At Mangalore Refinery and Petrochemicals Limited (MRPL), Artificial Intelligence and digital technologies are leveraged as a core business strategy to achieve operational excellence, strengthen business performance, and enhance safety, while enabling sustainable value creation through automation and advanced analytics. As a large and complex refining and petrochemical organization, MRPL leverages AI not as a stand-alone technology initiative, but as a systematic enabler of data-driven decision-making, operational transparency, and sustainable performance.

MRPL’s AI initiatives are primarily focused on three high-impact domains, process optimisation, asset reliability, and energy and environmental efficiency, which directly influence refinery safety, operational stability, sustainability, and competitiveness. In addition to these core areas, AI is being applied to other critical business challenges wherever data-driven insights can enhance performance, responsiveness, and decision quality. A data-centric approach has been institutionalized across operations and management, enabling timely, objective, and consistent decisions across the organization.

A distinctive strength of MRPL’s AI journey is its in-house AI and digital team, comprising domain experts who combine deep domain knowledge with advanced analytics and AI capabilities. All major AI solutions on process optimisation, asset reliability, and energy efficiency have been conceptualized, developed, and deployed internally under MRPL’s REVEAL (Refinery Efficiency and Value Enhancement through Analytic Lens) AI product suite.

1. Flagship AI use case: Hybrid AI with First Principles and Domain Knowledge

The common foundation across all AI deployments at MRPL is the use of hybrid AI models, where engineering fundamentals and domain expertise are tightly integrated with machine learning. This approach ensures that AI outputs remain interpretable, safe for OT environments, and aligned with operational realities—an essential requirement for mission-critical industrial systems.

1.1. AI for Real-Time Process Optimisation

MRPL has successfully deployed multiple Real-Time Optimisation (RTO) applications across complex process units, enabling continuous optimisation under dynamic operating conditions.

a) PolyOptima – Real-Time Optimizer for Polypropylene Unit

It is a first of its kind hybrid AI–first-principles Real-Time Optimization (RTO) system successfully deployed in a commercial-scale polyolefin unit, combining engineering science with artificial intelligence for continuous, sensor-less quality prediction.

It predicts critical quality parameters such as Melt Flow Index (MFI) and Xylene Solubility (XS) directly from reactor data—parameters that traditionally require delayed laboratory analysis due to the absence of reliable online analyzers.

By achieving laboratory-grade accuracy in real time, PolyOptima enables predictive control, smoother grade transitions, and consistent product quality. This breakthrough innovation was recognized with the “Best Innovation Award” by the Government of India in September 2022, underscoring its industrial relevance and technological originality.

b) Power Optima – Real-Time Optimizer for Captive Power Plant

MRPL’s Captive Power Plant Optimizer addresses the complexity of real-time steam and power management in a large refinery. Hybrid digital twins of turbines and auxiliary systems continuously evaluate operating efficiencies and compute optimal loading combinations across turbines, pressure-reducing stations, and grid interfaces. The application computes the optimal loading pattern among the grid, STGs and PRDS for steam and power generation in a real time basis. Real time advisory is sent to operations team through web and mobile interface. Fuel saving in terms of monetary benefits are shown real time on the interface for the operator.

The system provides real-time advisories to operations teams through digital dashboards, improving energy efficiency, operational transparency, and decision consistency during dynamic demand conditions.

c) Real-Time Optimizer for Petro-Fluidized Cracker (PFCC) Unit

The PFCC Real-Time Optimizer is among the early examples of AI being integrated into the control of a large commercial cracking unit. Using hybrid yield-prediction models, it identifies optimal operating conditions while respecting complex operational constraints.

The application was first deployed in advisory mode and subsequently transitioned to closed-loop control, demonstrating a governed and safe pathway for embedding AI into Distributed Control Systems (DCS).

1.2. AI for Asset Reliability

Rotary Sentinel AI is an advanced agentic AI-based framework developed entirely in-house under MRPL’s REVEAL AI suite. Designed as a one-stop solution for rotary equipment reliability, it covers standalone critical rotary assets that are critical to refinery continuity and safety. The system is architected as a collaborative multi-agent framework, where multiple AI agents work in coordination to perform anomaly detection, automated condition monitoring, efficiency tracking, Inventory monitoring, root-cause analysis, and prescriptive recommendation. A key differentiator is the integrated interactive Expert Advisor, which provides consolidated health reports of all critical machines, supports trend analysis with contextual reasoning, recommends operational or maintenance actions, and links predicted failure modes with spare requirements through inventory monitoring and alerts. All models run locally, ensuring complete data privacy and cybersecurity for OT environments.

This innovation received the “Special Innovation Award” from the Ministry of Petroleum and Natural Gas, Government of India in November 2024.

1.3. AI for Energy efficiency

FLARE REVEAL is a first-of-its-kind, fully in-house AI solution that identifies the root cause of flaring events in real time using existing process data. Conventional systems typically indicate only total flare quantity, offering limited actionable insight. No commercial solution exists with this capability.

FLARE REVEAL applies causal AI grounded in process understanding, dynamically accounting for time lags across interconnected units to accurately attribute flaring to specific sources. This solution has been successfully deployed across the entire refinery complex, operating at 15 MMTPA capacity. Refinery-wide deployment has uncovered numerous recurring contributors to flaring and reduced unaccounted flaring through faster, data-driven interventions.\

Also Read | India Aims to Build Indigenous AI Models, Expand Data Infrastructure

By providing precise attribution, real-time alerts, and unified dashboards, FLARE REVEAL enables targeted operational interventions, strengthens environmental governance, and directly supports MRPL’s net-zero and emissions-reduction objectives. The solution is scalable and applicable across refineries, petrochemical complexes, and gas processing facilities.

2. AI Priorities for the Next Five Years

MRPL’s five-year AI roadmap focuses on scaling proven AI solutions, institutionalizing digital platforms, and embedding AI into core operational workflows. Priority areas include expansion of hybrid AI–based process optimisation across additional refinery and petrochemical units, wider deployment of AI-driven asset health and reliability systems, and strengthening enterprise-wide data foundations. MRPL has also initiated a Refinery Virtual Assistant, envisioned as an expert digital guide to support operations teams during routine and abnormal conditions through contextual, real-time insights.

MRPL’s future focus is on scaling hybrid and agentic AI frameworks across more process units and asset classes, and progressively integrating intelligence across optimisation, reliability, and environmental applications. By connecting these systems, MRPL aims to move toward a more autonomous, resilient, and intelligent refinery ecosystem—while retaining strong human oversight and governance.

Equally important is the governance dimension of AI adoption. Change management, workforce upskilling, and building organizational confidence in AI-assisted decision-making remain central priorities as AI transitions from individual applications to enterprise-scale adoption.

Through sustained in-house innovation, deep domain integration, and industrial-scale deployment, MRPL continues to position itself as a leading AI powerhouse in the refining and petrochemical sector, contributing meaningfully to India’s technology-led industrial transformation.

Views Expressed By: Shri Mundkur Shyamprasad Kamath, Managing Director, Mangalore Refinery and Petrochemicals Limited (MRPL)

Be a part of Elets Collaborative Initiatives. Join Us for Upcoming Events and explore business opportunities. Like us on Facebook , connect with us on LinkedIn and follow us on Twitter, Instagram.

"Exciting news! Elets technomedia is now on WhatsApp Channels Subscribe today by clicking the link and stay updated with the latest insights!" Click here!