Best Contact Model for Simulating a Partially Molten Cohesive Zone in EDEM

JT_Wijnker_11 · 2025-06-30T14:53:14+00:00

There was an error rendering this rich post.

Hi everyone,

My questions ia about choosing the most suitable modelling approach. I’m currently working on an EDEM simulation where DRI particles fall into a slag and molten metal bath inside a furnace. The particles form a feed pile that slowly melts to produce slag and hot metal. My goal is to accurately simulate the cohesive zone at the interface where the cold feed pile contacts the hot slag (or metal) bath.

This is the zone where particles start melting, stick together, and are eventually removed from the simulation. A diagram of the situation and the expected feed pile shape (with cohesive zone) is provided below

Current Setup:

I’ve developed two API plugins:

Buoyancy Plugin: Defines slag and metal bath regions. Particles inside these zones experience buoyancy and drag forces, effectively creating a simplified fluid–solid interaction without CFD.
Shrinkage and removal Plugin: Tracks the residence time of DRI particles inside the slag bath (below the slag surface). Once particles exceed a threshold residence time, they shrink, and are removed upon reaching a minimum size. This mimics the melting behavior inside the cohesive zone.

During simulation, a steady-state pile shape emerges as the feed rate balances with the removal rate. Below is a snapshot of the pile with the slag in light grey and metal in dark grey.

The region where particles have resided in the slag for over 2 seconds is visualized as the cohesive zone. These particles are colored red in the image below. This is the region I aim to model as a physically cohesive zone inside EDEM.

The shrinkage and removal plugin already works accordingly. The red depicted particles start to melt and are removed when a certain diameter is reached (naturally at the outside of the pile). My next step is to incorporate cohesive forces between these particles to more accurately mimic the cohesive zone.

What I want to model now:
Inside the cohesive zone, particles should:

Feel cohesive forces, even if not in direct contact.
Still be able to move slowly relative to each other, so that particles can gradually reposition and migrate through the cohesive zone. This is important because, as particles at the edge of the cohesive zone melt and are removed, new particles must be able to move in and replace them to maintain a steady pile structure.
Behave as if they are surrounded by a viscous medium or thin glue-like film. (Partial melting / sintering)

In other words, particles in the cohesive zone should remain clustered and resist spreading due to cohesive forces, but not become completely fixed. They should still be able to slowly move and reposition, allowing them to progress through the zone and replace melting particles at the boundary.

What I tried:

I created a custom Variable Cohesion plugin (based on Linear Cohesion V2), where cohesion is applied as normal force proportional to the effective contact area A = π * R * δ if their residence time is above 2 seconds. However, this is not producing the desired behaviour:

Cohesion is only active during direct contact
Cohesion is weak (even for very large surface energy density) as the overlap of particles is also very low
Contact is easily broken
The pile collapses, and no cohesive zone shell is formed keeping the pile together

Thus, this model does not accurately mimic the cohesive zone.

What I am considering next
I am evaluating two alternative contact models to base my next plugin on:

Liquid Bridge Model:
1. Allows cohesion at distance (rupture length), not just on contact.
2. In my mind a liquid bridge seems to mimic the sticky, viscous nature of the cohesive zone. (I does not model viscous forces really)
3. Concern: This model only implements a normal force which is very similar to my current cohesion plugin. I am worried if forces will be strong enough and if behaviour will be different from my current implementation.
Bonding V2 Model:
1. Particles form bonds with normal and shear stiffness, and breakage thresholds.
2. Bonding is strong enough to form this very cohesive zone / solidified shell.
3. Concern: the bonds may be too rigid, potentially preventing particles from shifting or repositioning as they melt and new particles move into the cohesive zone.
4. The default Bonding V2 model only forms bonds once at the moment of contact. In my case, I plan to modify the logic so that at every time step, the plugin checks whether two particles are close enough and at least one of them has a residence time above 2 seconds. If these conditions are met and the particles dont have a bond already, a new bond would be created, even if a previous one was broken. By assigning low bond stiffness, the model could allow some relative motion while keeping particles connected. Although Bonding V2 does not include a built-in rupture distance, but I feel like this could mimic the same behaviour as the liquid bridge model. I dont know if it will work as I imagine.

My question:

Which contacct model is best suited to simulate the partially molten / sintered behaviour inside the cohesive zone?
Should I use the Liquid Bridge model, Bonding V2, or are there better alternatives?

Any suggestions are very welcome. Thanks in advance for the help!

Note: Please dont worry about how to integrate the cohesive model work with the shrinkage plugin. I am purely interested in what contact model conceptually is best suitable to mimic the cohesive zone behaviour. I am sure this can be handled.

