Introduction

The Indian Navy (IN) has undergone a significant transformation over the years from a fledgling navy at independence, to its current avatar as a blue water navy. In order to meet India’s national maritime security objectives, it is imperative that the IN continues to grow not only in scale, but also acquires superior technologies to maintain an edge over its adversaries. Accordingly, the IN has invested in state-of-the-art technologies towards achieving and maintaining optimal force levels required to undertake its envisaged roles, missions, objectives and tasks.

The maritime domain is intrinsically linked with national prosperity, and accretion of sea-borne assets is a significant aspect of maritime power. Operational health of sea borne assets is, therefore, one of the most significant aspects for achieving optimal asset availability, resulting in reliability and optimal performance of onboard equipment and machinery emerging as critical factors. The limited availability of INS Vikrant in the 1971 Indo-Pak War, due to a defective boiler, serves as a reminder about how equipment condition can impact operational readiness.1

Lately, IN ships have been deployed far and wide, on ‘Mission Based Deployments,’ posing daunting maintenance challenges. Deployment and maintenance share a linear relationship, wherein deployment is the dependent variable. In the classical maintenance model, an increase in deployments necessitates a proportionate increase in maintenance schedules. However, the ratio of operational availability to maintenance periods can be improved by adopting methodology of ‘Reliability Assessment,’ a maintenance model that takes into consideration equipment history and present status to predict future performance, based on monitoring of parameters.

All by itself, ‘Reliability Assessment’ has limited functionality in the dynamic environment of a warship. But, when supported with Artificial Intelligence (AI), the outcome is extraordinary. The methodology has the potential to keep onboard equipment, both machinery and weapon systems, in optimum condition whilst minimizing maintenance overheads, thus, substantially improving asset availability. Reliability Centred Maintenance (RCM) is a concept based on the ‘Reliability Assessment’ model, which can be further enhanced using AI and Big Data Analytics (BDA). This article discusses a concept of integrating these aspects to enhance reliability, while minimising maintenance.

Gaps in Current Life Cycle Management of Shipboard Equipment

The IN’s current maintenance strategy is based on Planned Preventive Maintenance (PPM). Condition Based Predictive Maintenance (CBPM) was envisaged to replace/ optimise PPM on a progressive basis. However, it has, more or less, been accepted as another supplementary methodology, in addition to PPM.

As per current practices, decision-making on shipboard equipment life cycle management, including induction, operation and maintenance, are largely human centric. This persists despite the rich and long experience of handling hi-tech equipment available with the IN’s maintenance support agencies. A consolidated database of curated and usable data, to aid and bring a level of AI enabled automation into the decision-making process, remains a capability gap. Such a database would require analysis and archiving, as per a common accepted protocol, and would form the core of the proposed concept. The requirement of an efficient and effective organisation using state-of-the-art technologies, including AI and BDA, for life cycle management of machinery and equipment of warships cannot be overemphasised. Some gaps in the current life cycle management of shipboard equipment are summarised as follows: -

  • Limited onboard defect analysis capability.

  • Manual recording of equipment parameters, leading to limitations in data analysis for predictive maintenance.

  • Non-availability of real-time ship equipment data at shore based agencies for real-time analysis.

  • Non-availability of quantifiable equipment reliability metrics, to aid operational decision-making processes.

  • Non-availability of statistical analysis of a large amount of data for ascertaining optimal operating parameters, which could facilitate in optimisation of endurance and costs.

  • Traditional manual methods of failure detection and analysis, which are time consuming and inefficient as compared to methods employed in RCM.

  • Continued dependency on Original Equipment Manufacturers (OEM) for formulating maintenance routines, despite the possibility of routines being inflated to serve vested interests.

  • Reliance on a human centric approach for resolution of maintenance problems/ situations, rather than relying on statistical tools involving advanced computing.

Maintenance Optimisation - Industry Trends

Maintenance of shipboard engines and machinery is evolving rapidly, with communications and IT playing an increasingly significant role. Implementation of these new concepts is reducing maintenance costs for ship owners and vessel operators in the merchant marine, by extending lifecycles of machinery. Vessel operators and manufacturers can remotely monitor machinery condition, modify maintenance programs, and identify as well as predict potential failures. Examples of such technologies are General Electric’s (GE) ‘Predix-2’ and ‘Digital Twin’ as shown in Fig 1, wherein gas turbines across the globe are monitored and analysed for continuous optimisation using AI.

