# Ensuring acoustic performance of windows and curtain walls

November 4, 2016

by Steve Fronek, PE
By comparing exterior noise levels with interior requirements, and considering background noise levels, a project’s required level of window or curtain wall acoustic performance can be inferred either in specific frequency bands or by using one of the single-number rating systems described in this article.

Sound pressure level and frequency
Proper acoustic design begins with an understanding of the fundamentals of sound.

Sound pressure level
Sound pressure level (SPL) is a common sound measurement typically expressed in decibels (dB). A decibel is a logarithmic ratio of the pressure of the measured sound to a reference pressure. Normal conversational speech is approximately 60 dB.

The threshold of human hearing, or the smallest audible pressure level, is usually assumed to be
20 µPa (20 x 10-6Pa). This is the reference pressure for calculating SPL, corresponding to an SPL of 0 dB.

The equation is:
SPL (dB) = 10 log10(p2/p2ref) = 20 log10(p/pref)

Where:
p = SPL in Pa (psi)

pref = reference SPL in Pa (psi) = 20 μPa (2.9 x 10−10 psi)

Sound is also characterized by frequency, which is measured in cycles per second or Hertz (Hz). Normal human hearing can sense frequencies from about 20 Hz to 20,000 Hz. This audible frequency range is often further divided into smaller octave bands (OBs) or third OBs, which are commonly denoted by their ‘center frequency,’ so the third OB centered at 31.5 Hz would correspond to a frequency range of 28.2 to 35.5 Hz. A full OB also is denoted by its center frequency. Therefore, the 31.5 Hz full octave corresponds to a range of 22.4 to 44.7 Hz.

Each full OB comprises a low-, middle-, and high-frequency third OB. The lower third OB of the 1000 Hz full OB is centered at 800 Hz, and includes the frequencies from about 708 to 891 Hz. Similarly, the middle third OB of the 1000 Hz full OB also is centered at 1000 Hz and includes the frequencies from about 891 to 1122 Hz. Finally, the higher third OB of the 1000 Hz full OB is centered at 1250 Hz and includes the frequencies from about 1122 to 1413 Hz.

Sounds in the range of 20 to 500 Hz tend to be categorized as low-frequency sounds. Sounds like rumbles and roars carry much more sound energy than high frequency squeaks and buzzes, and therefore, are more difficult to absorb. Similarly, low-frequency sounds are also heard more easily by typical human hearing than high frequency ones at the same distance from a given sound source.

Since the human ear is the most sensitive to midrange frequencies of 500 to 8000 Hz, sound-level meters usually include a filter to weight frequencies in a way that approximates human perception of loudness. This is called A-weighting; the A-weighted logarithmic sum of all such weighted frequencies is designated as dBA.

Generally, glass decreases sound transmittance at higher frequencies better than it does at lower frequencies. One should be especially cautious if acoustic expectations for windows or curtain walls dictate a certain performance level for any frequencies below 1000 Hz.

Transmission loss testing
As stated earlier, the decibel scale is logarithmic and not linear. Consequently, SPLs cannot be added or subtracted linearly. For example, some specifications call for minimum transmission losses (TL) in specific full-octave frequency bands. However, test results are reported in third OBs. To average three tested data points for checking compliance, the following formula is used:

TLOB =
10 x 10.5 – {log[(10(10-(TL1/10)) + (10(10-(TL2/10)) + (10(10-(TL3/10 ))]}

Where:
TLOB = octave band transmission loss;
TL1 = transmission loss of the lower third octave band;
TL2 = transmission loss of the center third octave band; and
TL3 = transmission loss of the upper third octave band.

Figure 1 lists the third OB groupings and the full octave groupings.

Technically speaking, lower noise levels are factored into calculations, but in practice, “total noise” is dependent only on the loudest source of many. An 80-dB sound is twice as loud as a 70-dB sound. (This is due to the exponential difference in sound pressure level (SPL) implied by the decibel (dB) scale.) Performance is governed by the weakest link—in other words, a 70-dB sound will be almost completely drowned out by an 80-dB one. An opaque wall transmitting 30 dB will make little difference to the occupants of a room in which the window transmits 40 dB. Small differences of less than 3 dB in SPL, TL, sound transmission class (STC) or outdoor/indoor transmission class (OITC) barely are perceptible to normal human hearing.

