University of Southern California Dissertations and Theses

Airflow investigation of fabric membrane forms: a fluid dynamic analysis for thermal comfort

(USC Thesis Other)

Airflow investigation of fabric membrane forms: a fluid dynamic analysis for thermal comfort

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content

AIRFLOW INVESTIGATION OF FABRIC MEMBRANE FORMS:

A fluid dynamic analysis for thermal comfort

by

Stephanie E. Egger

A Thesis Presented to the
FACULTY OF THE USC SCHOOL OF ARCHITECTURE
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
MASTER OF BUILDING SCIENCE

August 2013

Copyright 2013 Stephanie E. Egger
ii

ACKNOWLEDGEMENTS

I would first and foremost like to thank my committee members, Professor G G Schierle,
Simon K. Chiu, Professor Peter Simmonds, and Professor Karen Kensek. Without their
guidance and support my research would not have embodied the final form it did.

Additional thanks are due to the Discipline Head for Building Science, Professor Douglas
Noble who provided valuable support and served as a thought partner in the initial phases
of the research. Ilaria Mazzoleni and Jeff Landreth also deserve recognition for their
advice and insight offered during the later phases of the research.

I must express gratitude to Professor Robert Wehdorn-Roithmayr, creator of Formfinder,
for providing software licensing. I am also grateful to Jun Ortega from Autodesk for
reviewing the simulation models. Without their generosity this research would have been
much more difficult.

Adam Menter and the rest of the Autodesk Sustainability Team provided guidance and
flexibility in my desire to study and work simultaneously. I owe them many thanks for
their patience and confidence in my work.

I would also like to thank my fellow MBS students, past and present, who over the past
two years provided kind words of support. Our conversations and laughs have been a
grounding force through our academic challenges.

My family, for their lifelong belief in me. Early on they gave me the tools to succeed in
whatever I pursue and I am here today thanks to that.

Finally, I owe thanks and appreciation to Sean Kelly, who over the past two years has
encouraged me, listened to me, believed in me, and provided a support system when the
rest of my friends and family were far away. I would not be the person I am today
without you.

iii

TABLE OF CONTENTS
LIST OF TABLES ........................................................................................................... vi
LIST OF FIGURES ........................................................................................................ vii
ABSTRACT ...................................................................................................................... ix
Chapter 1 : INTRODUCTION TO AIRFLOW FOR FABRIC MEMBRANE
FORMS .............................................................................................................................. 1
TERMINOLOGY: Language use for this thesis .................................................................... 1
1.1 HYPOTHESIS STATEMENT: Informed Form .............................................................. 3
1.2 IMPORTANCE: Why study airflow for fabric membrane structures? ........................ 3
1.2.1 Current Environmental Research ................................................................................... 4
1.2.2 Research Contributions .................................................................................................. 6
1.3 STUDY BOUNDARIES: Defining the research limits .................................................... 6
1.3.1 Static Behavior of the Fabric ......................................................................................... 9
1.4 DELIVERABLE: A process for designing and analyzing fabric forms ....................... 10
1.5 CHAPTER STRUCTURE: Outline of the study ........................................................... 10
Chapter 2 : BACKGROUND FOR FABRIC MEMBRANES, THERMAL
COMFORT AND CFD ANALYSIS.............................................................................. 11
2.1 FABRIC MEMBRANES: Defining the form ................................................................. 11
2.1.1 Origins of Fabric Membranes ...................................................................................... 11
2.1.2 Classifications .............................................................................................................. 14
2.1.3 Components of fabric membranes ............................................................................... 19
2.1.4 Benefits to designing with fabric ................................................................................. 20
2.2 THERMAL COMFORT: Measuring occupant comfort............................................... 20
2.2.1 Predicted Mean Vote (PMV) ....................................................................................... 22
2.2.2 Predicted Percentage of Dissatisfied (PPD) ................................................................. 24
2.2.3 Adaptive Thermal Comfort .......................................................................................... 24
2.3 CFD ANALYSIS: For airflow and human comfort ....................................................... 26
2.3.1 Introduction to CFD ..................................................................................................... 26
2.3.2 Equations for Solving a Fluid Flow Problem .............................................................. 30
2.4 APPLICATIONS: CFD with fabric membranes ........................................................... 34
2.5 CONCLUSIONS: Background research contributing to the study ............................. 37
Chapter 3 : METHOD & PROCESS OF STUDYING AIRFLOW FOR FABRIC
MEMBRANE FORMS THROUGH CFD .................................................................... 38
3.1 INTRODUCTION: Overview of Method & Process Development .............................. 38
3.2 SOFTWARE WORKFLOW: Using software interoperability .................................... 38
3.2.1 Formfinder ................................................................................................................... 40
3.2.2 Autodesk
®
AutoCAD
®
Civil 3D
®
and Autodesk
®
Inventor
®
....................................... 41
3.2.3 Autodesk
®
Revit
®
........................................................................................................ 42
3.2.4 Autodesk
®
Simulation CFD ......................................................................................... 43
3.3 WORKFLOW TESTING: Preventative measures ........................................................ 49
3.4 SENSITIVITY ANALYSIS: Testing variable influences .............................................. 51
3.5 CLIMATE CONDITIONS: Different scenarios ............................................................ 53
3.5.1 Kuala Lumpur: Hot and humid .................................................................................... 55
3.5.2 Munich: Cool and temperate ........................................................................................ 55
iv

