Published Originally in Chemical Industry Digest, September 2023
Abstract
This article highlights the transformative impact of Artificial Intelligence (AI) technology in the chemical industry, enabling significant energy efficiency improvements of 20% to 40%. Through case studies, it illustrates how AI-driven digital twins and optimization algorithms are driving sustainability and carbon footprint reduction efforts, making it a crucial investment for chemical companies.
Introduction
The chemical industry is one of the oldest and largest manufacturing industries in the world. It is but natural that all technological advancements in the past have found their place in this industry to improve the process control, safety and productivity. The adoption of AI however has been relatively slow. With increased competition and dwindling profit margins, chemical companies are looking at every opportunity to improve their energy efficiency and become cost competitive. Another critical aspect is the thrust on Sustainability and carbon footprint reduction. AI/ Digital Solutions that can provide improvement in above two areas are now widely being explored.
Chemical Industries are energy intensive, hazardous, risk to environment and highly complex. The high safety risk in the process has made companies tread carefully in automation technologies. Most chemical companies have some level of automation. They have a robust DCS system, that can capture the process data from the plant and can also execute controls in the field including safety control actions. The industry need is of an AI Technology that can build on top of this infrastructure and start giving benefits immediately on deployment.
There is a need for manifold interventions from multiple angles at a low cost to save on margins and increase profitability keeping quality and safety in mind. One such powerful yet niche field is advent of artificial intelligence (AI). AI, automation and optimization go hand in hand. AI mimics decision taking and problem-solving capability using high end Deep Learning algorithms, the crucial requirement of course being data.
Situation Analysis and Importance of AI
Many chemical companies in India are still conventionally running with data captured at frequent intervals but not being tracked and stored properly. Few of the companies have good data capturing system but are not storing data. Hence the challenge for AI companies is to get quality data with high granularity for them to build models, which can be immensely useful for their operations. A hybrid model of first principles chemical-based process modelling combined with data-based machine learning models can bridge the gap of process data not being available to great extent. Through this a Digital Twin of the process can be built. This is an accurate prediction model that predicts the process behaviour in the next instance and in the future. Once the future state is known, powerful AI models can work on the predicted state and build an optimization strategy for achieving the energy efficiency and in turn, reduce carbon footprint.
Effective Digitalisation – A Pre-Requisite
It is critical for chemical companies to build their digital transformation infrastructure and adopt AI into their routine practice. It brings about seamless uninterrupted automated execution bringing about energy optimization thereby saving cost. Digital Transformation infrastructure should
(a) Sense i.e. real time capture of high granularity, high coverage, high quality, high variety, high volume of all relevant monitoring and controllable data points, operational, health, energy of process and utility equipment data and process, production and quality data.
(b) Connect i.e. real time transmission of upstream and downstream data with no cost towards control and network cables and
(c) Act i.e. real time actuation of control commands, along with centralized location for real time and historical data storage, real time data processing for business intelligence and report generation &highly interactive, intuitive visual analytics. The level of Digital Transformation infrastructure will depend on the current digital state of the plant. The beauty of centralized data repository is that it can store all the data that is generated from the process, and can provide integrated energy dashboards and reports customized to the end stakeholder like a Plant Head, Production Head, CXO’s and CEO.
“A typical road map for deployment of AI solution in chemical plants needs an initial assessment of the plant by AI/ chemical experts where the current state of the plant like existing data capturing system, data storage policy, existing process controllers in place like PLC, the control system architecture and process constraints are assessed”.
The AI models deployed on the process plant learn continuously from the real time data and improve their performance over the years. Thus, the Plant continues to benefit and can see reduction in energy consumption, and achieve its sustainability goals. Thus, it the right time now for all chemical companies to invest in AI technologies which can provide both the Digital Infrastructure and the AI technology that can make a huge impact in the way the plant operates and brings in all round benefits. The early adopters of this technology stand to gain immensely and will have a great first mover advantage in this highly competitive market.
Case Study 1: AI in Electrolysis Plant.
