Fully Automatic Hex Dominant Mesher Paul Gilfrin Sharc Ltd
Sharc Ltd UK About Sharc Developer of Harpoon Founded in 1997 Distributors of Ensight Engineers with CFD/FEA experience Specialise in the integration of Harpoon within customers workflow to reduce lead time for the design process 25/10/2007 www.sharc.co.uk 2
Design Process Identifying a Need Defining the Problem Conducting Research Finding Alternative Solutions Narrowing the Research Analysing Criteria Analysing Alternative Solutions Making a Decision Presenting and Selling the Product 25/10/2007 www.sharc.co.uk 3
Empowering the Engineer Analysing alternative solutions is an iterative process Amplifies decision making time Finding Alternative Solutions Full CFD analysis = 45 weeks Design change Source: EACC 2007 The CFD Process for Aerodynamics at Volvo Cars using HARPOON-FLUENT Zenitha Chroneer Volvo Car Corporation = 2-3 1-2 days weeks Analysing Alternative Solutions Assumptions Analysing Criteria Making a Decision 25/10/2007 www.sharc.co.uk 4
Meshing Problem Meshing of complex geometries is one of the most labour intensive, repetitive processes in simulation today. Therefore meshing is the primary bottleneck to productivity gains 25/10/2007 www.sharc.co.uk 5
Preparing input geometry Generating surface mesh Growing volume mesh Meshing Problem Requires educated operator Extensive software training Hours of repetitive work 25/10/2007 www.sharc.co.uk 6
Harpoon Extreme engineering problems require extreme solutions. Enter Harpoon, a fully automatic hex-dominant mesher specifically designed to cope with the most complex geometric entities. It s speed and versatility give companies the edge when it comes to shortening the time to market 25/10/2007 www.sharc.co.uk 7
Harpoon Process Standard Pre-Processing Hours Days Weeks Original CAD CAD Clean-up Surface Mesh Original CAD Volume Mesh Harpoon Surface + Volume Mesh Minutes - Hours 25/10/2007 www.sharc.co.uk 8
Harpoon Methodology Approach consists of a modified Octree method Body fitting mesh by means of node-movement and cell manipulation/reconstruction. No nasty cut cells.
Why Change Your Process? Harpoon = Mesh faster Mesh more Mesh automatically Mesh the previously impossible 25/10/2007 www.sharc.co.uk 10
Mesh Faster Mesh Size(Millions) Memory(GB) Time (Minutes) Million Cells/Minute 1.4 0.5 0.95 1.4 3.4 1 1.92 1.8 6.9 1.8 3.41 2 18.8 4.4 8.07 2.3 60.3 14.6 23.09 2.5 117 25.2 41.75 2.8 45 40 35 30 25 20 15 10 5 0 0 20 40 60 80 100 120 140 Memory Time Million Cells/Minute Source data: Jaguar XK8 model including refinement zones 25/10/2007 www.sharc.co.uk 11
CAD Loaders Mesh Faster Load CATIA, Unigraphics and IGES files directly Harpoon will retain layer association and separate into distinct parts Control tesselation settings through preferences 25/10/2007 www.sharc.co.uk 12
Dirty Geometry Mesh Faster Mesh directly from CAD data 25/10/2007 www.sharc.co.uk 13
Dirty Geometry Mesh Faster Tolerant to geometry errrors 25/10/2007 www.sharc.co.uk 14
Dirty Geometry Mesh Faster Automatic hole detection/automatic filling 25/10/2007 www.sharc.co.uk 15
Dirty Geometry Leak Detection Mesh Faster 25/10/2007 www.sharc.co.uk 16
Dirty Geometry Mesh Faster Leak Detection/Mesh Trace 25/10/2007 www.sharc.co.uk 17
Mesh Faster Simple setup 25/10/2007 Choose baselevel Apply levels to capture parts Name boundary surfaces Create refinement regions Mesh Check mesh Export to solver of choice www.sharc.co.uk 18
Mesh Faster Clearly, true to its word, Harpoon is very fast. It builds the grid fast, but, frankly, more importantly than that, you can set it up faster. You can give it these.stl files and it just builds the grid. If you don t like it, you can change it. And, one minute later, change it again. That kind of iteration takes weeks with a conventional unstructured grid software package for these geometries. Robert Kunz, Ph.D., Head of the Computational Mechanics Division at Pennsylvania State University s Applied Research Laboratory 25/10/2007 www.sharc.co.uk 19
Faster Convergence Mesh Faster The hexa-dominant mesh improves the convergence quality & rate of the model, even with reduced mesh resolution Mass conservation residuals dropped by up 90% Kinetic Energy residuals dropped by up to 20% Convergence time was reduced by more than 30% for the initial condition case, up to 80% for the consecutive cases Source: www.sharc.co.uk Daimler Chrysler Pump study. Comparing Harpoon mesh with previous full tet method 25/10/2007 www.sharc.co.uk 20
Ducati Example 4.6 Million cell mesh 10 minute set up 3 minutes to mesh Solved using ANSYS Fluent 6.3 K epsilon at 35m/s Mesh Faster 25/10/2007 www.sharc.co.uk 21
Mesh More Harpoon Speed up = More runs 30 millions cell mesh. Less than 2 hours to generate on HP C800 The software enables us to undertake more complex meshes in a much faster time, Anders Jönson, Senior CFD Engineer at the Volvo Car Corporation Mesh at an earlier stage of the design process The value of Harpoon, is that it provides a very quick grid adequate for the level of fidelity of prediction that we re interested in to support our preliminary design work. T. Alan Egolf, Supervisor of Aerodynamic Methodology at Sikorsky 25/10/2007 www.sharc.co.uk 22
Mesh More Results Daimler Chrysler pump study Excellent agreement was obtained using a Harpoon mesh, containing approximately 3M cells, at a fraction of the simulation effort required for a full tet mesh Source: www.sharc.co.uk 25/10/2007 www.sharc.co.uk 23
Mesh More Results Achieve comparable results in considerably less time Cd No Under Body Panel Under Body Panel ICEM/Tgrid 0.393 0.388 Harpoon 0.377 0.372 Comparing to test results made in wind tunnel, the separation line for the Harpoon mesh looks more realistic Zenitha Chroneer, Volvo Car Corporation Source: EACC 2007 The CFD Process for Aerodynamics at Volvo Cars using HARPOON-FLUENT Zenitha Chroneer Volvo Car Corporation 25/10/2007 www.sharc.co.uk 24
Simple to Use Mesh Automatically Minimal training required Intuitive GUI Config files for repetitive meshing Mesh then remesh to achieve the desired grid Harpoon, compared to other mesh generation software, doesn t require a steep learning curve maybe a week or two, Professor Len Imas Stevens Institute of Technology 25/10/2007 www.sharc.co.uk 25
Batch Mesh Automatically All features/settings can be invoked in batch. Including surface cell sizes, refinement areas, boundary/prism layers Config file is used to retain optimal settings Command line features for advanced mesh control 90% of meshes converge first time Perfect for optimisation projects 25/10/2007 www.sharc.co.uk 26
Mesh Automatically Harpoon 3.0 even better convergence! New meshing algorithms New cell treatment improvements Extreme optimisation for difficult geometries Mercedes C class 20m/s K-E Realisable Moving Road 25/10/2007 www.sharc.co.uk 27
Harpoon 2.4 Cell Based Harpoon 2.4 Node Based Harpoon 3.0 Cell Based Harpoon 3.0 Node Based 25/10/2007 www.sharc.co.uk 28
Mesh the previously impossible See results earlier in the design process If the designer changes something we want to be able to regrid quickly, which Harpoon allows us to do, and get results in a timely manner so we can have an impact on our schedule. T. Alan Egolf Supervisor of Aerodynamic Methodology at Sikorsky 25/10/2007 www.sharc.co.uk 29
Mesh the previously impossible Wrap really dirty geometry De-feature Close large holes 25/10/2007 www.sharc.co.uk 30
Mesh the previously impossible Match to a predefined surface mesh 25/10/2007 www.sharc.co.uk 31
Mesh the previously impossible Harpoon also enables us to support analysis in new areas where the complexity of the required meshing would have prevented our participation in the past. Anders Jönson, Senior CFD Engineer at the Volvo Car Corporation Image Source: EACC 2007 The CFD Process for Aerodynamics at Volvo Cars using HARPOON-FLUENT Zenitha Chroneer Volvo Car Corporation 25/10/2007 www.sharc.co.uk 32
Mesh the previously impossible While using panel methods is reasonably fast and inexpensive, it also allows a considerable amount of the engineer s subjectivity and speculation to creep into the calculations. That s because panel methods cannot account for flow separation or lift without user-selected models of these phenomena. Harpoon grids.. are cost-effective and provide quick results that, when combined with FLUENT, are far superior to traditional panel methods T. Alan Egolf Supervisor of Aerodynamic Methodology at Sikorsky 25/10/2007 www.sharc.co.uk 33
Harpoon s assets No need for a surface mesh No heavy CAD manipulation Tolerant to geometry errors : the input geometry is only used as supporting geometry Reliable (always produces a mesh) Extremely fast (takes only a few minutes) Good quality valid cells Able to handle massive changes in scale Very easy to use 25/10/2007 www.sharc.co.uk 34
Supported hardware platforms 64 bit architectures Linux SuSe 9.3 Linux Red Hat WS HP-UX 11 Solaris 8 IRIX 6.5 Itanium 2 Windows 64,Vista 32 bit architectures Linux Red Hat 8 Mac OSX 10.3 Windows XP, 2000,Vista 25/10/2007 www.sharc.co.uk 35
Output formats Geometries : STL (save modified/wrapped geometry) Meshes : DTF Ensight Fluent Star-CD (CCM+ Soon) CFD++ Cobalt/AVUS CFX-5 Gambit NASTRAN Dytran LS-Dyna ANSYS ABAQUS STL Acusolve FOAM 25/10/2007 www.sharc.co.uk 36
Who Can Benefit From Harpoon? Any company looking for significant improvements in development time Extremely quick 3D meshing Complex geometries Design optimisation Applicable to all areas of scientific computing A huge improvement in terms of time and effort spent compared to conventional methods 25/10/2007 www.sharc.co.uk 37
Why Choose Sharc? Sharc offers a personal approach which makes it a very easy company to work with. Over the past two years the team has always been very helpful, demonstrating a willingness and ability to meet our technical and business needs and responding quickly when necessary. Anders Jönson, Senior CFD Engineer at the Volvo Car Corporation 25/10/2007 www.sharc.co.uk 38
Thank you Website www.sharc.co.uk Technical Enquiries paul@sharc.co.uk Tel: 0161 877 5071 25/10/2007 www.sharc.co.uk 39