Jump to content

zentgrafbw

Members
  • Content Count

    13
  • Joined

  • Last visited

About zentgrafbw

  • Rank
    Beginner

Recent Profile Visitors

The recent visitors block is disabled and is not being shown to other users.

  1. I have successfully been able to apply a motion curve to a rigid body such that it will rotate about it's center of mass. However I am unsure what the curve sees in determining how much rotation is applied. I have done some trial and error and noticed that having a curve that goes from 0 to 0.1 yields a relatively small amount of rotation. Is this 0.1 value indicative of radians? Thanks for any help. Brad
  2. I would like to prescribe a rotational motion to a rigid body such that during a simulation a rigid plate will rotate about a particular axis. I know how to prescribe translational motion to rigid bodies (i.e. define a motion curve and then assign this curve to a rigid body and make the appropriate degree of freedom adjustments in the x, y, or z directions.) I have been unsuccessful in prescribing a rotational motion to a rigid body. I have included a link to a picture of a simple model for clarification. I would like to have the rigid plate on the right of the blue box rotate about the x-axis some angle per second and then interact with the blue box using contact algorithms. Any information on how this might be accomplished would be greatly appreciated. Brad http://imageshack.us/photo/my-images/710/rotation.jpg/ Uploaded with ImageShack.us
  3. My name is Bradley Zentgraf. I work creating FE models of traumatic brain injury. I am currently trying to take a deformation seen in an element on the macroscale and apply this same deformation to an entire plane of elements on the microscale. I use Hypermesh as my preprocessor and run my simulations with LS Dyna The main issue I am facing is that I would like to tell an entire plane of nodes on the outer surface of my model to deform in a manner in which the nodal motion is not uniform throughout the plane. (i.e. the top nodes will deform proportionally more than the bottom nodes - this can be seen in the attached jpeg image). Also it should be noted that the material I am deforming is viscoelastic. I know how to create sets of nodes, create motion curves, and apply these curves to sets of nodes such that they deform in a particular manner. But I wish to control the motion of each node (on a particular plane) not just a subset of nodes on the plane. However, the deformation of the nodes at the top of the plane is more severe than the nodes at the bottom of the plane. This would mean that each node would have to be given it's own motion curve. There are hundreds of nodes on a plane so it is not practical to create and assign curves for each node. I have researched some possible solutions to this matter, but I am unsure if these are feasible solutions. It seems that I should be able to tell the plane to act as a rigid body and then prescribe some motion to the top nodes which would then be evenly distributed throughout the plane. But I do not know how this is to be done. Some of the cards that I thought might be useful included the CONSTRAINED_LINEAR and CONSTRAINED_NODAL_RIGID_BODY, which could perhaps tell the nodes on the plane to deform proportionally to one another. Then this could possibly be used in conjunction with the BOUNDARY_PRESCRIBED_MOTION_NODE(DISP) card to tell the top nodes how to deform letting the rest of the nodes "proportionally deform". Any help you could provide me would be greatly appreciated. Bradley Zentgraf Uploaded with ImageShack.us
  4. Image of Model that Needs Meshing I should probably add that the geometry that is trying to be meshed is not a simple geometry. It is not a simple cylinder or rectangle. The geometry is comprised of 2 cylinders, one which branches off of the other. A picture has been included for clarity. http://imageshack.us/photo/my-images/191/branchi.jpg/ Uploaded with ImageShack.us
  5. Currently I am facing problem where one ALE group “leaks” into another. I am almost positive that the leaking from the inner ALE group occurs because the inner mesh is a tetrahedral mesh while the outer mesh is comprised of hex elements. I think that even though nodal positions are shared between the tetrahedral and hexahedral elements, the fact that the tetrahedral elements are essentially half the size of their neighboring hex elements is the cause for extreme leakage when simulations are performed. I have a few questions about some meshing techniques that I feel could resolve my current troubles. Below is a brief description of what it is I want to accomplish, but cannot yet figure out. I need to find a way to create a 3D hex mesh for the inner ALE component that can have the surface elements be extruded out from the inner ALE component to the outer ALE component. I realize that the “shrinkwrap” option can create a 3D hex mesh, but the surface elements in this option cannot be extruded outward. In looking at the tetramesh panel, it would seem to me that there should be an option to perform this same method of meshing, but create a 3D mesh that matches surface the hex elements. I have not yet found a way to do this. Is there a way to do this? Below is a step by step description of my current methodology for meshing my model. The concerns are restated at the end of this section. 