Embrace Industry 4.0: Uncover substantial opportunities with AI-based, real-time production optimization across hydrocarbon value chains

Jennifer Savard's picture
Jennifer Savard, Program Manager, Natural Resources, IBM Industry Platform Unit, IBM Canada
Comments: 0

Understanding, leveraging and applying Industry 4.0 principles to oil and gas facilities is important to an organization’s performance, resiliency and adaptiveness.  In today’s volatile environment, managing production infrastructure and supply chains in an adaptable and secure manner is crucial to remain competitive and deliver business value.

The convergence of technologies is a much talked about element of Industry 4.0.  The rapid proliferation of massive volumes of data and technologies like Artificial Intelligence (AI), the Internet of Things (IoT), Cloud, blockchain etc., and the effects it has on an organization’s culture, people, processes and models of engagement is truly transformative.  

Industry 4.0 interconnects and converges IT (Information Technology) and OT (Operational Technology).

Selecting, implementing and ultimately harnessing digital platforms, business intelligence layers and complex data science fabrics into data-driven insights to impact the bottom line is a critical effort across IT and OT business lines.  Plant operators and decision-makers, industrial engineers, asset and maintenance personnel and schedulers all need data insights to manage continuous performance of highly complex operations - that are non-linear in nature - where changes and disruptions on one end adversely affect the next link on the chain or several links downstream. 

Complex hydrocarbon plants have various software packages for asset performance, quality, scheduling, inventory, ERP etc. along with supporting infrastructure, as well as additional customized and best-in-class plant management technologies.  Common challenges are data silos, questionable integrity of structured information, reduced visibility into the factors and causal relationships to reduce downtime or maximize throughput and quality1, which ultimately leads to missed plan performance and process optimization opportunities across production processes.     

A first step toward realizing the Plant 4.0 promise is the critical step in identifying a high-value use case.  Once selected, the goal is to achieve proven value in steps, iterate quickly, apply agile sprints, test, fail and pivot when necessary to iterate again.  This staged value creation is crucial to scaling the Industry 4.0 advantage.   Selection of a high-value, high-impact use case should be a comprehensive approach to innovation and business transformation.  It should bring designers and developers together with business and IT stakeholders to scale ideas through agile and Design Thinking practices, including mapping out current vs. future state scenarios, to build cohesive alignment around a set of key use cases that are fundamental to your business value.  The IBM Garage™ method is one example. 

Identification of a high-value use case targeting plant operations that is directly coupled with efficiency, throughput and availability of processes and assets is an excellent value case.  IBM recently worked with one of Canada’s largest integrated energy producers on an Industry 4.0 project which set out to apply the latest cutting-edge data science methodologies, artificial intelligence (AI) and dynamic optimization techniques to significantly improve performance of a complex continuous production process across several interconnected facilities in upstream operations.  This high-value use case was selected for its business impact and significant complexity to prove robustness and reliability of the technology solution.

The goal of the project was to provide a real-time, AI-enabled decision support system that achieved optimal production across multiple interconnected and interdependent facilities while minimizing sub-optimization driven by individual plant performance indicators. This was achieved by opening-up operational data siloes, enabling prediction of upcoming process upsets and rapidly recommending practical remediation action, i.e. setpoint changes leading to either upset avoidance or rapid recovery of production.  Another important function performed by the solution project was the continuous and constant monitoring of the on-going steady state process to enable the identification of production improvement opportunities and to provide real-time production schedule changes to maximize business value.

Harnessing advanced technologies like AI to create digital advisors that are designed to automate the complex data science involved in machine learning based analytical models is necessary to tackle the challenge.  Digital advisors are often capable of outperforming even the most knowledgeable experts and have the added benefit of embodying knowledge regarding processes or assets in a replicable form. 

The scale of the project included a 130 Km2 area with over 35 individual plants which spanned over  130,000 individual asset tags operated by thousands of front-line operators, engineers and contractors.   The solution – Production Optimization Advisor – employed over one hundred machine learning models in a multi-layered approach to achieve predictive capabilities, optimization models and opportunity awareness including over 58 variations of process upset flagging models.1 

Technologies underpinning projects of this nature, include Open Source Technologies, advanced applications of AI, IoT data integration and Cloud technologies which can be public, private or on-premise hardware as well as data-driven digital twin models of the full system. Additionally, innovative UX design features are deployed to enable anywhere, anytime access to production optimization insights across devices.  

The production optimization advisor enables plant decision makers to generate optimal production schedules and create new plans in less than 10 minutes.  Plant operators are placed in full control of process upset management and opportunity awareness, enabling them to utilize AI-powered predictions and minimize plant upsets, taking action to optimize processes, quality, volume and inventory levels for improved profitability and energy use.  The solution currently predicts upsets 20 minutes beforehand with more than 84% accuracy. 

Value realization on Industry 4.0 projects requires a true commitment by executive leadership, identification of the right organizational stakeholders and selection of a high impact business case.  A willingness to manage risk adversity, accept operational project hurdles as well as challenge data availability and accuracy is also necessary.   The pay off is 2X value in less than 14 months and approximately ~2% production improvement across the end-to-end hydrocarbon value chain. 

The future holds many opportunities for extending and realizing advanced process optimization across upstream, downstream and process manufacturing industries to push the current boundaries of Industry 4.0 and advanced AI and digital twin projects in Energy and beyond.