Module 4 · Advanced physics
This module is still locked
To unlock it, pass the Module 3 quiz with at least 70%.
Go to the quiz · Module 3Cohesion
Lesson transcript
In dry, coarse material, nothing sticks to anything. But with moisture or fine powder, everything changes. The flow rate drops. Arches and blockages appear. And the material forms agglomerates. Cohesion is switched on in the interaction, for each pair. Up here you choose the pair: Vidro with Vidro. Then check Enable Cohesion. Cohesion Model has three options. Let's look at each one. Linear cohesion is the simplest, the SJKR model. It adds an attractive force proportional to the contact area. The only field is the energy density k, in joules per cubic meter. The text just below suggests the range, from a hundred to a million. It works with linear and Hertz contact, and it is cheap. JKR is the full adhesion model. The field is the surface energy gamma, in joules per square meter. It sets the pull-off force. To separate, you have to pull past the point of contact. It only works with Hertz-Mindlin, and it is the choice for fine powders. It applies to particle-particle and particle-wall pairs. And Liquid bridge is the liquid bridge, for wet material. It has a video of its own. After choosing, click Save Interaction, and repeat for every pair that needs cohesion. A quick guide. Want a simple, cheap attraction? Linear cohesion. A fine powder that adheres? JKR. Wet material? Liquid bridge. And with all of them, review the time step, with a Rayleigh factor down to zero point one.
Check what you learned
5 multiple-choice questions. Get at least 4 right to complete the module. You can try as many times as you like.
Transcripts of every lesson in this module
1. Beyond the sphere
So far, everything was done with spheres. Now let's get closer to real equipment. Grains with shape. Parts that are pushed by the material. Cases with too many grains. Sticky powder and wet material. And grains that break, and parts that wear.
2. Clumps
A real grain has corners, and that is why it locks. With spheres, we mimic that with rolling resistance. But sometimes shape drives the result. An elongated material interlocks, forms bridges, blocks. That is when we need real shape. The solution is the clump: several spheres glued into one rigid body. Contact is still between spheres, so the math stays cheap. But the group moves as a single grain. A clump is the shape of the material. Under Shape, I click Edit. The editor opens. On the left there are three tabs, which are three ways to build the grain. In the first one, Preset, you click a ready-made shape. Below the buttons, four fields adjust the shape. Sphere radius is the radius of each sphere. Count is how many spheres. Relative overlap is how much one sinks into the other: more overlap, a smoother grain. Radius ratio changes the size ratio between the two spheres of the dimer. The 3D preview is in the center. Show STL and Show spheres choose what is displayed, and Fit view reframes it. On the right, the program computes the number of spheres, the mass, the equivalent radius and the smallest sphere by itself. That smallest sphere matters, because it controls the time step. Calculated details shows the volume and the inertia. The second tab fills a closed STL mesh with spheres. You control the maximum and minimum radius, the overlap and the maximum number of spheres. The third one is manual: a table with the position and radius of each sphere. You can type it in, or import a CSV file. Apply shape saves the clump into the material. After that, the inlet injects it like any other material. The more spheres per grain, the better the shape, and the more expensive the run. And Use sphere goes back to a plain sphere. Four limits to remember. Each shape takes up to four thousand ninety-six spheres. On the GPU, one thousand twenty-four. The wall material cannot be a clump. And an inlet with a mix of materials does not accept clumps: use one inlet for each.
3. Free body
Until now, parts followed a motion that you prescribed. A free body is different: gravity and the material move the part. Think of a gate being pushed open by the material, or a loose bucket. Select the part, open the list under Motion, and choose Free 6-DOF. Now gravity and the material move the part. Start with Quick setup. Free leaves all six degrees of freedom open. Fixed body locks everything, and is useful to test the case before releasing the part. Rotation only lets it turn around a single axis, like a gate. Choose one, and click Apply preset. The green line shows which degrees of freedom are free. This button computes mass, center of mass and inertia from the mesh and the material density. The mesh must be closed. Always prefer this calculation. Mass holds the kilograms, and Center of mass is the point where the part is integrated, in world coordinates. Initial orientation is the starting pose, in degrees. Initial linear velocity and angular velocity are the starting speeds. Leave them at zero to start from rest. The inertia tensor, in kilogram square meters, normally comes from the automatic calculation. Excitations and constraints holds the loads and the restraints. External force and torque add constant values, on top of gravity and contacts. Do not use them to imitate a hinge. Axis locks lock each translation and each rotation, one by one. Physical hinge is the real hinge: fix a pivot in the world, enter the axis, and the part only turns around it. Hybrid drive holds a velocity with a limited force, leaving the rest free. Springs and dampers creates an elastic reaction around a position. And Advanced elastic limits are the end stops. None of them enters the time step calculation. Finally, Apply Motion.
