You are on page 1of 6

Analysis of Weapons Bay Pressure Distribution

Along the Walls Using Artificial Compressibility By

Approximating It As a 2-D Lid Driven Cavity
Shakeel Ahmed
Department of Aeronautics and Astronautics
Institute of Space Technology
Islamabad, Pakistan

AbstractThe acoustic environment of a weapons bay causes

large pressure oscillations. Many oscillations suppression
techniques have been applied with varying degrees of success in
the past. However, the understanding of why the pressure
oscillations exists or how the suppression methods work have not
been investigated as thoroughly. Advances in CFD permits
modelling of the bay environment to be performed and to reveal
details about the flow which is difficult to obtain from
experiments. In this paper, pressure distribution along the walls
of a new subsonic bomber aircraft has been analyzed in detail by
approximating the problem with a 2-D lid driven cavity flow.
Artificial Compressibility method was applied on Incompressible
Navier Stokes Equation to study the behavior of flow inside the
weapons bay. The code developed for this approach is first
validated for the well-known benchmark problem of
Computational Fluid Dynamics i.e. the 2-D lid driven cavity. The
problem was analyzed for a range of of Reynolds number and
grid independent studies was also performed. Results obtained
from the developed code match well with the benchmark results.
Finally the problem in hand of weapons bay pressure distribution
along the walls was solved with the validated algorithm. Results
obtained for the weapons bay have been discussed in detail while
identifying the location of maximum and minimum pressure
zones. Finally a recommendation is also made for optimally
placing the loads in the weapons bay.
Index Terms Artificial Compressibility, CFD, Weapons Bay,
2-D Lid Driven Cavity.

Over the years aircrafts have been developed with varying
methods of weapon carriage. It has been shown that external
weapons carriage may contribute as much as 30% of the total
drag of an air vehicle and in addition, substantially increase the
radar signature of the aircraft. Motivated by these
considerations, efforts to improve aircraft and stealth
technology have led to the desire to stow weapons in an
internal cavity-type bay. Unfortunately flight experience with
aircraft equipped with internal weapons bay has shown that
high speed flow over an open bay can produce a turbulent
environment within the bay that can damage the weapon(s) or
aircraft structure. Excessive structural vibrations may occur if
the acoustic frequency matches the characteristic structural

frequency of the cavity. Various studies have been conducted

to define this environment and means to improve or suppress
the aero-acoustic energy of the turbulent flow. In this paper,
pressure distribution along the walls of a new subsonic bomber
aircraft has been analyzed in detail. Due to the similar nature of
the weapons bay with an inverted 2-D lid driven cavity
problem, the problem in hand has been approximated by a 2-D
lid driven cavity flow. The solution to the incompressible
Navier Stokes Equations were found for the problem using the
Artificial Compressibility method. The code developed has
been tested for the well-known benchmark problem of
Computational Fluid Dynamics i.e. the 2-D lid driven cavity
and the results obtained have been compared with the results
by Ghia, Ghia and Shin. After validation of the developed
algorithm, it has been applied to solve the actual problem in
hand of finding the pressure distribution along the walls of the
internal weapons bay of a new bomber aircraft.
A. Governing Equations
The incompressible Navier Stokes Equations are the
governing equations to study the motion of fluids. The
equation for a 2-D case is given as below: u/t = -u2/x - uv/y - p/x + 1/Re(2u/x2 + 2v/y2) (1)
v/t = -v2/y - uv/x - p/y + 1/Re(2u/x2 + 2v/y2) (2)
u/x + v/y = 0
Artificial Compressibility methods, developed by A.
Chorin, provide a mechanism to march in pseudo-time towards
the divergence free velocity field such that the mass and
momentum are converved in the pseudo steady state. The
principle of this method lies in the introduction of an artificial
compressibility into the equations of motion in such a way that
the final results do not depend on this introduced factor. The
modified continuity equation after introduction of the artificial
compressibility factor is given as below: p/t + 1/(u/x + v/y) = 0


