AI: How AI and machine learning benefit refineries and petrochemical plants
Embedding machines with artificial intelligence (AI) allows them to learn from their past experiences.
Embedding machines with artificial intelligence (AI) allows them to learn from their past experiences. This means that machines at a petrochemical plant can use past data to predict future events. These predictions could include equipment maintenance or future disruptions that may occur at the plant. One of the best uses of AI is enabling computers to optimize plant operations based on historical trends. The overall automation of refineries helps plant managers minimize downtime and provide better returns on investment for the owners.
Petrochemical companies and refineries around the globe are partnering with leading research institutes to introduce machine learning at their plants. The following are case studies from top chemical engineering companies that explain the importance of AI and machine learning in this sector.
BASF
This German producer of paints for commercial fleets and collision repair recently partnered with the Massachusetts Institute of Technology (MIT) to implement the latest machine learning techniques at its plant in Germany. By utilizing software that uses neural networks to make efficient decisions, the company can prepare the right compositions of paint as per their clients’ requirements.
The task of preparing a color that exactly matches their client’s car is a time-consuming and complex process. However, due to machine learning, this process is automatic, thereby reducing the turnaround time and service quality at the plant.
BASF has also partnered with Hewlett Packard to develop the world’s largest supercomputer in the petrochemical industry. This machine will allow engineers to perform complex modeling and virtual experiments to improve the design of new polymers.
The Dow Chemical Co
This company has partnered with 1QBit, a Canadian-based firm that specializes in machine learning techniques. Through this collaboration, Dow Chemical wants to build quantum computing tools that will help in chemical research. These AI tools would be a dream come true for scientists who wish to perform complex experiments to improve their products.
Dow Chemical has also collaborated with TeselaGen to design cutting-edge experimental software to aid microbiology researchers in finding new ways to protect crops and increase agricultural growth at farms.
Shell
The company has been successful in introducing machine learning at its workplaces in several ways, including:
- Virtual assistants (VAs)—Operating in more than 151 locations, VAs can speak in multiple languages, including Chinese, German and Russian. They learn from their experiences with customers and optimize their systems to cater to future clients. Some of the tasks that these VAs can perform include giving directions to local stores and providing technical data sheets. According to sources at the company, VAs can also reduce call volumes to live agents by 40%, which is a great example of how using AI can reduce the workload on company staff.
- Smart downstream solutions—Shell’s Downstream Commercial business, which is responsible for supplying oil and gas to the end consumer, is using AI solutions to:
- Predict consumer demand for petroleum
- Measure supply shortages, if any
- Recommend a correct mix of oil for a refining process.
How is machine learning made possible in the downstream processing industry?
Processing plants are complex structures that use a variety of chemical components to give consumers their desired products. To introduce machine learning and AI at these facilities, organizations use the most complex computing techniques available, including neural networks and quantum computing.
Neural networks. These are a sequence of decision threads that take data as input. At every level, each thread performs complex calculations to arrive at a decision for selecting the next thread. Each level of decision thread is one step deeper and narrower than the previous thread. Hence, the inputs reach a final decision as they flow through the thread sequence.
Quantum computing. No matter how powerful they are, computing machines read a code in the form of 0s and 1s. This binary format of computing has been in use since the first computer came into existence. However, quantum computing has the capability of revolutionizing the way computers perform operations. Quantum computers use technology that can read data in the form of 0, 1 or both. This means that there can be an uncountable number of permutations and combinations of binary units—making the computing process faster than ever before.
Takeaway
Machine learning is needed for companies in the petrochemical and refining sector, as it allows companies to minimize downtime, reduce costs and maximize production levels—benefits that help companies stay ahead of their competition. HP
The Author
Welsh, T. - AX Control, Fuquay Varina, North Carolina
Taylor Welsh is a writer/programmer for AX Control, a GE Speedtronic control system reseller located in Fuquay Varina, North Carolina.
Related Articles
From the Archive