• Using the nonsmooth dynamics method for modeling both granular material and machines enables fast full-system simulations. Sometimes even with realtime performance.

    Martin Servin

    Umeå University

  • The goal is to make granular simulation accessible and easy-to-use for the persons with deep knowledge and innovative ideas about the physical system.

    Kjell-Ove Mickelsson


  • Many costly and time consuming full-scale experiments are now replaced by virtual experiments. In some cases, this has speeded up our work with several years, mainly due to less need for access to the real plant.

    Stefan Rönnbäck

    Optimation AB, Luleå

  • The solutions support rapid experimentation by interconnecting standard software and hardware for 3D CAD modeling, control algorithms and multibody dynamics with millions degrees of freedom.

    Tomas Berglund

    Algoryx Simulation AB, Umeå

Methods and Tools
  • Granular simulation - nonsmooth discrete element method (NDEM)

  • Control - interoparability through Functional Mockup Interface (FMI)

  • Modeling and visualization - 3D CAD and WebGL

A complex but important matter

Granular materials consist of many macroscopic particles, from microns to meters in size, which interact strongly with each other through frictional contacts. Examples are natural grains of cereals, rocks and minerals, manufactured pharmaceutical pills and pellets. It is estimated that more than 50% of the traded goods in the world involve handling of granular materials at some stage. These materials are rich in complex phenomena, like jamming, avalanches and size segregation. Granular media can switch quickly between solid, liquid, and gaseous form. Especially characteristic is the presence of strong force chains. These chains extend through the material and provide structural strength and determine the flow behavior. Understanding and simulation capability of these phenomena is fundamental for the design and control of systems for transportation, storage and processing of bulk materials.

Project GranuTrack

The feasibility of creating model-based machine vision by combining real-time physics simulation of granular materials with image sensors is investigated. The goal is robust tracking and determination of the state of granular matter for autonomous and optimal process control.

The technique is tested in three specific applications:

1) Detect and track individual objects for autonomous manipulation and loading. With particular focus on segmentation of ore rock piles and detection and tracking of boulders for autonomous control of rock breakers.

2) Determine the contents of bulk volumes based on surface imagery data. With particular focus on determining the particle size distribution of material on conveyor belts.

3) Tracking of material mixing and size-segregation in complex transport and storage processes. With particular focus on gravity reclaim stockpiles and silos for better control of the the material composition fed into subsequent processes.

Project partners include ABB, Algoryx Simulation, Boliden, LKAB, Optimation and Umeå University. The project, named, run 2016-09 to 2017-04.

Project GranuReg

The project “Control of granular processes” (GranuReg) aims to new knowledge, methods and tools for design, control and optimization of industrial processes with granular flows. The goals include new control methods for granular processes; solutions for interconnecting software tools for control and granular simulation; algorithms for accelerating the computational performance in granular simulation. The project run 2014-2016.

New control methods are developed and tested for transportation and processing of iron ore and pellets. The expected long-term effects are reduced energy consumption and maintenance, improved product quality and increased production capacity.

Interoperability between control software and granular flow simulation is created. The solution include modular distributed co-simulation based on the Functional Mockup Interface (FMI) standard over networks using TCP. The interface is agnostic to whether sensors and actuators are virtual or physical. This make the software and hardware components interchangeable with other options and allows for faster and easier transfer from virtual to physical systems, e.g., test-drive the simulator with operator in the loop and using different control solutions.

The project also include research and development to dramatically increase the computational performance large-scale granular systems. The nonsmooth discrete element method (NDEM) is accelerated through new algorithms and models designed for quasi-static materials.