Page |1

AE-706
ASSIGNMENT REPORT

Grid generation using

Elliptic PDE, Method
By
Patel Jaydip L.

Guided By
Prof. J.C. Mandal

INDIAN INSTITUTE OF TECHNOLOGY

BOMBAY

24M0004
24M0004@[Link]
 Page |2

Contents
1. Flow chart.......................................................................................................................................3
2. Introduction.....................................................................................................................................4
3. Transfinite Interpolation Method [1]...............................................................................................4
4. Elliptic Grid Generation..................................................................................................................7
5. Conclusion.....................................................................................................................................11
6. References.....................................................................................................................................11

24M0004
24M0004@[Link]
 Page |3

1. Flow chart

24M0004
24M0004@[Link]
 Page |4

2. Introduction
This assignment aims to implement the Elliptic Grid method to generate a structured computational
grid for a given physical domain. The domain consists of an Airfoil surrounded by a far-field boundary
located at a radius R. The grid mapping should transform this physical domain into a unit square in
the computational domain.
The initial step is to generate the Airfoil. The Airfoil generated is NACA 6-digit series and its
geometry is obtained using the coordinates downloaded from Airfoil tools website, then after the
smoothening of the Airfoil is obtained using the Cubic-spine.
The TFI method is applied over the generated Airfoil, and the structure grid is obtained, the output of
which is used as the initial data for the Elliptic Grid Generation Method.

3. Transfinite Interpolation Method [1]

TFI is a technique to generate a structured grid by mapping a simple computational domain onto a
more complex physical domain. TFI is an algebraic grid generation method that interpolates between
boundary curves to construct interior grid points. It provides a smooth transformation between a
computational plane (ξ, η) and a physical plane (x, y) without solving differential equations (unlike
PDE-based methods).
TFI works by blending multiple 1D interpolations along the edges of a domain to construct a smooth
2D or 3D grid. It is particularly useful when:
The computational domain is a simple square/rectangular space.
The physical domain has complex boundary shapes like an Ellipse, an Airfoil inside a circle.
A structured grid is needed without requiring a high computational cost.
The airfoil is NACA 63-412, and it is inside the rectangular domain of size 20*20 as shown in below
figure:

Figure 2.1 Physical and Computational Domain

24M0004
24M0004@[Link]
 Page |5

function 𝐹 (ξ, η).

To generate the grid in physical plane one must use computational domain by applying mapping

Linear Interpolation along ξ:

P ξ F [(η)]=(1−ξ)F (0 ,η)+ξF (1 , η)

Linear Interpolation along η:

P η F [(ξ)]=(1−η) F (ξ , 0)+η F (ξ , 1)
; (1−ξ),(1−η), ξ , η = Univariate Functions

The above general function contains the interpolation for x and y (for 2D), i.e., P ξ F [(η)] has
X ξ F [(η)] and Y ξ F [(η)] similarly for P η F [( ξ)] has X η F [(ξ)] and Y η F[(ξ)].
The quadrilateral domain has 4 boundary curves, i.e., the Bottom (at η = 0), Top (at η = 1), Left
(at ξ = 0), and Right (at ξ = 1).
The simple addition of P ξ F [(η)] and P η F [( ξ)] results in a double count of four corner
points. To correct this, the product projection term is being subtracted from that addition.
Product Projection Term

P ξ P η(F )=(1−ξ)(1−η)F (0 , 0)+ξ η F (1 ,1)

+ξ (1−η)F (1 ,0)
+η (1−ξ )F (0 , 1)
For this given problem, the Boundary Conditions will be:

Bottom, i.e., F [(ξ, 0)]:

F [ξ , 0]=
[ ]
x airfoil
y airfoil

x airfoil
; =Coordinates of Airfoil Geometry
y airfoil

Top, i.e., F [(ξ ,1)]:

F [ξ , 1]= [ Conditions ¿create Domain ]
This loop has divided the rectangle in five equal parts and for each case the boundary condition is applied
in computational plane to map in physical plane.

