The Sound Absorption Properties of Walls

While it is well known that lightweight stud walls absorb some sound, the specifics of their absorption behavior are, at times, hotly debated.

There are many questions about walls and absorption, including:

Does adding more mass – like double drywall – to a wall reduce its sound absorbing properties?
Do higher performance, high transmission loss walls have poor sound absorption properties because they don’t let low frequencies “pass through”?
Do resilient mounts, such as resilient channel improve the sound absorption of walls by allowing them to flex?
Does viscoelastic damping improve the sound absorption of walls?
What type of walls are the best low frequency absorbers?
What is the absorption coefficient at some given frequency for some given wall?
Does absorption peak at the low frequency resonance of the walls?

Answers to these questions could prove a great asset to persons designing high performance rooms, and indeed, a wall with low frequency sound absorption properties is desirable, as typical absorption treatments (bass traps, broadband absorbers) have diminished efficiency at very low frequencies. If a wall could be designed which would assist with absorption at subwoofer frequencies, higher sound quality studios and theaters could be constructed.

In a series of lab sessions in 2005, The Green Glue Company set out on perhaps the most ambitious project ever aimed at studying the sound absorbing properties of walls. The project spanned dozens of walls, of many different configurations, with a series of experiments designed to explore the questions listed above.

In the pages below we will present the fruits of this work. While the series of experiments were not perfect, a great deal was learned, and answers to all but one of the questions listed above seem to be attained.

Preface 1 – The Test Procedure.

The tests were run in the transmission loss test suite at Orfield Laboratories in Minneapolis, Minnesota. The test suite is arranged as shown below. The specimens were installed in the standard opening utilized for transmission loss tests, and each wall was tested for transmission loss in addition to absorption.

Transmission loss testing is conducted in accordance with ASTM E90, and involves making noise in the source room, measuring its level, and then measuring the noise level in the receive room. After some adjustments for various things, the transmission loss is (in essence) the sound level in one room minus the sound level in the other. Transmission loss is a measure of how much a wall reduces sound levels at different frequencies.

Typically, absorption is tested in accordance with ASTM C423. In this test process, you first test how long it takes sound to decay in the “empty” room, then add whatever absorption panel you wish to test and test it again. By comparing the differences in the decay rate, absorption coefficients can be calculated. When testing these decay rates, many individual decays are averaged over many microphone positions. The samples were tested for absorption in the receive room of Orfield’s facility, which exceeds the requirements for facilities given in C423.

Absorption testing is a standard part of testing transmission loss, and follows the same basic protocol as C423, except the empty room does not need to be tested, and ASTM E90 requires far less decays to be averaged. The absorption in the receive room is used for one of the adjustments mentioned above. The absorption tests in this case were conducted by this same process – averaging of the decay rate in the room over many individual decays and microphone positions.

In other words, the tests run on walls and discussed herein were conducted in a formal manner, consistent with other historical absorption tests.

Preface 2 - One big challenge and the absence of absorption coefficients.

A fundamental challenge involved in the testing of the absorption of walls in this manner is that it is extremely expensive, laborious, and time-consuming to construct an “empty” room to use as a reference.

To produce a simulation of the “empty” room used in C423 requires the installation of a wall with extremely low sound absorption – typically, this would be a solid concrete wall. Then, assuming that the absorption of the concrete wall was near zero, you could use the absorption present in the room as a reference, and attain true estimates of absorption coefficients.

The challenge – and limitation – is that for C423 one can test the “empty” reference room, and the room with absorption panels installed one immediately after the other.
The time, labor, and cost involved with installing a solid masonry wall into the test hole makes this both wildly impractical, and simply impossible. Therefore, if one is to build a concrete reference wall, it is realistically possible to do this only once or twice during the entire project. This is not an ideal situation.

One such reference wall constructed, and the absorption in this reference room recorded, but some unavoidable changes to the construction of the facility have resulted in a situation where (at least at this time) true absorption coefficients cannot be calculated.

Simply put, the reference data taken does not represent a “reference” room’s absorption after the facility modifications.That one drawback noted, on all other counts the project seems to have been a rousing success. So let’s proceed to discussion of some of the questions we raised above.

Preface 3- Understanding the graphs presented in this document.

Rules for construction of a high In this document, absorption coefficients are not presented. Rather, we present graphs showing the reverb time in the test room.

Reverb time relates to absorption like this – the lower the reverb time, the higher the absorption in any given room, all other things (like room dimensions) equal. So, in these graphs, a line that is lower on the graph represents higher absorption, like this:

Here we have a graph showing the reverb time with 3 different walls installed. They lines are intentionally not labeled – the data will be presented at a later time.

