Predictive Analytics – The future of operation efficiency

Achieving high availability of vessels is critical to commercial owners, as it directly impacts profitability for the vessel operators. Likewise, for naval vessels, the operational readiness and lifetime of the fleet is critical. To operate vessels beyond its intended lifetime, regular preventive maintenance is key.

With the advance of technology, particularly in the areas of data analytics and machine learning, the Marine arm of ST Engineering was quick to apply predictive maintenance in ships. Since 2012, ST Engineering has used data analytics and machine learning to narrow down areas to service and repair in vessels before it actually breaks down.

ST Engineering’s NERVA Ship Management System and Sensemaking System (SMS2) integrates predictive maintenance into the Ship Management Systems. Over the past years, NERVA SMS2 has been deployed on small naval crafts and government vessels, and expanded to yard cranes and water treatment plants.

Figure 1 below shows a timeline of the major milestones on the development of NERVA SMS2, through the use of data analytics.

Maritime maintenance – from reactive to predictive with NERVA SMS

The NERVA Ship Management System is a centralised control and monitoring system that provide situational awareness of the machinery status onboard the vessel. Integrating sensors with system machineries, the NERVA Ship Management System allows remote control and monitoring of the various systems from the bridge or a machinery control room.

The distinguishing feature of the NERVA SMS solution is the Sensemaking module. Effectively incorporating data analytics and machine learning technologies, it reduces unplanned downtime through predictive diagnostics and greatly enhances operational efficiencies by enabling maintenance to switch from reactive to predictive.
With data acquired from machinery onboard through the NERVA Ship Management System, the Sensemaking System builds an evolving algorithm based on the intervention carried out that enhances predictive maintenance, further minimising downtime from potential breakdowns in fleets.

Data Analytics and Machine Learning in NERVA SMS2

The core of predictive analytics is machine learning. Designed to predict failure over time, NERVA SMS2 uses machine learning to train the models to derive and enhance the inter-relationship between sensor readings and failure modes.

The truth is, predictive accuracy is very data dependent. NERVA SMS2 uses a combination of sensor data, OEM data and failure modes to create data models that mimics and predicts the performance of the system through machine learning. With time or data logs from multiple vessels, the data models constantly enhances itself, offering new depth for predictive maintenance that is customised for the fleets that has it installed.

Predictive Maintenance through NERVA SMS2

Predictive maintenance through NERVA SMS2 provides three types of alerts to operators, namely diagnostics, prognostics, and predictive alerts.

  1. 1) Diagnostics alerts aids the operator in finding the root cause of a failure.
  2. 2) Prognostics alerts gives the operator advanced warnings before the actual occurrence of a failure, allowing operators to consider timely maintenance before the failure occur or before the failure escalate and impact more subsystems.
  3. 3) Predictive alerts provide condition-based maintenance through real-time monitoring to customise the maintenance schedule.

The recommended maintenance schedule is typically generalised and may not relate to the actual condition of the machinery. With the predictive alerts, NERVA SMS2 analyses the real-time condition of the machinery, and tailor the maintenance interval to the actual worn out condition of the machinery.

With NERVA SMS2 in place, operators can be alerted when maintenance is required, before a failure occurs, and provide diagnostic insights when a failure occur. It offers operators a choice to act on the appropriate corrective or preventive maintenance, translating into higher vessel availability, and lower maintenance cost to fleet owners.

Case Study: Smart Maritime Autonomous Vessel (SMAV)

NERVA SMS with Sensemaking module was installed for a tugboat retrofit project. The added NERVA SMS monitors sensors values deviation and divergence as well as critical alarms. Prediction of machinery failure for the tugboat’s diesel generator lube oil system, as well as its main engine lube oil system, fuel system and turbocharger. At the component level, these predictions includes failure on lube oil pump, lube oil bearing degradation, fuel intake valve, fuel exhaust valve, fuel rack, fuel injection nozzle, turbocharger bearing degradation and misalignment.  

Case Study: Police Coast Guard 5th Generation Patrol Craft

NERVA Sensemaking module can also be tailored to better support the Police Coast Guard in the operations by predicting critical alarms that could cause engine shutdown hours before the occurrence of these critical alarm. This will allow Police Coast Guard to plan their operations better.   

What’s Next for Maritime Predictive Maintenance

A further development of NERVA SMS2 is an enhanced decision support tool, providing real-time situational prescriptive actions that the Operator can choose, to avoid unplanned downtime. The enhancement into an intelligent prescriptive system would help Operators with quick decision-making and minimise operation stoppers.

With the 5G cellular network technologies and coverage in Singapore expected nationwide by 2025, NERVA SMS2 can pipe data back from vessels to centralised maintenance operations centre onshore faster. With improved real-time overview of the machinery health, the commercial readiness of the fleet can be optimised for ship owners.

Predictive Analytics Beyond Maritime Applications
Beyond maritime, ST Engineering also extended data analytics applications to the predictive maintenance of water treatment plant. AquaNERVA is the ST Engineering product for the management of treatment processes for water and wastewater. In close collaboration with Singapore’s national water agency, PUB, AquaNERVA covers data analytics and machine learning model simulation for condition monitoring, predictive diagnostics and prognostics for the predictive maintenance of blowers at water treatment plant. These includes the early prediction of common failures of aeration blowers such as high differential winding temperature, reduced blower efficiency, high differential pressure and high discharge pressure.

With the advancement of technologies and the greater availability of data through Internet-of-things (IoT), predictive analytics can only get more accurate and would present a more compelling case for operators of fleets who are looking to improve the operational readiness and extend the lifecycle of their fleets.