Where, in equation (4) is the Artificial Compressibility

parameter. It is equal to the inverse of square of speed of sound
a. The introduction of this factor and the p/t term in the
continuity equation has simplified the incompressibile Navier
Stokes equations to get the required values of pressure, the u
and v components of velocity. The final non-dimensionalized
incompressible Navier Stokes equations in pseudo-transient
forms are re-written as below: u/t = 1/Re(2u/x2 + 2v/y2) - u2/x - uv/y - p/x (5)
v/t = 1/Re(2u/x2 + 2v/y2) - v2/y - uv/x - p/y (6)
p/t + a2(u/x + v/y) = 0
In order to solve the equations (5), (6) and (7) to get the
values of pressure, u and v components of velocity, the
equations were discretized using 2nd order Forward Euler
Methods (Forward in Time and Center in Space). All three
equations obtained after discretization are given below: -



Fig. 2. v-Velocity Plot

C. Grid Independence Study
Grid independence study was carried out by using different
grid sizes for Reynolds number of 100. In order to compare
the values of u component and v component of velocity for
varrying grid sizes, the values of u component along the
vertical centerline at five equally spaced points were noted and
similarly the v component along the horizontal centerline at
five equally spaced points were noted. The values of u
component and v component are compared for different grid
sizes in table 1 and table 2 respectively.

Where equation (8) will give the u component of velocity,
equation (9) will give the v component of vecolity and equation
(10) will give the pressure. Next the method and the code
developed is validated by performing validation runs as
explained in coming paragraphs.
B. Plot of Velocity Components
In order to validate the results, the u velocity distribution
along the vertical centre line and v-velocity distribution along
the horizontal centre line were plotted for Re=100 on a 81x 81
uniform grid with a tolerance limit of 10-4. The plots obtained
were compared with available results in literature and were
found to be in complete harmony with them. The plots are
shown in figure 1 and figure 2.

Grid Size

U Component of Velocity
(0.5, 0)

(0.5, 0.25)

(0.5, 0.5)

(0.5, 0.75)

(0.5, 1)

21 x 21




41 x 41




61 x 61




81 x 81




101 x 101





Grid Size

V Component of Velocity
(0, 0.5)

(0.25, 0.5)

(0.5, 0.5)

(0.75, 0.5)

(1, 0.5)

21 x 21




41 x 41




61 x 61




81 x 81




101 x 101




It is clear from the comparison tables that the values are not
changing much after a certain grid sizes, i.e. the results for the
grid sizes are same for higher number of grid sizes.

Fig. 1. u-Velocity Plot

D. Contour Plots
The horizontal and vertical velocity contours, spanwise
vorticity contours and velocity vector plots are shown in the
figures below: -

From the figure 5, primary vorticity is clearly seen to be

forming the center of the cavity, whereas two small secondary
vortices are also visible at the bottom right and bottom left
cornors of the cavity. Velocity vectors are also in complete
agreement with the predicted results.
E. CPU Time vs Grid Size
CPU time vs grid size was studied for Reynolds number of
100 with a tolerence of 10-4 for the grid sizes used in the grid
independence study was also carried out. The CPU time
increases with increasing grid sizes.
Fig. 3. u-Velocity Contours

Fig. 7. CPU Time vs Grid Size

Fig. 4. v-Velocity Contours
It is clear from the velocity contours that u-component of
velocity is maximum at the top where the bourdary is also
moving with the flow, whereas, at the center of the cavity it is
having the negative maximum component. Similarly, the vcomponent of velocity is pointing downwards at the right
cornor of the cavity and pointing upward in the left cornor of
the cavity which are in complete harmony with the predicted
results. Contours of spanwise vorticity and velocity vector
plots are also given in figure 5 and figure 6 respectively.

F. CPU Time vs Tolerance

Similarly CPU time vs tolerance was also studies for
Reynolds number of 100 on a grid size of 81 x 81 with
varrying tolerence limits. The CPU time increases
exponentially with decreasing the tolerance limit. A semi-log
plot of CPU time vs tolerance is shown in figure 8.

Fig. 8. CPU Time vs Tolerance Limit

Fig. 5. Vorticity Contours

Fig. 6. Velocity Vector Plot

G. Flow Quality for Different Reynolds Number

Flow quality was studied with three different Reynolds
number (100, 400 and 1000) for a grid size of 81 x 81 and a
tolerance limit of 10-4. The contours of spanwise vorticity for
all the three different Renolds number are shown in figure 9,
10 and 11.

Fig. 9. Vorticity Contours for Re = 100

1000 on a 81 x 81 grid size with a tolerance limit of 10-4.

