# Balancing report: what it should contain and how to check it

> The engineer has left, the fan is noticeably quieter, and you signed a one-line completion report: 'balancing performed, vibration within norm.' Two months later the vibration comes back, and you don't have a single number to tell whether it has returned to the old level or already gone past it. A balancing report exists for exactly this moment, not for the filing cabinet. Below we cover what sections it should contain, why the before-and-after conditions must match, and how to tell, in five minutes, a job done properly from an empty document dressed up to look like one.

**In short:** A balancing report should let another person repeat your measurement and get the same numbers. The minimum content: identification of the machine and assembly, date and location, actual RPM and operating mode, support/bearing type, measurement points and directions, the measured quantity with units and frequency band, before-balancing values (overall vibration and the 1x running-speed component — vibration at the rotation frequency, which is what imbalance produces — with phase), trial-run data, the correction applied with mass, radius and angular position, after-balancing values under the same conditions, the acceptance criterion with its source, limitations, and the technician and instrument used. If the document has no RPM, no phase, no frequency band and no trial-run data, it isn't a report — it's a note confirming that someone showed up.

Source: https://axiline.pt/en/articles/balancing-report/  
Publisher: AXILINE · Vila Nova de Gaia, Portugal · +351 931 831 229 · axilinegeral@gmail.com

## Why the document matters more than a verbal 'everything's fine'

A verbal conclusion does one job: it helps you decide what to do in the next hour. After that it's useless, because a month later no one will remember what operating mode the measurement was taken under, where the sensor was placed, or what the mm/s reading was. And the decisions you'll make about this machine later depend on exactly those details.

A report does three jobs at once. First, it proves the vibration reduction with numbers that can be checked. Second, it sets a baseline for trending — a reference point against which you'll see the rise over the following weeks and months. Third, it supplies data for a repeat visit without trial runs: it already records the influence coefficients (the machine's measured response to an added weight), the RPM, and the radii and correction planes — the rotor cross-sections where weights are attached.

There's a fourth function people usually remember too late. If the balancing result feeds into a contractual acceptance or a dispute with a contractor, the document is what the case will be argued from. The phrase 'vibration within norm' doesn't survive a single follow-up question: within what norm, against which standard, at which point, at what RPM, and in what frequency band.

There's one more thing worth knowing as a customer. A good report benefits the technician just as much as it benefits you. When vibration rises again a month later, a document with complete source data lets a single follow-up measurement separate 'the job was done badly' from 'something else in the machine has changed.' Without a report, that conversation turns into a swap of opinions.

> Ask for the report template before work starts, not after. It's easiest to agree the outline in advance: the machine's designation on your own schematic, the names of the measurement points, the operating mode, and the target vibration value. Then the document comes out the same day, instead of two weeks later by email.

## Required sections: a table of contents

Below is the full content of a field balancing report. The form can be anything: a company template, a printout from the software, a two-page table. What matters is the content, because every line answers a specific question about reproducibility.

