What is a Threshold? Decision Boundary Value • Portable balancer, vibration analyzer "Balanset" for dynamic balancing crushers, fans, mulchers, augers on combines, shafts, centrifuges, turbines, and many others rotors What is a Threshold? Decision Boundary Value • Portable balancer, vibration analyzer "Balanset" for dynamic balancing crushers, fans, mulchers, augers on combines, shafts, centrifuges, turbines, and many others rotors

What is a Threshold? Decision Boundary Value

Published by admin on

Understanding Thresholds in Condition Monitoring

Definition: What is a Threshold?

Threshold (also called limit, setpoint, or trigger value) is a predefined value that separates normal from abnormal conditions in condition monitoring systems. When a measured parameter (vibration, temperature, pressure, etc.) crosses a threshold, it triggers an action—alarm notification, data capture, work order generation, or equipment shutdown. Thresholds are the decision boundaries that convert continuous measurement data into discrete actionable events, enabling automated monitoring systems to identify exceptions requiring human attention.

Effective threshold setting is fundamental to monitoring program success, balancing sensitivity (catching problems early) with specificity (avoiding false alarms). Thresholds embody the program’s decision criteria, reflecting equipment criticality, failure mode understanding, and operational risk tolerance.

Types of Thresholds

Absolute Thresholds

  • Fixed values in engineering units (mm/s, °C, bar)
  • Example: Alarm if vibration > 7.1 mm/s
  • Based on standards (ISO 20816), specifications, or experience
  • Same threshold applies regardless of history
  • Simple to understand and implement

Relative Thresholds

  • Defined relative to baseline or reference
  • Example: Alarm if vibration > 3× baseline
  • Adapts to individual machine characteristics
  • More sensitive to changes
  • Requires good baseline data

Rate-of-Change Thresholds

  • Based on how fast parameter changing
  • Example: Alarm if vibration increases > 50% in one week
  • Detects rapid deterioration early
  • Independent of absolute level
  • Catches accelerating problems

Statistical Thresholds

  • Based on statistical analysis of historical data
  • Example: Alarm if value > mean + 3 standard deviations
  • Accounts for normal variability
  • Requires sufficient historical data
  • Adaptive to process variations

Threshold Setting Approaches

Standards-Based

  • Use ISO 20816 zone boundaries
  • Industry-specific standards (API, NEMA)
  • Advantages: Proven, documented, defensible
  • Limitations: Generic, may not fit all situations

Experience-Based

  • Based on historical failures and successes
  • Institutional knowledge
  • Refined over time
  • Advantages: Site and equipment specific
  • Limitations: Requires experience to develop

Risk-Based

  • Threshold selection based on failure consequence
  • High-consequence equipment: tighter thresholds
  • Low-consequence equipment: looser thresholds
  • Optimizes total program cost and risk

Common Pitfalls

Too Tight (Sensitive)

  • Result: Excessive false alarms
  • Effect: Alarm fatigue, wasted investigation time
  • Risk: Real alarms ignored among false alarms
  • Solution: Relax thresholds based on false alarm rate

Too Loose (Lenient)

  • Result: Problems detected late
  • Effect: Reduced lead time, higher repair costs
  • Risk: Failures before detection
  • Solution: Tighten thresholds, increase monitoring frequency

One-Size-Fits-All

  • Same threshold for dissimilar equipment
  • Doesn’t account for machine differences
  • Either too tight for some, too loose for others
  • Equipment-specific thresholds preferred

Threshold Optimization

Initial Setting

  • Start with standards or conservative estimates
  • Document rationale
  • Plan to refine based on experience

Tuning Process

  1. Track Performance: Count true vs. false alarms
  2. Target Metrics: < 10% false alarms, > 90% true problem detection
  3. Adjust: Tighten if missing problems, loosen if too many false alarms
  4. Document: Changes and reasons
  5. Iterate: Continuous improvement over months/years

Validation

  • Compare to actual failure events
  • Did thresholds provide adequate warning?
  • Were there false alarms that wasted resources?
  • Adjust based on outcomes

Multiple Parameter Thresholds

Overall Vibration

  • Primary threshold for general condition
  • Simplest and most common

Specific Frequencies

  • Bearing frequency thresholds
  • 1×, 2× component thresholds
  • More specific fault detection

Derived Parameters

  • Crest factor thresholds
  • Kurtosis thresholds
  • High-frequency acceleration bands
  • Advanced early detection

Documentation

Threshold Database

  • All thresholds for all equipment
  • Current values and history of changes
  • Rationale for each threshold
  • Approval and review documentation

Change Control

  • Formal process for threshold changes
  • Engineering review and approval
  • Communication to operations
  • Update monitoring system configuration

Thresholds are the decision boundaries that enable automated condition monitoring systems to identify equipment requiring attention. Effective threshold setting and continuous optimization based on performance metrics—balancing early detection with acceptable false alarm rates—is fundamental to condition monitoring program success and operator confidence in the system’s reliability.

← Back to Main Index

Categories:
WhatsApp