Choose PPV when the applicable standard evaluates peak vibration velocity. RMS describes the effective value over a defined time and band. FFT shows which frequencies make up the event, while third-octave bands organise energy or maxima into standard intervals. First define the decision, the standard, the physical quantity and the measurement point. Only then choose the result.
In short
- PPV, RMS, FFT and third-octave bands are not four versions of the same number. They answer different questions.
- PPV depends on peak velocity; RMS also depends on time and the selected window; FFT depends on the record, sampling and analysis window; third-octave bands depend on filters and the aggregation method.
- A result without unit, direction, measurement point, time and band is incomplete.
- A standard for structural damage may require a different metric than a standard for human comfort. Do not transfer one threshold between purposes.
- A platform can calculate a spectrum and third-octave bands, but the presence of a chart does not prove that the entire measurement chain complies with the standard.
On the market, you will find recorders sold under one acronym: PPV, RMS or FFT. That is convenient, because one large number looks good in a datasheet. The problem appears when you need to answer a business question: can construction work damage a neighbouring building, will people feel the vibration, which machine is generating resonance, or should the alarm respond to a short impulse.
One metric does not solve all of those tasks. A short peak can have high PPV and low RMS in a long window. Two signals with similar root mean square value can carry energy at completely different frequencies. A spectrum identifies the dominant frequency, but by itself it does not say whether the limit for a specific building has been exceeded. Third-octave bands let you compare the result with a banded criterion, but they require a correct measurement chain, filters and input quantity.
That is why the choice starts with the question, not with the device function.
Four concepts, one sentence each
PPV (peak particle velocity) is the greatest absolute value of particle vibration velocity in the considered waveform, determined according to the rules of the applicable evaluation method. In practice, you need to specify the direction, the way the components are combined, the frequency range and the event time. "PPV = 8 mm/s" without that information is incomplete.
RMS (root mean square) is the square root of the mean of the square of a signal in a defined time window and band. For a waveform x(t), it is written as:
RMS = √[(1/T) · ∫ x²(t) dt]
The square prevents positive and negative values from cancelling each other out. At the same time, the window length T, frequency weighting and the decision whether you calculate acceleration, velocity or another quantity directly change the result.
FFT (fast Fourier transform) is an algorithm that converts a discrete time waveform into a description of its frequency components. FFT is not a measure of damage potential. It is a diagnostic tool. It shows dominant frequencies, harmonics and bandwidth, provided the record, sampling, scaling and window are selected correctly.
A third-octave band is a frequency interval equal to one third of an octave, where consecutive centre frequencies have a constant ratio of about 2^(1/3). The limits of the ideal band lie around the centre frequency at a ratio of 2^(±1/6). IEC 61260-1 defines requirements for analogue and digital band-pass filters and their performance classes.
These definitions reveal the first difference: PPV and RMS are numbers calculated from a waveform, FFT is a frequency-domain representation method, and third-octave bands are a way to divide and filter frequencies.
PPV: when peak velocity is the right answer
PPV is widely used in the assessment of vibration from construction work, blasting, traffic or other sources acting on structures. DIN 4150-3 concerns the effects of vibration on structures designed primarily for static loads and directly takes peak vibration velocity into account. BS 7385-2 gives the procedure for measuring, recording and analysing groundborne vibration when assessing the potential for building damage.
You can find the practical context for national requirements and sensor placement in the guide on building vibration monitoring according to PN-B-02170.
That does not mean that every PPV value can be compared with every table. A standard may distinguish between building type, frequency, duration, transient and continuous vibration, measurement location and direction. A value calculated as the vector maximum can differ from the largest axis component. A low-pass or high-pass filter can also change the peak.
If the sensor measures acceleration and you need velocity, integration is required. That has consequences: offset and very low-frequency components can create drift, so explicit filters and unit control are needed. The label "accelerometer with PPV" does not explain the process.
PPV answers well the question of the highest instantaneous velocity in a specific event. It describes exposure time less well. Two waveforms can have the same peak: one for a fraction of a second, the other for many cycles. If the evaluation goal includes time or perceptibility, you will need another metric or additional information.
RMS: effective value always has a window and a band
RMS responds both to amplitude and to how long the signal remains within the window. It is used, among other things, in the assessment of human exposure, machine vibration and levels in frequency bands. The condition is a precise definition of the calculation method.
Illustrative example. Assume a synthetic square waveform with a value of 10 mm/s for 0.1 s and zero for the remaining 9.9 s. PPV is 10 mm/s, and RMS over the full 10 s window is 1 mm/s. If the same 10 mm/s level lasts for the full 10 s, RMS will be 10 mm/s. The example does not represent a real event; it shows the influence of time on the result.
That is why a report stating "RMS = 1 mm/s" must include at least:
- the measured quantity and unit;
- the start, end and length of the window;
- the frequency band;
- filtering and weighting;
- direction and measurement point;
- the way gaps and startup segments are handled.
RMS in the full band is not equal to the sum of RMS values from third-octave bands. For independent energy components, root-sum-square addition is used, but the detail depends on filters, band overlap and definitions. Do not add 21 results linearly just because they are in one table.
