ENGG386 – Computational Methods in Engineering
Final Coursework Description
1 Marking Overview
For the final coursework in ENGG386 Computational Methods in Engineering,
worth 40% of the module, you are asked to investigate three test cases.
Please see the attached tables for your bespoke parameters.
Each test case is worth 30% and these 30% are split into:
10% for producing high-quality meshes and delivering the following files
through Canvas: mesh in the SU2 native format (.su2) together with the
Ansys ICEM CFD production files (.tin geometry and .blk blocking)
AND for setting up and running the requested simulations and delivering
a functioning configuration (.cfg) file through Canvas and in the report
10% for producing the report with all required information that is presented in
a professional way, including data post-processing and preparing high-
quality (i.e. publishable) illustrations that include reference to theory
5% for a mesh-refinement (at least coarse/medium/fine grids) and analysis
(e.g. spatial order of CFD code and Richardson extrapolation)
5% (cases 2 & 3) for establishing the analytical or low-fidelity results for
validation purposes to compare your numerical CFD data with
(case 1) for a time-step refinement study
The remaining 10% require you to think independently, investigate through
simulation, and report on one topic of your choice. Here is a non-exhaustive
list of suitable topics:
• Investigation of choice of spatial discretisation on results
• Investigation of time-stepping setup for optimised rate of convergence
• Investigation of choice of turbulence model and y+ value on results
• Assessment of mesh quality on results
• Assessment of far-field domain extent on force/moment coefficients
• Development of enhanced MATLAB meshing tool (beyond a channel)
• Production of a sophisticated illustration with enhanced data analysis
and appropriate reference to theory
2 The Three Test Cases
Unsteady laminar square cylinder
Besides the common assessed elements listed above, the unsteady laminar
flow past a square cylinder focusses on data analysis for extracting dominant
frequencies and coherent structures (either from the force coefficients or flow
Page 1
field variables or both). Useful plots and analyses show the development of
the vortex street and frequency spectrum. Explore when the flow becomes
unsteady through varying the Reynolds number and compare with literature.
The underlying theory, first introduced either in AERO316 Aerodynamics or
MECH326 Engineering Fluid Mechanics, will be revised during a lecture.
Inviscid supersonic diamond- (or double-) wedge aerofoil
Besides the common assessed elements listed above, the study of supersonic
inviscid flow focusses on reinforcing supersonic shock and expansion theory.
As such, you are required to do the appropriate hand calculations to establish
the local Mach number, static pressure, etc. for each deflection of the flow and
compare with the numerical results. Inviscid flow implies solving the Euler
equations which offer correct solutions for shock and expansion. Consider the
wake angle carefully. In addition, lift and drag coefficients can be extracted
from the numerical data and compared with the theory. Useful additional
simulations would include establishing the minimum free-stream Mach number
for a given deflection angle to obtain an attached shock at the leading edge
(i.e. to first find a detached shock). Alternatively, find the required opening
angle 𝜀 (see figure 1) for a given Mach number to observe a detached shock.
The underlying theories, first introduced either in AERO316 Aerodynamics or
MECH326 Engineering Fluid Mechanics, will be revised during a lecture.
Turbulent NACA 4-digit aerofoil
Besides the common assessed elements listed above, the required activities
for the turbulent aerofoil study include the geometry generation based on the
analytical definition of the NACA 4-digit series (a blunt trailing edge is
required!) and the additional data generation using low-fidelity aerodynamic
theories (using the XFOIL tool and/or, if you are an Aerospace Engineering
student, hand calculations based on thin airfoil theory). Useful plots would
include the curves of lift coefficient over angle of attack, the drag polar (i.e. lift
coefficient over drag coefficient), as well as selected pressure/skin-friction
distributions (i.e. pressure/skin-friction coefficient over x-coordinate).
Additional material for producing the geometry of the aerofoil and the low-
fidelity reference data is provided below in the appendices.
3 Reporting Expectations
While there is no hard page limit for the report, the recommendation is to not
exceed three pages per test case and two pages for the independent study,
which brings the total maximum to 14 pages, including a cover page, a blank
page after the cover page and a table of content on a separate page, but
excluding the appendix which lists the configuration (.cfg) files for each case.
Page 2
ENGG386 – Computational Methods in Engineering
The information that must be included in the report is:
• Case overview in a sketch form that includes the chosen domain
boundary conditions, principal domain dimensions, leading flow
parameters (as exemplarily shown in figure 1).
• Summary of your theoretical/analytical work and results. Your report
must sufficiently demonstrate validation efforts of the CFD results.
• Figures of mesh overview and zoomed-in regions to reveal important
detail (case-dependent). Numbers of vertices/elements in the volume
and for the surface/wall discretisation must be stated.
• Summary table of the simulation setup, showing important information
such as the selected SU2 solver (e.g. EULER, INC_RANS), turbulence
model (where applicable), spatial scheme, and time-stepper (including
CFL number), and also selected results (case dependent)
• Briefly justify your choices (e.g. chosen solver, boundary conditions)
and summarise your observations per test case in one paragraph.
