Published on: February 7, 2023
Particle mechanics simulation is becoming more and more important in engineering across industries, particularly in solids handling. At a recent virtual roundtable moderated by Dr. Ahmad Haidari (Rocky/ESSS), ESSS invited Dr. Jörg Theuerkauf (Dow Chemical), Dr. Yidong Xia (Idaho National Laboratory), and Dr. Lucilla Almeida (Rocky/ESSS) to share their perspectives on how simulation helps engineers in these situations, what are the limitations and benefits, and what are the considerations, parameters, and options when using this method.
Lucilla Almeida, PhD, Rocky/ESSS
DEM is a numerical technique that simulates interactions between particles and boundaries, mainly to predict particle flow dynamics and bulk solids behavior. Today, many industries use DEM, such as geomechanics, agriculture, pharma, chemical processing, process engineering, oil & gas, and environmental and biological systems. More specifically, DEM is used in applications like grain transport, powder coating of medication, and water purification.
DEM is an explicit method that tracks every particle by following the laws of motion, allowing us to track every particle at every time step. It accounts for body forces, such as gravity, fluid, and electrostatic or magnetic fields. The core of DEM code is to detect particle collisions and compute contact forces. We can add additional physics to model adhesive and cohesive materials or breakage.
To help address any large-scale computational concerns, Rocky can use GPU processing. For example, using 4x A100, it is possible to solve a project 700 times faster than with 1 CPU core. Even if you have GPU processing, it may not be feasible or practical to get accurate results if you have too many particles. Commonly, in these cases people use coarse-grain modeling to reduce the total number of particles by using larger particles to represent a group of smaller particles.
Particle shape can affect the flow of particles. Spheres allow for a simple contact detection and a single point of contact, but sphere shapes are pretty rare in the real world, so we need to be able to model other shapes, which can be done in Rocky. Real particle shapes — like polyhedrons for example — differ in their packing density, the linear and rotational modes of transport, the dilating during shear and interlocking, and the strength of the materials.
Another way DEM is powerful: When coupled with CFD or SPH. When we need to account for fluid forces and flow on particles, CFD can be coupled with DEM. This coupling can be used with mesh-based approaches and with any particle shape. A SPH-DEM coupling is being increasingly used in hydrodynamics. SPH captures flow dynamics by discretizing it into a set of fluid elements.
It’s powerful because it can capture the complex-free surface flows with no diffusion errors, and dynamic body interaction is easily handled. This approach is suitable when splashing or surface fragmentation is involved, for example. SPH-DEM is often used with a Lagrangian meshless approach. Particles are interpolated using a kernel function to compute smooth fields for local variables.
Jörg Theuerkauf, Dr-Ing, Dow Chemical Company
In the chemical industry, DEM can help with many different particle system processes:
DEM is helpful in particle-particle interactions, such as when you need to understand the forces and stresses developing on those particles and possibly even describing the fracture mechanisms of the particles. The method can analyze a large number of particles — as many as 100 million. For analyses with even larger numbers of particles, you can build surrogate models by extracting the physics and then scaling up. Dow has worked with Professor R. Lueptow at Northwestern University to research particle segregation by using this method first experimentally, then computationally via DEM, then scaling up.
DEM’s benefits include direct access to bulk info away from boundaries and the ability to quantitatively measure key properties like packing fractions. With DEM, you can also optimize experimental designs. Keep in mind that DEM offers relatively simple physics, small time scales, and simplified particle geometry, which can sometimes be limitations.
For larger scales, surrogate approaches in DEM can offer a better understanding of physics for innovative solutions, the ability to scale up, and the possibility of coupling with other areas of interest, like gas flow, heat transfer, and reaction kinetics at full industrial scale.
Since DEM can be computationally expensive, it helps to think outside of the box as you’re planning an analysis. Do you need a full-scale model? What granularity is needed to solve the problem? Also, carefully consider your input parameters, and consider augmenting DEM with a two-fluid model or MPPIC.
Yidong Xia, phD, Idaho National Laboratory
At Idaho National Laboratory, scientists are studying the processing of pine particles and corn stover, which is the corn plant’s leaves, stalks, and cobs. These particles are particularly challenging since the surface and mechanical properties can change with the environmental moisture content. The goal is to reduce the risk of process disruptions and minimize downtime — for example, to reduce the instance of material clogs in a screw conveyor and jams in a hammer mill grinder.
The scientists at Idaho National Laboratory defined the shapes of the milled pine particles in four ways: extreme-poly (not usable for DEM in this case), high-poly, low-poly, or sphero-poly. Then they characterized the particles’ bulk compressibility and stiffness. The low-poly particle was able to capture the compressibility of the physical samples.
When the scientists applied the parameters from the loading test to a sliding friction shear tester, they were able to match the friction coefficient of the model particles very well with experimental data, which shows the importance of using the right particle shape and model.
Custom polyhedral particles are computationally expensive, so further reduce the computational cost, so they used the breakage model in Rocky. They tried to evaluate whether the actual physical particles could be made in a more regular shape during comminution processing of pine wood chips. With the calibrated shapes and their surface properties, they could measure the properties in the ratio tester in the lab. Also at the lab, scientists are currently modeling a screw conveyor feeder to obtain an instantaneous analysis of the risk of shaft damage. They’ll present results soon.
DEM is also used in preprocessing in a hammer mill. The hammer mill is prone to jamming, such as when the feed is too rapid, which leads to downtime. DEM can capture the flexible fiber shapes to determine the safe operation range (the feed rate, rotation speed, and so on). The scientists also used DEM to model a knife mill to determine the optimum RPM and energy for corn processing without the stalks becoming stuck. If the RPM is too low, the corn stalk could become deformed, but not comminuted.
At Idaho National Laboratory, scientists also use DEM-CFD coupling to simulate air blowing processes to determine how to filter certain types of particles like bark and needles. Finally, scientists are performing pellet durability tests in a rotation drum.
Brute-force methods don’t really work in DEM analysis, but you can consider several factors to ensure proper calibration. Ask yourself: Do you need a map of individual particles or a large system? How many particles are there? Do you want to use periodic boundaries?
Also, Rocky provides a calibration suite with examples. Simulating shear testers, like Jenike, can also work. Not all particle properties are represented in the scientific literature, so choose the tester that best resembles the material stresses or conditions you’re analyzing. You can also do an experiment first to preserve any unique behavior of the particles you’re interested in.
Finally, find a set of problems that you can recreate to mimic the actual behavior of your particle. The goal is to find a way to make sure the set of coefficients reproduce what you see in reality.
In summary, the main benefit of using DEM is the ability to model complex and multi-particle systems that are difficult to replicate in real-life experiments. This leads to a better understanding of the behavior of these systems and can provide valuable information for optimizing processes and improving product quality.
Additionally, the high accuracy of DEM simulations allows for the prediction of particle-particle interactions, particle-wall interactions, and other relevant phenomena. This can lead to reduced trial and error in the design and operation of process systems, resulting in significant cost savings and increased efficiency.
However, there are also some challenges associated with the use of DEM in process industries. One of the main challenges is the computational requirements of the simulations, which can be quite intensive and time-consuming. This requires high-performance computing capabilities, which can be difficult to obtain and can also add to the cost of the simulation.
Also, there can be limitations in the accuracy of the results due to the approximations and assumptions used in the simulation. Furthermore, the interpretation of the results can also be difficult and requires expert knowledge in the field of DEM. To overcome these challenges, it is important to have a strong understanding of the limitations of the method, as well as the strengths and weaknesses of the simulation software used.