Dr. Indresan Govender, professor and researcher at University of Cape Town (UCT) in South Africa, recently gave us an interview about how he and his team have been using Rocky DEM and the advantages they are discovering. Read the complete interview below.
Rocky DEM: Why did UCT decide to start researching with simulation software, and with Discrete Element Modeling (DEM) software in particular?
Dr. Govender: A large portion of our research is focused on mineral processing devices (mills, cyclones, screens, crushes, flotation, etc.). A key to understanding the flow and dissipation within these mineral beneficiation systems is contained in the underlying rheology. To better understand the rheology requires detailed measurements of the stresses and strain rates governing the flow. DEM is arguably the best tool for providing such “measurements” as traditional measurements are impossible due to the aggressive nature of these systems.
Rocky DEM: How can DEM simulations contribute to research and innovation at UCT?
Dr. Govender: We are currently developing a description of granular rheology using DEM and complimentary flow measurements via nuclear imaging. DEM provides the key information on stress, strain, and power density dissipation. Rocky DEM will be instrumental in extending our theories to include realistic shapes.
Rocky DEM: Why did UCT choose Rocky DEM?
Dr. Govender: Very simply, Rocky is the only DEM package that handles proper shapes. Other packages artificially achieve this by clumping spheres together. Rocky’s main advantages are realistic particle shapes and extension to GPU computing. The latter will certainly facilitate realistic simulations from a time perspective. For example, I can now run my mill simulations for 100s of revolutions using GPU in the same time it used to take to run 4s on a CPU.
Rocky DEM: What are the significant gains that you have obtained by using Rocky DEM in your projects?
Dr. Govender: A key objective in many of my research projects is to understand the stress distribution in granular flow systems. To get realistic and accurate estimates of the stress distribution, it is critical to know how the different constituents of the granular assembly are distributed. For example, in a system comprising steel balls (different sizes) and sea shells (also different sizes), the correct prediction of the stresses experienced by the sea shells can only be achieved if we can predict the spatial distribution of the sea shells. However, the spatial distribution is intimately connected to segregation, which in turn, is connected to particle shape. In this regard, Rocky is superior to all other commercial codes in that it can accurately describe particle shapes. Even super-quadrics fails to capture the asymmetric shapes and contours of sea shells. I have then used these outputs in combination with my current continuum models to predict realistic distributions of stresses, which are the precursors to breakage.
Rocky DEM: How do you validate numerical models?
Dr. Govender: My group is arguably the leading group in the world in measuring kinematics and flow of granular and fluid systems. We employ two nuclear imaging techniques: Positron Emission Particle Tracking (PEPT) and bi-planar X-ray imaging. These nuclear imaging techniques allow me to track the 3D trajectory field, velocity field, and concentration distribution of any type of particles (> 20 micrometers in diameter) within a granular flow system.
Typical PEPT measurements of granular flow in tumbling mills
These particles are real particles, which makes the information more valuable than even DEM data. Using the derived information, it is possible to validate kinematic information from equivalent DEM simulations. The technique and analysis schemes are well published by myself and other researchers that have used my PEPT facility. Beyond validation of the kinematics, I then combine the measured kinematic into my current continuum models to determine the stress/strain distributions of real granular flow systems. All of this information can be also be derived from equivalent DEM simulations. Comparing the two data sets leads to a natural and rigorous validation exercise. Validated DEM is arguably the main reason that a potential customer will by Rocky.
Rocky DEM: What is the importance of having a commercial software package that allows you customize some features by using the Python programming language?
Dr. Govender: Most DEM codes have very simplistic or rigid formats. They also suffer from lack of a flexible user interface to fully customize and control the simulation. Python is certainly a major improvement to user control. With the Python interface, I should be able to integrate my continuum models directly into Rocky for the extraction of the stress/strain field. Currently, no other commercial software package offers this level of control. I am very excited to work in collaboration with Qfinsoft to add our continuum modeling framework into Rocky.
Rocky DEM: What is the importance of simulation tools to engineering professionals and students, especially DEM software?
Dr. Govender: Safe design and scale-up rules can be investigated from the micromechanical picture described by DEM. This is ideal for consulting engineers. Another equally valuable aspect is that simulation tools allow academics to teach and illustrate several fundamental concepts without having to leave the classroom. I hope to utilize this extensively in my future teaching of courses in minerals processing and granular flow.
Dr. Govender graduated with a BSc in Maths & Physics at the University of Durban-Westville. At UCT he completed physics honors degree and then his PhD. He has been involved in comminution research for over 10 years. He currently heads up the Discrete Element Modelling (DEM) computational research within the Centre for Minerals Research. The computational modelling is strongly coupled to in-situ measurement techniques like Positron Emission Particle Tracking (PEPT) and X-ray tomography. He is currently overseeing the operations of the PEPT Cape Town facility at iThemba labs.