Diffusers: Characterization for CFD

Diffusers are common in many AEC applications. In large spaces within a building, there may be so many diffusers that the simulation becomes impractical due to the amount of small details. The process of characterizing a typical diffuser will be described in detail here.

A diffuser for an HVAC supply can be comprised of a lot of very small details (e.g., holes, slots, vanes) with respect to the room that it is supplying, which makes it a good candidate for simplification through a more generic characterization.


The diffuser above has many small air gaps and thin metal vanes that would require a small element size to capture accurately.  When used in a large room, explicitly modeling this diffuser would add a lot of local detail to the simulation and increase the solver time.

Identify Goals of Characterization

At the supply from the HVAC system, the air has already been conditioned and is entering the space at a known flow rate and temperature.  The purpose of the diffuser is to distribute the air by adjusting the throw pattern, so it has a significant impact on the flow fields around it.

Defining Characterization Method

Knowing that the flow is an inlet to the system, the diffuser can be characterized using simpler geometry with boundary conditions.  The angle (i.e. vector) of the incoming flow will be controlled using component velocity boundary conditions.


Changing the Velocity Method from Normal to Component provides access to individual velocity directions.

Gathering the Necessary Inputs

If diffuser performance specifications are available, hand calculations can be used to capture the component velocities given some of the following information:

  • Angle of flow leaving diffuser
  • Incoming flow rate
  • Area that the flow will be coming from

Vx and Vy are the input velocity components to produce the resultant vector, Vr

Vx and Vy are the input velocity components to produce the resultant vector, Vr

For example, assuming a flow rate of 100 cfm, an area of 2 square feet and a throw angle of 30 degrees from horizontal:

  • Vy = Q/A = 100/2 = 50 fpm
  • Vx = Vy/tan(theta) = 50/tan(30) = 86.6 fpm

If specifications are not available, a simple test model, with the diffuser explicitly modeled can be simulated.  The results from the simulation are then used to determine the inputs needed for the diffuser characterization.


A simple test model with the diffuser explicitly modeled.

The preferred method of obtaining the component velocities is to position a Plane in the middle of the diffuser and then use an XY-Plot to see the velocities.

The black dashed line (top) shows the location of the XY-Plot path.  Notice that it is placed where the velocity averages out away from the diffuser fins.  The component velocities can then be interrogated (middle and bottom) using the plot.

The black dashed line (top) shows the location of the XY-Plot path.  Notice that it is placed where the velocity averages out away from the diffuser fins.  The component velocities can then be interrogated (middle and bottom) using the plot.  Vx has an average of about 31 ft/min (positive for one side negative for the other) and Vy has an average of 78 ft/min (bottom).


Validation

A simple validation study comparing the actual and characterized versions of the diffuser model can be used to verify that the inputs and assumptions are correct.  Note that the characterized version will rarely match the actual version with exact precision; the objective here is to only get a reasonably accurate representation.

Small extrusions should be used to bring the diffuser inlet just off of the primary air domain a short distance, to help the solver resolve the mass flow balance. The velocity component boundary conditions are placed at the top of these extrusions.

Small extrusions should be used to bring the diffuser inlet just off of the primary air domain a short distance, to help the solver resolve the mass flow balance. The velocity component boundary conditions are placed at the top of these extrusions.  Note that the left extrusion will have a negative horizontal velocity component while the right extrusion will be positive, assuming that the positive global direction is to the right.


When the characterization technique is used correctly, a direct comparison of the two versions should reveal similar flow velocity profiles.

The velocity profiles from the fully detailed diffuser (top) are very similar to those of characterized version (bottom) only a short distance away from the diffuser.

The velocity profiles from the fully detailed diffuser (top) are very similar to those of characterized version (bottom) only a short distance away from the diffuser.

Mesh from a detailed diffuser model.

Mesh from a detailed diffuser model. 


Comparing the mesh between the detailed diffuser (top) and the characterized diffuser (bottom) reveals how the characterized model uses less than half of the elements.  This has a significant impact for AEC spaces that have many diffusers.