Acoustics and Wind Turbine Noise

By Chris Kapsambelis

Introduction

In Massachusetts, local acousticians are using the wrong metric to evaluate compliance with the MassDEP policy on environmental noise pollution. To understand this, we first must know what parameters are used to assess the magnitude of noise with different characteristics.

Noise can be categorized by two major characteristics–continuity and impulsiveness. Noise from a diesel engine is characterized as continuous with a steady state of volume, while noise from a pile driver is impulsive where the volume peaks with each hammer blow followed by intervals of relative silence. Wind turbine noise contains elements of both characteristics.

For a true assessment of its magnitude we need to understand how the background ambient sound is derived and what to measure when the wind turbine is operating.

In the field of acoustics there are several statistical descriptors that are used as metrics in the measurement of sound. These are very well explained in this link. Be sure to click on the buttons marked “DEMO” and “NEXT” to get the full demonstration. The two important values from the graphic below are L90 and the Maximum.

The L90 is the threshold level exceeded for 90% of the time. For 90% of the time, the noise volume is above this level, and for 10% of the time, the noise volume is below this level. The quietest portion of noise, what sound is heard 10% of the time, is generally accepted to be the background or ambient level of a noise environment.

The Maximum, also referred to as Lmax, is the instantaneous absolute maximum noise level we can hear during the time period under consideration.

It should be noted that if we were to evaluate noise conditions between midnight and 1:00 AM, and someone nearby sets off a firecracker at 12:30 AM the L90 ambient will remain unchanged, because the short duration firecracker noise will be quiet for more than 10% of the time between midnight and 1:00 AM, and only the ambient (L90) will be present. On the other hand, the Maximum, Lmax volume will skyrocket as an impulse of noise for a very short time at 12:30 AM. Doubtless, the noise will be noticeable and annoying, but the ambient (L90) measurement will be unaffected.

Acousticians and the MassDEP Policy Standard

It has become common practice for acoustical consultants to use the wrong statistical descriptor metrics to calculate the increase in noise emitted by industrial wind turbines, even though a document published by the MassDEP in 2003 provides a step by step procedure on how to measure the offending wind turbine noise. The MassDEP instructions for taking field background ambient (L90) plus sound source readings is:

“Take an instantaneous reading every 10 seconds. Continue taking readings for as long as necessary to determine the maximum sound level from the sound source.” (MassDEP Guidance, 2003, p. 12).

The word maximum is highlighted to signify that the method describes the metric descriptor Lmax. Instead of comparing Lmax to L90 to determine MassDEP compliance, local acousticians have been comparing the increase in the L90 level when the turbine is operating–comparing the before and after L90 instead of the L90 to Lmax as instructed. The telltale sign is contained in the “Executive Summary” of the study commissioned by the town of Falmouth in response to numerous complaints shortly after Wind-1 began operating. Below is the excerpt containing the highlighted language which misstates MassDEP policy by claiming to use an average of L90 readings to determine ambient and an increase in the L90 average readings to determine compliance.

“The average Mass DEP L90 background sound levels were measured to be 29 dBA in the early morning hours and 41 to 46 dBA during the day at the two long-term measurement locations. The average increases in background noise levels with the Wind-1 turbine operating were measured to be about 1 Dba during the day and 4 to 8 dBA at night; therefore, we conclude that Wind-1 is operating within Mass DEP requirements.”

An examination of Table 2 in the main report further confirms the fact that the compliance study makes use of the L90 metric not only to measure the ambient, for which it was intended, but to also measure the offending noise as an increase in the ambient which was not and is still not the intent of the MassDEP policy. Furthermore, the 2003 MassDEP Guidance makes use of absolute minimum values to measure the ambient (L90) and the absolute maximum (Lmax) to measure background plus sound source. The local acousticians use of average readings further understates the before and after comparison. To illustrate the absurdity of relying on the L90 metric for both measurements, let’s use the example of the pile driver. The link below demonstrates what it sounds like.

in the page that launches, click on the button in the orange tab to start the recording

A graph of the pile driver noise may look like this:

Notice that there is a very loud impulse of sound followed by a period of quiet with only the ambient present, which is at least 10% of the time. The Lmax metric will show the peak value of the impulse (90 dB(A)). The L90 will show the existing ambient (30 dB(A)). Again, because the noise is not present for more than 10% of the time the L90 ambient will not show any increase. If you are having trouble understanding the above example, please visit this link again for a better understanding of how these metric descriptors are derived.