AI is commonly used to help remote inspection and maintenance of difficult to reach places onboard ships, reduce risk to human life, cut costs, improve fuel efficiency of vessels by providing greater insights into ship performance, and support predictive maintenance, enabling early detection of equipment failures in real-time. ‘Expert Insight’ by M/s Wartsila is another example. Vessel maintenance is also being improved using 3D models of ships, known as ‘Digital Twins.’Availability of AI tools has aided Research and Development (R&D). Optimisation of Gas Turbines has been done, wherein a variety of models and software tools have been developed for real time performance adaptation, with different levels of complexity, fidelity, accuracy and computer performance requirements. Numerical Propulsion System Simulation (NPSS), developed by NASA, is one such powerful gas turbine simulation tool with impressive capabilities.2

Fig 1: Operations and Maintenance Optimisation Model by GE3

RCM

RCM is the process of determining the most effective maintenance approach to achieve desired reliability.4 The RCM philosophy employs PPM, CBPM, ‘Real Time Monitoring’ (RTM), ‘Run-to-Failure’ (RTF)/ ‘Reactive Maintenance’ and ‘Proactive Maintenance’ techniques in an integrated manner. The process increases the probability of achieving desired equipment performance over its designed life cycle, while minimizing maintenance.5,6

Reliability and performance of each equipment onboard is quantified using reliability engineering and mathematical methods. Historical data of each equipment, which includes repairs/ maintenance, spares usage and performance characteristics of each sub-equipment/ component is utilized in quantifying reliability.7,8,9 Reliability Analysis can be extended fromindividual equipment to the parent system and, even further, to mathematically provide a quantifiable measure of reliability. Propulsion, power generation, communication, sub-surface warfare, etc., are some examples of equipment that could benefit from RCM. Theoretically, in future, the data can be further integrated to quantify the reliability of an entire warship or even a fleet. Enhancements in safety and pollution control could be additional benefits of RCM.10

AI - Significance in Maintenance Management

As is evident, reliability quantification would be highly dependent on the collected and processed equipment performance data.11,12 The accuracy of results would improve consistently with regular population of data. The RCM approach of quantifying reliability is impossible using manual methods, due to the need for multi-level integration and involvement of large amounts of data. With the advent of high computation power, time is ripe for implementing AI to provide a practical solution through establishment of an ‘Integrated Maintenance Management Organization’ (IMMO).13

Fig 2: Types of AI14

Of the various types of AI (see Fig 2), narrow AI, which has reached an adequate level of technological maturity, is suitable for the proposed application. From the functionality perspective, reactive or limited memory concepts would be utilized to achieve the proposed solution along with BDA. BDA is a complex process of examining large amounts of data to uncover hidden patterns, correlations and parameter trends. The most common predictive techniques that use BDA are regression techniques to predict equipment failure, and machine learning techniques, which emulate humans learning by detecting patterns amongst variables that have, in the past, led to equipment failure. Challenges include data capturing, analysis, curation, search, sharing, storage, transfer, visualisation and information security.

IMMO

An ‘IMMO,’ which will be primarily based on AI applications, may be the way ahead to harness the benefits of RCM. Key aspects of the concept which could be considered for the IN are as follows: -

  • Data Collection. Equipment performance data, like running parameters, process values, and recordings from each equipment would be consolidated at the ship’s ‘Integrated Platform Management System’ (IPMS). The sampling rate of data can be adjusted to optimize and match available computational resources.

  • Data Transmission. The collated data can be encrypted onboard and transmitted to a shore-based facility, hypothetically designated as ‘Indian Navy Reliability Assessment Centre’ (INRAC), using existing satellite communication networks or hired satellite bandwidth. The data would be decrypted at INRAC and added to the existing data base.

INRAC

INRAC would be equipped with state-of-the-art infrastructure comprising advanced computing hardware. The data received from ships would be analysed using automated algorithms, with minimal human intervention, to predict key performance deviation indicators aimed at quantification of reliability of each equipment, as well as overall functional capability of the ship. The data analysis would provide maintenance alerts based on time, as well as condition based predictive methods, using current and past data. Other spin-offs from INRAC could be:-

  • Data Assessment and Alerts. Parameter deviations of equipment can be forwarded to concerned trial agencies, where decisions on remedial measures against impending failures/ breakdowns can be deliberated. The predictive maintenance alerts and decisions/ analysis can, thereafter, be communicated to ships for remedial measure implementation.

  • Database Sharing. Equipment reliability assessment can be made available in real time at all operation centres, wherein assessment of the actual material state of a ship can become part of the decision-making process. Real time, or near real time data, in a segregated manner would be available to select stakeholders, like trial agencies and professional directorates. The data could also be used to take informed decisions on equipment-fit selection for future acquisitions.

  • Data for Posterity. INRAC could also function as a data repository centre for the entire life cycle of an equipment. This data, with the help of AI, would cater for retention of institutional knowledge. Repository of such a database would be a prime asset for the IN as well as OEMs, for design optimisation.