Results obtained for various parameters including contours of
u-component of velocity, v-component of velocity, spanwise
vorticity, vector plots are discussed in subsequent paragraphs.
A. Horizontal Velocity Contours
Horizontal velocity contours for the weapons bay is shown
in the figure below: -

Fig. 10. Vorticity Contours for Re = 400

Fig. 12. u-Velocity Contours

Fig. 11. Vorticity Contours for Re = 1000

Comparing the flow quality for the three different
Reynolds number, it is clear that the flow characteristics are
more prominent in the contour for Re =1000. Although all the
figures are showing one primary vortex however, as the
Reynolds number is increasing the primary vortices are getting
more refined in the center, and the secondary vortices at the
bottom right and bottom left cornors are also getting more

It is clear from figure 12 that the horizontal velocity is

maximum along the bottom of the weapons bay where the
boundary is moving along the flow. Minimum horizontal
velocity is somewhere near the center of the bay. At the
minimum horizontal velocity the negative sign of the velocity
suggests that it is actually moving in opposite direction to the
flow along the moving boundry.
B. Vertical Velocity Contours
Vertical velocity contours for the weapons bay is shown in
the figure below: -


Renolds Number

Minimum U velocity




(0.6375, 0.4875)



(0.6502, 0.4012)



(0.675, 0.3875)

The numerical scheme used and the code developed has
been validated by the comparisons and studies explained in the
numerical methodology section. The actual problem in hand
i.e. to study the pressure distribution along the walls of the
intenal weapons bay was analyzed with the validated numerical
code. The only difference between the 2-D lid driven cavity
problem analyzed before and the weapons bay is that the
moving wall is now along the bottom of the cavity instead of
the top as studied before and also the horizontal length of the
weapons bay under consideration is four times as compared to
the cavity used for validation of the code. The necessary
changes in the boundary conditions and the cavity size has
been made to get the results for the weapons bay. The analysis
of the weapons bay has been run with Reynolds number of

Fig. 13. v-Velocity Contours

The red colours towards the right bottom cornor of the
weapons bay suggests that the flow is moving upwards at that
location and the flow at the bottom left cornor is showing a
downward direction. This is an indication that the flow inside
the weapons bay is rotating in a conter clockwise direction.
This rotation of the flow will in turn form vortices and
turbulence inside the bay which will effect the location of
weapons deployment.
C. Vorticity Contours
Vorticity contours inside the weapons bay is shown in the
figure below: -


Fig. 14. Vorticity Contours

Contours of the vorticity in figure 14 suggests a strong
primary vortice in the center and two small secondary vortices
at the top right and top left cornor of the bay.
D. Pressure Distribution Contours
Pressure contours inside the weapons bay is shown in the
figure below: -

Fig. 15. Pressure Contours

It is clear from the pressure distribution shown in figure 15
that the maximum pressure is near the right bottom cornor of
the bay and roughly nearly uniform pressure distribution is
seen in the rest of the bay. Therefore, it may be concluded that
the weapons may be deployed in the center and towards the
forward boundary of the weapons bay avoiding the maximum
pressure zones.
E. Velocity Vector Plot
The velocity vector plot for the weapons bay under
examination is shown in the figure below: -

Fig. 16. Velocity Vector Plot

The velocity vector shown in figure 16 also suggests the
same location of the primary vortice at near the center of the
weapons bay. So the location of weapons bay suggested in
previous paragraph is also supported by the velocity vector