Wind Turbine Noise

This first chart results from an acoustic study conducted by a Falmouth resident on March 26, 2011.

The characteristics of wind turbine noise are not as severe as the previous example of the pile driver, but depending on your relative distance, wind angle and the wind turbine power output, the graph above is not uncharacteristic. Notice that the waveform has impulse peaks and valleys as might be expected from the pile driver example. This impulsive characteristic of wind turbine noise is the result of the wind rushing around the blades. The peaks and valleys of sound volume are the result of each blade passing near the ground–which occurs at a little less than once per second. Researchers have named this characteristic Aerodynamic Amplitude Modulation (AAM). Resident complaints have described this noise as “whoosh, whoosh, whoosh…,” but the noise has been found to be more annoying than other noise sources of equal magnitude. The L90 measurement can be estimated at about 36 dB(A). The Lmax value is 46 dB(A). It is, therefore, obvious that even without measuring the ambient noise level with the turbine off, the difference between L90 and Lmax approaches the MassDEP 10 dB(A) limit. Here are additional graphs from data collected during the recent compliance testing of Falmouth’s Wind-1. The graph below shows the noise levels just before and just after turbine shut down–which occurs in the middle of the graph. A cursory examination shows that the difference in the ambient L90 levels, which has been the erroneous metric used by local acousticians, is a little less than 10 dB(A). However, the L90 to Lmax comparison, for compliance with the MassDEP policy, shows a difference of about 15 dB(A). The larger noise impulses are excluded because they are caused by the mechanics used to bring the wind turbine to a stop.

The graph also points out that the wind turbine noise exceeds the absolute 37 dB(A) threshold recommended by the recent MassDEP/DPH  Wind Turbine Health Impact Study for night time operation (p. 60). It also shows that Falmouth’s own absolute limit of 40 dB(A), called for in the town bylaws, is also exceeded. The graph below is a 30 second time expanded close up of the impulsive nature of wind turbine noise.

Notice that there 7 to 8 noise peaks in a 10 second period to coincide with the timing of each blade passing near the ground.

Wind turbine noise is made up of at least two components. One might be described as “normal” continuous noise generated by various mechanical components such as gear trains, and the other is the AAM which is riding on top in the graph above. The gear noise component does in fact result in an increase of the L90 ambient, in the same way traffic noise is louder at rush hour than at 4:00 am. But including the industrial noise of the turbine in the ambient not only distorts the fact that they are not natural sounds, or even sounds of ordinary urban life. It also masks the absolute peaks in the sound that occur when AAM is included.

The graph below which shows the reduction in noise as Wind-1 as turned off, illustrates another subtle but significant difference in instrument use and settings. Modern sound meters sample the sound waves and calculate the dB(A) value with at least two conventional time constants. These are known as “Slow” and “Fast.” For impulsive characteristics like AAM the “Fast” setting is more accurate. The pink color trace in the graph is the result of the “Slow” setting.

Below is an expanded 30 second time interval of the previous graph:

The blue is from the “Fast” setting. As can be observed, there is about a 4 to 5 dB(A) difference between “Fast” and “Slow.” It should be noted that the MassDEP in Falmouth elected to use the “Slow” setting which tends to understate the peak value of the AAM sound level. The “Fast” setting would most likely have found more instances of non-compliance.

Compliance Testing Prot0col

Collecting data to prove compliance is challenging to the point where it may not be possible or practical. Instead, testing is used to assess lack of compliance.

