Project

Rotating Machinery Vibration Commissioning and Reliability Baseline Project

Rotating machinery vibration project for tachometer reference, sensor locations, speed sweep, resonance screen, FFT setup, baselines, alarms, risk review, and release evidence.

Branch: Mechanical Engineering
Content: Project
Updated: Jun 23, 2026
Revision: v1.0.0 · reviewed

This project builds a commissioning package for rotating-machinery vibration measurement and reliability baselining. The goal is to release a machine into service with defensible evidence: sensor locations, tachometer reference, speed map, sampling plan, resonance screen, baseline spectra, alarm logic, risk review, and follow-up actions.

The example is a variable-speed industrial fan skid, but the workflow transfers to pumps, blowers, compressors, conveyors, marine auxiliaries, propulsion support equipment, test rigs, and production machines. The project does not replace site standards, machine-specific acceptance criteria, API or ISO rules, OEM limits, or safety procedures. It shows how an engineer can assemble a coherent baseline package instead of saving an unexplained spectrum screenshot.

Project Objective

Produce a vibration commissioning and reliability baseline dossier that includes:

machine boundary, operating envelope, and measurement locations;
tachometer or encoder reference plan;
sensor selection, mounting, calibration, range, and bandwidth checks;
speed and order map for 1x, 2x, blade-passing, gear-mesh, and known structural modes;
speed sweep or controlled operating-point survey;
FFT setup and anti-aliasing requirements;
baseline vibration values with units, axes, and operating state;
resonance and isolation screening calculations;
initial alarm and follow-up logic tied to risk;
release decision, limitations, and action list.

The deliverable is a controlled engineering record. A machine is not “commissioned” because vibration was measured once. It is commissioned when the evidence proves what was measured, when, under which operating conditions, with which instrument setup, and how the result supports release or further work.

Project Scenario

An industrial fan skid has been installed after impeller cleaning, belt replacement, and alignment. The fan is driven by a VFD and must enter service after a reliability review.

Item	Value
normal operating speed	$1785\ \text{rpm}$
controlled speed range	$1200$ to $1800\ \text{rpm}$
fan blades	12
motor rating	$90\ \text{kW}$
supported skid mass	$1600\ \text{kg}$
installed vertical mount stiffness	$12.8\ \text{MN/m}$
estimated damping ratio	$\zeta=0.08$
measured structural mode	$34\ \text{Hz}$
site action level at bearing housing	$7.1\ \text{mm/s RMS}$
site trip or immediate-review level	$11.0\ \text{mm/s RMS}$
measured normal-speed radial vibration	$5.2\ \text{mm/s RMS}$
measured axial vibration	$1.4\ \text{mm/s RMS}$
measurement bandwidth for commissioning spectra	$0$ to $1000\ \text{Hz}$

The numbers are simplified but realistic enough to build the project. Replace them with site-specific machine class, foundation, bearing type, process load, and standard limits when using the workflow in practice.

Deliverable Structure

The final dossier should have these sections.

Section	Required evidence
machine definition	asset tag, drawing reference, driver, driven equipment, coupling or belt, bearing locations, operating envelope
measurement plan	sensor type, serial number, calibration status, mounting method, axis convention, tachometer location
speed map	calculated order frequencies and known mode frequencies
test procedure	safe speed steps, dwell time, process condition, VFD mode, interlocks, stop criteria
measured baseline	vibration values, spectra, phase if available, bearing temperature, motor current, flow or load condition
calculations	natural frequency, separation margin, transmissibility screen, sampling settings, trend thresholds
risk review	likely failure modes, RPN or equivalent priority logic, unresolved risks
release decision	accepted, accepted with monitoring, hold for correction, or reject for redesign

Step 1: Define Measurement Points

Use repeatable names. A common fan-skid layout is:

Point	Location	Axis
M1H	motor drive-end bearing	horizontal radial
M1V	motor drive-end bearing	vertical radial
M1A	motor drive-end bearing	axial
F1H	fan inboard bearing	horizontal radial
F1V	fan inboard bearing	vertical radial
F2H	fan outboard bearing	horizontal radial
F2V	fan outboard bearing	vertical radial
BASE-V	skid base near isolator	vertical

Record whether the sensor is stud-mounted, magnet-mounted, adhesive-mounted, or hand-held. Mounting stiffness and surface preparation affect high-frequency response. For commissioning evidence, stud or adhesive mounting is preferred where practical and safe.

Step 2: Build the Speed and Order Map

At normal speed:

\displaystyle f_{1x}=\frac{1785}{60}=29.75\ \text{Hz}

The second order is:

f_{2x}=2(29.75)=59.5\ \text{Hz}

Blade-passing frequency is:

f_{BPF}=12(29.75)=357\ \text{Hz}

At the low end of the VFD range:

\displaystyle f_{1x,low}=\frac{1200}{60}=20.0\ \text{Hz}

At the high end:

\displaystyle f_{1x,high}=\frac{1800}{60}=30.0\ \text{Hz}

The 1x component therefore sweeps from $20.0$ to $30.0\ \text{Hz}$ , while blade passing sweeps from:

12(20.0)=240\ \text{Hz}

to:

12(30.0)=360\ \text{Hz}

Engineering comment: the measured structural mode at $34\ \text{Hz}$ is close to the top of the speed range. The commissioning plan must include a controlled speed sweep, not only a normal-speed measurement.

