So, you’ve built a DEM model, but how do you know it’s any good? It is similar to commissioning a new instrument in the field. The manufacturer’s specifications may be correct, and the installation may be done perfectly, but until the instrument is calibrated against a known reference, its readings cannot be trusted. This article will walk you through how to validate your DEM model using real-world tests, making sure your simulations actually reflect what happens in your plant. We’ll look at why this step is so important and how you can get it done.
Key Takeaways
- Bulk material characterisation testing: Laboratory tests are used to establish baseline material properties that directly inform DEM inputs. These typically include bulk density, particle size distribution, angle of repose, shear strength, cohesion, and wall friction. DEM results can then be checked against these measured properties to confirm that the simulated material exhibits comparable flow and strength behaviour.
- Angle of repose and heap Formation tests: Simple heap or pile tests provide a clear visual and quantitative comparison between physical material behaviour and DEM predictions. The angle, shape, and stability of a formed pile are measured physically and then replicated in DEM. Close agreement indicates that inter-particle friction, rolling resistance, and cohesion parameters are appropriately calibrated.
- Conveyor or Chute Flow Trials: Short-scale conveyor, chute, or transfer point trials allow direct comparison of flow patterns, velocities, material trajectories, and build-up zones. Observations such as spillage locations, stream breakup, and impact zones can be compared to DEM outputs to validate flow behaviour and trajectory predictions under realistic operating conditions.
- Wear and Impact Measurement: Physical inspections of liners, chutes, and impact zones provide real-world evidence of wear patterns and severity. These observations can be correlated with DEM-predicted impact energy, contact frequency, and abrasion intensity. Consistency between measured wear and DEM wear indices builds confidence in liner life predictions and design decisions.
- Throughput and Residence Time Measurements: By measuring actual mass flow rates, fill levels, and residence times in bins, feeders, or transfer points, engineers can compare system performance against DEM results. Discrepancies highlight issues such as incorrect material stiffness, damping, or boundary conditions, enabling targeted refinement of the model.
Understanding The Need To Validate DEM Model
Developing a Discrete Element Method (DEM) model requires careful selection of particle representations, contact models, material properties, and system geometry. While these inputs may be based on sound engineering judgement and published data, they do not, on their own, confirm that the model accurately represents reality. Validation is the step that establishes confidence. Without it, simulation results remain theoretical and may not translate reliably to real-world operation. A DEM model must be validated before its predictions can be used to inform design or operational decisions.
Why Physical Testing Is Crucial For DEM Accuracy
A DEM model is, at its core, a numerical representation of a physical process. If that representation does not capture actual material behaviour, the simulation outputs will be misleading. Physical testing provides the reference data needed to assess model accuracy. By comparing DEM results with measured behaviour, such as flow patterns, discharge rates, or blockage tendencies, it becomes possible to identify where assumptions or parameters are incorrect. This process is not about forcing agreement in isolated results, but about ensuring the model responds to conditions in the same way the real material does.
Bridging The Gap Between Simulation And Reality
All DEM models rely on simplifications and assumptions regarding particle shape, friction, stiffness, and interaction behaviour. These simplifications can introduce discrepancies between simulated and real systems. Physical testing provides the mechanism to close this gap by enabling direct comparison under controlled conditions. Observations such as flow rates, material trajectories, packing behaviour, and impact response can be evaluated against DEM predictions, allowing targeted adjustment of model parameters until realistic behaviour is achieved.
Ensuring Reliable Predictions Through Validation
The primary value of a DEM model lies in its predictive capability. Whether applied to bulk materials handling, mining, agriculture, or process industries, decision-makers must be confident that the model reflects actual system behaviour. Systematic validation through physical testing allows errors to be identified and corrected through an iterative process of testing, comparison, and refinement. This approach results in a robust model that can be relied upon for design optimisation, risk reduction, and performance improvement. Without validation, DEM outputs carry significant uncertainty, increasing the risk of costly design errors and operational failures.
Designing Physical Tests To Validate DEM Model
Conducting Meaningful Drop Tests
Drop tests are a practical and widely used method for validating particle behaviour in DEM models, particularly for transfer points, chutes, and hoppers. These tests aim to replicate how material behaves during free fall and impact. Key parameters such as drop height, impact angle, surface material, and confinement must be matched as closely as possible to the simulation.
During testing, measurements should focus on particle dispersion after impact, rebound behaviour, and final resting profiles. High-speed video or repeatable photographic methods can assist with capturing this data. Comparing these observations with DEM predictions provides a clear indication of whether particle stiffness, damping, and friction parameters are appropriately defined, establishing a baseline for model credibility.
