EUROGEN 2017 will be the 12th of a series of International Conferences
Multi-Objective Design Exploration - Fusion of Optimization and Data Mining
Prof. Shigeru Obayashi
Director, Institute of Fluid Science | Tohoku University | Sendai, Japan
Optimization is a powerful computational tool. The optimization problem can be mathematically defined by design variables and design objectives. If multiple design objectives are specified, design space can be regarded as a multi-objective optimization problem. Among the multiple design objectives, conflicting objectives are essential and they will provide Pareto-optimal solutions, which form a Pareto front, the key feature in the design space.
The shape of the Pareto front tells various aspects of design tradeoffs. Multi-Objective Design Exploration (MODE) has been developed to reveal the structure of the design space from trade-off information and to visualize it as a panorama for a decision maker. Data mining techniques are very useful for extracting design features from the design space. The process represents the fusion of optimization and data mining.
We consider a designer often needs a map of design space to draw an inference to the best design. If computational tools create a Bird’s-eye view of the design space, the result literally provides the map that support design activities. In this presentation, the collaboration with Mitsubishi Heavy Industries will be reviewed to reveal how MODE assisted the design activities to develop Mitsubishi Regional Jet (MRJ). In addition, other industrial applications of MODE in Japan will be reviewed.
Professor Shigeru Obayashi is a full professor and director of Institute of Fluid Science (IFS) at Tohoku University, Japan. His research interests include Computational Fluid Dynamics (CFD), Multidisciplinary Design Optimization (MDO), Multi-Objective Design Exploration (MODE), Evolutionary Computation, Data Mining, Experimental Fluid Dynamics and Data Assimilation.
Prof. Obayashi received his PhD degree in Engineering at University of Tokyo in 1987. He was a senior researcher at NASA Ames Research Center and helped to develop the three-dimensional unsteady Navier-Stokes numerical algorithms. He joined the department of aeronautics and space engineering at Tohoku University since 1994, promoted to a full professor in 2003 and served as the director of Transdisciplinary Fluid Integration Research Center under IFS from 2008-2013. He is Ex-President of Japan Society for Evolutionary Computation, associate fellow of AIAA, fellow of JSME, Japan Society of Fluid Dynamics and Japan Society for Aeronautical and Space Sciences (JSASS), and currently President Elect of JSASS. He received several research awards, including JSME’s 2007 Funai award and The Commendation for Science and Technology by the Minister of Education, Culture, Sports, Science and Technology, 2014.
Adjoint-based optimization of wind-farm control in large-eddy simulations
Prof. Johan Meyers
Department of Mechanical Engineering, University of Leuven, Leuven, Belgium
Modern wind farms grow in size: Europe’s largest operational offshore wind farm today is the London Array with 175 Siemens turbines for an installed capacity of 630MW. Recent offshore tenders such as Borselle I & II in the Netherlands, surpass the GW size. Onshore, wind farms are even larger, with the largest farms nowadays being built in the Gansu province in China. At these scales, wind farms interact with the atmospheric boundary layer through the accumulated effect of turbine wakes. As a result, the efficiency of individual turbines in the farm can drop drastically (up to 80% in some extreme cases).
Turbulence plays an important role in the interaction of a wind farm with the atmospheric boundary layer, and dominates the transport of momentum towards the turbines. Moreover, turbulent scales directly interact with the operational time scales in a wind farm. Typical time scales of large turbulent eddies in the boundary layer are in the order of 10 to 100 seconds. This overlaps with dominant wind turbine and wind farm scales such as blade revolution and turbine-to-turbine advection times. Since turbines can be dynamically regulated in time and throughout the farm, there are a lot of degrees of freedom for the wind farm to interact with the turbulent atmospheric boundary layer.
We investigate the optimization of these control degrees of freedom with the aim of increasing overall wind-farm efficiency. To this end, adjoint-based optimization of wind farm control is considered in large-eddy simulations (LES), allowing us to explore new mechanisms in which wind farms can influence their turbulent environment. Controls are optimized in a receding horizon control framework. Two main control mechanisms are considered: induction control, in which the turbine thrust is dynamically regulated in time, leading to increased wake mixing, and yaw control, in which the turbine yaw angle is dynamically adapted, leading to wake redirection.
