Calculation of the unit normal vector for wall shear stress in the lattice Boltzmann model
Introduction
Aortic aneurysm and dissection have high morbidity and mortality rates [1,2]. Their pathologies are thought to be related to vessel morphology, vessel mechanical properties, and hemodynamic forces [3]. Among the hemodynamic forces, wall shear stress (WSS) is a research emphasis. This stress affects the endothelium function, and a large WSS will induce the initiation of aortic aneurysm and dissection [4], [5], [6], [7]. The in vivo measurement of WSS is relatively difficult owing to the complexity of vessel geometry and the disturbance of other organs [8,9]. However, the distribution of WSS can be obtained by computational fluid dynamics (CFD) [10], [11], [12].
In traditional CFD methods such as the finite element and the finite volume methods, mesh generation for a given computational geometry is a time-consuming task [8,[13], [14], [15]. In the lattice Boltzmann method (LBM), the computational geometry is discretized to equal size voxels; thus, mesh generation can be completed entirely automatically. Easy implementation of the bounce-back boundary condition allows LBM to handle complex geometries, and the parallel nature of LBM makes it possible to solve flow problems at relatively large scales. Besides, WSS in LBM can be computed locally and does require calculation of the velocity gradient.
In LBM, the main problem in the calculation of WSS is obtaining the unit normal vector of computational geometry. Stahl et al. recommended calculating the unit normal vector as the vector that is perpendicular to the local velocity near wall [16]; this method has been adopted widely [17], [18], [19]. However, the capacity of this method in turbulent flow is questionable; moreover, its calculation procedure is relatively complex and does not provide the direction of the unit normal vector [15]. To address these problems, Matyka et al. proposed a further method in which the unit normal vector is calculated using only computational geometry information and is ‘geometric normal’ [15].
Herein, we propose a method to calculate the unit normal vector using a lower number of input parameters and significantly reduced computational costs. In the following, a brief introduction to the LBM and WSS is given in Section 2. An introduction to the proposed method, together with its comparison with the geometric normal in simple geometries, is given in Section 3. Numerical simulations for a realistic aortic arch are given Section 4. Discussion and conclusions are given in Section 5.
Section snippets
Lattice Boltzmann method
The LBM is used to solve the discrete Boltzmann equation, which can be retrieved from the incompressible Navier–Stokes equations at a small Mach number. The time and space domains in the LBM model are discretized into equal time intervals and a regular mesh lattice with time step δt and space step δx. δt and δx are usually set as 1.0 in lattice units, which can be transformed to physical units. Herein, the physical variables are transferred in the lattice unit system if they are not in their
Unit normal vector
This section gives the proposed method for the unit normal vector and tests this method in a simple geometry.
Wall shear stress calculation
In this section, the accuracies of the geometric normal and STL normal are compared in a realistic hemodynamic geometry. The WSS distribution is also calculated.
Aortic aneurysms and dissections have high morbidity rates because of an abnormal WSS distribution, which is induced by complex geometries such as artery twists and branch vessels. Herein, a realistic geometry of an aortic arch is reconstructed from computed tomography medical images and the surface mesh is exported using Mimics
Discussion and conclusion
In hemodynamic research, WSS is critical to the diagnosis and/or cure of an aneurysm. In LBM, the calculation of WSS is very simple; however, it requires the unit normal vector of computational geometry. A method for the STL normal is proposed herein, the accuracy of which is compared with that of the geometric normal. In the simple geometries of a sphere and a donut, the geometric normal exhibits better accuracy as the extra input parameters are set to recommended values. The accuracy of the
CRediT authorship contribution statement
Li Min: Conceptualization, Methodology, Software, Writing - original draft. Huang Jingcong: Data curation. Zhang Yang: Supervision, Investigation, Funding acquisition, Writing - review & editing. Wang Yuan: Supervision. Wu Changsong: Formal analysis. Qu Lefeng: Resources.
Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
Acknowledgments
This work is funded by National Natural Science Foundation of China (11402190) and China Postdoctoral Science Foundation (2014M552443).
References (26)
- et al.
Biomech Model Mechanobiol
(2007) - et al.
Ann Thorac Surg
(2002) - et al.
Surg
(1998) - et al.
Comput Fluids
(2013) - et al.
Comput Fluids
(2010) - et al.
Comput Math Appl
(2016) - et al.
Comput Fluids
(2018) Comput Fluids
(2014)- et al.
Med Eng Phys
(2014) - et al.
Coll Surg Engl
(2002)