As an Amazon Associate, Akermin earns from qualifying purchases. Product links on this page are affiliate links.
Because what harms hearing and what disturbs sleep is the total sound energy received over a period, not the level at any one instant. Two decibel figures can be identical at the moment you glance at the display and describe completely different exposures once duration is accounted for. The time averaged level, usually written as an equivalent continuous level, solves this by converting every instant back into energy, averaging that energy across the whole period, and expressing the result as the steady level that would have delivered the same total. It is a single number that carries both loudness and duration, which is precisely what a rule needs and precisely what an instantaneous reading cannot provide.
A single instant tells you nothing about exposure
Picture two workshops. In the first, a machine runs continuously at eighty five decibels for eight hours. In the second, a press operates at one hundred decibels for twelve minutes and the room is otherwise near silent. Walk into either with a meter at the wrong moment and you will record a figure that misrepresents the day entirely. Walk into the second during the press operation and the reading suggests a severe problem; walk in ten minutes later and it suggests none at all.
The energy arithmetic says those two situations are far closer than they look. Sound energy is proportional to the time spent at a level, so a short burst at a high level can carry as much total energy as a long period at a moderate one. A rule that set a limit on instantaneous level would have to ignore this, and would end up either banning brief loud events that cause little cumulative exposure or permitting all day exposure at levels just under the threshold.
Environmental noise has the same structure with different numbers. A road is quiet between vehicles and loud during each pass, a railway is silent apart from trains, and a condenser unit runs continuously at a low level. Judging any of these on an instantaneous reading produces an answer that depends on when you happened to be standing there, which is why outdoor limits are written around periods such as a day, an evening and a night rather than around a moment.
How Leq adds up energy rather than averaging the display
The equivalent continuous level, written as Leq, is defined as the steady level that would deliver the same sound energy over the stated period as the varying level actually did. Forming it takes three steps. Each measured level is converted from decibels back into a relative energy by raising ten to the power of the level divided by ten. Those energies are averaged over the period, weighted by how long each one lasted. The average is converted back into decibels by taking ten times its logarithm to base ten.
The conversion step is what makes the result behave sensibly, and it is also what makes a naive average wrong. Because the scale is logarithmic, a level ten decibels higher than another represents ten times the energy, and twenty decibels higher represents a hundred times. Averaging the displayed numbers treats a ten decibel difference as though it were a small one, which systematically understates any period containing loud episodes.
The period has to be stated with the figure, because Leq is meaningless without it. An Leq over eight hours, an Leq over one hour and an Leq over sixteen hours are three different quantities, and regulations name the period explicitly for that reason. A notation such as an eight hour equivalent level is a complete statement; the bare word average is not.
Exchange rates, and why 3 dB halves the allowed time
Once exposure is expressed as energy, the trade between level and duration follows automatically. Three decibels is a doubling of energy, so a level three decibels higher delivers the permitted total in half the time. This is called an equal energy exchange rate and it is the basis of most occupational noise rules outside the United States, where a five decibel exchange rate is still used in some regulations. Against a criterion of eighty five decibels over eight hours, the equal energy rule produces this ladder.
- Eighty five decibels for eight hours
- Eighty eight decibels for four hours
- Ninety one decibels for two hours
- Ninety four decibels for one hour
- Ninety seven decibels for thirty minutes
- One hundred decibels for fifteen minutes
Read down that list and the practical lesson is that small sounding increments are not small. A machine that measures six decibels above another is delivering four times the energy, and it consumes the allowance four times as fast. It also means that reducing a level by a few decibels, which rarely sounds dramatic to the ear, multiplies the time somebody can safely spend there.
The same arithmetic explains why the loudest part of a day usually dominates the whole of it. Energy adds, so quiet periods contribute almost nothing once a loud period is present. Spending effort on the quiet phases of a process is wasted; the entire result is set by the noisy minutes.
Worked example: an eight hour shift built from four noisy periods
Take a shift divided into four blocks and work the energy average through to a single figure. Each level is converted into relative energy, weighted by its duration, summed and converted back.
| Period | Duration | Level in decibels | Relative energy contributed | Share of the total energy |
|---|---|---|---|---|
| Machining | 2 hours | 95 | 6.32 units | 87 percent |
| Assembly | 1 hour | 88 | 0.63 units | 9 percent |
| Inspection | 3 hours | 80 | 0.30 units | 4 percent |
| Paperwork | 2 hours | 70 | 0.02 units | Under 1 percent |
| Whole shift | 8 hours | Equivalent level 89.6 | 7.27 units | 100 percent |
The eight hour equivalent level comes out at 89.6 decibels, which is well above the eighty five decibel criterion even though five of the eight hours were spent at or below eighty. Two hours of machining supply about seven eighths of the total energy, and the two hours of paperwork are statistically invisible. If the machining noise were reduced by five decibels, the shift average would drop to roughly eighty five and a half, which is a larger improvement than eliminating every other task in the day.
