请问，如果想让DOE中的试验方案在一台机器中实现分布计算，也就是四个核心，每个核心处理一个方案，该怎么做？--【匿名用户】:E-works热心网友 spred
不错！！--卢玉琴
那么请问isight&fd能不能进行类似的mdol语言开发呢？有没有相关资料？--【匿名用户】:E-works热心网友
能否较详细介绍一下开发的一个简单实例啊。--【匿名用户】:E-works热心网友
为什么语言介绍并不更新啊？--【匿名用户】:E-works热心网友
博主您好，我初次接触这种抽样法，感觉一头雾水。很幸运看到您的博文，请您能详细介绍一下拉丁超立方抽样好吗？最好举例说明。谢谢！--【匿名用户】:E-works热心网友
我用的是isight8.0可以实现并行计算吗--【匿名用户】:E-works热心网友
太经典了！对于我这样刚刚迈出校园的学生来说，真是醍醐灌顶啊！多谢楼主的教导之言！--【匿名用户】:E-works热心网友
很好的资料
请问你有包含图片的完整文档吗？
可否发一份networm_
谢谢--【匿名用户】:E-works热心网友
logs中写的错误类型是stderr：&Estimated&disk=1.2MB
stderr：&Estimated&DOF=80
stderr：&Estimated&memory=32MB
之类的错误，不知道是什么原因
--【匿名用户】:E-works热心网友
请问一下，做过isight集成nastran的案例吗，我用的是isight-fd版本，集成nastran2007，结果总是出错，不知道什么原因，烦请高手指点一下--【匿名用户】:E-works热心网友
好的&谢谢--【匿名用户】:E-works热心网友
说得是没错，但那些政府管员，当管的都拿老百姓的呀，大家一起努力吧，改变中国现在的样子吧，--【匿名用户】:E-works热心网友
好!--【匿名用户】:E-works热心网友
清华大学有N多个大学校长，俺想知道这五句话是哪个校长说的？&
--【匿名用户】:E-works热心网友
谢谢答疑解惑
--【匿名用户】:E-works热心网友
--【匿名用户】:E-works热心网友
请问Isight&for&Abaqus——by&hannah在abaqus的哪个版本中有啊？--【匿名用户】:E-works热心网友
Re&2楼：
如果想让DOE中的各个实验方案分到不同的机器上并行计算的话，单机版的Isight-FD是不行的，必须在FIPER并行分布环境中才能实现。在这个FIPER环境中，Isight-FD只是它的一个客户端，另外还有一个客户端叫Station，这样一旦Isight-FD把各个方案提交到FIPER环境中后，FIPER环境的服务器端ASCS就会自动的把任务分到各个机器的Station上来并行或分布执行。--赛特达
Isight&for&Abaqus是我们的另外一个产品，优化方面功能和Isight-FD基本是一样的，只不过这个产品只能集成Abaqus，不能集成其它的软件。
阅读排行榜
评论排行榜
本文标签: &&&&&
Airfoil Shape Optimization with Engineous Software’s eBlade
Overview of the Airfoil Multidisciplinary Design Optimization (MDO) Process
To remain competitive in the global marketplace, manufacturers of steam turbines, industrial gas turbines, and aircraft engines must continually strive to develop highly efficient, reliable, and cost-effective machines in the shortest possible time. To meet these objectives, engineers face an inherently multidisciplinary optimization problem with many conflicting design objectives and constraints. Because traditional manual trial-and-error design processes are no longer adequate to meet this challenge, many turbomachinery companies are working to automate their simulation-based design processes and use numerical optimization methods to develop better designs much faster than is possible with manual methods.
A key focus of these efforts is in blade design, because the blading is often the critical path item in a new design, and it has the greatest impact on overall machine efficiency and reliability. In the early 1990s, a number of turbomachinery companies began to automate their simulation-based blade design processes and apply numerical optimization methods to improve performance and reliability while at the same time reducing design cycle time and cost. Early examples that addressed the aerodynamic design process are presented in References 1-3. Because these automated systems can evaluate many more design alternatives than can be done manually, they have proven to be ideal tools for finding nonintuitive aerodynamic designs that can result in significant performance benefits (Refs. 4, 5).
