Structural and Flow Characteristics of Panax notoginseng Xylem

The detailed structural parameters of the Panax notoginseng xylem were shown in the scanning electron microscopy images. The flow resistance characteristics of the vessel were analyzed by the computational fluid dynamics method [11 (link), 18 (link)]. Based on the microscopic images of the cross-section and axial-section, the structural parameters of the annular thickening and pitted thickening vessel were measured. The measurements were taken for each character listed in Table 1.
The types of vessel cross-sections were shown in Fig 2. The terms of the annular thickening and pitted thickening vessel were shown in Fig 3, which R was inscribed circle diameter, W was width, S was spacing, H was height, L was length.
The actual structural parameters of the xylem were obtained from the Panax notoginseng samples (Table 1), and the computational domain model of the annular thickening and pitted thickening vessel was established based in SolidWorks. The hexagon vessels of Panax notoginseng were shown in Fig 4.
The computational domain models contained a flow area with a secondary wall thickening pattern of 250 μm in length. To avoid effects at the entrance and exit, an extended smooth segment with length 25 μm was added at both ends of the vessel (Fig 5).
The boundary conditions were that the pressure was zero at the model outlet, and the flow velocity was 0.3 mm/s at the model inlet. The irregularity of the vessel structure was generated by tetrahedral and hexahedral unstructured meshes. The maximum and minimum of the unit size were 4.4×10^-6m and 4.4×10^-8m, respectively. In this part, the scale of the mesh was based on the prediction accuracy of the inlet/outlet pressure drop, and the mesh size independence test was performed (Table 2). The pressure drop difference between the standard mesh and the fine mesh was 0.22%. The mesh number has no effect on the calculation results, so the standard mesh number was used to analyze flow resistance characteristics, and the total number of meshes in the model was approximately 728680 (Fig 6). The PowerCube-S01 with a high-performance computing system was used for the simulation.