Simulating 30 million particles to help design a dry powder mixer
Share this
Published on: August 19, 2021
This content is part of the “ Breaking Barriers with Rocky DEM” blog series. No matter your industry or application, there are many ways to leverage the power of discrete element modeling tools. Read on to discover how Rocky solves industrial problems in engineering workflow. In this blog post, we highlight the simulation of powder mixing that contains millions of particles.
Thirty million particles. It used to take weeks, maybe months, to complete a DEM simulation and find a solution for a project with that many particles. Now Rocky DEM is transforming complex problems involving multiple particles into opportunities to develop optimal equipment. And even better, Rocky get you there faster than ever before.
Case study of a ribbon mixer design
Our case study is a ribbon mixer design used in a solid mixing application involving 30 million multi-group particles–a high solid-loading and intricate analysis from anyone’s perspective. In this study, Rocky DEM evaluated ribbon mixer performance for a food industry application that stirred flour and two additives: a total of 30 million non-rolling spherical particles containing both dry and wet materials. The goal was to study the mixing efficiency of each design in days rather than weeks. The challenge was to simulate mixing a large number of small particles and to run the simulation for 30 minutes of process time. Successful application of Rocky to powder mixing opens the possibilities for engineers and designers to simulate and improve the quality and performance of a wide range of mixing applications while accounting for particles, processes, and types of mixers.
A simulation of a ribbon mixer.
DEM is well suited to optimize products and processes dealing with discrete and particle-laden flows. This is because from the get-go, DEM simulations are based on simple Newtonian laws of motion. For understanding particle-to-particle interaction, the workflow is intuitive. With a 3-D geometry design, you can set up an analysis quickly and gain a better understanding of products and processes. Discrete element modeling can be applied with a high degree of success. DEM’s advantages include accurate, detailed dynamic particle prediction. Many particle-level engineering phenomena (mixing, segregation, etc.) can be described and studied easily. The calculation time depends on a number of parameters. This includes the number of particles and the speed and scalability of the DEM solver. That is why Rocky can solve engineering problems with large particles in a quick time frame, even if the problem has 30 million or more particles.
Powder Mixing
Powders are mixed using many different types of mixers via three main mechanisms – convection, shear, and diffusion. For example, convection involves large-scale movement of particles from one location to another. Shear creates slip planes within a powder, usually when particles are interchanged between layers. Diffusion can occur when particles with random motion roll over a sloping powder surface, redistributing the mixture.
In contrast, segregation is the process of unmixing, a result of too much agitation. This occurs naturally when particles are poured in a heap, which causes larger particles to run down the cone’s edge, as well as when mixing via vibration, which forces larger particles to rise to the surface.
Ribbon Mixer in Action
The double-spiral ribbon mixer design incorporates an outer component that moves material in one direction and an inner part that transports matter in the opposite direction. This movement provides a fast but gentle mixing action; the design of the double spiral decreases overall mix time.
The process complexity increases as the ribbons rotate and particles interact with each other. Rocky addresses the problem using the latest hardware solutions, all within a workable time frame.
DEM Simulation Workflow
Model various physical components and their motions
Define particle properties and configure injection
Compute various particle–particle interactions as well as particle–wall interactions over time
Report particle data at various time intervals for positions, velocities, and forces
Analyze process effectiveness
Modify process parameters
Simulate and analyze process effectiveness
A rotating double-spiral ribbon mixer design incorporates an outer component that moves the product in one direction and an inner ribbon that moves it in the opposite direction. Simulation resolution time frame depends greatly on particle number, selected hardware and GPU.
Rocky’s multi-GPU capabilities make all the difference. It is possible to achieve substantial performance improvement–as much as 70%– by simply using a more advanced GPU (8x Nvidia A100). Rocky scalability was also evident as the simulations were moved from single to multi-GPUs. We simulated the mixer and compared the performance of two types of ribbon mixer design using Rocky to meet engineering projects’ deadlines.
Case study particle count: 30 million Ribbon rotation speed: 42 rpm Hardware resource: GPU: 2x Nvidia V100 Simulation total time:1.45 days per physical second of simulation.
Computational comparison: Using 8x Nvidia A100 GPUs (compared to 2x Tesla V100) speeds up solution time.
Rocky’s post-processing capability to evaluate the simulation results
When evaluating the simulation results, Rocky’s post-processing capability can easily and seamlessly help you analyze a complex project that involves millions of particles.
How do we actually determine a mixer’s performance? The Lacey mixing index assesses mixing uniformity by measuring mass fraction variance, providing a rating between zero and 1. The upper measure signifies complete homogeneity. To arrive at a measurement, we repeatedly sampled the mixture to measure the mass fraction variance across all samples, determining an average additive particle mass fraction. This process is difficult to perform using bench testing. But Rocky DEM provides particle-level data access. So it is convenient to find the point at which mixing is optimized — and learn when to stop mixing!
The Lacey mixing index evaluates the mixing quality of particles based on the standard.
Engineers can also obtain very detailed information, such as ribbon mixing dynamics, for each of the slicing sections.
Comparing Ribbon Mixing Design Candidates
During the equipment design phase, different geometries can produce vastly diverse particle behavior, and simulation can map the results. For example, the double helixes in mixer Design 1 are symmetrical, and we would expect the inner ribbon to move particles toward the center and the external ribbon to move particles away from center. However, design 2 is non-symmetrical, with the theory that the internal and external ribbons will nudge particles rightward and leftward, respectively. Simulation provides data based on actual particle behavior, and the results of this investigation confirmed our expectations.
Symmetrical and non-symmetrical helix/ribbon designs.
In mixer Design 1, particles are being dragged outward and then back to center. In Design 2, the components actually shear particles as the inner ribbon drives particles from one side to the other, and the outer ribbon drives particles in the opposite directions. Comparing the two studies shows that Design 1 is more efficient, since it reaches a higher homogeneity level faster.
Watch the video below for more information about this ribbon mixer case study, and visit our youtube channel.
Rocky DEM simulation with Lacey mixing index unfolding over time. The mixer takes 16 seconds to reach homogeneity. (Grey represents flour; olive green represents the initial additive; rust red represents the cohesive additive)
DEM tools enable engineers to easily obtain deeper insights into the process “micro-environment” at a fraction of the cost needed to run an experiment. These simulation insights are useful to achieve reliable process performance and assured product quality.
Ahmad Haghnegahdar
Applications Engineer, Rocky DEM Business Unit at ESSS
Ahmad holds a Master of Science degree in Chemical Engineering from Oklahoma State University with experience in analyzing multiphase and turbulent flows in the field of
biomechanics. His prior roles included developing and optimizing chemical processes through data analysis tools in the oil and gas industries. Given his diverse background from petroleum to pharmaceuticals, Ahmad recognizes and understands the challenges ahead of simulating multiphysics systems across many industries.
