Methodology

Basic CFD Approach

AeroCloud uses a type of CFD known as the finite volume method which involves breaking up the space around your model into small volumes called cells. Fluid flow equations are then solved across all of these cells to predict what the fluid will do when it interacts with your geometry.

The solution process is iterative. It starts from an initial condition & works its way towards the final flow field in small steps. At each step, it takes the current solution & uses it to predict where the flow field needs to change.

After many iterations (& lots of equation solving) the flow field will approach a steady state, which we use as the basis for your results.

Quality Levels

The number of cells in your simulation is determined by the quality level you select.

• Basic simulations contain approximately 5 million cells;
• Standard simulations contain approximately 15 million cells;
• Pro simulations contain approximately 30 million cells;

In general, as cell count increases, so does the level of detail captured in the results.

Pro simulations also use a higher-fidelity method to capture the flow near the surfaces of your model & should give a better prediction of flow separation.

As the cell count increases, so does the time taken to generate the solution & the amount of compute resources required to complete the calculations. As such, basic simulations are the cheapest, followed by standard & finally pro.

Recommended Use Cases

Whilst you’re free to select any quality level for any simulation, here are our recommended use cases:

• Basic simulations are best suited to early-stage design studies where the level of detail in your models is low & you just want an indication of their relative aerodynamic performance.
• Standard simulations will capture more detail than basic (but less than pro) & are a good choice for product development projects & more detailed design studies.
• Pro simulations are best suited to late-stage product evaluation & sign-off when you need to capture smaller geometric features &/or more detailed flow fields. They’re especially appropriate for models that feature (or are defined by) regions of separated flow.

Comparing simulations across quality levels can be difficult as they resolve different amounts of detail & capture different flow features.

If you’re looking to make direct comparison between two simulations we suggest choosing the same quality level for each.

Cost

You can find out more about the cost of simulations, along with an explanation of the credit system here

CFD details

All AeroCloud simulations are:

• steady-state - we don’t capture how the flow field changes with time;
• isothermal - we don’t simulate the effect of temperature changes within the fluid;
• incompressible - the fluid density is constant throughout the simulation;
• fully-turbulent - the fluid is considered to be turbulent throughout the model. Laminar flow is not modelled, nor is the transition from laminar to turbulent flow.

This makes AeroCloud ideally suited for the simulation of low-speed external aerodynamics, such as those encountered by cars, buses, trucks, trains, cyclists, drones, athletes & many more.

Porosity

Porous media are modelled as an additional isotropic pressure drop within one or more zones defined by closed 3D-geometries.

The pressure drop per unit length is specified by the Darcy-Forchheimer equation:

$$ {\Delta p \over \Delta x} = D \mu U + F {1 \over 2} \rho U^2 $$

where D and F are the Darcy and Forchheimer coefficients, $\mu$ is dynamic viscosity, $\rho$ is fluid density and U is freestream velocity. The first term represents the viscous effects and the second term represents inertial effects. The coefficients can be derived from experiments or estimated from models available for specific types of porosity.

Experimental method: Pressure drop vs. velocity

From experiments, the velocity dependent pressure drop function can be found using a quadratic polynomial curve fit.

$$ \Delta p = A U + B U^2 $$

$$ D = {A \over \mu L} \qquad F= {2B \over \rho L}$$

where $ \Delta p $ is the pressure drop across a length L.

Troubleshooting

Many problems have their root cause in the input geometry, particularly around scale, orientation & holes.

You can mitigate many issues by:

Using the model preview window to ensure that your model is facing into the wind & is at the right scale before launching your simulation. This is a difficult one for us to detect automatically as we don’t know which way around or how big your model should be.

Trying to make your input geometry as simple as possible. this could include removing unnecessary detail (such as internal components & internal walls), making your components solid/watertight & plugging open holes. This all helps to ensure that our automated meshing process can spend the cell budget from your chosen quality level as efficiently as possible.

Adding cells inside your model (when you’re just interested in the external flow) reduces the number of cells that can be used to capture the interesting bits.