How do you cope with the computational and data demands of FEM in civil engineering design?

1 Choose the right software

When selecting the right software to cope with the computational and data demands of FEM, there are many factors to consider. These include the type and complexity of the problem being solved, the accuracy and reliability of the desired results, the compatibility and integration with other software and data sources, the cost and availability of the license and support, as well as the user-friendliness and learning curve of the interface and documentation. It may also be beneficial to compare different software options by testing them on a sample problem or a benchmark case to evaluate their performance and suitability.

2 Simplify the model

To cope with the computational and data demands of FEM, simplifying the model is essential to ensure accuracy and validity of the results. This involves reducing the number and size of elements, degrees of freedom, boundary conditions, and material properties. Techniques such as using symmetry or periodicity to reduce the domain size, coarse elements in regions where the solution is not sensitive or critical, adaptive mesh refinement to concentrate elements where the solution is highly variable or complex, linear or simplified material models instead of nonlinear or advanced ones, and substructuring or superposition to divide the problem into smaller or simpler parts can help simplify the model. Doing so can save time, memory, disk space, as well as improve the stability and convergence of the solution.

3 Parallelize the computation

Parallelizing the computation of FEM is a third way to cope with the computational and data demands. This process distributes the workload among multiple processors or cores, which can speed up the solution process and enable larger or more complex problems to be handled. When parallelizing, you should consider the type and architecture of the hardware you are using, such as CPU, GPU, or cloud; the scalability and efficiency of the software you are using, such as its parallel algorithms and libraries; the partitioning and communication of the data among the processors; and the monitoring and debugging of the parallel process. It is also important to adjust some parameters or settings in your software or hardware to optimize the parallel performance and avoid bottlenecks or conflicts.

4 Manage the data

To cope with the computational and data demands of FEM, managing the data is essential. This means organizing, storing, transferring, and analyzing the data generated by the FEM process. Doing so can help avoid data loss, corruption, or duplication and enhance your understanding and interpretation of the results. To manage the data effectively, use consistent and meaningful naming conventions and formats for your files and folders. Compression and encryption techniques can reduce the size and protect the security of your data. Backup and recovery methods should be employed to safeguard your data from accidental deletion or damage. Visualization and post-processing tools can be used to display and explore your data in different ways. Additionally, statistical and numerical methods can be used to quantify and compare your data with other sources or criteria. Ultimately, managing the data can help you improve your productivity, quality, and confidence in your FEM design.

5 Here’s what else to consider

This is a space to share examples, stories, or insights that don’t fit into any of the previous sections. What else would you like to add?