Lower on these graphs is more absorption. As the red line above shows, the lower the reverb time, the higher the absorption in the test room.

So, for example, at 400 Hz the blue line has notably more absorption than the black or green lines. At 80 Hz, on the other hand, the black line has the most absorption.

Preface 4 – About this data (test sessions and low frequencies)

Test session labels. This data is from two separate test sessions – one conducted in August of 2005, and one conducted in October of 2005. A third test session was conducted in April of 2005, but this was before the facility revisions, and the data is not presented.

If a graph is labeled “August”, that means that both tests were from the August session. The same applies for October. If a graph features data from both test sessions, this is noted below the graph.

Low frequency reverb time data. The data at 31.5 Hz is not reliable and should be used for conclusion of any type. It is included simply in the hope of being thorough.

The data at 40 Hz is consistent enough to be useful, but as with all tests of this nature, reliability falls as frequency falls.

On a similar note, the transmission loss data from Orfield’s test suite is useful down to the

31.5 Hz band, and very consistent down to the 40 Hz band.

Now onto the data...

PART 1- Does low frequency absorption in walls correlate to the low frequency resonant behavior of the walls?

To answer this question before we present the data – yes, it does. Perhaps the most important lesson that is taught from this data is that there does appear to be a very strong correlation between the location of resonance in any given wall and the location of the absorption peak of that wall.

This should be expected as walls are extremely analogous to typical panel-type bass traps, which are well known to have absorption peaks that correlate directly to the location of the primary mass-spring resonance of the walls.

For the purpose of this discussion, let’s assume that there are two basic types of low-frequency resonances in walls. The first are mass-spring-mass resonances, or MSM’s. These involve the air cavity and have traditionally (and incorrectly) been described at times as mass-air-mass resonances. This resonance involves two masses – these being the weight of the panels on either side of the wall – resonating on a spring. The spring being the air in the cavity of the wall. These sketches outline the basic behavior of different walls at MSM.

Low frequency resonances in walls.

In a staggered or double stud wall – which will always be constrained at the top or bottom by the plates (staggered wall) or attachments to frame or foundation (double stud wall) – the motion at resonance must involve bending of the studs/drywall panels.

Motion at resonance is, therefore, controlled by a combination of the air spring stiffness and the bending stiffness of the structure.

In a wall with a resilient mount like a sound clip or resilient channel, with freely floating edges, the motion at resonance involves the compression and extension of the “spring” – of the rubber in the clip or the metal channel.

Motion at resonance is, therefore, controlled by a combination of the air spring stiffness and the spring stiffness of the clip or channel.

While in lab tests, these type of walls often approach this “free edge” condition, in the real world edge stiffness is present, and motion at resonance is a combination of bending behavior and this idealized behavior.
In this type of wall (clip or channel), bending will occur on the opposite side, but the dominant motion is as shown.

In this type of wall (clip or channel), bending will occur on the opposite side, but the dominant motion is as shown.

The second major class of resonance is the primary structural resonance, first predicted by Lin and Garrelick in their brilliant 1977 JASA paper “sound transmission through periodically framed parallel plates”. This resonance This type of resonance is most severe when the two sides of the wall are mechanically coupled – i.e., its most severe in single rigid stud walls.

It is worth mentioning that the low frequency behavior of some walls – like flexible steel stud walls – exhibits a combination of these two behaviors. Other walls exhibit both behaviors independently. Staggered and double stud walls are examples of this – both situations can occur (and do) in those types of construction.

So with that short brief on resonance behavior in walls lets just postulate that resonance location correlations to the location of the absorption peak. Rather than present data here, we will, with each graph presented in the sections below, offer observations on where the resonance of each wall lies.

PART 2 – Do Resilient Mounts affect the absorption of walls?

To address this common question, we ran a series of tests in which the frame, drywall, and insulation were unchanged. We constructed a conventional wall, with drywall screwed directly to the framing members, and then removed the drywall in the test room, and re-mounted it onto some type of resilient decoupling device – sound clips or resilient channel.

Comparison 1 - single layer of 1/2” drywall on single wood studs.

Construction was 1/2” standard weight drywall on 2x4 wood studs, 16” OC, with R13 fiberglass in the wall cavity. Screws were 16” on center. Resilient channel was installed on the test room (for resilient channel absorption data only) 24” on center. Channel was verified in the lab to be 25 gauge.

Here we see that the absorption of the resilient channel wall is not actually better than that of the conventional wall, but it is different.

The conventional wall is considerably higher in absorption over a range centered at 125 Hz, while the RC wall is somewhat higher over a lower frequency range.