FFT: diagnosis of source and frequency, not a ready verdict
A time plot says "when", while an FFT spectrum helps answer "at which frequencies". A clear dominant peak can correspond to a machine operating frequency, passing traffic, resonance of a component or an artefact. Harmonics can reveal nonlinearity or the nature of excitation. A broad spectrum fits a short impulse better than a single narrow line.
Interpretation depends on the analysis parameters. The frequency resolution of a simple record is approximately Δf = 1/T, where T is the length of the analysed segment. A shorter record distinguishes close frequencies less effectively. When the record does not contain an integer number of periods, energy "spreads" into adjacent bins. The window function limits this leakage, but it changes amplitude and peak width, so it must be described.
FFT does not replace the time waveform. Two events with similar amplitude spectra can differ in impulse order and duration. It also does not replace the third-octave bands required by the evaluation method. A spectrum with fine lines and a third-octave filter result are different representations.
The detailed selection of sampling frequency, triggering and pre-event buffer belongs to a separate stage. It is described in the article on how to choose sampling frequency and pre-trigger. Here, the rule is enough: without a preserved, calibrated waveform, you cannot later change the analysis reliably.
Third-octave bands: a bridge between spectrum and criterion
Third-octave bands group frequencies into intervals of constant relative bandwidth. As a result, the output does not depend on a single FFT bin and can be compared with a band-based criterion. Polish SWD scales for buildings use frequency criteria in third-octave bands. In assessing the effect of vibration on people, weighted root mean square values or band results are also used, but the objective, measurement point and thresholds are different.
A third-octave filter gives a separate waveform for each band. From it, you can determine RMS or maximum. These two metrics answer different questions: RMS describes the effective value in the window, while maximum gives the highest response moment after filtering. An envelope alarm can compare each band with its own reference value and react to the worst ratio.
IEC 61260-1 defines requirements for band-pass filter characteristics and two performance classes. The fact that software shows columns labelled "third-octave bands" does not prove compliance of the meter with IEC. For a normative assessment, the entire chain matters: sensor, mounting, conditioning, sampling, filter, calibration, environmental range and uncertainty.
Selection matrix: metric × question × data × limitation
| Tool | Decision question | Required data | Main limitation |
|---|---|---|---|
| PPV | what was the highest velocity peak | calibrated velocity waveform or correct integration | does not describe exposure time well |
| RMS | what was the effective value in the window and band | full waveform, time, filter or weighting | the result changes with the window |
| FFT | where are the dominant peaks and harmonics | uniformly sampled waveform and metadata | not a stand-alone damage criterion |
| third-octave RMS | how much effective value falls within a band | third-octave filters and a defined window | requires method compliance with the criterion |
| third-octave MAX | what was the largest result in each band | filtered waveform and maximum definition rule | sensitivity to short peaks and implementation |
| time waveform | when did the impulse occur and how long did it last | raw, calibrated record | difficult to compare many events quickly |
Most of the time you need a set, not a winner. For work near a building, that may be the time waveform, the correct normative metric and third-octave bands. For machine diagnosis, it may be the waveform, RMS and FFT. For human comfort, it may be the weighted quantity and exposure time according to the applicable standard. Separating structural damage from user comfort prevents the most expensive interpretation error.
Selection procedure before purchase or configuration
1. Write the decision in one sentence
"I want to stop vibration before the criterion for the neighbouring building is exceeded" is better than "I want to measure vibration". The sentence defines the object, the source, the response time and the recipient of the result.
2. Choose the assessment document and its edition
ISO 4866 establishes rules for measurement and data processing for evaluating vibration effects on structures, but it does not give one universal limit for all objects. DIN 4150-3, BS 7385-2, PN-B-02170 or a design criterion may lead to different procedures. Record the title and edition before configuration.
3. Define the physical quantity and location
Velocity, acceleration and displacement are not interchangeable without assumptions about frequency and filtering. A point on a foundation, floor slab or machine measures a different response. Axis directions must be documented.
4. Define the calculation
For PPV, record the component or the method of combining axes. For RMS, record the window, band and weighting. For FFT, record the record length, window, scaling and unit. For third-octave bands, record the centre frequencies, filter type, RMS or MAX measure and the required chain class.
5. Preserve the evidence
An aggregated result without the waveform does not allow you to verify a peak, change the window or calculate a different metric. In the contract, specify retention of the waveform, calibration metadata, time and analysis profile version. Exporting only the alarm table is not enough.
Acceptance checklist for vibration metrics
- [ ] The evaluation objective and object are recorded in one sentence.
- [ ] The standard, edition and measurement point are stated.
- [ ] The input and output unit are unambiguous.
- [ ] The axes have physical orientation and sign.
- [ ] PPV has a defined peak definition and filtering described.
- [ ] RMS has a defined window, band and weighting.
- [ ] FFT has record length, window, scaling and resolution.
- [ ] Third-octave bands have centre frequencies, limits and the RMS/MAX method.
- [ ] The waveform and analysis metadata are retained.
- [ ] The test includes a sinusoid of known frequency, a short impulse and an out-of-band signal.
