machinery-condition-monitoring

name	machinery-condition-monitoring
description	Machinery condition monitoring — vibration analysis (FFT, envelope, cepstrum), bearing fault frequencies (BPFO/BPFI/BSF/FTF), ISO 10816 severity criteria, oil analysis, thermography, ultrasound, motor current analysis (MCSA), alarm levels, online monitoring systems, predictive maintenance integration, ISO 13373.
metadata	{"priority":7,"promptSignals":{"phrases":["condition monitoring","vibration monitoring","bearing fault detection","predictive maintenance","machinery monitoring","ISO 10816 vibration"],"minScore":3}}

Machinery Condition Monitoring — Complete Skill

Vibration Analysis

Signal Acquisition

Transducers: Accelerometer (piezoelectric): most common; 10 Hz–20 kHz; measures g; mount with stud (best) or magnet Velocity transducer: 10 Hz–1,000 Hz; integrates internally; legacy systems; 4–20 mA output Proximity probe (eddy current): shaft relative motion; 0–10 kHz; shaft rider; journal bearings Microphone/hydrophone: airborne/waterborne noise; useful for flow-induced vibration

Signal conditioning: ICP (Integrated Circuit Piezoelectric): constant current signal conditioning; long cable capable; 2–20 mA Charge amplifier: for charge output accelerometers; high impedance input Anti-aliasing filter: low-pass at < f_sample/2 (Nyquist) before digitizer

Sampling: Fmax = highest frequency of interest (bearing faults: up to 20× shaft speed × number of rolling elements) f_sample ≥ 2.56 × Fmax (typical; 2.56 × oversampling) Resolution: Δf = f_sample / N_FFT; N_FFT = number of samples; typical 1,024–4,096 samples

FFT and Spectrum Analysis

Vibration signature: 1× (1 per revolution): unbalance dominant; always present 2×: misalignment (parallel or angular); looseness Fractional ×: rubs (partial arc contact); 0.5×, 0.333× n× (high multiples): gear mesh, blade passing, structural resonance

FFT analysis: A(f) = FFT{a(t)} [converts time-domain acceleration to frequency-domain amplitude spectrum] Units: g RMS vs. Hz; or velocity mm/s RMS vs. Hz; or displacement μm pp vs. Hz

Windowing (reduce spectral leakage): Hanning window: best for general rotating machinery (reduces side lobes) Flat-top window: more accurate amplitude measurement for single-frequency calibration Rectangular: only for transients (impulse testing, impact hammer)

Averaging: Linear averaging: N spectra summed/N → reduces random noise by √N; N = 4–32 typical Synchronous time averaging: trigger on 1× → keeps shaft-frequency events; cancels random noise → best for gear faults with speed reference

Bearing Fault Frequencies

Rolling element bearing fault frequencies from geometry: BPFO (Ball Pass Frequency Outer Race): BPFO = (N_balls/2) × f_shaft × (1 - d_ball/d_pitch × cosα) [Hz; d_ball = ball diameter; d_pitch = pitch diameter; α = contact angle]

BPFI (Ball Pass Frequency Inner Race): BPFI = (N_balls/2) × f_shaft × (1 + d_ball/d_pitch × cosα)

BSF (Ball Spin Frequency): BSF = (d_pitch/2d_ball) × f_shaft × (1 - (d_ball/d_pitch)^2 × cos²α)

FTF (Fundamental Train Frequency = cage speed): FTF = (f_shaft/2) × (1 - d_ball/d_pitch × cosα)

Typical range: BPFO = 3–7 × f_shaft; BPFI = 5–9 × f_shaft; BSF = 1.5–3 × f_shaft Fault identification: look for peaks at BPFO, BPFI, BSF, FTF ± harmonics; sidebands at f_shaft intervals around bearing frequency = inner race defect

Example bearing (6205: N=9, d_ball=7.938mm, d_pitch=38.5mm, α=0°, f_shaft=25 Hz): BPFO = 4.5 × 25 × (1 - 0.206) = 4.5 × 25 × 0.794 = 89.3 Hz BPFI = 4.5 × 25 × (1 + 0.206) = 135.7 Hz