Acoustic test chambers in U.S. laboratories consist of two reverberant rooms—a source room and a receiving room. A high-transmission-loss filler wall separates these rooms. The test specimen, whether glass-only or a composite frame/glass assembly, is mounted in an opening in this filler wall. The sound-transmission loss tests are conducted in accordance with ASTM E90, Standard Test Method for Laboratory Measurement of Airborne Sound Transmission Loss of Building Partitions and Elements.

Loudspeakers in the source room generate broadband noise in third OBs, at frequencies ranging from 80 to 5000 Hz. Microphones connected to a data-collection system will measure SPLs within the receiving room, providing the measured values used to calculate TL provided by the test specimen. TL data are plotted on a logarithmic axis with respect to the third OB frequencies to create a TL curve.

When needed for project specifications, this TL curve is usually converted to a single-number rating, either STC or OITC.

Sound transmission class
Originally developed for interior wall partitions, the STC rating is determined per ASTM E413, Classification for Rating Sound Insulation. It involves fitting a pre-defined standard contour curve to the measured TL curve. STC is a single-number rating system for acoustic performance, developed primarily for typical interior noise frequencies, and is the most commonly specified measure of acoustic performance. The higher the STC, the more sound a given partition blocks.

STC is determined by a three-step process:

1. Test the glazed window assembly in a highly insulated test buck, generating a TL curve that characterizes overall performance.

2. Overlay the criterion curve best representing the TL data. This ‘best fit’ is defined as no more than 32 dBA of total ‘deficiencies’ between the criterion curve and specimen TL curve, and no more than a 8-dBA deficiency at any one third OB.

3. STC is reported as the criterion curve TL at 500 Hz.

Due to the peculiarities in this calculation methodology, there may be very little difference in tested performance between a ‘strong’ STC of 35 and a ‘weak’ STC of 37. STCs of 48 or higher are past the point of diminishing returns for window and curtain wall improvements, and can be very expensive to achieve.

Outdoor/indoor transmission class
The OITC rating was devised to more accurately quantify the amount of a reference exterior sound signal attenuated by a given partition. The reference sound approximates typical noise sources commonly occurring in urban areas. The OITC rating (referenced in ASTM E1332, Standard Classification for Determination of Outdoor/Indoor Transmission Class) is calculated from the following formula, using a logarithmic summation of individual third OB TL measurements:

OITC = 100.14 − 10log10åf10((Lf-TLf+Af)/10)

Where:
Lf = reference source spectrum level at each third octave band frequency;
Af = A-weighting adjustment at each third octave band frequency; and
TLf = specimen TL at each third octave band frequency.

There is no way to convert from OITC to STC (or vice versa), without access to the underlying TL data. OITC is predominantly a measure of low-frequency attenuation, as low-frequency sounds are generally the more prevalent in typical urban environments. In one typical test on high-performance laminated insulating glass with a makeup of 8-mm (5/16-in.) exterior, 25-mm (1-in.) air space, and 11-mm (7/16-in.) laminated inboard, 85 percent of the total sound pressure transmitted was less than or equal to 200 Hz, as derived from OITC calculations per ASTM E1332. (See Element Materials Technology’s test report, ESP017522P-5.)

Test repeatability and performance
Test-to-test and lab-to-lab results for STC and OITC of similar specimens can vary by as much as three dBA, due to inconsistent installation practices, varying temperature and humidity, size differences, and aspect ratio changes. Attachments and sealants used around the perimeter can provide unrepresentative damping. Both the specimen size and the glass-to-frame ratio also affect the sound transmission loss. Assessment of incremental improvements must be made in the context of test-to-test variation, which often ‘masks’ small changes in TL resulting from design changes. (Field test procedures for sound transmission also exist, and the field results can vary from the laboratory results by as much as five points due to installation accessories, sound flanking though adjacent building elements, and interior noise sources. Most project specifications allow for a difference between the lab and field test results. ‘Flanking’ refers to sound transmission through areas outside the glazed area being tested, and can be caused by excess air infiltration, or lower performance of spandrel areas or adjacent wall construction.)

Many laminated glass fabricators publish acoustic data on ‘glass-only’ prototypes, based on testing loosely supported lites without framing. While somewhat useful for glass-to-glass comparison purposes, it is not recommended to use such results in project specifications, as any rigidly framed system tests significantly lower.

Comparable glass-only acoustic data was available for 26 of the 80 acoustic test reports found in one manufacturer’s library of acoustic test reports.

