Volume 8, Issue 7, Juiy 2023 International Journal of Innovative Science and Research Technology

ISSN No:-2456-2165

Numerical Methodology of Mesh-based

Simulation Technique
Adamu Musa Mohammed1; Ali Musa Mohammed2; Mohammed Dahiru Buhari3; Abdulrahman A. Abdulrasheed4
1,4
Department of Chemical Engineering, Abubakar Tafawa Balewa University Bauchi, Bauchi State – Nigeria
2
Department of Electrical Engineering, University of Birmingham, Birmingham United Kingdom
3
Department of Electrical and Electronics Engineering, Abubakar Tafawa Balewa University Bauchi, Bauchi
State – Nigeria

Abstract— A clear methodology for mesh-based II. MODEL, GEOMETRY AND CALCULATION
simulation technique is presented in this study. The
principles from creating the model to creating geometry, A. Partial Differential Equations
mesh generation, calculation, and result analysis (post- In numerical studies, a problem is represented in form of
processing) were highlighted. Some of the results picked differential equations (partial differential equation). The
from previous works that employed the mesh-based Partial differential equation (PDE) is an equation that relates
technique were presented and that demonstrates how a function of several variables to its partial derivatives. Many
robust this technique is. It is obvious that today the use of problems in engineering and sciences are presented in form
computer simulations, in general, is paramount and that PDE, e.g., wave equation, mass transfer equation, heart
with computer simulations, different solutions can be transfer equation, Schrödinger equation etc. Mostly, physical
evaluated numerically, and only promising designs are phenomena are modelled by equations that relate several
sent to the laboratory or production line for scale-up partial derivatives of physical quantities such as momentum,
production. velocity, forces, temperature, etc [4]. Equation 1 and 2 shows
the example of first and second order partial differential
Keywords:- Models; simulation; geometry; mesh; mesh equation in three dimensions (3D).
generation; Elmer.
𝜕𝑢 𝜕𝑢 𝜕𝑢 𝜕𝑢
= 𝜕𝑥 + 𝜕𝑦 + 𝜕𝑧 (1)
𝜕𝑡
I. INTRODUCTION
𝜕2𝑢 𝜕2𝑢 𝜕2𝑢 𝜕2𝑢
Traditionally, computer simulations are carried out over = 𝜕𝑥 2 + 𝜕𝑦2 + 𝜕𝑧 2 (2)
𝜕𝑡 2
system domains made of meshed (grid) surfaces. According
to [1], “traditional modelling approaches make use of The typical way to solve PDEs is to discretize them
parametric patches, implicit surface, or subdivision surfaces since they constitute a large sparse of matrix problem.
that have been well integrated into 2D or 3D software”. The However, there are many ways of discretizing PDEs and the
procedure for mesh-based simulation includes building a simplest is finite difference approximations for partial
geometry, meshing and solution of the model equations using operators. The finite element method replaces the original
a given software. The system is first translated into a function (overall equation of the global domain) with a
geometry which is latter discretized into mesh of a given size function (sub-equations) that has some degree of smoothness
and finally the model, which is normally a differential but is piecewise polynomial on simple cells such as rectangle,
equation, is solved. After building the geometry the operation squares, or triangles. These triangles or rectangles are called
is usually carried out in three stages (mesh generation, the grid or mesh in 2D or 3D.
solution and postprocessing) and this, sometimes, require
three independent software to do the job. For example, in With numerical PDEs, the conditions at one or both
computational study of mass transfer at surfaces with reactive extremes need to be specified in other to set limit or
nanocones [2] the geometry and meshing were done by Gmsh boundary. Therefore, initial value and boundary value
software, whereas solution and postprocessing were carried problems are included as constraints equations in the solution
out by Elmer and Paraview, respectively. Each of the stages of PDEs. Boundary conditions are practically essential for
mentioned above is as important as the other and the error in defining computational problems because the resultant
one stage will significantly lead to an error in the other. quality of computations can critically be decided on how
However, some software (e.g. COMSOL) were developed in those boundary conditions are numerically treated [5].
such a way that it can handle all the three operations.
 Boundary value and initial value problems
When complex geometries were involved in a When a solution and derivatives value of a system of
quantitative study of a physical phenomenon, numerical differential equations (ordinary or partial) is specified at more
methods (such as finite element method) that discretize space than one point, it is said to be a boundary value problem
are often the best method [3]. The procedure for (BVP). The points at which these derivatives are specified are
discretization in Multiphysics is not much different with the called the boundaries, which they occur at the extreme. Since
numerical simulation that is based on partial differential differential equation represent problems in engineering, it is
equation discretization. very important to know the boundary condition for such
problems so as to proffer solutions. For instance, in mass
transfer problems on the surface of a block [2] , the initial
concentration at surface ‘A’ and the final concentration at the

