Finite element analysis (FEA) is a powerful tool for engineering design, but it can also be time-consuming and computationally intensive. If you want to optimize your FEA simulations and reduce your waiting time, you need to apply some smart strategies and techniques. In this article, we will share six tips that can help you speed up FEA simulations without compromising accuracy or quality.
Depending on the type and complexity of your FEA problem, you may have different options for choosing the solver that performs the numerical calculations. Some solvers are faster, more robust, or more suitable for certain types of problems than others. For example, direct solvers are generally more reliable and accurate, but they require more memory and CPU time than iterative solvers, which are more efficient for large and sparse systems. You should also consider the parallelization capabilities of your solver, as some solvers can take advantage of multiple cores or clusters to speed up the computations.
Choosing the right solver in Finite Element Analysis (FEA) depends on the nature of your problem and the capabilities of the solvers available in your FEA software. Here are some common types of solvers and when they might be suitable: 1. Direct solver - for smaller models computational cost is less and for larger models computational cost is more. 2. Iterative Solver - mostly used for non linear analysis. 3. Implicit Solver - best for static problems. 4. Explicit Solver - best for dynamic problems. 5. Parallel Solver - if parallel computing is available. Solution time could be significantly reduced. Always refer to the documentation of your FEA software to understand the strengths and limitations of each solver.
Choosing the right solver is important & effective. For example, if a thin plate (2D) element gives you the same result as of a cubed element, then it's better to go with thin plate as it will take less computational effort & give similar results. This depends on the problem one is trying to solve and what type of data is needed.
CAE Engineer
Selection of the right solver for finite element analysis (FEA) depends on various factors, including the type of problem you're solving, the complexity of the geometry, material properties, and the desired level of accuracy. FEA software suite that covers a broad range of applications, including structural, thermal, fluid dynamics, and electromagnetic analyses is more applicable. Software, that is suitable for complex and large-scale simulation preferable.
Model order reduction (MOR) is a technique for reducing the computational complexity of mathematical models in numerical simulations.
choose the right mesh size, solver, and material properties. simplify the geometry like 2D shell type or 3D model. The memory size of the CPU is a higher configuration to be selected.
One of the most effective ways to speed up FEA simulations is to simplify the geometry of your model as much as possible, without losing the essential features and characteristics of your design. You can do this by removing unnecessary details, such as fillets, chamfers, holes, or bolts, that do not affect the structural behavior or the boundary conditions of your model. You can also use symmetry or periodicity to reduce the size of your model and replicate the results for the full domain. Simplifying the geometry can significantly reduce the number of elements and nodes in your mesh, which in turn reduces the computational time and memory requirements.
Right idealization is key even today despite so much of imrpovents in computational efforts. And, in some scenarios, more than the accurate end results, behaviour of model gives out many orientations to design/redesign further.
To accelerate finite element analysis (FEA) simulations in the construction industry, several strategies can be employed. Mesh optimization, parallel computing, and selecting the appropriate solver are key. Simplifying the model, load balancing, and adaptive meshing help streamline computations. Reduced integration, efficient preprocessing, and parameter tuning further enhance speed. Additionally, consider modeling simplifications like 2D simulations or beam elements. Striking a balance between accuracy and efficiency is paramount. Regularly updating techniques based on advancements is advised to achieve faster and reliable FEA analyses.
Perform CAD geometry simplification like deleting the features that are not bearing any stress. Some features could be fillets, chamfers, holes etc. Deleting these features will reduce the mesh count and hence reduces computational cost. As the calculations are performed over these nodes. Try to discuss with the product owner (CAD Engineer) about these features like why they have provided these features and what is the role. Also, there might be some cases where you will find that the geometry is symmetric in some way. Divide the geometry. This will further reduce mesh count. However, plot your results correctly considering that symmetric approach is used. Provide washer wherever there is a bolting. This helps to capture precise results.
