A question that comes up often during conversations with our customers and partners relates to CAD simulation. In particular, the question is, “How does CAD simulation compare to SmartSlice™ for FFF parts?”. In this post, we will define what CAD simulation is, detail what needs to be accounted for when simulating FFF parts, and see how CAD simulation stacks up to SmartSlice™.
To start, it must be said that we are very supportive of anyone using simulation to virtually design and prototype their parts. Simulation, when used properly, helps designers, engineers, and analysts quickly predict how their parts will function in the real-world and can drastically shorten the design cycle by slashing iteration times. That said, we also advocate using simulation responsibly. Using the right simulation tool for the right application is paramount to success.
Here, we will focus on structural simulation for the FFF (or FDM) manufacturing method. FFF (fused-filament fabrication) is a very common additive manufacturing process employed by a wide variety of 3D printer manufactures. Using simulation for FFF presents some unique and challenging considerations and certain simulation tools are better suited for FFF than others. Many people are curious to know if they can use CAD simulation tools to analyze FFF parts. But as you will see in this post, CAD simulation is simply not accurate for predicting as-printed structural behavior of FFF parts.
In the context of this post, CAD simulation refers to structural simulation tools that operate within a given CAD program. In traditional CAD-to-FEA workflows the part geometry is created in the CAD program, exported as a BREP file, and then imported into a commercial FEA (finite element analysis) program, where the structural analysis takes place. With CAD simulation, the FEA tools exist inside the CAD program, which is convenient for the user because it removes the data transfer step between the CAD and FEA programs. In addition to providing FEA tools for structural and thermal analysis, some CAD programs offer access to CFD and even injection molding simulation tools.
When doing structural analysis of FFF parts, there are a variety of considerations that must be accounted for to get meaningful answers from the simulation. Let’s look at what needs to be considered:
The materials used in polymer-based filaments are either isotropic or anisotropic. Isotropic materials have the same properties in all directions. Anisotropic materials have properties that are directionally dependent. This is best understood with an illustration. Consider Figure 1 below which shows cylindrical shaped representations of ABS (isotropic) and carbon fiber nylon (anisotropic) materials. The ABS cylinder is isotropic meaning that the stiffness and the strength in the axial and radial directions are equal. The CF nylon cylinder is anisotropic due to the presence and arrangement of the carbon fibers. For anisotropic materials, the stiffness and strength in the axial direction is not equal to the stiffness and strength in the radial direction.
After you determine if your material is isotropic or anisotropic, the next step is to figure out what the properties (stiffness(es), strength(s), etc.) of the material are. Material properties are the foundation of any structural simulation. If you do not know what these properties are, then the results from the simulation will be questionable at best.
Figure 1. Isotropic and anisotropic material illustration.
The FFF method facilitates considerable flexibility in terms of what types of “regions” are printed within parts. In general, the types of regions that are printed can be classified as the following: layer, wall, and infill (see Figure 2). Accounting for these regions in a simulation is one of the keys to analyzing FFF parts. But how does one go about accounting for these regions? Questions that must be asked and answered are:
These questions have no easy answers when using traditional FEA tools.
The hard answer to the question, “what are the material properties of the different regions”, is that, in lieu of some experimentally measured properties for the different regions, micromechanics models need to be created and used to predict the behavior of these regions. Micromechanics models (also referred to as Representative Volume Elements) are essentially parametric FEA models of the smallest repeating unit in a microstructure (region). Building these models requires more expertise and patience than the vast majority of FEA users have, even analysts. But they really are the best way to predict the mechanical properties of the different regions in FFF parts.
Figure 2. General regions types for FFF parts.
The material orientation defines the direction of the material properties at every point in the FEA model. This is particularly important for FFF parts because the properties of the regions are directionally dependent, even when the filament is isotropic (ABS or PLA, for example). Consider the 2 print paths shown in Figure 3. Print path A shows the direction that the horizontal walls are printed and print path B shows the direction that the vertical walls are printed. The simulation tool needs to know what direction these regions were printed to properly orient the material properties. Without a material orientation, the results from the simulation will be essentially meaningless.
Figure 3. Print paths for wall regions.
Meshing and element type
One of the most challenging aspects of simulating parts with different material regions is meshing. First, the geometry must be divided (or partitioned) into the different regions and for 3D geometries, these boundaries are surfaces. For example, there must be boundaries in the geometry between the infill, wall, and layer regions. Once these boundaries are defined, then the geometry must be meshed. Meshing is the process of discretizing (dividing) the geometry into small elements. These are the elements in Finite Element Analysis.
The user must also take into consideration what element type to use. Most CAD simulation tools use tetrahedral (tet) elements for the mesh. Tet elements are commonly used because it is very easy to mesh complex geometry using these element shapes.