The numerical scheme used for analysis of the weapons bay

is Artificial Compressibility. The developed code was given a
series of validation run on the famous benchmark problem of
CFD i.e. 2-D lid driven cavity problem. Figure 1 and 2 shows
the u-component and v-component of velocity for the 2-D lid
driven cavity which were compared with the results available
in literature. Figure 3, 4 and 5 are the u-Before you begin to
format your paper, first write and save the content as a separate
text file. Keep your text and graphic files separate until after
the text has been formatted and styled. Do not u-velocity, vvelocity and vorticity contours. These figures suggests the
presence of a strong primary vortice in the center and two
secondary vortices at the bottom right and left cornors of the
cavity. CPU run times againts different grid sizes and tolerance
limtis are plotted in figures no 7 and 8 respectively. Finally the
validation run was performed for Renolds number of 100, 400
and 1000 as shown in figures no 9, 10 and 11 respectively.
After completion of validation runs, the actual problem of
weapons bay was analyzed with this validated algorithm. The
u-velocity contours of the weapons bay shown in figure 12
suggests the maximum velocity along the bottom wall and
negative component near the center of the cavity where the
primary vortice is forming as shown in figure 14. Analyzing
the different velocity and vorticity contours shown in figure 12,
13 and 14 and pressure contours shown in figure 15, we can
deduce that pressure along the rear boundary wall is maximum.
The weapons therefore can be deployed in the center of the bay
and towards the forward boundary wall. If deployment of
weapons near the rear boundary and the bottom surface then
the unwanted structural vibrations which may cause damage to
the weapons and the aircraft itself may be avoided to a
considerable extent.
In this paper numerical solutions were obtained for laminar
flow inside weapons bay, which was approximated as an
inverted 2-D lid driven cavity. The study has enabled us to
identify the maximum and minimum pressure locations inside
the weapons bay. The weapons are recommended to be loaded
towards the front and center of the cavity as these are the
minimum pressure zones. By placing the weapons in the
recommended locations will help in avoiding unnecessary
structural vibrations which would result in better aircraft
manuverability and agility.
The study can further be extended by modelling the
weapons inside the weapons bay and then performing the CFD
analysis of the weapons bay again. In this way a more realistic
study of the weapons and aircraft integration may be carried
out and also their effect on aircraft structural vibration may be
studied through numerical simulations.
I would like to thank my classmate Mr. Afreen Abbas, who
helped me in understanding the code provided to us for the
final project. His support enabled me to better understand and

manipulate the Matlab code for performing the analysis on the

weapons bay pressure distribution problem.
[1] Jason Handerson, Investigation of cavity flow aerodynamics
using Computational Fluid Dynamics Thesis submitted to
Department of Aerospace Engineering, University of Glasgow,
August 2001.
[2] U. Ghia, K.N. Ghia, C.T. Shin, High Re solutions for
incompressible flow using the Navier Stoke equations and a
multi-grid method Journal of Computational Physics, vol. 48
issue 3, December, 1982, pp.387-411.
[3] Klaus A. Hoffmann, Steve T. Chiang, Computational Fluid
Dynamics vol. I, edition. IV, Engineering Education Systems,
Witicha, USA, August 2000.
[4] Alexandra Joel Chorin, A Numerical Method for Solving
Incompressible Viscous Flow Problems Journal of
Computational Physics, vol. 2, April 1967, pp.118-125.
[5] Charles-Henri Bruneau, Mazen Saad, The 2-D Lid Driven
Cavity Problem Re-visited Computers and Fluids, vol. 35 issue
3, 2006, pp. 326-348.
[6] W. D. McQuain, C. J. Ribbens, C. Y. Wang and L. T. Watson,
Steady Viscous Flow in a Trapezoidal Cavity Computers and
Fluids, vol. 23, 1994, pp. 613-626.

[7] Reyad Omari, CFD Simulations of Lid Driven Cavity Flow at

Moderate Reynolds Number Europeon Scientific Journal,
vol.9, May, 2003, pp.22-35.
[8] A. J. Chumkha, Hydromagnetic Combined Convection Flow in
a Vertical Lid Driven Cavity with Internal Heat Generation or
Absorption Num. Heat Transfer, Part A, vol. 41, 2002, pp. 529546.
[9] N. A. C. Sidik, S. M. R. Attarzadeh, An Accurate Numerical
Prediction of Solid Particle Fluid Flow in a Lid Driven Cavity
International Journal of Mechanics, vol. 5 issue 3, 2011, pp.
[10] G. K. Batchlelor, On Steady Laminar Flow with Closed
Streamlines at Large Reynolds Number Journal of Fluid
Mechanics, vol. 1, 1956, pp. 177-190.
[11] S. Chen, A Large Eddy Based Lattice Boltzmann Model for
Turbulent Flow Simulation Applied Mathematics and
Computation, 2009, pp. 591-595.
[12] A. Jignesh, P. Thaker, B. Jyotirmay Banerjee, Numerical
Simulation of Flow in Lid-Driven Cavity Using OpenFOAM
International Conference on Current Trends in Technology,
Institute of Technology, Nirma University, Ahmedabad, 03-08
December, 2011.