Simplifying the geometry is the key factor in successful simulation. Never try to shoot a fly with a tank. Sophisticated model with detailed geometry may look so fancy, but after several hours of accurate modeling it may not run at all. There are several components that could be replaced with simple elements or operations, such as bolts and nuts. Also, choosing the best element type is other key factor, some elements would give the same results when modeled as frame, shell or, solid element. But frame elements would be much easier to use. Last but not least mesh type is one important factor, if there's no much discontinuations in geometry (e.g. holes, cuts, etc.), triangular mesh would be sufficient and will give accurate results.
Simplify geometry for FEA by removing intricate details, utilizing symmetry, replacing complex shapes with simpler representations, merging similar parts, and employing idealization techniques such as using beams or shells. Additionally, consider mesh decimation to reduce element count and lumping masses or stiffness to key points. Validate simplifications by comparing results with the original model to ensure essential behavior is retained while improving computational efficiency.
Another important factor that affects the speed and accuracy of your FEA simulations is the mesh, which is the discretization of your geometry into finite elements. You want to have a mesh that is fine enough to capture the stress gradients and the boundary conditions, but not too fine that it increases the computational cost and the error. You can use mesh refinement techniques, such as adaptive meshing or local meshing, to adjust the mesh size according to the level of detail and accuracy needed in different regions of your model. You can also use higher-order elements, such as quadratic or cubic elements, to improve the accuracy with fewer elements than linear elements.
Also, for analysis like modal analysis or any other dynamic analysis (such as harmonic response analysis) it is okay to have a coarse mesh as the idea is to capture the stiffness of the part/assembly.
Mesh refinement is a key factor in crack propagation and fracture mechanics relevant problems. The mesh refinement at the crack tip, where the crack propagation occurs, is the dominant issue in fracture mechanics, preserving course meshes in other regions surrounding the cracked region to save time for analysis.
Mesh refinement is vital in FEA to ensure accurate representation of complex structures under various conditions. The quality and density of the mesh directly impact result reliability. Some strategies for mesh refinement: 1. Local refinement 2. Gradual refinement 3. Adaptive refinement 4. Convergence study 5. Mesh quality In conclusion, mesh refinement in FEA is a nuanced process that involves strategic decision-making to balance computational efficiency and result accuracy. By employing above mesh strategic techniques, engineers can ensure that their FEA simulations yield reliable and precise insights into structural behavior. Validation through experimental data further enhances confidence in the accuracy of the refined mesh.
?FEA Enthusiastic ?Mechanical Engineer ?CAD ?CFD ?Us Permanent Resident
Right mesh selection drastically speeds up the FEA simulation. There are tons of mesh types available: solid elements (tetrahedron, prism, Hex, pyramid, etc), Beam elements, Shell element, line elements, pipe elements, 1D, 2D element, etc. Each element has unique advantages and disadvantages. You should know which element has to be used based on the structure. Let's say, you need to model a large thin structure. You can generate a lot of solid elements to simulate it. In that case, simulation will be very time-consuming as well as expensive. Alternatively, by using the Shell element ( less element), you can capture the same physics very fast with more accurately. Selecting the right mesh type is a pivotal step to speed up FEA simulation!
Coarse FEA meshes are ideal for initial design exploration or dynamic analysis of large structures, offering faster computations. Fine meshes excel in detailed simulations such as stress concentration areas or contact/friction analysis, ensuring accuracy in localized effects. The choice depends on the analysis goals, with coarse meshes balancing efficiency and fine meshes providing precision in critical areas like thermal analysis of intricate geometries.
Submodeling is a technique that allows you to focus on a specific region of interest in your model, without having to analyze the whole model at the same level of detail. You can use submodeling to perform a more refined analysis of a critical or complex part of your model, while using a coarser mesh or a simplified geometry for the rest of the model. Submodeling can save you time and resources by reducing the size and complexity of your FEA problem, while still providing accurate results for the region of interest.
Start with smaller, simpler models to test settings and identify potential bottlenecks before running large-scale simulations.
In my experience, I will cut a large part into small parts and mesh individually. This will speed up the FEA simulation. or we call it submodeling
Submodelling reduces the convergence time period by focusing only on the part of the model that is of interest. This methodology is applicable to large complex models.
