Density of Sample, particle shape and Contact Model

Hi Mohammad,

1. the density is not externally prescribed, so whatever the packing geometrically allows, that will be the density. The friction angle is set initially to zero to reach the highest density possible. Tuning the PSD will have an influence on porosity as well. As far as I remember, there was no direct control of posirosity in Yade either; radius expansion is just an inverted process of boundary compression (minus inertia). Can elaborate on your idea how to control the porosity?

2. &3. Different particle shape and different contact law are currently not possible. If you look at the TriaxTest example (the one without membrane), it allows you to specify different particle shapes and contact laws (by assigning generator and model). I will, if you want, modify the CylTriax preprocessor in the same way. Then you will be able to use capsules or spheres or clumps. There is no way to do it in a backwards-compatible manner unfortunately, so you will have to modify your older saved preprocessors to reflect the change. Would you be interested?

(Ellipsoids will still not work because the routine to collide facet with ellipsoid is missing; I have not yet found a way to do it efficiently, basically I need an efficient algorithm to compute the distance between an ellipsoid and cylinder).

Cheers, Václav

Hi Vaclav,
Thank you very much for your answer.

Density:
In sample preparation stage of Toyoura sand in conventional triaxial tests, just before applying shear relative density , I have followed the following step to get relative density

1)minimum index density (row_min)=m_s1/v_c
 here, m_s1=mass of mold with soil placed loose-mass of mold
        v_c= calibrated volume of the mold

2)maximum index density(row_max)=m_s2/v
 here, m_s2=mass of mold with soil after vibration loose-mass of mold
        v_c= volume of tested dry soil = v_c-A_c*H
3) e_min=((row_w*G_s)/row_min)-1

  e_max=((row_w*G_s)/row_max)-1

4) relative density D_r=(e_max-e)/(e_max-e_min)

In DEM,before radius expansion, porisity is minimum. Adopting radius expansion, I got the target porosity. To get relative density, how other quantity can be achieved?

Shape and Contact Law:
I have started to work this issue. If I need any help let you know.

Thanks in advance for your support.

Regards,
Mohammad Nurul Islam

Hi Mihammad,

thanks for the explanation of relative density. It looks like the same as https://en.wikipedia.org/wiki/Soil_mechanics#Relative_density? Now, controlling something like this is difficult with DEM, as long as you maintain spherical particle shape. Some people decrease (to zero) friction angle during compaction so that the sample is as dense as possible. So you could perhaps keep friciton angle higher during compaction to partially prevent rearranging of spheres, but it is not easily controllable what the result will be. Radius expansion is like compression, it will not change relative density significantly (unless you allow big overlaps of particles; but the same could be done with regular compression).

WIth DEM, it is tempting to think of spheres as grains of soil, and then calibrate properties like porosity or relative density to what you see in reality; but I like more to see it as a model which looks similar, but has different mechanisms involved, and the microscopic parameters don't match 1:1 to what you can physically observe or measure with grains. I am sure there is literature on triaxial in DEM, which will also discuss relative density and so on, I am by no means expert here.

Have you solved to contact law issue? It should really not be difficult to adjust the preprocessor (if you do it, push your branch to launchpad and I will be happy to merge that to trunk), but I will do it if you need (will be away till next Monday, so then).

Cheers & all the best for 2015.

Václav

Hi Vaclav,
Thank you very much for your explanation. At present, would you be kind to extend your co-operations how to achieve target porosity in Woo? I have copied and pasted the part which I have used earlier in PFC3D. I am wondering, is there any way to achieve target porosity? I am trying to compare conventional triaxial test result of Toyoura sand using Woo. For qunatitative research,except "Tunables"parameters, I am interested to explore other properties from experiment and samples.

// PFC3D_et3 (3D Triaxial environment, cubic box). Same procedure is also applicable in Cylindrical sample

def et3_genballs
;
; ----- Generate initial particle assembly contained within walls.
; Ball and wall material properties are not installed.
;
; INPUT: slx, sly, slz - specimen dimensions
; _rmin - minimum ball radius
; _rmax - maximum ball radius
; _n - target porosity
; _ntries - number of tries for the GENERATE command
; (defaults to 125000)
; _et3_poros - initial porosity
;
; --- Compute number of balls to generate
; using the expression:
; N = 3V(1 - n) / (4*pi*Rbar^3) (PFC3D)
;
   _rbar = 0.5*( _rmin + _rmax )
   _numballs = 3.0*slx*sly*slz*(1.0 - _n) / (4.0*pi*_rbar^3)
   _rlo = 0.5 * _rmin
   _rhi = 0.5 * _rmax
   _xl = -0.5*slx
   _xu = 0.5*slx
   _yl = -0.5*sly
   _yu = 0.5*sly
   _zl = -0.5*slz
   _zu = 0.5*slz
   command
     GENERATE x=(@_xl,@_xu) y=(@_yl,@_yu) z=(@_zl,@_zu) &
              rad=(@_rlo,@_rhi) &
              id=(1,@_numballs) &
              tries=@_ntries
   end_command
;
; --- Determine radius multiplier, [_m], so that we
; achieve desired porosity, [_n].
;
  _m = ( (1.0 - _n) / (1.0 - _et3_poros) )^(1.0/3.0)
  command
    change rad mult @_m
  end_command
;
  if et3_cell = 1 then
  ; --- Delete balls that lie outside of the cylindrical region.
  ;
    command
      delete ball range cylinder end1=(0.0,@_yl,0.0) &
                                 end2=(0.0,@_yu,0.0) &
                                 rad=@et3_rlen not
    end_command
  end_if
end

//

About Contact Law: Let me try first, I will inform you the update.

Thanks again.

Regards,
Mohammad Nurul Islam

Hi, if I understand correctly, you generate random packing of spheres (which are not overlapping), then you rescale it such that you have the desired porosity. But the spheres are allowed to overlap, because you increase their radius by the multiplier even if they are very close to each other, for example, do I get this right? What you the obtain is a packing with overlaps, so when you run the simulation afterwards, they might be enormous forces even just in the initial state, is that right? I don't understand what is it good for, but if it is really what you need, I help you to script that, no problem. Cheers!