high strain rate effect

As I know, to reduce the simulation time of a simple tensile test by MD, high strain rate loading used with npt or nvt ensemble. but there is a question to me about the effect of high strain rate in MD simulation on result. I have tried to understand this and search more in internet about it but I couldnt find appropreate answer. I have done a simple tensile test simulation at strain rate of 1e9 and 1e10 (1/s) and there was no significant difference at results. it will be very helpfull to me if someone describe the effect of high strain rate in MD simulation.

The issue is whether the strain rate is slow enough to allow physically-relevant deformation processes to occur. If the deformation is purely elastic, then very little time is need (100 fs), because the displacement field is completely uniform, and is sampled entirely by local vibrations of the atoms. However, if the deformation involves some kind of plasticity, this will usually take a lot longer, say 10 ps. If the total strain magnitude is 1%, then this corresponds to a deformation rate of 0.01/10ps = 1e9 1/s.


Dear Aidan.

Thanks for your reply.

As I understand from your description:

1- The elastic deformation need 100 fs and plastic deformation need about 10 ps to occur, and also the MD simulations with plastic deformation at strain rate lower than 1e9(1/s) and any total strain such as 10% should show same result, due to the plastic deformation had enough time to occur at these strain rates?

2- Is ensemble (such as NPT and NVT) have a role in neglecting of the strain rate effect at MD simulations? for example, keeping constant temperature during the simulation.

3- The 1e9 (1/s) is very high strain rate in experimental and finite element simulations that the effect of the strain rate is evident on results. What is the difference of strain rate definition at MD and experimental test or finite element simulations?


  1. The numbers I gave are just examples. In particular, the timescale for plastic deformation can range over many orders of magnitude and is different for different materials, different temperatures, different loading modes, etc. You will have to do your own testing to figure out what is appropriate for your research project.

  2. The Nose-Hoover barostat and thermostat should not play a role, unless they malfunction. This is particularly true for NPT. It is very important to monitor the simulation (especially at the beginning) to ensure that the system is not undergoing large unphysical changes in temperature, pressure, volume, etc.

  3. The big difference between engineering tests and MD is scale. A 10 nm cross-section nanocrystal of copper atoms has very little in common with a 1/2" copper pin.

For more information, you will need to start studying the scientific literature on plasticity.