You are on page 1of 9

2023 Coursework for 16429‐CFD

1. Introduction

Chimneys or smokestacks are a typical feature of modern society’s fluid waste disposal methods as
witnessed by the various visible gaseous emissions into the atmosphere from domestic and industrial
smokestacks (McGrattan et al. [1]) and from cooling towers or mobile exhausts. Other ‘non‐visible’
examples are represented by the releases of liquid into coastal water, rivers and lakes from a variety
of (industrial, municipal and agricultural) sources, mining and oil extraction operations (Jirka [2];
Tomàs et al. [3]), chemical reactors and various plants for waste treatment and desalination facilities
(Oliver et al. [4]). Many other variants can be found in the specific field of energy production, where
such configurations are a characteristic feature of thermal discharges from nuclear and fossil‐fueled
electricity generation plants (Martineau et al. [5]; Lee and Asce [6]; Fregni et al. [7]). They can also
exist at a smaller scale in typical problems relating to the cooling of computer mother boards and
related CPUs and memories (Sun and Jaluria [8]; Biswas et al. [9]) or in combustion chambers as a
result of the presence of holes or orifices for fuel injection and dilution (Issac and Jakubowski [10];
Baltasar et al. [11]). Similar concepts also apply to the manufacturing industry, where gas furnaces
are commonly used for the heat treatment of metals (Viskanta [12]). In the built (civil engineering)
environment, such configurations are widespread in emergency ventilation and air conditioning
systems in buildings (Venkatasubbaiah and Jaluria [13]; Subudhi et al. [14]; Morsli et al. [15,16];
Harish [17]) and can be found, in general, in every technological situation in which a heat exchanger
is required.

Given the diversity and rich spectrum of circumstances in which such configurations can be
encountered and the myriad technological applications briefly reviewed above, generalizations are
rather difficult. Many situations are possible in principle depending on the specific case considered.
However, all these cases share a common factor, namely, the existence of localized regions where
the temperature is higher than that of the surroundings (typically the area located at the bottom of
the considered chimney or smokestack configuration). This typically leads to the onset of “thermal
convection” [18, 19] (heated fluid tends to become lighter and therefore it rises through the
chimney). Two fundamental situations are possible in principle: Fluid rises in the chimney only due to
thermal convection (completely “natural” phenomenon), or a fixed mass flow rate is forced through
the chimney (this leading to “mixed” natural‐forced convection through the chimney). Here, we will
concentrate on natural convection (no forced flow).
The considered configuration can be seen in Figure 1. This geometry has a symmetry plane. It
consists of a region (called “heat chamber”) delimited externally by walls with a constant (high)
temperature and an upper region of larger horizontal extension (the chimney) delimited externally
by adiabatic (no heat exchange) solid walls. The symmetry axis is indicated with a dot‐dashed line.
The distance of the walls from the symmetry axis for the heat chamber and the chimney is denoted
by b and B, respectively (with B>b). Due to the heating effect of the walls, the fluid inside the heat
chamber expands (its density reduces) and therefore it becomes “lighter”. The lighter hot fluid tends
to rise in the heat chamber and enter the chimney region. As the velocity of these naturally‐induced
currents is in general relatively small, this flow can be considered incompressible and the Boussinesq
approximation can be used. As a result of the fluid rising through the heat chamber and the chimney,

1
new colder fluid (having temperature Tc) tends to be sucked into the heat chamber through the inlet
region located at the bottom of the heat chamber. In this way a continuous flow is obtained. As the
flow is incompressible (the amount of fluid entering the system per unit time must be equal to the
amount of fluid leaving the system per unit time), at any instant an amount of hot fluid equal to the
amount of cold fluid entering the system from the bottom, leaves the domain through the outlet section
(at the top).
As shown in Fig.1, the proper numerical simulation of this problem requires that the numerical domain
(the overall mesh) is not limited to the heat chamber and the chimney, It must also include two extra
areas corresponding to a certain portion of the external environment (located under the heat chamber
and over the chimney, respectively).

Figure 1 Schematisation of the Heat Chamber (channel)‐Chimney system. The vector is the gravity
acceleration

2
The characteristic number for natural (buoyant) flow occurring in this system is the Rayleigh number,
i.e.:

g T TL3
Ra 


where L is a reference length.