A full discussion on meshing and element types is beyond the scope of this blog post but an internet search will yield many educational results.
Now that we know what needs to be accounted for when doing structural simulation for FFF parts, we can finally answer the question, “How does CAD simulation compare to SmartSlice™ for FFF parts?”. Let’s look at each of the considerations mentioned above.
Material properties and type
Most CAD simulation tools are limited to isotropic materials and do not offer the ability to define anisotropic material properties. And while many include a database of material properties, the materials in the database are generic materials and do not include specific FFF materials.
One of the primary benefits of SmartSlice™ is that it includes a database of experimentally tested FFF materials. SmartSlice™ users do not have to spend time tracking down material properties or paying a company to test the material for them. And the SmartSlice™ database includes a wide variety of isotropic (ASA, ABS, PLA, nylon, etc.) and anisotropic (CF nylon, GF nylon, CF PET, etc.) filaments so users can quickly and easily use SmartSlice™ to explore how their part performs when printed with different materials.
This is where CAD simulation completely breaks down for FFF parts. There is no method for creating boundaries between the infill, wall, and layer regions. And even if there was a method, it would be completely manual and would be tedious at best, and borderline impossible on parts with complex geometry. Further, CAD simulation does not consider things like layer height, line width, infill density, infill pattern, wall thickness, etc. so users have no idea what the properties of a given region are. For example, what is the stiffness and strength of an infill region that has a grid pattern, 45% density, and is made from CF nylon? That is an extremely challenging question to answer with CAD simulation.
SmartSlice™ was designed specifically to simulate FFF parts. The technology that powers SmartSlice™ can quickly compute the properties of the different regions based on the material and print settings using the micromechanics models discussed earlier. In addition, SmartSlice™ uses information from the slicer to precisely locate the boundaries between the different regions.
As discussed earlier, defining the orientation for the material properties at every point in the model is necessary for FFF parts because the properties of the infill, wall, and layer regions are anisotropic. In most CAD simulation packages, defining a material orientation is not an option. If it is an option, then the functionality is limited, making it difficult or impossible to define for all the regions in an FFF part.
Defining precise material orientations with SmartSlice™ is done automatically using data from the slicer.
Meshing and element type
Meshing in CAD simulation programs is generally automatic and uses tetrahedral (tet) elements. This method works well for isotropic parts like machined metal parts and many injection molded parts but is not appropriate for FFF parts because FFF parts have different material regions. Without boundaries between these regions, there will be a high percentage of elements that overlap these regions.
SmartSlice™ uses a proprietary meshing algorithm and element type that was designed to handle the complexities of FFF parts. This technology includes a region assigner that accurately locates all the different region in the part.
Perhaps the best way to demonstrate the differences between CAD simulation and SmartSlice™ is with an example. Consider the part shown in Figure 4, which is a lever that is printed from ABS. The 2 cylinders on the left are fixed and a load of 100 N is applied in the direction shown. This is a lightweight part and only has 2 walls, 2 top/bottom layers, and 20% triangles infill. Using a CAD simulation tool, the constraints and load are defined, ABS is selected as the material, and the default mesh settings are used. With SmartSlice™, the constraints and loads are defined, ABS is selected as the material, and the previously listed print settings are defined.
Figure 4. Part with constraints (red) and load (blue).
The results from the simulations are compared to experimental results in Table 1. For full details on the experiment, refer to this blog post. The CAD simulation predicts a part stiffness that is 3.9 times greater than the experimental stiffness and a part yield load that is 4 times greater than the experimental yield load. SmartSlice™, on the other hand, is within 8% of the experimental part stiffness and 5% of the experimental part yield load. The differences are staggering.
Why is there such a big difference between the CAD simulation and SmartSlice™? The CAD simulation model assumes the lever is made from solid ABS while the SmartSlice™ model is constructed using detailed information from the slicer so SmartSlice™ model is able to generate a model that accurately represents the as-printed part.
What would happen to the results if the print settings changed? The experimental results would change, the SmartSlice™ results would change, but the CAD simulation results would remain the same.
Table 1. Part stiffnesses and yield loads.
Simulating the behavior of FFF parts is very complicated and presents many unique challenges compared to simulating traditionally manufactured parts. CAD simulation is simply not accurate for parts made from the FFF process due to all the considerations mentioned in this post. In short, user who want to simulate FFF part behavior must:
These requirements are not realistic in CAD simulation tools. There is a common phrase used by simulation users that says, “garbage in, garbage out”. In other words, you can’t expect to get good results from your simulation if you are using bad inputs and assumptions or if your simulation tool isn’t equipped for your process.
SmartSlice™ is designed specifically for the FFF process and takes into consideration all the issues that CAD simulation does not address. It does this in a highly automated and efficient manner which takes the burden off the user, providing them with realistic simulations of their as-printed FFF parts.