H-element and P-elements in Autodesk Simulation Mechanical

[tibor121774]

Hi Simulators!!!

Is H-element and P-element a feature in Sim Mech?...If yes how it can be triggered in the software?....Many thanks....

[PipePakPat]

Hello Tibor,

H-element and P-element are words which I have more formally heard described as "H-Method" or "P-Method".  These options from my understanding do the following:

• H-Method, automatically refines the mesh in regions of the model where stress is significant and continues to re-run the analysis until mesh convergence has been reached
• P-Method, automatically changes the element integration order (polynomial order), on a per element bases, to describe the element stiffness and stresses more accurately

The Autodesk Simulation Mechanical software offers neither of these options as described above. 

Instead, one can add refinement points manually and remesh the model, then manually perform the analysis to perform a mesh convergence study.  Or one could perform the automatic Mesh Study Wizard (which does not consider results during its iterations) to approximate the effects of the H-method. 

In regard to the P-Method, one can enable "Mid-Side Nodes" for any part in a model prior to running the analysis.  Activating mid-side nodes raises the element integration order which produces the higher order element advantage offered by the P-Method.  However, there is no automatic determination as to which elements should or should not have this integration order increased.  Mid-side nodes raises the integration order for all elements in that part.

[tibor121774]

Thanks Pat for that info....Is it normal in a solid element to have a Von Mises (smoothed option default setting) of 149 Mpa and unsmoothed result (maximum value) of 351 Mpa....It's more than twice difference...I was thinking this H-method and P-method will solve this problem....As of now I'm running the system with a very fine mesh of 5mm and takes about an hour....Maybe this P-method will run faster....

[PipePakPat]

Hi Tibor,

The difference in stress between the smoothed and unsmoothed display can be reflective of whether the mesh size has converged in that region of the geometry.  A vague overview of this is provided from the following link.  The is also the chance the some stresses don't converge.  In some cases, a mesh convergence study uncovers a singularity.  This is where the stresses continue to climb even though the mesh size is reduced.  Singularities can occur if the same load is applied over a smaller area, such as a nodal force applied to a node.  A boundary conditions can do the same thing, since a reaction force occurs at a boundary condition.  Finally, unresolved geometry can be a source for a singularity.  Sometimes a round, or filet needs to be modeled to resolve the stresses appropriately.

Sometimes it is problematic to create a finer mesh on the entire geometry, since it results in a huge number of elements which slow the solution and slow the review of results.  Other times, local mesh refinement is a problem since it can complicate the job of the solid mesher to generate transitional elements from this refined region to the larger mesh regions of the geometry.

Ultimately, the solution is subjective and requires us to consider these factors when working towards a solution.

[tibor121774]

Thanks Pat for that valuable information...Really FEA is a good tool if interpreted correctly and a dangerous one if misinterpreted incorrectly...That's why I think it's always nice to have some good old tools for approximating stress calculations to check if the fea result makes sense...Anyway as one of my Senior Engineer says when in doubt make it stout...Need more reading for me to master the art and science of FEA....I've just started it this year....

Cheers!!!!

[dmp]

---

@PipePakPat wrote:

Hello Tibor,

H-element and P-element are words which I have more formally heard described as "H-Method" or "P-Method".  These options from my understanding do the following:

• H-Method, automatically refines the mesh in regions of the model where stress is significant and continues to re-run the analysis until mesh convergence has been reached
• P-Method, automatically changes the element integration order (polynomial order), on a per element bases, to describe the element stiffness and stresses more accurately

The Autodesk Simulation Mechanical software offers neither of these options as described above. 

Hello,

Sorry to tag onto an old thread like this, but I am seeking some clarification.  Under what circumstances should I be using Simulation Mechanical vs Inventor for simple analysis tasks? 

I have a very simple test of deformations under static loads, and Inventor 14's easy-to-use convergence options allow me to have confidence that my mesh is as refined as it needs to be for stable results.  I was under the impression that SimMech was more advanced than the built-in stress tools of Inventor.  This makes the lack of a convergence option puzzling.  What about SimMech is preferred for simple static tests like I describe (I know it also does thermal expansion, etc - but I don't need them for my case)

Thanks!

[PipePakPat]

Hello DMP,

The simulation tool offered in Inventor Simulation contains 3 levels of P-method optimization and perhaps 2 levels of H-methos optimization.  This is convenient and should produce good results for linear static analyses.  Just because Simulation Mechanical offers more features than Inventor Simulation does not mean that the features in Inventor Simulation are not satisfactory, or even superior to those equivalent features in Simulation Mechanical.  If your goal is to perform a static stress analysis and you can do so successfully within Inventor Simulation, then that would be the most convient option.  However, if you find that you cannot perform your simulation with enough accuracy based on the available features and capabilities, then it would be time to start looking into either the newly acquired NEi Nastran or Simulation Mechanical.

Most of Autodesk's analysis capabilities were acquired from other companies.  These individual companies did not design their software in a manner that would offer an incremental increase in functionality between software packages at a later date.  Inventor Simuilation is a good tool and can be used confidently within its limitations.  Simulation Mechanical can perform non-linear, time-dependent analyses with motion adn non-linear materials.  Inventor Simulation cannot.  Both interfaces are different.  Inventor Simulation is simple, because it does not have many analysis types or features to contend with.  Simulation Mechanical is a general finite element analysis tool.  It would take some time to learn either interface and they were not design by the same people, so the learning curve is different.  Please let me know if you have further questions or comments regading this issue.

Pat

[dmp]

Hi Pat,

Thanks for the response - you answered my question.  I had been worried that the Inventor FEA was somehow intrinsically less accurate than that performed by SimMech.  I will continue to use Inventor+convergence with a clear conscious and only fire up SimMech when I need the additional features it offers.