I would like to ask You - material science experts - I think you might know the answer. I would like to generate / simulate porous material microstructure, something similar to polyurethane or other polymer foam (open and/or closed foams). A prerequisuite is my knowing or being able to determine the size and boundaries of each ,pore,. I found powerful tool LAMMPS but I think the entry level is too high for me, unfortunately. I know that there is lots of tutorials but I havent found similar cases or I am unable to understand that.
Do you know other tools for such task ? Maybe something more beginner-friendly? Or maybe you have some example LAMMPS scripts for foam material-like microstructures? I am familiar with Python.
@Dwal I suspect that you are making the same mistakes, that many beginners make: you are moving too fast.
MD simulations is not something that you can do instantly, and if you want accurate results, the is no escape from spending a significant amount of time learning how to correctly use an MD software package and do simple text-book example simulations first and learn how to analyze those.
Moreover, you should not do this on your own. There is far too much to learn (like statistical mechanics and thermodynamics) and a lot of practice needed, until you can trust your skills sufficiently. For that you need a proper tutor and supervisor that can review you training and guide you. No software, no internet forum, no chatbot can do that for you.
Only after you have developed sufficient skills, you should consider starting with your simulation project and that usually requires a proper survey of the relevant published literature, to learn from others that have succeeded in the kind of simulation studies that you want to do. But to understand them properly, you need the basic skills I was talking about.
Sorry, or to put it differently there ain’t no escape from the blues!
Hi @akohlmey , I think you are absoletuly right. But I can’t find an experienced tutor locally to be honest who could devote his time. What is more I need to verify if MD simulations could generate structures similar to real polymer foam microstructures. If yes I will try to find a good tutor and deeply get familiar with simulations. If the satisfatory results are not available using MD methods I will try to find another method. For example my initial idea for 2D microstructure like image was to use simple voronoi tesselation but they definetely do not capture the complexity of the microstructure of porous materials.
You have to understand the dynamics of an online forum. This is not a place where the people with expertise in all kinds of research hang out and then can solve all your problems. Rather, you have some enthusiasts that enjoy working with the LAMMPS software so much, that they want to help others. So your question is completely misguided (and the forum guidelines already explain that), since it is not about LAMMPS, but about your research in general.
The advice for your kind of issue is always the same: thoroughly study the published literature and figure out what others have done to simulate the same kinds or similar materials and what kind of results they got. Only that will inform you about what is a viable method to perform your research.
Starting a new research project without competent local training and supervision is a doomed project.
I’ve generated porous structures before, but the first thing that comes to mind is the prefix “micro”, settle on a scale that you want to study and what it is that you want to learn from the simulation itself. Then ask yourself what kind of model is appropriate for your system and if it can capture the physics you’re interested in. Put another way, do you care about the polymer’s behavior at the chain level?
To generate a porous microstructure, you can use a tool like PoresPy that is capable of generating various microstructures of interest with user-specified features like porosity, different levels of noise added, etc. Then, you can take that 3D array and run an algorithm called Marching-Cubes (available in scikit image) to generate a .stl file. That .stl file is basically just a list of vertices of triangles which can be read directly into LAMMPS with the create_atoms command with the “mesh” keyword. That will place atoms at points on the faces of those triangles with a density that you prescribe.
From a practical standpoint, will you have the computing power to complete the simulation that you want to do in any reasonable amount of time? These microstructures are inherently statistical in nature, you’ll probably need to run this simulation many times with different microstructures with the same statistics to draw any meaningful conclusions from your modeling.
This seems like a pretty ambitious problem to tackle for someone brand new to LAMMPS, but I can definitely sympathize with not having the specific hands-on training when starting out.
Note that this forum is generally staffed by people (volunteering!) who have experience using LAMMPS, but not necessarily using LAMMPS for polymer simulations (or any other specific topic, other than the ones we happen to do professional research in). As such, you will not be able to get a lot of specific information from us about how to simulate polymers.
However, if you have tried to perform some task XYZ that you expect to give results ABC but instead you are seeing results DEF, there is at least some chance that the specific issue will be one that some of us have seen before. Such queries are usually your best bet for getting help.
To satisfy my curiosity, I also did a bit of a search and it looks to me that CFD is more commonly used to model foams. There are even several CFD programs that have FOAM in their name.