31.5 Hz data can be discounted.

These walls were tested in the August session.

The most interesting aspect of this is that the directly screwed wall shows notably higher absorption over a range centered at 125 Hz. This region is largely non-resonant in the resilient channel wall (which exhibits an MSM centered at around 85 Hz), but is the center of the primary structural resonance in the conventional wall. Observe the graph below, which shows the transmission loss of each wall.

Note that the regions where the absorption of either wall is higher than the other correlate to the shifting in resonance frequency.

Comparison 2 – single wood studs with single 5/8” drywall

The construction for this comparison was single 2x4 wood studs, with R13 fiberglass insulation and a single layer of 5/8” drywall on each side with screws 12” on center. Resilient channel was as above, and the procedure was as above – same frame, same drywall, only the channel as a variable.

Chart 2
October test session
As above, each wall is superior to the other at its primary low frequency resonance. Again, resonance location is strongly tied to absorption peak.

With a large advantage at 80 Hz (anti-resonance in the conventional wall, MSM in the RC wall) aside, no gains in absorption due to resilient mounting are noted.

The transmission loss of these test walls, showing the shifting location of the primary low frequency resonance.

Comparison 3 – A heavier wall with sound clips as the resilient mount

In this case the test was exactly as above – single 2x4 wood studs, 24” OC, with R13 fiberglass in the cavity, screws spaced 12” OC, but both walls had double 5/8” drywall on each side. The resilient channel was replaced with modern sound clips.

Chart 3
October test session

The combination of sound clips (inherently more resilient than channel, therefore yielding a lower resonance frequency) and extra mass move the resonance of the resilient wall well down – to just over 40 Hz. At lower frequencies, the clip assembly clearly has higher absorption than the conventional wall, but the conventional wall has higher absorption around its primary low frequency resonance.

The clips do not raise absorption; they shift it to a lower frequency (just as they shift the resonance to a lower frequency). The same thing is observed in the transmission loss chart below – while the clips yield the expected large gains at higher frequencies, at low frequencies they are much better at some frequencies, and much worse at others.

Resilient Mount Conclusion:

The addition of resilient mounts to a wall results in fundamental changes in the low-frequency resonance behavior. The new resonant behavior isn’t necessarily better, but it will be different.

It appears that the same can be said for the impact of resilient mounts on absorption. They change the absorption behavior – with the absorption of each wall being centered at its resonance point.

It does not appear that resilient mounts improve absorption – the just change it. A very nice positive aspect of resilient mounts may be that the ability to reasonably predict the location of resonance in a resiliently decoupled wall, such as those discussed above, grants one perhaps some level of absorption prediction. This is countered by the need to have resonance as low as possible to attain good low frequency isolation in resilient walls. Tuning the resonance to, for example, 70 Hz may have a severely negative impact on isolation when compared to a similar system with a resonance of 40 Hz.

Additional data on resilient mounts can be found throughout this document. In the next section, we will take a look at the impact of another type of product – viscoelastic-damping compounds – on absorption.

PART 3 – The Effect of Viscoelastic Damping on Absorption

Comparison 1 – Flexible damped and undamped panels on a wood stud wall.

In this comparison, the test wall was single 2x4 wood studs, 24” OC, with R13 fiberglass insulation and screws 12” on center. In one test, two layers of 1/4” drywall were laminated with a viscoelastic damping compound (Green Glue, but applied at 120 square feet per gallon). In the other test, two layers of 1/4” drywall were laminated with a conventional adhesive having very low damping properties.

The conventional adhesive was somewhat more rigid, but either of the panels were considerably more flexible than standard drywall of the same thickness.

Chart 4
October session
As always, the 31.5 Hz data can be discounted

The transmission loss plots show that resonance falls in nearly the same location. The absorption of the damped panels appears to be higher in magnitude, and broader in scope than of the undamped panel.

This is perhaps analogous to the inclusion of fiberglass or other damping material in a common panel trap – absorption typically rises and broadens from the damping effect of the fiberglass.

The absorption peak of the flexible low-damping panel doesn’t appear to be as narrow as was seen with the stiffer panels discussed above.

Comparison 2 – Viscoelastic Damping in a Conventional Wall

This comparison features two walls identical in construction with the exception of the inclusion of Green Glue in one wall. Each wall was single 2x4 wood studs, 24” on center, with R13 fiberglass insulation, and screws 12” on center.

The conventional adhesive was somewhat more rigid, but either of the panels were considerably more flexible than standard drywall of the same thickness.

