<html><head><meta http-equiv="Content-Type" content="text/html charset=utf-8"></head><body style="word-wrap: break-word; -webkit-nbsp-mode: space; -webkit-line-break: after-white-space; "><br><div><div>Le 2 mai 2013 à 11:50, Matthijs Suijlen <<a href="mailto:matthijs.suijlen@innoluce.com">matthijs.suijlen@innoluce.com</a>> a écrit :</div><br class="Apple-interchange-newline"><blockquote type="cite"><div dir="ltr">Dear Mr Geuzaine and Mr Remacle,<br clear="all"><div><br></div></div></blockquote><div><br></div><div>Hello,</div><br><blockquote type="cite"><div dir="ltr"><div>Please let me introduce myself and the company I am working for (Innoluce). Innoluce is a high-tech starter which has spun off from Royal Philips Electronics three years ago. The core activities of Innoluce are designing and manufacturing micro MEMS mirrors for different markets where miniature and accurate laser scanning is required e.g. laser printers, optical medical devices and laser radar/gesture control. I am responsible for the mechanical (FEM) analysis of these mirror designs. The reason for writing this message to you is that we are facing some mesh generation difficulties/problems and we would appreciate your comments and advice very much.<br>
<br></div></div></blockquote><div><br></div><div>Great !</div><br><blockquote type="cite"><div dir="ltr"><div>For the finite element analysis we use Elmer FEM from the Finnish institute CSC (<cite><a href="http://www.csc.fi/english/pages/">www.csc.fi/english/pages/</a><b>elmer</b></cite>).<br></div></div></blockquote><div><br></div><div>I know well Elmer, we use it here for education purposes, together with gmsh.</div><br><blockquote type="cite"><div dir="ltr"><div>With the Alibre CAD software we model the actual designs. These design files in STP format are then used as input for the mesh generation in GMSH and NETGEN. For the post-processing we use own programs written in Numerical Python and we use Paraview a lot for inspecting output in 3D projection.<br><br></div></div></blockquote><blockquote type="cite"><div dir="ltr"><div>The typical geometries we are facing have all very thin membranes (~10 μm thick) attached to large anchor blocks (~300 μm thick). Currently we use unstructured mesh element distributions. For the stress calculations we would like to use at least two element layers over the thickness of these thin membranes. The difficulty is to combine coarseness of the mesh with the double element layer over the thickness. I have attached a model file which represents this typical geometry. Currently we use attractor fields to locally refine the mesh but this results in meshes with very large number of nodes. We believe this could be done more smartly and efficiently in GMSH. However we do not know how to proceed.<br>
<br></div></div></blockquote><div><br></div><div>Gmsh will have 3D boundary layer capabilities in less than a year (we are working on it). For now the best option (as you say) is to generate</div><div>extruded meshes in the thin layer and unstructured meshes in the bulk.</div><div><br></div><div>Here is an example you can look at that mimics what you need.</div></div></body></html>