Questions about the load plate test for sandy soil

Asked by Xue

Hello everyone,
    I am trying to calibrate the contact parameters of sand with the plate load tests in article [1].
    However, due to the large size of the subsequent model I want to build, the number of particles may be too high if using the real sand particle size (maximum around 2mm), so I choose to enlarge the particle size to 3~15mm.
    After that, I have built the model according to the whole size in [1], where the script after sample making is shown in(I am sorry, I have no idea how to upload the dat file after making the sample).The script of the plate load tests model is as follows:

from yade import pack,qt,plot,utils,polyhedra_utils,ymport,export,pack,timing
from yade import *
import numpy
from pprint import pprint
import random
from random import uniform
from random import randint
from math import *

steel = PolyhedraMat() #the material of wall the parameter is identical with the model in chaning.
steel.density = 7800 #kg/m^3
steel.young = 8E8 #inital steel was ,try
steel.poisson = 0.21
steel.frictionAngle = 0.3 #rad

steelp = PolyhedraMat() #the material of plate
steel.density = 2000 #kg/m^3 in order to ignore the gravity of plate
steel.young = 20E9 #inital steel was ,try
steel.poisson = 0.21
steel.frictionAngle = 0.3 #rad

spheresand = CohFrictMat()
spheresand.isCohesive = False
spheresand.alphaKr = 1.7 # 0.7
spheresand.alphaKtw = 0
spheresand.momentRotationLaw = True
spheresand.density = 2600
spheresand.young = E7 #in October 3th
spheresand.possion = 0.33
spheresand.frictionAngle = 0.977 #rad 0.77

'''spheresand = FrictMat()
spheresand.density = 2600
spheresand.young = 8E7 #in10.3
spheresand.possion = 0.33
spheresand.frictionAngle = 0.68 #rad

global a
#length unit: m; using stander unit



  [Law2_PolyhedraGeom_PolyhedraPhys_Volumetric(),Law2_ScGeom_FrictPhys_CundallStrack(),Law2_ScGeom6D_CohFrictPhys_CohesionMoment()], # contact law -- apply forces #Bo1_Cylinder_Aabb()

def cycle():
 for b in O.bodies:
  if isinstance(b.shape,Sphere):
   if b.state.pos[0] < -0.2 or b.state.pos[0] > 0.2:
   elif b.state.pos[1] < -0.26 or b.state.pos[1] > 0.26:
   elif b.state.pos[2] < -0.2 or b.state.pos[2] > 0.079:
 #for b in O.bodies:
  #if isinstance(b.shape,Sphere):

def Compact():
 global a
 if a == 1:
  ldpltheight=max([b.state.pos[2] for b in O.bodies if isinstance(b.shape,Sphere)])
  print ("pos=%.5f"%(ldpltheight))
  if O.iter < 20: return
  ldplt=polyhedra_utils.polyhedra(steel,v=((0.045,0.05,ldpltheight),(0.045,-0.05,ldpltheight),(-0.045,-0.05,ldpltheight),(-0.045,0.05,ldpltheight),(0.045,0.05,ldpltheight+0.02),(0.045,-0.05,ldpltheight+0.02),(-0.045,-0.05,ldpltheight+0.02),(-0.045,0.05,ldpltheight+0.02)),fixed=False, color=(0.75,0.65,0.65))
 elif a == 2:
 elif a == 3:
  print ("Load= %.5f"%(plateF))
  print ("h=%.5f"%(h))

def save():

    According to my preliminary calculations: under a smaller load (50kPa), the loading plate sank into the sand, which implied that the sand strength is too small. However, in the model, the friction coefficient of particles have raised to a surprisingly high level.
   Now, my questions are as follows:
    1. For the simulation, for the sand particles contact model, I have chosen Cohfrictmat to to simulate the anti-rotation effect between unbonded sand particles, but I am not quite sure if the relevant settings are correct. (May I ask how to check whether the anti-rotational moment is generated between the sand particles)
    2. For the calculation, is there an fatal error in the enlargement of the particles? According to the relevant literature, I found that after the model and particle size greater than 30, it seems that the particle size has less effect on the strength, but here the size ratio of loading plate and particle does not meet this requirement. (For this, I doubled the loading plate size, but again, under a 50 kPa load, the loading plate sank into the sandy soil.)

   Any help and advice would be greatly appreciated!

[1] Behavior of Geocell-Reinforced Sand Under a Vertical Load.

Question information

English Edit question
Yade Edit question
No assignee Edit question
Solved by:
Karol Brzezinski
Last query:
Last reply:
Revision history for this message
Jérôme Duriez (jduriez) said :


> May I ask how to check whether the anti-rotational moment is generated between the sand particles

Checking for a non-zero cont.phys.moment_bending and .moment_twist [*] could make sense, I guess. With cont an interaction of CohFrictPhys type.

Generally speaking, checking could also help for asking easier-to-answer questions.


Revision history for this message
Xue (1q12) said :

Dear Jérôme.
     Thank you very much for your answer, thank you!
     But unfortunately I haven't touched the querying of information in iphy before, so I still want to ask: how can I see the non-zerocont.phys.moment_bending you mentioned in yade?
     Thanks again for your patient answer.

Revision history for this message
Xue (1q12) said :

Moreover, I would like to apologize for the disturbance caused by the overly long statement.
     In fact, I am also concerned about the solution for the loading plate sinking into the soil specimen.
     I am not sure if it is appropriate to just scale up the particle size in the model, but leave the specimen and loading plate dimensions unchanged.

Revision history for this message
Best Karol Brzezinski (kbrzezinski) said :

Hello Xue,

upscaling particles is an acceptable technique in DEM modeling (in the case of modeling bulk properties material of relatively large volume). However, you should always analyze whether your case is appropriate. I think yes. You can also do a parametric study.

When it comes to the sinking of your plate, there may be many different reasons (material properties, particles shape, etc.). Unfortunately, I cannot run your simulation, because you load the spheres from an external file. Please look at this example [1]. It doesn't look like the rod is sinking, despite the fact that both the material and particles shape is simple. You can give more rotation resistance by a more complex constitutive model (it seems that you tried it) or by more complete shapes (e.g. clumps).

Please read this chapter to learn a little bit about "Data Mining" from simulation [2]. You can select interactions in different ways, e.g. loop over O.interactions (all interactions in model), loop over interactions of given body (e.g. body with id = 0 it would be O.bodies[0].intrs()) or you can select "manually" (eg. interaction between bodies 0 and 1 would be O.interactions[0,1]).

Once you select your interaction (let's call it your_interaction) you can read its property by 'your_interaction.phys.moment_bending'. One of the things that are great about Yade is that most of the objects are serializable. You can study your interaction step by step by using dict() method. For example your_interaction.dict() will give you "the first level set of properties" including the fact that interaction has 'phys' property. Next, you can study this phys by command your_interaction.phys.dict(), etcetera etcetera :)



Revision history for this message
Xue (1q12) said :

Thanks Karol Brzezinski, that solved my question.

Revision history for this message
Suyu (suyuu) said :
Revision history for this message
Suyu (suyuu) said :
Revision history for this message
Suyu (suyuu) said :
Revision history for this message
Suyu (suyuu) said :
Revision history for this message
Suyu (suyuu) said :
Revision history for this message
Suyu (suyuu) said :
Revision history for this message
Suyu (suyuu) said :