Jump to content


  • Content Count

  • Joined

  • Last visited

Posts posted by Amar17

  1. Hello Everybody,

    We keep eye on element quality criteria in order to get a good quality mesh. Anybody would please tell me in what way does every element criteria (such as Warpage, Aspect ratio, Jacobian, skew etc) affect the results? I know that ideal element shapes are square and equilateral triangle (for 2D elements) and any diversion from the ideal shape would lead to negative effect on the results. But what effect does every criterion has on the result? 

  2. Hello Everybody,

    To verify the results and to check the mesh quality, one method we use is to refine the mesh and check the results. To increase the mesh density, isn't it that we have to re-mesh the model again ?

    Particularly for the case of analysis of large assemblies, the task may become a cumbersome one. And for the case of explicit analysis it becomes worse since we also need to take care of the time step. Then how is it possible to refine the mesh with least possible efforts? Please guide.

  3. 6 hours ago, George P Johnson said:


    A good mesh depends on various factors like element type used, element size, element quality....etc. So it is difficult to comment how good is this mesh, but based on the quality parameters and mesh it seems fine.

    The critical regions are where we expect high stresses and those areas has to be captured well. In those regions choose a mesh size that is small enough to accurately follow the shape of the parts, particularly around small features in critical regions. Mesh density is a significant metric used to control accuracy of results. A high-density mesh will produce results with high accuracy. However, if a mesh is too dense, the overall number of elements and resulting degrees of freedom will be high, requiring a large amount of computer memory and long run times.

    The most basic and accurate way to evaluate mesh quality is to refine the mesh until a critical result such as the maximum stress in a specific location doesn’t change significantly as the mesh is refined.


    Thank you sir! 

    The minimum length in my case here is 1.5 mm and with that I am getting the aforementioned aspect ratios. Since I am solving the problem in Radioss, with a transient load, I also need to take care of the time step. Apart from the element type used is my mesh still okay?

  4. Hello Everybody,

    I am trying to bring a good quality hexa mesh. The mesh I am getting is good in almost all the quality criteria except 'aspect ratio'. I am getting the aspect ratio values in range 3 to 5. I have read that the acceptable aspect ratio is below 5, so my mesh should work. But what I am expecting is  to get a good quality mesh in the area since it may be the region of high stress.  My question is how good my mesh is if it is to rate between 1 to 10? (I am getting a very good mesh for rest of the quality criteria) And to what extent my results may get affected if I fail to bring a proper mesh in the region?

    The attached illustrative image may be referred. The 3D mesh will be connected to 2D mesh and the the analysis is explicit (Radioss)

    Please advise.




  5. Hello everybody, my question is, what yields more stress static or dynamic (transient) analysis? 

    According to what I could know up til now, it is dynamic analysis in which the stress  values are expected to be higher in case of the static analysis.


    It it always likely to happen that stress values will be higher in case of the transient analysis if I perform both analyses for the same structure?  Please guide.


  6. On 9/12/2017 at 10:07 AM, George P Johnson said:

    Hi Amar,

    The nodal time step equation is : sqrt (2m/k), where m is nodal mass and k is equivalent nodal stiffness. A scale factor will be also used to calculate the time step.

    Please go through https://altairuniversity.com/learning-library/considerations-about-time-step/ which is a webinar showing time step calculation in RADIOSS.

    RADIOSS will calculate nodal and elemental time steps of the model and will choose the highest of these to run.

    From Check Elems (F10) panel it is possible to determine the approximate minimum time step for explicit solvers based on element dimensions. 

    Sir, I tried checking time step with Check elements Panel, but an error popped up:- "HM_CALC_TIMESTEP_func: Division by zero error"

    Does the F10 panel calculates considering both the elemental and nodal time steps?

    Also, What care we should take to keep the nodal time step maximum?  

  7. Hello everybody,

    I want to perform an explicit analysis in radioss. I have read that we follow a time step of 1E-06 sec for steel with minimum element size of 5 mm. In my case the element size goes up to   1 mm.

    I have gone through the book "Crash Analysis with Radioss", according to which we have to look for two formulae to calculate the time step- one is elemental time step and the other is nodal time step. With an approximate sound speed of 5000 m/s through steel, I calculated the time step of 2E-07 sec [i.e., 1 mm / 5000 m per sec]. But I am not able to figure out how to calculate the nodal time step? Would somebody help me out with this?

    My question is - how much smaller time step would I require to take (for steel) in this case? [Please suggest me an optimum- rough estimate for the element size of 1 mm  for steel by taking into consideration the nodal time step also].

  8. 21 hours ago, Q.Nguyen-Dai said:

    I think you can do it in shell with variable thickness.

    Thank you for that sir! But what about connection between the two bodies? Should we merge the nodes at the connection or provide some connection? For the latter case what type of connection you would suggest me for an explicit analysis (in Radioss)?