4. Coarse-graining
Sometimes the problem has more grains than the computer can handle. Coarse-graining replaces several real grains with one larger particle. With a factor of two, each particle stands for eight. With a factor of three, twenty-seven. Coarse-graining is also under Materials. I open Configure coarse-graining factor. Scale factor is the CGF. It multiplies the radius. With two, each particle stands for eight. With three, twenty-seven. The box below, Reduce E by CGF, divides Young's modulus by the factor squared. That helps the time step, but changes the contact stiffness. Leave it unchecked unless you recalibrate. The gain is huge. With a factor of two, the particle count drops to about one eighth. With a factor of three, to one twenty-seventh. The price is losing what happens at the grain scale, such as fine segregation and arching. Calibrate at the real size: the program scales restitution, friction and cohesion on its own. And whenever you can, compare with a fine case.
5. Cohesion
In dry, coarse material, nothing sticks to anything. But with moisture or fine powder, everything changes. The flow rate drops. Arches and blockages appear. And the material forms agglomerates. Cohesion is switched on in the interaction, for each pair. Up here you choose the pair: Vidro with Vidro. Then check Enable Cohesion. Cohesion Model has three options. Let's look at each one. Linear cohesion is the simplest, the SJKR model. It adds an attractive force proportional to the contact area. The only field is the energy density k, in joules per cubic meter. The text just below suggests the range, from a hundred to a million. It works with linear and Hertz contact, and it is cheap. JKR is the full adhesion model. The field is the surface energy gamma, in joules per square meter. It sets the pull-off force. To separate, you have to pull past the point of contact. It only works with Hertz-Mindlin, and it is the choice for fine powders. It applies to particle-particle and particle-wall pairs. And Liquid bridge is the liquid bridge, for wet material. It has a video of its own. After choosing, click Save Interaction, and repeat for every pair that needs cohesion. A quick guide. Want a simple, cheap attraction? Linear cohesion. A fine powder that adheres? JKR. Wet material? Liquid bridge. And with all of them, review the time step, with a Rayleigh factor down to zero point one.
6. Liquid bridge
Take damp sand: you can build a sandcastle. What holds it is the liquid bridge. A small amount of liquid joins two grains. Surface tension pulls one toward the other. When the grains move apart, the bridge stretches until it breaks. And unlike dry contact, the attraction exists even without touching. The liquid bridge is set up in three places. First, in the material. Open Advanced material models. Liquid content is the amount of liquid on the particle surface, as a percentage of dry mass. One percent is ten grams per kilogram of solid. Do not count liquid absorbed inside the grain. In Liquid properties, pick the liquid: water, oil or custom. These values are global, they apply to the whole case. Surface tension scales the capillary force. Density converts liquid mass into volume. And Viscosity only acts if the viscous part is switched on. Water is close to zero point zero seven two newtons per meter, and one millipascal second. Careful: entering the liquid content alone does not switch the force on. The second step is in the pair: select Liquid bridge under cohesion. Contact angle is the wetting angle of the liquid on the surface. Zero degrees is full wetting. Keep it at zero, the recommended range for the program's correlation, and do not confuse it with a friction angle. Viscous contribution switches on the liquid's viscous resistance, which slows the motion while the bridge exists. Switched off, only the capillary attraction remains. Advanced holds two fields. The volume limit per bridge, with a default of zero point zero five. It is dimensionless: volume divided by the radius cubed. It is not five percent moisture. And the minimum separation S zero, one micrometer, which avoids an infinite viscous resistance at touch. Golden rule: do not tune only the liquid content to hit a force. Also check the liquid balance in Results. And remember the limits: the liquid bridge does not work with clumps, and it represents small, isolated bridges, not soaked material.