All simulations and adjoint-based optimizations are performed using SP-Wind, which is a dedicated LES and DNS code developed at KU Leuven over the last decade. The code is based on pseudo-spectral discretization in horizontal directions, and a fourth-order energy-conserving finite-difference discretization is vertical direction. Inflow turbulence is generated using a concurrent-precursor method, and wind turbines are represented using different turbine models, ranging from simple actuator disk models to fully nonlinear flexible multi-body actuator line models.
This talk is sponsored by CCS:
Johan Meyers is an associate professor at the department of Mechanical Engineering of KU Leuven (Belgium). He heads a research group on the simulation and optimal control of turbulent flows, and a lot or research efforts in recent years have focused on wind energy applications.
Before he joined KU Leuven’s faculty in 2009, J.M. obtained a PhD at the same university in 2004. Afterwards, he was a postdoctoral researcher at the Université Pierre et Marie Curie in Paris France, and a postdoctoral fellow of the Flemish Science foundation. In recent years, J.M has also been a visiting professor at Johns Hopkins University, collaborating with C. Meneveau on common wind-energy interests. In 2012, J.M. obtained an ERC research grant (Active Wind Farms) on wind-farm turbulence interaction.
Currently, Johan Meyers is a member of the Optimization in Engineering Excellence Center of KU Leuven (OPTEC), where he chairs a working group on PDE-constrained optimization. He is a reviewer for numerous scientific journals, as well as an associate editor for Computers and Fluids, and Wind Energy Science.
Supersonic Low-Boom Aircraft Design Using the Open-Source SU2 Framework
Prof. Juan J. Alonso
Department of Aeronautics & Astronautics | Stanford University | Stanford, CA 94305 | U.S.A.
The current ICAO/CAEP ban on supersonic flight over land is slowing down the commercial development and deployment of supersonic aircraft that, based on the careful tailoring of the streamwise lift and volume distributions, can be extremely quiet (under 70 dB). In our opinion, it makes no sense to prohibit supersonic flight over land and, instead, we favor the establishment of noise regulations that, if met or exceeded, will make such overland flights legal.
In order to collect the necessary boom impact data so that ICAO/CAEP can establish rational boom noise regulations, Stanford University teamed up with Lockheed-Martin to design a low-boom flight demonstrator that will permit us to carry out an assessment of the impact of low intensity shaped sonic booms. This lecture will discuss the development and application of advanced adjoint optimization techniques, within the context of the SU2 framework, to the analysis and design of low-boom supersonic aircraft that are simultaneously able to achieve very high-levels of aerodynamic performance. We begin with the results of the N+2 Supersonic studies (for long-range, large passenger aircraft) and conclude with details of the design of the NASA Low-Boom Flight Demonstrator (LBFD). All the necessary methods and algorithms used in these designs are contained within the SU2 framework: an open-source collection of software tools written in C++ for performing CFD analysis and design, and PDE-constrained optimization on unstructured meshes that was started at Stanford in the Aerospace Design Laboratory and that is now under development at many partner institutions around the world.
This talk is sponsored by ANSYS:
Juan J. Alonso is an associate professor in the Department of Aeronautics & Astronautics at Stanford University. He joined the faculty in 1997 shortly after receiving a PhD degree in Mechanical and Aerospace Engineering from Princeton University. He is the founder and director of the Aerospace Design Laboratory (ADL) where he specializes in the development of high-fidelity computational design methodologies to enable the creation of realizable and efficient aerospace systems. Prof. Alonso’s research involves a large number of different applications including transonic, supersonic, and hypersonic aircraft, helicopters, turbomachinery, and launch and re-entry vehicles. Prof. Alonso was one of the main PIs of the Stanford University ASC Center for Integrated Turbulence Simulations (CITS) sponsored by the Department of Energy to create the computational solution methodologies to solve the flow through entire jet engines. He is the author of over 100 technical publications on the topics of computational aircraft and spacecraft design, multi-disciplinary optimization, fundamental numerical methods, and high performance parallel computing.