Envelope Analysis (Demodulation)

Purpose: detect early bearing faults from modulated high-frequency impulses Process:

1. Bandpass filter signal around high-frequency carrier (resonance frequency 5–50 kHz)
2. Rectify → demodulate (envelope signal)
3. FFT of envelope signal → shows bearing fault frequencies clearly Advantage: works at very early fault stage when FFT vibration spectrum shows no visible peak

KURTOSIS (impulsiveness indicator): K = E[(x - μ)⁴] / σ⁴ [dimensionless; normal distribution K=3; impulsive bearing fault K >> 3] K > 4: possible fault; K > 6: confirmed fault developing; K > 10: advanced fault Kurtosis deteriorates at advanced fault (many defects → signal less impulsive → K decreases); use with trending

Cepstrum Analysis

Cepstrum: log(|FFT(x(t))|²) inverse-FFT → converts spectrum to cepstrum domain Quefrency axis [s]: reciprocal of frequency → peaks at 1/f correspond to families of harmonics at f Application: gear fault detection (harmonics of gear mesh frequency visible in cepstrum); bearing fault families

ISO 10816 Vibration Severity

ISO 10816 (Mechanical Vibration — Evaluation of Machine Vibration)

Velocity RMS (10 Hz–1 kHz) as overall severity indicator: ISO 10816-3: industrial machines with rated power > 15 kW and nominal speed 120–15,000 rpm

Zone	Velocity RMS [mm/s]	Condition
A	≤ 2.3	New machine, acceptable
B	2.3–4.5	Acceptable for long-term
C	4.5–11.2	Unsatisfactory; short-term only
D	> 11.2	Dangerous; risk of damage

ISO 10816-1 (general): machines up to 15 kW; different severity boundaries by machine class (I–IV) ISO 20816-3 (2016): updated replacement for ISO 10816-3; adds rolling element bearing absolute limits

Alert and danger levels: Alert: zone B/C boundary; investigate; increase monitoring frequency Danger: zone C/D boundary; take out of service if safety critical

Oil Analysis

Ferrography and Particle Count

Oil sample collection: Live sampling (running machine): representative of circulating oil contamination Drain port: biased sample (concentrated particles); avoid Standard volume: 100–250 mL per ASTM D4057

ICP (Inductively Coupled Plasma) spectroscopy — ASTM D5185: Measures dissolved metals [ppm]: Fe, Cu, Cr, Ni, Al, Pb, Sn Fe increase: steel wear (bearings, gears, cylinders) Cu increase: bronze bearings, thrust washers Pb/Sn increase: babbitt bearing overlay wear Cr increase: chrome-plated ring or liner wear

Trend alarming (ppm change rate): Absolute: Fe > 100 ppm, Cu > 50 ppm → investigate Rate: Fe increasing > 25 ppm per interval → accelerating wear → action

PQ Index (Particle Quantifier): Measures total ferromagnetic particle mass; sensitive to large particles (> 10 μm) PQ > 100: advanced wear (compared to baseline when machine was new)

Water content (Karl Fischer, ASTM D1533): < 100 ppm: acceptable; > 1,000 ppm: free water risk; > 0.1%: emulsification → oil change

Viscosity (ASTM D445): Change > ±10% from new oil: oxidation (increase) or fuel dilution (decrease) or wrong oil (either)

Vibration Correlation with Oil Analysis

Early fault: oil analysis shows Fe increase, vibration normal → oil sample more sensitive for inner machinery Advanced fault: oil analysis plateaus (particles filter out), vibration amplitude grows Use both: oil analysis first-warning, vibration confirmation; never replace one with the other

Thermography (Infrared)

Application: overheating bearings, electrical connections, insulation failures, heat exchanger blockage

Camera specifications: Thermal resolution: 0.05–0.1°C NETD (noise equivalent temperature difference) Spatial resolution: 640×480 detector (standard); higher for distant targets Lens: standard 25° lens; telephoto for access-restricted areas

Alarm criteria: Bearing: > 25°C above baseline → investigate; > 40°C → immediate action Electrical connection: > 30°C ΔT → poor contact; > 50°C ΔT → high risk

