12.17. Node Based Averaging

Various DPM parcel quantities can affect the continuous fluid flow by being integrated into the flow solver equations. By default, Ansys Fluent applies the effects of a DPM parcel only to the fluid cell that contains the parcel. As an alternative, mesh node averaging can be used to distribute the parcel’s effects to neighboring mesh nodes. This reduces the grid dependency of DPM simulations, since the parcel’s effects on the flow solver are distributed more smoothly across neighboring cells.

A variable is averaged on a mesh node using the following equation:

(12–515)

Here is the particle variable, is the number in the parcel, and is the accumulation of the particle variable on the node for all parcels . The function is a weighting function, or “kernel.”

The following kernels are available in Ansys Fluent:

  • nodes-per-cell

    (12–516)

  • shortest-distance

    (12–517)

  • inverse-distance

    (12–518)

  • gaussian

    (12–519)

    where is a characteristic length scale of the cell containing the parcel.

    For the Gaussian kernel, an additional parameter is necessary to control the width of the Gaussian distribution. Kaufmann et al. [294] had good success with a small width and using , while Apte et al. [25] used a broad distribution and a value of . The default value is .

    For information on using node based averaging, refer to Node Based Averaging of Particle Data in the User’s Guide.