| Section | What exactly gets recorded | Why it matters |
| --- | --- | --- |
| Machine and assembly identification | Machine type, position on your own schematic, asset or serial number, drive power, and what exactly was balanced: impeller, fan wheel, pulley, drum, motor rotor | So that a year later the document matches your own records, not just 'the fan in bay three' |
| Date, location, participants | Date and time of the work, the shop/site, and who from the customer's side was present and started the machine | A timeline for trending, and traceability — who to follow up with if figures don't match |
| Operating mode and RPM | Actual rotation speed in RPM, not the nameplate value, the load, damper or valve position, temperature, and, for a VFD-driven machine, the set frequency | Imbalance-driven vibration rises roughly as the square of RPM. A measurement without a recorded mode has nothing to be compared against |
| Support and foundation type | Rolling-element or plain (sleeve) bearings, a rigid or flexible foundation, the machine on anchor bolts or on vibration isolators | This determines which evaluation zone applies. A flexible foundation allows a higher level than a rigid one |
| Rotor layout and correction planes | Rotor mounted between bearings or overhung, number of correction planes, weight-mounting radii, number of fixed positions | Lets the job be repeated, and shows whether the correction was full (dynamic) or static only |
| Measurement points and directions | Label for each point (bearing 1, bearing 2), direction — horizontal, vertical, axial — and how the sensor was mounted: magnet or stud | A different point or a different direction gives a different number. This is the main source of measurements that can't be compared |
| Measured quantity, units, band | Vibration velocity in mm/s RMS (root-mean-square), the band used for the overall-level evaluation (10-1000 Hz as a working reference), and the 1x running-speed component reported separately | Without a stated frequency band, an overall value is meaningless — almost any number can be 'produced' |
| Analysis settings | Fmax, number of spectral lines, window function, number of averages — if spectra are attached to the report | A spectrum without its settings can't be reproduced or compared against a later one |
| Before-balancing values | For each point: overall vibration, 1x amplitude and its phase in degrees, rotation speed | This is what you're reducing. Balancing lowers the 1x component, not everything at once |
| Trial-run data | Plane, trial-weight mass, mounting radius, angular position or fixed-position number, and the resulting response in amplitude and phase | Proof that the calculation was based on the system's measured response, not a guess |
| Correction applied | For each plane: mass, radius, angle from the zero mark with the direction of reference, or the position number, the mounting method, and what happened to the trial weight. If metal was removed, where, at what drill diameter, and to what depth | Only this data lets you locate the weight on the rotor a year later and tell whether it has shifted |
| After-balancing values | Same points, same directions, same mode, same instrument: overall level, 1x, phase, RPM | A comparison only means something if the conditions match |
| Residual unbalance and criterion | Residual unbalance in g·mm or g·mm/kg, balance quality grade G, the target vibration value, and the source of the criterion: the specific standard part and edition, or the customer's own requirement | Answers the question 'is this good or bad' with a reference, not an opinion |
| Limitations and open issues | What could not be done, and why: second plane inaccessible, rotor not cleaned, unstable RPM, signs of a bearing defect, suspected shaft misalignment | The most valuable section for the customer, and the rarest one in other people's reports |
| Technician and instrument | Who performed the work and signs the document, the instrument and sensor model, serial number, software version | Accountability for the data, and a way to judge how far the conclusions can be trusted |
| Attachments | Before-and-after spectra and time waveform, polar plot, photos of the installed weights and measurement points, the run-history file | Lets another specialist double-check the work, and serves as an archive for future comparison |

> This list isn't tailored to any particular instrument. The requirement to record the point, direction, measured parameter, operating mode, date and instrument used comes from vibration-measurement documentation practice under ISO 13373-1, and the need to cite the exact standard part and edition follows from the fact that the evaluation criteria in ISO 20816 are tied to the measurement conditions. Check that the specific part actually applies to your machine.

Sources: [ISO 13373-3:2015](https://www.iso.org/standard/40840.html) · [ISO 20816-1:2016](https://www.iso.org/standard/63180.html)

## Why the before-and-after conditions must match

Measured vibration isn't a property of the rotor. It's a property of the 'rotor-bearing supports-foundation' system, at a specific point, in a specific direction, under a specific operating mode. Change any one of these, and the number will shift with no help from imbalance at all.

Moving the sensor just 10 cm along the same bearing housing can easily change the reading by a factor of one and a half. Horizontal and vertical directions on the same bearing can differ several-fold, because the system's stiffness is different along each axis. A magnet on bare, cleaned metal and a magnet on top of paint and grime give a different frequency response, and a sensor held by hand gives nothing usable at all.

RPM is even stricter. Centrifugal force from an unbalanced mass is proportional to the square of the rotation speed: a 10% increase in RPM raises the force by roughly 21%. A 'before' reading at 1450 RPM and an 'after' reading at 1380 RPM already can't be compared, and on a VFD-driven machine that kind of mismatch happens on its own.

Operating mode matters just as much. A partly closed damper changes the aerodynamic load and the fan's operating point, a warmed-up machine changes clearances and stiffness, and a different crusher load changes everything at once. That's why the mode is recorded before the first run and left untouched until the verification run.

A separate point about the instrument and the frequency band. The same actual vibration will read differently in a 10-1000 Hz band than in a 2-2000 Hz band: the wider band picks up more high-frequency energy from bearings and gear meshing. If 'before' was measured on one instrument and 'after' on another with different settings, the comparison is invalid even if both instruments are accurate.

