Dr. Jan Cervenka, founder of Cervenka Consulting, explains that there are no general capabilities an engineering firm needs in a CAE software.
“It depends on their specific needs, requirements and the type of structures they are designing or analyzing,” he said. “However, some general features should include ease-of-use, user support, validation and robustness of the software.”
Srinivasa (Ravi) Shankar, director, Global Simulation Product Marketing at Siemens PLM Software, agreed. He said, “the ability of the software to handle the types of models that the firm needs to build and simulate is key.”
Some large organizations may need a workhorse – a full CAE platform complete with a glut of structural analysis types and some multiphysics structural interactions, among other features.
All your organization might need is the usual suspects of FEA structural simulation. But, if you need a more in-depth CAE tool, you will have to investigate which vendor will make your team say, “Roger.” (Image courtesy of Bad Hat Harry Productions and Gramercy Pictures.)
No matter your budget, or finite element analysis (FEA) lineup, you might need to perform a verbal interrogation of vendors to walk away with a true mastermind of a package for simulation capabilities.
“Structural FEA has the capability to influence engineering at multiple levels – from mainstream solutions that provide trends and insights to guide product development, to high end solutions that aim to match real-world data,” explained Vikram Vedantham, senior manager, simulation business strategy at Autodesk. “Picking features and capabilities is determined by the time of use, the persona involved, the level of depth, the geometry, the nature of the design, its use case and the size of the firm.”
So where do you begin?
1. Does the Simulation Software Perform All the Assessments You Need?
Linear statics is great and all, but when do you need more? No one can answer that question better than your organization, keeping the future in mind.
Autodesk Nastran in-CAD. (Image courtesy of Autodesk.)
As your organization becomes larger you might need a more robust simulation platform, complete with additional features you never considered necessary during the purchasing process.
Industry trends will give you an idea of how the needs of other organizations are changing the CAE industry. These needs might be your needs in the future.
For instance, take the boom of 3D printing. Your organization may not be designing products with this process today, but to assume you never will might be far-fetched.
Bjorn Sjodin, VP product management at COMSOL, suggests why you might want to look at CAE software with an additive manufacturing flair:
“Additive manufacturing is becoming extremely important. With it follows a demand for a wide range of simulation tools that can be used in additive manufacturing simulations. This is still an emerging industry so it is necessary for simulation tools to be general and extensible to cover all new manufacturing methods that are constantly being developed in this area.”
If you are still not sure which features you will need in the future then pick an FEA software with a pay-as-you-go option or one that integrates into your computer-aided design (CAD) tool. These will typically have a lower price tag and licenses that won’t weigh you down if you decide to jump ship.
2. Is the CAE Software’s Materials Library Customizable or Include Composites?
Composites are everywhere these days. Don’t limit your material capabilities if you want to get those bids.
Visualization of a composite’s ply orientation in HyperWorks’ HyperMesh. (Image courtesy of Altair.)
“Engineering of materials to fulfill design needs is being employed more and more,” said Richard Mitchell, lead product marketing manager at ANSYS. “Accommodating the properties and behaviors of these materials is essential to optimize performance and reliability of new designs.”
Mitchell notes that the number of materials used in simulations now is getting steep. Everything from human body parts, to rocks, wood, elastomers and more are being used. Engineers will need extensive material libraries to accommodate these simulations.
Additionally, users will need to define their own material properties to represent engineered materials and composites.
When including new materials into your simulations there are two schools of thought: you can either test the materials in the real world or you can simulate the part or composite to virtually determine its properties.
“Composite definition is complex with most software solutions and takes a lot of time,” said Vikrant Srivastav, junior FEA applications engineer for SimScale. “Delamination can be especially tough to model and not very robust in case of high deformation. Challenges can be overcome by good software designs and/or model simplification.”
The challenge engineers face when defining a composite from experimentation is that these materials are anisotropic. Therefore, they will have different material properties along different directions and load paths. Most other materials used in simulations are isotropic so they behave the same in all directions. This makes them difficult to test, especially for the part’s geometry.
