THE FIRST GRAIN THAT BROKE THE SILO
The alarm blared at 3:17 a.m. Sarah’s phone lit up with a single line: “Silo 7 discharge jammed. Production halted.” She pulled on her boots, her breath fogging in the cold plant air. Inside the control room, the HMI screen pulsed red—500 tons of corn kernels frozen mid-fall. The maintenance crew had already tried the usual fixes: vibrators, air cannons, even a sledgehammer. Nothing worked.
Sarah remembered the DEM simulation she’d run last month. A tiny digital twin of the silo, grain by grain, tumbling under gravity. She’d seen the exact moment the flow channel collapsed into a rat-hole. That night, she’d dismissed it as “just a model.” Now, staring at the real silo’s dead screen, she realized the model had been screaming the truth all along.
She pulled up the simulation files on her laptop, fingers flying over the keyboard. The fix was buried in the particle-scale data: a 2-degree tilt in the hopper angle would break the arch. She sent the coordinates to the foreman. Thirty minutes later, the silo groaned, then roared as 500 tons of corn surged back to life. The plant restarted. Sarah exhaled, watching the grains flow like pixels on a screen—real and digital finally in sync.
That night, she learned the first rule of DEM simulation: trust the grains, not the guesses.
—
WHAT IS DEM SIMULATION, REALLY?
DEM stands for Discrete Element Method. It’s a way to simulate how individual particles—grains, tablets, rocks, even coffee beans—move and interact. Unlike fluid dynamics, which treats everything as a smooth continuum, DEM tracks every single particle. Each one has size, shape, mass, and friction. When two particles collide, the simulation calculates the exact force, bounce, and rotation.
Think of it like a super-detailed video game. Instead of rendering trees and cars, you’re rendering sand, pills, or fertilizer pellets. The physics engine doesn’t cheat—it follows Newton’s laws for every single particle, every single time step. That’s why DEM is perfect for problems where individual particle behavior matters: clogging, segregation, wear, or mixing.
—
HOW TO SET UP YOUR FIRST DEM SIMULATION: A STEP-BY-STEP WALKTHROUGH
You don’t need a supercomputer. A decent gaming PC with a good GPU can run small DEM models. Here’s how to start.
STEP 1: PICK YOUR SOFTWARE
Free options: LIGGGHTS (open-source), Blaze-DEM (GPU-accelerated), or Yade (Python-friendly).
Paid options: EDEM (Altair), Rocky DEM (ESSS), or STAR-CCM+ (Siemens).
For beginners, start with EDEM or Rocky DEM—they have intuitive GUIs and built-in tutorials.
STEP 2: DEFINE YOUR PARTICLES
Shape: Spheres are easiest. Real particles aren’t perfect spheres, but spheres run fast and teach you the basics.
Size distribution: Use a normal or log-normal distribution. For example, if simulating corn, set mean diameter to 8 mm with 10% standard deviation.
Transfer Chute Design properties: Density (750 kg/m³ for corn), Young’s modulus (1 GPa), Poisson’s ratio (0.3), and friction coefficients (0.3 for particle-particle, 0.5 for particle-wall).
STEP 3: BUILD YOUR GEOMETRY
Start simple: a box, a chute, or a small hopper. Import STEP or STL files from CAD software.
In EDEM, use the “Geometry” tab to create or import shapes. Assign material properties to walls (steel, rubber, etc.).
STEP 4: SET UP THE PHYSICS
Time step: Critical. Too large, and particles tunnel through each other. Too small, and the simulation takes forever.
Rule of thumb: Time step = 20% of the Rayleigh time (the time it takes a wave to travel across a particle). For 8 mm corn, that’s about 1e-5 seconds.
Solver: Use the default Hertz-Mindlin contact model. It’s accurate for most granular flows.
STEP 5: DEFINE BOUNDARY CONDITIONS
Gravity: Always on (9.81 m/s² downward).
Initial conditions: Fill the geometry with particles. Use “Factory” tools to generate particles inside a volume or on a surface.
Motion: Rotate a drum, vibrate a screen, or move a conveyor. In EDEM, use “Motion” to define rotational or translational speeds.
STEP 6: RUN THE SIMULATION
Start with 10,000 particles. Monitor the simulation in real-time. Watch for unphysical behavior: particles flying through walls, excessive overlap, or unrealistic bouncing.
If the simulation crashes, reduce the time step or check for geometry errors.
—
HOW TO VALIDATE YOUR DEM SIMULATION: DON’T TRUST THE PRETTY PICTURES
A simulation is useless if it doesn’t match reality. Validation is how you prove your model isn’t just a digital fantasy.
METHOD 1: COMPARE TO EXPERIMENTS
Start with a simple experiment. Fill a clear acrylic box with real particles. Measure the angle of repose (the slope formed when particles pile up). Run the same setup in DEM. Adjust friction coefficients until the angles match.
For example, if real sand forms a 35-degree pile but your DEM sand forms 45 degrees, increase the particle-particle friction coefficient.
METHOD 2: USE VIDEO ANALYSIS
Record high-speed video of real particle flow (e.g., a chute or mixer). Track individual particles using software like Tracker or PTVlab