The evolution of process optimisation
Published by Callum O'Reilly,
Senior Editor
Hydrocarbon Engineering,
Launched in 2015, the World Economic Forum initiative has determined digitalisation as an indisputable requirement, resulting in the oil and gas industry taking a fresh look at its boundaries. The pandemic has accelerated digitalisation, and companies quickly learned that they had to become more agile to respond to significant disruptions and demand fluctuations when the COVID-19 crisis diminished demand for oil and gas almost instantly.
At present, the oil and gas sector must be flexible enough to respond immediately to the changes in feedstock product demand as a result of the changing global economy. The hydrocarbons industry has become more sophisticated, with heavy investment in cleaner fuel production, molecular recycling, plastic waste reduction efforts, and more efficient and environmentally-friendly production methods. Economic process optimisation is now a fundamental requirement and the only solution to survive in competitive markets.
In order to fully exploit the economic optimisation potential of a technological process, the approach of working in the paradigm of optimising control should be taken, as opposed to addressing optimisation and control problems as two separate tasks. This article demonstrates how the trend for convergence of process optimisation and multivariate control has evolved, and discusses the computational roadblock that is standing in its way. It also explores how artificial intelligence (AI) can overcome these constraints without compromising the process modelling fidelity.
Multivariate process control
The first computer usage for multivariate process control can be traced to the 1959 deployment of the RW-300 computer by Ramo-Wooldridge Co. at the Texaco Port Arthur refinery polymerisation unit in Texas, US. The project included the digitisation of sensor readings and control outputs. The algorithm approximated the economically-optimal process parameters using the measurement from the sensors. The implemented algorithm performed multivariate control, but a significant component was missing.
The next major step in multivariate process control was to consider the delay between a change in the process parameters and the delayed response of the process. This technology was named Model Predictive Control (MPC), and its commercial deployment began in the 1970s. The first industrial application of MPC algorithms could be traced back to Shell Oil’s internally-developed controller in 1973. The linearised model of the dependencies between the manipulated and controlled variables and a quadratic form of the optimised cost made its implementation feasible with the computers available at the time. This notation of the optimisation problem allowed for a straightforward and achievable solution...
Written by Gregory Shahnovsky, Ariel Kigel and Gadi Briskman, Modcon Systems
This article was originally published in the April 2022 issue of Hydrocarbon Engineering magazine. To read the full article, sign in here or register for a free trial subscription.
Read the article online at: https://www.hydrocarbonengineering.com/special-reports/19042022/the-evolution-of-process-optimisation/
You might also like
Kinaxis and ExxonMobil announce co-development deal
Kinaxis® has announced a co-development deal with ExxonMobil to create supply chain technology solutions designed specifically for the energy sector.