A bonded discrete element approach to simulate loading with hydraulic mining excavators

Abstract: When operating hydraulic mining excavators the loading equipment is exposed to harsh conditions which lead to extensive wear of the equipment, especially the bucket and bucket teeth. Simulations are used to better understand the wear development and to evaluate new methods to operate excavators more efficiently. At the Aitik mine, operated by the high-tech metal company Boliden Mines, hydraulic excavators are used when loading the mined ore. One of the hydraulic excavators used at Aitik is the Komatsu PC7000. In this master thesis, a simulation model for the hydraulic excavator Komatsu PC7000 was developed with the simulation software LS-DYNA. This model consists of multi rigid body dynamics to describe the motion of the excavator and a granular material model to describe the rocks loaded into the bucket of the excavator. Simulations with two different types of granular material models have been utilized to study the wear development of the bucket. One of the models (bonded DE model) uses bonded discrete elements to describe the large rocks and single discrete elements are used to describe smaller rocks. This model is compared to the current FE-DE model which is being used today at Boliden. This model uses finite elements (FE) to model the larger rocks and discrete element spheres (DES) for smaller rocks. By using the bonded DE method a 71\% reduction in simulation time could be achieved. This can be partly explained by the reduction of the number of elements included in the rock pile.  Archard's wear law was used to numerically describe the wear development of the bucket. When simulating the wear a total of 30 bucket fillings were performed with the excavator. This was done with both the bonded DE method and the FE-DE method. In this wear study, the inside of the bucket was of interest. The resulting simulated wear map was compared to experimental measurements from which the plate thickness of the bucket had been measured two times to obtain the wear depth of some points inside the bucket. The experimental measurements and two 3D scanned point clouds were used to determine the wear depth inside the bucket. Results from the simulation showed that the wear is concentrated to the center of the bucket while less wear is concentrated to the sides of the bucket. With the bonded DE method the wear appeared to be more evenly distributed inside the bucket while the wear from the FE-DE method appeared in spots inside the bucket. The experimental results also showed that the wear was more extensive in the center of the bucket and also in the back of the bucket. Both simulation methods also showed that the wear was concentrated to the back of the bucket. From the simulations, it was also seen that the behavior of the material flow differed between the two methods. In the bonded DE method the material flow had more sliding behavior while the material flow in the FE-DE method had more rolling behavior. This could also be the reason why the bonded DE method captures the wear more evenly. The rolling behavior seen in the FE-DE method leads to more impact wear which is not captured by Archard's wear law. Overall, the bonded DE method leads to a big reduction in simulation time which is favorable when it comes to simulation. The larger rocks will have simpler shapes without sharp corners. However, the method allows for a more complex shape than just an ordinary sphere which is the simplest and most common shape to describe granular material. The bonded DE method also allows for easier configuration of contact definition since fewer contact interfaces must be added to the model. Furthermore, the post-processing of wear in LS-DYNA was facilitated since the wear does not have to be divided into two wear collectors for FE elements and DE elements.