Fact Sheet: Combustion Pulsation & Noise

All flames make noise and larger ones make more. You can’t hear the noise from a pocket lighter, but you know when your gas water heater or gas fired central heat burner starts. And any industrial size furnace makes enough noise that you can tell if it is on without looking.

Usually combustion noise can be ignored – it seldom causes problems.

In rare cases it’s loud enough to draw complaints from the neighbors and even vibrate equipment to destruction. Here we will consider cases like that.

A Little Theory:

Any noise has a frequency and an amplitude.

You can hear noise over a frequency range of about 20 to 20,000 Hz or cycles per second. Noise at a frequency less than 20 Hz might be felt but not heard – you might notice the floor or windows rattling even though you can’t hear the sound.

If you have good hearing, the quietest sound you can hear is slightly louder than zero decibels (db). The loudest (I’m told) is 120 db, at which point your ears are quickly damaged. (http://en.wikipedia.org/wiki/Psychoacoustics)

In the combustion world, there seem to be two categories of noise:

  • Noise formed by the resonant characteristics of furnaces, ducting, stacks, etc.
  • Noise formed by a flame only.
  • Actually noise in the first category is seldom a problem unless a flame is also involved. Noise inherent in the system appears to be amplified by the flame energy. A few field reports illustrate this:

    Field Report No. 1 "Claus Plant Incinerator"

    A new Claus tail gas incinerator was built at a refinery in the Houston area. It included a relatively long tail gas duct running from the last sulfur condenser down to the incinerator burner. The burner was natural gas fired and included a centrifugal air blower to overcome the pressure drop across a water tube waste heat boiler positioned at the incinerator furnace outlet and exhausting to a carbon steel stack.

    This system was designed for future plant expansion, so throughput was only a fraction of design. Immediately upon startup, the operators heard a very loud, relatively high frequency noise. The noise disappeared when the burner was fired at its design rate, and of course it went away when the burner was off.

    Attempts to modify the burner fuel gas injection geometry had little effect on the noise, even though this step often eliminates or at least reduces similar problems.

    An acoustics expert was called in. Microphones were installed at several locations throughout the system but the problem source remained elusive until, with the burner turned off, low level noise at the problem frequency was detected. This meant that the noise was being triggered somewhere in the system and the burner was simply amplifying the signal into a container accidentally sized for resonance.

    Ideas to alter the resonant frequency of the equipment were considered. For instance, shortening or lengthening the stack would have changed the resonant point and cut the noise amplitude. The fix finally selected consisted of installing a perforated stainless steel plate across the boiler outlet flange. The perforations were sized to create as much flue gas pressure drop as the blower and Claus plant could handle. That fixed the noise problem by interrupting the resonance.

    Field Report No. 2 "Chemical Plant Thermal Oxidizer"

    At a chemical plant, also near Houston, a new thermal oxidizer system was installed to dispose of offgas from an acrylonitrile plant. The system was to have a waste heat boiler installed later, so the horizontal furnace was connected via a long duct to the exhaust stack. A forced draft natural gas burner was used to bring the furnace up to temperature. Once the refractory was cured out, the operator switched the vent gas stream away from atmospheric vent and into the burner. Immediately a very loud, high pitched noise was heard. Apparently the furnace system was sized just right for resonance and the burner provided the amplification, while the noise was started at the inlet valve. Fortunately, a simple change to the inlet valve position changed the frequency enough to reduce the noise to acceptable levels. In this case the change was easy to accomplish and worked "like magic".

    Field Report No. 3 "Elevated Flare"

    A refinery in California installed a large, steam assist, elevated flare. Steam injection was necessary for smokeless burning of the waste gases. Upon startup, a very loud, low frequency (2 Hz) noise was produced. It resulted in complaints of vibrating floors and walls from workers well away from the flare site and steps were immediately taken to find the cause and eliminate it. With trial and error and a few lucky guesses, the cause was determined to be the steam jet position in relation to the waste gas injection passage. By raising the steam jet elevation about 2 inches, the problem noise was eliminated and smokeless operation maintained.

    Field Report No. 4 "North Dakota Hydrogen Vent Incinerator"

    A petrochemical plant in North Dakota produced a waste gas rich in hydrogen. It was to be used to fire a process heater. The heater was equipped with a number of floor mounted burners, each with a central fuel gas gun mounted inside a refractory throat. Upon startup, a very loud noise was produced. In this case there were operator complaints, and the amplitude was so great that it was loosening the nuts holding the furnace platforms and other hardware in place. With nuts falling, the furnace could not operated while personnel were below. This problem was solved by adding a central gas passage through the center of each burner gas gun. This changed the flame geometry enough to "detune" the burners and reduce the noise to acceptable levels.