The turbine airfoil design process is very complex and time-consuming, and it requires skilled aero/thermo designers to achieve the optimum balance between efficiency requirements, mechanical reliability, and manufacturing cost. Many of our turbomachinery customers have used Engineous technology to develop integrated, multidisciplinary, multi-objective optimization systems for the aerodynamic and mechanical design of turbine blades that produce substantial reductions in design cycle time (for example, see Ref. 6). Our iSIGHT and iSIGHT-FD process integration and design optimization software is used to link together simulation codes to automate the various design and analysis tasks, including airfoil shape generation, CFD analysis, stress and vibration analysis, and post-processing of results. Expert design rules and constraints are captured in these systems, and various optimization strategies are employed to automatically search the design space to find optimal designs that meet goals for performance and stresses. Recent efforts have focused on achieving robust designs (Ref. 7), or designs that are insensitive to uncertainties and variability in such things as manufacturing tolerances, material properties, and loading conditions, and on tracking multiple objectives independently to better understand the tradeoffs and ensure that the best design decisions are made (Refs. 8, 9).
A fairly representative MDO process currently used for blade design by many steam and gas turbine manufacturers is shown in Figure 1 (from Ref. 10). The aerodynamic design process is shown on the left side of the figure, and the mechanical design process is shown on the right side of the figure. The blade design tasks are:
Data defining the radial distribution of inlet and exit flow angles and thermodynamic operating conditions are extracted from the output file of a quasi-3D throughflow analysis at radial heights specified by the designer (i.e. for the “design sections”). Data defining the radial distribution of section properties such as axial chord, cross-section area, and moments of inertia also can be extracted from an external file so that first-order mechanical requirements can be met. An initial value for the optimum number of blades is calculated based on the Zweifel solidity criterion (Ref. 11). The resulting pitch and blade throat is fixed during the optimization process. A shape generation program is used to generate airfoil shapes for each design section using a variety of built-in parametric representations. (These runs can be launched in sequence or in parallel, depending on available computer resources.) Before optimization runs are launched, the airfoil design engineer will usually import an existing blade to use as a starting point and set constraints on the allowable ranges of design variables such as stagger angle, maximum thickness, leading and trailing edge metal angles, thicknesses, and wedge angles, leading edge radius or ellipse ratio, and unguided turning angle. Constraints are also placed on output quantities such as incidence angle, the shape of the surface Mach number distributions on pressure and suction sides, and the amount of diffusion past the throat. Airfoil section properties are calculated to check if they meet the requirements defined in Task 1 within a specified tolerance. If the requirements are not met, a new airfoil shape is chosen. If they are met, the process proceeds to Task 6. The blade passage defined by the airfoil geometry generated in Task 3 is meshed and a 2D channel flow analysis is performed. If the optimizer has achieved an optimum aerodynamic design for a given section, Tasks 3 through 7 are repeated for additional design sections. If not, the process returns to Task 3 and a new shape is generated. A typical optimization objective is to maximize profile efficiency while meeting a number of aerodynamic and mechanical design rules and constraints. Once all of the individual design sections have been optimized, they are stacked relative to one another to create a 3D shape. An optimizer-driven smoothing process can be invoked to make small adjustments to the profile shapes to eliminate ripples in the 3D surface. If desired, a full 3D Navier-Stokes solution can be performed to check the final 3D stackup. If all aerodynamic and first-order mechanical requirements have been met, the process proceeds to Task 11 for a more detailed mechanical analysis. If they are not, the process returns to Task 3 and a new stackup is generated. Geometries for the shroud, platform, fillets, and dovetail or fastener are parametrically generated and attached to the main airfoil shape to create a model of the complete blade. Aerodynamic loads and boundary conditions for the FEA analysis are captured from the CFD results, and material properties are selected from a database. A structured mesh is generated and an input file is created for the FEA code chosen for the blade stress and vibration analysis. The FEA solver is run to perform a detailed structural analysis. A postprocessing tool reads the FEA code output files and generates a Goodman diagram to assess cyclic stresses, and a Campbell diagram to assess blade vibration. The optimizer assesses information in the FEA code output files and the plot files for the Goodman and Campbell diagrams to determine if detailed stress and vibration requirements have been met. Many specific requirements can be included, such as bending stresses, centrifugal stresses, vibratory stresses (axial, tangential, torsional), thermal stresses, response to stimulus (nozzle passing frequency, low per rev), damping (shrouds, lacing/tie wires, mid-span snubbers), and high and low cycle fatigue. If all requirements are met, the design is completed. If not, the process returns to Task 2 for further iterations.