- Measurement point. The same spot on the same bearing housing, marked with paint or a marker.
- Direction. Horizontal, vertical, or axial, stated in words in the report.
- Sensor mounting. A magnet on a cleaned pad, or a stud — the same in both measurements.
- Rotation speed. Actual, in RPM, recorded separately for each run.
- Mode and load. Damper, valve, loading, temperature, state of the process line.
- Measured quantity and band. Vibration velocity in mm/s RMS, and the same frequency band throughout.
- Instrument and channel. The same instrument, the same channel, the same analysis settings.

> Sometimes the conditions have to change for good reason — the process engineer won't allow the previous load, or the machine won't reach the same RPM. That's not a disaster. The disaster is when the change isn't written into the report, and for years you keep comparing two different states of the machine while believing you're looking at the same thing.

## Trial-run data: the section that's most often missing

A trial (calibration) weight is a temporary weight with precisely known mass, radius, and angular position. From the resulting change in amplitude and phase of the running-speed component, the software derives an influence coefficient — the measured response of this particular system to an added mass. Without a trial run there's no correction to calculate: the instrument doesn't know the rotor's mass, the stiffness of the supports, or the contribution of the frame and belt drive.

So trial-run data is the foundation of the whole calculation. Its absence from the report means one of two things: either the technician worked from influence coefficients saved from a previous balancing job — in which case that should be stated in plain words — or the correction was fitted by guesswork.

There's a simple quality check for a trial run, and it too should be visible from the numbers. The 1x amplitude should change by at least 20-30% after the trial weight is fitted, or the phase should shift by at least 20-30°. If the response is weaker than that, the weight is increased and the run repeated. A calculation based on readings that barely moved produces a near-random result, and you'll see this from the number of attempts listed in the correction section.

Trial-run data is useful to you personally, not just to the technician. The recorded mass, radius, and resulting response let another engineer recalculate the influence coefficient and check the arithmetic. And on the next visit to the same machine, saved coefficients can eliminate trial runs altogether — saving you two shutdowns out of four.

- The plane the weight was fitted to, and the run number.
- The actual weight mass in grams, weighed, not 'about ten.'
- The mounting radius in millimetres. Unbalance is mass multiplied by radius.
- The angular position or fixed-position number, which the correction angle is later measured from.
- Post-run readings: 1x amplitude and phase on both channels, rotation speed.
- A note on whether the trial weight was removed or left on the rotor as part of the correction.
- If saved influence coefficients were used, a reference to the report they came from, and the date.

> Single-plane balancing needs at least three runs: initial, trial, and verification. Two-plane balancing needs at least four. The number of runs in the report is an honest indicator of how the job went. Eight runs isn't a sign of poor work, but it is a reason to read the limitations section.

Sources: [Balanset-1A operation manual](https://vibromera.eu/balanset-1a-operation-manual/)

## What should be recorded about the correction applied

The correction section isn't written for the sake of paperwork — it's written for the person who opens the cover a year from now and needs to understand what's fitted inside. Five parameters make the entry unambiguous.

### Mass and radius

'A 25 g weight' means nothing on its own. Unbalance is defined as mass times radius, so 25 g at a 200 mm radius and 25 g at a 350 mm radius are two different effects. Both values are recorded, and ideally the radius matches the one the trial weight used.

### Angle or position number

If the correction is set by angle, the report needs three things: the angle value, the zero reference point (usually the trial-weight location or a factory mark), and the direction of measurement — with or against rotation. Without the direction, the weight can end up mirrored, and vibration will rise instead of falling. If the work used fixed positions (blades, bolt holes, spokes), the position number and mass are recorded instead, and no angle is needed at all.

### Mounting method

Bolted, welded, clamped, or a standard balance weight in a slot. Tape, putty, a magnet, or a cable tie do not count as mounting for either a trial or a permanent weight. The method matters for future inspection: a welded weight doesn't shift, while a bolted one needs its tightness checked periodically.

### Plane and mass split

The report states which correction plane each weight is fitted in. If the calculated mass was split across two adjacent fixed positions, or recalculated for a different set of available planes, that's recorded too — otherwise the next engineer won't follow the logic of the layout.

### Metal removal

When correction is done by drilling or grinding, the report records the location, diameter and depth of the holes, the number of holes, and the calculated mass removed. Metal can't be put back, and this entry is the only trace the operation leaves behind.

