← Back to team overview

yade-users team mailing list archive

[Question #707888]: Why do particles in the concrete cube compression test disperse directly?


New question #707888 on Yade:

Dear all, 

In the Yade official examples, I found the concrete Brazilian splitting test [1], which is a simple process where two walls approach the specimen at a constant speed, resulting in tensile failure of the specimen. Below is the code:
# -*- encoding=utf-8 -*-
from __future__ import print_function
from yade import plot,pack
A simple script of a Brazilian splitting test.
A cylinder with x axis form 0 to specimenLength.
Load in z direction by z-perpendicular walls.
Code strongly inspired by uniax.py
# default parameters or from table
readParamsFromTable(noTableOk=True, # unknownOk=True,
	young = 24e9,
	poisson = .2,
	sigmaT = 3.5e6,
	frictionAngle = atan(0.8),
	epsCrackOnset = 1e-4,
	relDuctility = 30,
	intRadius = 1.5,
	dtSafety = .8,
	strainRate = 1,
	specimenLength = .15,
	specimenRadius = .05,
	sphereRadius = 3.5e-3,
from yade.params.table import *
# material
concreteId = O.materials.append(CpmMat(
	young = young,
	poisson = poisson,
	frictionAngle = frictionAngle,
	epsCrackOnset = epsCrackOnset,
	relDuctility = relDuctility,
	sigmaT = sigmaT,
# spheres
	radius = sphereRadius,
	spheresInCell = 500,
	memoizeDb = '/tmp/packing-brazilian.db',
	returnSpherePack = True
# walls
zMin,zMax=[pt[2] for pt in aabbExtrema()]
wallIDs = O.bodies.append([wall((0,0,z),2) for z in (zMin,zMax)])
walls = wallMin,wallMax = [O.bodies[i] for i in wallIDs]
v = strainRate * 2*specimenRadius
wallMin.state.vel = (0,0,+v)
wallMax.state.vel = (0,0,-v)
# engines
# stop condition
def stopIfDamaged():
	if O.iter < 1000: # do nothing at the beginning
	fMax = max(plot.data["f"])
	f = plot.data["f"][-1]
	if f/fMax < .6:
		print("Damaged, stopping.")
		print("ft = ",max(plot.data["stress"]))
# plot stuff
def addPlotData():
	# forces of walls. f1 is "down", f2 is "up" (f1 needs to be negated for evlauation)
	f1,f2 = [O.forces.f(i)[2] for i in wallIDs]
	f1 *= -1
	# average force
	f = .5*(f1+f2)
	# displacement (2 times each wall)
	wall = O.bodies[wallIDs[0]]
	dspl = 2*wall.state.displ()[2]
	# stress (according to standard brazilian test evaluation formula)
	stress = f/(pi*specimenRadius*specimenLength)
	# store values
		t = O.time,
		i = O.iter,
		dspl = dspl,
		f1 = f1,
		f2 = f2,
		f = f,
		stress = stress,
# plot dspl on x axis, stress on y1 axis and f,f1,f2 in y2 axis
O.dt = 0.
O.step(); # to create initial contacts
# now reset the interaction radius and go ahead
# time step
O.dt = dtSafety*PWaveTimeStep()
# run simulation

Regarding this process, I have two questions, and I would appreciate any assistance or insights from the community.

1. According to the model's micro-scale parameters, this should simulate a concrete specimen, but the stress-displacement curve indicates that the displacement is too large, roughly an order of magnitude greater than real-world conditions. Why is there such a significant difference between micro-scale parameters and macroscopic behavior, and how should I adjust it? (I believe that as an example, the micro-scale parameters in this code are reasonably set, at least in the same order of magnitude.)
2. I would like to simulate a uniaxial compression test, so I changed the original specimen shape to a cube. However, the process becomes very strange, with a sudden increase in stress, and the specimen composed of particles eventually explodes. What is causing this? Did I miss setting some parameters? 
  I have replaced the #spheres section of the original code with the following code:
# spheres
	radius = sphereRadius,
	spheresInCell = 500,
	memoizeDb = '/tmp/packing-brazilian.db',
	returnSpherePack = True


[1] https://gitlab.com/yade-dev/trunk/-/blob/master/examples/concrete/brazilian.py?ref_type=heads

You received this question notification because your team yade-users is
an answer contact for Yade.