During the period spanning August 2006-October 2008, Prof. Alonso was the Director of the NASA Fundamental Aeronautics Program in Washington, DC (~1,500 civil servants and contractors and an annual budget of approximately $500M.) In that position he was responsible for the entire portfolio of aerospace vehicle and vehicle technology research for the agency in the subsonic rotary wing, subsonic fixed wing, supersonic, and hypersonic regimes, with particular emphasis on the energy and fuel efficiency and sufficiency of the aviation enterprise and its environmental impact. As Director of the Fundamental Aeronautics Program, he also oversaw a large number of interactions with academia, industry, and other government agencies including the FAA, the Department of Defense (USAF, Army, Navy), Department of Energy, DARPA, and the JPDO. He is also the recipient of several awards and fellowships including being a three-consecutive-time recipient of the AIAA Best Paper Award in Multi-Disciplinary Optimization, the NASA 2009 Exceptional Public Service Medal, the Stanford Chapter AIAA Professor of the Year Award, the Ray Grimm Memorial Prize in Computational Physics, and the Terman and Princeton University Honorific fellowships. Prof. Alonso is deeply interested in the development of an advanced curriculum for the training of future engineers and scientists and has participated actively in the curriculum committee for the Institute for Computational and Mathematical Engineering (ICME) at Stanford University. He holds a Bachelor of Science in Aeronautics & Astronautics from the Massachusetts Institute of Technology (MIT 1991) where he was a member of the team that currently holds the world speed record for human powered vehicles over water.
Prof. Alonso serves in the AIAA Multidisciplinary Optimization Technical Committee, the CGNS Steering Committee and the Center for Turbulence Research Steering Committee and he is a reviewer for a number of archival journals. He has also served in the NASA Advisory Council (Aeronautics Committee), the VAATE Steering Committee, the Fixed Wing Vehicle Executive Council, and the FAA Office of Environment & Energy REDAC. More recently (2010), Prof. Alonso was a member of the Secretary of Transportation’s Future of Aviation Advisory Council and in December 2010 he was appointed to the FAA Administrator’s Management Advisory Council for a term of 3 years. In the past, his research work has been funded by DARPA, AFOSR, the Department of Energy, NASA, Boeing, and Raytheon Aircraft among others
How surrogate models support aerodynamic optimization in industry – An example with methodologies and applications
Graduate physicist and expert in aerodynamic optimization
- CAD and geometry parameterization and how a surrogate model supports the adjoint aerodynamic optimization chain
- Mesh and geometry adjoint method
- Design space exploration using POD together with interpolation
- Applications at Airbus
- Requirements towards model based system engineering
This talk is sponsored by AIRBUS:
Holger Barnewitz joined Airbus in 1998 as research and capability development engineer in the aerodynamics department. He has a wide expertise in computational fluid dynamics, numerical algorithms, software development and how to apply research results in aeronautical industrial context. He works on the interface between methodologies invented at research institutions and bringing this into applications used by flight physics design engineers at Airbus.
Institution: Airbus, Germany, Bremen
Position: Airbus, Flight Physics Department, Capability Development, Expert Aerodynamic Optimisation
Research and Application:
Flight physics capability development, surrogate modeling, reduced order models, aerodynamic optimization, parametric geometry and mesh deformation, computational fluid dynamics, high performance computing
Practical wing design via numerical optimization—Are we there yet?
Joaquim R. R. A. Martins
Professor of Aerospace Engineering | University of Michigan
Wing shape is a crucial aircraft component that has a large impact performance. Wing design optimization has been an active area of research for several decades, but achieving practical designs has been a challenge. One of the main challenges is the wing flexibility, which requires the consideration of both aerodynamics and structures. To address this, we proposed the simultaneous optimization of the outer mold line of a wing and its structural sizing.
