Despite powering the world, refining and petrochemicals still hold vast untapped potential. Digital transformation and applied AI solutions offer a proven path to higher profits and lower emissions, but change management remains a bottleneck.

Executive Summary 

Refineries are central to the global energy system, processing an estimated 100 million barrels of crude per day. Yet most run far below potential. Only 8% of refinery data is used in decision-making, while unplanned downtime erases $20–50 million annually per site. The opportunity is proven. End-to-end optimisation can add $0.50–$1.00 per barrel, or $30–85 million per mid-size refinery annually. Indian Oil Corporation (IOC) deployed digital twins to cut losses of $25–35 million per unit’s interruptions, while Aramco’s Yanbu refinery achieved an 18% increase in throughput, a 35% boost to profitability, and a 14% reduction in emissions by deploying optimisation algorithms. The implications are clear. Leaders who embrace AI-powered transformations are adding tens of millions in free cash flow to assets that sustain entire economies. 

Refining: An Industry Running Below Potential 

Human civilisation runs on refined energy. The cars we drive, the planes that lift us across continents, and the chemicals that sustain our industries all flow from the quiet labour of refineries. These industrial complexes, largely invisible to the public, are the cogs that keep modern life in motion. 

Most of them, however, remain relics of an earlier era. They were built in the bygone century to perform a simple task: converting crude oil into usable fuels. For decades, these refineries functioned with remarkable reliability. But what was once sufficient has now become an example of inefficiency. In a time when artificial intelligence and advanced analytics make optimisation easier than at any point in history, these critical assets still operate with old control stacks and fragmented practices. The paradox is clear: the world’s most essential machinery is also among its most neglected. 

This article examines why under-optimisation persists by looking at state-owned refineries in India and the Middle East, and how practical technologies are beginning to transform raw data into stability, profitability, and lower emissions. 

A Paradox at the Heart of the Energy Economy 

The scale of inefficiency in downstream energy facilities is staggering. Oil refineries process on the order of 100 million barrels per day globally, an industrial achievement without which the modern economy would collapse. And yet, the very sector that carries such global implications continues to run far below its potential. 

The reasons are visible to anyone who has stood inside a control room. Much of the equipment has long outlived its inception age. Data is scattered across historian systems and rarely flows into decisions. Management remains inert, with strategies that advocate for a ‘just keep it running’ approach to avoid operational disruption at the cost of value-generating improvements. 

The numbers reveal the real impact. Only a fraction of available data, about 8 per cent, is ever used in decision-making. A mid-size refinery loses $20–50 million each year to unplanned downtime, while a single unit failure can erase $100,000 to $1 million in an hour. Yet where optimisation has been attempted, the results are dramatic. According to McKinsey, end-to-end improvements add $0.50-1.00 per barrel, the equivalent of $30-85 million semi-annually. Indian Oil Corporation and Saudi Aramco’s Yanbu refinery have already shown that this is not theory but practice. 

The takeaway is obvious. The bottleneck is not technology; the tools are already here. What is missing is the willingness to govern these immense assets with the same rigour that is transforming other industries.

National Implications of Downstream Under-Optimisation 

The true weight of refining is measured not only in barrels, but also in consequences. A refinery’s inefficiency does not remain within its fences; it ripples outward across economies. Transport costs rise, supply chains strain, and energy security wavers. 

For state-owned facilities in India or the Middle East, a single underperforming refinery can reshape the country’s domestic energy markets, import requirements, and indirectly its foreign policy strategy. What seems like a technical inefficiency inside a control room can, at scale, alter the balance sheets of governments and the stability of markets. 

In this sense, refining is not merely another branch of heavy industry. It is a geopolitical lever with national implications.

The Blind Spot: 90% of Data Goes Unused 

If inefficiency carries national weight, the roots of that inefficiency lie in a less visible failure: the failure to use information. Refineries are among the most instrumented industrial systems on earth, recording streams of measurements from every valve, pump, and reactor. Yet almost 90 per cent of this information remains idle, buried in data historians or scattered across spreadsheets. 

 This infrastructure was once understandable. When most of the control architectures were built, the computational tools to make sense of such torrents of data simply did not exist. The priority was to keep the plant running, not data interpretation. In that context, deprioritising the flood of data was not negligence. It was practicality. 

But that world has changed. Today, applied intelligence for the energy sector can translate these datapoints into process optimisations, enhanced financial margins, and lower emissions. What holds plants back is no longer technology, but change management. This leadership inertia leads to energy-use drifts trending upward that drive up costs. 

The central challenge, then, is not a shortage of data, but the willingness to turn it into AI-empowered action. Dormant information continues to accumulate while plants operate as if nothing has changed. This reluctance explains why, despite their scale and importance, refineries remain an under-optimised industry. 