IJISRT23JUL476 www.ijisrt.com 703

Volume 8, Issue 7, Juiy 2023 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165
opposite surface ‘B’ must be specify. This knowledge will B. Geometry and mesh generation
help in understanding what happens in-between at a A geometry of a computational domain represents a
particular time and the whole profile can be figured out. specific problem of interest by its physical features which
include all details of its shape, size, and dimension. The
In the order hand, the initial value problem (IVP) is a problem may be presented in two or three dimensions
differential equation along with a specified condition at initial depending on the precision requirement and the cost of
state or with an appropriate number of initial conditions. The simulation. After creating (building) the geometry, the next
IVP does not require to specify the value at boundaries rather step is geometric discretization, in other word, meshing or
it needs the values at initial condition. This usually apply for grid formation. Meshing is usually performed so as to divide
dynamic system that is changing over time. In summary, in the models into cells and the mesh is generated inform of
IVP the conditions at one extreme of the interval are known polygons or polyhedral that connect in a series of lines and
whereas in BVP the conditions at both extremes of the points. There are varieties of software for geometry and mesh
interval are known and that is why all PDEs are BVP and generation, and these includes open-sourced software such as
IVP. For instance, we need the initial position and velocity of Gmsh, NETGEN, GridPro, MeshGenC++, etc and
a particle in motion to be able to solve for its position under commercial software like ANSYS, Gambit, Grid Pro, CFD,
differential equation at time t. etc. The mesh generated could be structured or unstructured.
 Discretization
There are two major approaches to discretisation namely,
Langrangian and Eulerian. Where the former has it’s
coordinate and properties defined for every local field, the
latter has it’s coordinate and properties defined for the overall
system.

Fig. 2: Structured mesh (A) and unstructured mesh (B)

The structured meshes are meshes with implicit

connectivity whose structure allows for easy identification of
elements and nodes (Fig. 2A). In the other hand unstructured
meshes (Fig. 2B) has general connectivity with an arbitrary
structure and the connectivity of its elements must be defined
and stored [9]. Structured mesh requires less memory
compared to the unstructured mesh. The choice depends on
many factors such as memory requirement, numerical
Fig. 1: Structured mesh (A) and unstructured mesh (B) accuracy, ease of generation, computation time, among
others. Moreover, it is very difficult or impossible to compute
In Eulerian approach, discretisation is completed in a structure mesh for a complex geometric domain or a
fixed space and no movement of nodes and elements is geometry that requires high resolution mesh at a localized
required (Fig. 1A). This approach is suitable for simple region.
geometries and for fast calculation, it is however, not suitable
for high deformable and complex geometries. In the other  Gmsh
hand, the Langrangian approach has it’s nodes associated Gmsh is the software used for mesh generation in one of
with material particles thereby representing free surfaces the case studies reported in this thesis. Gmsh is a three-
(Fig. 1B), this helps to evaluate the deformation and define dimensional finite element grid generator, it has a built-in
the interfaces of different materials [6]. Therefore, CAD engine and post-processor. It is a user-friendly mesh
Langrangian approach can handle high deformation model tool that has an advanced visualisation capability (Gmsh
due to non-restriction of a closed domain [7]. manual). Basically, Gmsh is built around four modules:
geometry, mesh, solver and post-processing. In building a
There is also an approach called Arbitrary Langrangian- geometry the constituents are points, lines, surfaces and
Eulerian (ALE) that combines the two approaches and bodies. These constituents form a complete solid model
harnessed the advantage of both Langrangian mesh and the where lines are defined from points, surfaces from lines, and
Eulerian mesh. In this approach, the nodes in the mesh can be bodies from surfaces. The mesh generation is performed in
flexible and move as the material deforms like the similar way as the geometry: lines are discretized first; the
Langrangian mesh or fixed in a space and eliminate the issue mesh of the line is then used to mesh the surfaces; then the
of mesh distortion as in Eulerian mesh or it can move in some mesh of the surfaces is used to mesh the volumes. Fig. 3 is an
arbitrary manner [8]. This flexibility offered by ALE has a example of solid geometry (cube) with unstructured mesh.
great advantage of handling large distortion problems.

IJISRT23JUL476 www.ijisrt.com 704

Volume 8, Issue 7, Juiy 2023 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165
 ParaView
This a post-processing software to read the Elmer output
file in the format ‘vtu’. “It is an open-source, multi-platform
scientific data analysis and visualization tool that enables
analysis and visualisation of extremely large datasets” [12].
ParaView has a collection of libraries for various applications
including web visualization, in-situ analysis with catalyst and
scripting using python [13]. To provide backbone for
visualization and data processing ParaView uses the
Visualization Toolkit (VTK). The VTK supports various
visualization algorithms which includes vector, scalar,
tensor, texture and volumetric methods. It also includes mesh
Fig. 3: A cubic geometry with unstructured mesh smoothing, contouring, polygon reduction and Delaunay
triangulation [14].
Gmsh software is robust in building and meshing any
kind of geometry which can be exported to finite element
solver like Elmer for calculation and post-processing. Detail
guide on how Gmsh is use in building geometry and meshing
could be found in the user manual [10].

