This ongoing technology transfer demonstration project will prove the performance and viability of the integration of the CASTeC Core Modules within other IT systems based on standard architectures and products for the energy domain.

Objectives

The project aims at exploiting the technology previously developed for the space domain in a different application domain, which focuses on the control and supervision of petrochemical plants or large hydrocarbon production processes and transportation facilities.

In fact, these are characterised by similarities with the satellites operations according to the following features:

  • co-existence of several interacting units and equipment components in a single facility
  • similarity among different component within a single facility or among facilities of a same operator (which could be similar to satellites constellations)
  • need of high and long-term reliability and safety
  • high level and increasing automation of the operation control
  • high number of interconnected equipment and sensors logging huge amounts of data.
  • sensitivity to failures or unplanned malfunctions of critical components, which may lead to service downtimes and / or severe environmental or health impacts
  • need for remote monitoring/control/operations due to harsh environment

Benefits

Reduction of operators’ workload

Reduction of operators’ workload

Automation of monitoring process reduces operators’ overload and potential stress due to the many signals that could be symptom of anomalies.

Improved identification of subtle anomalies

Improved identification of subtle anomalies

Identification of hidden anomalies within thresholds:

– signals within acceptable thresholds that might be anomalous under certain contexts

– trends going to exceed thresholds within a critical time horizon

Understanding the monitored behavior

Understanding the monitored behavior

Reduction of potentially late/delayed understanding of novel behaviour / anomalies and supporting of suited reaction / remedial intervention.

Improved predictive capability

Improved predictive capability

Improving the predictive capability on the system performance / integrity degradation by the supervision system and team.

Results interpretability

Results interpretability

Methods implemented produce results that can be easily interpreted. This feature is hardly covered by State-of-the-Art Deep Learning or, more generally, Artificial Intelligence techniques typically exploited in this field.

Features

Contextualised anomaly detection

Contextualised anomaly detection

Anomaly detection based on specific operative conditions of the components under analysis, defined by the operators.

Identification of correlations among events

Identification of correlations among events

Identification of potential cause/effect relationships or correlations among events and trends of different signals.

Estimate of the Remaining Useful Life

Estimate of the Remaining Useful Life

Computation of the estimated Remaining Useful Life (RUL) of a component. This helps the engineers in scheduling the maintenance, optimizing operating efficiency, and avoiding unplanned downtime.

Health status indications

Health status indications

The anomaly detection process returns measures of the health status at different levels (from the single telemetry parameter to the whole subsystem investigated), providing a priority score that allows the operators focussing only on potentially critical situations.

System Architecture

The system is integrated with the monitoring platforms or dynamic digital twin already available at end users premises.

In this scenario, the system exchanges information with the end users database by means of dedicated APIs having the goals of:

– retrieving the input data from the plant database and making them available to the algorithms;

– executing queries for writing the results of the algorithms within the plant database.

End users have access to the results saved in their database and they can represent them in the existing monitoring platforms and dynamic digital twin as they prefer.

The system is deployed as a set of portable interconnected Docker images, in which each Docker image contains the basic operating system (Linux based) and basic COTS software applications, producing a set of runnable entities for each algorithm implemented, exploiting HTTP interfaces.

The solution is made available on a software-as-a-service (SaaS) basis.

Related Content