

Comments are closed