regarding MSST

Dear Users,

Recently I tried to do some shock simulation of metals through EMD by applying fix msst and nphug. As per my understanding through literature and documents , the results of msst should show same behavior which I get from nemd shock simulation (not really as the gradient of the TD variables in NEMD ). I have read the documentation and also the previous discussion minutely. Initially I choose Cu perfect crystal as I have the known results from nemd and also lit. So I guess at Us = 6 km/s (assuming Up = 2.8 km/s ), it should compress significantly as I am setting shock direction as [100] which is most sensitive for Cu. But during MSST, there is not significant compression, as per Aiden’s suggestion in this regard I was trying to choose “q” the mass like parameter, the range I have tested 10-50 with almost all values in the range and as per the tutorial is concerned I also was set “tscale” smaller non-zero, the range I used .01, .05,.001… .0001…and many but no compression is there. My sample was well equlibriated and I have not set p0,v0 and e0 it was calculated in the first step. my fix was " fix 2 all msst x 60.0 q 20 tscale .001 "

I know there is no hard and fast rule to choose for selecting “q” and “tscale”, at least what Steve and Aiden said before but I really wonder as i have tested a long range of values but no significant compression. Is there any other parameters which may cause this problem ? And also can you suggest me a range at least which value is smaller for “q” and “tscale” (where default values are 10 and 0.01) from your experience ?

Thanks for your time.

Rgds,

Rajom

The manual suggests to use the “smallest value of tscale that results in compression”, so I’d suggest you increase the value of tscale instead of decreasing it. Decreasing the mass-like parameter q would help as well. To make sure you have everything set correctly, I’d suggest you start with a stronger shock, e.g., 10 km/s.

Ray

Hi Ray and all,

I have trouble suited as you suggested and I got the values of “q” and “tscale”, for which the steady state pressure and temperature is now comparable with NEMD results. But initially the P value is varying drastically but after a long time simulation it attains the steady pressure at 320 Gpa and T = 8000 K which is comparable with experimental results for Cu(100). My question is, is this the right way to be sure that the parameter I have set is correct ?

Another question is that as the volume is compressing and increasing periodically in this simulation, how can I get the hugoniot curve P Vs (V/V0) ? How can I plot the values “dhugoniot” and “dreleigh” i.e to see the hugoniot curve shift from the actual hugoniot of Cu, which should be plot against the “dhugoniot” and “dreleigh” ?

The computed "lagrangian_speed " is the particle velocity as described by docs, so should it be reasonable to take the average value of this computed values as particle velocity when I am defining Us at a particular steady value ? I saw that the average value of this v_lgr_vel differs from the literature value. i was tried to calculate the Up from RH jump condition i.e from after and before shock density profile, but problem is that as the simulation domain is expanded and compressed during run so naturally density is also fluctuating drastically how can I take the after shock density ?

I have read many paper including Ray’s publication JPC, regarding MSST application in shock simulation and previous discussion of MSST in this forum. I would be obliged and it would be a big help if you share your experience and suggest me something regarding this.

Thank you

Regards

Ravi Rajom

Hi Ravi,

Replies in-line, thanks.

Ray

Hi Ray and all,

I have trouble suited as you suggested and I got the values of “q” and “tscale”, for which the steady state pressure and temperature is now comparable with NEMD results. But initially the P value is varying drastically but after a long time simulation it attains the steady pressure at 320 Gpa and T = 8000 K which is comparable with experimental results for Cu(100). My question is, is this the right way to be sure that the parameter I have set is correct ?

If the parameters are reasonable and the results are convincing, why should not it be a reasonable way? By the way, the pressure seems very high…

Another question is that as the volume is compressing and increasing periodically in this simulation, how can I get the hugoniot curve P Vs (V/V0) ? How can I plot the values “dhugoniot” and “dreleigh” i.e to see the hugoniot curve shift from the actual hugoniot of Cu, which should be plot against the “dhugoniot” and “dreleigh” ?

For each U_S you specify, you obtain one point on the Hugoniot. To get a Hugoniot curve, you would need a series os shocks with different U_S. Departures from Hugoniot and Rayleigh conditions should be as small as possible in all steady-state shocks.

The computed "lagrangian_speed " is the particle velocity as described by docs, so should it be reasonable to take the average value of this computed values as particle velocity when I am defining Us at a particular steady value ? I saw that the average value of this v_lgr_vel differs from the literature value.

Yes, but only the portion that exhibits minimal fluctuation should be taken to average.

i was tried to calculate the Up from RH jump condition i.e from after and before shock density profile, but problem is that as the simulation domain is expanded and compressed during run so naturally density is also fluctuating drastically how can I take the after shock density ?

Again, only one density is obtained from one shock.

Dear all,
I have done a series of MSST simulation, to be sure about the results. I have some questions regarding MSST and as well as regarding Ray’s and Thomptons’s JPC (2012); so i’m cc’ed to lammps forum.

  1. I have checked almost all possible values of “q” and “tscale” and also optional “mu” of msst; q = (2-10),tscale = (.1-.001); mu = (1e-03 and 1e-04). Comparing the results, it seems to me that the attached results (please find the plots in attachments ) are better than others. So from the results, I see that almost 4-5 ps time have required to reduce the volume and pressure fluctuations, after that it’s showing steady-state pressure and temperature almost comparable with exp results. Is this a usual experience of MSST, that need some significant time to equlibriate the shock hugoniot and Releigh line ?
    From your results in JPC in case of pent, it’s showing that it does’t take so much time to equlibriate, can you please help me to understand how can I fasted this equilibrium process of shock-hugoniot ?

  2. from docs I saw that in case of special high symmetry condition this fluctuation have physical significance. Could you be so pleased to give me a clue to understand this lines ?

  3. During this long and drastic fluctuation period, my system geometry (when I was trying to do that msst for poly-crystalline and pre-defined defects in Cu) was completely collapsed before reaching the steady state shock condition, so as a consequence the results are affected severely. So this will be overcome if the fluctuation can be reduced. Can you please suggest something which is the best way to reduce this time of fluctions ?

  4. From your paper I have attached one P vs specific volume hugoniot: my question is, is that curve was taken from one simulation i mean one shock velocity ?
    I think this was taken from one shock velocity simulation because P-V hugoniot usually for one particular shock speed.
    If so how can you calculate V/V0, when time-dependence of volume was fluctuating and after reach steady condition fluctuation is nearly negligible?
    and also where from you got the steadily increasing pressure for this curve ?

surely I’m missing something fundamental please make your expertise comment.

Please bear me out for such lengthy letter. Thank you very much for your attention.

Cu_msst_dHug_time.jpg

cu_msst_vol_time.jpg

msst_pent_ray.png

sorry the curve from Ray’s paper was not that which I attached before, here it is plz find that.

Rgds,

Rajom

P_specific_volume_msst_ray_for_pent.jpg