eBlade – Engineous Blade Design Framework
Most turbomachinery companies use a combination of in-house and commercial design and analysis codes in the MDO process in Figure 1, but it is becoming increasing difficult to maintain in-house codes, many of which are decades old. It can also be problematic to manage the interactions between old in-house codes and newer commercial codes, causing designers to spend a great deal of time manually transferring data from one code to another. To help alleviate these problems, Engineous has developed a blade design framework called “eBlade” to address the aerodynamic design tasks shown in Figure 1. (Further work is in progress with some of our software partners to integrate the mechanical side of the MDO process in iSIGHT-FD, as described in Ref. 10.)
Figure 1. Airfoil aero/mechanical MDO process (Ref. 10)
eBlade packages a comprehensive set of blade design tools within a powerful, flexible, and easy-to-use graphical user interface to allow complete 3D airfoil shapes to be optimized. These tools include parametric models for airfoil shapes, a special Windows version of the well-known MISES 2D flow solver (Ref. 12) created by Analytical Methods Inc. just for eBlade, 3D stackup utilities, a unique optimizer-driven 3D stackup smoothing operation, and a built-in 3D viewer that allows 3D airfoils to be visualized in the eBlade GUI without having to export geometry data to an external CAD program. eBlade is designed to be very flexible and customizable so that it can be integrated seamlessly into any industrial design environment. It can be linked to other tools and simulation codes, such as CAD, grid generation, CFD, or FEA, using the integration capabilities in iSIGHT or iSIGHT-FD. Custom GUIs can be developed to incorporate specific company terminology and geometric parameters to make it easy and familiar for engineers to work with. In short, it allows companies to dramatically reduce airfoil design cycle time and development costs, while at the same time improving performance and reliability.
The aerodynamic blade shape optimization process captured in eBlade is outlined in the flow chart shown in Figure 2. The tasks shown in this flow chart have been integrated and automated with iSIGHT v10.0, and they will be integrated with iSIGHT-FD later in 2007. iSIGHT drives the process by manipulating airfoil design parameters and assessing program outputs to find the optimum airfoil design based on aerodynamic and mechanical design criteria defined by the airfoil designers. Optimization objectives can include minimizing profile losses or maximizing profile efficiency, achieving desired surface velocity or Mach number distributions on suction and pressure sides, minimizing incidence angle losses, controlling flow diffusion downstream of the throat to avoid flow separation on the pressure side, and avoiding flow separation inside the blade passage.
In the flow chart, boxes indicate design calculations that are made or simulation codes that are run, while diamonds indicate decisions that must be made. These decisions are based on design rules and expert knowledge captured in iSIGHT as either constraints (i.e. upper and/or lower bounds on specific input variables or output quantities) or as conditional logic. An iSIGHT task plan has been preconfigured to control the airfoil shape optimization process, but designers can modify the plan in iSIGHT if they so desire.
