Modelling Brake Dust in EDEM

Carles
Carles
Altair Employee
edited February 12 in Altair HyperWorks

The environment is an increasingly important concern today and no economy is unaffected. Brake dust particle emission is becoming more and more important during brake development due to upcoming legislation and manufacturer responsibility to develop “green products”.

As braking is a complex process, depending on speed, vehicle weight, level of deceleration, as well as environmental conditions such as temperature and humidity which influence the type and nature of the particles emitted, ranging from very fine to coarse. Depending on their nature, these emissions can end up in the environment in the draining water or in the air we breathe. In the literature there are more and more studies about airborne particles; they all show that mainly emitted particles during braking have a distribution which varies with the braking conditions from nano to micro scale particles and are classified in different sizes. The smallest size classification of these particles is divided between PM10 (coarse dust) and PM2.5 (fine dust) which are capable of penetrating deep into the lungs.

Now regulations such as EURO-7 in EU and EMFAC-2021 in US force manufacturers to proof that the number of particles emitted while braking, are under certain limits. To ensure this, facilities are going to be used and the brake system will be tested under a reference load most likely the Worldwide harmonized Light-Vehicle Test Procedure (WLTP). In this procedure which take about 6 hours, there are about 300 brake events, and the overall pollution of the brake system will be measured. Such facilities will enable the measurement of a total emission for long term test cycle but will not give any inside of the behavior of the dust while emitted in a braking event within few seconds.

At Altair we have been working on a discrete element method solution to investigate how particles behave just after they have been emitted. Combined with computational fluid dynamics analysis, it is possible to create a virtual testing environment of a brake system. The aim of the simulation is to understand how the particles behave just after been emitted around the brake system. This capability enables a virtual trial-out of countermeasure and the investigation of different braking scenario. One first publication has been presented by our customer Mercedes Benz AG [ (Kartik, 2022)] a the last EuroBrake 2022. In this blogpost, we will summarize the numerical solution implemented as well as give an indication of the HPC infrastructure needed to compute this kind of task in an appropriate way.

Attached you can find a brake dust model setup in EDEM. The key steps to set it up are:

Step 1 – Materials

In general, EDEM materials should be calibrated. However, there are no clear validated procedures to adjust a dust material model in a DEM framework and in this example it has been decided to focus on adjusting particle size, density and shear modulus.

Although particle density has been defined as per the solid density values found in literature, Shear modulus and particle size have been adjusted in order to find a reasonable computational time while maintaining numerical stability. Hence, size and shear modulus adjustments may vary depending on the computational power available.

Step 2 – CAD

EDEM supports many file types such as STL, STEP, IGES. Geometries are imported to EDEM via the Creator > Geometries > Import Geometry section.  You can choose either the default mesh, which is optimized to capture the geometry shape while minimizing the mesh elements, or choose a manual mesh size. You may want to manually select the mesh if your analysis includes wear. Wear is plotted per triangular mesh element, as such the quality of the meshed surface influences the analysis.

Step 3 – Introduce Material

Starting simple, the user can prescribe the quantity, initial velocity and direction of the dust by utilizing a dynamic factory. It is key, as seen in the attached example, to define the factory following the perimeter of the brake pad. By doing that, realism on the dust origin is achieved.

Step 4 – Assign Kinematics

For that example, defining kinematics is quite simple as there is only the entire wheel rotation to be followed by the corresponding CAD bodies (disk, wheel rim and tire rubber).

Step 5 – Preparing the CFD model

In the specific case of brake wear particulate emissions, the amount of particles in a volume of fluid is always small compared to the amount of air. Hence, it can be considered that the influence of particles on the fluid is negligible. Therefore, in all cases of application, the principle of uni-directional coupling can be used, where only the influence of the velocity vector field is used to calculate the drag force.

Two different types of simulation can be performed with AcuSolveTM for this application. If we accept the assumption that the flow can be simplified with a steady-state calculation and that the turbulences can be represented as mean velocity flow.The advantage of this approach is of course the speed because the computation time is much shorter than in a transient approach. This allows to compute several scenarios very quickly (e.g. different speeds). In this case it is quite easy to export the velocity vector field in a format like CGNS and import it into EDEMTM.

The disadvantage is of course the simplification of the flow, because in reality the turbulences around the brake caliper are very present.

The second type of simulation is a transient calculation with a rotating disc, which implies a sliding mesh in the fluid side. The CFD model should be capable to resolve a problem with a moving rigid boundary. By using the one way coupling the velocity vector field can be transmitted to EDEMTM at given time. In the demo application a coupling between AcuSolveTM and EDEMTM has been used.

Step 6 – Running the simulation

For this co-simulation, the CFD part may run as an MPI-parallel simulation across multiple dual-CPU-socket compute servers, while the DEM calculation may run on a GPU-accelerated server. Reducing the time to communicate between parallel servers and between the solver steps is achieved by using a high-speed interconnect.

Step 7 – Analyzing results

Once the simulation is finished it is straight forward to visualize dust trajectories and locations of contact with the rest of the parts in the simulation. With this data, comparison between different driving speeds is already possible, enabling engineers to understand all the possibilities of dust distribution in the wheel casing for different route scenarios.

However, more advanced postprocessing is possible using the Analyst tools available in EDEM:

  • Relative Wear: If this model is selected for the simulation EDEM will display accumulated impact energy (normal and tangential) in each mesh element of the tire, wheel rim, disk and brake pad. This output allows for clear understanding of most frequent contact locations and pathways of the dust from the inside of the wheel casing to the . Wear calibration is unnecessary for the Relative Wear model.
  • Selections: In our case we use objects called Grid Bins that split the model domain into lattice. Each grid is known as a bin group and each cell is known as a bin. Anything in or moving through each bin can be monitored like the number of particles contained like show on the right picture in Fig. 5. In this case the bin is colored with the monitored value, which enable a visualization of the cloud of particles

For further information please see the following:

See how EDEM can be coupled to Altair Acusolve: https://learn.altair.com/enrol/index.php?id=157

Welcome!

It looks like you're new here. Sign in or register to get started.

Welcome!

It looks like you're new here. Sign in or register to get started.