  9. This means we can store different element types of a single solid in different component collectors and assign them different formulations. Wouldn't it create any problem?  

    Because, once when I was analyzing an assembly, I stored all the component  in a single collector and wasn't able to find any penetration. Please tell me if I need to take care of any such probable concern while defining element formulation for different element types in a single solid? 

  10. I have a component meshed with different elements (Hexa, Penta and tetra). The problem I am facing is while defining the element formulation of the component. How one can define element formulation in such case? 


    The problem is the same for the case of 2D plates when we encounter quad and tri elements in a component.


    I also searched the answer for this in the book "Practical aspects of FEA" but couldn't find it. Please guide me through the problem.

  11. Hello Everybody, I am trying to mesh a stepped cylinder. In HyperMesh we find the surfaces of cylinder get divided into to halves. In my case, as these halves are connected, I am not able to get an uniform 2D mesh at the cross section. I have tried to toggle the partition lines but couldn't solve the issue. Please tell me, how may I get this resolved. 


    I am attaching the images of my problem



  12. On 8/18/2017 at 1:56 AM, Amar17 said:

    Thank you Prakash sir,

    I have shared two illustrative images of my question. Please tell me if I am correct- "For 2D-2D T-connection I need to get all the nodes connected at the joint".

    Also, which type of RBEs are used for such connections in explicit analysis (Radioss 12.0)? 




    @Prakash Pagadala Sir, please tell me what to for 2D to 3D mesh connection in cases as illustrated in aforementioned images. 

  13. On 8/20/2017 at 1:03 PM, George P Johnson said:

    Hi Amar,

    I think for 'good mesh' you are meaning a fine mesh. Fine mesh does not mean good mesh.The purpose of meshing is to get solutions in discrete locations. Good mesh is the mesh that serves your project objectives. So, as long as your results are accurate enough for your project, your mesh is sufficiently good. The most fundamental and accurate method for evaluating mesh quality is to refine the mesh until a critical result, such as the maximum stress in a specific location converges (i.e. it doesn’t change significantly with each refinement).

    Normally  if a mesh is too dense, it will require a large amount of computer memory and long run times, especially for nonlinear and transient analyses. In such cases the critical regions are meshed fine so that the results at those region are well captured. The user has to judge on this based on the model, analysis and experience.

    Please go through Practical Aspects of Finite Element Simulation (A Study Guide) ebook which contains all practical concepts on FEA with examples.

    Thank you sir for your reply! But I mean what I wrote. Also I have gone through the book you suggested but I was unable to get what I need to know. I am putting my question once again here -

    1) Does the meshing in one portion of a single component affects the results in other portion of the same component?

    2) Does the meshing in one component of an assembly affects the results in other components of the same assembly?

                        What do I want is to mesh critical portions with good mesh quality and remaining portions with the mesh having quality only good enough to capture the features of the component. Please give me guidance for the approach I am thinking of.

  14. Dear all,

    I have following doubts in meshing:

    1.  What if I do good meshing in one portion of solid (say critical portion) and poor quality mesh in some other portion, do the results in the good mesh region are affected by the results of poor mesh region. Or do they remain unchanged? If possible, please explain the reason for the answer. 
    2. What if we couldn't maintain uniform mesh size in non linear analysis?, If I use small elements to capture the geometry and large elements to keep nodes as less as possible then it causes sudden transition between the two. Does it affects the results badly or increases the solving time?

  15. Dear all, I have following questions regarding the tetra element checks


    1) Which quality criterion (skew or vol skew or equia skew)  is more important for tetra mesh. I am using HM 12.0 version. Are all the criteria essential to be satisfied.

    2) In general, out of the many criteria (warpage, aspect, skew, tet collapse, cell squish, jacobian, equia skew, vol skew, vol AR, min-max length) which one we need to focus more to get a good quality mesh and which we may ignore?

  16. On 8/18/2017 at 11:15 AM, George P Johnson said:

    HI Amar,

    As a general note it was mentioned. Ensure the connectivity and you can continue.

    I am using HW 12.0 Radioss. When I go to Toolbar->mesh->assign->element type, I see that only pyramid5 elements (in 2D and 3D page) are available (See the attached image).  Does it mean that only 1st order pyramids could be generated at the interface of the hex and tet mesh?


  17. On 8/16/2017 at 9:15 AM, Prakash Pagadala said:

    Hi Amar,


    No you don't need skin elements for 2D to 2D connections. You can either do a nodal connectivity or use RBEs


    Can you share an image? You can also have freeze contact between 2D and 3D mesh. 

    Thank you Prakash sir,

    I have shared two illustrative images of my question. Please tell me if I am correct- "For 2D-2D T-connection I need to get all the nodes connected at the joint".

    Also, which type of RBEs are used for such connections in explicit analysis (Radioss 12.0)? 



  • Create New...