Modelling, Parameterization, and Validation of Simulation Approaches for Deformable Packages in Conveyor-Based Bulk Scenarios

Abstract

Introduction:              Non-rigid packages such as polybags pose significant challenges for operators of sorting and conveyor systems, as their motion and contact behaviour is more complex than that of rigid cardboard parcels [1]. Simulation models are a key tool for analysing and evaluating the motion and deformation behaviour of deformable packages in the early stages of system design, as also highlighted by recent work on deformable package simulation in robotics  [2]. While the simulation of individual packages using Multi-Flexible Body Dynamics (MFBD) [3] has already yielded promising results [4-6], a robust validation for bulk scenarios involving multiple deformable packages is still missing. The aim of this work is therefore to model, parameterise and validate two MFBD approaches, as well as to investigate the trade-off between simulation accuracy and computation time.

Method:                       The investigation is designed as a comparative validation study. It is based on an experimental dataset comprising 100 repetitions of an obstacle scenario involving four types of packages [7]. Within the simulation, two modelling approaches – a detailed shell element model and a simplified volume element model – are compared using the MFBD formulation described in [8] and a contact model for interactions between complex geometries [9]. The individual models are parameterised on the basis of targeted real-world tests to determine relevant material and contact parameters, following the methodology described in  [6], as illustrated in Figure 1. The simulations are initialised with the initial conditions measured in the real-world tests (see Figure 2), so that experimental and simulated results for 100 runs under identical starting conditions are compared. The evaluation is multi-criteria, based on transition times, lateral end positions and occupancy heatmaps, as well as, additionally, on numerical stability and computation time.

Result:                         A comparison of the experimental data and the simulations enables a detailed evaluation of both MFBD approaches in terms of agreement with experiments, numerical stability and computation time. The analysis shows that both models are generally suitable for representing the motion behaviour of deformable packages in a bulk scenario. However, they differ in the trade-off between level of detail and computational cost. The shell element model is characterised by a higher physical resolution in the description of local deformations and the package contents. In contrast, the volume element model offers potential for reducing computational costs.

Conclusion:                This work provides a basis for a methodological selection of suitable approaches for deformable packages depending on the application objective. For investigations with high demands on physical accuracy, the shell element approach proves to be particularly suitable. Volume element models offer advantages when computational time and scalability are the primary considerations. The work thus contributes to the simulation-supported development of future sorting and conveyor technology, as well as to further multi-fidelity approaches for modelling a large number of packages [10].