The project concerned the process of loading coking coal and iron
ore into the hopper. These different materials had formed piles and
pitches, leading to a reduced efficiency of the plant. The uneven
distribution of the material inside the hopper caused variations in
the temperature profile, gas flow, and gas composition.
To solve this problem, modeFRONTIER was coupled with Rocky
DEM to better understand the behavior of the materials and
to optimize the design of the deflector. The modeFRONTIER
integration also made it possible to meet the time constraints by
reducing the computational time for each simulation.
Fig. 1 - The workflow set up in modeFRONTIER to automate the simulation process in
Rocky DEM and run the design optimization strategies.
This project was developed in two phases. The first phase
concerned the calibration of the Rocky DEM parameters and
the simulation of the loading of the hopper. The second phase
consisted of optimizing the geometry of a new deflector for the
The discrete element method (DEM) parameters, such as static
particle-particle friction and rolling resistance, were used as inputs
to modeFRONTIER for the calibration process. The simulated
material’s angle of repose was used as an output.
For the optimization of the equipment, a Uniform Latin Hypercube
sensitivity analysis ran 90 designs and identified the most
important design variables. The engineers then optimized three
different geometries, using these geometric variables as inputs.
The outputs were based on the distribution of the material, which
was calculated by virtually dividing the hopper into 12 sectors and
performing a statistical analysis of the particles found in each one.
These values were then used to define the two objectives and the
The department used ESTECO’s proprietary pilOPT algorithm
to run the three optimization studies. By using the autonomous
mode, they were able to evaluate more than 1,000 designs in
just a few weeks, without having to set any parameters and with
considerable time advantages.
With its user-friendly graphical user interface, modeFRONTIER
helped automate the simulation process. Without modeFRONTIER,
engineers would have had to manually modify the geometry of
the deflector for each simulation, wasting a significant amount
of time. With modeFRONTIER they were able to save up to 130 hours of computation time. Lastly, by automating the process,
the design engineers could launch the optimization and avoid
the painstaking process of manually combining the output from
Newsletter EnginSoft Year 17 n°3