2. Geometry Size and Fluid Properties

Figure 2 Dimensions of the Heat Chamber‐Chimney System (lengths are measured in millimetres,
they shall be multiplied by 10‐ 3in order to have the corresponding ones in metres)

3
Table 1 Fluid properties for the Boussinesq model

Fluid Air
Density Model Boussinesq
Density  1.225 kg/m3
Specific Heat (Cp) Model Constant
Specific Heat (Cp) 1006 J/kgK
Thermal Conductivity Model constant
Thermal Conductivity  0.0242 W/mK
Viscosity Model constant
Viscosity 1.7894x10‐5 kg/ms
Thermal Expansion Coeff. Model constant
Thermal Expansion Coeff. T 0.0033 K‐1
Kinematic viscosity 1.46x10‐5 m2/s
Thermal diffusivity 1.96x10‐5 m2/s

3. Set‐up of the simulation

In Tutorial 3‐Heat Chamber‐Chimney System we have seen how the geometry and the mesh for such
a two‐dimensional problem can be created. Relevant information has also been provided about the
steps required to set‐up the Boussinesq approximation (to simulate buoyancy phenomena), set the
correct boundary conditions and, eventually, select the correct solution algorithm (solver) and
related integration schemes.

4. Analysis and final report

Note: Before starting to deal with the analysis, You are advised to read the Guide for the submission
of the simulation (you can find it on myPlace) where we report fundamental information about the
organisation and naming of your Ansys Workbench project.
Note: Submissions of the coursework without the simulation files will result in a 60% penalisation.

For the coursework, the following questions shall be addressed in the form of a technical report.
1) Generate a mesh as explained in the frame of Tutorial 3. Refer to the information shown in
Fig. 2. A (large) figure showing the mesh must be reported. Describe (briefly, one sentence only)
the mesh "properties" (overall number of grid points, shape of computational cells). Entitle the
section where you report these results “Basic Mesh” (5 Marks). See Note1

4
2) Set up the simulation following the instructions provided in the frame of Tutorial 3 and in the
preceding text (pressure‐based solver + Boussinesq approximation). Assume Tair=300 K and Thot =
310 K. Using the height of the heat chamber as the reference length (L), determine the Rayleigh
number for the considered problem. Run the simulation. Include plots showing the streamlines
and the Temperature distribution. Entitle the section where you report these results “Basic
Simulation” (10 Marks). See Note2

3) Discuss the “implications” about using a density‐based solver in place of the pressure‐based
solver for the chimney problem. Explain how the three equations on which projection methods
rely work in the context of pressure‐based solvers. Explain in your own words the main difference
between pressure‐based solvers and density‐based solvers with regard to the approach they use
to deal with the fluid “pressure”. Explain how a pressure‐based solver can deal with natural
convection induced by temperature variations although density is assumed to be a constant
quantity. (maximum 1 page). Entitle the section where you report this description “Equations and
Method description” (10 Marks). See Note3

4) Following the concepts illustrated in the frame of tutorial 2 (video part c), conduct a mesh
refinement study. For this study use as control parameters (i) the maximum value of the velocity
in x and y direction (ii) the heat flux on the walls of the heat chamber. Consider at least 4
different meshes in addition to that used for point (1) and run a simulation with each of them
(add approximately 15k nodes or elements in each step, until you find a convergent behaviour:
the easiest way to see this is by plotting the control parameter as a function of the number of
elements or nodes and verify that the curve tends to a plateau). Report the results in a table (the
table must report the number of elements of the mesh as well, i.e. it must show how the
maximum value of the velocity in x and y direction and the heat flux on the walls of the heat
chamber change as a function of the overall number of used computational cells). Attach a figure
showing in detail the mesh for which grid independence has been obtained (i.e. the mesh that
provides a similar result to the one with a smaller number of elements). The figure shall show a
detail of the inlet region, the heat chamber region, the chimney region and the outlet region.
Entitle the section where you report the outcomes of this assessment “Mesh Refinement Study”
(20 Marks). See Note4

5) Using the original geometry (question (4)) and a basic mesh of around 30,000 nodes, modify
the value of Thot. You should simulate at least 6 different values of Thot in addition to the original
one. You should remain in the range 310 Thot350 K and determine at least one value of Thot for
which the flow becomes “oscillatory”, i.e. time‐dependent. Use a “probe” located in the chimney
region to produce a signal about the evolution of temperature or velocity in time and show that
the flow is oscillatory (the difference between the maximum and minimum temperature in the
oscillatory signal should be at least 1K). Record the maximum and minimum variations in both
temperature and velocity signals (measured by the probe) over a period of approximately 10
seconds. Refer to Tutorial 3 and, first, run the simulation using the steady state solver (to develop
the flow field) and then use the transient solver (see Note 5). Moreover, for this (oscillatory) case,
plot the temperature contour and the streamlines at two or more different times, which clearly
show that the flow is changing in time. Determine the Rayleigh number for each considered value
of Thot and prepare a plot showing the maximum velocity as a function of the Rayleigh number.
Entitle the section where you report such results “Rayleigh number Effect” (20 Marks). (see Note
5 ).
5
6) With regard to the simulation related to question (5) for which oscillatory flow has been
found, discuss critically which time and space scheme would be more suitable to simulate a
problem like that. Implicit or explicit approach? Central differences or upwind? Explain why
(maximum half a page). Then, pick the simulation conducted for question (2) (steady flow) and
run again the simulation changing the spatial discretisation schemes (only momentum and energy
equations for one simulation). Describe briefly the modifications (even if they are extremely
small) induced by such a change in the velocity and temperature field and other related
differences (concerning, e.g., the calculation time, the convergence of the residuals, etc…).
(maximum 1/2 page). Entitle the section where you report such results “Discretisation scheme
assessment” (15 Marks). (see Note 6)