1.) I create a 2D QUAD mesh through the 2D automatic panel on the surface of the inner most component 2.) I extrude this 2D surface mesh to the outer ALE component through the 3D --> Solid Map --> Ends Only Panel, by selecting “source geom - surf” (inner, surface mesh elements) and “dest geom - surf”. When these new 3D elements are created, a new “component” is also made. 3.) I repeat this extrusion step for all of the surfaces that need to be meshed. 4.) I then inspect all regions where these newly extruded meshes come together. Often the case is that the nodes of these elements do not line up. It is absolutely paramount that these nodes line up or else models will not run. 5.) When elements are found that do not have adjoining nodes, I delete the layers of elements that need to be edited (through the Tools -->Delete panel) and then create new elements that fit perfectly by accessing the 3D --> Edit Element -->Create Element via Node List. a. This is a tedious process, especially when there are a lot of elements to adjust but it is the only way I have found that works, and that I know how to do. 6.) Once the outer ALE mesh is set, I try to create the 3D mesh for the inner ALE group. a. THIS IS REALLY WHERE MY CURRENT PROBLEM OCCURS **** I have not yet found a way other than the tetramesh option to create this 3D mesh. None of the solidmeshing techniques seem to fit my needs. It seems that there should be a similar option to the tetramesh panel in which I could say “take this surface mesh, match it, and create only hex elements” but as of now I know of no such option. **** Also I have looked into the “shrinkwrap” meshing option which will generate a 3D hex mesh, but the elements created in the shrinkwrap option cannot be extruded using the solid meshing option I outlined above. I have noticed that there are options in Hypermesh where one can create partitions/divisions that make a solid part “mappable”. I am not very familiar with these options, but they seem like maybe they would be useful in creating a 3D hex mesh. If this is a viable option, please let me know how I can accomplish my aims through creating partitions/divisions. Yours, Bradley Zentgraf zentgrafbw@gmail.com
  6. This method did not work when I turned off both the source and destination geometry tabs, and then selected the inner mesh and tried to drag them out to match the outer mesh. An error comes up which reads: "Auto Completing Input: Found No Node Loop on Destination Face" I have found some success using the ENDS ONLY tab in the 3D SOLID MESH. This has allowed me to project out elements which line up perfectly with the source surface. However, a "new" solid is created that I have to assign MAT props and Section props and assign these new components part numbers. I have noticed that these "new" components have no select-able surfaces. And in simulations that I have tried to run using these parts, the simulations stall and do not complete. Is there any way to create a solid component with surfaces that I can select from the solid meshed "new component"? Thank you
  7. I am trying to create a solid mesh of elements that connects two surfaces (an inner surface of a cylinder and an outer face of a cylinder). It it paramount that the inner and outer surfaces which have been meshed with a specific number of elements remain the same. Is there a way to select the inner 2D mesh and outer 2D mesh and create a mesh of elements that connects the two? I have tried to include a picture of the two meshed surfaces with part of the model masked so that the inner and outer surfaces can be visualized. Thank you for your help. -Brad
  8. I'm sorry, this last post was not at all helpful. Of course correct boundary conditions and behaviors must be modeled! That is a given. But as to merging nodes of to adjacent surfaces (which was my question) there was no pertinent information. Any knowledge about merging the nodes of two adjacent surfaces would be appreciated.
  9. I am attempting to correctly set up a fluid structure interaction simulation that implements an ALE group. I have created a "work piece" that is modeled as a SOLID ALE Material. This work piece is allowed to deform/flow into a surrounding "void space" which is also modeled as a SOLID ALE Material. The void space is a component with a space exactly matching the work piece's dimensions cut out of it, such that the work piece fits perfectly within the void space. In other words the void space perfectly surrounds the work piece. Both of these components have been 3D meshed using quad elements. And the element sizes of both of these components are exactly the same. I have been informed that in order for these components to interact together properly the meshes of the void space and the work piece must share the same Node ID's on their common boundaries. I.E. these nodes must be merged. How is this accomplished? Thank you for your help! Brad