Figure 2. Airfoil aerodynamic shape optimization process captured in eBlade
A customized GUI is employed as the front end gateway to the eBlade airfoil design optimization environment, as shown in Figure 3. It allows the designer to select the desired design parameters to be manipulated by iSIGHT during the optimization process, and it displays results from the parametric airfoil shape generation and MISES flow solution. Many airfoil design parameters can be manually adjusted by the user in the eBlade GUI by dragging slider bars. This is very useful for setting up an initial airfoil to use as a starting point for optimization, and it allows the user to set constraints on the design parameters to ensure that the optimizer will always produce reasonable sections.
In the upper left, the main Airfoil Panel shows the airfoil cross-sections, with each design section on a separate tab. Two adjacent airfoils can be shown to visualize the passage shape by clicking on the Passage button, as shown in Figure 4 in full screen mode. Buttons located in the toolbar above the main Airfoil Panel can be selected to display a variety of special graphics information superimposed on the airfoil shape. For example, Figure 4 shows “passage circles” superimposed over the blade passage so that the designer can judge the smoothness of changes in the passage width and the location of the minimum throat, which can have a major impact on the airfoil efficiency.
In the lower left, the Parameters Panel shows a group of slider bars that can be used to manually adjust the values of airfoil design variables and set upper and lower bounds on the design variables. The cross-section shape changes in real time as the slider bars are moved. The designer can use these slider bars to manually generate a decent airfoil shape to use as a starting point for the optimization run.
In the upper right, a Plot Panel displays thumbnail plots showing output data from the various simulation codes and calculations used in the process, including output from the MISES flow solution. Any of these plots can be displayed in enlarged form in the main Airfoils Panel by double-clicking on it.
In the lower right, a Program Data Panel shows a variety of different tabs with detailed blade geometry or program output. For example, Figure 3 shows tabs related to MISES output.
Figure 3. eBlade GUI
The eBlade GUI can be integrated with output files from quasi-3D throughflow codes to extract the proper data needed for input to the airfoil optimization system. Several parametric methods for generating airfoil geometry have been integrated in the eBlade GUI and in iSIGHT. Engineous has developed a unique “New Pritchard” parameterization scheme based on the well-known Pritchard parametric model (Ref. 13) modified with Bezier polynomials to allow more flexibility for the optimizer to fine tune the shape. This method can produce virtually any airfoil shape required for steam and industrial gas turbine blades. eBlade can easily be customized to incorporate proprietary parameterization methods used by individual companies, to incorporate specific company terminology, and to integrate in-house 2D flow solvers. The system will support a variety of formats for blade shape definition to be saved in output files (IGES, tables of xyz points, etc.).
The process of generating new shapes, importing them into MISES, and launching the MISES solver has been automated. Blades can be designed on horizontal surfaces or on slant stream surfaces. Typical optimization runs may take 100 to 300 trials to achieve an optimized airfoil that meets all requirements and constraints. However, MISES runs will take only 15-20 seconds on a reasonably fast PC, so an optimized airfoil can be found in about 30 to 45 minutes. The x, y plotting capability in iSIGHT can be used to display the airfoil shape and surface Mach number distribution during each iteration in the optimization process. Mechanical issues are addressed in that the system is set up to meet user-defined requirements for cross-section area, Imax and Imin within a specified tolerance. This will ensure that the blades will be reasonably close to meeting mechanical requirements, pending higher fidelity 3D finite element analysis.
Figure 4. Special graphics buttons and passage circles showing minimum throat location
Airfoil designers can import x,y points defining existing blade cross-section shapes into eBlade and “parameterize” them to use as a starting point for optimization. The built-in Fitting Reviewer utility launches a fitting process driven by the Pointer optimizer to determine the values of New Pritchard parameters that reproduce the imported blade shape as closely as possible. As shown in Figure 5, the imported shape is defined as the target airfoil, and the designer manually manipulates the parameter slider bars to define a reasonable parametric shape that the fitting process will use as a starting point. Pointer then automatically adjusts the parameters until a very close fit is obtained, usually within a few minutes. This process can be used to fit almost any blade shape quite closely, as shown in Figure 6. In this figure, the imported target shape is shown in red, while the final fitted shape is shown in black.
