FORUM

P E E R- R E V I E W E D T E C H N I C A L PA P E R S

Sources of errors for

indoor air CFD simulations (Part 2)
Kemal Gungor, M.AIRAH Norman Disney & Young

ABSTRACT
Computational fluid dynamics (CFD) has emerged as a viable design tool in the industry over the past decade.
CFD is a powerful technique for estimating air motions within buildings; however, the accuracy of CFD simulations
strongly depends on the appropriate setting of boundary conditions and numerical simulation parameters. Thermal
comfort and indoor air quality simulations are the most common CFD applications within the construction industry.
Lack of any common guideline/protocol causes diverse settings for the CFD simulations. This also causes difference
in simulation outcomes for indoor air CFD simulations. This study summarises the factors affecting the indoor air flow
simulation and demonstrates the effect of using simple settings upon the results.

THE BACKGROUND
The first part of this paper, published in September Ecolibrium,
provided a brief historical background on the use of Computational
Fluid Dynamics (CFD), explaining that it solves and analyses
fluid-flow and heat-transfer problems by using numerical methods
and algorithms.
The paper detailed how in the past few years, CFD
has played an increasingly important role in building design,
assessment of HVAC system performance, indoor air quality,
air-change effectiveness, fire safety and thermal comfort.
Indeed, CFD models account for 70 per cent of the ventilation
performance studies published in 2007 [2].
During the past few years, there has been a substantial Figure 8: Geometric model for the air-change
increase in CFD simulations for the construction industry, effectiveness simulation.
predominately for indoor air quality and/or thermal comfort.
There is anecdotal evidence that not all of the simulations are A total of 2190 L/s fresh air was supplied at 14°C and 16°C
performed to the same standards, guidelines, detail, and level to the occupied zone through 11 swirl diffusers. Three different
of accuracy. Trust and quality issues in CFD have become flow rates were used. Thermal simulation was carried out
important, as it is widely used in design and analysis. to determine the facade loads, i.e. wall boundary condition.
Direct solar gains were applied to a 4m wide perimeter zone floor
This study is an overview of the CFD simulation parameters for each orientation.
affecting quality/accuracy of CFD analysis in general and more
specifically the thermal comfort and air-change effectiveness Conduction heat loads were applied individually to visual
analysis. and spandrel sections and also vary with their orientation.
The semicircle section of the building has full height glass.
Part 1 of the paper explored general uncertainties and errors
Sitting occupants were represented by 1200 x 500 x 400mm
in CFD simulations, the effects of software packages, user errors
prism and 75W sensible heat gains are applied to occupant
in CFD simulation, CAD-related errors, errors due to improper
surface. 150W equipment loads were applied to the computers
grid generation, inappropriate setting of solver and boundary
located on desks for each occupant. For each luminaries,
conditions, and post-processing user errors.
49W light loads applied. Maximum age of air should be 369 s
to achieve 0.95 air- change effectiveness, ACE.
AIR-CHANGE
ANSYS CFX Version 15.0 was used to carry out the CFD analysis.
EFFECTIVENESS SIMULATION The shear stress transport (SST) turbulence model was used
An air-change effectiveness simulation was performed because it offers significant advantages for non-equilibrium
to demonstrate how a very simple user input can change turbulent boundary layer flows and heat-transfer predictions.
the simulation outcome. Figure 8 shows the geometric model The SST model is as economical as the widely used k-ε model, but it
of a 279.3 m2 office used for the CFD simulation. offers much higher fidelity, especially for separated flows, providing