I recommend analyzing the case, sometimes the way we propose a solution is not the best, so a good way to start is to divide a large problem into smaller problems, that is, simplify the models and reduce the sections to be analyzed by separate
Submodeling is not available in all fea codes. When available it can perform greatly if special care is taken on the influence of the “light mesh” or simplification applied on the model alters the results
Before running your FEA simulations, you should always validate and verify your model to ensure that it represents the physical reality and that it satisfies the mathematical assumptions and criteria. Validation is the process of comparing your FEA results with experimental data or analytical solutions, to check if your model is realistic and reliable. Verification is the process of checking if your model is free of errors, such as incorrect inputs, boundary conditions, units, or mesh quality. Validating and verifying your model can help you avoid wasting time and resources on running simulations that are inaccurate or invalid.
Model verification and validation is a crucial stage to ensure the accuracy and dependability of results. ASME Sec VIII Div 2 recommends to achieve this by mesh sensitivity review and equilibrium check for finite element analysis, e.g., check of hoop stress in a component away from structural discontinuity and a check to ensure that global equilibrium is achieved between applied loads and reactions at specified boundary conditions.
The model boundary conditions and loadings should be the first things checked on a new model. Until these items are right the rest of the analysis is suspect. If later you change the model geometry or loads make sure the reactions at the boundary conditions are still valid.
Validate and verify your model. The final check of the calculation model is a necessary and obligatory matter of course. However, this is not enough, it is necessary to look from the point of view of the customer, who does not have to be an expert in calculations, whether the calculation and the calculation protocol are transparent, easy to read and sufficiently informative. Jet such a calculation protocol is successful.
Yes sure checking the coherence of the results is a best practice. Before that the question of the model relevance must be asked. Is my model in line with the reality of the physics ? Are my boundary conditions relevant ?
Nunca esta de más efectuar una verificación manual simplificada, así uno sabe que se puede esperar como resultado en la simulación con fea. Ahora bien, no siempre es adecuado una verificación mediante el soler, porque eso puede arrastrar errores originados por el ingeniero, generando una falsa sensación de que todo está bien
Finally, you should monitor the convergence of your FEA simulations, which is the measure of how close your numerical solution is to the true solution. You can use different criteria to monitor the convergence, such as the residual norm, the displacement norm, the energy norm, or the error estimate. You should set a reasonable tolerance for the convergence, which is the maximum acceptable difference between the numerical solution and the true solution. You should also use appropriate convergence methods, such as direct or iterative methods, to ensure that your FEA simulations converge in a reasonable number of iterations and do not diverge or oscillate.
Tighten convergence criteria if you need highly accurate results, but loosen them if acceptable accuracy can be achieved with faster run times.
Speed up FEA simulation: Assuming gadgets are perfectly high speed processor and high resolution screen, I can suggest to perform parity check to detect errors in input data to ensure a secure data transmission process.
Using HPC (High Performance Computing) is a great way to speed-up the calculations. Distributing the workload across multiple processors, especially for complex or computationally intensive analyses can significantly accelerate the simulations.
Some of the ways one can speed up FEA Simulations is by optimizing the mesh quality of a design. Furthermore, a designer can utilize joint selections instead of contact-that is if it is more reliable and feasible. Lastly, a good computer system with a higher processing power, RAM and SSD storage goes a long way in ensuring faster and efficient simulations.
Leaning heavily on model simplification is my go to. If there is ever an ability to use symmetry or even a 2d plane that is representative of a section of a round model (think along the lines of Ansys Fluent CFD as applied to rocket thrusters). By simplifying the model down to a symmetric and/or a simplified 2d planar model the solver is able to complete the analysis in a substantially shorter timeframe.
Use the correct element type for the given physics. Not everything needs to be a tet or brick. Beams and shells exist for a reason.
Boundary conditions! When reviewing a very large ANSYS model for compressor impeller fatigue analysis by a renowned billion dollar consultant results showed an odd frequency response. All it took was having them change the exit annulus model to give a slight change to give outward accelerating flow. Saw the same possible error in a NASA paper and informed them. Also let one of my consulting customers know for performing a successful blade analysis with ANSYS.