Engineering Applications of Artificial Intelligence & Machine Learning
August 13 & 14, 2020
9:00 am – 12:00 pm PDT
Shahab D. Mohaghegh
CEO, Intelligent Solutions, Inc. &
Professor, West Virginia University, USA
Successful and realistic application of Artificial Intelligence and Machine Learning (AI & ML) in engineering disciplines requires engineering domain expertise. This fact challenges the current state of hypes and marketing schemes of this technology in multiple engineering disciplines.
Artificial Intelligence and Machine Learning (AI & ML) will have a flourishing future as the new generation of engineers and scientists are exposed to, and start using, this technology in their everyday life. The solution to clarify and distinguish the application of this technology to physics-based disciplines and to demonstrate the useful and game-changing applications of AI & ML in engineering and industrial applications is to develop a new generation of engineers and scientists who are well versed in the application of this technology. In other words, the objective should be to train and develop engineers who understand and are capable of efficiently applying data-driven analytics to engineering problem-solving.
The industrial revolution simulated human muscles. The industrial revolution required about two centuries to completely control and change the evolution of societies. The currently known bio-tech and AI revolution simulates human brains. This revolution requires not centuries but only decades to change our societies. Engineering-domain experts who become highly skilled AL & ML practitioners are the ones who will control the future of engineering disciplines. Becoming an engineering-related AI & ML expert practitioner requires extensive experience using AI & ML to solve engineering related problems. This will not happen in a short period, similar to the fact that becoming an expert engineer requires much more than taking the a few Engineering courses at the university.
AI and ML have a game-changing contribution to industrial and engineering-related problems. This technology will completely change the future of many industries through a transformational increase of the efficiency and accuracy of the problem solving. The contributions of AI & ML to many industries can be summarized in two classes:
Class One: Minimization or avoidance of assumptions and simplifications in order to build highly realistic models of the physical phenomena.
Class Two: Minimization of computational footprint of the numerical models such that they can act in a realistic and practical manner.
You Will Learn:
- Brief History of Artificial Intelligence & Machine Learning (AI & ML)
- Theoretical Background of AI & ML
- Artificial Neural Networks
- Fuzzy Set Theory
- Genetic Algorithms
- The Difference Between Engineering & Non-Engineering Problem Solving using AI & ML
- The Difference Between Traditional Statistics and AI & ML
- Engineering Applications of AI & ML
- Smart Proxy Modeling of Computational Fluid Dynamics (CFD)
- Contribution of AI & ML during the aftermath of Deepwater Horizon (Gulf of Mexico)
- Reservoir Simulation Models for Offshore Oilfields Using AI & ML
Verification & Validation of Industrial CFD
July 27 & 28, 2020
9:00 am – 12:00 pm PDT
Assistant Professor, IST
CFD simulations have become an Engineering tool that complements model testing. As for physical models, such capability requires the assessment of the quality of the results, which depends on the mathematical model (basin for physical models) and its numerical solution (instrumentation for experiments). This course teaches CFD practitioners to distinguish numerical and modelling errors. It presents the definitions of the different contributions to the numerical error of steady and unsteady flow simulations. Techniques to quantify numerical (Verification) and modelling errors (Validation) in industrial CFD Simulations are presented including examples from practical simulations. The course provides a framework for the establishment of the credibility of simulations so that they can be safely used for engineering decisions
You Will Learn:
You will learn to how to demonstrate the quality of your CFD simulations and evaluate the accuracy of the mathematical models behind those simulations”
9:00 - 9:30
Introduction: Errors and Uncertainties
9:30 - 10:30
10:30 - 11:00
11:00 - 12:30
Estimation of Numerical Uncertainty in steady and unsteady flows
12:30 - 13:30
13:30 - 15:00
Code and Solution Verification Examples
1500: - 15:30
15:30 - 17:00
Developing a Common Basis for Designing Unmanned Air and Underwater Vehicles
August 17 & 18, 2020 9:00 am – 12:00 pm PDT
Sumon K. Sinha, Ph.D.
According to Markets and Markets the Unmanned Aerial Vehicle, or UAV market was valued at USD 18.14 Billion in 2017 and is projected to reach USD 52.30 Billion by 2025, at a CAGR of 14.15% from 2018 to 2025. According to Globenewswire.com, the global unmanned underwater vehicles (UUV) market was around USD 2.92 billion in 2018 and is expected to reach approximately USD 7.64 billion by 2025, at a CAGR of slightly above 14.91% between 2018 and 2025.