Chart 5
October session
31.5Hz data can be ignored

The damped wall does not exhibit the sharp dip/peak in absorption observed in the conventional 2x4 wall. Overall absorption is slightly better, and much broader in frequency.

The transmission loss plots show the enormous gain in performance at the low frequency resonance that results from damping with Green Glue.

This gain in TL did not come with a loss of absorption, but rather with a slight improvement and broadening of the apparent absorption curve.

As will be demonstrated below, the absorption of a single wood stud wall with a single layer of conventional drywall is nearly identical to one with double drywall – so it could be said that the addition of damping and mass has not a negative, but a positive effect on wall absorption.

Comparison 3 - Heavier damped wall -vs- conventional wall from the April data set

These walls were again a single 2x4 stud frame, 24” OC, with R13 insulation. The conventional wall had a single layer of 5/8” drywall on each side, and the damped wall had 5/8” and 1/2” drywall, damped with a typical field application of Green Glue.

Chart 6
April session

Despite higher mass and enormous gains in transmission loss, the absorption of the damped wall is superior over a broad frequency range.

The transmission loss plots for these two walls.

Viscoelastic damping conclusion: The addition of viscoelastic damping appears to have some positive impact on absorption, and it also appears to broaden the range of absorption.These gains in absorption – and in-room acoustics – come without any sound isolation tradeoff or penalty whatsoever.

PART 4 – The Effect of Adding Mass

A discussion of the background of this question – does the mass of a wall affect sound absorption – is found in the appendices. For now, let’s just look at some comparisons and examples.

Comparison 1 – adding mass to a simple panel.In this comparison, the absorption of a simple panel consisting of 2 layers of 1/2” drywall bonded to each, and another assembly featuring 2 sets of such bonded panels was made. So, the absorption of 2 layers of 1/2” drywall (typical of the mass of a common interior wall) was compared to the absorption of 4 layers of 1/2” drywall (a heavier, improved interior wall).

Chart 7
August session

31.5 Hz data may be ignored

For all practical purposes, the absorption of these samples is identical, despite large differences in mass (double).

The absorption peak in the 4 layer sample at about 630 Hz probably corresponds with a resonance that resulted from the small air cavity that was created when the second bonded set of panels was applied to the wall.

The mass addition had no effect. A very large panel (8’ x 8’ in this case) is not resonant in the low-frequency region of these tests.

Comparison 2 – adding mass to a conventional wood stud wall

Here we track the absorption of a single 2x4 stud wall with R13 insulation as we move from one layer of 5/8” drywall on each side to two layers of 5/8” drywall on each side. We compare 2 total layers, 3 total layers, and 4 total layers.

Chart 8
October session

31.5 Hz data may be ignored

While the transmission loss plot shows the expected incremental gains that result from fractional increases in mass, the absorption is almost unchanged. Adding mass to this type of wall does not change the location or nature of its resonance, as resonance is almost completely controlled by the stiffness of the panels. A rigid adhesive between the layers would raise resonance frequency, and shift the absorption curve.

Again, as above, the addition of mass has basically no effect on the absorption of these walls.

On somewhat of a side note, this is as good of a time as any to repeat the basic lesson that simply adding mass to a common wood stud wall is a very inefficient way of improving performance. To improve performance on this wall, you have to do something that makes a fundamental change in the behavior of the construction.

Comparisons 3, 4, and 5 – adding mass to a resiliently decoupled wall

While we saw above that adding mass to a simple panel or a conventional 2x4 wall has no effect on absorption, what will happen when we add mass to a decoupled wall? Well, a decoupled wall (unlike the two structures discussed above) exhibits a mass-spring-mass resonance (MSM), and the frequency of that resonance will go down as we add mass. So we should probably expect at least some change… lets see what we get.

Chart 9
August session, 31.5 Hz may be disregarded

Here we have a 2x4 wood stud wall, 16” OC, with resilient channel and R13 insulation. One test has a single layer of 1/2” drywall on each side, and one test has a double layer of 1/2” drywall on each side.

The addition of mass has a negative effect on absorption except around the MSM of the heavier wall (~63 Hz). While the peak absorption does shift down, as expected, it is surprising that it never gets better than the lighter wall.

Chart 10
October session, 31.5 Hz may be disregarded

Here we show the changes in absorption resulting from adding a 2nd layer of drywall to the clip side of a sound clip wall. Both walls are 2x4 walls, with R19 insulation, and double 5/8” drywall on the non-clip side. The “1+2” wall has a single layter of 5/8” on the clip side, while the “2+2” wall has a double layer.

The same basic effect that was noted above occurs – absorption falls over a broad frequency range from the addition of mass to the resiliently decoupled wall.