7. Particle breakage
In mills and crushers, breaking is the goal. In crushing and grinding, it is what you want. In handling, it is a problem: the material degrades. And the fines produced change everything downstream. BrixDEM handles this with the Ab t ten model. The idea in four steps. Each impact delivers energy per kilogram to the grain. Even without breaking, that energy builds up damage. With enough damage, the grain breaks. The strength of each particle is drawn at random, because real grains are not all alike. And t ten tells how fine it got: it is the percentage of mass that passes one tenth of the original size. Breakage is also a property of the particle material. It is under Advanced material models, Breakage and fragmentation. The model is Ab t ten. I click Configure. Mode has three levels. Disabled computes nothing. Damage and fracture detection tracks damage and counts breakage events, but does not replace the particle. And Damage plus fragment replacement swaps the particle for spherical fragments. Reference material loads published examples: copper ore, glass, rice, coal and feed pellets. But careful: picking an example does not calibrate your case. It is only a starting point. Now the strength. S, the selection coefficient, in kilograms per joule, tells how sensitive the particle is to energy: a larger S breaks more easily. d ref is the reference size for that value. And E min ref is the energy below which there is no damage at all. Then, the fragments. A is the maximum t ten: the largest percentage of mass that passes one tenth of the original size. It is not the probability of breaking. Max fragments limits how many spheres are created per event. And d min is the size below which a fragment does not break again. On the right, the preview shows the response to a single impact. E min is where damage starts. E fifty is the energy at which half of the fresh particles break. And E ninety, nine out of ten. Use this chart to check the numbers, not to validate. Advanced holds fine-tuning choices, such as damage inheritance and the random seed. To save, click Apply. Here I cancel, because this case does not use breakage. Keep CGF at one, because breakage does not accept coarse-graining. After the run, fragmentation results appear in Results. Four warnings. Breakage requires a CGF of one. Small fragments call for a smaller time step. Check it before running. The reference materials are examples, not calibration. And Module 5 shows the calibration path, with drop-weight tests.
8. Wear
Wear comes from two mechanisms. In abrasion, the grain slides over the wall and scratches it. In impact erosion, the grain hits and tears material away. Abrasion depends on force and sliding. Impact depends on speed and angle. Wear is switched on per surface. Select the part and open Properties. There are two independent boxes: abrasion, with the Archard model, and impact, with the Oka model. I check abrasion. A single field appears, Abrasive coef. It is K over H, in one over pascal. It combines the material and the wall hardness into one number. Wear grows linearly with this value. The default is ten to the minus seventeen. Now I check impact. The Oka model fields appear. Vickers hardness is the wall hardness, in gigapascals. K, the master coefficient, scales the whole result: double K, double the wear. That is where you apply the rule of three. k one is the hardness exponent. k two is the velocity exponent: the default, two point three, is what makes speed dominate. k three is the particle diameter exponent. k one and k three come switched off, at zero. n one and n two shape the curve with the impact angle. They decide whether the worst hit is shallow or head-on. And v ref and d ref are the speed and diameter of the reference test: one hundred and four meters per second and three hundred and twenty-six micrometers. They are not from your case. Do not change them without a new test. Coefficients per material pair opens a table with one row per particle material that touches this surface. An empty field uses the global value above. That way, ore and limestone can have their own coefficients in the same case. The exponents and the hardness still belong to the surface. Finally, Apply Properties saves everything. Only then does wear enter the simulation. One recommendation. With the default values, the map shows with good confidence where it wears most. To predict service life in hours, you need to calibrate. But comparing two designs is already very valuable, and you can do it from day one. The special module goes deeper into all of this.
9. Gate and mixer in practice
This case brings together what we have seen. Material leaves a chute. And hits a hinged gate, which is a free body. The gate opens on its own, pushed by the load. And in real equipment, these features show up together. In this twin-shaft mixer, the shape and the motion of the parts decide the result.
10. How far the model goes
Every model has limits, and it is good to know where they are. There is no air or water around the particles, only contact. The wall is rigid. And the part does not change shape as it wears. If any of this is essential to your problem, take it into account when you use the result.
11. Which feature to use
To close, a quick guide. A simple pile or flow? Spheres with rolling resistance. Does shape drive the result? Clumps. Does the part react to the material? Free body. Too many grains? Coarse-graining. Does the grain break? Ab t ten. Always with the simplest case that answers your question.