This reference text will provide a systematic review of state-of-the-art AI technologies, high-performance computing and their applications in process engineering. It will introduce the development of traditional process simulators in the field of process engineering and new numerical solvers based on data-driven and physics-informed neural networks approaches.

It will also cover applications in science and engineering, including but not limited to multiscale modelling and optimal design problem in reverse osmosis desalination, model-based experimental analysis in enhancement of boiling heat transfer, and inverse identification of high-resolution spatiotemporal contaminant source distributions in atmospheric pollution.

Chapter 1: Artificial Intelligence: from Predictive to Prescriptive and Beyond

Author: James H. Chappell, VP, Global Head of AI & Advanced Analytics, AVEVA (United States)

This chapter will discuss, from an industrial software perspective, how state-of-the-art artificial intelligence technologies improve industrial processes, proactively detect, and solve problems, and provide guidance for risk-based decisions, resulting in significant cost savings and improved competitiveness for the enterprise.

Chapter 2: Artificial Intelligence and Process Engineering

Author: Dr. Nariman Piroozan, HPC AI/ML, Intel Corporation (United States)

In recent years, the application of artificial intelligence towards long-standing engineering problems has experienced a boom thanks in large part to tremendous increases in computing power. Utilizing various AI solutions such as deep neural networks, graphical neural networks, and so forth, can help to improve accuracy of solutions as well as to decrease time to solution across the spectrum of fields within this domain.

Chapter 3: The Rise of Time-Travelers: Are Transformer-Based Models the Key to Unlocking a New Paradigm in Surrogate Modeling for Dynamic Systems?

Author: Dr. Joseph Kwon, Artie McFerrin Department of Chemical Engineering, Texas A&M University (United States)

This chapter will provide an overview of surrogate modeling and its importance in control, optimization, and monitoring applications. The recent advances in surrogate models, including recurrent neural network (RNN), convolutional neural network (CNN), long short-term memory network (LSTM), and transformer-inspired natural language processing (NLP) models, will be discussed. The performance of the developed TST models will be compared with other existing machine learning approaches on non-trivial systems.

Chapter 4: Optimization-based Algorithm for Solving Inverse Problems of Parabolic Partial Differential Equation

Authors: Yi Heng, Chen Wang, Qingqing Yang and Junxuan Deng, Sun Yat-sen University (China)

Inverse problems of partial differential equation have been widely applied in science and engineering. When direct measurements are not practical due to environmental barriers or high experimental costs, it is more efficient and economical to use inverse solution techniques to estimate the unknown values of indirect observations. This chapter will introduce the idea of optimization-based algorithm for solving the heat conduction inverse problem and the practical procedure, which is generalized and has been successfully applied to heat flux estimation for pool boiling, steel cooling, and chip heat dissipation.

Chapter 5: Deep Learning-based Approach for Solving Forward and Inverse Partial Differential Equation Problems

Authors: Yi Heng, Jianghang Gu and Xie Guohong, Sun Yat-sen University (China)

The forward and inverse problems of partial differential equations in complex systems are widely used in multidisciplinary fields, such as atmospheric science, earth science, and chemical engineering, which have important application potential and academic research value. In this chapter, both established and novel deep learning-based methods for solving forward problems will be illustrated.

Chapter 6: Efficient Control Invariant Set Approximation Algorithms: A Graph-based Approach

Author: Dr. Jinfeng Liu, Department of Chemical & Materials Engineering, University of Alberta (Canada)

Increasingly faced with sustainability and profitability objectives, chemical process plants have become very complex with many operating constraints. To achieve these objectives, a high level of automation is required. Unfortunately, the ability of the automated controllers to ensure that the control objectives are met at all future times is complicated by constraints on the available control energy and uncertainties in the control system. This chapter will present efficient control invariant set approximation algorithms developed based on graph theory.

Chapter 7: Active-subspace-based Swarm Intelligence Method with its Application in Optimal Design Problem

Authors: Jiu Luo, Ke Chen, Junzhi Chen and Junxuan Deng, Sun Yat-sen University (China)

Today's model of the physical world could be very complicated and described by plentiful factors due to the rapid development of computing power. Directly solving these sophisticated high-dimensional model constrained optimal design problems is not only extremely time-consuming, but also very difficult to converge. This chapter will focus on the dimensionality reduction optimization method, namely supervised active subspace approach, where the searching process of intelligent swarm would become more efficient.

Chapter 8: Multiscale Modeling and Optimal Design of Reverse Osmosis Desalination Systems using Machine Learning And High Throughput Computing

Authors: Jiu Luo, Sun Yat-sen University (China) and Xiaojuan Tang, Guangzhou City Polytechnic (China)

This chapter will present optimal design of brackish water and seawater reverse osmosis desalination systems following an interdisciplinary approach, utilizing artificial intelligence, computational fluid dynamics, field synergy principle, dissipation extremum principle, global optimization, and high-performance computing. The developed model-based prediction and optimization tools can be extended to guide the design and optimization of other membrane-based processes.

Chapter 9: Size-Controlled Continuous Manufacturing of Perovskite Quantum Dots through Multiscale Modeling and Optimal Operation of Millifluidic Synthesis

Author: Dr. Joseph Kwon, Artie McFerrin Department of Chemical Engineering, Texas A&M University (United States)

Inorganic lead halide perovskite quantum dots (QDs) are a promising semiconducting nanomaterial for various applications. The optoelectronic properties of colloidal QDs depend primarily on their size, so it is essential to fine-tune their size while achieving fast and continuous production. However, the use of batch reactors, which suffer from mass and heat transfer limitations and batch-to-batch variations, hinders precise control over the size-dependent optoelectronic properties of QDs. To address this gap in knowledge, we propose a multiscale model for continuous flow manufacturing of colloidal perovskite QDs.

Chapter 10: Supercomputing-based Inverse Identification of High-Resolution Atmospheric Pollutant Source Intensity Distributions

Authors: Mingming Huang, Junzhi Chen, Yinan Han and Yi Heng, Sun Yat-sen University (China)

Atmospheric pollution research plays an important role in monitoring emissions of pollutants by industrial processes, understanding the behavior of pollutants in the atmosphere and developing strategies to reduce their impact on the environment and human health. Considering mathematical ill-posedness and high computational costs, a high-throughput parallel computing framework is established for solving the nonlinear inverse problem of estimating high-resolution spatiotemporal atmospheric contaminant sources distributions. The computational tools and methods are of great theoretical and practical value for designing environmentally friendly industrial processes, exploring the mechanism of pollutant degradation, and revealing the causes of global climate change.

Chapter 11: Enhancing Boiling Heat Transfer via Model-based Experimental Analysis

Authors: Min Hong, Yi Heng and Dongchuan Mo, Sun Yat-sen University (China)

Boiling heat transfer is an efficient energy transfer method and can be used to develop high-heat-flux dissipation key technologies with small volume and strong temperature-control ability. The experimental and numerical studies of pool boiling have been a very active field of research for many decades. In this chapter, two heuristic modelling approaches for model-based experimental analysis will be respectively proposed to mimic the dynamic boiling processes.

Chapter 12: Conclusions and Outlook

Authors: Mingheng Li, California State Polytechnic University (USA) and Yi Heng, Sun Yat-sen University (China)

This chapter will present some closing thoughts on the concepts and applications of HPC and AI/ML in process engineering covered in the preceding chapters of this book. It will also provide several promising research avenues that will hopefully help advance this promising field.