“When modelling anisotropic materials, you must define the material orientations and there may be some complicated meshing and partitioning of the geometry that is involving during pre-processing, depending on the type of composite material,” explained Vedantham.
Experts like Dominique Lefebvre, head of product management at ESI Group, argue that simulating the manufacturing process is the only way to truly capture the material properties of the composite.
“The manufacturing history, inherited properties and assembly should be chained step-by-step to accurately simulate the behavior of composites or composite/metallic structures,” said Lefebvre.
Brent Lancaster, principal support engineer at ESRD added: “If the fiber directions are discontinuous, the stresses/strains at those locations will be discontinuous and are not very useful in detailed analysis and prediction of failure. It’s important to identify [FEA] tools which support very thin elements with high aspect ratios, as required for a ply-by-ply analysis and can represent the material directions continuously. These tools must also have inherent solution verification capabilities to ensure accurate solutions.”
Long story short, be it composites produced layer-by-layer or through injection molding, engineers will need to know how the fibers and weaves are oriented. This will require some multi-scale thinking and perhaps experimentation. Either way, you will need the opportunity to expand the material and failure mode libraries.
3. Does the CAE Software have a Simple UI or other Democratization Tools?
Simulation apps can create a little tool for non-CAE exports to play in without fear of breaking anything. The CAE expert puts in checks and limits to the app’s abilities. This will give the power of simulation to more people in the industry.
Simulation app being shared on the COMSOL Server. (Image courtesy of COMSOL.)
The idea is that most simulation tools are far too complex. The hope is that new tools, or ways to simplify complex CAE solutions will boost the number of simulation users.
“Firms need to democratize the use of FEA tools, which will reduce the time wasted by the analysis specialists downstream,” said Jose Coronado, product manager of Creo manufacturing and simulation applications at PTC. “The FEA software should enable a simulation-driven design process. It should be a tool that allows any engineer/designer to perform sensitivity analysis early, analyze trade-offs, remove excess material, leverage optimization algorithms and more.”
Shankar agreed, adding, “since in most companies the number of designers far outnumber the number of dedicated CAE specialists, there is a need to explore strategies whereby more designers are using simulation methods to help in the design task.”
Therefore, if you are a designer that is new to the game and shopping around for a CAE tool, then democratized versions must sound appealing.
These democratized tools typically come in the form of a more sandbox-like simulation in-CAD environment or more targeted options like simulation apps, simulation templates and job-specific simulation tools.
Simulation-in-CAD acts as a familiar environment for designers to work in, whereas apps, templates and job-specific tools are designed for a team of users working on a specific simulation or task.
“Vertical apps and wizards within software can help ensure that designers, when using simulation, can do so reliably and with confidence,” said Mitchell.
The elephant in the room for much of these tools is that by bringing non-experts into the realm of CAE you can bring in a lot of room for error. Unfortunately, the engineers in charge of making these apps, templates and job-specific tools may not catch all the potential input permutations where the simulation can break.
“What is needed is more intelligence and functionality in the analysis to avoid user error and this means more analysis capabilities and flexibility,” noted Cary O’Connor, VP of marketing at IronCAD. “For example, for large scale assembly performance, an automatic part-to-part contact or automatic tied analysis without any user intervention is needed.”
Given the chances of human error, will democratization really increase the number of simulation users? Will the fear of mistakes scare people away? Will It be useful for your team?
Well, contrary to the industry creed, Reinhard Helfrich of INTES notes that the democratization concept isn’t a new one and that it has failed in the past. He remembers a time when it was said that a secretary would one day be doing an engineer’s simulations. To Helfrich, democratization is little more than a buzzword to drum up CAE business.
Lancaster, however, takes the middle ground:
“The goal of democratizing simulations for greater use by non-experts, such as design engineers, is indeed possible, but very difficult to accomplish with the complexity and fragility of legacy finite element methods and codes. Democratization of simulation cannot be safely and reliably attained unless a new framework is used that is based on predictive computational science using numerical simulation and less on the art and craft of finite element modeling, as has been practiced over the past four decades.”