3.5.3 Abu Dhabi: Hot and arid .............................................................................................. 56
3.6 MODEL GEOMETRY: Realistic external scenario ...................................................... 57
3.6.1 Context Geometry ........................................................................................................ 57
3.6.2 Fabric Membrane Shade Geometry ............................................................................. 59
3.6.3 Human Models ............................................................................................................. 60
3.6.4 Flow Geometry and Ground Geometry ....................................................................... 61
3.7 CFD SIMULATION SETTINGS: Building & assigning each simulation model ....... 63
3.7.1 Geometry Tools (1) ...................................................................................................... 64
3.7.2 Assign Materials (2) .................................................................................................... 65
3.7.3 Boundary Conditions (3) ............................................................................................. 71
3.7.4 Assign and Modify the Mesh (4) ................................................................................. 75
3.7.5 Solution Conditions (5) ................................................................................................ 78
3.8 DATA COLLECTION: How the data was extracted .................................................... 80
3.8.1 Airflow Patterns ........................................................................................................... 80
3.8.2 Air Velocity ................................................................................................................. 82
3.8.3 Temperature ................................................................................................................. 83
3.8.4 Relative Humidity ........................................................................................................ 84
3.8.5 Human Models ............................................................................................................. 84
3.9 CONCLUSIONS: Development of process to yield results ........................................... 85
Chapter 4 : RESULTS FROM AIRFLOW STUDY ................................................... 87
4.1 AIRFLOW PATTERNS ................................................................................................... 87
4.2 AIR VELOCITY ............................................................................................................... 89
4.3 AIR TEMPERATURE ..................................................................................................... 92
4.4 RELATIVE HUMIDITY.................................................................................................. 94
4.5 HUMAN MODEL DATA ................................................................................................. 96
4.6 CONCLUSIONS: Raw data representation ................................................................... 97
Chapter 5 : THERMAL COMFORT ANALYSIS OF AIRFLOW STUDY ............ 99
5.1 ANALYSIS BY VARIABLE: Velocity, temperature, and relative humidity .............. 99
5.1.1 Velocity...................................................................................................................... 100
5.1.2 Temperature ............................................................................................................... 103
5.1.3 Relative Humidity ...................................................................................................... 105
5.2 PMV & PPD THERMAL COMFORT ANALYSIS .................................................... 108
5.3 ADAPTIVE THERMAL COMFORT ANALYSIS ..................................................... 117
5.4 FORM ANALYSES ........................................................................................................ 120
5.4.1 Saddle Shape .............................................................................................................. 120
5.4.2 Wave Shape ............................................................................................................... 121
5.4.3 Arch Shape ................................................................................................................. 123
5.4.4 Point Shape ................................................................................................................ 124
Chapter 6 : CONCLUSIONS FROM AIRFLOW CFD STUDY AND THERMAL
COMFORT ANALYSIS .............................................................................................. 126
6.1 THERMAL COMFORT: Improved by fabric membrane shades ............................. 127
6.2 VELOCITY: Decreased by wave and arch shades ...................................................... 128
6.3 TEMPERATURE AND RELATIVE HUMIDITY: Unknown correlation to form . 128
6.4 APPLICATION OF CONCLUSIONS .......................................................................... 129
6.5 VALIDITY OF STUDY .................................................................................................. 130
6.6 PROCESS OVERVIEW ................................................................................................. 132
v

6.7 SUMMARY ..................................................................................................................... 133
Chapter 7 : FUTURE WORK FOR FABRIC MEMBRANE AIRFLOW AND
THERMAL COMFORT ANALYSES ........................................................................ 135
7.1 CONTINUING WORK .................................................................................................. 135
7.2 OTHER WORK .............................................................................................................. 136
7.3 CONCLUSIONS ............................................................................................................. 138
BIBLIOGRAPHY ......................................................................................................... 139
Appendix A: Additional Tables for Human Comfort Metrics ................................. 143
Table A.1 ................................................................................................................................ 143
Table A.2 ................................................................................................................................ 144
Appendix B: Detailed Climate Data ............................................................................ 145
Kuala Lumpur Climate Data ............................................................................................... 145
Munich Climate Data ............................................................................................................ 149
Abu Dhabi Climate Data ...................................................................................................... 153
Appendix C: Fabric Shade Dimensions ...................................................................... 157
Figure C.1 .............................................................................................................................. 157
Figure C.2 .............................................................................................................................. 158
Figure C.3 .............................................................................................................................. 159
Figure C.4 .............................................................................................................................. 160
Appendix D: Additional Airflow Patterns .................................................................. 161
Figure D.1 .............................................................................................................................. 161
Figure D.1 .............................................................................................................................. 162
Appendix E: Additional PMV and PPD Color Maps ................................................ 163
Figure E.1 ............................................................................................................................... 163
Figure E.2 ............................................................................................................................... 164

vi

LIST OF TABLES

Table 2.1: Predicted Mean V ote sensation scale. ............................................................. 22
Table 3.1: Test scenarios for different geometry and solution parameters. ...................... 51
Table 3.2: Results for sensitivity analysis.. ...................................................................... 53
Table 3.3: Description of site conditions used for CFD simulations. .............................. 54
Table 3.4: Material assignment for each simulation by geometry component. ............... 66
Table 3.5: Boundary conditions as assigned to the model. .............................................. 71
Table 4.1: Velocity averages for all three climate scenarios. ........................................... 90
Table 4.2: Air temperature averages for all three climate scenarios. ............................... 92
Table 4.3: Relative humidity averages for all three climate scenarios. ............................ 94
Table 5.1: Results ordered from lowest to highest. ........................................................ 100
Table 5.2: Percent difference charts for velocity.. .......................................................... 101
Table 5.3: Percent difference charts for temperature. .................................................... 104
Table 5.4: Percent difference charts for relative humidity. ............................................ 107
Table 5.5: Dew point temperature averages. .................................................................. 108
Table 5.6: Values and derivations of variables used to calculate PMV and PPD. ..........110
Table 5.7: PMV and PPD averages for each scenario. ....................................................114
Table 5.8: Mean monthly outdoor temperature and indoor operative temperatures. ......118