Figure 5. Blade fitting showing starting and target airfoil shapes
Figure 6. Fitting Reviewer results shown in full screen mode After all of the 2D design sections have been generated, a 3D stackup can be created and viewed using the 3D Viewer shown in Figure 7. The 3D Viewer allows 3D pictures of the optimized airfoil stackup to be displayed without the need to export the blade data to an external CAD program. The 3D Viewer GUI is composed of four panels:
3D View Panel: A 3D picture of the blade is displayed in this panel. The blade can be rotated, panned, and zoomed using mouse commands. A single blade can be displayed, but 2, 3, or all the blades on the row can also be displayed in the proper orientation in space relative to one another. Parameters Panel: This panel shows slider bars or other features depending on the selection made in the Parameter Selection Panel. Parameter Selection Panel: This panel displays a tree of items that can be selected to determine what is displayed in the Parameters Panel. It includes tools for defining and displaying the stacking axis (which can be a straight line or several different types of curved line), adjusting the section shapes (particularly useful for manually smoothing or refining leading and trailing edge shapes), defining the x,y point distribution for the sections, generating additional interpolated sections, implementing stackup smoothing, and saving the x,y,z data to a file. Color Selection Panel: This panel allows the user to select the colors displayed for the airfoil section shapes, the airfoil section fill color, the airfoil skin, and the radial wire filaments between sections. It also allows the user to turn these features on and off to present different views of the blade shape. For example, Figure 8 shows a wire frame view of a 3D airfoil with the outer skin turned off to show the straight line stacking axis. Figure 9 shows an airfoil with a curved stacking axis composed of a parabola-line-parabola.
An automated procedure, driven by iSIGHT’s Pointer optimization algorithm, can be launched from the 3D Viewer to assess the smoothness of stacked 3D airfoils and to make local modifications to individual airfoil cross-sections to achieve the desired smoothness. This process typically takes 5 to 10 minutes.
Figure 7. eBlade 3D Viewer GUI
Figure 8. 3D view showing stacking axis
Figure 9. Blade stacking using Parabola/Line/Parabola method
Customization for Wind Turbine Airfoils
With increasing emphasis being placed on finding renewable power generation technologies that minimize impact on the environment, wind power has exploded in popularity. Wind turbines present similar multidisciplinary design challenges as more conventional forms of power generation, and, as with other turbomachines, the blading is critically important to the performance, reliability, and acoustics of wind turbines. Engineous Software is looking for wind turbine partners to develop a version of eBlade to be called “eWind(TM)” that can be used to automate the wind turbine blade design optimization process. To ensure that the eWind environment will meet the requirements and expectations of our customers, our development partners will provide direct guidance on the look/feel/functionality of the eWind GUIs. Our partners will also have access to development versions of the software for testing, and we will rely heavily on their feedback. Development partners will receive substantial discounts on eWind licenses for the initial commercial release.
With the acquisition of Synaps in 2004, Engineous Software acquired state-of-the-art shape optimization technology that enables consistent, high quality aerodynamic shape generation through intelligent parameterization. This proprietary “shape function” technology was originally developed by Synaps for the aerodynamic and structural optimization of aircraft wings and fuselages, working with such clients as Airbus (Ref. 14), Bombardier, Cessna, and Lockheed Martin. It has also been used to design hydrodynamic shapes for marine applications and aerodynamic components for race cars (wings, bodies, wheel fairings).