Not convinced in democratization tools or think you won’t need them? Well chances are that even the most seasoned of simulation users will eventually have a question or two about their FEA software.
If you opted out of the democratized simulation then you might need to look for a tool with an active user community that can help your team out in a pinch. These online communities can manifest via forums, YouTube tutorials, Reddit threads and more.
Sometimes getting help directly form the vendor is a pain. In this case you may want to ask the all-seeing eye of Google. It’s possible your question has already been answered in an active community online. You are also likely to find someone who understands your problem. This can be the difference between meeting or missing a tight deadline.
4. What User Friendly Meshing Tools are Available in CAE Software?
Automated meshing around features and curves is made easier with automation. (Image courtesy of ANSYS.)
However, current technology can help reduce much of the trial and error associated with mesh sizing and refinements.
“Automatic mesh refinement help to refine the mesh in regions where there are fine details, small gaps, big curvature and more, to avoid low-quality elements in the first place,” said Mitchell. “Adaptive meshing helps refine the mesh during the run. As elements reach a certain baseline or minimum quality, they distort.”
The automatic meshing tool can update the whole mesh, or focus on a specific region. This way, engineers can keep the computational cost of the simulation down by creating a dense mesh at a point of interest.
“The areas where the mesh is automatically refined can also reveal regions of unexpected high stress that were missed in a first analysis where the mesh was manually created,” explained Sjodin.
Dale Berry, senior technical director at Dassault Systèmes SIMULIA, explains that these advanced technologies allow non-expert users and designers to run simulation and obtain accurate results without having to be a meshing expert.
However, errors can still crop up in an automatically generated mesh.
“Adaptive and automatic meshing take away the burden from the user to figure out what areas often need mesh refinement for reliable results, but it can also chase singularities in a model,” warned Uwe Schramm, CTO of Altair. “So, caution needs to be applied along with engineering judgment to determine if the automatic refinement is desirable or not.”
Although automatic meshing can create a singularity, it can also prevent the creation of a mesh that would take forever to compute and stall the design process. These snail-paced meshes can be caused by large meshes or errors in element definitions.
The good news is that you will be hard pressed to find CAE software that doesn’t have some form of user friendly meshing.
“Auto meshing has been the standard in CAE with automatic geometry feature refinements,” said O’Connor. “Adaptive meshing has been gradually replaced by automatic mesh refinement and is hardly used in design analysis nowadays.”
Additionally, solver and high-performance computing (HPC) technology can lessen the need of a perfect mesh.
“There are certain situations where high deformation needs mesh refinement,” said Schramm. “On the other hand, modern solvers that are programmed for an HPC environment with a fine mesh actually lowers the need for adaptive meshing.”
Unfortunately, not every simulation user will have access to the top of the line solvers or HPC due to their high cost. For these users, automatic meshing is a good option.
An alternative to user friendly meshing tools is using simulation software that relies on p-elements. Instead of mesh refinements, these simulation tools, like Creo Simulate or ESRD, change the degree of freedom of the polynomial that defines each element. As result, engineers can produce an accurate simulation with a coarse mesh.
“P-element technology and auto meshing eliminate the need to understand element types,” argued Coronado. “Among many advantages, it automatically creates the mesh model, accurately captures geometry contours and the mesh is automatically refined during the solving process.”
5. Does the FEA Tool Integrate with Internal and Third Party IoT, PLM, 3D Printing, CAD and More?
Platforms like the 3DEXPERIENCE integrates simulations with many non-CAE tools within the Dassault Systèmes Library. This, and third-party integration, will become more important as the potential for interesting interactions between technologies grows. (Image courtesy of Dassault Systèmes).
“Digitalization is a trend sweeping many industries,” explained Berry. “Simulation practice and tools need to keep up with this trend. Simulation data, methods, practices and results will need to be ‘digitalized’ in the corporate sense. This will require expanded thinking about the role of simulation and tools used to perform that simulation.”