vii

LIST OF FIGURES

Figure 1.1: Rolf Gutbrodt and Frei Otto’s West Germany Pavilion ................................. 4
Figure 1.2: Wind ventilation, stack ventilation, and Bernoulli’s Principle ........................ 5
Figure 1.3: The four fabric membrane forms studied. ....................................................... 7
Figure 1.4: Fabric membrane structures in Kuala Lumpur, Munich, and Abu Dhabi. ...... 8
Figure 2.1: An illustration from 1878 of Bedouin tents (Image Ebers 1878, 88). ........... 12
Figure 2.2: The 1972 Olympic Stadium in Munich (Image Mojtahedi 2006). ................ 13
Figure 2.3: Two different anticlastic structures. .............................................................. 15
Figure 2.4: Surface conditions of membrane structures alternative variations. ............... 17
Figure 2.5: Saddle shape, wave shape, arch shape, and point shape ............................... 18
Figure 2.6: PPD as a function of PMV (Image ASHRAE 2010, 5). ................................ 24
Figure 2.7: Adaptive thermal comfort (Image ASHRAE 2010, 9). ................................. 25
Figure 2.8: 3D view of analysis model (Image Palmer 2003, 996). ................................ 35
Figure 2.9: 3D view of existing structure (Image Palmer 2003, 997). ............................ 36
Figure 2.10: Plot of air speed on a horizontal plane (Image Palmer 2003, 997-8). ......... 36
Figure 3.1: Software workflow developed ...................................................................... 40
Figure 3.2: Screenshot from Formfinder. ......................................................................... 41
Figure 3.3: The screenshot as exported from Formfinder into AutoCAD®. ................... 42
Figure 3.4: The ability to launch the geometry into Simulation CFD from Revit®. ....... 43
Figure 3.5: Suggested flow geometry dimensions for external airflow. .......................... 47
Figure 3.6: Refinement region ......................................................................................... 48
Figure 3.7: Convergence plot after 356 iterations achieving a steady state solution. ...... 49
Figure 3.8: Saddle shape within the flow geometry in Simulation CFD. ........................ 50
Figure 3.9: Examples of outdoor areas with fabric membrane shades. ........................... 58
Figure 3.10: Plan, section, and axon view of site context................................................ 58
Figure 3.11: Saddle, wave, arch, and point shades over the pedestrian walkway. ........... 60
Figure 3.12: Human model family from Revit® 2013. ................................................... 61
Figure 3.13: Shown are the flow directions and flow geometries ................................... 62
Figure 3.14: Complete model geometry that was launched into Simulation CFD .......... 62
Figure 3.15: Screenshot from Simulation CFD. .............................................................. 64
Figure 3.16: All six shade conditions were simulated in the three climate scenarios. ..... 64
Figure 3.17: Geometry Tools dialogue box ..................................................................... 65
Figure 3.18: Model during the material selection process. .............................................. 66
Figure 3.19: Material selection dialogue box .................................................................. 67
Figure 3.20: Material properties of Moist Air. ................................................................. 67
Figure 3.21: Material selection dialogue box .................................................................. 68
Figure 3.22: Material selection dialogue box .................................................................. 69
Figure 3.23: Material selection dialogue box .................................................................. 70
Figure 3.24: Model during the boundary condition assignment process.. ....................... 71
Figure 3.25: Boundary condition dialogue boxes. ........................................................... 72
Figure 3.26: Each of the climate scenarios ...................................................................... 73
Figure 3.27: Boundary condition dialogue box for temperature. .................................... 74
viii

Figure 3.28: Temperature boundary condition assignments to the flow geometry. ......... 74
Figure 3.29: Humidity boundary condition dialogue box. ............................................... 74
Figure 3.30: Total heat generation boundary condition dialogue box settings. ............... 75
Figure 3.31: Mesh editing dialogue box. ......................................................................... 76
Figure 3.32: Refinement region of the pedestrian area. ................................................... 77
Figure 3.33: Final mesh pattern of the model. ................................................................. 77
Figure 3.34: Solve dialogue box with results quantities. ................................................. 78
Figure 3.35: Solve dialogue box with Advanced and Solar Heating options. ................. 79
Figure 3.36: Example of an angled plane with velocity results. ...................................... 81
Figure 3.37: Example of velocity flow vectors denoting airflow patterns. ..................... 81
Figure 3.38: Air velocity exported to Excel. .................................................................... 82
Figure 3.39: Example of air velocity data mapped on a 2D plane. .................................. 83
Figure 3.40: Air temperature exported to Excel. ............................................................. 83
Figure 3.41: Relative humidity exported to Excel. .......................................................... 84
Figure 3.42: PPD, PMV , and Temperature represented through color mapping ............. 85
Figure 3.43: Output of the temperatures for each human model volume. ....................... 85
Figure 4.1: Cross section of data from the x-y plane. ...................................................... 87
Figure 4.2: Cross section of data from the x-z plane and from the from the y-z plane. .. 88
Figure 4.3: Airflow patterns from the x-y plane .............................................................. 89
Figure 4.4: Analyses levels where air velocity was measured. ........................................ 90
Figure 4.5: Velocity averages for Kuala Lumpur, Munich, and Abu Dhabi. ................... 91
Figure 4.6: Temperature averages for Kuala Lumpur, Munich, and Abu Dhabi. ............ 93
Figure 4.7: Relative humidity averages for Kuala Lumpur, Munich, and Abu Dhabi..... 95
Figure 4.8: Locations and labels of human models in site context. ................................. 96
Figure 4.9: PPD (left) and PMV (right) data on human models ...................................... 97
Figure 5.1: Temperature offset graph (Image ASHRAE 2010, 6). ................................ 102
Figure 5.2: Color maps . ................................................................................................. 111
Figure 5.3: Color map inserted in the model plan. ......................................................... 111
Figure 5.4: Color maps of PPD for all climate scenarios. ..............................................113
Figure 5.5: Percent improvement for PPD measurements. .............................................115
Figure 5.6: Temperatures mapped over adaptable thermal comfort model. ...................119
Figure 5.7: Overlapping pattern of the saddle shades. ................................................... 120
Figure 5.8: Zigzag profile of the wave shade outlined in red. ....................................... 122
Figure 5.9: Zigzag profile of the arch shade outlined in red. ......................................... 124
Figure 5.10: Smooth profile of the point shade outlined in red. .................................... 125

ix

ABSTRACT

In the current building industry, fabric membranes are often a choice construction when
structural behavior, geometry, and cost are determining factors. Environmental
performance is only considered an added benefit, and the research and analysis is
commonly an afterthought. As fabric membranes become more present in the modern
building industry, their environmental performance qualities need to be further evaluated
and measured against energy standards. They offer unique performance qualities in the
metrics of daylighting, solar radiation, and possibly acoustics. A limited amount of
research has been conducted in how fabric membranes can enhance natural ventilation
and occupant thermal comfort, but it can be proposed that when studying airflow
patterns, fabric membrane structures will also behave differently from a static building
material. Furthermore, they will improve human thermal comfort as a result of their
unique anticlastic geometries.

This thesis examines what airflow patterns occur within fabric membrane structures and
how architects and engineers can use these forms to enhance occupant thermal comfort in
their designs. The study assesses four basic fabric forms serving as shades over a
pedestrian area through airflow patterns, air velocities, temperatures, humidity, Predicted
Mean V ote (PMV), and Predicted Percentage of Dissatisfied (PPD) in three different
climates: hot and humid, temperate, and dry and arid. The metrics were measured in
terms of human comfort defined by ASHRAE Standard 55 and ISO 7730.

1

CHAPTER 1 : INTRODUCTION TO AIRFLOW FOR FABRIC MEMBRANE
FORMS

High performance and large-scale fabric membranes are becoming increasingly popular
in the modern building industry. They are lightweight and possess many unique qualities
when compared to traditional building solutions such as concrete and steel. Partly as a
result of the distinctive material properties of fabric, there is still an abundant amount of
research to be done on how these membranes behave structurally and environmentally.
These unique forms, serving as shading structures, affect airflow patterns, air velocities,
air temperatures, and relative humidity and these parameters are analyzed based on their
improvements in thermal comfort.

