Testing the acoustical noise of a generator set and interpreting the data

October 31st, 2017, Published in Articles: EE Publishers, Articles: Energize, Featured: Energize

A growing number of businesses find that large, heavy-duty generator sets are critically important to their operations, whether for continuous, prime or standby power. Not surprisingly, large generator sets produce high levels of noise that often must be mitigated to meet local, state and federal regulations or directives.

This usually involves key decisions about the optimal generator set location and installation, how to design noise control systems, and other factors. That is why it is important to have an accurate assessment of noise levels early on, before investments are made and facility and enclosure designs are locked in. However, it can be difficult to obtain precise, consistent data which can be easily compared.

Not all manufacturers measure generator noise the same way, and some use less precise standards for their data. As generator set acoustic performance grows in importance, it is important to understand how to compare the data in a way that’s “apples to apples”.

Why precision acoustical testing matters

Noise levels at workspaces are regulated by federal governments to protect workers’ safety. In addition, communities are increasingly enacting ordinances aimed at reducing sounds that are disruptive to nearby residents.

Fig. 1: Generator undergoing acoustic testing at Cummins ATC.

For example, in North America, typically permitted noise levels during the day and at night are around 60 dB(A) and 50 dB(A), respectively. However, in certain special zones, local authorities have set even lower noise limits. That’s why it is very important to measure noise levels using appropriate and applicable national and international standards in order to establish an accurate platform for comparison.

These precisely measured noise levels are also useful for understanding the noise characteristics of generator sets so that noise control strategies, such as acoustic enclosures and acoustic walls, can be successfully implemented.

In order to bring a generator’s noise levels into compliance with local ordinances and federal regulations, facility design engineers must carefully evaluate where to locate the generator and what noise reduction strategies will be most effective. These strategies may include the construction of sound-attenuating enclosures or the addition of acoustic insulation or barriers, exhaust silencers or isolation mounts. By equipping power system designers with precise acoustic data, they’re better able to design a noise reduction system that reliably meets local noise restrictions.

Understanding acoustical data measurements

When comparing generator noise and sound data, it helps to understand some of the most common units of measurement used for industrial products. Sound levels are often expressed in terms of sound pressure level and sound power level. Both are important measurements, but they’re quite different.


A-weighted sound pressure level (SPLA or LA) is specific to the distance of the receiver from the sound source. SPL decreases as distance from the sound source increases, and vice versa. Without the context of distance from the source, SPL data is meaningless.

Fig. 2: Location of microphones to achieve 360° baseline noise data.

A-weighted sound power level (SPWLA or LWA) is a measure of acoustical energy produced by a sound source, with no regard to its distance from the point of observation. Sound power level is calculated based on SPL measured by using parallelepiped or hemispherical array methods as described by ISO or ANSI standards.

For electric power generator sets, noise levels are also dependent on two major operating parameters: engine speed and load. Since noise levels increase as speed and load increase, it’s important to know at what speeds and loads the noise levels were measured.

Cummins conducts noise testing at its acoustical testing centre (ATC) in Minnesota, USA. This carefully engineered and environmentally controlled facility provides exceptionally reliable acoustic data through the use of precision-grade acoustical equipment and rigorous measurement techniques.

The ATC is among the largest in the diesel and power generation industries, featuring a hemi-anechoic chamber certified as “precision grade,” the maximum accuracy level according to ISO 3745:2009. Its vast size is needed to accommodate large generator sets and the many microphone positions required for precision-grade testing, including those at seven meters from the face of the largest generator set the company manufactures.

By using the microphone setup according to ISO and ANSI standards, measurements are conducted all around the generator set so that all unique noise characteristics of its components are captured and also the directionality of the noise sources studied. This precision-grade noise data increases the operator’s confidence level and eventually helps in devising noise control strategies quickly and economically.

A glossary of acoustical terms

  • A-weighted sound power level (SPWLA or LWA): A measure of the acoustical energy produced by a sound source, no matter what its distance from the point of observation.
  • A-weighted sound pressure level (SPLA or LA): A relative measure of the impact of a sound wave at a specific distance from the source. For example, a generator set producing an average SPL of “X” dBA will be much quieter when heard from 50 feet away than from 5 feet away. Therefore, the average SPL will be lower at 50 feet than at 5 feet, although the sound power level has not changed. Both SPWLA and SPLA are represented by the same unit of measure — dBA — but they are two different metrics.
  • Cycle: The complete oscillation of pressure above and below the atmospheric static pressure.
  • dBA: A-weighted variation of the standard decibels measurement. Since the human ear has different sensitivity to different frequencies, noise levels are often measured with an “A-weighted” filter that has a frequency response similar to that of the human ear, stated as dB(A) or dBA.
  • Free field: A sound field in which there are no or negligible sound reflections.
  • Hertz (Hz): Frequency of sound expressed by cycles per second.
  • Noise: Unwanted sound that is annoying or uncomfortable.
  • Parallelepiped array: A three-dimensional arrangement of microphones at a certain distance (most commonly at 1 m) from an imaginary parallelepiped that encloses the testing object.

