R&D teams today increasingly turn to system-level digital twins to develop, improve, and analyze their equipment and processes — especially in structural, fluid and electromagnetic domains.
But the mining industry has not had such opportunity concerning bulk flow applications, even though it faces similar mandates: Reduce downtime, better predict maintenance, tweak efficiency, increase output, and improve safety. Until now.
ANSYS and Rocky DEM software companies have developed a way to leverage mechanical, design-of-experiment, and discrete element modeling (DEM) tools to create a digital twin for the mining, bulk material and process industries.
This solution uses reduced-order models (ROMs) that minimize compute cost/time yet are accurate in predicting real-world operation.
Beyond financial and efficiency benefits, digital twins offer the industry an opportunity to improve its safety record.
With automation, companies can reduce risk to bulk material handlers; minimal equipment breakage (a result of better maintenance scheduling) means that fewer workers get hurt.
How a Digital Twin Works
For computational efficiency, a physics-based digital twin starts with ROMs based on full 3-D physics. These are connected to a real, on-site asset via an internet of things (IOT) platform that inputs sensor data.
The resulting digital twin provides additional data that cannot be measured.
There is no need to refer to or rely on historical data. Digital twins have been used to accelerate product R&D by simulating product components, to identify and troubleshoot potential process issues, and to predict/improve product performance.
ANSYS® products lend powerful capabilities to building the digital twin. For a mining industry application, the simulation also pulls in Rocky DEM reduced-order models (Figure 1).
These ROMs estimate behavior in continuous space by using quick evaluations, rather than doing a full-physics solve. Additionally, Rocky DEM is integrated into both the ANSYS Workbench™ and parameter (ANSYS DesignXplorer™) environments, which streamlines the analysis process.
Mining Industry Case Study
In this real-world mining test case, shuttle cars carry mined coal from the rock face to an intake feeder. The raw material is dumped onto a conveyor that delivers it to a crushing mechanism located downstream from the intake.
As the cars dump their loads (sometimes too quickly), spillage inadvertently occurs and builds up, affecting where the next shuttle can park to unload.
When more than one shuttle offloads at the same time, the mass flow rate along the conveyor is higher than the downstream crushing system can handle, causing clogging and subsequent equipment damage.
Operators can control the system in case of overload or run it on automatic — but any change affects one end of the system or the other.
The engineering team concluded that analysis was needed to assess how all the variables in the system affect the feeder’s mass flow rate (Figure 2).
The goal was to optimize all components working together and learn how much material can be fed downstream before overflow occurs.
The main variables studied were feeder conveyor speed and raw material mass. DEM analysis alone would take too long to run, requiring a new solve for each set of inputs. On the other hand, a ROM approach provides instant results and is best for analyzing a highly variable system. However, it provides only an estimated answer.
The Final Approach
The best approach, engineers decided, called for a system digital twin, adding a final simulation with Rocky DEM for validation.
Rather than develop a single-use-reduced order model, the team set up two separate ROMs — shuttle car and feeder — to handle all the variations (Figure 3).
The shuttle car test-case setup involved 20 tons of raw material running on the conveyor at different speeds.
The team conducted four simulations of the shuttle car and feeder each. These simulations served as input to DesignXplorer, tracking several time points (Figure 4).
Digital Twin Validation
Because the team couldn’t physically record data from the mining process in operation, it ran another Rocky DEM simulation on points that weren’t included in the ROM models, then compared the two.
This Rocky DEM analysis functioned as a stand-in for on-site-based measurements. The validation case involved two 20-ton shuttle cars at a conveyor speed of 0.35 m/s; feeder conveyor speed was 0.46 m/s.
Since it was run in Rocky, the data was accurate for that given point (and not an estimate).
Comparing Rocky DEM results to the Twin Builder simulation, the validation results matched very closely (Figure 5) — indicating that it was prudent to use this technique to build reduced-order models as well as to predict accurate on-site results.
The benefits of the digital twin far outweigh the effort involved. Possibilities include:
Link to the IOT platform to predict operations and avoid bad outcomes
Use ROM to immediately see effects of variables on a system, in an effort to make better-informed changes
Modify ROM to include calculation of spilled mass
Use ROM to optimize the system
Embed the ROM onto a controller (PLC), which can then automate conveyor speeds
As you can imagine, it holds great potential!
Senior FEA and DEM engineer at Qfinsoft, South Africa
Wynand Prinsloo provides technical support to clients who use Rocky DEM and ANSYS Mechanical software. This entails training and supporting current customers, providing pre-sales support to the company‘s sales team, and performing consulting work for clients when needed. Qfinsoft focuses on staying on top of new technologies and features made available in Rocky DEM and ANSYS software, bringing these technologies to the local engineering and product development sector.