Quadrupole Mass Spectrometer
Images made using COMSOL Multiphysics® and are

provided courtesy of COMSOL®

Biased Resonator

​​​Electromagnetics and Multiphysics
​Modeling and Simulation

Establishing Simulation and Modeling Capabilities

Establishing in-house electromagnetic modeling and simulation capabilities 

requires careful planning and proper execution. Scroll down to review some consideration that can help.​​​

Define your modeling needs and objectives 

Define the types of problems that must be modeled

Determine, if possible, the most time-consuming modeling tasks

Define budgets

Identify available or required staffing resources

​Research computational electromagnetics (CEM) methods

Select the right CEM software

Evaluate best-suited software

Understand license types and prices

Select available and required computer hardware

Investigate training options

Imitation is the sincerest form of flattery

​​It has come to our attention that the contents of the NablaWorks site have been copied almost completely in their entirety, and in violations of copyright protection, by a company claiming to be located in China. Please, do not confuse other sites with NablaWorks. Enjoy your visit to the Original, and we hope that you find the information contained here to be useful and relevant to your endeavors. 

​

Define your modeling needs and objectives (back to the top)​​

​Defining your needs and objectives will greatly assist in the software evaluation process.

• ​Identify the types of problems that need to be modeled
  
• Estimate model sizes, both in terms of the number of points or elements and in terms of degrees of freedom (DOFs)
  
• Determine the frequency of occurrence when modeling is needed (occasional vs. ongoing)
  
• Define requirements for optimization problems and parametric studies
  
• Determine typical deadline requirements
  
• Determine complexity of typical geometries and how they will be created them​
  

Define the types of problems that must be modeled (back to the top)

Understanding the types of problems that must be modeled is critical in the software evaluation process.

Every software package has its strengths and weaknesses. Many codes are specialized or tailored for specific applications while others are very general and suited to a wide variety of problems. Here are some important considerations to take into account,

• Static, quasistatic, transient, harmonic (low-frequency or high-frequency)
  
• Bounded or unbounded (open boundaries)
  
• Just EM or EM coupled with other types of problems (mechanics, heat transfer, fluid dynamics, etc.)
  
• Types of materials involved
  
  •   Nonlinear
  •   Inhomogeneous
    
  •   Anisotropic
    
  •   Dispersive
    
  •   Lossy
    
  •   With dependence on past environment (hysteresis)
    
• Geometry type: 2D, 2D-axisymmetric, 3D​
  

​​​Determine, if possible, the most time-consuming modeling tasks (back to the top)

Determining the most time-consuming modeling tasks will help you set proper priorities for various software features, hardware choices, and license costs. Quite frequently users underestimate the amount of time it takes to create and validate a model. Execution time, even if it takes many hours, can be easily eclipsed by many days or even weeks of geometry creation, model creation, and model validation. On the other hand, sometimes a single, relatively simple model can be followed by weeks of parametric studies, so computer execution times become very critical. The three main tasks in modeling and simulation are:

• ​Creating models and getting them to run properly
  
• Running model(s)
  
• Processing results
  

​

​Define budgets (back to the top)

Depending on your requirements, in-house modeling can demand significant capital investment in hardware and software along with ongoing need for staff funding and software license maintenance.

​Identify available or required staffing resources (back to the top)

​Availability of qualified staff is imperative to the success of in-house modeling capabilities. Understanding of both electromagnetics and numerical techniques is necessary, and the learning curve for software usage can be steep and lengthy. If new staff must be hired, it can take a significant amount of time to recruit qualified individuals.

Research computational electromagnetics (CEM) methods (back to the top)

Take a little time to research and understand CEM methods. Researching the types of available methods can make it easier to understand the capabilities and the limitations of electromagnetic simulation software.​

Select the right CEM software (back to the top)

Appropriate choice of CEM software will make your modeling more efficient and your simulations more accurate. It will also keep you from spending unnecessarily on features and capabilities you may not need.

​​Evaluate best-suited software (back to the top)

• Evaluate more than one software package
  • Narrow down the search to two or three. Thorough investigation of the software is time consuming and evaluating more than two vendors may not be practical.​
• Arrange for a software demonstration and for an evaluation copy of the software.
  • Most vendors will allow you to try their software over a period of weeks or months. If you need more time, ask for it.​
• Try out major software features
  • Geometry creation
    • Built-in geometry creation
    • Ability to link to relevant CAD software (Pro/E, SolidWorks, Inventor, etc.)
    • Ability to import major CAD software formats
  • Mesh generation and control (automatic generation, manual controls)
  • Physical problem definition, including boundary conditions
  • Support for material properties database
  • Control and flexibility of numerical methods
  • Scripting and automation support
  • Post-processing capabilities
• Try out software functionality as it relates to important aspects for your problems. Here are some examples of problems that can lead to challenges during simulations,​
  • Unbounded space
  • Skin effect
  • Multiscale
  • Electrically-large simulation space (simulation space much larger than a wavelength)
  • Nonlinearity
  • Coupled physics
• Create and verify validation models
  • Ask the vendor for availability of validation models that it already has
  • Create your own simple models with known analytical solutions
• Evaluate documentation
• Try out technical support
  • Availibility
  • Knowledge
    • Asking questions to which you already know the answers
  • Types of communications (e-mail, chat, phone, ability to view your model via remote connection)
  • Response time
  • Time needed to resolve a problem
  • User forums and groups​

Understand license types and prices (back to the top)

Licensing practices vary widely. If you want to avoid expensive surprises after you make a commitment to a particular vendor, make sure you thoroughly understand the vendor's licensing options. Here is a list of common licensing practices:

• Single-CPU licenses
  
• Single-user licenses
  
• Floating (network) licenses
  
• Multicore CPU licenses
  
• Multi-CPU licenses
  
• Cluster licenses
  
• Distributed computing licenses
  
• Accelerator board licenses
  
• Dongles and license files
  
• Perpetual licenses
  
• Leased licenses
  
• Annual maintenance fees

License management issues can be important as implementation vary. Customer service considerations are also important, for example, the vendor's ability and willingness to transfer license between machines.

​Select available and required computer hardware (back to the top)

Here are some factors to consider when selecting hardware:

• ​Software compatibility
  • Software vendors typically test their products on a few (sometimes very few) hardware platforms. You have to make sure your hardware is supported.
• CPU performance
  • Keep in mind that greater CPU performance does not always result in increased execution time. For example, if your models are limited by storage system performance (hard disk or solid state), increasing CPU performance will not help in improving execution time.
• Memory
  • Make sure the software can make use of all available memory. During software evaluation it is very important to run very large models in order to understand the limitations of the software and hardware. Software stability problems can also be discovered when memory limits are tested.
• Storage configurations
  • Consider RAID with stripping because it will improve storage system performance
• Support for parallel processing (multicore, multi-CPU, distributed and GPU-accelerated)
  • Consider all possible options for accelerating your hardware, but beware -- execution time usually does not scale linearly with multiple cores, CPUs, nodes or GPU boards. Make sure to ask the vendor for benchmark tests.

Investigate training options (back to the top)

Training can be very important in mitigating steep learning curves that are associated with establishing CEM modeling capabilities. Many companies offer such training options as instructor-lead courses (in-person or online), free workshops, online video courses, self-paced training exercises and modeling examples. For instructor-led courses it is worth while learning about course availability, frequency, duration, variety (short and long, introductory and in-depth, application-related, etc.), and, of course, cost.