Measuring noise: instruments and surveys

You cannot manage what you have not measured. A hearing conservation program depends on knowing which areas are noisy, which workers are exposed and by how much, which sources dominate, and whether controls are working. Measurements answer those questions, but only if they are planned for a purpose, taken with suitable instruments, and interpreted by someone who understands their limits. This lesson covers the instruments, the main types of noise survey and the practical steps that make measurements trustworthy. Lesson 5 then shows how the results are turned into a daily exposure.

Why measure?

Noise measurements are taken for different reasons, and each reason calls for a different approach.

A common error is to use one kind of measurement for a purpose it cannot serve. A single sound level meter reading at a machine is useful for a noise map, but it says little about the daily exposure of a mechanic who spends ten minutes there and the rest of the day elsewhere. Conversely, a dosimeter result tells you a worker's dose but not which machine to fix.

Screening

Before calling in instruments, a simple rule of thumb helps decide whether a full assessment is needed. If people standing about one metre apart have to raise their voices to be understood, the level is probably 85 dBA or more. If they have to shout, it is probably above 90 dBA. Other signs are ringing ears or muffled hearing at the end of a shift, and machinery known to be noisy, such as grinders, saws, impact tools, presses, compressors and chippers. Any of these signs means a proper assessment should follow.

Sound level meters

A sound level meter measures the sound pressure at its microphone and displays it in decibels with a chosen frequency weighting and time response. Modern integrating-averaging meters also calculate the equivalent continuous level, Leq, over a measurement period, the maximum and minimum levels, and the peak level. Many can record octave or one-third octave bands.

Meters are classified by accuracy under the international standard IEC 61672. Class 1 meters have tighter tolerances and are used for detailed and legal work; Class 2 meters are adequate for most occupational surveys. CSA Z107.56, Measurement of noise exposure, which several provinces reference, sets out the instrument requirements and procedures for occupational measurements.

Noise dosimeters

A noise dosimeter is a small integrating meter worn by the worker, with its microphone mounted on the shoulder near the ear. It records the A-weighted level continuously and calculates the dose, the Leq and usually a time history showing how the level changed during the shift. Because it goes wherever the worker goes, it captures exposure for mobile jobs such as maintenance, material handling and supervision that would be very hard to estimate from area readings.

Dosimeters must be set up correctly for the rule being assessed. The key settings are the criterion level, the exchange rate, the threshold below which sound is ignored, and the frequency and time weightings. A dosimeter set for an OSHA 90 dBA, 5 dB rule will give a very different dose from one set for an 85 dBA, 3 dB rule for exactly the same noise. Modern instruments can run several virtual dosimeters at once with different settings, which is useful for employers working across jurisdictions.

Calibration and quality

Every measurement session should begin and end with a field calibration check using an acoustic calibrator, a small device that produces a known level, typically 94 or 114 dB at 1,000 Hz, when fitted over the microphone. If the readings drift by more than a small tolerance, usually about 0.5 to 1 dB depending on the method, the measurements made since the last good check are suspect. Instruments and calibrators should also be sent for laboratory calibration at the interval the manufacturer and the measurement standard recommend, commonly once a year or every two years, and the certificates kept on file.

Other quality issues are just as important. The microphone should be fitted with a windscreen outdoors or near air movement, kept away from the body when hand-held, and on a dosimeter mounted on the shoulder of the worker's dominant side or the side facing the main source, not covered by clothing or a hood. Workers wearing dosimeters may bump, shout into or tap the microphone, which produces false peaks; a short conversation about the purpose of the measurement and a check of the time history reduce this problem. Electromagnetic interference from welding and radio transmitters can also cause spurious readings.

Planning an exposure assessment

A good exposure assessment follows a plan. International standard ISO 9612 and CSA Z107.56 describe three main strategies, and the choice depends on the work pattern.

Whichever strategy is used, the plan should identify similar exposure groups, workers who do the same job in the same area and are expected to have similar exposure, so that not every worker needs to be measured every time. It should cover typical days and the noisiest normal days, not just quiet ones. It should note the production rate, machines running, doors open or closed and any unusual events, because these explain differences later. And it should record shift length, because long shifts change the comparison with an 8-hour limit.

Energy averaging of readings

When several sound level readings are taken at a workstation over time, they must be averaged on an energy basis, the same way levels are added in Lesson 2. Simple arithmetic averaging understates the exposure, because it treats a reading of 100 dBA as only a little worse than 90 dBA, when it carries ten times the energy.

import math, random

random.seed(4)
# One-minute LAeq readings at a cutting station over an hour (illustrative):
# mostly 86 to 90 dBA with a few minutes of very loud cutting.
readings = [random.uniform(86, 90) for _ in range(52)] + [random.uniform(100, 104) for _ in range(8)]
random.shuffle(readings)

arith = sum(readings) / len(readings)
energy = 10 * math.log10(sum(10 ** (L / 10) for L in readings) / len(readings))
quiet = 10 * math.log10(sum(10 ** (L / 10) for L in readings if L < 95) / 52)
print(f"Arithmetic average:   {arith:.1f} dBA")
print(f"Energy average (Leq): {energy:.1f} dBA")
print(f"Leq without the 8 loud minutes: {quiet:.1f} dBA")

Arithmetic average:   89.8 dBA
Energy average (Leq): 94.2 dBA
Leq without the 8 loud minutes: 88.1 dBA

Eight minutes of loud cutting raise the hourly Leq by several decibels, and the arithmetic average misses most of that effect. The same principle explains why short, very loud tasks deserve so much attention in a survey.

Area surveys and noise maps

An area survey records sound levels across a workplace to identify noisy zones, decide where signs and mandatory protector areas are needed and show where workers should not linger. Readings are taken at the height of a standing or seated worker's ears, typically at workstations and on a grid through aisles and open areas, with the main machines running at normal production. The results are often drawn as a noise map, with contour lines or coloured zones at 80, 85, 90 and 95 dBA.

Noise maps are excellent communication tools. Workers can see at a glance where hearing protection is needed, and managers can see how much of the floor is affected. They should be dated and updated when equipment or layout changes. A map is not an exposure assessment, though: it shows levels at locations, not doses received by people.

Source measurements

To control noise, you need to know which sources and which parts of each source matter. Engineers use close measurements around a machine, switch-off tests, octave band and narrow-band frequency analysis, and sometimes sound intensity probes or acoustic cameras that show where sound radiates from. Even a simple sound level meter used carefully, combined with the source-ranking calculation in Lesson 2, will usually identify the main contributors: an air exhaust, a vibrating guard panel, a gear set, parts dropping into a metal bin.

Reporting results

A noise report should allow someone else to understand, check and repeat the measurements. At a minimum it should include the purpose, date and conditions, the instruments with serial numbers and calibration records, the settings used, the locations or workers measured and their tasks, the results with the shift length, a comparison with the applicable limit, the main contributing sources and clear recommendations. Results should be communicated to the workers measured and to the joint health and safety committee or representative.

Several provinces require records of noise exposure assessments to be kept and made available to workers. In Quebec, assessments must be repeated at least every five years where limits are exceeded, and within 30 days of changes that could increase exposure. Even where no frequency is set, reassess whenever processes, equipment, layout, shift patterns or production rates change.