At this point MassDEP is bound by the protocol contained in the 2003 MassDEP Guidance document, which was reaffirmed  in the  letter to Falmouth dated January 24, 2011, requiring an extensive amount of attended measurements. In the letter, MassDEP calls for data to be collected

• for each hour of the day
• for each increment of wind speed
• from turbine cut-in to cut-off
• with the wind turbine off to establish the L90 ambient levels
• repeating the process with the turbine operating (to determine the corresponding Lmax).

Because wind turbine noise is also a function of wind direction, one could include the additional requirement for each variation in wind direction.

There is no question that this is the protocol that is needed to prove compliance beyond reasonable doubt. However, it has become obvious that collecting the data will takes years of monitoring, and the monitoring will require a technician in attendance to manually remove transient instances of noise not attributable to the wind turbine.

In a second letter to Falmouth dated June 30, 2011, MassDEP recognized  Falmouth’s unattended study by an outside consultant, and reduced the requirements to an additional short term attended study to collect data at three operating wind speeds:

1. at or near the cut-in speed where background will be the lowest;
2. at the wind speed where manufacturer data indicates there will be the greatest sound power level from the turbine; and
3. at the maximum wind speed where the turbine will be operating.

During the month of March, 2012 MassDEP performed an Attended Sampling of Sound , using a shortened version of the 2003 Guidance to determine compliance of Falmouth’s Wind -1 turbine. It found Wind-1 noise exceeded the 10 dB(A) limit at several wind speeds during the night time at a nearby residence 1,320 feet from the turbine. Wind-1 was declared to be in violation. The MassDEP then accepted Falmouth’s decision to turn off both of its turbines from 7:00 PM to 7:00 AM as a mitigation measure. Further testing to determine daytime compliance is underway.

The Falmouth short-term attended study deviated from the 2003 Guideline by averaging the Lmax readings of three different data collection runs. The absolute maximum would be more representative of what one hears and what therefore produces annoyance and complaints.

As was demonstrated by the previous graphs the meter setting of “Fast” is more accurate in reading the effects of Aerodynamic Modulation (AAM). In Falmouth, MassDEP elected to use the “Slow” setting which understates the Lmax readings by 4 to 5 dB(A) in that particular instance. In future compliance testing, this leaves the MassDEP in the challenging position of having to “guess” the conditions under which worst case noise will be generated in setting up the test locations and wind conditions for testing.

While it is clear that any results showing non-compliance can be relied upon, the limited amount of data and the “Slow” meter setting will not be sufficient to declare any wind turbine in compliance, without extending the data collection cycle to the initial long term protocol, and using a meter setting of “Fast.” The MassDEP is now in a position to declare the conditions under which wind turbines are not in compliance with MassDEP policy. However, the shortened version of the 2003 Guidance is not robust enough to determine compliance under all possible conditions leaving us with no way to verify compliance in cases where wind turbine noise complaints are filed.

Conclusion

It appears that MassDEP has now settled on the shortened version of the 2003 Guidance used in Falmouth for testing in Fairhaven and Kingston.  This is a major deviation from what the MassDEP has proposed in the past. While it is clear that any results showing non-compliance can be relied upon, the limited amount of data and the “Slow” meter setting will not be sufficient to declare any wind turbine in compliance, without extending the data collection cycle to the initial long term protocol, and using a meter setting of “Fast.” The MassDEP is now in a position to declare the conditions under which wind turbines are not in compliance with MassDEP policy. However, the shortened version of the 2003 Guidance is not robust enough to determine compliance under all possible conditions leaving us with no way to verify compliance in cases where wind turbine noise complaints are filed.

References:

Harris Miller Miller & Hanson, Inc. Falmouth Wind Turbine Noise Study, Falmouth MA. September, 2010.

Massachusetts Department of Environmental Protection. Attended Sampling of Sound from Wind Turbine #1, Falmouth MA. May, 2012.

Massachusetts Department of Environmental Protection. Noise Guideline Document. April, 2003.

Wind Turbine Health Impact Study: Report of Independent Expert Panel. January, 2012.