Left, i.e., 𝐹 [(0, η)]:

F [0 ,η ]=
[ ( 1−η ) 1+ η∗10
( 1−η ) 0+η∗0 ]
Interpolating for line AD

Right, i.e., 𝐹 [(1, η)]:

F [0 ,η ]=
[ ( 1−η ) 1+ η∗10
( 1−η ) 0+η∗0 ]
24M0004
24M0004@[Link]
 Page |6

Interpolating for line BC

Now the coordinates for the physical geometry can be written as:

F [0 ,0 ]= 1
0 []
F [1 , 0]=
[ 10]
F [ 0 , 1]=
[100]
F [1 , 1]=
[ 100]
Thus, the final expression can be written as:

𝐹[(ξ,η)]¿ ( 1 – ξ ) F
[ ] [
( 1 – η ) X A +η∗X D
( 1 – η ) Y A +η∗Y D
+ξF
( 1 – η ) X B +η∗X C
( 1 – η ) Y B +η∗Y C ] x
[ ]
+ ( 1 – η ) F airfoil +¿
y airfoil

η [ Conditions¿ create Domain ] −( 1−ξ )∗( 1−η )

[10 ]−ξ∗( 1−η) [ 10 ]−η∗( 1−ξ) [ 100]−η∗ξ [100 ]

24M0004
24M0004@[Link]
 Page |7

Figure 2.2 Grid in physical plane – TFI

Figure 2.3 Grid in physical plane – TFI-zoomed

4. Elliptic Grid Generation

This method uses the following elliptic PDEs, which can be solved by any iterative techniques like
Point Gauss-Seidel or line Gauss-Seidel.
ξ XX +ξ YY =0

η XX + ηYY =0

; η∧ξ arethe coordinates∈the computational domain

To transform the elliptic PDEs, the dependent and independent variables are interchanged. The
mathematical expressions employed for this purpose are
aX ξξ−2 b X ξη +C X ηη =0

aY ξξ−2 b Y ξ η +C Y ηη =0

2 2 2 2
; a=X η +Y η , b=X ξ X η +Y ξ Y η , c=X ξ +Y ξ

24M0004
24M0004@[Link]
 Page |8

The X ξ , X η, Y ξ ,Y η are obtained by applying the gradient of a respective variable over the output of
the TFI grid generation.
After getting the coefficients a, b, and c, the iterative method is applied:

1
F ( i , j )= ∗{ p∗[ Fi +1 , j+ F i−1 , j ] + q∗[ Fi , j+ 1+ F i , j−1 ] −r∗[ F i+1 , j+ 1+ F i−1 , j−1 ] −[ F i+ 1, j−1 + F i−1 , j+ 1 ]}
denom

a
; p= 2
(∆ ξ)
c
; q= 2
(∆ η)
c
; r=
2∆η ∆ξ

;denom=2∗
[ a
2
+
c
(∆ ξ) (∆ η)
2
]
The iterative elliptic grid generator updates the X and Y coordinates at the left boundary using the
following algebraic expressions:

1
F ( 0 , j )= ∗{ p∗[ F 1 , j+ F−2, j ] +q∗[ F 0 , j+1 + F 0 , j−1 ]−r∗[ F 1 , j+1 + F1 , j−1 ]−[ F−2 , j−1+ F−2 , j+1 ]}
denom

The calculation is continued until the error reaches the predefined tolerance value.

24M0004
24M0004@[Link]
 Page |9

Figure 3.1 Grid in physical plane – Elliptic Grid Generation

Figure 3.2 Grid in physical plane – Elliptic Grid Generation-zoomed

Instead of rectangular domain if it is circular one then the results would be like shown below:

24M0004
24M0004@[Link]
 P a g e | 10

Figure 3.3 Grid in physical plane – TFI Grid Generation-Circular domain

Figure 3.4 Grid in physical plane – zoomed

TFI Grid Generation-Circular domain

24M0004
24M0004@[Link]
 P a g e | 11

Figure 3.5 Grid in physical plane –ELLIPTIC Grid Generation-Circular domain

Figure 3.6 Grid in physical plane – zoomed Elliptic grid Generation-Circular domain

5. Conclusion
From the obtained results following conclusion can be:
Elliptic Grid Generation produces a smooth, well-graded grid without abrupt changes in spacing.
Elliptic Grid Generation ensures near-orthogonal grids at boundaries, reducing numerical error.
Elliptical Grid Generation minimizes skewness and distortions, improving grid quality.

6. References
 [Link]
 Computational Fluid Dynamics; By T.J. Chung; ISBN 0 521 59416 2 hardback
 An Introduction to Computational Fluid Dynamics; By H K Versteeg and W Malalasekera; ISBN:
978-0-13-127498-3
Computational Fluid Dynamics Volume I , Klaus A. Hoffmann, Steve T. Chiang

24M0004
24M0004@[Link]