Tools and technologies like the Internet of Things (IoT), digital twin, project lifecycle management (PLM), CAD, augmented/virtual reality (AR/VR), systems engineering and 3D Printing can all have an interesting symbiotic relationship with simulations.
Schramm notes that thanks to the IoT, physical data can be collected by sensors and used by simulations in real-time. “That’s the new thing in the context of simulation with data analytics,” he said. “Not just design, but maintenance decisions will be based not on physical inspection, but from data obtained in operation. A simulation model is combined with physical data offering feedbacks for future design directions as well as current maintenance.”
Lefebvre agreed praising, “the use of big data and machine learning algorithms combined with CAE and the coupling of CAE analysis and in-service data acquisition for predictive maintenance and other in-service operations.”
Divesh Mittal, senior project engineer at Engineering Technology Associates, Inc. (ETA), notes that bringing systems engineering into the analysis world would be a good initiative, advocating that model-based system engineering will allow teams to better manage changes to ensure proper updates and traceability.
Lefebvre agreed, adding, “it is now possible to couple realistic 3D models at the system engineering level, if components or sub-systems are being re-used from one product to the other. This allows for more realistic models to be used at a much earlier stage of the product design.”
Simulation tools should not exist and work in isolation, but rather be the cornerstone of your software portfolio.
6. How much does the Software’s Solvers and User Collaboration Scale with HPC and the Cloud?
When the computation gets tough, turn to the cloud or HPC. (Image courtesy of Rescale.)
In theory, HPC helps engineers solve simulations faster. It also gives them the ability to increase the complexity of a simulation, in terms of a larger mesh or multiphysics, while maintaining a fast turn-around
However, sometimes more computational power doesn’t mean a faster solution. This is because solvers may not scale the same way as others with more HPC. “As models grow larger and solution accuracy is critical, scalability of solutions is important and we continue to place R&D effort to allow our solutions to scale effectively on HPC,” said Shankar.
Therefore, if you plan on using HPC, it’s best to ask vendors how their software will scale with added computational power. This way you maximize the most time at the most affordable HPC or Cloud rate.
Assuming the software does scale well to your added computation power, you might find yourself saving time in the simulation process in ways you didn’t expect.
“One of the less obvious time savings is that less time needs to be spent on cleaning models and simplifying geometry,” said Mitchell. “More computational resources mean bigger models can be run without too much impact on the solve time.”
Unfortunately, not everyone has access to HPC as acquiring the infrastructure can be a very expensive endeavour. This is where the Cloud, particularly pay-as-you-go offerings or software-as-a-service models, come in handy. The Cloud can also supplement the HPC you already have. Many Cloud and HPC server options will also include options for data storage at either a public or private level.
“All the benefits of the Cloud are applicable for CAE solutions, especially the idea of using thin clients and browser-based access, to powerful engineering analysis tools, representing a pool of resources available on-demand,” said Coronado.
But the Cloud isn’t all about computational power and data storage according to Vedantham. He said. “Processing and storage is nice, but in simulation, we can handle most of this with a decent desktop rig. The real benefits come with collaboration, continuous updates, new business models and security.”
Unfortunately, not everyone is sold on the security aspect of the Cloud. “Many organizations are still nervous about utilizing Cloud resources for sensitive data and models,” noted Mitchell.
Schramm, on the other hand, argues that the Cloud “offers accessibility to computational resources, more flexibility and better data security.”
Lefebvre argues that, “if larger size companies are still being cautious with public Cloud usage because they fear cybersecurity threats then private Cloud solutions might bring in a lot of flexibility to their simulation needs.”
Schramm predicts that “stand alone and Cloud resources will blend and mix in the future and there will be virtually no distinction between these two. The Cloud is a logical development of interconnection between computing resources.”
One thing to keep in mind when using the Cloud is the time it takes transferring data between big data files. Mitchell argues that, “large files present challenges in downloading models. After all, a 5GB movie is very different than a 5TB model file.”
To that point, Helfrich notes that “Using cloud services only makes sense if all product data from design, to CAD, to FEA and other simulations are stored on the Cloud. These files should be used without any explicit data transfer to a local computer.”