  10. The work I do involves creating models that simulation traumatic brain injuries in mouse specimens. Naturally, this type of work requires the modeling very small geometries (on the order of at least 200 micrometers). Because our main interest is the developing stress and strain within the blood vessels found in the brain, very fine meshes are required. I have been utilizing the automesh feature found in the 2D panel to get a surface mesh and then I apply this 2D mesh throughout the rest of the 3-dimensional model. I have noted that the "ELEMENT SIZE" window can only sustain elements of a particular size, and that if a desired element size that is too small is requested, it will not be allowed to be entered into the ELEMENT SIZE field. I.E. if I want an element size of 0.5e^-2 and enter this value into the field, it will not be accepted and it is replaced with 1.0e^-2. How am I able to change the settings on how small of elements I can use? I have looked through the tutorials and not found much help. I have found that clicking in the display window and then clicking "o" will bring up the "options panel" where I can adjust the "zoom factor" from 2 to 1 (thereby allowing for slightly smaller element sizes) but even the zoom factor can only by set to a minimal of 1 (making 1.0e^2 my smallest element size option). Is there another way to be able to decrease element sizes so that I can get a finer mesh within an area approximately 200 micrometers wide? Thank you for any help provided. Brad
  11. Thank you for you initial response. I want to make sure I understand a few of the things you wrote and perhaps added details of what I am doing will prove useful to answering this question. I should mention that I use LS DYNA as the solver. The model itself is still very basic at this point. I am trying to include two parts in the model. One, the inner part, that is comprised of MAT NULL to represent fluid. And second, a part that fully encases the MAT NULL component. This second part will be elastic or viscoelastic in nature. These two parts will subsequently undergo a compression test. In reviewing some of the online demos on the LS DYNA website, it appears that the components used in demos where ALE (arbitrary lagrangian eulerian) interactions are incorporated utilize HEX elements exclusively. When I attempt to 3D volume mesh my components I always end up with one face of the component that has QUAD elements and the rest of the model that is composed of HEX elements. How is it that I can prevent the creation of these QUAD elements? I have noted that shell elements can be used from the 2D automesh option to "drive" a solid HEX mesh through out the rest of a model. But the same issue of having a mixture of shell and solid elements persists. I have resorted to deleting all of the QUAD elements using the TOOLS_DELETE panel after creating my HEX mesh. But I would like to know if there is a better more simple way of creating a HEX-only mesh. Another point about trying to simulate Fluid with Hypermesh/LS DYNA. The demos online that use ALE seem to use rigid materials only when interacting with a fluid. Is it even possible to use material cards that aren't RIGID in these simulations? Another question about "contact interactions". In the two demos that are most pertinent to my work, one uses a AUTO_NODES_TO_SURFACE contact algorithm; while the other uses a CONSTRAINED LAGRANGE_IN_SOLID card and does not incorporate any contact algorithm. What, if any, advantages are gained by using one of these options. And are there any particular situations in which one option is more desirable than the other? I know this is a lengthy post, but any information on the subject is greatly appreciated. Brad
  12. I have been trying to create a solid mesh comprised of only HEX8 elements. I have been using the 3D solid mesh tab. In this tab I choose the ONE VOLUME option. I pick a solid to mesh, and subsequently mesh it. But I noticed that there is a SOURCE SHELL tab where I can choose quads, trias, or mixed. I am able to mesh the entire construct, but in so doing a layer of shell elements is created. This layer of shells is unwanted because I feel that a mixture of shell and solid elements in a single part will yield an error when I run the simulation. Is there a way to avoid this creation of shell elements and attain a mesh that is composed solely of HEX elements? Additionally, this HEX mesh will be treated as a fluid. To do so are there other meshing parameters that must be set? Or is this done in the ALE_REF_SYS_NODE and GROUP panels only? Thank you, any help would be greatly appreciated. Brad
  13. Hello, I am using Hypermesh to try and simulate a compression test of an elastic (or viscoelastic) material encases a fluid. I am new to Hypermesh, and I have been working through many of the tutorials and to familiarize myself with the software. I will be using LS DYNA as the FE solver. I am unable to find any tutorials or information that describe the process of creating a mesh of elements that will be treated as a fluid. Am I to created a solid mesh (my model is currently in 3D) and then be able to assign some sort of parameters to this meshed component that would make it a fluid? Or must I use the SPH capabilities of Hypermesh to have "particles" that are treated as a fluid. I am currently stuck and would greatly appreciate any information that might lead me in the correct direction. Thank you, Brad
×
×
  • Create New...