Predictive Maintenance of Industrial machine with Amazon Monitron and What is “ISO 20816” ?
In this article, learn about Amazon Monitron, an AWS service that enables predictive maintenance for industrial machinery, and the ISO 20816 standard used in the service.
> This article is an English translation of a Japanese article.
The importance and challenges of predictive maintenance in industry
In the industrial world, many factory automation (FA) and other industrial machines are involved in controlling machine and are indispensable to modern society. Examples include opening and closing automatic doors, operating production lines, and controlling air conditioning.
Their importance grows every year, and they must operate safely and reliably. Inspection and maintenance are essential to achieve this, but it is also true that regular maintenance is becoming increasingly difficult due to strict downtime requirements and the sheer number of plants. Predictive maintenance, which uses sensors to reduce the labor required for daily inspections, detects abnormal conditions based on data, and performs active maintenance, is gaining attention.
While predictive maintenance has become a easy for newer industrial machines equipped with sensors from the start, existing(retro) machines without such a system face the challenge of being left behind.
Amazon Monitron and ISO 20816
Amazon Monitron is an AWS service for predictive maintenance of industrial machine. It provides not only the cloud, but also hardware such as sensors, gateways (repeaters), and a smartphone app for on-site management all in one place.
Vibration is measured by an Amazon Monitron sensor with built-in acceleration and evaluated using machine learning models and the rules of ISO 20816, the international standard for machine vibration, to detect anomalies and make maintenance operations more active and efficient based on the data. One of the features of this system is that the sensor can be retrofitted, enabling predictive maintenance not only for new but also for existing industrial machines.
Machine learning models learn, infer, and evaluate based on sensor values. So what is ISO 20816? Let’s dig a little deeper.
What is ISO 20816 ?
ISO 20816 is an international standard for the measurement and evaluation of machine vibration. It defines measurement methods and values for evaluating the condition of machines based on the vibrations they emit during operation.
As a machine ages, its parts wear out and a different vibration pattern is generated. ISO 20816 defines the condition of a machine by measuring and evaluating vibration displacement, vibration velocity and vibration acceleration.
As an aside, ISO standards are revised periodically, and Life Cycle shows that ISO 20816 is preceded by ISO 7919 (shaft vibration) and ISO 10816 (stand vibration). Check the part before you buy.
ISO 20816’s “Machine types” and “Evaluation zones”.
ISO 20816 evaluates the condition of a machine by matching the machine types under consideration to the evaluation zone.
First, the machine type is divided into several parts. The parts of ISO 20816 that I am currently aware of are as follows;
- Non-rotating parts (*1)
- Group 2: Medium-sized machines *2 (rated power above 15 kW up to and including 300 kW) (*2)
- Group 1: Large machines (rated power above 300 kW) (*2)
The status of a machine is determined by “Evaluation zones (*3)”. Evaluation zones have four definitions.
- Zone A: Newly commissioned machines
- Zone B: Acceptable for unrestricted long-term operation
- Zone C: Unsatisfactory for long-term operation
- Zone D: Sufficient severity to cause damage
As you can see, Zones A and B are the safe zones.
The classification of the machine under consideration and the values for the evaluation zones are defined in the standard. For example, for non-rotating parts, Zone A/B has a vibration velocity of 0.71 mm/s ~ 4.50 mm/s, Zone B/C 1.80 mm/s ~ 9.30 mm/s, and Zone C/D 4.50 mm/s ~ 14.7 mm/s (*4). Group 1 and Group 2 of the target machine have similar definitions.
In short, “the louder the machine vibrates, the more problematic”. BTW, the reason the values overlap is to provide a range for evaluation.
- *1: ISO 20816–1:2016 — Annex C
- *2: ISO 20816–3:2022 — 5.2 Classification according to machine type, rated power or shaft height
- *3: ISO 20816–1:2016 — 6.3.2.3 Evaluation zones
- *4: ISO 20816–1:2016 — Annex C
Evaluate by Amazon Monitron with ISO 20816
Amazon Monitron uses ISO 20816 mapping with “machine class” as the machine type and “ISO warnings and alarms” as the evaluation zones. Once the machine class is set, the evaluation zones are set by the application. This is easy because there is no need to know detailed values.
The machine class is selected from the four classes defined in ISO 10816–1:1995 (predecessor to ISO 20816). Specifically, when setting up a sensor in the Amazon Monitron application, you will see the following;
This selection sets the “ISO Warnings” and “ISO Alarms” corresponding to the evaluation zones. Warnings correspond to Zone C and Alarms correspond to Zone D.
Note: You can create your own definitions in Amazon Monitron’s “Custom Classes”.
The following is an example of the display in the Amazon Monitron app. It shows data configured as Class I. ISO warning is set to 1.80 mm/s and alarm to 4.50 mm/s.
What machine class should I select in Amazon Monitron?
The machine class is basically selected based on the rated power. If you do not know the specifications, compare with similar machines.
For example, an electric motorcycle in the 50 cc displacement category would be classified in Class I with a maximum power rating of 0.6 kW. An electric motorcycle with a power rating of 20 kW, which corresponds to Class II (Group 2 for ISO 20816), is the power required for a large motorcycle.
Note: This example is based in Japan and may not apply to your country. The important thing is to refer to machine that is familiar to you.
Settings that differ from the specifications do not significantly affect measurements or evaluation.The class can be changed after commissioning. Therefore, the class can be changed based on actual measurements or a custom class can be created and applied. If the rated power is unknown, one method is to set Class I first.
Common to all machine classes is the mounting of the sensor. The Amazon Monitron documentation describes where and how to mount the sensor according to ISO 20816. This is to ensure accurate vibration, so be sure to install them properly.
Conclusion
Amazon Monitron has the advantage of being ISO 20816 compliant right out of the box, so you can get started right away.
On the other hand, measurements of industrial machinery, especially existing machinery, can vary greatly depending on the operating performance of each individual machine and the installation conditions. Therefore, it is advisable to check against previous maintenance records before making a decision, even if warnings or alarms occur immediately.
In addition, once machine learning inference is initiated, a more situationally accurate assessment can be obtained.
Measurement and predictive maintenance with retrofit sensors has the potential to transform operations.
Check and Try the Amazon Monitron today!!
― Kohei “Max” MATSUSHITA / @ma2shita
