Differential immission criterion

The differential criterion is a check in Italian noise-pollution law: it verifies that the noise produced by a specific source — a plant, a commercial or industrial activity — does not cause an excessive increase over the background noise, measured inside the dwelling of the affected receiver. It is an additional control on top of the absolute zone limits, and it has no direct equivalent in most European legal systems.

What it is for

The absolute limits of the municipal zoning plan (DPCM 14/11/97) say how much noise an area may contain. They say nothing, however, about how much a single new source may worsen an already quiet situation. A quiet residential area with a night-time background of 30 dB(A) stays within the absolute limits even if a plant adds noise up to 40 dB(A) — yet those 10 dB of increase are clearly audible and disturbing. The differential criterion fills exactly this gap: it limits the increase, not the absolute value.

How article 4 is built

The whole criterion sits in one article of three commas, and knowing which comma says what avoids the most common mistakes in technical reports:

The Class VI exclusion sits in comma 1, not among the exemptions: it is not a case of “negligible noise”, it is an area where the criterion does not exist. The two are often cited together and are different in nature.

The formula

For the day and night periods you compute the difference between the ambient level and the residual level:

$$\Delta L_r=L_{r,\text{amb}}-L_{r,\text{res}}$$ $$\begin{array}{rl}{L}_{r,\text{amb}}& =\text{level with the disturbing source running, inside the dwelling}\\ L_{r,\text{res}}& =\text{residual level WITHOUT the disturbing source}\end{array}$$

A clarification about the quantities, because it decides whether a later step is legitimate. The difference is between two levels, and DM 16/03/1998 provides for penalty terms to be added to the ambient level for impulsive, tonal or low-frequency components. Where those corrections are present the figure is no longer a purely energetic level: the energetic subtraction shown further down — the one that recovers the energy of the disturbing source alone — is valid on uncorrected levels, and applying it to a penalised level does not give “the source” but a hybrid with no physical meaning. If both are needed, keep two columns: the energetic levels for the decomposition, the corrected ones for comparison with the limit.

The ΔL_r difference is not explicitly defined by the decree: it is technical practice derived from DM 16/03/1998 and from reading L_r as the rating level, including the corrections for tonal (K_T), impulsive (K_I) and low-frequency (K_B) components. DPCM 14/11/97 sets the limit values, not the formula.

The limits in Art. 4 comma 1 are:

$$\begin{array}{rl}\text{day}(06\text{:}00-22\text{:}00)& :\Delta L_r\le 5\text{dB}\\ \text{night}(22\text{:}00-06\text{:}00)& :\Delta L_r\le 3\text{dB}\end{array}$$

What 3 and 5 dB actually mean

The limit is expressed on a difference of levels, which makes it look like a thin margin. In energy terms it is not. The ambient level is the sum of the source and the residual, so:

$$\Delta L_r=10{\log }_{10}\left(1+\frac{E_{\text{src}}}{E_{\text{res}}}\right)⟺L_{\text{src}}-L_{\text{res}}=10{\log }_{10}\left(10^{\Delta L_r/10}-1\right)$$

Inverting the formula tells you how loud the source may be on its own relative to the pre-existing background:

The 3 dB night limit therefore allows the source to be worth, on its own, exactly as much as all the pre-existing noise of the place: it doubles the acoustic energy of the environment. The 5 dB day limit allows more than double that. Put this way, the differential criterion is not severe at all — it is severe only relative to the absolute limits, which in an already quiet area would leave far wider margins.

This reading is useful at design stage too: if a receiver’s night-time background is 32 dB(A), the source cannot exceed 32 dB(A) at the receiver. Not “35”, not “the class limit”: 32, that is, the background itself.

Uncertainty against the limit

There is a practical problem that reports rarely confront: ΔL_r is the difference of two measurements, so its uncertainty is larger than either one. For independent readings with the same standard uncertainty u:

$$u(\Delta L_r)=u\sqrt{2}$$

With a realistic uncertainty of ±1 dB per measurement — which covers microphone positioning, the repeatability of the source’s duty cycle and the stationarity of the background — the expanded uncertainty on ΔL_r is 2.8 dB, almost the entire night limit. A measured ΔL_r of 3.5 dB does not prove an exceedance, and one of 2.5 dB does not prove compliance: both fall inside the uncertainty band. This is why a night-time differential check requires repeated measurements and a declared uncertainty budget rather than a single reading. The method for building one is in the measurement uncertainty (GUM) wiki.