  • Data Errors. With data collation and assessment at shorter data sampling rates, incorrect data/ sensor errors can be picked up by an AI enabled system and the concerned entity can, thereafter, be alerted. Central collation of large amount of performance data would also aid in continuous improvement of reliability assessment results.

Operational Optimisation

The ‘IMMO’ concept links equipment condition assessment with operations and deployment. This link provides an ideal feedback loop amongst various agencies, through a realistic assessment of quantified equipment reliability metrics. Relative assessment of each capability aspect of the ship, from main propulsion, power generation, auxiliaries, weapons and sensors, etc., can be utilised to choose the ‘most suited’ platform available for a particular mission. Performance assessment can also be used to optimise varied functions, like the most optimal cruising speed of a ship, which could not only result in optimisation of fuel consumption, but could also enhance endurance and reach.

Way Ahead

A project on prototype implementation of the concept can be initiated on a small scale through a navy-industry partnership, including collaboration with a ‘start-up.’ The concept would involve quantification of reliability of a sample equipment, using tenets of reliability engineering and creation of a simulated set up of IMMO. The application of the concept could then be implemented on an IPMS ship. Evaluation of hardware and software requirements for implementation on a larger scale in the IN could then follow in due course, post successful technology demonstration.

Conclusion

The proposed concept is based on a ‘maintenance-management organization,’ which will cater to current requirements and capabilities of our rapidly growing Navy. Employment of cutting edge technology, including BDA, AI and RCM methods, would optimize reliability while significantly reducing maintenance overheads and mean time between failures. Additional benefits that could accrue would be greater self-reliance, improved safety, and environmental protection.

Naval Equipment Reliability Analyser (NETRA) project, initiated by the Indian Naval Ship Maintenance Authority (INSMA), reiterates the significance and need to migrate towards RCM philosophy.15 However, for effectively utilizing RCM, implementation of the ‘IMMO’ concept is considered as the ‘way ahead.’ The proposed concept, using AI, BDA and RCM, could revolutionise maintenance management of IN ships by minimising maintenance and enhancing combat potential, through accurately predictable and quantified reliability.

Endnotes

  1. VAdm N. Krishnan (Retd), A Sailor’s Story, ed. Arjun Krishnan (New Delhi: Punya Publishing, 2011).

  2. M. Tahan et al., “Performance -Based Health Monitoring, Diagnostics and Prognostics for Condition-Based Maintenance of Gas Turbines: a Review,” Applied Energy no.198 (2017), 122-144.

  3. Ibid.

  4. Chief of Naval Operations, US Navy, “Maintenance Policy for Navy Ships,” OPNAV Instruction 4700.7M, 08 May 2019.

  5. A. Taher, I. Lazakis, O. Turan, “Integration of Business and Technical Aspects of Reliability and Maintenance,” ICMT2014, 7-9 July 2014, Glasgow, UK.

  6. O. Turan, et al., “Maintenance/Repair and Production-Oriented Life Cycle Cost/Earning Model for Ship Structural Optimization during Conceptual Design Stage,” J. Ships Offshore Structures vol.4, no.2 (2009), 107-125.

  7. AA Serap, A. Seref and O. Turan, “Reliability and Availability of Pod Propulsion Systems,” Journal of Quality Reliability Engineering International vol. 22 (2006), 41-58.

  8. JJF Espiritu, “Component Criticality Importance Measures for the Power Industry,” Electric Power Systems Research vol.77 (2007), 407-420.

  9. I Lazakis et al., “Effective Ship Maintenance Strategy using a Risk and Criticality-Based Approach,” Proceedings of the 13th Congress of International Maritime Association of the Mediterranean, Istanbul, vol.3 (12-15 October 2009), 1113-1120.

  10. O. Turan, A. Olcer and PL Martin, “Risk Assessment of Loss of Life for Fishing Vessels in Fuzzy Environment,” Safety and Reliability vol.23, no.2 (2003), 19-39.

  11. I. Lazakis et al., “Advanced Ship Systems Condition Monitoring for Enhanced Inspection, Maintenance and Decision Making in Ship Operations,” 6th Transport Research Arena, 18-21 April 2016.

  12. CK Pil, M. Rausand and J. Vatn, “Reliability Assessment of Reliquefication Systems on LNG Carriers,” Reliability Engineering and System Safety vol.93 (2008), 1345–1353.

  13. AHK Khairy, “Application of Artificial Intelligence in Maintenance Modelling and Management,” 2nd IFAC Workshop on Advanced Maintenance Engineering, Services and Technology, Spain, 22-23 November 2012.

  14. Ibid.

  15. S. Gupta, M. Goel, and J. Tom, “NETRA- Naval Equipment Reliability Analyzer – Development of Analytical Tool for Reliability Prediction,” Journal of Marine Engineering vol.77, 8-20