Step 3: Screen Natural Frequency and Transmissibility

For the simplified mounted skid:

\displaystyle f_n=\frac{1}{2\pi}\sqrt{\frac{k}{m}}

with $k=12.8\times10^6\ \text{N/m}$ and $m=1600\ \text{kg}$ :

\displaystyle f_n=\frac{1}{2\pi}\sqrt{\frac{12.8\times10^6}{1600}}=14.2\ \text{Hz}

The normal-speed forcing ratio is:

\displaystyle r=\frac{15.0}{14.2}=1.06

for a $900\ \text{rpm}$ operating condition, or:

\displaystyle r=\frac{29.75}{14.2}=2.10

for the $1785\ \text{rpm}$ normal-speed condition.

For force transmissibility:

\displaystyle T_R=\frac{\sqrt{1+(2\zeta r)^2}}{\sqrt{(1-r^2)^2+(2\zeta r)^2}}

At $1785\ \text{rpm}$ with $\zeta=0.08$ :

\displaystyle T_R=\frac{\sqrt{1+(2(0.08)(2.10))^2}}{\sqrt{(1-2.10^2)^2+(2(0.08)(2.10))^2}}=0.31

The simplified isolation model predicts attenuation at normal speed. It also predicts amplification near $900\ \text{rpm}$ , where $r$ is close to 1. This matters if the machine starts, stops, ramps, or operates for long periods near that speed. The project should therefore include ramp observations and a minimum dwell-time restriction near resonance if measurements confirm a high response.

Step 4: Check Separation from a Measured Structural Mode

The structural mode is at $34\ \text{Hz}$ , and normal 1x speed is $29.75\ \text{Hz}$ :

\displaystyle M_{sep}=\frac{|34-29.75|}{29.75}=0.143=14.3\%

This is not a universal pass or fail. It is a risk flag. If the mode is lightly damped and participates strongly at the bearing housings or base, 14.3 percent separation may be too small. The test plan should include vibration measurement at several speeds near the upper operating range and a note that permanent operation near any response peak requires engineering review.

Step 5: Select Sampling and FFT Settings

The commissioning spectrum must cover up to $1000\ \text{Hz}$ .

Nyquist requires:

f_s>2(1000)=2000\ \text{Hz}

Choose:

f_s=2560\ \text{Hz}

For $0.5\ \text{Hz}$ frequency resolution:

\displaystyle T_{record}=\frac{1}{\Delta f}=\frac{1}{0.5}=2.0\ \text{s}

The sample count is:

N=f_sT_{record}=2560(2.0)=5120

At normal speed, the number of revolutions in the record is:

N_{rev}=29.75(2.0)=59.5\ \text{rev}

Engineering comment: this setup is reasonable for steady-state commissioning spectra. It is not enough for fast run-up order tracking unless the tachometer signal is recorded and the analysis follows shaft order rather than fixed frequency bins.

Step 6: Convert Baseline Velocity to Displacement

The measured normal-speed radial vibration is $5.2\ \text{mm/s RMS}$ at 1x.

For a sinusoidal component:

v_{peak}=\sqrt{2}v_{RMS}=\sqrt{2}(5.2)=7.35\ \text{mm/s}

\omega=2\pi(29.75)=186.9\ \text{rad/s}

Displacement amplitude is:

\displaystyle X=\frac{v_{peak}}{\omega}=\frac{7.35}{186.9}=0.0393\ \text{mm}

The synchronous displacement amplitude is about $0.039\ \text{mm}$ . This helps compare casing motion with runout, clearances, and shaft-relative probes. It should not be applied to broadband vibration without using the spectrum because different frequencies convert differently.

Step 7: Build Initial Alarm Logic

Use site standards, OEM limits, and machine class first. If the project must define temporary commissioning thresholds, make them traceable and conservative.

For this example:

Condition	Rule
release baseline	all bearing points below $7.1\ \text{mm/s RMS}$ , no unexplained narrowband peak, no rising trend during dwell
action review	any bearing point from $7.1$ to $11.0\ \text{mm/s RMS}$ or dominant peak near a known mode
hold release	any bearing point above $11.0\ \text{mm/s RMS}$ , unstable phase, abnormal bearing temperature, rub, looseness, or unsafe process condition
follow-up	repeat measurements after 24 to 72 operating hours and after belt tension or alignment changes

Trend ratio from baseline:

\displaystyle R_{trend}=\frac{x_{current}}{x_{baseline}}

If the baseline is $5.2\ \text{mm/s RMS}$ and a later measurement is $6.4\ \text{mm/s RMS}$ :

\displaystyle R_{trend}=\frac{6.4}{5.2}=1.23

The machine may still be below the action level, but a 23 percent increase is worth reviewing if it is repeatable at the same speed, load, sensor location, and bandwidth. Trend alarms should not compare different operating states.

