At Robert Bosch Packaging Technology, optiSLang was used in conjunction with Rocky DEM to obtain accurate models for the simulation of vertical filling of granular foods
Vertical filling is a flexible process and commonly used in industrial packaging of granular foods, such as candy, snacks and bakery goods. By increasing the frequency of drops of granulate portions, the output rate can be easily increased. However, the time distance between the portions must be kept large enough so that there is enough time to perform sealing. Otherwise, particles get caught between the sealing jaws, which often results in need for maintenance. Thus, compact falling of the portions is important for keeping the process reliable.
The Discrete Element Method (DEM) simplifies contacts by assuming particles to be stiff. Deformation is implemented by allowing a small overlap between particles. Contact forces are then calculated with simple relations to the current overlap. A variety of contact models are available in different DEM implementations.
2. Model Calibration
Identifying model parameters for DEM simulations is challenging. An attractive and commonly used method is numerical model calibration, which consists of varying the model parameters while comparing the simulations to experimental results until reality is reproduced to a satisfactory extend. Calibration is usually performed in a relatively simple representative experiment. A consecutive validation step can be then performed to verify if the model parameters hold up in the actual process of interest.
3. Solver Noise
Since granular systems are highly chaotic, small variations in initial conditions, such as the precise positions of individual particles in the collection bin before the drop, can dramatically affect the process outcome. Physical randomness, just as process design, can be of great importance in achieving a desirable outcome and avoiding unfavorable ones. This is true for the physical process as well as for the simulations.
For this study, model parameters for a granular sample food had to be found. The chosen good was sugar-coated, bite-size chocolate candy with a porous cookie core. As calibration trial, a drop test that is very similar to the industrial process was used representing in-situ calibration. Further, the necessity to incorporate the physical randomness in the DEM simulations and their effect on the calibration was evaluated. Finally, the methods were compared with regard to their feasibility, robustness and accuracy.