36 E CO L I B R I U M • O C TO B E R 2 0 1 5
 FORUM

excellent answers on a wide range of flows and near-wall mesh

conditions. The turbulence intensity at the diffuser discharge
was set to 5%.
Non-conformal meshing techniques have been used to generate
the computational grid. The maximum mesh face size is 100mm.
As the mesh approaches the objects, the mesh size progressively
reduces. In the immediate vicinity of the diffusers, finer mesh
sizes were used to resolve the detail of the ventilation discharges
and intakes. A maximum 5mm mesh face size was used at the
discharges.
RMS of residuals was used as the global convergence criterion.
The simulation was considered to be converged when the global
convergence criterion of 0.0001 was achieved.
Again, three air-change effectiveness simulations were Figure 9: Mean age of air contours for the Scenario 1. Inset shows
performed to demonstrate how a single user input can alter area that MAA>369 s, 4.8 %.
the entire CFD simulation.
• Scenario 1: Base simulation as defined above
• Scenario 2: Same as Scenario 1, only the radiation
calculations were turned off
• Scenario 3: Workstations were eliminated from the
geometric model and occupant and equipment
heat loads were applied to the floor as uniform
heat flux. This is a very common approach in
order to control the number of meshing elements
for very large floor plate simulations. All the
other inputs and set-ups are same as Scenario 1.
Figure 9 to 11 shows the age of air contours for three scenarios.
Table 1 gives the areas that achieve less than 0.95 air-change
effectiveness, i.e. has age of air greater than 369 s. These
areas are also illustrated in Figure 9–11 insets. Mean age
of air contours are plotted at 1000mm AFFL.
Figure 10: Mean age of air contours for Scenario 2. Inset shows area
that MAA>369 s.
Area less than 0.95 ACE
Scenario
m2 %
Scenario 1
13.4 4.8
(Base simulation)

Scenario 2
22.4 8.0
(No radiation heat transfer)

Scenario 3
(Uniform occupant 34.1 12.2
and equipment loads)

Table 1: Areas that have less than 0.95 ACE, I.e. MAA>369 s.

IS CFD AN “IMPERFECT TOOL”?

Considering that the first general-purpose CFD software,
CHAM (Prof. Brian Spalding) has been available since 1974,
some might think that there are various issues with CFD. One
might even find some early publications that show discrepancies
between simulation results and experimental results. In such
publications an estimate of experimental uncertainty is also not
quoted, and in most cases it is not even available. In science and
engineering, CFD was one of the first fields to seriously begin
developing concepts for validation methodology and validation
experiments.
Figure 11: Mean age of air contours for Scenario 3.
Inset shows area that MAA>369 s.
In June 2000, Oberkampf and Trucano [21] published an
extensive review of the literature in computational validation,
quoting 159 works in the field. The NASA [22] website provides
a list of references related to CFD verification, and validation
of CFD codes. In the literature there are industry-specific
best-practice guidelines.