Electrolysis plant is an energy guzzler and companies are looking at ways to bring down energy costs in this unit. AI/ IoT technology can bring in energy efficiency in the process without violating the constraints and operating limits. Digital Twin of the Electrolysis process can give accurate predictions for the in process and output parameters like chlorine & hydrogen pressures, caustic flow rates and concentration, differential pressure across ion exchange membrane. The Reinforcement Learning Agent Optimization Algorithm will use the predicted values and give the best combination of control parameters like voltage applied, brine flow rate, brine temperature to bring power reduction for the same output values. The results are savings of at least 30% in power consumption.
Deployment of AI in Chemical Industry
A typical road map for deployment of AI solution in chemical plants needs an initial assessment of the plant by AI/ chemical experts where the current state of the plant like existing data capturing system, data storage policy, existing process controllers in place like PLC, the control system architecture and process constraints are assessed. The baseline figures of energy consumption, production, yields, maintenance costs are recorded. This assessment helps in finalizing the digital transformation infrastructure requirements for the plant.
The number of IoT devices and the wireless network architecture are finalized, and Big Data Lake Enterprise Software Platform for Data management is architected on the cloud/ on premises server.
The plant design data and historical operational/ process data is collected from the plant for model building. Through design data and first principles, chemical-based modelling is done and for historic data neural network deep learning data-based modelling done. The integration of the above two models gives the operational digital twin of the plant. This is a predictive model and replicates the physical plant environment with accuracy greater than 95%. On top of this model, for optimisation, deep reinforcement learning agents are trained for bringing about the energy, production efficiency improvement. Once these models are verified and tested, they are ready to be deployed. These models are deployed on the enterprise software platform hosted on the cloud/ on premises server.
Case Study 2: Ammonia Recovery Plant.
Ammonia Recovery uses both low pressure and medium pressure steam. Bringing down steam consumption can immensely reduce costs in this unit. AI/ IoT technology can bring in energy efficiencies the process without violating the constraints and operating limits. Digital Twin of the ammonia process can give accurate predictions for the in process and output parameters like Ammonia recovered, Column Temperature, Column Pressure, Ammonia in Column bottom. The Reinforcement Learning Agent Optimization Algorithm will use the predicted values and give the best combination of control parameters like LP Steam flow, MP Steam Flow, Cooling water flow, Lime addition to bring Steam reduction for the same output values. The results are savings in the range of 40% in Steam consumption. This has direct effect on the reduction of fuel in boilers.
Case Study 3: Evaporator Plant
Evaporators are used in many chemical industries as a part of the chemical recovery process. There are two types co-current and counter current evaporators and they usually have multiple effects/calandria. It uses low pressures team to improve the concentration of the feed liquor. Optimizing steam consumption can immensely reduce costs in this unit. AI/ IoT technology can bring in energy efficiency the process without violating the constraints and operating limits. Digital Twin of the evaporator process can give accurate predictions for the in process and output parameters like product concentrations in each effect, temperature and pressure values across each effect, condenser flow rates, product flow rates, something not being currently measured. The Reinforcement Learning Agent Optimization Algorithm will use the predicted values and give the best combination of control parameters like LP Steam flow, Steam Pressure, Vacuum Flow to bring Steam reduction for the same output values. The results are savings of to the tune of 20% in Steam consumption. This has direct effect on the reduction of fuel consumption in boilers.
Clearly advancements in AI Technology are impacting the energy efficiencies of chemical industries from 20% to 40% and thus help companies in their sustainability journey for achieving carbon footprint reduction targets.
Conclusion
In conclusion, there can be no doubt that the relentless march of AI is poised to revolutionize the chemical industry, offering a beacon of hope for better energy management on a global scale. AI’s ability to optimize chemical processes, predict energy consumption patterns, and identify energy-saving opportunities is nothing short of remarkable. It promises to break down the barriers that have long hindered efficient energy management in the chemical sector. By harnessing the power of machine learning, we are equipped to reduce waste, cut emissions, and drive innovation at an unprecedented pace. The chemical industry must wholeheartedly embrace AI, not as an optional luxury but as an essential tool for survival and progress. The future belongs to those who harness AI’s potential to transform chemical processes, ushering in an era of sustainable energy management.
