What this means for you
If you come from a vibration analysis background, you're likely familiar with ISO 20816-3 and its severity zones (A, B, C, D). You may have expected Tractian to use those same thresholds to trigger alerts.
Tractian takes a different approach. Instead of applying fixed, universal vibration limits, the platform compares each machine against its own historical behavior to generate Insights, and against similar assets in the Tractian database to generate Initial Health Reports.
This article explains the technical reasoning behind that decision and why it results in earlier, more accurate fault detection.
Important: Tractian does not discard ISO 20816-3 entirely. Tractian fully adopts the standard's physical guidelines, such as sensor placement philosophy and mounting rigidity recommendations. The difference applies exclusively to the vibration severity thresholds.
The problem with fixed ISO thresholds
How analysts traditionally use thresholds
In traditional condition monitoring, analysts hard-code these values into their software as Caution (Yellow) and Warning/Critical (Red) limits. It works like a traffic light: Zone A/B is green (acceptable), Zone C is yellow (caution), Zone D is red (shut down).
This is a practical approach for route-based, human-led analysis. If you manage 2,000 machines with a portable data collector, you need a universal filter to tell you which 50 machines require immediate attention. The ISO limits serve that purpose well.
Why fixed limits fall short for continuous monitoring
The ISO limits are a "one-size-fits-all" approach. They must be broad enough to cover everything from a massive paper machine roll to a small cooling tower fan. In doing so, they miss the behavioral nuance of each individual asset.
This limitation was identified as early as 1971 by researchers Downham and Woods, who demonstrated that two machines with the exact same internal rotor problem can show wildly different vibration levels on the outside due to differences in casing stiffness. Their recommendation: look at each machine's individual baseline rather than relying on a generic chart.
With continuous IoT monitoring and AI, that route-based filtering bottleneck no longer exists, and neither does the need for a universal, static threshold.
How Tractian detects failures instead
Insights: comparing the machine against itself
Every machine has a unique vibration fingerprint. Even two identical motors from the same assembly line vibrate differently depending on mounting, ambient temperature, and load.
Tractian's AI follows a two-step process:
Detect the anomaly: "Is this machine behaving differently than its established baseline?"
Diagnose the cause: "If yes, what specific problem does this machine have?"
Why this matters in practice: Imagine a motor that historically runs at 1.0 mm/s. Over two weeks, it ramps up to 3.5 mm/s. Under ISO 20816-3, this machine is still in Zone A/B (Good/Acceptable) no alert would trigger. But Tractian's AI detects a 250% increase in vibration energy and flags the deviation as an Insight, potentially weeks or months before the machine ever reaches the fixed 4.5 mm/s threshold.
Initial Health Reports: comparing against similar assets
When a sensor first comes online, it hasn't built its own baseline yet. Traditionally, ISO thresholds fill this gap with a generic classification (e.g., "Group 1, Rigid Foundation").
Tractian uses a different approach: it compares the new asset against its database of similar machines. Instead of placing a 50 HP motor driving a centrifugal pump into a broad ISO bucket, Tractian compares it to thousands of other 50 HP motors driving centrifugal pumps.
This peer-to-peer comparison uses three vibration variables:
Variable | What it captures |
Velocity RMS | In broader bands than ISO, allowing us to see a fuller spectrum |
Acceleration RMS | No filters applied |
Acceleration Peak-to-Peak | Impulsive events like bearing defects |
The Tractian database currently monitors over 50,000 motors, categorized by rotation speed (low, mid, high) and power rating (low, mid, high), resulting in 9 comparison groups.
What this means in practice
Fewer missed failures
Tractian catches machines that are degrading rapidly but still technically below the fixed ISO limit. A rigid threshold system would let these machines run until they cross the line. Tractian's AI identifies the failure curve earlier by tracking relative change from each machine's baseline, allowing earlier action from the maintenance team.
Fewer unnecessary alarms
Some machines naturally vibrate at higher levels by design. ISO limits would flag these machines as constantly critical, causing alarm fatigue. Tractian's AI learns that high vibration is the normal state for that specific asset and only alerts you when it deviates from its own normal.
Immediate assessment for new machines
When you install a new sensor, you don't have to wait weeks for a baseline to build: Tractian's peer comparison provides an immediate health assessment based on real-world data from similar assets, from the biggest asset library in the world.
Frequently Asked Questions
Does Tractian ignore ISO 20816-3 completely?
No. Tractian follows the standard's physical guidelines (sensor positioning, mounting rigidity), and many other suggestions from it. The difference is only in the vibration severity thresholds: instead of fixed ISO limits, Tractian uses dynamic, AI-driven baselines.
Can I still see raw vibration values in mm/s on the platform?
Yes. The vibration data is fully visible in the Analytics tab. You can view Velocity RMS, Acceleration RMS, and Acceleration Peak-to-Peak values and compare them against any reference you choose.
What happens during the first days after sensor installation, before a baseline exists?
Tractian generates an Initial Health Report by comparing your asset against similar machines in its database. This provides an immediate health assessment while the machine's individual baseline is being established. If you need further information about the Initial Health Report, refer to this article.