The solution of such design optimization problems is made possible by a framework for high-fidelity aerostructural optimization that uses state-of-the-art numerical methods. This framework combines a three-dimensional CFD solver, a finite-element structural model of the wingbox, a geometry modeler, and a gradient-based optimizer. This framework computes the flying shape of a wing and is able to optimize aircraft configurations with respect to hundreds of aerodynamic shape and internal structural sizes. The theoretical developments include coupled-adjoint sensitivity analysis, and an automatic differentiation adjoint approach. The algorithms resulting from these developments are all implemented to take advantage of massively parallel computers.
Applications to the optimization of aircraft configurations demonstrate the effectiveness of these approaches in designing aircraft wings for minimum fuel burn. The results show optimal tradeoffs with respect to wing span and sweep, which was previously not possible with high-fidelity models.
This talk is sponsored by ESTECO:
Joaquim R. R. A. Martins is a Professor at the University of Michigan, where he heads the Multidisciplinary Design Optimization Laboratory (MDO Lab) in the Department of Aerospace Engineering. His research involves the development and application of MDO methodologies to the design of aircraft configurations, with a focus on high-fidelity simulations that take advantage of high-performance parallel computing.
Before joining the University of Michigan faculty in September 2009, he was an Associate Professor at the University of Toronto Institute for Aerospace Studies, where from 2002 he held a Tier II Canada Research Chair in Multidisciplinary Optimization. Prof. Martins received his undergraduate degree in Aeronautical Engineering from Imperial College, London, with a British Aerospace Award. He obtained both his M.Sc. and Ph.D. degrees from Stanford University, where he was awarded the Ballhaus prize for best thesis in the Department of Aeronautics and Astronautics. He was a keynote speaker at the International Forum on Aeroelasticity and Structural Dynamics (2007), the Aircraft Structural Design Conference (2010), the SIAM Conference on Optimization (2014), and the Congress on Numerical Methods in Engineering (2015).
He has received the Best Paper Award in the AIAA Multidisciplinary Analysis and Optimization Conference four times (2002, 2006, 2012, and 2014). He is a member of the AIAA MDO Technical Committee and was the technical co-chair for the 2008 AIAA Multidisciplinary Analysis and Optimization Conference. He has served as Associate Editor for the AIAA Journal, and is currently an Associate Editor for the Journal of Aircraft, Optimization and Engineering, and Structural and Multidisciplinary Optimization.
Machine Learning algorithms for prediction problems in energy applications.
Dr. Sancho Salcedo Sanz
Dpto. de Teoría de la Señal y Comunicaciones, Universidad de Alcalá, Madrid, Spain.
The application of Machine Learning models and algorithms in Energy applications is currently a hot topic in different engineering fields.
Renewable and clean energy sources are specially important in the sustainable development of current societies, but they have important problems for their introduction to the energy mix, due to their intermittent nature. This is a continuous source of interesting problems, which in many cases are better solved by applying novel machine learning algorithms.
Demand prediction at different levels is another problem in which soft-computing algorithms have been successfully applied. In this talk we review some of the most important problems in energy applications, including renewable energy resource prediction, energy demand prediction and optimization problems that arise in micro-grid design with renewable generation. In all cases we will show how Machine Learning algorithms are able to obtain excellent results in these problems.
Dr. Sancho Salcedo-Sanz was born in Madrid, Spain, in 1974. He received the B.S degree in Physics from the Universidad Complutense de Madrid, Spain, in 1998, and the Ph.D. degree in Telecommunications Engineering from the Universidad Carlos III de Madrid, Spain, in 2002. He spent one year in the School of Computer Science, The University of Birmingham, U.K, as postdoctoral Research Fellow.
Currently, he is an associate professor at the department of Signal Processing and Communications, Universidad de Alcalá, Spain. He has co-authored more than 150 international journal and conference papers in the field of machine learning and soft-computing.
Dr. Salcedo-Sanz has received different research awards in his career, such as the Universidad de Alcalá's Best Young Researcher award in 2009, the 3M Innovation award in 2010 and the Price of the Social Council of University of Alcalá award to technology transfer in 2011. His current interests deal with Soft-computing techniques, hybrid algorithms and neural networks in different applications of Science and Engineering.