That is the argument for logging rather than spot checking. The DANOPLUS decibel meter data logger states storage for up to forty three thousand records with export to a spreadsheet file over a computer connection, a measuring range of thirty to one hundred and thirty decibels, a stated accuracy of plus or minus one and a half decibels, A and C frequency weighting with adjustable fast and slow response, and a three colour alarm indicator. A record of that density can be reprocessed into an average for any period you choose after the fact.
What to log so an average can be defended
An average is only as good as the record behind it, and the record has to answer the questions a sceptical reader will ask. Log the level at a fixed interval rather than noting figures by hand, so the sampling is even and the raw data survives. Log the whole period, or two clearly representative portions of it with the activity in each noted, rather than a convenient window. Note the frequency weighting and the time weighting used throughout, since mixing settings mid record invalidates the average.
Keep the maximum alongside the average, because the two protect against different things and a regulation may contain limits on both. An instrument that retains extremes is useful here even without logging: the Extech 407730-NIST states recording of maximum and minimum values over time, two decibel accuracy with a tenth of a decibel resolution and A plus C weighting. Note also the position, the distance from each source and the orientation of the microphone, and whether a windshield was fitted outdoors.
Finally, be explicit about what the instrument itself computed. Some meters display an equivalent level directly and some only show the instantaneous reading, leaving the averaging to be done from logged data. The RISEPRO HT-80A states a range of thirty to one hundred and thirty decibels, a frequency range of thirty one and a half hertz to four kilohertz, an accuracy of plus or minus one and a half decibels, A weighting, a fast time weighting of one hundred and twenty five milliseconds, and maximum and minimum capture; whether it computes an equivalent continuous level is not stated in the listing bullets. If a rule asks for an averaged quantity, that capability is the specification to settle first. Background on the instrument and the quantity it reports is in what a sound level meter is, the logging and averaging end of the market is surveyed in the noise meter overview and the sound level meter overview, and the order in which to settle range, class and averaging is set out in how to choose a sound level meter.
Frequently Asked Questions
What is the difference between an energy average and an average of the readings?
A decibel is a logarithmic quantity, so adding decibels and dividing by the count is arithmetically meaningless. An energy average converts each level back into the sound energy it represents, averages those energies over the period, and converts the result back into decibels. The two methods give very different answers whenever levels vary. A period split between ninety five and seventy decibels has an energy average close to ninety two and an arithmetic mean of the displayed numbers of eighty two and a half, which would understate the exposure dramatically.
Why is three decibels such a big step?
Because the decibel scale is logarithmic and three decibels corresponds to a doubling of sound energy. Two identical machines running together produce three decibels more than one of them, not twice the number. Under an equal energy rule the allowed exposure time therefore halves for every three decibel increase, which is why a level that seems only slightly higher can cut the permitted duration from a full shift to half of one. The jump sounds modest and the energy behind it is not.
Does the meter have to compute the average itself?
Not necessarily, though it is far more convenient when it does. An instrument that logs levels at a regular interval lets the average be computed afterwards from the stored data, which has the advantage of keeping the raw record available for scrutiny. An instrument that computes the average internally gives an immediate answer and may not retain what it was built from. The combination of both is ideal, and where only one is available, logged raw data is the more defensible choice because it can always be reprocessed.
What is the difference between an energy average and a maximum reading?
They answer unrelated questions and both appear in regulations for that reason. The energy average describes accumulated exposure over a period and is the quantity most occupational and environmental limits are written around. A maximum, or a peak, describes the single loudest event and exists because very loud impulses can cause immediate damage that an average would smooth away entirely. A short hammer strike barely moves an eight hour average while easily exceeding a peak limit.
How long must a measurement run to represent a shift?
Long enough to include every distinct phase of activity in proportion to its real duration. If a job cycles through loud and quiet tasks, a sample that covers two full cycles will usually represent it, while twenty minutes that happen to fall inside the loud phase will not. The practical test is to measure two separate representative periods and compare: if they agree closely the sampling is adequate, and if they do not, the activity is too variable to be captured by sampling and needs continuous logging instead.