Industry as we know it is changing. Technological developments such as advances in internet connectivity, data analytics, artificial intelligence and machine learning are revolutionizing refinery operations. Refiners can track production operations second-by-second with reliable, networked digital equipment. Sensors allow companies to monitor assets and vehicle locations, both globally and within a facility, allowing more efficient route-planning and company operations. Powerful artificial intelligence algorithms can be used to optimize a production process for different strategic goals, allowing companies to shift focus intelligently and confidently.
Within the energy industry, these changes have given rise to the smart refinery. A smart refinery is a refinery that use plant digitalization, data-driven decision making and extensive simulation in their refining operations.
Chemical refineries have been a part of the industrial landscape for over a century. Petroleum refineries are practically synonymous with manufacturing and images of fractional distillation equipment are used to communicate the concept of industry in visual media. With much of the world relying on the array of petrochemicals they produce through complex processes drawing on costly equipment, it’s no wonder that refineries have been designed to be robust. For example, a distillation column is designed so that as long as it has a steady temperature and sufficient feed, no further adjustments are needed to separate a non-ionic chemical mixture according to constituent boiling points. This type of stability and equipment reliability is one of the factors that allows refineries to continuously operate.
As well-planned as this arrangement may be, however, it does not inherently capture data. The energy input and product yield must be measured separately, and unless something goes wrong, the processing may as well take place in a black box. In contrast, a smart refinery has the entire process wired with sensors, so the operation is constantly generating data. Smart refineries easily monitor a host of factors that provide the company useful information, such as energy use, temperatures at every stack in the distillation column, and the volume, density, and even the purity of the outputs.
The information coming from networked sensors can be used to create digital models through refinery modeling using process simulation software. These models serve as faithful, living representations of the refinery, allowing teams to analyze the impact of proposed changes digitally before implementing them in the real world.
The flood of data plant digitalization technology generates isn’t just a bonus of smart refinery practices; this torrent of information actually drives decision-making. A company that is accustomed to adjusting a process, waiting for information about how those changes affected production, then making new adjustments, will initially use this information to simply accelerate their existing decision-making process. Yet this approach fails to tap into many of the benefits of the industrial internet of things.
The new data coming in may satisfy the desire to validate if targets are being met, but to truly become a smart refinery, facilities must rely on data to inform their decisions across all aspects of production.
Demonstrating the impact of changes via software tightens the decision loop significantly by allowing companies to explore new parameters without changing real world assets, but does not use information to its full potential. Some of the most effective refinery optimization and value chain optimization strategies let artificial intelligence software identify what changes to make to the refining process to achieve specific business goals.
Machine learning and artificial intelligence algorithms crunch enormous volumes of data to create abstract models of the underlying processes. The models are known as neural networks, as they are modeled on how animal neurons are wired together. By iteratively adjusting the various weights each neuron is given, as well as which neurons it should communicate with, the software is able to discover links in the data.
For example, it would quickly become clear to both a human observer and an artificial intelligence algorithm that when the flow rate of the energy source increases, the temperature in the distillation stack increases. Where artificial intelligence excels is in finding the subtle connections between seemingly disparate pieces of information. For instance, perhaps a small wobble in the pressure reading on a valve is actually a sure sign that the valve will fail. A human operator looking at the design parameters might not notice or consider the disturbance noteworthy, let alone connect it to a failure that occurs weeks later. But an AI for oil and gas solution will figure out if it is consistently associated with another phenomenon.
A smart refinery can actually self-optimize by letting AI control the parameters of the physical assets. Programmed with the expectations and goals that the organization has for the refinery process under control, the algorithm can adjust refinery equipment and processes to meet those goals. Utilizing the neural network the algorithm built by analyzing operations data, AI can automatically optimize a smart refinery.
How can a smart refinery help a company comply with environmental goals?
Equipment failure and exceptional circumstances account for more than a third of total methane emissions in the United States every year. By monitoring asset performance in real-time and analyzing the data, companies can identify failure profiles that result in unintended releases of greenhouse gasses.
Can a smart refinery predict what will happen in the petroleum markets?
The commodities markets are as closely scrutinized as the stock and futures markets, where any inefficiency is quickly exploited. A smart refinery can make sure a company is nimble enough to keep their operations aligned with market conditions without requiring a slow chain of command to translate market information into refinery operations.
How does a smart refinery relate to smart enterprise?
A refinery can be operated as a smart refinery within a traditional company. This may serve as a test for the larger adoption of smart enterprise, or it may remain an isolated case of data-led production within the organization. A refining operation within a company that uses smart enterprise will be a smart refinery.
The Self-Optimizing Plant-A New Era of Autonomy Powered by Industrial AI
Hybrid Modeling AI and Domain Expertise Combine to Optimize Assets