Standards: ISO 18434-1 (thermographic inspection of machinery); IEC 60068-2-1 for cold environments

Ultrasound (Acoustic Emission)

Ultrasound range: 20 kHz–100 kHz; beyond human hearing Application: detect early-stage bearing defects (ultrasound precedes vibration fault signature by weeks) Detect: steam traps, compressed air leaks, electrical arcing (ultrasound emission)

Bearing detection: Healthy bearing: background hiss level (dBμV); trending comparison Defect: increased dBμV + characteristic patterns at BPFO, BPFI frequencies

Methods: Contact: threaded transducer on housing; frequency: 30–40 kHz Airborne: microphone-type; for leak detection; 40 kHz

Motor Current Signature Analysis (MCSA)

Principle: broken rotor bars, eccentricity, bearing faults modulate motor current at specific sideband frequencies Current spectrum: measure current with clamp-on CT → FFT → analyze sidebands

Broken rotor bar sidebands: f_broken_bar = f_line ± 2 × s × f_line [s = slip frequency; typically 0.01–0.05; f_line = 50 or 60 Hz] Amplitude > -40 dBc (40 dB below fundamental): investigate; > -30 dBc: likely broken bar

Eccentricity sidebands: f_ecc = f_line ± k × f_shaft [k = 1, 2, 3...]

Bearing fault in motor (from current): f_bearing_current = f_line ± n × f_bearing [n = 1, 2; f_bearing = BPFO/BPFI] Less sensitive than vibration but non-contact; useful for inaccessible motors

Online Monitoring Systems

Architecture

Continuous online monitoring: permanent sensors + data acquisition + rule-based alarms Periodic monitoring: route-based data collector; manual rounds

Sensor networks: Wired (4–20 mA or industrial Ethernet): reliable; high cost for large plants Wireless (IEEE 802.11 or ISA100.11a): flexible; battery life 1–5 years; latency acceptable for vibration Fieldbus (Profibus PA, HART): process integration; moderate bandwidth

Edge processing: Preprocessing at sensor node: FFT, RMS, peak; reduces bandwidth; filter pre-process data before cloud Alert on deviation: send alarm event; stream raw data only when alarmed

CMMS Integration (Computerized Maintenance Management System): Work order generation from alarm → tracks repair; links to history; cost tracking SAP PM, Maximo, IBM Maximo: typical enterprise CMMS platforms

Alarm Philosophy

Two-level alarm: Alert: amplitude > 1.5× baseline; investigate; increase monitoring interval Danger/Shutdown: amplitude > 2.5× baseline; stop machine; inspect Alert should not be set to shutdown (nuisance trips → alarm silencing → missed real alarms)

ISO Standards for Condition Monitoring

Standard	Scope
ISO 10816-1/-3	Vibration severity criteria for machines
ISO 20816-1	Updated machinery vibration standard (2016)
ISO 13373-1/-2/-3	Condition monitoring and diagnostics of machines — vibration
ISO 17359	General guidelines for condition monitoring
ISO 15243	Rolling bearings — failure modes (identification guide)
ASTM D5185	ICP spectroscopy for oil metals
ASTM D4057	Oil sampling from pipelines and tanks
ISO 18434-1	Thermographic condition monitoring

Output

Provide: machine type (pump/compressor/motor/gearbox), rated speed [rpm] and power [kW], bearing type and fault frequencies (BPFO, BPFI, BSF, FTF [Hz]), vibration overall level [mm/s RMS] and ISO 10816 zone (A/B/C/D), spectrum peaks identified (1×, 2×, gear mesh, bearing fault [Hz] and amplitude [mm/s or g]), envelope analysis result (fault frequency identified? amplitude?), kurtosis [dimensionless] and interpretation, oil analysis (Fe, Cu, Cr ppm; water ppm; viscosity change %), thermography result (ΔT above baseline [°C]), alert/danger levels set (velocity [mm/s RMS]), monitoring type (online/route-based; interval), CMMS integration, and applicable standard (ISO 10816-3, ISO 13373, ISO 17359).