TERMINOLOGY: Language use for this thesis
Below are common terms related to fabric membrane forms and thermal comfort.
Additional terms can be found in Chapter 2 that are more specific to the background and
process of the research topic.

 Airflow: This term refers to the air movement as a result of an input wind
direction and speed. It also takes into account the behavior of the airflow
properties in response interacting with the model geometry. Airflow properties
include airflow patterns, air velocities, and air temperatures.

2

 Computational Fluid Dynamics (CFD): CFD is a method of solving fluid flow
problems with computer software. The CFD software referenced is Autodesk
®

Simulation CFD (Autodesk
®
Simulation CFD 360 2012). The software is used to
measure airflow patterns, air velocities, air temperatures, relative humidity, and
human comfort metrics for a designed scenario. More information about CFD can
be found in section 2.2.
 Fabric (membrane) form or shape: This term describes the materiality and
geometry of the architectural components studied. It specifically refers to the
anticlastic geometry of the shading device used in the simulation models.
 Thermal comfort: Also known as human or occupant comfort, thermal comfort
refers to an occupant’s satisfaction with the surrounding thermal conditions. Its
determining factors include metabolic rate, clothing insulation, air temperature,
radiant temperature, air velocity, and relative humidity. Thermal comfort will be
evaluated through air velocities, air temperatures, and relative humidity for this
study. Additionally, through CFD analysis, thermal comfort will be measured
from Predicted Mean Vote and Predicted Percentage of Dissatisfied results.
 Predicted Mean Vote (PMV): PMV is the prediction of the average vote of a
group of people. The vote is on a seven point thermal sensation scale described as
the ASHRAE thermal sensation scale.
 Predicted Percentage of Dissatisfied (PPD): PPD predicts the percentage of
occupants that will be dissatisfied with the thermal conditions. It is a function of
PMV, given that as PMV moves further from 0, or neutral, PPD increases.

3

1.1 HYPOTHESIS STATEMENT: Informed Form
Through a computational fluid dynamic (CFD) study of fabric membrane shade
structures in an exterior environment, it will be shown that

i) Fabric membrane forms positively influence the airflow patterns, air velocity,
temperature, and relative humidity of the space they are covering.
ii) The positive influence will improve occupant thermal comfort conditions as
measured by PMV and PPD defined by ASHRAE 55 and ISO 7730 when
compared to scenarios with no shades or flat shades.

1.2 IMPORTANCE: Why study airflow for fabric membrane structures?
Fabric membrane structures have long history, but they were recently reintroduced with
great force in the second half of the 20
th
century (Siedel 2009, 2). Since then, they have
become increasing popular within the building industry and have been the focus of new
research and development. The benefits of utilizing a fabric membrane structure in
comparison to traditional building solutions and processes include structural
performance, environmental performance, and cost effectiveness.

4

Figure 1.1: Rolf Gutbrodt and Frei Otto’ s West Germany Pavilion at the Montreal Expo in 1967. This is
considered the first modern fabric membrane structure and kick started the renewal of fabric membrane
architecture (Image Dutfield 2008).

Due to their structural performance, membranes allow for large uninterrupted spans
(Siedel 2009, 5). They are also very lightweight, which is becoming more relevant as the
building industry shifts focus to sustainable design principles (Bechthold 2008, 2). Their
lightweight aspect contributes to a low embodied energy while the material’s
translucency allows natural daylight to enter the covered or enclosed space but can still
have insulating properties (BirdAir 2011). Reduced material usage and relying on passive
design strategies contributes to fabric membranes’ cost effectiveness when compared
with traditional building materials and methods.

1.2.1 Current Environmental Research
Current research in the realm of environmental performance includes focused studies on
solar radiation, daylighting, and acoustic analysis (BirdAir 2012, Croome 1985,
Hernandez 2006, Huntington 2004, Koch 2004, Mark 2011, Roh 2008). However,
research and information linking natural ventilation, thermal comfort, and fabric

5

membrane structures is lacking. Design sheets and propaganda for projects featuring
fabric membranes quite often mention the fabric membrane “encourages natural
ventilation,” but it is rare to find evidence of such research or analysis supporting the
claims (Urban Environmental Programme 2010). Especially when designing for thermal
comfort, natural ventilation cannot be used as a casual feature to mention in design briefs.
Natural ventilation is also not something that can always be assumed. Wind can provide
natural ventilation, but so can air temperature differences caused by solar radiation,
pressure differences known as stack ventilation, and Bernoulli’s Principle (Mitchell
Science 2011). Wind ventilation takes advantage of the natural force of the wind to move
air through a space. Stack ventilation occurs when solar radiation heats up the air. Hot air
has lower pressure, and therefore it rises, moving the air as a result of temperature
difference. Bernoulli’s Principle is air movement as a result of pressure differences. The
greater the distance from the ground, the faster air will move. Faster moving air has lower
pressure, which allows for the air to be drawn up through a space (Autodesk
®

Sustainability Workshop).

Figure 1.2: Wind ventilation, stack ventilation, and Bernoulli’ s Principle from left to right.

6

1.2.2 Research Contributions
More research in this field is valuable because it currently only exists in very limited
forms. With regards to architecture, airflow and thermal comfort analyses for fabric
structures is nearly nonexistent. As fabric membranes become more popular and widely
used in designs, it is essential that architects and engineers understand how their forms
and design decisions are influencing their occupants.

This research will shape the way architects and engineers design and form fabric
structures. If thermal comfort is of high importance, and it almost always is, the results
will inform the form of the structure. Additionally, given a specific climate, the resultant
delivery of the research could prove useful in the design phase. Most importantly, the
research will serve as a stepping stone for future thermal comfort analyses for fabric
membrane forms.

1.3 STUDY BOUNDARIES: Defining the research limits
The purpose is to evaluate natural ventilation opportunities for fabric membrane shade
structures and their influence on thermal comfort for an external environment. Natural
ventilation opportunities include air movement caused by wind and the stack effect. The
study does not include the influence of mechanical HVAC systems. Fabric membrane
structures include tensile surface structures in which the geometry curvature is anticlastic
and fabric is the surface membrane material. The scope of this work encompasses studies
four fabric shade forms (Figure 1.3) and evaluates their airflow capabilities through

7

airflow pattern, air velocity, temperature, and relative humidity results in addition to the
human comfort metrics of Predicted Mean Vote and Percent Persons Dissatisfied.

Figure 1.3: The four fabric membrane forms studied. From left to right: saddle shape, wave shape, arch
shape, and point shape.