> A useful habit: photograph every installed weight with the position markings visible, and attach the pictures to the report. A year later, a photo answers 'has the weight shifted, or was it never there' faster than any amount of text.

## The criterion and its source: the word 'normal' means nothing without a reference

In balancing, the word 'tolerance' covers three different things, and a report must not blur them together. First: the residual running-speed component below a target value entered into the software. Second: the machine's overall vibration against the ISO evaluation zones. Third: the rotor's residual unbalance in g·mm per kilogram of rotor mass, judged against a balance quality grade G. A separate article covers these three tolerances in detail — what matters here is that the report must state which of the three criteria was applied.

A 'within tolerance' line from the software means exactly one thing: the residual 1x is below the entered target value. It does not confirm a balance quality grade, and it does not evaluate the machine's overall condition. If that phrase is the only thing in the report, you effectively have no criterion at all — the target value was typed in by a person, and it could have been set to anything.

Evaluating machine condition in mm/s RMS relies on ISO 20816, and the applicable part and edition must be stated. The parts differ by machine type, power, and RPM range, and there are exceptions. For small fans, mulchers, and belt-driven units, applicability needs to be checked case by case, not assumed from the first table found online.

The quality of the rotor's own balancing relies on ISO 21940-11 and a balance quality grade G chosen for the machine type and operating speed. That's what puts a calculated permissible residual unbalance in g·mm, alongside the actual value achieved, into the report — a claim that can actually be checked.

- The measured quantity and frequency band the criterion applies to are stated.
- The source is stated: the standard's part and edition, or the relevant clause of the customer's technical requirements.
- The machine group and foundation type are stated, if zone-based evaluation of overall vibration is used.
- For grade G, the rotor mass, RPM, the chosen grade, and the calculated tolerance in g·mm are all stated.
- The two conclusions are kept separate: 'residual 1x reached the target value' and 'machine condition was evaluated from overall vibration.'

> We work in the EU and cite ISO as a working reference, not as our own regulatory framework. If the result feeds into a contractual acceptance, agree the exact part, edition, measurement points, frequency band, operating mode, and support type before work starts. Arguing about it after the fact is expensive.

Sources: [ISO 20816-1:2016](https://www.iso.org/standard/63180.html) · [ISO 21940-11:2016](https://www.iso.org/standard/54074.html)

## How to read a report in five minutes

You don't need to be a vibration analyst to check the document. Six steps, done in the right order, are enough. The order matters more than expertise here: start not with the final number, but with the conditions it was obtained under.

1. **Check that before and after conditions match** — First, compare the RPM, points, directions, and operating mode across the two sets of measurements. If they match, keep reading. If they don't, you can skip the rest — there's no comparison to make.
2. **Look at 1x, not the overall level** — Balancing reduces the running-speed component. Its drop is what shows the job was done. Overall vibration will drop by less, because it still contains everything unrelated to imbalance — bearings, looseness, aerodynamics.
3. **Check the gap between overall and 1x after the job** — If 1x is small after balancing but the overall level has barely moved, something else in the machine is driving the vibration. That isn't a failed balancing job — it's diagnostic information, and it should be reflected in the limitations section.
4. **Find the phase** — Phase is the angle showing where in each rotation the rotor swings hardest. The 1x phase before balancing must be in the report — it's needed for comparison at the next measurement. A changed phase at a similar amplitude signals a mechanical event, such as a shifted weight or debris breaking loose. After balancing, phase often becomes unstable at low amplitude, and that's normal.
5. **Check the trial runs and the correction** — Is there a mass, radius, and response for every trial run? A mass, radius, and angle or position number for every weight fitted? If so, the job can be reproduced and checked.
6. **Read the limitations section** — The most informative text in the document. A technician who wrote 'rotor not cleaned, result will hold until the next buildup' or 'signs of bearing wear at bearing 2, recommend a follow-up measurement in a month' did an honest job.

| Point and direction | Before: overall / 1x / phase | After: overall / 1x / phase |
| --- | --- | --- |
| Bearing 1, horizontal | 8.4 mm/s / 7.9 mm/s / 62° | 1.7 mm/s / 0.9 mm/s / 214° |
| Bearing 1, vertical | 4.1 mm/s / 3.6 mm/s / 148° | 1.2 mm/s / 0.5 mm/s / 96° |
| Bearing 2, horizontal | 6.8 mm/s / 6.2 mm/s / 41° | 1.9 mm/s / 1.1 mm/s / 187° |
| Rotation speed | 1478 RPM | 1476 RPM |

> The figures in the table above are illustrative, meant to show the structure, not taken from an actual job. Pay attention to the structure itself: overall and 1x sit side by side, phase is given for every point, and RPM is recorded on its own line for both states. This layout lets the work be checked without knowing anything else about the machine.