• Figures of convergence histories for residual norms and (where
applicable) for force/moment coefficients
• Figures of pressure/skin-friction distributions and meaningful flow fields.
Also report on the y+ values to further assess your meshing quality.
• Append the configuration file (.cfg) for each case (properly formatted!)
All figures must be legible with meaningful captions, legends, axis labels, etc.
Besides the report, you are required to upload to Canvas your medium mesh
files (in .su2 format, plus .tin/.blk meshing files) and associated configuration
file (in .cfg format) for each test case. The configuration file must be set up for
the medium mesh and must run on execution. We will check your work using
an automated script to run your cases. The same script will analyse your files
to identify collusion.
Figure 1: Representative sketch of simulation setup for diamond-wedge flow.
Page 3
Appendices
NACA 4-digit Aerofoil Geometry Generation (MATLAB code)
N=125; % number of points
% example NACA 2412
m=0.02; %camber in percent chord
p=0.40; %max camber location in tenth chord
t=0.12; % max thickness in percent chord
% 1-cos mapping to cluster points at LE/TE
x = 0.5*(1.0-cos( linspace(0,pi(),N)) );
%aerofoil thickness distribution
yt = 5*t*(0.2969*sqrt(x)-0.1260*x-0.3516*x.^2+0.2843*x.^3-0.1015*x.^4);
%camber line and rotation argument
yc = zeros(size(yt)); ar = zeros(size(yt));
for i=1:N
if(x(i)<p)
yc(i) = m/p^2*(2*p*x(i)-x(i)*x(i));
ar(i) = atan(2*m/p^2*(p-x(i)));
else
yc(i) = m/(1-p)^2*((1-2*p)+2*p*x(i)-x(i)*x(i));
ar(i) = atan(2*m/(1-p)^2*(p-x(i)));
end
end
%assemble geometry
xU = x - yt.*sin( ar );
yU = yc + yt.*cos( ar );
xL = x + yt.*sin( ar );
yL = yc - yt.*cos( ar );
%plot
figure(1);plot(xL,yL,'k',xU,yU,'r',x,yc,'g')
axis equal, grid on
XFOIL Crash Course for Producing Validation Data
Using the program XFOIL to analyse aerofoil sections is straightforward and
mostly self-explaining following the instructions given in the user interface.
Here are some basic guidelines on how to run the program to produce data;
Installing and starting XFOIL
Download XFOIL directly from the website and unpack the archive (i.e. right
mouse click on the file and extract). Start the application of XFOIL named
xfoil.exe. This will open a DOS-type text console window. Maximise this
window. Do not close any windows as this will crash XFOIL.
Listing all current program options
Typing ? in the console will always list all possible commands at the current
stage of the program. Alternatively, there is a more detailed user guide at
http://web.mit.edu/drela/Public/web/xfoil/xfoil_doc.txt.
Page 4
ENGG386 – Computational Methods in Engineering
Aerofoil geometry and discretisation parameters
Type naca to set an aerofoil geometry of the NACA 4- or 5-digit series.
Following the instruction in the text console, you could type, for instance, 0012
to create the NACA 0012 aerofoil of the 4-digit series. Next type ppar to
inspect the geometry. This will open another window. Information is also
provided in the text console window. To change the number of surface panels
to discretise the aerofoil geometry, type N followed by the desired number. If
there is nothing else to change, press the return/enter key to navigate back to
the main menu.
Note that the length of the aerofoil is scaled between 0 and 1, where the
leading edge of the aerofoil is at the origin of the coordinate system.
Setting reference point for moment calculation
While you are in the main menu, type xycm and then give the x and y
coordinates of the reference location when prompted to change the reference
point for the moment calculation.
Running simulations
We only want to use the option of directly computing operating points, i.e.
defining angle of attack or lift coefficient for a specified Mach number,
Reynolds number and a given geometry. Thus, type oper followed by ? to list
the options. To change the Mach number for the simulation, type mach. To
toggle between viscous and inviscid simulations, type visc followed by the
required Reynolds number. Typing alfa will prompt you to give an angle of
attack which will start a simulation at that fixed angle of attack. Typing cl will
prompt you to give a target lift coefficient which will start a simulation to
calculate the required angle of attack to obtain the target lift. The results are
displayed in the second window.
Alternatively to speed things up, type pacc which will prompt you to enter a
name for a file where the results will be written to. When you see polar
save filename, choose a name such as naca0012.dat. No input is required
for polar dump filename, thus just press enter when prompted. Then you
can use aseq or cseq to run a sequence of angles of attack or lift coefficients
at once by specifying minimum, maximum and increment of the sequence as
prompted in the user interface. The results will be written automatically to the
chosen file, which is normally located where you started xfoil.exe.
Page 5