Although there are many aspects of fabric membranes that are unique not all of them are
covered in this study. The following features of fabric membranes are not covered:
thermal comfort influenced by material properties (insulation values), daylighting
opportunities, acoustic properties, and structural properties. Solar radiation is only
applicable in its relevance to the stack effect.

In regards to structural properties, only the geometric forms of the fabric shade structures
are considered. Supports, connections, and structural forces are not considered. Material
properties of the fabric are not analyzed. The reasons for this are discussed in Chapter 3.

The fabric forms were analyzed in three different climates; hot and humid, temperate, and
dry and arid. The climate data was modeled after the following cities: Kuala Lumpur,
Malaysia (hot and humid); Munich, Germany (temperate); Abu Dhabi, United Arab
Emirates (dry and arid).

8

The selection of these climates was based on where current and future plans for fabric
membrane structures exist. These climates were also selected to produce variation
amongst the input values of air speed, air direction, air temperature, and relative
humidity. Only the annual mean wind speeds and directions are considered. No wind
speeds that would be classified as extreme, such as hurricane force winds, are simulated.

Bukit Nanas Monorail Station, Kuala Lumpur
(Image Wikimedia Commons 2006)

Tierpark Hellabrunn, Munich
(Image Archi Expot 2009)

Yas Island Gateway Park, Abu Dhabi
(Image Light Weight Structures Architectural Solutions 2011)

Figure 1.4: Fabric membrane structures in Kuala Lumpur, Munich, and Abu Dhabi.

9

1.3.1 Static Behavior of the Fabric
As demonstrated by sails working with the wind, fabric behaves dynamically when it
comes in contact with airflow. This phenomena gives fabric a unique quality not seen in
other traditional building materials and can influence the behavior of airflow within
fabric membrane structures. However, given the methodology of performing a CFD
simulation in this research, the dynamic response of fabric is not included. The rationale
from this decision comes from wide-ranging research of how to execute such an analysis.
The initial background study into research methods did not yield encouraging results.
Within the building sector, a dynamic simulation of fabric membranes could not be
found. In fact, there was evidence supporting this claim stating that “at this time no
software with sufficient functionality and efficiencies is available to generate a fabric
model for the CFD simulation to determine the fluid flow properties of the textile” (Reif
2011, 78).

One example of dynamic responsive fabric analysis was thought to be found. North Sails,
an international sail making company, uses a proprietary software set known as the North
Design Suite that appears to have dynamic response capabilities. However, since the suite
uses 15 modules and access to the suite is for internal use only, the extent to which the
suite can measure dynamic performance is questionable (North Sails 2013).

Since the focus of this thesis is on the human comfort scale, wind speeds are already so
low that dynamic response from the fabric will be insignificant, if not negligible.

10

1.4 DELIVERABLE: A process for designing and analyzing fabric forms
The airflow study results for fabric membranes will be presented as a deliverable that not
only provides new research findings to the building industry, but it also serves as a set of
reference guidelines for designers and engineers when designing fabric membrane
projects. The intent of the guidelines is to serve as an available resource that can be used
to inform fabric membrane design decisions given set environment and climate scenarios.
Results also include a process workflow that architects and engineers can use to design
and test fabric membrane structures in CFD simulation software.

1.5 CHAPTER STRUCTURE: Outline of the study
Chapter 1: Introduction to Airflow Investigation for Fabric Membrane Forms
Chapter 2: Background for Fabric Membranes, Thermal Comfort, and CFD Analysis
Chapter 3: Method and Process of Studying Airflow for Fabric Membrane Forms through
CFD
Chapter 4: Results from Airflow Study
Chapter 5: Thermal Comfort Analysis of Airflow Study
Chapter 6: Conclusions from Airflow Study and Thermal Comfort Analysis
Chapter 7: Future Work for Fabric Membrane Airflow and Thermal Comfort Analyses

11

CHAPTER 2 : BACKGROUND FOR FABRIC MEMBRANES, THERMAL
COMFORT AND CFD ANALYSIS

In order to study the link between fabric membrane forms and thermal comfort properties
their respective backgrounds must first be considered. This chapter documents how and
why fabric membranes exhibit unique geometries (2.1 Fabric Membranes), the basis of
thermal comfort analysis (2.2 Thermal Comfort), and how computational fluid dynamics
can be used to study airflow properties and thermal comfort (2.3 CFD Analysis for
Airflow & Thermal Comfort).

2.1 FABRIC MEMBRANES: Defining the form
This section describes fabric membranes with respect to history, classification, material
properties, and benefits. This information can be used to understand the unique properties
of fabric membranes and how the forms can influence airflow properties.

2.1.1 Origins of Fabric Membranes
Human use of fabric shelter structures can be dated back several thousand years. These
tents depicted in Figure 2.1 were woven from animal hair and used by Arabian Bedouins
as well as nomads from Mauritania, Libya, Saudi Arabia, and Tibet (Bechthold 2008, 19).
These structures had some similarities with modern fabric membrane structures. They
consisted of multiple panels sewn together and supported with wooden poles and tied
down with ropes. Some ancient tents featured the double curvature that classifies fabric

12

structures as anticlastic structure systems.

Figure 2.1: An illustration from 1878 of Bedouin tents (Image Ebers 1878, 88).

Another historical example of vernacular fabric structures is the tipi dwellings of the
Native Americans. The tipi’s featured animal hides wrapped around poles leaning against
each other. These structures featured ventilation flaps that could be opened or closed
depending upon the climate and oriented based on the season (Bechthold 2008, 20).

A notable use of fabric is sails on a ship. Sails are an early example of pneumatic
structures that get their shape from differing air pressures. The sails were moved and
adjusted based on the wind direction and speed to guide ships along their proper course.
As fabric properties and analysis technologies advanced with time, sails became more
refined and powerful.

The type and style of modern fabric membrane structures found their beginnings in the
second half of the twentieth century through the work and research of Frei Otto and the

13

Institute of Lightweight Structures in Stuttgart, Germany. An important contributions of
international acclaim include first large scale use of fabric with a cable net for Expo 64,
in Lausanne, Switzerland (Schierle 2008, 337) and the 1967 Montreal World Expo
pavilion with Rolf Gutbrodt featuring a cable net roof with a membrane cover. The 1972
Olympic Stadium in Munich, designed by Günther Behnisch in collaboration with Otto,
was the first permanent roof structure to use a pre-stress cable net on a large-scale
project and set the stage for today’s modern fabric membrane structures (Figure 2.2).