## What should raise a red flag

A poor report rarely looks poor. It's usually a tidy document with a logo on it that simply lacks the data needed to check anything. Below are the signs that give it away.

- No rotation speed. Imbalance-driven vibration depends on RPM as a square. Without RPM, no measurement can be repeated or compared.
- No phase. You have an amplitude and no vector. Next time vibration rises, you won't be able to tell a new imbalance from a weight that has shifted.
- No frequency band and no units. 'Vibration 2.1' could be mm/s RMS, mm/s peak, microns of displacement, or g of acceleration. Those are four different statements.
- Only overall values, no 1x. Then it's impossible to tell what balancing actually reduced, and how much vibration remains from other causes.
- No trial-run data and no reference to saved influence coefficients. There's nothing to back up the calculation.
- Weight mass with no radius and no angle or position number. A year later, that correction can't be found or checked.
- Different points or directions before and after. A classic way to 'improve' the result — sometimes without any bad intent, the sensor was simply moved.
- Different RPM or operating mode before and after, with no explanation.
- No limitations section at all. A real machine always has at least one remark worth noting, even when the job went perfectly.
- The phrase 'within tolerance' with no stated criterion or source.
- No technician, instrument, or serial number. The document isn't tied to a person or to a measuring instrument.
- No spectrum or time waveform attached, even though the text draws conclusions about bearings or shaft misalignment. A conclusion without supporting data remains an opinion.

> One missing item is a reason to ask a question, not to make a scene. A technician can usually fill in the RPM or radius from their own notes the same day. What should raise a flag is different: when a request to clarify the measurement points and trial-run data gets the answer that 'this is internal information.'

## Document acceptance checklist

Go through this list before signing off on the work. It takes five minutes and settles most future questions about this machine.

### What this looks like on our side

The Balanset-1A software keeps an archive for every rotor: rotor designation, installation location, vibration and residual-unbalance tolerances, measurement date. Each job gets its own folder with charts and a report file, and the run history is time-stamped into a text log, so the number and sequence of runs never has to be reconstructed from memory. The report prints straight from the archive and can be edited in the built-in editor, and saved influence coefficients stay in the database for trim balancing (a quick re-correction without trial runs) on the next visit.

### If you need a site visit

Send us the machine designation, RPM, power, rotor mass, support type, and photos of the bearing housings with the correction planes. AXILINE's engineers, who design and manufacture the Balanset instruments and use them to balance rotors on site themselves, will agree the report format and criterion with you before the visit. That way you get the document on the day of the job, complete with before-and-after figures taken at the same points.

### If you're doing it yourself

The Balanset-1A gives you everything a proper report needs: overall vibration and the running-speed component with phase, RPM, spectrum and time waveform, single- and two-plane balancing by the influence-coefficient method, tolerance calculation by grade G, fixed-position work and drilling calculations, a polar plot, and an archive with report printing. You get methodology support from the same engineers who take these instruments out to sites themselves.

- [x] The machine and assembly are identified unambiguously: type, position on your own schematic, asset number, and exactly what was balanced.
- [x] Date, location, and participants from both sides are stated.
- [x] The actual rotation speed for each run is recorded, not the nameplate value.
- [x] The operating mode is recorded: load, damper or valve position, temperature.
- [x] The bearing-support and foundation type is stated: rigid or flexible.
- [x] Measurement points are named and tied to specific bearings, with a direction given for each.
- [x] The sensor-mounting method is stated.
- [x] The measured quantity, units, and evaluation frequency band are stated.
- [x] For each point there's an overall vibration value, a 1x amplitude, and a phase, all measured before balancing.
- [x] Data for every trial run is given: plane, mass, radius, position, resulting response.
- [x] The correction applied is described: mass, radius, angle with its reference direction or position number, mounting method, plane.
- [x] After-balancing values were taken at the same points, in the same directions, under the same mode, and with the same instrument.
- [x] The criterion and its source are stated, separately for residual 1x and for the machine-condition evaluation.
- [x] There's a limitations and open-issues section, written in plain language.
- [x] The technician, instrument, sensors, serial number, and software version are stated.
- [x] Before-and-after spectra and time waveform are attached, along with a polar plot and photos of the weights and measurement points.
- [x] The report is handed over in a format you'll be able to open a year from now, and it's filed with this specific machine's records, not dropped into a shared folder.