In only a handful of cases were the glass-only results the same as the whole-window results. In typical cases, where glass-only results were better than whole window results, the difference in STC ranged from 1 to 4 dBA. The difference in OITC ranged from 1 to 5 dBA. There was no discernible pattern to the differences when comparing higher-and lower-performing systems.

The significant variance associated with this data means whole-window STC or OITC cannot be systematically calculated or predicted from published glass-only data by the use of an adjustment factor to account for the framing. Subsequent test results could vary by 5 dBA or more, resulting in the need for significantly more expensive glass to meet requirements, or the perception of costly over-design.

There are a number of relatively cost-effective means of improving acoustic performance of conventional window and curtain walls. Specifiers should identify the specific OBs needing improvement before deciding on a design strategy.

Bigger is not better
Large glass panels can vibrate at higher amplitude than their smaller counterparts, causing a dip in the TL at the natural frequency of the glass. Square lites with an aspect ratio (the ratio of height-to-width) close to 1.0 are more prone to resonance than rectangular lites with aspect ratios of 1.5 or greater. Results of previously tested glass-frame combinations should be reviewed in that context. Additional framing members added to reduce glass surface area may be sufficient to improve acoustic performance to targeted levels. The effects of size and aspect ratio are less pronounced when laminated glass is used.

Most standard product acoustic tests are run at the standard sizes cited in ASTM E1425 Standard Practice for Determining the Acoustical Performance of Windows, Doors, Skylight, and Glazed Wall Systems, which calls for:

• window test specimens neither less than 1.9 m2 (20 sf) nor more than 2.2 m2 (24 sf), with neither dimension less than 1070 mm (42 in.);
• single-door test specimens neither less than 1.8 m2 (19 sf) nor more than 2 m2 (22 sf), with neither dimension being less than 910 mm (36 in.); and
• double-door specimens neither less than 3.5 m2 (38 sf) nor more than 4 m (44 sf).

There is no category in ASTM E1425 for window wall, storefront, or curtain wall systems.

Laminated glass
Laminated glass is often a cost-effective option for improving acoustic performance. Using laminated glass minimizes ‘coincidence’—a resonant frequency exhibited by rigidly supported lites of glass, usually occurring within the frequency range tested in ASTM E90. Coincidence can be the controlling factor for STC ratings. (Resonance appears in the test results as ‘dips’ in the TL curve.)

Resonances can be seen by increased sound transmission at a specific frequency or frequency range. Resonant frequency sometimes can be shifted out of the audible range by changing the size, thickness, and aspect ratio of glass lites. The location of laminated lites (interior or exterior) within the assembly makes no significant difference in acoustic performance, nor does heat-strengthening or tempering.

Temperature also can affect the sound transmission loss of laminated glass panels. Windows and/or curtain wall systems containing laminated glass perform better in warm environments than cold ones, since the interlayer’s damping characteristics improve at higher temperatures. American Architectural Manufacturers Association (AAMA) technical information report (TIR) A1-15, Sound Control for Fenestration Products, has shown the STC and OITC ratings can change by up to five and three dBA respectively, over a glass temperature range of 15 to 32 C (60 to 90 F).

Assemblies with more than one laminated lite may introduce unexpected visual distortion, coating limitations, or other design issues. It is important to always check with glass fabricators and review samples before specifying ‘double-laminated glass.’ Additionally, thin laminated glass can be subject to rather restrictive size limits in fabrication.

Laminated glass also offers significant benefits in ultraviolet (UV) light protection, making it particularly useful in environments with furnishings or carpeting that can fade, or in rooms with sun-vulnerable items, such as a museum gallery. When properly glazed, it can offer improved impact resistance over annealed glass. However, using laminated glass does little to improve thermal performance and can limit the choice of colors, coatings, and heat treatment options available to the specifier.

Increased air space
An alternative to laminated glass involves increasing the air space of insulating glass units (IGUs) with non-standard spacers or interior access doors. In general, adding another layer of glass at the expense of air space does not help much; a 38-mm (1 ½-in.) triple IGU performs similarly to a 38-mm double IGU. Air and argon in the space of an IGU perform essentially the same. The practical limit on sealed IG air space width is 19 mm (3/4 in.) to 25 mm (1 in) depending on makeup.

Having incremental changes in air space typically offer only marginal improvement in acoustic performance. In recent Element Materials Technology side-by-side tests of IGUs employing the same unequal glass thickness at the interior and exterior lite, increasing the air space from 13 to 25 mm (1/2 to 1 in.) improved STC by 1 dBA, with no change in reported OITC (Figure 2). As noted in AAMA TIR A1-15, mass-air-mass resonance across the air space of an insulating or dual-glazed assembly is also possible, usually evident at low frequency.