Step 8: Risk Priority Review

Use a simple RPN only as a prioritization aid.

Before baseline commissioning:

Failure mode	Severity	Occurrence	Detection	RPN
residual imbalance damages bearing	7	4	5	140
operation near structural mode amplifies base response	6	3	6	108
sensor setup misses blade-passing component	5	3	6	90

After the tachometer-based baseline and speed sweep:

Failure mode	Severity	Occurrence	Detection	RPN
residual imbalance damages bearing	7	3	2	42
operation near structural mode amplifies base response	6	2	3	36
sensor setup misses blade-passing component	5	2	2	20

The improvement comes mainly from better detection and controlled operating evidence. RPN is not proof of reliability. The release still depends on measured vibration, phase stability, bearing temperature, process condition, mechanical inspection, and follow-up trend data.

Step 9: Produce the Release Matrix

A concise release matrix should look like this.

Requirement	Evidence	Status
tachometer reference installed	optical tach on shaft keyway mark, stable pulse through speed sweep	pass
spectrum bandwidth adequate	$f_s=2560\ \text{Hz}$ , $0$ to $1000\ \text{Hz}$ commissioning spectrum, anti-alias filter enabled	pass
normal-speed vibration below action	max measured bearing radial $5.2\ \text{mm/s RMS}$ versus $7.1\ \text{mm/s RMS}$ action level	pass
axial vibration credible	$1.4\ \text{mm/s RMS}$ , no dominant axial 1x or 2x growth	pass
structural mode reviewed	$34\ \text{Hz}$ mode, 14.3 percent separation from normal 1x, speed sweep required in upper range	conditional
isolation screen documented	normal-speed $T_R\approx0.31$ , low-speed resonance risk noted near $900\ \text{rpm}$	conditional
follow-up measurement planned	repeat after 24 to 72 operating hours and after belt retension	pass

Recommended release decision:

Accepted with monitoring. Normal-speed vibration is below the site action level, but upper-range speed sweep evidence and low-speed resonance dwell restrictions must be retained in the commissioning record.

This is a better engineering decision than a simple “pass” because it separates current acceptance from unresolved operating-envelope risk.

Field Procedure

Use this sequence during the site test.

Confirm guarding, lockout rules, access, communication, and emergency stop authority.
Verify asset tag, driver, belt or coupling condition, alignment record, base bolts, flexible connections, lubrication, and bearing temperature before run.
Install accelerometers or velocity probes at defined locations and axes.
Install tachometer or encoder reference and confirm one pulse per revolution.
Record zero-speed noise and sensor bias.
Run at low speed and check for rubbing, looseness, abnormal current, abnormal temperature, or unsafe vibration.
Step through the speed range with stable dwell at each point.
Record spectra, overall values, phase if available, motor current, speed, flow or load condition, and bearing temperature.
Stop if vibration crosses the hold criterion, phase becomes unstable with high amplitude, or personnel safety is affected.
Save raw time waveform, spectrum, metadata, and final report in the maintenance or engineering record system.

Final Dossier Template

The delivered report should contain:

one-page release summary with decision and unresolved conditions;
machine boundary and operating envelope;
measurement point diagram or table;
instrument calibration list and sensor mounting notes;
speed map and calculated order frequencies;
sampling and FFT settings;
speed sweep table;
baseline spectra and overall vibration table;
resonance, isolation, and trend calculations;
RPN or risk review table;
corrective actions, monitoring interval, and owner;
approval record.

Acceptance Limits and Caveats

Do not use this project as a universal vibration standard. Acceptance depends on machine class, bearing type, foundation, speed, power, mounting, safety consequence, process service, site history, and governing standards.

Important limitations:

a clean commissioning spectrum does not prove long-term bearing life;
low casing vibration does not prove low shaft-relative motion;
a velocity alarm may miss high-frequency bearing or gear defects;
broadband values can hide narrowband forcing mechanisms;
measurements at one operating point do not cover the full VFD envelope;
a baseline is only useful if later measurements repeat the same setup;
vibration release does not replace inspection of guards, belts, couplings, lubrication, fasteners, foundations, duct loads, pipe loads, and electrical drive settings.

Project Outcome

At completion, the engineering team should have a defensible vibration baseline, not just a startup note. The package should answer:

what was measured;
where it was measured;
which operating state it represents;
which calculated risks were checked;
which limits were used;
what still requires monitoring;
why the machine was released, held, or corrected.

That is the level of evidence needed for rotating machinery that must remain reliable after commissioning.

REF

Disciplines