[lammps-users] simulating the pull-out of fiber from matrix with LAMMPS

Dear everyone,

I am planning to simulate the pull-out of the fiber from matrix with MD. Honest to say, I have no any idea of MD and LAMMPS as a beginner. Although I have read some papers about them, it is still difficult for me to detect whether it is proper to solve this problem. Would you like to give me some suggestions?

Thanks in advance.


MD or Molecular Dynamics (or Molecular Statics/Molecular Mechanics in the static case, there are a few names for it) is generally applied to problems where one is interested in the evolution of the atomic structure of a solid, liquid or gas due to some boundary and initial conditions. The fundamental idea is using Newton’s second law as the equations of motion for an ensemble of particles (points, actually) which represent the position of the center of idealized atomic nuclei and whose interactions (i.e. interatomic forces) are governed by some sort of potential energy function (also called simply a ‘potential’). It of course can get more complicated than that, however.

One of the major drawbacks to MD is that it is generally restricted to a length and time scale which is far smaller than that which is typically of interest to those interested in continuum-level behavior. In other words, unless the system you are interested in is on the order of a few nanometers or less and you are interested in studying it for less than a few nanoseconds… MD is probably not what you are looking for. And this is before getting into the hassle of finding a suitable potential for your system.

That said, it sounds to me from what you are talking about that you probably want to use some sort of continuum method to model the presumably polymer matrix and fiber, with a semi-emperical force/displacement relation to simulate the interface between the two (i.e. interaction forces). Depending on the particulars, MD may be of use in determining this relation.

Hope that helps,


Thanks very much for the words. It really helps me.
Based on the review of present documents, MD maybe the proper method to investigate the mechanical behavior of nanocomposites. As you said, the limit to time-scale and length-scale is the challenge for MD. That is why there are so many researchers shedding light on the multi-scale theory.
So,in order to observe the procedure of the pull-out of the nanofiller, maybe MD is one of the proper way. However,there are so many softwares on it, LAMMPS,AMBER, and etc. Actually, it is a bit difficult to compare them and select the one which is most suitable for your work.
When you are using LAMMPS to solve the problem, which part is the most difficult for you?

Thanks very much for the words. It really helps me.

Yes, there are a fair number of MD packages, each with their own benefits. For me, when choosing a package, I look for several things:

is the code currently maintained by some party, is there some sort of basic help available
Is the code written, distributed in such a manner that I can add functionality if need be
does the code have good parallel capabilities, is it portable to the machines I need to use

does it have the capacity to handle the geometry, loading I am interested in
does it have the potential(s) I need to use

The first one is actually usually the major deciding factor, as there are a number of packages mentioned in the literature that have more or less died or were never available to the public. The last 3 are also important - though the answer to these may vary depending on the problem being studied. I have used 2 packages in the last few years, Tahoe and LAMMPS - and while Tahoe has some nice features for implementing multiscale methods, it’s atomistics were a bit behind the continuum methods (particularly in parallelization, though that may have changed). LAMMPS on the other hand, I have found to be very capable for the problem I am currently working on since it relies entirely on large scale atomistic calculations.

Now, as to if MD (or empirical potential atomistics in general) will be the way to solve your problem - that will be something you will have to determine. It all depends on what you are interested in finding out.

Anyway, hope that helps,