How not to Give Up when Running a Large-Scale Simulation
We are living in an exciting time. Simulations have become a standard tool in almost every product development process, substantially speeding it up when compared to the traditional approach of physical testing.
Modern large-scale simulation models are, however, often very complex systems that require extensive knowledge, resources, and effort. So, how to make the most out of the simulation capabilities and minimize the trial and error pain – especially when dealing with a large scale simulation?
In this article, I will go through the most important best practices that will help you set up your simulation like a pro.
Do your research
Before starting with any setup, take some time to understand the problem at hand. Gather as much information as you can about the physical phenomenon that you are planning to simulate. At this stage, you don’t need to look for anything specific, but rather make sure you have a clear picture of the model you are using, its capabilities, and limitations.
Furthermore, try to find out if others have worked on similar problems in the past and identify the critical issues they had to deal with. This knowledge will automatically translate into new ideas and solutions while your simulation becomes more and more complex. Organize your findings and review them on a regular basis as you move forward.
It does not matter how complex the final simulation is going to be; you should start where you feel comfortable. This is generally a good practice for any type of simulation, but a MUST for any large-scale simulation. In a system with several complex components, it is virtually impossible to “get it” the first time. You could see it as building your simulation layer by layer. For example, start with a simplified geometry if you are not sure how capable your meshing tool is. Do enough iterations to get a good grip of it. Once you are confident, move on to the next step. Even though this might seem time-consuming, compared to the amount of time and resources you would have to invest in finding the problems within a large complicated project, your little extra effort is definitely well worth it.
Monitor your progress closely
As you make progress with your simulation, you will eventually have to make decisions that will greatly affect the future direction of your project: which model parameters should I choose? How should I modify the boundary conditions to get the results I want? How much computational resources I need to conclude the project? etc. Modeling tools will normally give you a lot of useful information in the course of a simulation run. If you monitor the outcome of your runs closely, you will be able to answer all these questions easily and accurately. This will help you identify potential issues in your setup, avoid trial-error approach while solving them, and make informed decisions about your project as a whole.
Manage your resources for the large-scale simulation
Performing a large-scale simulation means a great investment in terms of effort and money. Computational resources are limited and costly, there are normally several time constraints, and most importantly, your time as a simulation expert is extremely valuable. Therefore, avoid unnecessary overhead by managing your resources. Make sure to:
- Keep your simulation projects well-organized: you are typically dealing with hundreds of different setups in your simulation. Organize and document them in a way that is understandable when you come back to them in future.
- Keep your simulations portable: cloud computing is everything! There are several extremely useful solutions on the market that make your life easier (such as SimScale). Also, if your simulation is too large, try dividing it into well-defined independent components. This makes it easy to move your project around, make backups, and present it to others.
- Seek help regularly: there is tons of resources out there that could help with the problem you are working on. Find your solutions and add it to your project library. You will come back to these again and again in the future.
Do NOT get discouraged if you are at the beginning of the road or if you keep running into issues with your large-scale simulation. Follow these tips and you’ll be on your way to becoming a simulation expert. Happy simulating!
Do you have any additional advice for running a large-scale simulation? I’d love to hear in the comments.