C. Calculation and post-processing

After building the geometry of a particular problem and
generating the mesh, the project file is then opened/imported
to a solver for calculations where the model has to be setup
and appropriate equation defined. In this work Elmer
Graphical User Interface solver (ElmerGUI) is used for the
solutions of the mesh-based approach problems and a
Visualization Toolkit ParaView is used for post-processing.

 Elmer
Elmer is an open-source finite element multiphysical
simulation software. The software has a graphical user
interface (ElmerGUI) which is capable of importing finite
Fig. 4: Para View application window
element mesh files in various formats [11]. It can also set up
PDE systems for solution and export the model data and In this work, ParaView 5.9.1 version was used, and the
results for Elmer Solver. The stages in running the simulation application window is as shown in Fig. 4. The mass transfer
with Elmer are: contours, von Mises stresses and displacement were analysis
 Model setup using ParaView. The files are in ‘vtu’ format for both
 Add equation structured and unstructured grids. After uploading, then
 Add material ‘apply’ button is pressed. Analysis of various properties of
 Add body force the model is done using different filters. Details on how
 Set initial conditions ParaView is used for visualization and post-processing can
 Set boundary conditions be found in [14], [12].

To setup the model, the simulation type that is either III. RESULTS
steady-state or unsteady-state need to be specified. Maximum
iteration, maximum output level, and time-stepping method The numerical methodology for mesh-based simulation
also need to be specified. In addition to the simulation is used to solve a variety of problems in engineering and
parameters, some constants, e.g gravity, vacuum permittivity, sciences yielding promising and near-perfect results [2],
Stefan Boltzmann, and unit charge, are being specified. There [15]–[18].
are handful of equations to be used depending on the type of
For instance, in [2], the mesh-based simulation was
problem one has. In our case, ‘model PDE’ and ‘linear
used to compute a mass transfer on surfaces structured with
elasticity’ equations were used for mass transfer and linear
reactive nanocones where current densities were computed
elastic problem respectively.
on the surfaces of the conical electrodes, greatly advancing
The idea behind choosing a material in the material the field of electrochemistry. In that work, Gmsh software
library is to use the specific properties of the material in was used to build and mesh the geometry of the conical
question e.g density, heat capacity, specific heat capacity etc electrode surface. However, the simulation is done using
for accurate computation and results. Initial and boundary ELMER and then post-processed with ParaView. Some of
conditions are set, and the simulation is run after saving the the results were shown in Fig. 5 and Fig. 6.
model. The result file is then open with a post-processing
software ParaView in our case.

IJISRT23JUL476 www.ijisrt.com 705

Volume 8, Issue 7, Juiy 2023 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165
as it affects the prospective results. Figure 8 shows the extent
of the mesh displacement at maximum angle of attack [16].

Fig. 5: (a); the computational domain and (b); conical

boundary section [2].

Fig. 8: Mesh displacement magnitude at root palne (top) and

tip plane (bottom) [16].