    Field Report No. 5 "Houston Chemical Plant Waste Heat Recovery Plant"

    A petrochemical plant near Houston operated a large horizontally fired incinerator mated to a water tube heat recovery boiler and economizer. The flue gas was ducted to atmosphere through a 100 ft. stack close coupled to the economizer. The waste gas had very low heating value and was injected, along with the combustion air, through a series of stainless pipes installed through the furnace refractory lining. A natural gas burner started the combustion and two waste liquids were sprayed in between the burner and the waste gas injectors. With increased plant production, a "noise" began to appear. In this case, the frequency was only 1 or 2 cycles per second, and was detected by rhythmic swelling of the fabric expansion joints connecting the economizer to the stack. Expansion joint life was reduced to about 6 months, and the system had to be taken down for fabric replacement (total shutdown about 3 days each time). Previously, the plant could be run about 2 years between shutdowns. In this case, modifying the waste distribution across the pipe injectors resulted in enough amplitude reduction to solve the problem. The low frequency noise was still present, but flexing of the expansion joint fabric was eliminated.

    More Theory:

    How can a fuel gas burner can act as an amplifier for noise? It appears that the pressure changes which cause the noise are created when the fuel and air streams are combined improperly. If the two streams are combined incorrectly, pockets of rich or lean mixtures are created which are not "flammable". As additional fuel or air enters the pocket, the mixture becomes flammable and burning continues, increasing the volume of the pocket, which displaces the air and fuel streams, forming a nonflammable pocket again. As this cycle continues, the on/off nature of the combustion is seen as pressure fluctuations. With many industrial size furnaces, the pulsation frequency is in the range of 1 or 2 per second, but with hydrogen rich fuel gas or fuel gas under high pressure, the frequency can be greater.

    Premixing the fuel and air before ignition eliminates the problem. This "premix" burner method is found in the small pilot burners used to ignite an adjacent larger burner. This is also the type of burner used in kitchen ranges, propane grills and gas fired central heat units.

    Larger burners are typically "diffusion type" burners, meaning that the fuel (gas or oil or pulverized coal) is injected adjacent to the combustion air stream. As the fuel diffuses into the air stream (and vice versa) the burner flame develops and grows until all of the fuel is oxidized.

    So how would a fuel burner "amplify" an existing noise, as apparently happened in Report Nos. 1, 2, 3 and 5 above? In Report No. 1, the inert waste gas was injected through an annular gap surrounding the flame zone. Apparently the pressure fluctuations in the waste gas distorted the fuel gas / air mixing process at the existing frequency, and the energy already available from the flammable mixture pockets was redirected from standard combustion noise to the new frequency, which happened to be the same as the resonant frequency of the furnace system. Bad luck.

    The other cases involved a similar effect, although the circumstances look quite different.


    The negative effects of combustion pulsation include;

  • Rhythmic flame body displacement (flame detection device may lose sight of the flame),
  • Brief loss of combustion air flow as furnace pressure peaks, briefly reducing air flow (low flow switch may trip),
  • "Breathing" of the furnace shell and expansion joints leading to material failure through fatigue.
  • Structural damage to the furnace if the amplitude is high enough.
  • Extremely irritated operators and neighbors.
  • Eliminating combustion pulsation and excessive noise is possible through changes in burner operation (reduce or increase air flow, change steam flow, etc.), but often some sort of hardware change is required. One approach involves changing the location or velocity of fuel injection. Another involves changing the location direction of combustion air flow, steam flow or waste flow. And in some cases, steps must be taken to change the resonant frequency of the furnace/stack assembly. These changes are typically accomplished in an experimental manner - change only one thing at a time, record results and move on till all options have been tried.  Sometimes observation of the flame body will give a clue as to the best approach. There is some hope from acoustic analysis of the resonating chamber(s) as in Report No. 1, but this is presently very difficult and typically too expensive to use prior to a problem showing up.  

    Combustion noise is inevitable, but excessive noise can always be controlled without resorting to magic. 

    Dan Banks, P.E.

    Banks Engineering Inc.

    Tulsa, OK


    Phone 877-747-2354