> Keep one file per machine and add each report to it in sequence. After three balancing jobs you'll have your own trend for that unit, and the question of 'how long does the result hold on our induced-draft fan' will stop being a question.

Sources: [Balanset-1A operation manual](https://vibromera.eu/balanset-1a-operation-manual/) · [Balanset-1A manufacturer specification](https://vibromera.eu/product/balanset-1/)

## Frequently asked questions

**Is there a mandatory balancing-report format required by the standards?**

There's no single mandatory template for field balancing, and that's fine — you agree the form with the technician. What's mandatory is the content, and that follows from the requirement of reproducibility. Vibration-measurement documentation practice under ISO 13373-1 calls for recording the measurement point and direction, the measured parameter, the machine's operating mode, the date, and the instrument used. The overall-vibration evaluation criteria in ISO 20816 and the residual-unbalance tolerances by grade G in ISO 21940-11 only work under the stated measurement conditions, so those conditions are the skeleton of the report. Check the applicable part and edition for your own machine — there are exceptions by power, RPM, and unit type.

**We were given a document with overall vibration before and after, and a signature. Is that enough?**

Sometimes it's enough for a completion sign-off, but not for operating the machine. Two overall numbers won't tell you what actually went down: balancing reduces the running-speed component, while the overall level also includes bearings, looseness, misalignment, and aerodynamics. Without 1x and phase, you won't be able to tell, next time vibration rises, whether it's a new imbalance, a weight that has shifted, or a developing bearing fault. Ask for the RPM, the points with directions, the frequency band, the 1x values with phase, and the trial-run data to be added. The technician usually has this data in the software's archive, and adding it to the report is an hour's work.

**The software says 'within tolerance.' Can the job be considered accepted?**

Only if you know which tolerance is meant. That line from the software means one thing: the residual running-speed component is below the target value the operator entered. It does not confirm a balance quality grade G for residual unbalance, and it does not evaluate the machine's condition from overall vibration. For proper acceptance, the report needs both assessments stated separately: the residual 1x achieved against the set target value, and the overall vibration against the evaluation zone, with the applicable standard part, machine group, and foundation type all stated.

**Why does the report need the weight's mounting radius if the mass is already given?**

Because unbalance is defined by mass times radius, not by mass alone. A 30 g weight at a 180 mm radius and a 30 g weight at a 340 mm radius produce an effect on the rotor that differs by almost double. Without the radius, you can't recalculate the correction if the weight is moved to a different location, can't check the calculation's arithmetic, and can't compare the achieved residual unbalance against a grade-G tolerance, which is expressed in g·mm. State both radii — the trial weight's and the correction weight's — since the influence coefficient is calculated for a specific radius.

**What if the report has no phase, and the job has already been signed off?**

Ask the technician for it: 1x amplitude and phase values for every run are usually kept in the software's archive along with the charts and run history, so they can normally be recovered. If the data is genuinely lost, take your own measurement as soon as possible, while the machine is still in good condition, and record overall vibration, 1x, phase, and RPM at the same points. That becomes your own baseline for future comparison. One honest measurement of your own is worth more than someone else's report with no phase in it.

**How do you use the report six months later, once vibration has risen?**

Take a repeat measurement strictly under the conditions from the report: the same points, the same directions, the same sensor mounting, the same mode, and the same RPM. Then compare against three patterns. Only 1x has risen while phase is stable — more likely product buildup on the impeller or uneven wear, and you have time to plan a shutdown. Phase has shifted at a similar amplitude — look for a mechanical event: a shifted weight, debris that broke loose, a chipped blade edge. The high-frequency content has risen while 1x stays calm — plan for a bearing, not a balancing job. Recorded influence coefficients will let you get away with trim balancing, without trial runs, if needed.