Chart 11
October session, 31.5 Hz may be disregarded

Here we show the changes in absorption starting with 1 layer of drywall on each side, and adding layers until we have 2 on each side. The base wall was a resilient channel wall with 2x4 wood studs, 24” OC, and R13 fiberglass. The drywall was all 5/8”.

As expected, we see clear evidence of absorption peak shifting down, but in this case the overall loss of absorption from the addition of mass is more subtle than in the cases above. Nonetheless, absorption does fall as mass is added.

Mass Addition Conclusion:

The addition of mass to a single panel without an air cavity or a conventional single-wood-stud wall has virtually no effect on absorption in these tests.

Adding mass to resiliently decoupled walls, on the other hand, has an interesting effect. While the expected downward (in frequency) shift in apparent absorption peak occurs, it is also observed that the overall level of absorption falls. No apparent theory as to why this may be presents itself, and we are left to report what is observed without explanation.

Regardless, this data seems to strongly support the mathematically minded persons who suggest that no practical wall can transmit enough sound to represent a high level of absorption, and as such the absorption properties of walls must be related to sympathetic motions at resonance. Again, see the appendices below for a more elaborate discussion of this topic.

PART 5 - Does the absorption of walls have an inverse correlation to transmission loss?

To address this question, we simply compare the wall with the highest (by far) low frequency transmission loss observed during these tests to a variety of lower-TL walls.

The compared walls were:

A double wood stud wall with 2x4 studs 16” on center and 8” cavity depth. R13 fiberglass was present on both sides, and Green Glue with double 1/2” drywall was utilized. One side of the wall had an extra layer of Green Glue/Drywall. (Note: This is an August test, the remainder are from October).

A single wood stud wall with 2x4 studs 24” OC, R19, sound clips, and double 5/8” drywall on both sides.	A single wood stud wall with 2x4 studs 24” OC, R13, and double 5/8” drywall on both sides.
A single wood stud wall with 2x4 studs 24” OC, R13, resilient channel, and double 5/8” drywall on both sides.	A single wood stud wall with 2x4 studs 24” OC, R13, and soundboard + 5/8” drywall on both sides.

Each of these walls is roughly the same weight (soundboard wall is notably lighter), they have wildly differing transmission loss, with the damped double stud wall being – by an enormous margin – the highest performer at low frequencies, as we can see here:

The damped double stud wall is 10-25 dB better than any of the other walls over most of the subwoofer frequency range.

On the next page we’ll see if its absorption is similarly lower.

Despite the sizeable transmission loss advantages, the double stud wall does not have lower absorption than the others. In fact, look closely and you will see that at 60-80 Hz, it’s at or near the top, matching the RC wall which has it’s MSM in roughly this frequency range.

At ~50 Hz, it’s bested only by the sound clip wall whos MSM falls very close to this frequency. And at 100 Hz, the two walls with resonances at this frequency have better absorption, but the double stud wall is in the middle of the pack.

The reason for this broad absorption behavior is that this wall – a double stud wall – exhibits both an MSM and a primary structural resonance (and other activity as well). We hypothesize that this multiple-resonance behavior results in broader low frequency absorption than other wall types. A staggered wood stud wall will exhibit a similarly broad resonance behavior. For this reason, it seems reasonable to consider staggered and double studs as the ideal starting points if the absorption of walls is to be a consideration.

It seems that this might represent that rare situation where we can have our cake and eat it too. The highest TL walls are also the best absorbers.

CONCLUSION

We have taken a look at data that hopes to address some common questions about the absorption behavior of walls.

The most important conclusion is that absorption in walls is directly tied to the resonant behavior of the walls.

Beyond that, it appears that viscoelastic damping may have some benefits with respect to wall absorption, and that resilient decoupling methods such as resilient channel change the low frequency absorption of walls, but don’t necessarily improve it. They are better at some frequencies, and worse at others.

It is demonstrated that adding mass in and of itself does not result in reduced absorption in a simple panel or the common single wood stud wall, but does have a much more pronounced effect on resiliently decoupled walls.

It is demonstrated that transmission loss and absorption are not correlated.

Finally, it appears based on this work that the best starting points for sound absorption at low frequencies are staggered or double wood stud walls. These walls exhibit resonant behavior over a broader range than other wall types, and as a result appear to offer a broader range of effective absorption.

In the near future, The Green Glue Company will publish addition tests, and also a large database to help designers predict the location of resonances in various types of walls. This should prove helpful, since absorption and resonance are so closely related. Make sure to browse the appendices, where we take a look at the mass-vs-absorption debate, and try to estimate the absorption coefficient of some of these walls based on what data we have.