# Computational Fluid Dynamics Technology

In multi-function machines, airflow is created through exhaust ducts like the one shown in Fig. 1, in order to cool down and collect dust inside the machine. In designing the airflow through such exhaust ducts, it is important to properly distribute the limited volume of air to each inlet. At Fuji Xerox, we utilize computational fluid dynamics for designing branch ducts and optimizing the volume of air suctioned in from each inlet of the exhaust duct.

Computational fluid dynamics employed in designing exhaust ducts requires highly accurate analysis. Besides correctly reflecting the cooling fan's characteristics and the exhaust duct's shape in the analysis model, and selecting the appropriate mesh size^{Note1} (a parameter for computational fluid dynamics) and mesh aspect ratio,^{Note2} it is important to set analysis conditions for simulating turbulent flow, such as separation phenomenon. This phenomenon occurs at a curved part of a duct where the duct shape changes considerably, such as part C in Fig. 1. It is a phenomenon where fluid motion becomes locally turbulent and a drift-like vortex occurs. If a separation phenomenon affects airflow in this way, analysis conditions for simulating turbulent flow must be set, with the shape of the duct's curved part being correctly reflected in the model.

- Note1
- Mesh size: A parameter to be examined when configuring an analysis model. It denotes the length of a side (normally several mm in length) when the analyzing space is divided into small elements (tetrahedrons or cuboids).
- Note2
- Aspect ratio: Ratio of the long side and short side of an element when dividing analyzing space into tetrahedron or cuboidal elements of sides several mm in length.

The following introduces a case of optimizing the air volume balance of the exhaust duct in Fig. 1 by using computational fluid dynamics. First, we created a computational fluid dynamics model of the exhaust duct that simulates the duct form, separation phenomenon, and other factors, and then checked that the physical size of the exhaust duct is accurately reflected in the model. Next, we used that model to analyze the volume of air and the air volume balance of inlets A and B. The result shown in Fig. 2 clarified that the volume of air suctioned in from inlet A greatly exceeded the target, and that suctioned in from inlet B failed to reach the target.

Therefore, we utilized computational fluid dynamics in investigating ways to improve the air volume balance of inlets A and B without increasing pressure loss inside the duct. The result confirmed the effectiveness of setting separation plate D inside the duct to balance the volume of air. By optimizing the form and position of separation plate D, we succeeded in balancing the volume of air suctioned in from inlets A and B, even though the total volume of air is limited (sum of volume of air suctioned in from inlets A and B), as shown in Fig. 3.