How Often to Check Vibration — and When That Check Should Become a Balance Job

Setting the Right Monitoring Interval

There is no universal schedule. "Monthly" is not always right. "Quarterly" is not always wrong. The correct interval depends on one thing: how fast can a fault develop from first detectable symptom to functional failure? ISO 17359 calls this the "lead time to failure."

The rule is simple: measure at intervals shorter than half the lead time to failure. If a bearing typically takes two months from first spalling to seizure, measure at least monthly. If a fan impeller accumulates enough dust to shift vibration in three weeks, check every 10 days. The half-interval rule gives you at least two data points in the fault development window — enough to see the trend and plan action before failure.

Risk-based interval selection

ISO 17359 requires a criticality and risk audit before choosing intervals, but it does not prescribe numbers. The intervals below are our field defaults, built on that framework — start there and adjust as your own trending data shows how fast faults actually develop.

These are starting points. The moment you detect a change — a vibration level creeping up, a new frequency appearing in the spectrum — increase measurement frequency immediately. A machine that was "quarterly" becomes "weekly" the moment it shows a developing fault.

Continuous vs Periodic: Two Approaches, One Goal

Most plants use both. Critical assets get online systems. Everything else gets periodic routes with a portable instrument. The key is matching the approach to the criticality and fault development speed — not choosing one method for the entire plant.

For periodic routes, the Balanset-1A has a built-in Route Inspection mode (F9 in the software): define your measurement points once, walk the round, and the software stores every reading and plots the overall RMS trend by date with ISO 10816 zone bands, plus separate trends for the 1× and 2× harmonics. Comparing each reading to the baseline and to previous rounds happens in the chart, not on paper.

Vibration Trending: What to Track and How

Collecting data without tracking changes over time is pointless. Vibration trending means comparing each reading to a baseline and to previous readings — to see whether the machine is getting better, worse, or staying the same.

Establishing a baseline

Every machine needs a reference point. Record baseline vibration under stable, documented conditions: steady speed, normal load, stable temperature. For new machines, measure after commissioning. After overhaul, allow a short run-in period (24–72 hours) before locking the baseline — vibration may shift during bedding-in as bearings seat and components settle.

Record the operating conditions with the vibration data. A vibration reading without RPM, load, and temperature context is almost useless — you can't compare a reading taken at 60% load to one taken at 100% load.

What to track: three layers

Layer 1 — Overall RMS velocity (mm/s). The simplest and fastest check. Compare to ISO 10816 zone boundaries (see table below; the ISO 10816 series has been superseded by the ISO 20816 series — the zone values are unchanged). A single number that tells you "good, acceptable, investigate, or act now." Use this for route efficiency — it takes 30 seconds per measurement point.

Layer 2 — Key frequency components. When the overall level rises, you need to know why. Track the 1× RPM component (unbalance, looseness, buildup) and the 2× RPM component (misalignment, coupling). The Balanset-1A measures vibration velocity in the 5–1000 Hz band (its sensor response rolls off above roughly 550 Hz), so its FFT spectrum resolves the 1× component, its harmonics, and bearing tones that fall below 1000 Hz. Early-stage rolling-element bearing defects develop in a much higher frequency band and normally require envelope (demodulation) analysis, which the Balanset-1A does not provide — use it to catch unbalance, misalignment, and looseness, and a dedicated bearing analyzer for early bearing diagnostics.

Layer 3 — Rate of change. The growth rate matters as much as the absolute level. A machine at 4.5 mm/s that's been stable for 12 months is different from a machine at 4.5 mm/s that was at 2.0 mm/s three weeks ago. Rapid acceleration means fast-developing fault — shorten the interval and plan action immediately. Slow linear growth supports planned maintenance at the next convenient window.

When to Rebalance: 4 Condition-Based Triggers

Balancing is not a calendar task. Do not schedule balancing "every 6 months" or "every year" without evidence. Balance when the data says so — and only when you've confirmed that unbalance is the dominant fault.