The procedure, in order

An example: within by day, over by night

A ventilation plant serves a ground-floor business; the affected receiver lives on the first floor. Measurements inside the dwelling give:

• Day — ambient level (plant running) 52 dB(A), residual level (plant off) 48 dB(A): ΔL_r = 4 dB ≤ 5 → within the criterion.
• Night — ambient level 44 dB(A), residual level 40 dB(A): ΔL_r = 4 dB > 3 → over the criterion.

The same 4 dB increase respects the daytime limit and breaches the night-time one: a direct consequence of the stricter night threshold (3 dB against 5). Neither ambient level falls below the exemption thresholds with windows open (50 dB(A) day, 40 dB(A) night), so the criterion applies in both periods.

The night-time case is worth reading the other way round as well: at ΔL_r = 4 dB the plant alone is worth 41.8 dB(A) against a 40 dB(A) background. To come back within 3 dB it must drop to 40 dB(A), that is −1.8 dB: a small cut in absolute terms, which on a fan often means redesigning the discharge section. This is the asymmetry typical of the criterion — you exceed it by a little, and the fix is not proportionally little.

The exemptions

Comma 2 of Art. 4 provides that the criterion does not apply when the ambient noise is already so low that the differential is not meaningful. The thresholds distinguish measurements with windows open and closed (the decree says “lower than”, i.e. a strict comparison):

$$\begin{array}{rl}\text{windows open:}& \text{noise}<50\text{dB(A) day}\\ & \text{noise}<40\text{dB(A) night}\\ \text{windows closed:}& \text{ambient noise}<35\text{dB(A) day}\\ & \text{ambient noise}<25\text{dB(A) night}\end{array}$$

The windows-closed thresholds are very low — 25 dB(A) at night is a room where you hear your own breathing — and in practice they only trigger in well-insulated buildings in quiet areas. The measurement configuration is therefore not a detail to be recorded at will: it selects which pair of thresholds applies, and thus whether the criterion exists at all for that case.

When the criterion does not apply

Beyond the negligible-noise exemptions of comma 2, the differential criterion does not apply:

• to transport infrastructure (roads, railways), governed by specific decrees such as DPR 142/2004 — see the Italian road noise limits wiki;
• to activities and behaviours unrelated to production, commercial or professional needs;
• to building services shared by the building, for disturbance perceived within the same building where they are installed;
• in Class VI areas, exclusively industrial (comma 1).

The tool handles these cases as preliminary applicability checks.

Terms

Ambient level — noise present in the receiving environment, produced by all sources together, including the disturbing one. Residual level — noise present when the disturbing source is switched off. Acoustic class — class of the municipal zoning plan (I–VI) that sets the absolute limits and, in Class VI, the inapplicability of the criterion.

Limits of the tool

• The differential criterion is an additional condition on top of the absolute zone limits; it does not replace them.
• Identifying the reference and measurement time intervals requires analysing the source’s duty cycle: the tool assumes the levels supplied are already expressed as corrected L_r values.
• The K_T, K_I and K_B corrections must be applied before the comparison, during the processing of the measurement data.
• The comparison is deterministic: the tool does not propagate the uncertainty of the two measurements. The table above is there to do it by hand, because on a 3 dB limit a result without a declared uncertainty is not defensible.
• The residual level presupposes being able to switch the source off. When that is impossible — a business that cannot stop — the residual has to be estimated in idle time windows or by comparison with an unexposed point, and the assumption must be declared: it is the weakest link in many assessments.
• The checks cover the most common cases; borderline situations (mixed use, multiple sources) require assessment by a qualified engineer.

A demonstrative tool: it does not replace the measurements and expert reports of a qualified acoustics engineer.

Normative references

• L. 26/10/1995 n. 447 — Framework law on noise pollution.
• DPCM 14/11/1997 — Determination of the limit values of sound sources. Art. 4: comma 1 differential limit values and Class VI exclusion, comma 2 exemptions for negligible noise, comma 3 cases of non-application. Annex A: reference time intervals and the K_T, K_I, K_B corrections.
• DM 16/03/1998 — Techniques for detecting and measuring noise pollution, including the correction terms for impulsive, tonal and low-frequency noise.
• UNI/TR 11326:2009 and UNI/TS 11326-2:2015 — Assessment of uncertainty in acoustic measurements and calculations, and criteria for comparison against a limit value. They are the reference for turning a numerical comparison into a decision rule: where specific regulation exists, it prevails over the general method.
• DPR 142/2004 — Road traffic noise: acoustic buffer strips, in place of the differential criterion.

The tool is available as a runnable instrument at /en/tools/italian-differential-criterion/.