Kind regards,

Jan-Thijn Wijnker

Find more posts tagged with

EDEM

FAQ

How To

Scripts

Accepted answers

Stephen Cole

Hi Jan-Thijn,

I would consider using the liquid bridge model, or a variation of this as the bonding model doesn't typicalyl allow any motion relative to the oroignal bond position. The bond is elastic and always tries to return the particle to it's original bond position (relative to the particles it's bonded to).

Regards

Stephen

Stephen Cole

Hi Jan,

You can see a version of the liquid bridge model here -

https://community.altair.com/discussion/36811/edem-api-contact-model-example-liquid-bridge-cohesion

Its not identical to the one in-built to EDEM as the in-built version was implemented after this. The best way as you say is to combine the models so you can then trigger one or the other based on the condition you require.

Regards

Stephen

Stephen Cole

Hi Jan-Thijn,

I think some of the questions are covered in your other post, but just to confirm the API gives you Contact Overlap as well as Physical Overlap, so you can use either in the code:

Also i'm not sure if it's useful in your case but in 2025.0 and later there is a 'preserve contact' option. So once the contact is detected it can be kept as a contact, regardless of physical distance between the elements until you want to break it.

Just to note if you do use this there is a distance limit set in the Solver Advanced Settings also, so even if you use preserve contact it will break after 4x radius unless you modify the default settings:

All comments

Stephen Cole

Hi Jan-Thijn,

Regards

Stephen

JT_Wijnker_11

Hi Stephen,

Thank you for your response and clear answer. There is an additional challenge. Namely, I want to keep using the Hertz-Mindlin contact model as base model and "activate" the liquid bridge model based on certain temperature criterion. The reason for this is that the liquid bridge model should of course only be active in the cohesive zone where partial melting occurs. Now, my idea is to just make my own Contact Model plugin that adds this liquid bridge cohesion logic. However, I have noticed that there is no Liquid Bridge Model source code available correct? And my understanding is that the model in EDEM 2025 does not allow to set Temperature based behaviour.

I am currently using EDEM 2024.1 API but I could switch to 2025 if needed.

Do you maybe have an idea on how to best approach this or if the source code is provided anywhere?

Thanks for the support thusfar.

Kind regards,

Jan-Thijn Wijnker

Stephen Cole

Hi Jan,

You can see a version of the liquid bridge model here -

https://community.altair.com/discussion/36811/edem-api-contact-model-example-liquid-bridge-cohesion

Regards

Stephen

JT_Wijnker_11

Hi Stephen,

As I am working now on my cohesive model, I would like the add a non-contact force using the particle contact radius in EDEM.

My question is if this is compatible with using a particle size distribution as well. For example, if my physical diameter is 12mm and I set my contact radius to 14mm. When adding a particle size distribution, will this also scale the contact radius accordingly when creating particles?

Maybe some background:
I am creating a model in which I want to mimic a molten layer around the core of my particle. At the start of my model, the physical and contact radius of the particle are set equally, for example 12mm. Then, based on my heat transfer model, the physical diameter of the particle will shrink using setScale(). Now, I want the particle contact radius to remain the original size, namely 12mm. The difference between this contact radius and the physical radius represents a molten layer surrounding the solid core, gradually increasing when the particle shrinks.

Therefore, I am wondering if setScale() also updates the contact radius of the particle as that would result in wrong behaviour (shrinking the contact radius instead of keeping it 12mm). I want a model in which I can shrink the physical radius but keep the contact radius constant to mimic the molten layer around the "core". Is there a manual way to reassign only the contact radius for example? Then I can add that after setScale to counter this. So, is it possible to manually set contact radius if it is enabled, independently of the physical radius during runtime.

And I am also wondering if later on adding a particle size distribution will scale the set contact radius correctly upon creation. Or if I will have for example particles with 6mm physical radius and 12mm contact radius (which would be wrong of course for my purpose).

I am also assuming that contact-radius overlap is is not automitically calculated? EDEM only conducts actually physical overlap calculations? That would not be a big problem because I could ofcourse compute these myselfs, but just wondering if that is build in. I require both contact radius overlap to calculate the interaction of the molten layers as well as the physical radius overlap for core-core interactions.

And finally, can the contact radius also be visualised in the post processor? I assume no but would be nice haha.

Can you elaborate maybe on how this is handled within EDEM.

Thank you for your help.

Kind regards,

Jan-Thijn Wijnker

Stephen Cole

Hi Jan-Thijn,

I think some of the questions are covered in your other post, but just to confirm the API gives you Contact Overlap as well as Physical Overlap, so you can use either in the code:

JT_Wijnker_11

Thanks for answering my questions. Yes the preserve contact is also usefull in my case and does solve (a part) of the problem.

See my other post for my suggestion.