BET Disk

The BETDisks parameters can be challenging to configure, especially the nested lists of sectionalPolars.

BET Translators

To help generate these JSON entries, four translators are available to translate information from common propeller analysis data formats.

C81 to Flow360 JSON

The C81 2D airfoil data format is a common NASA format for storing airfoil data. This format only stores the airfoil’s 2D polars, so more information such as blade geometry, RPM, location in 3D space, etc. will need to be provided to the translator.

XFOIL to Flow360.json

XFOIL is one of the most common 2D airfoil analysis tools. If you have generated 2D polars with XFOIL, the translator will package them into the correct format. As for the C81 translator above, a variety of additional parameters will have to be provided in order to complete the required JSON entries.

XROTOR to Flow360.json

XROTOR is another common lower fidelity propeller analysis tool. XROTOR input files contain most of the information required to define a propeller, so few additional parameters are required for the translator.

DFDC to Flow360.json

DFDC is very similar to XROTOR except that it allows for ducted flows. The main difference is that while BET can easily simulate a virtual propeller, the duct(s) defined in DFDC will need to be generated in CAD, meshed and included in the simulation.

Note

Please contact Flexcompute customer support for access to any of the BET input data translators and examples. Tools to help check that BETDisks information is as expected are also available.

BET Sectional Polars

If you do decide to implement your own scripts to generate the BETDisks section of the Flow360.json input files, a number of aspects must be considered when generating the sectional polars for Flow360.json input. Firstly, continuity through the entire alpha range including at extreme angles (180 ° and -180 °) must be enforced to ensure numerical stability. Such high angles are not likely to occur at the end of the simulation, but may be encountered during the convergence history, especially near the hub region. Discontinuous forcing terms will increase the likelihood of solution divergence. To avoid this, the following setup is recommended:

• For each of the BETDisks add values for -180 ° at the beginning of the list and +180 ° at the end of the list

• For each of the sectional polars (number defined by number of radial stations within the sectional polar input) and each BET disk, set the lift (and drag) coefficients to equal values at the beginning and the end of the (inner-most) list (corresponding to alpha = -180 ° and +180 °). If not available, a value of zero can be used for lift and 1 for drag. See example polars Fig. 6.1.13 and Fig. 6.1.14 below

Fig. 6.1.13 Sample Cl and Cd polars for various mach numbers showing good continuity across the whole range of alphas = -180 ° to +180 °

Fig. 6.1.14 Same sample Cl and Cd polars as above but zoomed in on alphas = -45 ° to +45 ° for various Mach numbers.

The first index in the sectional polar nested lists corresponds to the MachNumbers, the second index corresponds to the ReynoldsNumbers, and the third index corresponds to the sectional alphas. If only a single value of Reynolds or Mach number is provided, the sectional coefficients will not vary with local Mach or Reynolds number. For multiple values given, the sectional coefficient (lift and drag) values are linearly interpolated according to the local alpha/Mach/Reynolds number in the solution field. If values are sparse in certain parts of the alpha/Mach/Reynolds space then the interpolation will not be accurate.

Another important aspect when generating the sectional polars is the post stall behavior of the lift and drag coefficients. As the lift and drag coefficients are also interpolated in a piecewise-linear fashion, high nonlinearity in these regions may cause numerical instability within the CFD simulation. The recommendation is to approximate these regions, by smoothing the sectional polar curves. This aspect will only become a consideration if post stall alphas are encountered within the BET region during the CFD simulation, which may occur throughout the convergence history.

The next consideration is the definition of the remaining BET required information, including the geometric data, rotational speed, etc. The available BET translators also process this information.

• The chord and twist distributions are typically obtained from the geometric information of the rotor. The chord affects the local lift and drag magnitude, whereas the twist affects the local angle of attack (which affects the local lift and drag).

• The radius associated with chords and twists do not need to coincide with the radial stations of the sectional polars, as the data will be linearly interpolated.

• It is recommended to add a point at a radius of zero with zero chord and 90 ° twist at the root to prevent a tip vortex from escaping there. For more information see this section.

See Fig. 6.1.15 and Fig. 6.1.16 below for an example of the twist and chord distributions of the XV15 propellers.

Fig. 6.1.15 Sample twist vs propeller span in mesh units

Fig. 6.1.16 Sample chord vs propeller span in mesh units