How to glue spheres to the box?

Asked by Chien-Cheng Hung

Dear all,

I am trying to simulate direct shear experiments.

In my simulation, the periodic boundary condition is applied and in addition, the top and bottom are confined by a box.
To simulate a rough boundary surface similar to the experiment conditions, I tried to glue some spheres on the box.
I thought "clump" spheres and boxes together could be a feasible way. However, it seems not to work (or maybe I did it wrong).
It turns out the spheres penetrate into the box during isotropic compression.
I also tried to glue them together by playing with cohesive, but the contact model I use is a non-cohesive Hertz contact model.

I am wondering if there are other ways I can glue the spheres and box together and therefore, the glued material can also act as confinement. Thanks!

Hereby I attach the simplified code:

##########################
from yade import pack
import math

sp=pack.SpherePack()

O.periodic=True

RADIUS=0.025
length=0.75
height=0.375
width=0.375
thickness=RADIUS
PI=1.e5
spFRIC=26.6

O.cell.hSize=Matrix3(length,0,0,0,3*height,0,0,0,width)

O.materials.append(FrictMat(density=2500,young=5.5e10,poisson=0.25,frictionAngle=radians(spFRIC),label='boxMat'))

upBoundary=[]
lowBoundary=[]

### Creat top and bottom box
upBoundary.append(O.bodies.append(utils.box(center=(length/2,2*height+thickness,1.5*width),orientation=Quaternion(1,0,0,0),extents=(2*length,thickness/2.,width),fixed=1,wire=True,color=(1,0,0),material='boxMat')))
lowBoundary.append(O.bodies.append(utils.box(center=(length/2,height-thickness,1.5*width),orientation=Quaternion(1,0,0,0),extents=(2*length,thickness/2.,width),fixed=1,wire=True,color=(1,0,0),material='boxMat')))

### Roughness setting
wallLength=15
wallWidth=10
roughnessPara=2

### top spheres
for c in range(1, wallLength):
    for d in range(0, wallWidth):
        upBoundary.append(O.bodies.append(sphere([c*RADIUS*roughnessPara,2*height-thickness,d*RADIUS],RADIUS,fixed=1,color=(72/255.,72/255.,147/255.),material='boxMat')))

### low spheres
for a in range(1, wallLength):
    for b in range(0, wallWidth):
        lowBoundary.append(O.bodies.append(sphere([a*RADIUS*roughnessPara,height+thickness,b*RADIUS],RADIUS,fixed=1,color=(72/255.,72/255.,147/255.),material='boxMat')))

### Clump spheres and box together
idClump1=O.bodies.clump(upBoundary)
idClump2=O.bodies.clump(lowBoundary)

sp.makeCloud((0,height+3*RADIUS,0),(length,2*height-3*RADIUS,width), rMean=0.025, periodic=True, seed =1)
sphere_id = O.bodies.append([utils.sphere(s[0],s[1],color=(0,0,1),material='boxMat') for s in sp])

effCellVol=(O.bodies[idClump1].state.pos[1]-O.bodies[idClump2].state.pos[1])*O.cell.hSize[0,0]*O.cell.hSize[2,2]
volRatio=(O.cell.hSize[0,0]*O.cell.hSize[1,1]*O.cell.hSize[2,2])/effCellVol

O.engines=[
 ForceResetter()
 ,InsertionSortCollider([Bo1_Box_Aabb(),Bo1_Sphere_Aabb()],verletDist=-0.1,allowBiggerThanPeriod=True)
 ,InteractionLoop(
  [Ig2_Sphere_Sphere_ScGeom6D(),Ig2_Box_Sphere_ScGeom6D()],
  [Ip2_FrictMat_FrictMat_MindlinPhys()],
  [Law2_ScGeom_MindlinPhys_Mindlin()]
 )
 ,GlobalStiffnessTimeStepper(active=1,timeStepUpdateInterval=1,timestepSafetyCoefficient=0.8,defaultDt=utils.PWaveTimeStep())
 ,PeriTriaxController(dynCell=True,mass=10,maxUnbalanced=1e-3,relStressTol=1e-4,stressMask=7,goal=(-PI/volRatio,-PI/volRatio,-PI/volRatio),globUpdate=1,maxStrainRate=(1,1,1),doneHook='triaxDone()',label='triax')
 ,NewtonIntegrator(gravity=(0,0,0),damping=0.3,label='newton')
 ]

def triaxDone():
 global phase
 vol=h*O.cell.hSize[0,0]*O.cell.hSize[2,2]
 contactStress=getStress(vol)
 vol_s=Rmean=Rmax=nbSph=0
 Rmin=1e6
 for o in O.bodies:
  if isinstance(o.shape,Sphere):
   nbSph+=1
   Rmean+=o.shape.radius
   if o.shape.radius>Rmax: Rmax=o.shape.radius
   if o.shape.radius<Rmin: Rmin=o.shape.radius
   vol_s += 4.*pi/3.*(o.shape.radius)**3
 Rmean=Rmean/nbSph
 O.pause()

#####################

Cheers,
Chien-Cheng

Question information

Language:
English Edit question
Status:
Answered
For:
Yade Edit question
Assignee:
No assignee Edit question
Last query:
Last reply:
Revision history for this message
Jan Stránský (honzik) said :
#1

Hi,

> However, it seems not to work

please be more specific.
I have tried your code, and it seemed to work ok (the box and boundary particles behaved like rigid body)

> It turns out the spheres penetrate into the box

if this is the problem, then it is probably due to stiffness and/or loading conditions

cheers
Jan

Revision history for this message
Chien-Cheng Hung (chiencheng) said :
#2

Hi Jan,

Thanks for your reply!

> please be more specific. I have tried your code, and it seemed to work ok (the box and boundary particles behaved like rigid body)

Sorry for the vague description. What I mean is after pressing the "play" button, isotropic compression begins and the spheres penetrate into the box. If I use walls as top and bottom boundaries instead of the clump, I can successfully confine the spheres within the middle space without any penetrating issue happen. It's not the thing whether the box and boundary particles behaved like a rigid body or not.

> if this is the problem, then it is probably due to stiffness and/or loading conditions

I am wondering whether I can "turn off" the periodic boundary condition in the Y direction to avoid this issue. Do you know how to do that?

Thanks again!

Chien-Cheng

Revision history for this message
Jan Stránský (honzik) said :
#3

> If I use walls as top and bottom boundaries instead of the clump, I can successfully confine the spheres within the middle space without any penetrating issue happen.

I **think**, that the boundary spheres are "more sparse" and the free particles are easier penetrated than in the case of walls

> I am wondering whether I can "turn off" the periodic boundary condition in the Y direction to avoid this issue. Do you know how to do that?

I am not sure if the problem is related to periodicity in y direction..
Anyway, "turn off" periodicity in one direction is achieved by enlarging sufficiently the periodic cell

cheers
Jan

Can you help with this problem?

Provide an answer of your own, or ask Chien-Cheng Hung for more information if necessary.

To post a message you must log in.