Even though both type of vehicles enable cost effective access to environments not possible with any other conventional approach the difference in the operating media necessitate different approaches in several areas. Firstly, for UUVs the much higher density of water enables the use of buoyancy for generating vertical lift with a much smaller size vehicle. On the other hand radio waves used to communicate with and control UAVs will not work with UUVs. Underwater acoustic waves are therefore the best communication medium for UUVs. Similar types of configurations of propulsion and fluid dynamic force and moment generation can be used for both water and airborne vehicles. Hence the basic dynamics for motion and control developed for UAVs can be applied to UUVs as well. Students of this short course will learn how to utilize the significant body of knowledge available from UAV design and operations to cost effectively analyze and develop UUVs. In this unified approach methods for conceptual stage testing of entire vehicles and subsystems will be introduced. Dr. Sinha has had significant experience in UAVs, application of innovative flow control techniques to practical problems and using quick diagnostics. These can be applied to both gas and liquid flows. Some of these will also be introduced here.
You Will Learn:
- Introduction to UUVs and UAVs; Applications of Each and Role of Operating Media; Cost Drivers
- Dynamics and Control of UAV vs UUV
- Power Plant, Energy Sources, Propulsors and Efficiency of Operations
- Remote Control: Radio Frequency vs Acoustic Communications
- What features of UAVs can be advantageously incorporated into UUVs and vice versa
- Use of Scaling Laws and Techniques for Rapid Evaluation
- Future Directions
Physics-Based Data-Driven Modeling and Machine Learning: Application to Marine/Offshore Engineering
August 15 & 16, 2020
9:00 am – 12:00 pm PDT
Dr. Rajeev K. Jaiman,
Associate Professor, Department of Mechanical Engineering
University of British Columbia, Vancouver, Canada
Advances in high-performance computing (HPC) have empowered us to perform large-scale simulations for millions of variables in coupled fluid-structure systems involving complex geometries and multiphase flows. These high-fidelity simulations via coupled nonlinear partial differential equations (PDE) have been providing invaluable physical insight for the development of new designs and devices in marine/offshore engineering. Despite efficient numerical methods and powerful supercomputers, state-of-the-art computational fluid dynamics (CFD) and coupled fluid-structure simulations are somewhat inefficient hence less attractive with regard to the design optimization, parameter space exploration and the development of control and monitoring strategies for marine and offshore structures. In this short course, I will cover some of our recent developments to integrate and to complement the HPC-based high-fidelity computations with the emerging field of data science and machine learning. The primary focus of this course is: (i) to develop simple and efficient reduced-order models for the physical modeling of fluid-structure systems, and (ii) to explore the integration of projection-based model reduction with deep neural networks. A series of canonical academic and industry test cases will be covered to elucidate the integration of standard CFD with model reduction and deep learning techniques for the prediction of vortex-induced loads and motion effects. Some efforts on the iterative optimization and feedback active control of unsteady wake flow and vortex-induced vibration will be explored. Such hybrid high-fidelity CFD with a data-driven computing framework is precisely aligned with the current marine/offshore industry needs on structural life prediction, feedback control and monitoring via digital twin.
You Will Learn:
- Data-Driven Modeling of Physical Systems: Using the input and output data streams, participants will be able to build a dynamic state-space model in continuous-time and discrete-time domains for generic physical systems. For the stability analysis and control of coupled fluid-structure systems, they will be able to utilize data-driven modeling techniques such as eigensystem realization algorithm (ERA), dynamic mode decomposition (DMD), sparse identification of nonlinear systems, and Koopman operator theory.
- Model Order Reduction: Participants will be able to reduce the order of the state-space model by using model reduction techniques based on balanced realization and proper orthogonal decomposition (POD) via the method of snapshots. Using the projection-based model reduction, they should be able to build some ideas to solve unsteady flows and fluid-structure interaction problems.
- Machine Learning and Neural Networks: Participants will get some understanding of the nuts and bolts of supervised neural networks (feedforward and recurrent). They will understand selected neural network architectures (e.g., convolutional neural network) and will be exposed to current research efforts. Especially, they will acquire a general understanding of optimization strategies to guide training Deep Architectures and practical aspects of ML/DL software tools. They will get exposure to unsteady fluid flow predictions via hybrid POD/CNN and convolutional recurrent autoencoder networks.