When I realized that I wanted to go back to school, I stopped to think about what I would need to do. It had been years since I cracked open a textbook, much less learned something new on a computer. However, I wanted to succeed, so I started to learn more about technology. I spent hours brushing up on my computer troubleshooting skills and when I was done, I felt a little more prepared. After I started school, I learned even more. I made this blog to teach other people a little about technology, so that you can keep up with your peers.
If you want to conduct finite element analysis, there's a good chance you'll need to contract with a third party for FEA services. When clients and FEA consulting companies are setting up a project, two of the biggest factors you need to consider are scale and processing power. Folks preparing for projects need to understand how scale and processing power influence the following issues:
Time, Performance, and Money
The main problem you'll see as you scale up and eat processing power on a project is how everything will take more time. FEA services center on computers that can model how different materials and environmental elements interact. The net effect is that adding anything to the model requires more processing power from the computers. That will slow down its performance, although it won't affect the quality of the analysis. Also, adding machines means more modeling and more electricity, and that usually means higher costs.
Going Small or Big
There are two common ways a project scales. First, you might want to get more granular with the analysis. Second, you might need to look at something very big.
A granular analysis is computationally intensive because it increases the number of elements in the analysis. If you're studying the performance of an alloy, for example, going more granular will give you a better sense of how the materials might develop microfractures. You're looking at adding more and smaller elements to the model, and that adds computation to the project.
Scaling up also creates more elements. If you add five structures to a wind-sheer analysis of urban development, for example, you have to add enough elements to meaningfully model those structures.
Can You Reduce the Element Count?
Like so many other questions in life, the answer is: "It depends." There is an argument for not analyzing anything at a level more granular than you need. However, you don't want to avoid doing a detailed analysis just to save a quick buck. That may cost you in the long run if a component fails. An FEA consulting firm can help you select the target based on experience.
You'll also want to think about how many simulations are necessary. Do you need a million runs? If you're looking at performance under conditions that are well within your control, you might be able to trim that down a few thousand. Conversely, if you're analyzing performance under many different conditions, you may need to add simulation runs to capture all of the possible variants and outcomes.
Share