7) Imagining that a large increase in Thot can make the flow turbulent (e.g. Thot = 700 K or 800K),
in your own words (and without reporting equations or running simulations) identify the specific
aspects of the considered problem which would make it more relevant to the RANS or vice versa
to the LES, if it had to be treated in the framework of a turbulence model. Support the required
discussion with a relevant rationale based on the main principles on which these two different
approaches rely. Moreover, evaluate the Kolmogorov length corresponding to Thot = 800 K, and
discuss the advantages and disadvantages related to the DNS approach (maximum one page and
half, 20 Marks). See Note 7

NOTE: By completing the coursework a student shall prove himself/herself that:


1) he/she is able to build relevant grids and run simulations using an existing commercial package
(in this case ANSYS Fluent)
2) he/she has understood the "general" equations, principles and methods on which ANSYS
Fluent (and essentially ALL other existing CFD packages) rely.

5. Notes/FAQ – other cues and suggestions to make your coursework successful !

Note1: The dimensions of the geometry are different with respect to Tutorial 3; the splitting of
the edges shall be changed accordingly. In order to obtain convergence in the simulation, it is
important to set up the bias and number of divisions that guarantee a good mesh quality for the
new dimensions of the geometry.

Note2: Only the figures and related captions have to be included. No need to include descriptions
or interpretations.

Note3: There is no need to describe exactly how “Fluent” works, you shall describe the used
category of solvers ‘in general’, refer to lecture 4. Using a density‐based solver would have led to
notable differences from a “practical” point of view. Discuss how and why.

Note4: There is no suggested or maximum length for the response. In any case your document
shall not exceed the limit of 14 pages. The student is expected to develop a brief overview of how
the mesh was improved with each pass (increasing number of divisions, changing bias growth
rate to have more elements and also mentioning the orthogonal quality and skewness). Note 5b:
When creating a new mesh, the bias growth rate should be modified appropriately to give a
smooth transition in the size of the elements. You could also increase the number of iterations
for the steady solution as you refine the size of the mesh up to a maximum of 3500 iterations. In

6
general, mention where you have changed the number of divisions and why (e.g. the number of
divisions have been changed and the bias has been adjusted accordingly where gradients
appeared in the contours of the temperature, velocity, and pressure). Note5c: Increments of less
than 15k nodes or elements in each step are acceptable. However, they will result in a larger
number of steps to achieve convergence.

Note5: Only figures, tables and related captions have to be included. No need to include
descriptions or interpretations. Make sure that the figures that you include for different time
clearly show that the flow is NOT steady (e.g. the position and shape of the thermal plume or
vortices are changing in time). Run the simulation in 3 stages starting from a steady state solver
to a fixed courant adaptive transient/fixed time step solver and finally running a fixed time step
transient solver monitoring the flow over a period of 10 seconds:

Stage1: To develop the flow field, run the case for 1000 iterations using the steady solver as
shown in Tutorial3 and monitor the temperature and probe velocity reports.

Stage2: Using a time‐step size of 0.001 seconds and 1000 time steps, run a 1 second and extract
the temperature and velocity probes for the flow. You can also use an adaptive transient solver
to monitor the flow. However it may take a longer time.

Stage 3: Increase the time step size to 0.01 seconds and plot the temperature and velocity
variations graphically using the probe signals for a flow time of approximately 10 seconds.

Suggested position for the probe: place the probe along the frontier of the central jet at a
distance from the bottom of the domain in the range between 0.19 and 0.21 m (a probe located
in the chimney area, e.g. at a distance 0.15 from the bottom would also be OK ).

Note6: Explain why space schemes pertaining to the central difference and upwind categories are
different and do the same for the explicit and implicit methods (Refer to lectures 2 and 3). In
particular, which spatial scheme would be more suitable for an oscillatory (or turbulent)
problem? Discuss the implications about using a central‐difference scheme to detect the
transition from steady to oscillatory flow. With regard to the computations, only a table (showing
the maximum velocity and/or the heat flux) has to be included. No figures are needed.

Note7: As a definitive or ‘perfect’ turbulence model does not exist, marks will be assigned only
on the basis of the quality of reflections provided by the student (Refer to lecture 5).

