From post-processing and visualization to simulation result exploration

We all know post-processing, because it is part of CAE.

moon-explorationAll text books teach you how important the post-processing is. For most Generation Y engineers, post-processing (and maybe even the whole CAE) almost equals to visualization as discussed in this post.

Yes, post-processing is important: it gives you the chance to justify your one-month (or one-day, or one-week, or one-year) work; it shows you something you want or something unexpected.

If you are lucky, you can even show how your simulation matches the experimental results. If you don’t have any experimental data (unfortunately, nowadays, this is not uncommon any more), you can still justify your results against common sense.

Of course, “right” or “accurate” results are not the goal of simulation.  The purposes of simulation usually  include:

  • to get some data or insights that are impossible, or impractical, or too expensive,  for experiments or physical tests (so that the process or system or physics can be better understood; for PhD students, more colorful figures in papers and thesis) ;
  • to diagnose the problems encountered of a system (or a component) in real-world operation (so that the possible causes or scapegoats can be identified);
  • to predicate the performance of a system (usually not manufactured yet) in ideal or real -world conditions (so that design can be improved before or during prototyping stage);

In the past, simulation, especially CFD simulation, is costly: you need spend at least a few days to prepare the geometry and mesh; then you need wait at least another a few days for the calculation. Of course, you also need time to do the post-processing. So, once you finish the project, you are exhausted.

Thanks to the advancement in computing hardware, and improvements on efficiency of parallel computing, the calculation time can significantly be shorted although the problem (e.g., the mesh) ia getting bigger and bigger.

In addition, because most of meshing tools now can import geometry from CAD directly, are more tolerate to dirty geometries, and are more automated, the pre-processing is no longer a bottleneck for most CAE projects. For FEA, this can be even transparent to the users: users just define loads, contacts  and other constraints on CAD entities, then click the solve button.

Such advancements mean more simulations can be done in the same time frame, and therefore it is practical to incorporate or integrate simulations into the design and even the product lifecycle management.

This means at least two sets of tools are needed: 1) simulation data management; 2). simulation data exploration.

For simulation data management (SDM), or simulation lifecycle management (SLM), most CAE or PLM vendors have some sorts of such tools. For example. MSC has SimManager; Dassault has SIMULIA SLM; ANSYS has EKMSiemens has Teamcenter for Simulation. Chad Jackson wrote several posts with detailed analysis on some of these tools on Engineering-matters.com (now moved to Engineering.com).

However, for simulation results exploration, it is quite disappointing . Most  SLM/SDM tools simply are incapable of exploration: they mainly can manage the simulation data and extract some meta data. Probably a few can provide some sorts of simulation template for new engineers.

Last year, Tecplot launched the Tecplot Chorus, which is a quite impressive and interesting product for CFD. But as an independent (from solver vendors) tool, it is impossible to directly compose a test case for immediately verification of the explored results. I haven’t seen any other similar products in the market.

Simulation data exploration is an extension of post-processing: deriving valuable correlations among independent variables or designs from results of multiple (often independent)  simulations.  This is the real value of having piles of simulation data.  Without simulation data exploration, SLM is merely a simple file repository for bunches of unrelated simulation files on the storage server.

Essentially, simulation data exploration is different from design optimization, a hot topic in CAE industry. Exploration is trying to discover the knowledge buried upon the massive amounts of simulation data. This is essential for improving the efficiency of simulation in an organization, and can be an indispensable part of IP portfolio.  Currently, most organizations don’t have such knowledge bases, and therefore have to rely on “experiences” of  some “old” engineers to train “new” staffs or to build some so-called best-practice templates. This is one of the major reasons why simulation has not realized its potential in most organizations.

In short, simulation results exploration is a necessity, not a luxury.

 

About shengwei

Dr. Ma Shengwei has spent last 20 years in numerical simulation, including writing CFD solver and applying numerical simulation in solving various real-world problems, including pollutant dispersion, sediment transport, heat transfer, membrane separation, water and waster treatment, centrifugal pump, natural ventilation, and separation systems.

You may get his detailed info on his expertise from Linkedin .

Comments

  1. Great stuff! I’m not sure I agree with the statement that the problem of meshing nasty geometry has been solved- but definitely agree that as we all move along the maturity curve with CAE, results exploration will be huge!

    • Hi Jeff, thanks for your comment.

      no, the meshing problem has not been solved. It is still one of the biggest obstacles for simulations with complex geometry.

      For designers, there are quite a few CAD-embedded simulation tools (not only FEA, even CFD) that hide the meshing options. Of course, for any serious analysis, this is surely not sufficient.

      But the trend is the meshing tools should be able to “understand” the geometry, and try to give sensible initial settings (to make it automated). The advancement in this aspect is impressive in last a few years. But the bigger challenger for meshing tools is how to understand the physics. It seems no meshing tools in the market can read the problem setup in downstream. Even experienced engineers often need trial-and-error.

      So, we still have a long journey to get fully automated meshing.

  2. Informative article. May I blog about this with a link back to you?

  3. I actually blog likewise and I am crafting a thing similar to this excellent post, “From post-processing and visualization to simulation result exploration | CAE industry analysis
    and technical resources”. Will you mind in the event
    that Iutilise a little of your personal suggestions? Thanks for your effort -Roy

Leave a Reply