Why Vibration Returns Soon After Balancing

If vibration comes back within days or weeks after a balance job, don't rebalance again — investigate. Recurring vibration means balancing is addressing a symptom, not the root cause.

Dirty rotor. Deposits shift or flake off, destroying the balance. If you balanced a dirty impeller, the correction weights compensated for the dirt. When the dirt moves, the weights become the new imbalance source. Solution: clean to bare metal before balancing.

Thermal distortion. The rotor bows or expands unevenly when hot, shifting the mass distribution. A motor balanced cold at 20°C winding temperature may vibrate badly at 80°C. Solution: balance at operating temperature.

Loose fits. The rotor shifts on the shaft, the hub slips, or a key loosens during starts and stops. Each start changes the position slightly, so the balance changes too. Solution: fix the mechanical fit before balancing.

Resonance. Running speed near a structural natural frequency amplifies small residual imbalance. The machine appears to "need rebalancing" constantly because tiny mass changes (thermal growth, deposit shifts) get amplified. Solution: change the speed or modify the structure to move the natural frequency — see our vibration isolation guide.

Field Report: 14 Months Without a Rebalance

A food processing plant in Central Europe had four identical 30 kW centrifugal fans on a drying line, each running at 2,920 RPM. The maintenance team was rebalancing all four every 3 months — a technician came in for a full day, balanced each fan, and left. Four campaigns a year, four fans each time: 16 scheduled balance jobs per year.

We set up a monthly monitoring route using the Balanset-1A in vibrometer mode. The first three months of data showed: Fan 1 and Fan 3 were stable at 1.8–2.2 mm/s overall (Zone B — acceptable for long-term operation, no action needed). Fan 2 was climbing slowly — 2.4 → 3.1 → 3.8 mm/s — with a rising 1× component indicating unbalance from product buildup on the impeller blades. Fan 4 had a strong 2× component suggesting coupling misalignment, not unbalance at all.

Result: we balanced Fan 2 (after cleaning) and aligned Fan 4's coupling. Fans 1 and 3 were left untouched. Fourteen months later, Fans 1 and 3 still don't need balancing — they're at 2.0 and 2.3 mm/s respectively.

The savings came from stopping unnecessary work. Two fans didn't need balancing at all. One needed alignment, not balancing. Only one actually had an unbalance problem. Monthly monitoring with a portable instrument cost 30 minutes per round — the data told the team exactly which machine needed what, and when.

ISO 10816 Severity Reference

ISO 10816-3 applies to industrial machines with a rated power above 15 kW and nominal operating speeds between 120 and 15,000 RPM (the often-quoted "up to 50 MW" is the cap of its Group 1 category, not the scope of the standard). The table below shows the zone boundaries for Group 2 machines (15–300 kW) on rigid foundations. Use these as reference thresholds for your trending program. Your plant may set tighter limits based on experience.

Note: the ISO 10816 series has been superseded by the ISO 20816 series; the numerical zone boundaries below are unchanged in ISO 20816-3.

These limits apply to broadband RMS velocity measured on the bearing housings or machine structure in the 10–1000 Hz band (2–1000 Hz for machines running at or below 600 RPM). Do not compare a single narrow-band component — for example the 1× peak — with these zone boundaries; they are overall values.

These values apply to Group 2 machines (15–300 kW) on rigid foundations. For Group 1 (>300 kW) and flexible foundations, thresholds differ — consult the full standard. The key point: Zone A/B = monitor normally. Zone C = investigate and plan. Zone D = act now.

Note that the Route Inspection trend chart on the Balanset-1A draws its zone bands from the ISO 10816-1 machine classes (I–IV), not from the ISO 10816-3 group scheme in the table above — so the zone shown on the screen and the zone from this table can differ for the same reading. Pick one scheme for your plant standard and stay with it.

Frequently Asked Questions