The Engineous Shape Functions provide a very efficient and “clean” parametric description of wing airfoil geometry that is tailor-made to be driven by optimizers. Fewer design variables are needed, and the feasible design space is governed by the parameterization itself rather than by numerous constraints and penalty functions. The result is that fewer function calls to the flow solver are required to achieve an optimized design, producing higher quality in less time. Since the Engineous Shape Functions were originally developed to model both subsonic and supersonic aircraft wing cross-section shapes, they can easily be adapted to model the variety of wing-like cross-section shapes used for wind turbine blading. eWind can be integrated with any in-house or commercial 2D CFD code applicable for isolated wing airfoils. For example, links to the widely used XFOIL panel code from Mark Drela at MIT are already in place (Ref. 15). It can also be integrated with acoustics models so that noise reduction can become an optimization goal.
Figure 10 shows a typical wing cross-section parametrically modeled with Engineous Shape Functions. The shape functions can be customized to capture the essential nature of the airfoil shapes used by specific companies, so that the unique features that give them their competitive advantage can be applied efficiently to all new designs. Other features in the eWind GUI can be customized to meet the needs of individual companies.
Figure 10. Wing airfoil shape generated by Engineous Shape Functions
References
Shelton, M.L., Gregory, B.A., Lamson, S.H., Moses, H.L., Doughty, R.L., and Kiss, T., “Optimization of a Transonic Turbine Airfoil Using Artificial Intelligence, CFD and Cascade Testing,” ASME Paper 93-GT-161, June 1993. Cofer, IV, J.I., “Advances in Steam Path Technology,” ASME Journal of Engineering for Gas Turbines and Power, Vol. 118, April 1996, pp. 337-352. Goel, S., Cofer, IV, J.I., and Singh, H., “Turbine Airfoil Design Optimization,” ASME paper 96-GT-158, June 1996. Lethander, A.T., and Thole, K.A., “Optimizing the Vane-Endwall Junction to Reduce Adiabatic Wall Temperatures in a Turbine Vane Passage,” ASME paper GT, June 2003.
Kohli, A., and Bogard, D.G., “Improving Film Cooling Performance Using Airfoil Contouring,” ASME paper GT, May 2006. Clark, J.P., Aggarwala, A.S., Velonis, M.A., Gacek, R.E., Magge, S.S., and Price, F.R., “Using CFD to Reduce Resonant Stresses on a Single-Stage, High Pressure Turbine Blade,” ASME paper GT-, June 2002.
Karl, A., May, G., Barcock, C., Webster, G., and Bayley, N., “Robust Design - Methods and Application to Real World Examples,” ASME paper GT, May 2006. Keskin, A., Dutta, A.K., and Bestle, D., “Modern Compressor Aerodynamic Blading Process Using Multi-Objective Optimization,” ASME paper GT, May 2006.
Huppertz, A., Flassig, P., Flassig, R., and Swoboda, M., “Knowledge-Based 2D Blade Design using Multi-Objective Aerodynamic Optimization and a Neural Network,” ASME paper GT, May 2007. Cofer, IV, J.I., “A Flexible, Automated System for Multidisciplinary Aero-Mechanical Design Optimization of Turbine Blading,” paper presented at Japan Society of Mechanical Engineers Design & System Division conference on November 16, 2006. Wilson, D.G., The Design of High-Efficiency Turbomachinery and Gas Turbines, The MIT Press, 1984, pp.241-243. Drela, M., and Youngren, H., “A User’s Guide to MISES 2.58,” MIT Computational Aerospace Sciences Laboratory, March 2005. Pritchard, L.J., “An Eleven Parameter Axial Turbine Airfoil Geometry Model,” ASME paper 85-GT-219. Van der Velden, A., Kelm, R., Kokan, D., and Mertens, J., “Application of MDO to Large Subsonic Transport Aircraft,” AIAA paper AIAA-.
Drela, M., “XFOIL: An Analysis and Design System for Low Reynolds Number Airfoils,” Conference on Low Reynolds Number Airfoil Aerodynamics, University of Notre Dame, June 1989. See also http://web.mit.edu/drela/Public/web/xfoil/.
17:19 赛特达 阅读(2086)