Figure 2.2: The 1972 Olympic Stadium in Munich is considered one of the early modern anticlastic
structures (Image Mojtahedi 2006).

Today’s fabric membrane structures are being used as temporary and long-term structures
for many reasons. Not only do they provide unique geometric forms, but they can also be
cost effective, have a short construction time, and have unique material properties such as
translucency to provide natural light entering the space.

14

2.1.2 Classifications
Fabric membranes are classified as synclastic and anticlastic structures that are nested
within tensile structures. The hierarchy and a description of qualities are explained further
in this section.

Tensile Structures
A tensile structure exhibits construction members that resist loads in tension. This
classifies tensile structures as form active systems (Siedel 2009, 1). They are more
efficient structures when compared to bending and compression structures because they
use the material to its full capacity. Under compression, tensile members become slack.
To avoid this, pre-stress can be introduced which will allow members to absorb
compressive stresses without becoming slack and unstable. They are often used as roofs
and can span long, uninterrupted distances (Schierle 2008, 308). Tensile structures
include the following:
 Stayed structures,
 Suspended structures,
 Cable trusses,
 Pneumatic structures, and
 Anticlastic structures.

The fabric membranes that are studied in this thesis are classified as anticlastic structures,
and therefore only this category will include further explanation.

15

2.1.2.1 Anticlastic Structures
Anticlastic structures are flexible membranes that feature principle curvatures in two
directions, which are mutually perpendicular; one direction concave, and the opposite
direction convex. The anticlastic curvature describes curves in opposing directions.
Figure 2.3 illustrates anticlastic curvature. Fabric membranes can be synclastic (air
supported) or anticlastically curved surfaces. These doubly curved surfaces are very
strong and resistant to tensile forces because the perpendicular opposing curvature
provides stability (Schierle 2008, 271). Since flexible membranes are tensile structures
that can become slack under compressive forces, the double curvature and pre-stress of
the membrane is essential for structural stability. Flat membranes are unstable because
they deform excessively under loads (Schierle 2008, 330).

Figure 2.3: Two different anticlastic structures with the perpendicular curvature directions in red arrows.

2.1.2.1.1 Minimal Surfaces
A minimal surface describes a surface that spans an area using the smallest possible area.
Soap films are an example of minimal surfaces (Brakke 2009). They can be flat or
anticlastic. They are formed by cable nets or fabric membranes. Minimal surfaces require
three conditions;

16

1) Minimum surface area between any boundary,
2) Equal and opposite curvature at any point, and
3) Uniform stress throughout the surface (Schierle 2008, 331).

2.1.2.1.2 Membrane Structures
Membrane structures are anticlastic structures in which the spanning material is fabric.
As described for tensile structure members, fabric does not carry loads in compression
because it is flexible. Fabric must be pre-stressed and the curvature taunt so it can support
gravity or wind uplift loads.

There are three different types of edge conditions that are found in fabric membrane
structures;
 Edge cables, which carry loads in only tension,
 Edge arches, in which the rigid arch transfers the load from the membrane
through compression, and
 Edge beams or truss, in which the rigid beam or truss transfers the load from the
membrane through bending or axial stress respectively (Schierle 2008, 333).

Similar to edge conditions, anticlastic membranes have unique surface conditions. The
surface conditions include:
 Saddle shape,
 Wave shape,

17

 Arch shape, and
 Point shape.

These shapes are common, with many variations existing in architecture. Figure 2.4
illustrates these surface conditions.

Figure 2.4: Surface conditions of membrane structures alternative variations.

Saddle Shape
Saddle shapes consist of a single surface of anticlastic curvature of various possible
plans. The edges alternate in high and low points, which generates the curvature. The
plan of the membrane can be any symmetrical or asymmetrical shape, such as a square,
hexagon, octagon, or other polygons, excluding triangles. It can be seen in architecture
applications as a single form or as multiple iterations.

18

Wave Shape
Wave shapes are repetition of anticlastic panels. The shape is characterized by ridges and
valleys to create a folded appearance.

Arch Shape
Arch shapes feature a rigid arch that spans the membrane. They may use pin supports so
that they can resist lateral forces.

Point Shape
Point shapes feature high or low points, typically in the center of the membrane that are
supported by a single or multiple masts. A tension ring, loop, or dish can be used at the
support point to assure the membrane does not puncture due to stress concentration.

Figure 2.5: Saddle shape (Image Structureflex), wave shape (Image BirdAir), arch shape (Image
Innovative Tensile), and point shape (Image Fabric Architecture).

19

2.1.3 Components of fabric membranes
2.1.3.1 Fabric Membrane
Fabric membranes are constructed with high strength fabric, also known as structural
fabric. The material spans between supports and carries the tensile stresses due to self-
weight and environmental loads such as wind and snow. For all permanent uses, the
material needs to be at least weatherproof and fire resistant. Temporary structures do not
always require fire rated fabric. For enclosed environments, the material also needs to be
air tight and water tight.

The material properties of structural fabric are rapidly advancing. Light transmission and
reflectance, heat transmission and reflectance, and sound insulation are all features of
different types of properties integrated into structural fabrics.

Structural fabrics are woven using fibers that are treated to represent the material
properties. Fiberglass or polyester is typically the nature of the fiber, with coatings or
laminates that can enhance structural performance, lifetime, light transmission and
reflectance, heat transmission and reflectance, sound insulation, and water proofing.

2.1.3.2 Connections
The connections of fabric membranes are an essential part to their unique anticlastic
curvature and dictate the way in which forces are transferred. The scope of this thesis

20

does not include connections for fabric membranes, however typical considerations are
listed below.

 Fabric to fabric connections
 Fabric to edge connections
 Fabric to supporting member connections

2.1.4 Benefits to designing with fabric
Fabric membranes are able to span long distances with minimal material. This allows for
large uninterrupted interior spaces. Additionally, fabric membranes are fast to erect and
cost effective. They offer various unique energy solutions. Fabric membranes are capable
of allowing light to penetrate them, to provide daylighting, decreasing the need for
artificial lighting. New technologies are also allowing fabric membranes to feature
thermal insulation that is broadening the types of climate they can be used in. Fabric
membranes are a unique structural system with growing popularity.