Larger non-hermetically sealed air spaces typically improve both acoustic and thermal performance, as well as solar heat gain coefficient (SHGC), but can introduce the need for periodic cleaning. Interior access doors can be fitted with integral between-glass venetian blinds, which are suitable for healthcare and educational facilities. Unlike their interior, free-hanging counterparts, integral blinds arc protected from damage and dirt, and usually are fitted with occupant tilt-control and custodial raise/lower. This prevents the checkerboard appearance of blinds hanging at all levels of the building.

The window designer must pay special attention to all expected seasonal temperature and humidity conditions within these air spaces to minimize the potential for between-glass condensation. Integral blinds often require heat-strengthening of interior/exterior glass to prevent thermal stress breakage. Dual glazing in not recommended for cold-climate hospitals; one should use exterior IGU with an interior single-glazed access door instead.

When existing windows are weathertight, and ventilation is unnecessary, interior accessory windows can improve sound, energy, air and light control economically, and with a minimum of occupant disruption.

Thicker glass
In general, use of ‘unbalanced’ IGUs—where the exterior and interior lite are of different thickness—improves both STC and OITC by reducing the effect of resonance. In one recent side-by-side test of an IGU employing unequal glass thickness at the interior and exterior lite, STC performance improved by 5 dBA and OITC by 3 dBA versus a balanced, symmetrical IGU. In this test, the outboard lite simply was increased from 6 to 8 mm (1/4 to 5/16 in.), which was a very cost-effective performance improvement strategy.

Once options for unbalancing glass thickness have been exhausted, adding mass by using significantly thicker glass can be the most expensive—yet most effective—means of improving acoustic performance for windows or curtain walls. It should be considered only when previously mentioned options do not suffice. Depending on the application, thicker glass can have either a positive or negative effect on coincidence effect and, therefore, acoustic performance.

Framing systems
Beyond ensuring an airtight assembly, there is little a curtain wall or window system designer can do to change the inherent acoustic performance of a given glass/air space combination at a given size. Acoustic performance—as measured by a low-frequency-dominated OITC rating—is generally unaffected by the following:

• framing mass;
• frame cavity insulation;
• presence/absence of thermal breaks and vents; and
• glazing gasket design. (High-frequency sound absorption using between-glass acoustical foam is helpful in increasing STC only when deficiencies are predominantly in frequencies above 1000 Hz. OITC is typically unaffected.)

For similar glass makeup, glass lite size, and aspect ratio, there will be similar performance for windows and curtain walls.

Since there are no commonly used computer modeling tools for estimating acoustic performance of frame-glass combinations, both specifiers and manufacturers are forced to make product-selection judgment calls from experience and archival evidence. Due to variations, existing test reports should be accepted by the specifier as proof of compliance for previously tested frame/glass combinations, even if results or exact compositions vary to a small degree.

Compliance verification
As no window or curtain wall manufacturer could practically pre-test every possible frame/glass combination for a broad product offering, provisions should be made for job-specific acoustic testing, or for evaluation of existing test reports by acoustical consultants. The complex physics associated with multi-layer fenestration systems of different sizes, containing many materials and complex three-dimensional geometries, makes an empirical approach necessary when required performance is compared with past test results.

To make acoustic performance guarantees relative to future job-specific test results, manufacturers can be forced into expensive overdesign. Such guarantees are not in the project’s best interests. Working together early on, the design professional, acoustical consultant, and manufacturer can select or design the right product for the needs of the project.

Conclusion
When designing a quality interior environment, excluding exterior noise can become a necessity. Windows and curtain walls are often the weakest link, acoustically speaking, in envelope design—therefore, they are the focus of sound attenuation strategies. A successful, cost-effective solution to the problem of excess noise requires detailed project specifications and a thorough understanding of the physics of different glass/air space combinations.

Steve Fronek, PE, leads Wausau Window and Wall Systems’ new product development, marketing, field service, technical support, and general research. A past-president of American Architectural Manufacturers Association (AAMA), he has served on many of its committees and task groups. Fronek chairs the Apogee Enterprises’ Technical Council and is a LEED Green Associate. He can be contacted via e-mail at sfronek@wausauwindow.com[3].

Endnotes: