Академический Документы
Профессиональный Документы
Культура Документы
2007
1 Problem description
The simulations described here were part of investigations on the homogenization of an optical fiber coil. This means, we wanted to know how a fiber coil with several hundreds of windings deforms "macroscopically", when external pressure is applied or when it's subjected to vibration. One way to do this is to look at one elementary cell of such a coil (called unidirectional fiber compound among mechanics people), apply pressure in certain ways and look how it deforms on the average. Let's better not dwell on the particulars of this subject. What you want is to learn how to use Elmer for YOUR tasks. What you need to now, however, is the geometry of this elementary cell and what is done to it:
210m
On the left you see a sketch of the standard cell: the innermost part is the optical fiber. Then come two layers of coating, which serve to protect the fiber against scratches, water or whatever else comes from outside. This is what you get from the fiber manufacturer. The outermost layer is silicone rubber which holds the fiber windings together. So much about the geometry. The loads (boundary conditions) are like this: No body forces like gravity or centripetal forces constant normal displacement on the outermost boundary. The tangential displacement is not prescribed.
uy=1m
ux=-1m
ux=1m
uy=-1m Keywords: elmer, finite element method, stress analysis, two dimensional elasticity, plane strain, young's modulus, poisson's ratio, rubber, optic fiber, cladding, coating
2 Geometry Description
Since Elmer seems to have SEVERE compatibility issues with any mesh generation software i've tried so far (gmsh, netgen and other less well-known software), and twodimensional geometries can be meshed using elmer's gui (ElmerFront), it seems sensible to use elmer for geometry definition and mesh generation. Writing the geometry file is not very conveniend, but it's not that hard either. Please copy the left column of the following spreadsheet into a text editor and save it as a file with the ending ".egf" (elmer geometry file). The right column contains only comments and mustn't be copied into the .egf-file.
Header Model Name "Faserelement" Dimension 2 !geometry dimension Unit 1e-6 ! dimensional unit: m End Vertices Points Size 4 2 Real -110 -110 ! 1. bottom left vertex 110 -110 ! 2. bottom right vertex 110 110 ! 3. top right vertex -110 110 ! 4. top left vertex End Edge 1 Name "Kante unten" ! bottom edge Vertices 1 2 End Edge 2 Name " Kante rechts" ! right edge Vertices 2 3 End Edge 3 Name " Kante oben" ! upper edge Vertices 3 4 End Edge 4 Name " Kante links" ! left edge Vertices 4 1 End Edge 5 Name "Circle3" ! boundary of aCoating Geometry "Circle" Radius 83
Edges from the outside to the inside! It's important to stick to this order!
Circle boundaries
Center 0 0 End Edge 6 Name "Kreis2" ! boundary of iCoating Geometry "Circle" Radius 65 Center 0 0 End Edge 7 Name "Kreis1" ! boundary of fiber cladding Geometry "Circle" Radius 40 Center 0 0 End Body 1 Name String "Kautschuk" Color 128 128 0 255 Edges 1 2 3 4 5 End Body 2 Name String "aCoating" Color 128 0 0 255 Edges 5 6 End Body 3 Name String "iCoating" Color 0 128 0 255 Edges 6 7 End Body 4 Name String "Glas" Color 128 128 128 255 Edges 7 End
Definition of different bodies which can be advised different materials, forces and even equations! I recommend to use different colors for different bodies to improve visibility.
What you need to do is to select the equation "STRESS ANALYSIS" in the field "Active equations", give it the name "stress", press "Add" and then "Attach" it to each of your 4 bodies. You're finished, if you see an entry like "[1] stress" next to the name of each body in the body list (upper left corner). Now press "OK".
Please proceed similarly for the other three materials. Don't forget to attach each material to the correct body!
The continuity conditions on the common boundaries of different bodies are obeyed by elmer automatically and thus don't need to be specified.
3.7 Simulation
Now everything should be prepared for a stress analysis, had we not used a layer of silicone rubber with a relatively high poisson's ratio close to 0.5 which leads to a high bulk modulus. The iterative solver which is used by default does not converge if you use materials like that, so you need to use another solver. In "Solver" "Solver Settings" please select "Solver type" "Direct". To be honest i never informed myself about solver algorithms, so i can't tell you which Solving method to choose. Another, perhaps more important switch is "Calculate Stresses" on the bottom of this dialog. If you don't activate this switch, elmer will only calculate displacements, which are probably not very usefull information for the user. After having done this you can start the simulation by clicking "Solve" on the push button column on the right hand side of ElmerFront.
4 Postprocessing
I won't go into too many details about the possibilities of ElmerPost but just show you how to watch your results as color maps. You can start ElmerPost by clicking on "Results" in ElmerFront. If you didn't specify the correct model directory however ElmerPost won't find the data it needs and breaks with an error message. In this case you either specify your working directory ("Problem" "Model name and directories" in ElmerFront) and press on "Results" again, or you load your data in ElmerPost ("File" "Open". Your results file is in the mesh directory that elmer created after mesh generation and has the ending .ep.). The two dialogs below allow you to specify which variable you want to inspect (in this case the modulus of displacement). It's important to choose the same variable in both dialogs. Otherwise the numbers of the color scale correspond to another variable than your color map image.
Since the rest is pretty self-explanatory (and i need to set off to work now) i won't mention other possibilities of ElmerPost. Just play with it :-)