At the end of the day, most experts agree that the Cloud has a bright future in the CAE space.
“It enables engineering groups without a dedicated IT infrastructure, HPC’s and additional hardware limitations to compete with any other engineering firm,” said Lancaster. “It levels the playing field in a logistical view. Also, software evaluations will benefit greatly from the Cloud because no license or installation may be needed to try the software.”
Srivastav believes not using the Cloud leaves users dependent on limited RAM and storage issues.
“Users must download and install every software update themselves,” Srivastav explained. “This is prone to mistakes and the user has to buy and update hardware as projects grow or as the hardware becomes obsolete. Also, the user is bound to one physical location or machine to perform the CAE. All this can be circumvented by using the Cloud for CAE as a standard.”
When looking for Cloud and CAE options, Bjorn Sjodin, VP product management at COMSOL, suggests that users consider looking at the software license agreement to ensure that the software can be used in cloud computing and various computer nodes.
7. Have Tools for Optimization, Design Space Exploration and/or Generative Design?
Topology optimization gives engineers a starting point based on the best possible shape for a part, based on load paths and packaging space. With 3D printing, many of these parts are now feasible to manufacture.
Inspire Unlimited topology optimization (Image courtesy of solidThinking.)
“All methods of optimization including sampling, sizing, topology and shape optimization, as well as reliability analysis and optimization for a robust design, will become more important than today,” said Helfrich. “Not to make an optimized product will become clearly disadvantageous for the manufacturer.”
The major enabler to DoE and parameter based optimizations is the proliferation of HPC. However, engineers need to pay attention to more than just their computational power when using an optimization tool. They must ensure that each iteration of the design is based on sound engineering.
“If there is control of the numerical error in the optimization process, meaning each solution has proven convergence in the data of interest before moving to the next optimization step, there is a tremendous value in these types of tools,” said Lancaster. “The decisions made by these tools are still controlled by the design requirements of the engineers. The optimization software cannot do its job without a clear set of goals and minimum criteria.”
For generative design, however, the biggest challenge facing its proliferation was the feasibility of manufacturing the organic looking products the tools created. Fortunately, with the creation of 3D printing technology many, though not all, of these designs can now become manifest.
There are two typical use cases for topology optimization tools; the first is to create a new design from scratch using only a 3D space, boundary conditions and loads. The second is to take an existing design that is underperforming or in need of light-weighting and redesign it for 3D printing.
“Topology and shape optimization are very important and helpful in identifying missing load path,” said Mittal. “Hence it is helpful in finding weight reductions and performance improvements.”
For topology optimization, much of the generative design is based on structural physics and loads.
Schramm notes that as these tools become more popular, supply and demand will likely see more physics and manufacturing constraints guiding the production of user’s designs. This will inevitably create better starting points for engineers to draw inspiration from and lead to better overall designs.
Mitchell explains that, “engineers and designers working with more traditional manufacturing methods can use the tool to develop designs that are suitable for casting, injection molding and even traditional subtractive processes. Topology Optimization will readily produce designs that can include manufacturing constraints that allow them to be molded or cast and the results can be used simply for guidance if parts are to be machined.”
When interested in optimization and generative design software, engineers should look at how these technologies will interreact with their CAD packages. “The main difficulty remaining is the optimization tool’s interoperability with the CAD definition,” said Lefebvre. “It’s about the amount of rework to create the new CAD definition.”
Coronado suggests that it is time that vendors change the output of their topology optimization software from a triangulated object, produced outside the CAD environment, to feature-based, associative parametric 3D geometry. Vendors that plan to make – or have already made – this shift will have an advantage.
Nonetheless, optimization and generative design tools will certainly give your team a leg up on that next bid. After all who doesn’t want to make a better part with less material?
To find some answers to these simulation software capacities, keep reading with us and check out our Design Software e-Book. Why not compare structural FEA software you might be interested in another e-Book: Comparison Charts for Structural FEA Simulation Software.