Newsletter 2015.11 Index
Theme : "Mechanical Engineering Congress, 2015 Japan (MECJ-15) Part 1"
|
Historical Perspective on Fluid Machinery Flow Optimization and a Message for the Future
Akira GOTO
|
Abstract
Fluid machineries such as pumps are the “Hearts” to support social infrastructures and protect civil lives, supporting reliable and efficient mechanical systems.ﾂ Contribution of fluid machineries to global issues such as environment, energy, water, and food is significant.ﾂ Commission Regulation (EU) has drawn strong attention of pump industries as it clarified Eco-design requirements (minimum efficiencies) for standard water pumps.ﾂ This sort of regulation would be the global trend for a variety of fluid machineries, leading to the strong demand to “design” better and innovative products.ﾂ Digital engineering, including the use of numerical simulation and design optimization, is the key to responding such demands.
The fluid-dynamic design improvements of pump flows had been heavily dependent on experiments, experience, and empiricism.ﾂ In 1980s, CFD (Computational Fluid Dynamics) based on RANS (Reynolds Averaged Navier Stokes) approximation using a simple turbulence model began to reveal the complex 3-D (three dimensional) flow physics in fluid machineries.ﾂ The fluid-dynamic design improvements of machine components were not very straightforward because of the complex three-dimensionality of flow fields (e.g. secondary flows), strong non-linearity between the flow phenomena and the complex 3-D flow passage configurations, etc.ﾂ In 1990s, an innovative 3-D inverse design method was applied successfully for controlling complex secondary flows in pumps, compressors, and turbines and had achieved drastic improvements of their fluid-dynamic performances.ﾂ In 2000s, the inverse design method was coupled with CFD and optimizer to achieve automatic single objective numerical optimization.ﾂ More recently, multi-point/multi-objective optimization has been proposed employing, for example, DoE (Design of Experiments), RSM (Response Surface Model), and MOGA (Multi-Objective Genetic Algorithm).ﾂ The challenge is continuing towards multi-disciplinary optimization having more objectives.ﾂ In case of having more than 4 design objectives, visualization of multi-dimensional objective space is another area of research for supporting efficient trade-off design selection.
The real world problem is often very complex and multi-disciplinary, but it is still not practical to challenge such problems relying only on numerical optimization using advanced simulation technologies.ﾂ The most important aspect in establishing effective design technology is to integrate a physical insight for improving flow fields, which ideally be predicted by lower level of approximation method (simple empirical equation, potential flow solver, RANS CFD, etc.).ﾂ Any physical insights obtained experimentally (using pressure probes, multi-color oil-film method, non-intrusive DPIV, etc.), numerically (using RANS, URANS, DES, LES, Adjoint, etc.), and empirically need to be combined with prediction method with minimum approximation level and implemented into multi-objective/multi-disciplinary optimization problems to challenge real-world problems in practical time frame.
Continuous challenges are envisaged in product innovation & work process innovation, employing numerical optimization, especially by young generations using digital engineering.
Key words
Design technology, Physical insights, Numerical optimization
Figures
Fig.1 History of simulation and design technologies.
Fig. 2ﾂ Image of workflow for product innovation.
Fig. 3ﾂ Trade-off selection for 4 design objectives using SOM (Self-Organizing Map).
Fig. 4ﾂ Multi-disciplinary optimization using empirical knowledge for design objectives.