O C TO B E R 2 0 1 5 • E CO L I B R I U M 37
 FORUM
Three prominent best-practice guidelines have been developed by
• the European Research Community on Flow, Turbulence,
and Combustion (ERCOFTAC)[4];
• OECD /Nuclear Energy Agency/Committee on the
Safety of Nuclear Installations CFD working groups[23];
• the American Society of Mechanical Engineers (ASME)[24].
These studies, the exponential increase in computer power, the
development of efficient and robust numerical algorithms and
the continuous improvements in physical modelling all enable
us to use CFD successfully and confidently in a wide range
of industries such as aerospace, nuclear energy, shipbuilding,
medical (biomedical flows), food, etc.
In the scientific literature, there are many case studies
highlighting the success of CFD simulations. Only one of
such successful CFD simulation results are shown in Figure 12
and Figure 13. Figure 12 demonstrates an excellent agreement
between the wind tunnel test and CFD simulation results
carried out by NASA. One should note that such an outstanding
simulation/comparison was performed in 1990.
Figure 12: Comparison of CFD simulation
and wind tunnel test from 20.
Figure 13 compares CFD simulation results against actual flight
measurements. A good agreement between the CFD simulation
and actual measurement can be noticed.
Figure 13: Comparison of CFD simulation results
against actual measurements from 20.
38 E CO L I B R I U M • O C TO B E R 2 0 1 5
The ERCOFTAC QNET-CFD knowledge base wiki introduces
a new concept for collecting, structuring and sharing knowledge,
best-practice advice and guidelines for all CFD applications.
The knowledge base is partitioned into sector disciplines such
as “built environment”, “chemical and process engineering”,
“external aerodynamics”, “turbomachinery”, “combustion”
and “heat transfer”, etc. Each sector disciplines (application
area) is comprised of application challenges. These are realistic
industrial test cases, which can be used to judge the competency
and limitations of CFD for a given application area.
The National Agency for Finite Element Methods and Standards
(NAFEMS), is an independent not-for-profit organisation that
has the aim of promoting the effective and reliable use of CFD.
NAFEMS publishes an/the International Journal of CFD Case
Studies. This journal presents a series of mainly industrial case
studies presenting comparisons with experiment as validation.
All these studies show that CFD can be used confidently
and successfully in very large industries/disciplines.
SUMMARY AND CONCLUSION
CFD indoor simulations are sensitive to a large number of input
variables [25] and minor changes that can significantly affect the
airflow pattern. We need to have a good understanding of where
errors and uncertainties can arise within the modelling and
solution process. Often, user inputs are the source of varying/
erroneous CFD simulation results. The potential for usage error
multiplies with the increased level of options available in a CFD
package. User errors can be minimised through proper training
and the accumulation of experience.
CFD can be an imperfect tool; however, this is only the case
when it is used inappropriately and with very limited resources
(CPU core and RAM) within an unrealistically short timeframe,
or when used by a user who has very limited experience.
Otherwise, it is no different than any other building thermal
simulation package or engineering tool.
All of the above mentioned works have proven that CFD is
an invaluable tool that has a lot to offer to design improvement.
Finding a solution to design challenges without the aid
of CFD would be highly time-consuming and costly.
CFD can identify the issues, and presents the solution
to the construction industry during the design stage.
CFD offers unprecedented energy-saving solutions on building
HVAC systems and its component design.
If we can use CFD to design spacecraft, cars, nuclear reactors,
and submarines, without any doubt we can confidently
and successfully apply it to the building industry. ❚
REFERENCES
1 Manickam Siva Kumar, Pichai Philominathan “Bringing out
Fluids Experiments from Laboratory to in Silico – A Journey
of Hundred Years”, American Journal of Computational
Mathematics, 2011, 1, 271–280
2 Qingyan Chen “Ventilation performance prediction for
buildings: A method overview and recent applications”,
Building and Environment 2009, 44(4), 848–858.