Comparing Particle Size Distributions
Particle size and shape have a big impact on how bulk solids flow. For your DEM calibration experiments, it’s important to accurately characterise the particles you’re using in your physical tests. This means measuring their size distribution. You can use sieving for granular materials or laser diffraction for finer powders. Once you have this data from your physical tests, you can compare it directly to the particle size distribution you’ve defined in your DEM model. Are they similar? If not, this is a clear area where your model might need adjustment. Small differences here can lead to quite different flow behaviours in your simulations, so paying attention to this detail is important for accurate testing calibration DEM results.
Observing and Matching Flow Patterns
Beyond just numbers, you need to watch how the particles actually move. Set up your physical tests so you can clearly see the flow patterns. Are the particles forming stable flow channels? Is there any bridging or ratholing occurring in your hopper? Record these observations. Then, run your DEM simulation under the same conditions and compare the visual flow patterns. Does your simulation show similar channel formation? Does it predict bridging where you observed it physically? This qualitative comparison is just as important as the quantitative data. It helps you understand if your model is capturing the complex interactions between particles that dictate bulk solid behaviour. Getting these flow patterns to match is a big step towards a reliable DEM model.
Integrating Physical Feedback Into Simulation Iterations
So, you’ve done your physical tests and gathered all that data. Now what? This is where the real magic happens – feeding that real-world information back into your DEM model. It’s not just about running simulations; it’s about making them work like the real thing. Your simulation is only as accurate as the test work information you add.
Calibrating DEM Parameters With Test Data
This is the core of it. Your physical tests give you actual numbers for things like how particles bounce off each other or how they slide. You can take these results and adjust the parameters within your DEM software. For instance, if your drop tests show particles breaking more than your simulation predicts, you might need to tweak the material’s fracture properties or the impact energy thresholds. It’s a bit of trial and error, but you’re guided by actual measurements, not just guesswork.
Refining Particle Properties For Accuracy
Beyond just the basic parameters, you might need to get more granular. Are your particles perfectly spherical, or do they have irregular shapes? Does their surface roughness vary? Physical testing can reveal these details. If your flow patterns in the test hopper don’t match the simulation, it could be because the particle shape or surface friction in the model isn’t quite right. You might need to adjust the shape factors or the coefficient of friction based on what you observed. Getting these particle-level details correct makes a big difference to the overall bulk behaviour.
Iterative Improvement For Robust Models
This isn’t a one-and-done deal. You’ll likely go through several cycles of testing, adjusting, and re-simulating. Each iteration gets you closer to a model that accurately reflects reality. You might run a simulation with your updated parameters, compare the results to another physical test (maybe a different type of flow or a different container), and then make further adjustments. This iterative process builds a more robust and reliable DEM model over time. It’s about building confidence in your simulation results because you know they’re grounded in physical reality.
Leveraging Real-World Data To Validate DEM Model
Case Studies: DEM Validation of a Conveyor Transfer Chute
A bulk materials handling operation experienced persistent spillage and accelerated liner wear at a critical conveyor transfer point. A DEM model was developed to assess material trajectories, impact zones, and wear mechanisms within the chute. While the initial simulation provided useful insights, validation was required before design changes could be implemented.
Physical testing was carried out using site measurements and targeted inspections. Conveyor belt speed, feed rate, and material properties were recorded, and wear patterns on existing liners were mapped during shutdown inspections. High-wear zones, impact locations, and material build-up areas were documented and used as reference data.
These observations were then compared with DEM outputs, including predicted impact energy, contact frequency, and flow paths. Model parameters such as particle stiffness, restitution, and wall friction were iteratively adjusted until the simulated impact and wear distributions aligned with the physical evidence observed on site.
Once validated, the DEM model was used to evaluate alternative chute geometries and liner configurations. The final design reduced spillage, improved flow, and significantly extended liner life. Because the model had been validated against real-world behaviour, the client was able to implement the design changes with confidence, avoiding trial-and-error modifications and unplanned downtime.
Learning From Calibration Successes
When you manage to correlate DEM with site data, you learn a lot. It’s not just about tweaking numbers; it’s about understanding the behaviour of your material. For example, if your simulations consistently show material bridging in the bin when the real-world data says it flows freely, you might need to look at your particle shape or cohesion settings. Perhaps the real ore particles are more rounded than you initially modelled, or the moisture content is affecting cohesion differently. Each successful calibration builds your knowledge base for future simulations. It’s about building trust in your digital twin.
Achieving High-Fidelity DEM Simulations
Getting your DEM model to accurately reflect real-world behaviour is the ultimate goal. It means your simulations can be relied upon for making important decisions. Think about designing chutes for bulk materials. If your DEM model, validated against physical tests and site data, shows that a certain chute angle causes excessive wear or blockages, you can trust that prediction. This saves time and money by avoiding problems before they happen in the field. It’s about making your simulations so good they’re practically a crystal ball for your material handling processes.