Route-Based Monitoring: The Foundation of Vibration Programs
Continuous online monitoring gets all the attention. Wireless sensors and IIoT platforms dominate the conference presentations. But for most industrial plants, route-based vibration data collection with a portable analyzer remains the most practical and cost-effective approach to vibration monitoring. A trained technician with a good analyzer can cover 30-50 machines in an 8-hour shift, collecting high-quality data that supports informed maintenance decisions.
The key word is “high-quality.” A poorly designed route produces inconsistent data that wastes analysis time and erodes confidence in the program. Getting the route setup right from the beginning avoids years of frustration.
Selecting Equipment for the Route
Start with your criticality ranking. Every machine in the plant doesn’t need vibration monitoring. As a rule of thumb, route-based monitoring makes sense for:
- Rotating equipment rated 5 HP and above on critical and important systems
- Equipment with a history of bearing or alignment failures
- Equipment without installed spares where failure stops production
- Equipment where bearing replacement requires significant downtime (specialty bearings, difficult access)
Small motors driving non-critical loads, equipment with installed and operational standby units, and machines that are inexpensive to replace can be excluded from routine vibration routes. Run-to-failure or simple operator checks (touch, listen) are more cost-effective for these assets.
For a typical manufacturing plant, you might end up with 100-300 machines on vibration routes. A food processing plant of similar size might have 150-400 machines due to the large number of conveyor drives, blower motors, and refrigeration compressors.
Defining Measurement Points
Measurement point selection directly determines what you can and cannot detect. Every bearing location on a monitored machine should have defined measurement points.
Number of Points Per Bearing
At minimum, take one radial reading (horizontal or vertical) at each bearing. For critical equipment, take three readings: horizontal (H), vertical (V), and axial (A). The horizontal reading is most sensitive to misalignment forces. The vertical reading is most sensitive to mechanical looseness. The axial reading catches angular misalignment, thrust bearing problems, and coupling issues.
Sensor Mounting
Mount the accelerometer as close to the bearing centerline as possible, on a surface that provides a solid transmission path. Avoid mounting on thin sheet metal covers, plastic guards, or flexible housing features — these act as mechanical filters that attenuate the high-frequency bearing defect signals you need most.
Mounting method matters. Stud mounting (threaded stud in a tapped hole) provides the highest frequency response — usable to 10,000 Hz or higher depending on the accelerometer. Flat magnets provide good response to about 2,000-3,000 Hz. Two-pole magnets on curved surfaces reduce the usable range further. Handheld probe tips are the worst option — inconsistent contact pressure introduces variability, and frequency response drops off above 1,000-1,500 Hz.
For route-based collection, a rare-earth flat magnet is the practical choice for most applications. It balances good frequency response with speed of attachment. Invest in quality magnets — cheap magnets with weak pull force don’t seat consistently.
Marking Measurement Points
This step is simple and critical. Mark every measurement point with a permanent identifier — paint pen, stamped tag, or engraved plate. The mark should indicate direction (H, V, A) and bearing position (DE for drive end, NDE for non-drive end). Without permanent marks, different technicians will place sensors in different locations, and trending becomes meaningless.
Prepare a flat, clean surface at each measurement point. A dime-sized area ground smooth with a right-angle grinder gives the magnet a consistent, solid seat. Some plants install permanent magnet pads — a thin steel disc epoxied to the bearing housing — for the ultimate in repeatability.
Data Collector Configuration
Frequency Range (Fmax)
Set Fmax high enough to capture bearing defect frequencies and their harmonics. A general starting point is:
Fmax = BPFO x 3 (to capture the third harmonic of the outer race defect frequency)
For most industrial equipment running at 1,800 or 3,600 RPM, Fmax of 1,000-2,000 Hz covers the relevant range. For high-speed equipment (turbo compressors, spindles), Fmax may need to be 5,000-10,000 Hz.
Don’t set Fmax higher than necessary. Higher Fmax means either longer data collection time (more lines of resolution required) or reduced frequency resolution (same number of lines spread across a wider range). Both have drawbacks.
Lines of Resolution
More lines of resolution means better ability to distinguish closely spaced frequency components. For general route collection, 1,600 lines provides adequate resolution. For detailed analysis of gearboxes or machines with closely spaced bearing frequencies, 3,200 or 6,400 lines may be needed.
Remember: collection time is proportional to lines of resolution and inversely proportional to Fmax. A 1,600-line spectrum with Fmax of 1,000 Hz takes about 1.6 seconds of data. At 6,400 lines, that’s 6.4 seconds per measurement. Multiply by 3-4 points per machine and 50 machines per route — the difference adds up.
Measurement Parameters
Collect both velocity and acceleration spectra at each point, or use a dual-channel measurement. Velocity (mm/s or in/s) best represents overall machine condition for frequencies between 10-1,000 Hz. Acceleration (g’s) is more sensitive to high-frequency bearing defect signals above 1,000 Hz.
Also collect a time waveform at each point. The time waveform reveals transient events, impacts, and modulation patterns that the averaged spectrum can miss. Set time waveform length to capture at least 6-10 shaft revolutions.
Averaging
Use 4-6 linear averages for route data. Averaging reduces random noise and improves the consistency of the spectrum. Fewer averages are faster but noisier. More averages are cleaner but take longer. Four averages is a good balance for most applications.
Walking the Route: Best Practices
- Consistency wins. Collect data at the same operating condition each time. A pump running at full flow produces a different vibration signature than the same pump running at half flow. Document operating conditions when they vary.
- Don’t rush. Give the sensor time to stabilize before triggering collection. Verify the data looks reasonable on the collector screen before moving on. Recollect if something looks wrong — a dropped sensor, a strange spike from an impact, or an obvious data quality issue.
- Take notes. If the machine sounds different, smells different, or the operator mentions a concern, note it on the route. Context information helps the analyst enormously.
- Maintain the schedule. Skipping routes or allowing the interval to stretch undermines trending. If the route is monthly, collect monthly. If a machine is down when you arrive, note it and come back to collect when it’s running.
Analysis Workflow
Download the route data and run your analysis software’s automated fault detection as a first pass. This catches obvious problems and flags machines that need attention. Then review each flagged machine manually — automated diagnostics have a significant false alarm rate and should never be trusted without human verification.
For the manual review, follow this sequence:
- Check the overall trend first. Is the machine stable, gradually increasing, or showing a step change?
- Compare the current spectrum to the baseline. New peaks or significant amplitude changes in known frequencies indicate developing faults.
- Look at the time waveform for impacts, modulation, or truncation (sensor bottoming out).
- If bearing defect frequencies are present, check for harmonics and sidebands that confirm a real defect versus a coincidental frequency match.
- Compare to previous spectra to verify the pattern is developing rather than being a one-time anomaly.
Generate a concise exception report listing only machines with actionable findings. Include the machine ID, fault description, severity assessment, recommended action, and a timeframe for repair. Send this report to the maintenance planner within one week of data collection — a two-month-old report has lost most of its value.
The discipline of regular route collection, prompt analysis, and timely reporting is what separates effective vibration programs from expensive data collection exercises. The technology is mature and proven. The human discipline to apply it consistently — that’s the differentiator.