IV. CONCLUSION

In this article, the procedures and ideas of mesh-based

simulation are described. The basics of building geometry
and meshing with Gmsh (a finite element mesh generator),
computing with Elmer (a finite element method solver), and
analysing the results obtained with Paraview are described in
depth. Several of the outcomes of this methodology were
Fig. 6: Local current densities along conical face for (a) α = highlighted, and base on that, it is clear that the meh-based
5° and (b) α = 135° , S* = 1 and S* = 10 [2]. technique is reliable and effective for using numerical
computation to address a wide range of problems in
In another research [15] the mesh-based approach was engineering and the sciences.
used to compute a linear elasticity problem on a wooden bar
that is subjected to a load of 200N at one end and fixed at the REFERENCES
other end. The bar geometry was built and meshed, and the
stress and displacement were computed. Fig. 7 shows the [1.] Z. Ji, L. Liu, and Y. Wang, “B-Mesh: A Modeling
stress distribution using colour code. System for Base Meshes of 3D Articulated Shapes,”
Computer Graphics Forum, vol. 29, no. 7, pp. 2169–
2177, Sep. 2010, doi: 10.1111/j.1467-
8659.2010.01805.x.
[2.] A. M. Mohammed, M. Iqbal, N. V. Rees, and A.
Alexiadis, “Computational study of mass transfer at
surfaces structured with reactive nanocones,” Applied
Mathematical Modelling, vol. 74, pp. 373–386, Oct.
2019, doi: 10.1016/j.apm.2019.04.057.
[3.] T. Fischbacher and H. Fangohr, “Continuum multi-
physics modeling with scripting languages: the Nsim
simulation compiler prototype for classical field
Fig. 7: The von Mises stress on a wooden bar Using mesh- theory,” arXiv:0907.1587 [cond-mat,
base technique [15]. physics:physics], Jul. 2009, Accessed: Apr. 30, 2019.
[Online]. Available: http://arxiv.org/abs/0907.1587
Additionally, a mesh-based approach can be used to [4.] Y. Saad, Iterative Methods for Sparse Linear Systems,
compute flow issues with shifting boundaries and interfaces; Second. Society for Industrial and Applied
however, this approach must account for the distortion caused Mathematics, 2003. doi: 10.1137/1.9780898718003.
by the mesh's movement. Tonon et al., 2021 conducted [5.] S. Koshizuka, K. Shibata, M. Kondo, and T.
research on a mesh-moving technique without cycle-to-cycle Matsunaga, Moving particle semi-implicit method: a
accumulation of distortion and clearly demonstrated how to meshfree particle method for fluid dynamics. London,
measure the mesh distortion as well as the quality of the mesh

IJISRT23JUL476 www.ijisrt.com 706

Volume 8, Issue 7, Juiy 2023 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165
United Kingdom: Academic Press, an imprint of
Elsevier, 2018.
[6.] Z. Bi, Finite element analysis applications: a
systematic and practical approach. London
[England]: Academic Press, 2018.
[7.] J. J. Monaghan, “A turbulence model for Smoothed
Particle Hydrodynamics,” European Journal of
Mechanics - B/Fluids, vol. 30, no. 4, pp. 360–370, Jul.
2011, doi: 10.1016/j.euromechflu.2011.04.002.
[8.] J. P. Davim, Ed., Computational methods and
production engineering: research and development. in
Woodhead publishing reviews: mechanical
engineering series. Duxford, United Kingdom:
Elsevier, 2017.
[9.] M. Bern and P. Plassmann, “Mesh Generation,” in
Handbook of Computational Geometry, Elsevier,
2000, pp. 291–332. doi: 10.1016/B978-044482537-
7/50007-3.
[10.] C. Geuzaine and J.-F. Remacle, “Gmsh: A 3-D finite
element mesh generator with built-in pre- and post-
processing facilities,” International Journal for
Numerical Methods in Engineering, vol. 79, no. 11,
pp. 1309–1331, Sep. 2009, doi: 10.1002/nme.2579.
[11.] Lyly et al, “ELMER - A finite element solver for
multiphysics.,” 2000, 1999.
[12.] U. Ayachit, The ParaView guide: updated for
ParaView version 4.3, Full color version. Clifton Park,
NY: Kitware Inc, 2015.
[13.] P. O’Leary, J. Ahrens, S. Jourdain, S. Wittenburg, D.
H. Rogers, and M. Petersen, “Cinema image-based in
situ analysis and visualization of MPAS-ocean
simulations,” Parallel Computing, vol. 55, pp. 43–48,
Jul. 2016, doi: 10.1016/j.parco.2015.10.005.
[14.] C. D. Hansen and C. R. Johnson, Eds., The
visualization handbook. Amsterdam ; Boston:
Elsevier-Butterworth Heinemann, 2005.
[15.] A. M. Mohammed, U. D. Hamza, and Abdulrahman
A. Abdulrasheed, “Computation of Three-
Dimensional Linear Elasticity Problem using Mesh
and Mesh-free Particle Modelling Techniques,” Jun.
2023, doi: 10.5281/ZENODO.8042289.
[16.] P. Tonon, R. A. K. Sanches, K. Takizawa, and T. E.
Tezduyar, “A linear-elasticity-based mesh moving
method with no cycle-to-cycle accumulated
distortion,” Comput Mech, vol. 67, no. 2, pp. 413–434,
Feb. 2021, doi: 10.1007/s00466-020-01941-y.
[17.] D. Violeau and R. Issa, “Numerical modelling of
complex turbulent free-surface flows with the SPH
method: an overview,” Int. J. Numer. Meth. Fluids,
vol. 53, no. 2, pp. 277–304, Jan. 2007, doi:
10.1002/fld.1292.
[18.] D. Pearce, S. Fischer, F. Huda, and A. Vahdati,
“Applications of Computer Modeling and Simulation
in Cartilage Tissue Engineering,” Tissue Eng Regen
Med, vol. 17, no. 1, pp. 1–13, Feb. 2020, doi:
10.1007/s13770-019-00216-9.

IJISRT23JUL476 www.ijisrt.com 707