In a software-led global industrial complex that’s going strong on digitalization, the mention of ‘big data’ has become routine. But process industries say it is ‘quality data’ and not randomly gathered digital data mountains that make the difference.
That’s the real clincher for industrial artificial intelligence, advanced analytics, and ultimately agile business decisions made by industries ranging from refining to fertilizer plants.
At the recently concluded, OPTIMIZE 26 conference in Houston, U.S., industry experts described high-quality datasets are those defined by business relevance, rather than sheer volume which makes the phrase big data a bit cliched. Getting the balance right is becoming even more critical as machine learning and agentic AI dawn on process industries.
Nuno Pacheco, senior optimization and control engineer at Repsol, and a professional deeply involved with the Spanish integrated energy company’s operations, said: “Poor quality or randomly dumped data will lead to bad insights, while perhaps smaller, strategically inputted well thought out datasets would enable more trustworthy results.
Pacheco added the industry remains only too aware of this whilst working on digital solutions, and often scoffs when it hears the phrase big data. “AI models trained on rubbish, random operations-wide data will produce rubbish results. High-quality data is crucial for effective AI, not just large amounts of it.”
Adriano Alfani, CEO of Versalis, a wholly-owned subsidiary of Italian energy supermajor Eni, said quality data gathering and its predictive analysis, in partnership with its key software vendor Emerson’s Aspen Technology business, underpins much of the company’s efficiency, safety, maintenance and sustainability drives.
But Alfani drew a direct connect with something more operationally profound in Versalis’ case. “It enabled us to respond to our operating climate in a more agile fashion. As high energy costs and other macroeconomic factors started hurting our headline base chemicals business in Europe, reliable enterprise data underpinned our subsequent transformation plan.”
That transformation involved Versalis restructuring its base chemicals operations in Europe toward biochemistry, circularity, and oilfield chemicals to ensure its competitiveness in the region.
Muhammad Zaghum Riaz, vice president at Engro Fertilizers, part of Pakistan’s Engro Group, said in a world where strategic assets such as petrochemical and fertilizer plants aim for full autonomy, the right kind of data gathering and analytics are mission critical.
“We are no exception. AI-enabled fully autonomous plants remain our ultimate goal. We embarked on this journey in 2021. Digitalization across our operations was a huge part of our group’s 10-year vision exercise triggered that year with two projects – one apiece in polymers and fertilizers using AI-enabled process control, with an emphasis on quality operations-wide data feeding into the system.”
A Multi-Billion Dollar Data Fabric Market
Software vendors appear to be relishing the global quest for quality data and related analytics feeding into autonomous pathways for process industries.
Claudio Fayad, chief technology officer of industrial solutions giant Emerson’s Aspen Technology business, said in an interview at OPTIMIZE 26 that the emphasis of quality data is here to stay, as whole industries amplify from unit-based process optimization to enterprise-wide optimization and rapidly get to grips with one critical factor. “That operational technology or “OT” data is fundamentally different from information technology or “IT” data.
“IT data often comprises of raw and unprocessed data that computers store. On the other hand, OT data is information generated by physical processes and plant assets, and prioritizes real-time operational process, telemetry and safety. We see many initiatives stumble on getting the OT-IT data treatment differentiators right.”
Unsurprisingly, Fayad sees rapidly rising business opportunities in gathering and analyzing quality OT data, and getting it right to the heart of customers’ distributed control systems or “DCS” – a computerized platform deployed to manage, and increasingly automate large-scale, continuous industrial processes.
AspenTech and its competitors’ vehicle for bringing this about remains the data fabric. In the software industry terms, it is a digital architecture that connects, automates and unifies data management across fragmented operational environments, alongside facilitating access and governance regardless of where it is stored – whether on premise or on the cloud.
The concept itself is just over a decade old, according to IBM, but has already become an “essential building block” as process industries embrace AI and automate at a canter, Fayad added.
Various forecasters put the current size of the data fabric market in the region of $3.5 billion to $4.1 billion (e.g. Fortune Business Insights and GVR), with energy, petrochemicals and pharmaceutical industries accounting for a sizeable chuck of this.
This market could potentially grow to $16 billion to $19 billion by the middle of the next decade, with a compound annual growth rate in excess of 20%.
Eyeing this space, AspenTech launched its enhanced Inmation OT data fabric product at OPTIMIZE 26. “It is at the core of how we unify and contextualize industrial data, and its new capabilities significantly strengthen that foundation,” Fayad, said.
“Inmation’s latest enhanced version will make the data fabric more scalable, flexible and enterprise-ready. In doing so we have created the technical underpinning required of our wider industrial data platform to support analytics, AI‑driven workflows and increasingly autonomous operations over time.”
AspenTech’s offering under Emerson’s corporate umbrella will compete directly with ABB’s Genix, Cognite’s Data Fusion, and others in the market including the likes of Honeywell, which offers proprietary data fabrics for specific industry segments.
Fayad admitted its a competitive yet lucrative market where vendors like AspenTech have to constantly innovate. “We believe we have the right mix. Our latest release simplifies how industrial data environments are deployed, scaled and secured.
“A new distributed node‑based architecture replaces rigid components with a modular foundation that delivers consistent behavior across sites while reducing operational complexity.”
It all goes back to the trend of industries graduating from unit or site-specific optimization to doing so across the enterprise, he added.
“It is why we have specifically designed a data fabric product that enables customers to expand from individual plants to global deployments using a common operating model with centralized security, governance and lifecycle management. We remain convinced that’s where the market is heading.”
With a 20%-plus CAGR market growth being forecasted, and the adoption of data fabric products by process industries deemed just below adoption levels by the financial services sector, things can only get bigger in the quest for quality data.