3 “Guide for the Verification and Validation of Computational
Fluid Dynamics Simulations”, American Institute of
Aeronautics and Astronautics, AIAA G-077-1998,
4 Casey, M., and Wintergerste, T., (ed.), European Research
Community on Flow, Turbulance and Cumbastion,
ERCOFTAC, Special Interest Group on Quality and Trust
in Industrial CFD “Best Practice Guidelines”, Version 1.0,
January 2000
5 CFD Online, http://www.cfd-online.com/Wiki/Codes
6 TenLinks, http://www.tenlinks.com/cae/products/cfd.HTM
7 Yoshihide Tominaga et al, “AIJ guidelines for practical
applications of CFD to pedestrian wind environment around
buildings”, Journal of Wind Engineering and Industrial
Aerodynamics 96 (2008) 1749–1761
8 Jörg Franke, Antti Hellsten, Heinke Schlünzen, Bertrand
Carissimo “Best Practise Guideline For the CFD Simulation
of flows in the Urban Environment” COST Action 732,
1 May 2007
9 Steve Owen, “An Introduction to Mesh Generation
Algorithms”, 14th International Meshing Roundtable
presentation, San Diego, California, USA,
September 11–14, 2005
10 Dimitri J. Mavriplis, NASA Langley Research Center,
”Unstructured Mesh Related Issues In Computational
Fluid Dynamics, (CFD) – Based Analysis And Design”,
11th International Meshing Roundtable, Ithaca New York,
USA, September 15–18, 2002,
11 Stanford University, ME469B/2/GI “Geometry Modeling &
Grid Generation” hand-out, http://web.stanford.edu/class/
me469b/handouts/geoan grid.pdf , last accessed, 3/3/2015
12 R. H. Nichols, “Turbulence Models and Their Application
to Complex Flows”, University of Alabama at Birmingham,
Revision 4.01
13 Christopher J. Roy and Frederick G. Blottner, “Review
and Assessment of Turbulence Models for Hypersonic Flows:
2D/Axisymmetric Cases”, 44th AIAA Aerospace Sciences
Meeting and Exhibit, 9¬–12 January 2006, Reno, Nevada
14 NASA, Langley Research Centre, Turbulence Modeling
Resource, http://turbmodels.larc.nasa.gov/index.html last
accesses on 7/3/2015.
15 Srebric, J. and Chen, Q. 2001. “A method of test to obtain
diffuser data for CFD modeling of room airflow,” ASHRAE
Transactions, 107(2), 108–116.
16 Eddy Rusly and Mirek Piechowski, “CFD Modelling For Swirl
Diffuser and its Implications on Air Change Effectiveness
Assessment to Green Star’s IEQ-2”, Proceedings of Building
Simulation 2011: 12th Conference of International Building
Performance Simulation Association, Sydney, 14–16 November.
17 Gery Einberg “Air Diffusion and Solid Contaminant
Behaviour in Room Ventilation – a CFD Based Integrated
Approach”, Doctoral Thesis, Kungliga Tekniska Högskolan,
June 2005, ISBN 91-7178-037-8
FORUM
18 Sean Badenhorst, “Floor swirl diffusers: types and
applications”, Ecolibrium, 40–50, October 2013,
19 “Tips & Tricks: How to interpret results for multiphase &
porous domains using true velocity and superficial velocity”,
Posted By LEAP CFD Team on Nov 22, 2012, –www.
computationalfluiddynamics.com.au/interpreting-results-
for-superficial-and-true-velocity/, last accessed on 16/3/2015.
20 Pieter G. Buning, Reynaldo J. Gomez, “20+ Years of Chimera
Grid Development for the Space Shuttle”, 10th Symposium
on Overset Composite Grid and Solution Technology
September 20–23, 2010, Moffett Field, CA
21 W. L. Oberkampf and T. G. Trucano, “Validation
Methodology in Computational Fluid Dynamics”, American
Institute of Aeronautics and Astronautics, AIAA 2000-2549,
Fluids 2000, 19–22 June 2000, Denver, CO
22 www.grc.nasa.gov/WWW/wind/valid/tutorial/bibliog.html,
last accesses 25/3/2015
23 Nuclear Energy Agency and The Committee on the Safety
of Nuclear Installations (CSNI) “Best Practice Guidelines
for the Use of CFD in Nuclear Reactor Safety Applications”,
NEA/CSNI/R(2014)11, February 2015
24 ASME, “Standard for Verification and Validation in
Computational Fluid Dynamics and Heat Transfer”,
V V 20 – 2009.
25 Kemal Gungor, “Guide to air change effectiveness”,
Ecolibrium, March 2013.
26 J. Shewchuk, “Lecture Notes on Delaunay Mesh Generation”,
The University of Iowa, 58:110 Computer-Aided Engineering,
Lectures 12 and 13: Mesh Generation Notes, 1999.
ABOUT THE AUTHOR
Kemal Gungor (PhD), M.AIRAH, is an ESD consultant at
Norman Disney & Young. Kemal’s expertise lies in heat
transfer, thermodynamic analysis, design and computer
modelling of thermal systems (i.e. energy modelling, facade
analysis). He is also specialised in computational fluid
dynamics analysis. These skills are being utilised to search
for, develop and implement ecologically sustainable design
solutions within the building industry.
O C TO B E R 2 0 1 5 • E CO L I B R I U M 39