Meshless Simulations Offer Lightening Fast Results for Large Assemblies

An engineer’s reaction to seeing how fast SIMSOLID solves large assemblies? After verifying the results of course. (Image courtesy of 20th Century Fox Television.)

Just the thought of meshless simulation is bound to turn many heads in the computer-aided engineering (CAE) community.

However, what will make a CAE engineer seriously consider throwing money at this technology is how lightening fast it produces results for massively large assemblies and complex geometries.

According to Ken Welch, CEO of SIMSOLID, once engineers are comfortable using the software by verifying and validating results, they will see how it complements their simulation portfolio by adding the ability to quickly assess large and complex geometries.

How to Run Simulations Without a Mesh

Modal Analysis of this 545-part assembly took only 49 seconds to run on a standard laptop. (Image courtesy of SIMSOLID.)

SIMSOLID works on a similar principle to p-version finite element analysis (FEA)  in that it is based on increasing the order of polynomial and non-polynomial functions through an iterative process until convergence.

What is different between the two simulation algorithms is how the geometry is broken down. Using p-elements still requires a mesh, albeit a coarse one that is bound to give an h-element CAE engineer a heart attack.

“We classify the geometry and features as regions,” said Welch. “Think of a part as thin and thick. Think of a part with a through hole. We can classify that and have equations that know what happens to these parts. Once we have the equations, we can increase the order of the polynomial for increased accuracy in the regions of interest.”

“In finite elements, you typically have to resolve it numerically. We are no different than anyone else,” added Welch. “The difference is that we are always adaptively enriching our equations. It’s automatically adaptive and looks at errors.”

In other words, SIMSOLID creates interface functions for the different regions. These functions try to define the region, as opposed to FEA, which defines a geometry broken up into simplified pieces. The simulation then iterates to discover a solution and reduce the errors, much like FEA.

Once the results are ready they are mapped back onto the 3D model. Though the results are volumetric, the mapping in SIMSOLID is currently only done on the surface. This means that there are no cutting planes available at this time. Bummer.

How to Use SIMSOLID and Its Capabilities

SIMSOLID can be used as a stand-alone program or as a fully integrated add-in for SOLIDWORKS or Fusion 360. It also supports standard STL files. Alternatively, the stand-alone program can pull down CAD files from the Onshape cloud and bring them down into the desktop.

SIMSOLID is also smart enough to know when an imported file is linked to a previous study. It will open up the new geometry within that study so that boundary conditions and loads don’t have to be reapplied. The program will also grab material properties if they are included within the original CAD file.

When a user downloads SIMSOLID they automatically get the standalone program and both of the add-ins.

Unfortunately, SIMSOLID is limited in its physics portfolio at this time. The software currently focuses most of its abilities in the structural analysis field. They include:

  • Static stress
  • Vibration/modal analysis
  • Thermal analysis
  • Thermal stress
  • Geometric non-linear static stress

However, Welch explained that the algorithms that govern SIMSOLID can be used for other physics. Therefore, other simulations might be available down the line.

A simulation tool that is so quick to solve such large assemblies almost begs to be accompanied with optimization and systems engineering technology. However, engineers will be shocked to see that this is not currently the case. Welch noted that these are a priority for the software’s development

“We are talking with a number of people that work in the optimization world,” said Welch. “It’s a little unnatural we don’t do it today. We are not optimization guys today, but it’s an easy way for us to partner.”

“Connecting to large scale 1D simulations is another avenue we can expand as well,” added Welch. “There are a lot of places we could go but we are a very young company.”

Benefits of Meshless Simulations

The obvious benefit to not needing a mesh is that the geometry doesn’t need to be simplified anymore. In SIMSOLID, the complete geometry is classified and studied as is. This will save a considerable amount of time in pre-processing.


A crank shaft given to Welch by a third party colleague that has many parts, faces and slivers in its CAD geometry. Without simplifying the geometry, SIMSOLID performed a modal analysis in 9 seconds. Welch’s buddy would take an afternoon performing the study using traditional FEA. (Image courtesy of SIMSOLID.)

However, where SIMSOLID truly shines is its solution speed for large assemblies. During Welch’s presentation to ENGINEERING.com, he would show the simulation of various assemblies ranging from 500 to 700 parts, nuts, bolts and all. The modal analysis would take 50 seconds. A few models solved faster than Usain Bolt on the 100-meter dash. All of this was done on a standard laptop.

The simulations are able to run quickly because of the model classification. Welch explained that the software can tell the difference between a plate and a bolt, which gives it a big advantage.  This allows the software to better define these parts with fewer equations. If the equation that SIMSOLID assigns to a part turns out to be incorrect, then this is assessed in the iteration process.

Additionally, Welch noted that, since SIMSOLID doesn’t have 50 years of legacy code, it has been written from the ground up for parallel processing.

“All CAD embedded simulation tools do a great job,” admitted Welch. “The challenge with CAD embedded is they break out single part. On that scale, the challenge is: what’s the load? FEA is a very exact solution to the wrong problem. You simplify it and break it up and worry about convergence but forget about the simplification and loads.”

Welch didn’t dodge the question everyone reading this is likely thinking: How do the results compare to traditional finite element analysis (FEA)? He explained that this comparison is often made when talking to big companies that are looking to use his product.

“We get similar answers to FEA, but we solve on a bigger level with large system. If you need to, you can then drill down more for more detail, but SIMSOLID isn’t a product to replace anything,” Welch said. “Its intended to allow designers to ask another question. Can you work in a design environment at the speed of design?”

“The problem with simulation in a design environment is that you have to break it out and apply analysis. This takes time and makes you think more like an analyst than a designer,” he added. “If I have a design question, an analyst would say ‘give me two to six weeks.’ I want the answer right now. The idea to work in design was to solve in a factor of seconds. It’s really a design directional tool.”

This still might all sound just a little too good to be true for a seasoned CAE engineer. However, Welch is prepared to put his software where his mouth is. Organizations that have shown interest in SIMSOLID typically send Welch some test models to compare with their known results. Considering how fast these simulations run, he invited any engineer to send him their models. Challenge accepted?

To learn more about SIMSOLID, read: SIMSOLID’s Meshless Simulation Software Releases Professional Edition.