# Vibration in buildings > Vibration in buildings: UNI 9614:2017 for disturbance to people, UNI 9916:2014 for effects on the building, Legislative Decree 81/08 for worker exposure. Quantities, weightings, measurement chain and validity conditions. Published: 2026-08-26 Practice: acustica Standard: Measurement and assessment of vibration in buildings Page: --- When a client says "there is vibration", they have not yet said which problem they have. "Vibration in buildings" is not one question: it is **four different questions**, each with a different standard, a different quantity and a different threshold. Conflating them is the most frequent mistake in technical reports on this subject, and it produces correct numbers that answer the wrong question. ## The four questions | Question | Reference | What is assessed | |---|---|---| | Are the people inside disturbed? | **[UNI 9614:2017](https://store.uni.com/uni-9614-2017)** | Disturbance to occupants | | Is the building being damaged? | **[UNI 9916:2014](https://store.uni.com/uni-9916-2014)** | Structural response and architectural integrity | | Are workers overexposed? | **[Legislative Decree 81/08, Title VIII Ch. III](https://www.normattiva.it/uri-res/N2Ls?urn:nir:stato:decreto.legislativo:2008-04-09;81)** | Occupational health and safety | | Does the sensitive equipment work? | Severity criteria (VC curves) | Environment for instruments and production | The four do not substitute for one another, and none follows from another. A level that does not damage the building can be thoroughly disturbing for the people living in it: the threshold of human perception is orders of magnitude below the threshold of damage. And a level harmless to people can be fatal to an electron microscope or a coordinate measuring machine. UNI 9614:2017 says so itself: it expressly excludes from its scope seismic vibration, structural damage, occupational hygiene and damage to sensitive equipment or instruments. It is a standard that answers **one** question, and it declares as much. ## The Italian regulatory gap, and what to put in its place One point has to come before any number: **in Italy there is no statutory limit for vibration in residential premises**. The framework law on noise pollution, Law 447/95, concerns noise. DPCM 14/11/1997 sets limit values for sound sources. Neither deals with vibration introduced into a building. What exists is **UNI 9614**, which is a voluntary technical standard. It is not binding in itself: it becomes binding when a contract invokes it, when a local regulation adopts it, or when a court takes it as the technical criterion of reference in assessing an emission under Article 844 of the Civil Code. The operational consequence is precise: **the assessment criterion must be stated in the report**, not left implicit. A report that presents a measured value and compares it to a threshold without saying where that threshold comes from, and on what basis it applies to the case, is contestable before the number is even discussed. ## Disturbance to people: UNI 9614:2017 The standard establishes how to measure vibration introduced into buildings — from internal sources such as building services and lifts, or external ones such as road traffic, railways and construction sites — and the criteria for assessing disturbance to occupants. It has been in force since 7 September 2017 and refers to **ISO 2631-2:2003** for measurement and evaluation methods. ### What changed with the 2017 edition The 2017 edition is not a cosmetic update of its predecessor: it introduces a different assessment approach, taken from the Norwegian standard **NS 8176**. Anyone who worked with the 1990s edition therefore cannot carry its practice over: the quantity the conclusion rests on changes, and so does the way events are treated. This has a practical consequence worth isolating: **a comparison with historic measurements is not automatic**. If a 2010 report has to be re-read, the edition it was produced under must be stated — exactly as for the transition between the two editions of ISO 9613-2 in acoustic calculation. ### The frequency range, and its lower limit The assessment covers the band in which the human body responds to vibration, which for buildings is conventionally **1–80 Hz**. UNI 9614 explicitly excludes vibration below the lower band limit of the one-third octave centred at 1 Hz. The lower limit has a practical consequence. Very slow oscillation — that of a tall building under wind, for instance — is outside this standard's perimeter: it is perceived, it can disturb, but it must be addressed with other tools. If the measurement shows energy below 1 Hz, declaring it and declaring it out of scope is more useful than folding it into an index that does not contemplate it. ### The weighting, and why it cannot be ignored The human body does not respond equally at all frequencies. The same acceleration, at 4 Hz and at 60 Hz, produces two different sensations: low frequencies excite resonances of the organs and trunk, high ones are attenuated in the tissue. Disturbance is therefore assessed on **frequency-weighted acceleration**, that is, filtered with a curve reproducing that sensitivity. The weighting also depends on **direction**: the response along the vertical axis (foot-to-head, for a standing person) differs from that along the horizontal axes, and the curves differ accordingly. Two consequences for the measurement follow: - a **triaxial accelerometer** is needed, or three uniaxial ones oriented; - the axes must be **in phase with each other**, because composing the three components is a vector operation (see the measurement chain section). A single vertical-axis measurement is not wrong in itself, but it holds only if the horizontal components have been verified to be negligible — and that verification requires measuring them anyway. ### Repeated events: why a single r.m.s. value is not enough Road traffic, railways and construction sites do not produce stationary vibration: they produce **events**. A train passes in twenty seconds, then for eight minutes nothing passes. An r.m.s. value averaged over the whole hour dilutes the event in the silence around it and returns a low number, which does not describe the experience of the people living there. This is why the modern approach to assessing repeated events does not use a simple average but a descriptor representing the set of events — typically a statistical maximum, such as a high percentile of the distribution of the individual event maxima. It is the direction the NS 8176 framework invoked by UNI 9614:2017 takes. A related family of descriptors, defined in ISO 2631-1, is the one based on the **vibration dose**, integrating the fourth power of weighted acceleration over time: $$ \mathrm{VDV} = \left[\int_{0}^{T} a_w^4(t)\,\mathrm{d}t\right]^{1/4} $$ The exponent 4 is not arbitrary in the sense of random: it weights short, intense events far more than a quadratic mean would, which is the intended behaviour when describing disturbance produced by occasional peaks. The price is that VDV grows with observation time, so **a VDV without its reference period means nothing**: the interval it was computed over must always be stated. ## Damage to the building: UNI 9916:2014 When the question is whether the building is being damaged, the reference is a different one: **UNI 9916:2014**, in force since 16 January 2014, which provides guidance on the choice of measurement methods, data treatment and evaluation of vibration phenomena, assessing their effects on buildings in terms of structural response and architectural integrity. Three substantive differences from disturbance assessment: - **The quantity changes.** For damage, the descriptor used in this domain is typically vibration **velocity**, and in its peak form, not weighted acceleration: damage is related to strain, and velocity is a better indicator of it. - **Weighting does not apply.** Weighting curves reproduce the sensitivity of the human body, which has nothing to do with masonry. - **The measurement point changes.** Disturbance is assessed where people are, typically on the floor slab at mid-span, where amplitude is greatest. Damage is assessed on the structure, at the base or on the suspect element, where the real stress matters. Thresholds depend on the **building category** — industrial, residential, of historic interest — and on **frequency**, because at equal velocity a slow and a rapid excitation produce different strains. A historic building has appreciably lower thresholds: it is the case where getting the distinction between the two questions wrong costs most, because the two answers can differ by an order of magnitude. ## Worker exposure: Legislative Decree 81/08 This is the third question, and it has a statutory answer, not a voluntary technical one. **Legislative Decree 81/08, Title VIII Chapter III** governs worker exposure to vibration with action values and limit values expressed as eight-hour normalised daily exposure, A(8): | System | Action value | Limit value | |---|---|---| | Hand-arm (HAV) | 2.5 m/s² | 5 m/s² | | Whole body (WBV) | 0.5 m/s² | 1.0 m/s² | Above the action value, obligations on information, training and health surveillance are triggered; the limit value may not be exceeded. Measurements refer to ISO 5349 for the hand-arm system and ISO 2631-1 for whole body. These figures concern **a worker's exposure to a tool or a vehicle**, not the disturbance of someone living in a building. They are not transferable to the residential setting in either direction: they are several orders of magnitude more permissive than disturbance thresholds, because they describe a risk of health damage from occupational exposure, not annoyance. ## Sensitive equipment The fourth question has no Italian standard. The severity of the vibration environment for instruments — microscopy, lithography, measuring machines, analytical balances — is expressed with severity criteria given as r.m.s. velocity per one-third octave band, the so-called **VC curves**, which descend decade by decade to values of the order of micrometres per second. It is mentioned here for one reason only: if a building houses sensitive instruments, that is almost always the governing question, and it has far lower thresholds than the other three. Designing for human comfort guarantees nothing for a microscope. ## The measurement chain, which is where the numbers are formed The standards say what to