- [ ] The alarm has been compared with a reference calculation.
- [ ] The report does not declare normative compliance without confirmation of the full chain.
How it works in Inclify
For stored vibration events, Inclify shows the time waveform, the FFT spectrum and 21 third-octave bands from 1 Hz to 100 Hz. A profile with 4th-order Butterworth filters determines RMS and MAX in each band; it requires an effective sampling rate above 224 Hz and at least 10 s of pre-trigger data. Calibration is set separately for each axis.
The dynamic alarm compares RMS and MAX in active bands with a reference envelope. It reduces the results to the worst ratio, and the WARNING and ALARM levels operate with hysteresis. This makes it possible to quickly identify the event and the band that requires analysis. The full waveform and dynamic results can be reviewed, and the signal summary can be exported to XLSX.
The platform does not declare a separate PPV result or automatic compliance with ISO 4866, DIN 4150-3, BS 7385-2 or PN-B-02170. If the decision requires PPV, you need to confirm whether the recorder provides velocity, how the peak is calculated and which chain meets the requirements of the chosen method. Inclify is an analysis and alarm layer, not a certificate for the entire measurement.
Each stored event retains the analysis profile used for the calculation, which makes it easier to reproduce the RMS and MAX determination method. Reference values and alarm thresholds remain a separate, approved configuration. However, this does not extend the available 1-100 Hz range.
Limitations and most common pitfalls
The biggest pitfall is comparing a correctly calculated number with the wrong threshold. PPV on a machine is not PPV at a building foundation. Acceleration RMS without weighting does not automatically match a comfort criterion. The maximum from third-octave bands is not the same as RMS in a band. A value on the vertical axis cannot be assigned to the horizontal axis after rotating the sensor without updating the documentation.
The second pitfall is losing the waveform. If you store only PPV or a band table, after a dispute you will not be able to assess whether the result came from an impulse, electrical interference, sensor saturation or incorrect triggering. Keeping the data increases storage cost, but it determines whether re-analysis is possible.
The third pitfall is treating a standard like a single threshold table. ISO 4866 emphasises the importance of frequency, duration and amplitude of structural response. BS 7385-2 also covers the nature of the vibration and the properties of the building. Proper assessment requires competence and the full text of the standard, not a graphic found online.
FAQ
Is PPV always the best metric for assessing the risk of building damage?
No. It is used in many criteria for groundborne vibration, but the correct method depends on the country, object, source and purpose. Some procedures include frequency, direction, duration or banded results. First choose the assessment document and confirm its scope, then configure PPV.
Can acceleration be converted to velocity?
Yes, by integrating the waveform, but the result is sensitive to offset, low-frequency noise, filtering, initial conditions and calibration. A faulty process can create apparent drift or inflate PPV. The transformation must be explicit and verified with a reference signal. A simple unit change in the report is not a physical conversion. The full procedure is described in the guide on integrating acceleration to velocity, filters and drift.
Why does RMS depend on the window length?
Because the mean square is calculated over the specified time. A short impulse is diluted in a long window that contains silence, and in a short window it has a larger share. That is why only results with the same definition of time, band and weighting can be compared. The standard or procedure should define those parameters.
Do FFT spectra and third-octave bands show the same thing?
Both describe frequency, but differently. FFT splits the record into narrow bins that depend on length and sampling. Third-octave bands aggregate the response into intervals of constant relative width, usually through filters. A dominant peak is easier to diagnose in FFT, while a band criterion is applied to third-octave results.
Are 21 bands from 1 Hz to 100 Hz enough for every assessment?
No. The range must cover the requirements of the chosen standard, source and object dynamics. Some phenomena lie below 1 Hz or above 100 Hz. The sensor and sampling must also properly carry the frequency range of interest. The number of columns in the report does not replace engineering analysis of the full measurement chain range.
What should be retained after a vibration event?
The time waveform in calibrated units, exact time, axis directions, location, recording parameters, calibration, analysis profile, PPV or RMS according to the adopted definition, FFT, band results and alarm state. Also the source context and the operator action. That package allows the conclusion to be reproduced without guesswork.
Sources and further reading
- ISO 4866:2010 - Vibration of fixed structures - rules for measurement and data processing for evaluating vibration effects on structures.
- IEC 61260-1:2014 - Octave-band and fractional-octave-band filters - requirements for band-pass filters, classes and operating conditions.
- DIN 4150-3:2016-12 - Effects on structures - assessment of vibration effects on structures designed primarily for static loads.
- BS 7385-2:1993 - Guide to damage levels from groundborne vibration - procedure for measurement, recording and analysis in damage assessment.
- NIST Technical Note 334 - An Introduction to Sampled Data and Switching Logic - fundamentals of sampling, signal recovery and aliasing.
- PKN - standards search - current editions of PN-B-02170 and PN-B-02171.
What next
Take one existing vibration report and try to add to each number: quantity, unit, point, direction, time, band, filter and standard. Every blank field is an acceptance risk. Talk to the Inclify team if you want to work through a sample waveform and choose a set of views and alarms for a specific decision, instead of starting with the acronym in the device datasheet.