Recommendations for evaluating acoustical data

With the growing focus on generator set noise and its impact on workers and nearby communities, noise mitigation is an essential component of generator set installation. Obtaining precise acoustical data allows for the design of effective and cost-effective noise control solutions that bring generator sets in compliance with community and federal noise limits.

However, comparing generator set acoustical data isn’t as simple as it would seem. Not all testing facilities meet the latest ISO, ANSI and SAE standards. In addition, various manufacturers may offer different types of data, which can make comparing “like to like” numbers difficult. Here are some questions to consider when comparing generator set acoustical data from different manufacturers:

  • Were the generators tested at the same engine load/power node?
  • Were the generators tested at the same engine speed?
  • Was the data collected with or without a set-mounted radiator?
  • Are the units of measurement the same? (SPLA or SPWLA)
  • Were sound pressure level measurements taken at the same distance?
  • Was the same standard or measurement method used for the noise measurements?
  • How reliable is the data?
  • Were the measurements conducted at precision- or engineering-grade test facilities? Precision grade is more precise than engineering grade.

Testing the Cummins QSK95 Series generator set

The QSK95 Series generator set is equipped with a 16-cylinder diesel engine capable of producing approximately 3,5 MW of power. It was tested at a variety of operating conditions and a range of acoustic parameters to determine how different aspects of normal operation would contribute to overall noise levels.

Testing focused on sounds which are most impactful to the human ear and structures surrounding the generator sets, and frequencies which have a higher potential to disturb people in surrounding communities.

The company strived to achieve high precision and reliability in its numbers, calculating average readings from an array of 71 microphones. This covered the entire human audible frequency range, from 20 to 20 000 Hz.

Engineers conducted 58 different test configurations over a total of 321 hours in the testing facility and acquired almost 360 GB of noise data. By using advanced noise measurement techniques developed at the ATC, the company was able to determine the noise contributions from individual components of the generator set to overall noise levels. This included the noise characteristics of the engine, cooling system, exhaust system, air intake system, fuel injection system and alternator.

Acquiring noise data in this manner helps ensure precise information is incorporated in product specification sheets. It is also helpful for quickly devising noise control strategies without the need for extensive engineering resources, whether it’s to meet regulation- or directive-based noise limits or customer-specific requirements.

Some highlights of the QSK95 Series generator set testing:

  • Baseline noise data was acquired at 1 m using the parallelepiped method as per ISO 3744:2010 with 59 microphone locations around the generator set (engine and alternator ends, sides and top of the generator set).
  • Baseline noise data was also acquired at 7 m using eight microphones (eight locations spanning 360° around the generator set, each microphone placed at 45° increments) based on internal noise measurement standards derived from the US Department of Defence power generator set noise measurement requirements. Noise data collection at 7 m from the surface of the generator set is important because sound levels are generally stable at that distance, which enables reliable extrapolation and prediction of noise levels beyond 7 m. Very few hemi-anechoic chambers in the world are large enough to conduct testing at that distance.
  • Cummins used 71 microphone locations to measure noise levels at 1 and 7 m from the generator set, 1 m from the line of exhaust and fuel pump, and at operator location to achieve a very high degree of reliability.
  • The average SPLAs at 1 and 7 meters reported in the data sheets is an average of 59 and eight microphone locations at a distance of 1 and 7 m from the generator set, respectively.
  • The SPWLA reported in the data sheets was calculated using the average SPLA measured by using 59 microphones placed at 1 m from the generator set, as per the parallelepiped method described in ISO 3744:2010.

These extensive testing configurations resulted in highly detailed, three-dimensional (sound field) sound data. This data provides a valuable resource to customers who seek to design effective and cost-efficient generator enclosures and noise attenuation systems.

Contact Kenneth Gaynor, Cummins, Tel 011 589-8400, kenneth.gaynor@cummins.com