School of Mechanical and Aerospace Engineering
MA4814 – Computational Fluid Dynamics
Final Project (Industry version): Thermal Laboratory
(AY 2023-2024, Semester 1)
Name: Tejas Prabodh Shreeshanth
Matriculation No: U2023584L
1.
Introduction ........................................................................................................................ 2
2.
Approach to Position Optimisation .................................................................................... 3
3.
Meshing .............................................................................................................................. 4
4.
CFD Modelling ................................................................................................................... 5
5.
Results and Discussions...................................................................................................... 6
6.
Conclusion ........................................................................................................................ 10
1. Introduction
Efficient air conditioning is critical for regulating temperatures in enclosed spaces such as
laboratories, control rooms, and data centres. The positioning of the air supply and
exhaust vents can significantly impact the airflow patterns and temperature uniformity
within the space. Optimally locating these vents is essential to maximize cooling
performance while minimizing energy usage. This project focuses on optimizing the
placement of a pair of air conditioner (A/C) inlet and outlet vents in a small thermal
laboratory using computational fluid dynamics (CFD) simulations.
The thermal lab considered is a cube-shaped room with dimensions 1.8 x 1.8 x 1.8 m3 (L
x W x H). It contains heat sources that must be cooled, including a heater, and two lights.
The goal is to identify the optimal positions for a single cooled air supply vent and warm
air return vent on the vertical walls that provide the most uniform and efficient cooling.
The project is constrained by fixed locations of the heat sources, and allowable positions
on the side walls for the two 19 cm diameter vents.
To determine the optimal vent configuration, an iterative CFD optimization approach is
taken. First, a CAD model of the lab is constructed in SolidWorks, with the vents
parameterized to enable positional changes. ANSYS Workbench is then used to mesh the
fluid volume and set up the CFD simulation in Fluent. The standard k-ε turbulence model
is utilized along with a buoyancy-driven ideal gas law density model. Appropriate
boundary conditions are imposed based on the problem statement.
The objective function guiding the optimization is defined as the average temperature in
the diagonal plane and the 75 cm centre plane. For each iteration, the vent positions are
updated in the CAD model, mesh regenerated, and CFD simulation solved to evaluate the
objective function.
This report documents the CFD modelling methodology, optimization results, and
quantitative comparison to the initial design. The optimal inlet and outlet positions are
shown to produce significantly lower temperatures than the initial configuration. Velocity
and temperature contour plots provide insight into the improved airflow patterns. The
project demonstrates how CFD can be leveraged to enhance the performance of air
distribution systems for electronics cooling and climate control applications.
2. Approach to Position Optimisation
Here are some points that I have taken into consideration based on natural phenomena in
the thermal lab,
1. Based on natural convection phenomena (i.e., cold air sinks and hot air rises), the
supply was placed at the highest point and exhaust at the lower point.
2. The inlet and outlet were placed diagonally to facilitate more airflow in the room.
3. To have the maximum circulation and with reference from cooling systems in
industries, the supply and exhaust were placed on the same wall.
4. The supply is facing the heater outflow to restrict the circulation of hot air in the room
and to reduce the temperature.
Supply Inlet (opposite wall)
AC outlet (closer wall)
3. Meshing
Meshing of the model was done using the meshing in Ansys Workbench. The parts of the
lab were named accordingly as inlets, outlets, walls and cases to account for the boundary
conditions.
The base mesh size was settled at 0.03m per cell as an ideal size, to achieve less
discrepancies while refining the mesh to get a more accurate solution.
An inflation layer was added with 10 layers and a growth rate of 1.2. Doing this helps the
solver to accurately capture the necessary velocity, pressure, and temperature near the
walls, which leads to a more accurate solution.
Figure 1: Mesh Statistics
The mesh was further refined by mesh adapt during the Fluent solver part by setting the
Y-plus values between 0 and 10 as a refinement criterion. The final range of Y-plus values
are given in the report and histogram below showing 99.8% of the elements having a
value between 0 and 6.2. (the optimal value for the SST k- ω model is <11).
Figure 2: Quality Report of Mesh
Figure 3: Y-plus values vs % of elements
4. CFD Modelling
The turbulence model used for this project is Shear-Stress-Transport k- ω (omega) model,
because of
1. Ideal for near wall regions
2. Blends with k-ε (epsilon model) which is ideal for free stream objects.
Thus, making it more reliable for more accurate solutions.
Based on the values provided in the project brief,
The velocity for the ac inlet was calculated as shown:
Given flow rate: 250m^3/hr.
Converting to m^3/s and dividing by the area of inlet: 0.0694/0.0284 = 2.44m/s
Hydraulic diameters were applied wherever possible, (inlets and outlets).
After running the project in Ansys Fluent, we get the shown scaled residual plot:
Figure 4: Scaled Residual Plot
The slopes of the plot (run for 250 iterations) indicate a smooth and non-divergent solution
with respect to the fluid domain and the turbulence model. Second order solution methods
were followed to improve the accuracy of the solution.
The mass flow rate shown by the below by the surface integral report implies that the mass
was conserved in the simulation, and there are no discrepancies with the theory.
Figure 5: Verification of Conservation of Mass
5. Results and Discussions
1. 2 planes, a diagonal to the XZ-axis and a plane parallel to the y axis at 0.75m are used
to measure the average temperature of the room.
Figure 6: temperature contour of y at 0.75m
Figure 7: temperature contour of diagonal plane
Figure 8: Surface integral report of Average Temperature in the planes
2. A plot was generated using the average temperature (obtained from the surface
integral) of y planes throughout the domain.
This shows a general trend of temperature reducing with increase in the Y axis but has
disturbances due to the heater from 0.2 to 0.6m.
Figure 9: Surface integral report of Average Temperature over Y axis
Y-axis vs Temperature
300
298
296
294
292
1
2
3
4
5
6
7
8
9
10
Temperature
Figure 10: Plot of Temperature change vs Y axis position
3. Velocity path lines, temperature vectors and temperature contours are used to gain
more insights into the airflow pattern and the temperature distribution in the lab.
They are:
1. Cool air circulation as shown in the velocity path line diagram showing the
efficient airflow pattern in the lab.
2. The temperature vector and contours across the wall show the temperature
distribution across the lab which shows a good temperature distribution across the
lab keeping every single corner cool.
3. Surface integral reports are shown to show the temperature at each point, and can
be useful in further optimisation of the vents.
Figure 11: Velocity path lines
Figure 12: Temperature Vectors
Figure 13: Temperature contour of interior of lab
Figure 14: Average Temperature of walls
6. Conclusion
The average temperature of the room has reduced 5.5 degrees Celsius from the room
temperature due to the air conditioning supply and outlet vents.
The placement of these vents is important in keeping thermal laboratories cooled, and
similar concepts can be used in data centres, cleanrooms, and precision engineering
facilities with temperature restrictions.
This can be further optimised by suing concepts like the Coanda effect, and studying the
airflow pattern and temperature distribution where supply and exhaust vents are placed on
the same side.