2.2 THERMAL COMFORT: Measuring occupant comfort
According to the ANSI/ASHRAE Standard 55-2010, thermal comfort is defined as “that
condition of mind which expresses satisfaction with the thermal environment and is
assessed by subjective evaluation” (ASHRAE 2010, 2). Also known as human comfort,
thermal comfort is the occupants’ satisfaction with the surrounding thermal conditions

21

and is essential to consider when designing a structure that will be occupied by people.
There are six factors to take into consideration when designing with thermal comfort. Its
determining factors include the following:

 Metabolic rate (met): the energy generated from the human body,
 Clothing insulation (clo): the amount of thermal insulation the person is wearing,
 Air temperature: temperature of the air surrounding the occupant,
 Radiant temperature: the weighted average of all the temperatures from surfaces
surrounding an occupant,
 Air velocity: rate of air movement given distance over time, and
 Relative humidity: percentage of water vapor in the air.

The environmental factors include temperature, radiant temperature, relative humidity,
and air velocity. The personal factors are activity level (metabolic rate) and clothing.
Tables with additional information for the personal factors can be found in Appendix A.

Thermal comfort is calculated as a heat transfer energy balance. Heat transfer through
radiation, convection, and conduction are balanced against the occupant’s metabolic rate.
The heat transfer occurs between the environment and the human body, which has an area
of 2.472 ft
3
. If the heat leaving the occupant is greater than the heat entering the
occupant, the thermal perception is “cold.” If the heat entering the occupant is greater
than the heat leaving the occupant, the thermal perception is “warm” or “hot.”

22

A method of describing thermal comfort was developed by Ole Fanger and is referred to
as Predicted Mean V ote (PMV) and Predicted Percentage of Dissatisfied (PPD).

2.2.1 Predicted Mean Vote (PMV)
PMV is the prediction of the average vote of a group of people. The vote is on a seven
point thermal sensation scale described as the ASHRAE thermal sensation scale. The
votes (
Table 2.1) result from relating the six thermal comfort factors to each other through heat
balance principles.

Value Sensation
-3 Cold
-2 Cool
-1 Slightly cool
0 Neutral
1 Slightly warm
2 Warm
3 Hot

Table 2.1: Predicted Mean Vote sensation scale.

Fanger’s thermal comfort model is used to calculate PMV in the following equation
(Fanger 1970, Fanger 1982).

23

𝑃𝑀𝑉 =,0.303𝑒 ;0.036𝑀 + 0.028-*(𝑀 −𝑊 )−3.96𝐸 ;8
𝑓 𝑐𝑙
,(𝑡 𝑐𝑙
+ 273)
4
−(𝑡 𝑟 + 273)
4
-
−𝑓 𝑐𝑙
𝑕 𝑐 (𝑡 𝑐𝑙
−𝑡 𝑎 )−3.05,5.73−0.007(𝑀 −𝑊 )−𝑝 𝑎 -−0.42,(𝑀 −𝑊 )
−58.15-−0.0173𝑀 (5.87−𝑝 𝑎 )−0.0014𝑀 (34−𝑡 𝑎 )+

With
𝑓 𝑐𝑙
=
1.0+ 0.2𝐼 𝑐𝑙
1.05+ 0.1𝐼 𝑐𝑙

𝑡 𝑐𝑙
= 35.7−0.0275(𝑀 −𝑊 )−𝑅 𝑐𝑙
*(𝑀 −𝑊 )−3.05,5.73−0.007(𝑀 −𝑊 )−𝑝 𝑎 -
−0.42,(𝑀 −𝑊 )−58.15-−0.0173𝑀 (5.87−𝑝 𝑎 )
−0.0014𝑀 (34−𝑡 𝑎 )+

𝑅 𝑐𝑙
= 0.155𝐼 𝑐𝑙

𝑕 𝑐 = 12.1(𝑉 )
1 2 ⁄

Where e Euler’s number (2.718)
f
cl
clothing factor
h
c
convective heat transfer coefficient
I
cl
clothing insulation [clo]
M metabolic rate [W/m
2
] 115 for all scenarios
p
a
vapor pressure of air [kPa]
R
cl
clothing thermal insulation
t
a
air temperature [°C]
t
cl
surface temperature of clothing [°C]
t
r
mean radiant temperature [°C]
V air velocity [m/s]
W external work (assumed = 0)

The recommended acceptable PMV range for thermal comfort from ASHRAE 55 is
between -0.5 and +0.5 for an interior space.

24

2.2.2 Predicted Percentage of Dissatisfied (PPD)
PPD predicts the percentage of occupants that will be dissatisfied with the thermal
conditions. It is a function of PMV , given that as PMV moves further from 0, or neutral,
PPD increases. The graphical relationship and equation can be seen in Figure 2.6: PPD as
a function of PMV (Image ASHRAE 2004, 5)..

𝑃𝑃𝐷 = 100−95𝑒 ,;(0.3353𝑃𝑀𝑉 4
:0.2179𝑃𝑀𝑉 2
)-

Figure 2.6: PPD as a function of PMV (Image ASHRAE 2004, 5).

The recommended acceptable PPD range for thermal comfort from ASHRAE 55 is less
than 10% persons dissatisfied for an interior space.

2.2.3 Adaptive Thermal Comfort
For interior spaces that rely on natural ventilation the occupants respond to climate
conditions rather than HV AC settings. The space being occupied must have operable
windows, no mechanical cooling system, and the occupants must be near sedentary and
have a metabolic rate between 1.0 and 1.3 met. Also, occupants have the option of adding
or removing clothing to adapt to the thermal conditions. Given these conditions, Figure

25

2.7 shows the adaptable comfort range. The range uses two levels to determine acceptable
indoor comfort; 80% acceptability is for typical applications, and 90% acceptability is
used for a higher standard of thermal comfort (ASHRAE 2004, 9).

Figure 2.7: Adaptive thermal comfort given outdoor air temperature and indoor operative temperature
(Image ASHRAE 2004, 9).

Outdoor temperatures beyond the extent of the graph should not be extrapolated. This is
because outdoor temperatures below 50°F and above 92.3°F are not eligible for the
adapted comfort model. Additionally, for this model there are no air velocity or humidity
limits.

The indoor operative temperature is the weighted average of air temperature and mean
radiant temperature. Air temperature is weighted by the convective heat transfer
coefficient (h
r
) and the mean radiant temperature is weighted by the linear radiant heat
transfer coefficient (h
c
) of the occupant.
𝑡 𝑜 =
(𝑕 𝑟 𝑡 𝑟 + 𝑕 𝑐 𝑡 𝑎 )
𝑕 𝑟 + 𝑕 𝑐

26

2.3 CFD ANALYSIS: For airflow and human comfort
This section describes the principles and theories behind computational fluid dynamics
and the application of using CFD analysis to study airflow and human comfort. There are
many CFD programs designers and engineers can use. Autodesk
®
Simulation CFD was
chosen; an explanation of why is in Chapter 3. Throughout the rest of this section, the
equations and calculation methods pertain to Autodesk
®
Simulation CFD.