assess. The value being assessed, however, comes out of an acquisition chain, and its defects do not show in the result: they show as a plausible number. ### The accelerometer and the IEPE interface Piezoelectric accelerometers for these measurements typically use the **IEPE** interface: two wires carry both the supply, as a constant current of a few milliamps, and the signal, as a voltage riding on a bias voltage. The advantage is cabling; the constraint is that the front-end must provide the excitation, decouple the DC component and — this matters in the field — **detect cable break and short circuit**, because an eight-hour measurement with a cable pulled at the second minute is eight hours wasted. ### Simultaneous sampling, and why triaxial analysis depends on it A triaxial accelerometer gives three signals. If the three channels are read in turn through a multiplexer, each axis is sampled at a different instant: the vector composition of the three components is then computed on data that do not belong to the same moment. The skew is small in absolute terms, but its effect **grows with frequency**: at 80 Hz, the upper limit of the band of interest, an inter-channel delay that is negligible at 1 Hz becomes a non-negligible fraction of a period. **Simultaneous** channel sampling is therefore the condition for the axes to stay in phase with each other, and it is a requirement of the acquisition hardware, not something corrected in post-processing. It is the reason the four-channel front-end for IEPE sensors we develop samples its channels simultaneously rather than in turn: [the board is on the electronics page](/en/hardware-firmware/), with 24-bit conversion over 20 kHz of bandwidth and on-board excitation. ### Resolution and dynamic range Vibration in buildings has a wide dynamic range: between the night-time floor of a quiet building and the passage of a heavy vehicle there are tens of decibels. A 24-bit conversion is not there to read the twentieth bit — the noise of the sensor and front-end arrives first — but to avoid having to change gain between night and transit, which in an unattended twenty-four-hour measurement cannot be done. ### Uncertainty, and what has to be declared A value without its uncertainty cannot be compared to a threshold. The contributions to put in the budget are at least: - **calibration** of the sensor-conditioner chain, with its date and its certificate; - **mounting** of the sensor: the mechanical coupling is the most underestimated contribution, and a wax or magnetic fixing has a cut-off frequency of its own that must be known; - **position**, which on a floor slab changes the result far from marginally between mid-span and support; - **temporal representativeness**, which for an event-driven phenomenon is often the dominant contribution: measuring two trains or twenty is not the same thing. The criterion is that of the [GUM guide](/en/wiki/measurement-uncertainty/): declare the budget, not only the result. And comparing against a threshold also requires a declared **decision rule**: with what margin conformity or non-conformity is concluded, given that uncertainty exists. ## Validity conditions of this page - The normative references are verified at source for **title, edition and scope**. The standards' threshold tables are not reproduced here: they are content of the standards themselves, which must be consulted in the edition in force. The Legislative Decree 81/08 values are statutory and are reported. - The page describes **how the four questions are framed**; it does not provide a ready-made conformity criterion: that depends on the edition of the applicable standard, on the intended use, on the period, and on the contractual or judicial criterion invoked in the specific case. - An assessment that carries a signature requires calibrated instrumentation, a justified sensor position, documented temporal representativeness and the responsibility of a qualified technician. No document and no program replace that responsibility. ## References - **UNI 9614:2017** — Measurement of vibration in buildings and criteria for assessing disturbance. In force since 7 September 2017. Refers to ISO 2631-2:2003 for measurement and evaluation methods; assessment approach taken from NS 8176. Excludes seismic vibration, structural damage, occupational hygiene and damage to equipment. [UNI record](https://store.uni.com/uni-9614-2017) - **UNI 9916:2014** — Criteria for the measurement and evaluation of the effects of vibration on buildings. In force since 16 January 2014. [UNI record](https://store.uni.com/uni-9916-2014) - **ISO 2631-1** — Evaluation of human exposure to whole-body vibration: general requirements, frequency weightings, descriptors including the vibration dose. - **ISO 2631-2** — Vibration in buildings, 1–80 Hz band. - **ISO 5349-1 / -2** — Hand-transmitted vibration: measurement and practical guidance. - **Legislative Decree 81 of 9 April 2008, Title VIII Chapter III** — Protection of workers from the risks of exposure to mechanical vibration. Action and limit values on A(8). - **Italian Civil Code, Article 844** — Emissions: the normal tolerability criterion, which is the frame in which a vibration measurement enters civil litigation.