Monitoring Asset Health

Objectives

After completing this lesson, you will be able to:

  • Explain Asset Health Monitoring

Sensors and Alerts

How Can you Monitor the Condition of your Physical Objects?

Physical objects are often equipped with sensors which monitor a specific physical property such as temperature, pressure etc. Sensors enable you to collect time-series data, and by means of this, detect anomalies and predict failures.

SAP Asset Performance Management (together with SAP IoT and SAP S/4HANA) allows you to work with these physical sensors. The sensors are represented as Indicators in SAP Asset Performance Management.

Indicators

In SAP Asset Performance Management, a physical object is represented by an equipment or a functional location. The equipment contains one or several Indicators which represent the sensors of the physical object. Functional Locations can also have indicators.

Note

The Indicator of an equipment in SAP Asset Performance Management is directly linked to the Measuring Point of the equipment in SAP S/4HANA.

One of the major objectives of SAP Asset Performance Management is to create warnings when certain thresholds are exceeded. Warnings and comparable messages are represented by Alerts which are based on Alert Types.

Alerts

Alerts notify users of anomalies, potential failures, or indicator threshold violations.

Alerts can either be created based on rules, or by alarms raised by equipment. The configuration for the equipment alarms mapping details allow the equipment to send alerts. In a scenario where validation is required with alarms to raise an alert, a trigger-based rule must be defined.

The alert type definition allows you to define the alerts that are based on equipment error codes or based on certain computation on the data, for example by applying rules on the data. The definition can also define associations with an indicator and possible failure modes. The association with failure mode data allows you to identify associations, such as instructions.

Rules

With rules, you can leave it up to the system to permanently keep an eye on each and every sensor data point coming in. You start with one or more technical objects that you want to monitor, pick the indicators and characteristics to be observed, and associate both with a rule.

A rule is the technical representation of simple decision logic that, once evaluated against live data, leads to a decision. In the Rules app, you set up the rules to be applied to incoming values. If a value matches the rule condition, the system triggers an action. With these preparations done, you can rest assured that none of the critical values that you might think go unnoticed, thus making sure that all necessary action can be taken without delay.

Available types of rules in SAP Asset Performance Management are, for example, calculation rules, aggregation rules, or scheduled rules.

Integration with SAP IoT

After an equipment is created and published, it is synchronized with SAP IoT (formerly: SAP Leonard IoT), subsequently, it is also synchronized, and objects are created in SAP BTP Internet of Things.

The physical objects with their sensors can be mapped to SAP IoT (Internet of Things): the physical object corresponds to a Device, the sensor is represented as a sensor in SAP IoT.

From SAP IoT, there is an integration to SAP Asset Performance Management: the Device is mapped to a Technical Object, the sensor is mapped to an Indicator.

The following points need to be considered while publishing to SAP IoT:

For every technical object, a Device is created which will be associated with capabilities corresponding to Measurement Point Positions. The characteristics of the measurement point is created as properties within the capabilities.

When you have created a technical object, for every measurement category, a sensor type is created. Also, only one sensor will be created for each device corresponding to a technical object.

For a successful synchronization, each measurement point position can have only one characteristics of type, Date.

The measurement point positions are reused across technical objects.

Failure Curve Analytics

Failure curve analytics lets you view the age at which your equipment may fail within an age range and the likeliness of the failure by using a failure curve with different insights. This helps you detect potential failure risks early and plan actions to prevent failures.

The failure curve with the different insights is calculated per failure mode for a group of equipment (fleet group) with similar operating conditions. The failure curve is displayed on the equipment page. On the failure curve, the following insights are displayed:

  • The current age of the piece of equipment in calendar days
  • The probability of failure (POF) or conditional probability of failure for the current age
  • The upper and lower confidence interval for the POF
  • The predicted failure date for the piece of equipment
  • The time to failure in days (remaining useful live) for the piece of equipment based on the predicted failure date

Failure Curve Analytics and Machine Learning

For calculating the failure curve and the insights, machine learning is used. In that process, model configurations are trained and scored using a Weibull distribution algorithm. The model configuration contains different input data that is used by the algorithm to produce outputs (results). You can have multiple model configurations with similar or different input data.

Input Data

Configuration Parameters:

  • Fleet Group
  • Age of the conditional probability of failure
  • Whether the installation date is maintained and reliable
  • Whether the equipment is repairable

Data Sets:

  • Equipment of fleet group
  • Failure modes of equipment
  • Breakdown notification for equipment and failure mode

Model Configuration Training and Scoring

The training and scoring happens successively in one go.

During the training of the model configuration, the Weibull distribution algorithm uses the breakdown notifications and failure modes together with the information of the first installation date and equipment repairability to calculate the uptime and downtime of the fleet group. The uptimes and downtimes are then used to produce a Weibull model. The Weibull model includes the shape and scale of the failure curve for the fleet group.

During the scoring of the model configuration, the algorithm uses the output of the training to calculate the probability of failure and the other insights, for example, the predicted failure date. With this, the calculation of the final curve and the insights is complete for the fleet group and the failure modes.

Log in to track your progress & complete quizzes