GENERAL NOTE: Your report shall be submitted in pdf format. Reports containing more than 14
pages (excluding the cover page, table of contents, references, appendix) will be penalized. The
report shall be prepared using the dedicated template (which also includes the cover sheet and
the initial table where the student shall indicate which questions have been addressed). No
need to include an abstract or an introduction. For each figure and table a dedicated caption
shall be prepared describing what is shown in the figure or table and the conditions for which
those results have been obtained. A plagiarism check will be conducted on the report through
comparison with the text and figures included in previous‐years reports (and this‐year reports
by other students). The original ANSYS Fluent files, shall also be submitted together with the
report (attach them separately as a single zip file). Refer to the Guide for the submission of the
simulation for detailed info about the files to be submitted and their format. Submissions of
7
the coursework without the simulation files will result in a 60% penalisation. The content of
these files will be checked to verify that they reflect what has been included in the report.

6. References

[01] McGrattan K.B., Baum H.R., Rehm R.G., (1998), Large eddy simulations of smoke movement, Fire
Saf. J. 30, 161–178.

[02] Jirka G.H., (2004), Integral Model for Turbulent Buoyant Jets in Unbounded Stratified Flows. Part
I: Single Round Jet, Environmental Fluid Mechanics 4: 1–56, 2004.
[03] Tomàs A.F., Poje A.C., Özgökmen T.M., and Dewar W.K., (2016), Dynamics of multiphase
turbulent plumes with hybrid buoyancy sources in stratified environments, Phys Fluids, 28,
095109 (24 pp).
[04] Oliver C. J. Davidson M. J. and Nokes R. I., (2013), Removing the boundary influence on
negatively buoyant jets, Environmental Fluid Mechanics, 13(6), 625–648

[05] Martineau R.C., Berry R.A., Esteve A., Hamman K.D., Knoll D.A., Park R., Taitano W., (2010),
Comparison of natural convection flows under VHTR type conditions modeled by both the
conservation and incompressible forms of the Navier–Stokes equations, Nucl. Eng. Des. 240,
1371–1385.

[06] Lee J. H.W. and Asce F., (2012), Mixing of Multiple Buoyant Jets, J. Hydraul. Eng., 138(12),
1008‐1021.

[07] Fregni A., Angeli D., Cimarelli A., Stalio E., (2019), Direct Numerical Simulation of a buoyant triple
jet at low‐Prandtl number, Int. J. Heat Mass Transfer, 143, 118466.
[08] Sun Z. and Jaluria Y., (2011), Conjugate thermal transport in gas flow in long rectangular
microchannel, J. Electron. Packag. 133(2), 021008 (11 pages).
[09] Biswas N., Mahapatra P. S., Manna N.K. and Roy P.C., (2015), Influence of Heater Aspect Ratio
on Natural Convection in a Rectangular Enclosure, Heat Transfer Engineering, 37(2), 125‐139.

[10] Isaac K.M. and Jakubouski A. K., (1985), Experimental study of the interaction of multiple jets
with a cross flow, AIAA Journal, 23(11), 1679‐1683.

[11] Baltasar J., Carvalho M.G., Coelho P., Costa M., (1997), Flue gas recirculation in a gas‐fired
laboratory furnace: measurements and modelling, Fuel , 76(10), 919–929.
[12] Viskanta R., (1993), Heat transfer to impinging isothermal gas and flame jets, Experimental
Thermal and Fluid Science, 6(2), 111‐134.
[13] Venkatasubbaiah K. and Jaluria Y., (2012), Numerical simulation of enclosure fires with horizontal
vents, Numer. Heat Tran. 62, 179–196.

[14] Subudhi S., Sreenivas K.R., Arakeri J.H., (2013), Study of buoyant jets in natural ventilation of a
model room, Int. J. Heat Mass Transf. 64, 91–97

8
[15] Morsli S., Ramenah H., El GanaouiM. and Bennacer R., (2018), Effect of aligned and misaligned
ventilation opening affecting energy demand and air quality in buildings, Eur. Phys. J. Appl. Phys.
83, 10901 (10 pages). https://doi.org/10.1051/epjap/2018180119

[16] Morsli S., Boussoufi M., Sabeur A., El Ganaoui M., Bennacer R., (2018) Small to large scale mixed
turbulent convection: buildings application, International Journal of Numerical Methods for Heat
& Fluid Flow, 28(1), 188‐205, https://doi.org/10.1108/HFF‐10‐2017‐0392

[17] Harish R., (2018), Effect of heat source aspect ratio on turbulent thermal stratification in a
naturally ventilated enclosure, Building and Environment , 143, 473–486.
[18] Lappa M., (2009), Thermal Convection: Patterns, Evolution and Stability, John Wiley & Sons, Ltd
(2009, Chichester, England).

[19] Lappa M., (2012), Rotating Thermal Flows in Natural and Industrial Processes, John Wiley & Sons,
Ltd (2012, Chichester, England).

You might also like