Only 8% of refinery data is used; 90% sits idle while value evaporates

Reliability: The First Frontier of Optimisation 

If unused data is the hidden bottleneck of refineries, unreliability is the visible one. Before any vision of artificial intelligence or advanced optimisation can take root, there is a simpler and more fundamental task: keeping the machines stable. A mid-size refinery loses $20–50 million every year to unplanned downtime. These losses occur before gains in yield, before energy optimisation, or before any futuristic scenario of digital transformation. They are the cost of instability. 

The economics of reliability are brutal. When a major unit, be it a crude distillation column, a catalytic cracker or a hydrocracker, shuts down, the clock begins ticking at $100,000 to $1 million per hour in lost contribution. In that moment, every hour of delay becomes a line on the balance sheet. This is why predictive maintenance and reliability analytics matter. They are not optional ‘digital add-ons’, but the frontline defence against catastrophic loss. 

When Theory Becomes Reality 

Praise of digital transformation can seem abstract until executed in practice. In downstream oil and gas, the evidence is already visible. Where leadership has confronted the costs of inefficiency directly, optimisation tools have moved from experiment to necessity. 

Indian Oil Corporation (IOC): The Price of Interruption  

IOC quantified the stakes with brutal clarity: a single day of disruption at one refinery costs about $25 million. By placing a hard number on risk, IOC reframed digital initiatives. They were no longer ‘projects of the future’, but defensive strategies to protect present value. The company rolled out digital twins and predictive analytics across 10 refineries, generating millions in annual savings through improved stability and optimisation. The lesson is simple: when risk is priced, implementation becomes a priority. 

How Yanbu Refinery unlocked Profit, Stability, and Emissions with AI Saudi Aramco’s Yanbu refinery, built in the 1980s, is a shining example of how digital transformation can deliver not only incremental but also systemic change. The strategic integration of AI, IoT sensors, and advanced analytics delivered three outcomes at once: 

• 18% increase in throughput 

• 35% increase in profitability 

• 14% reduction in greenhouse gas emissions 

The significance lies not in the numbers alone but also in their combination. Yanbu is a lighthouse case study on how margins and emissions can move together with the help of applied AI. It is a testament to the fact that economic resilience and environmental responsibility can co-exist as parallel gains. In an industry under pressure to cut carbon while sustaining growth, this is not a minor footnote. It is a roadmap.   

The Value of End-to-End Digital Optimisation 

The examples of IOC and Yanbu prove that digital transformation is no longer a matter of speculation, but a measurable reality. When planning systems are directly connected to execution, leakage disappears and value creation compounds. 

McKinsey’s work across upstream, midstream and downstream oil and gas shows that a well-designed value chain optimisation (VCO) process, integrated across commercial, operations, and finance teams, consistently unlocks $0.50 to $1.00 per barrel. For a mid-size refinery, this translates into $30–85 million in annualised value, often captured in less than six months. What makes this compelling is not just the scale of savings, but the repeatability across geographies and levels of complexity. 

Whether in a highly flexible plant chasing market arbitrage or a focused facility prioritising stable supply, end-to-end digital optimisation aligns operations with strategy, eliminates silos, and creates a seamless decision-making loop from feedstock selection to final product placement. The message is clear: with integrated execution, digital transformation is not a “future promise” but a present and bankable source of profit. 

End-to-end optimisation can add $0.50-$1.00 per barrel, translating to a $30-85M a year value unlock for a mid-size refinery

Why Most Refineries Still Hold Vast Untapped Potential 

Many of the world’s refineries still have immense opportunities to unlock greater efficiency. Tools exist. The economics are undeniable. The case studies from India and Saudi Arabia show that embracing digital transformation can unlock tremendous business value. Still, only about 4 per cent of companies have achieved AI at scale, while around 74 per cent struggle to extract value from even modest digital initiatives. The obstacles are not technological. They are psychological. Legacy control systems, designed for a simpler world, prevent rapid adaptation to volatile crude and product slates. Data lies scattered and siloed, like a library whose books no one reads. Reliability crises bias leaders toward caution, teaching them to value continuity over transformation.

The lesson is profound – a refinery’s bottleneck is not in its pipes or processors. It is an inertia towards change management that dictates how much potential a refinery can truly unlock. 

A Boardroom Intervention for Downstream Oil & Gas 

Optimisation in downstream oil and gas is less about equipment, data, or economics, and more about governance. The tools to stabilise, optimise, and decarbonise already exist, and the benefits are proven in practice. What keeps most refineries from extracting full operational value is not the complexity of the challenge, but a bias towards preserving established habits. In this sense, the refinery becomes a metaphor for larger systems: vast institutions, built in another age, pushing to adapt to a world that has evolved rapidly. Leaders who recognise this gap and act decisively will unlock enormous business and social value. 

The lesson, while simple, is not trivial: the future of refining will be decided not in control rooms, but in boardrooms. 

The future of refining will be decided not in control rooms, but in boardrooms