2.3.1 Introduction to CFD
Computational Fluid Dynamics (CFD) is a method of solving fluid flow problems
through computer software that uses numerical methods and algorithms. It is used to
analyze fluid flow in a wide variety of engineering industries, from the micro level of
measuring heat flow through a small electrical circuit, to the large scale aerodynamics of
airflow over an airplane. For building designers, CFD analysis can measure large scale
wind patterns through a city plan or analyzing airflow from natural ventilation and HV AC
systems within buildings. In some CFD programs, human models can even be evaluated
to measure human comfort within the rooms.

2.3.1.1 Terminology
The following terms are used to describe the fundamentals of the CFD analysis
conducted in this thesis. Further description of how they applied to the simulation
scenarios can be seen in Chapter 3. The terms are specific in regards to the process and

27

application of Autodesk
®
Simulation CFD.

 Boundary Conditions: The parameters that describe the fluid flow problem.
These include input velocities, pressures, temperatures, and humidity (Autodesk
2012).
 Compressible Flow: Flow in which there is a direct relationship between density
and pressure, in that a change in density will affect pressure, and vice versa.
Downstream pressures do not affect any part of the upstream flow. This yields a
hyperbolic pressure equation and requires only an upstream boundary condition.
Most scenarios concerning the flow of gases, particularly at high speeds, are
considered compressible flow problems (Autodesk 2012). Compare to
Incompressible Flow.
 Conduction: One of the three modes of heat transfer. Describes heat flow
through objects in physical contact. Caused by molecular motion traveling from a
high temperature to a lower temperature. Typically occurs in solid objects
(Engineering Toolbox). Compare to Convection and Radiation.
 Convection: One of the three modes of heat transfer. Describes heat flow through
an object and its environment as a result of fluid motion. Typically occurs in
liquids and gases (Engineering Toolbox). Compare to Conduction and
Radiation.
 Flow: A solving option that enables fluid pressure and momentum equations.
Disabling flow would be for a heat transfer analysis through conduction only.

28

Enabled with Heat Transfer it creates a natural convection scenario (Autodesk
2012).
 Flow Geometry: The 3D shape that defines the physical extents in which the
analysis will occur (Autodesk 2012).
 Incompressible Flow: Also known as isochoric flow. Flow in which the density
is constant. Downstream pressures do affect upstream flows, resulting in an
elliptic pressure equation. Additionally, downstream boundary conditions are
required. While no fluid is actually incompressible, density variance can be
assumed negligible when the Mach number (Ma) is less than 0.3. In most
scenarios, liquid flow is incompressible (Autodesk 2012). Compare to
Compressible Flow.
 Heat Transfer: A solving option that considers heat transfer effects. This
includes conduction, convection, and radiation if additionally enabled (Autodesk
2012).
 Laminar Flow: Smooth and steady fluid motion. Typically characterized by slow
speeds and a Reynolds number (see below) less than 2,500 (Engineering
Toolbox). Compare to Turbulent and Transitional Flow.
 Mach Number (Ma): Measure of compressibility. It is fluid velocity divided by
the speed of sound. Mach numbers less that 0.3 are considered incompressible
flows (Engineering Toolbox).
 Mesh: The division of a volume into discrete cells. When the model is meshed, its
volumes are broken up into elements. The elements are defined by nodes. The

29

nodes are where the calculations are done. The mesh can be uniform or non-
uniform and is defined by the user (Autodesk 2012).
 Natural Convection: Also known as buoyant-driven flow. Temperature
differences dictate the fluid motion by controlling the density (Autodesk 2012).
 Navier-Stokes Equations: Used to describe fluid motion. The equations apply
Newton’s second law to fluid motion. The general equation is
(
𝑣
𝑡
+ 𝑣 𝑣 ) = − 𝑝 + +

Where ρ density
v velocity
p pressure
T deviatoric total stress
f body forces
(Engineering Toolbox).

 Newtonian Fluid: Fluid that has a linear relationship between fluid shear and
strain. Viscosity is constant, or can be a function of temperature. Most
engineering flow problems concern Newtonian fluids. Air is also a Newtonian
fluid (Autodesk 2012).
 Radiation: One of the three modes of heat transfer. Describes heat flow through
the emission or absorption of electromagnetic radiation. When enabled as a
solution feature as in this thesis, it includes surface-to-surface radiation effects
(Engineering Toolbox). Compare to Conduction and Convection.
 Reynolds Number (Re): Dimensionless number used to determine between
laminar and turbulent flows. Calculated as

30

𝑅𝑒 =
𝑉𝐿 𝜇

Where ρ density
V velocity
L length traveled
μ viscosity
(Autodesk 2012).

 Stack Effect: Refers to air movement caused by temperature differences. Hot air
has a lower pressure, so it rises above the cool air. Also known as buoyancy
driven ventilation (Autodesk
®
Sustainability Workshop).
 Transitional Flow: A combination of turbulent and laminar flow, with turbulent
flow in the center and laminar flow near the edges (Engineering Toolbox).
Compare to Turbulent and Laminar Flow.
 Turbulent Flow: Unstable and changing fluid motion. Typically characterized by
high speeds and a Reynolds number greater than 2,300. Most engineering fluid
flow problems, and this thesis, have turbulent flows (Engineering Toolbox).
Compare to Laminar and Transitional Flow.
 Viscosity: A fluid’s resistance to flow (Engineering Toolbox).
 Wind Driven Flow: Refers to air movement caused by wind (Autodesk 2012).

2.3.2 Equations for Solving a Fluid Flow Problem
The fundamental equations of all fluid flow problems are the continuity equation, the
Navier-Stokes equations, and the energy equation. They describe fluid flow and heat

31

transfer. These equations are valid for single-phase fluid flow (which will be the only
kind of fluid flow considered in this thesis). The scenario modeled is for turbulent flow,
which requires a large number of finite elements, so much so that it can become
unreasonable to model turbulent flow with these equations. As a result, Autodesk
®

Simulation CFD uses time-averaged governing equations to solve for turbulent flow. This
assumes that a mean value and a fluctuating value about the mean can represent the
dependent variables. For example, given
u = the x-velocity component,
U = the mean velocity, and
u' = the fluctuation about the mean, the following equation can be generated.

𝑢 =𝑈 + 𝑢 ′

The averaged governing equations with this substitution are described as partial
differential equations. The result is more unknowns than equations to solve for, and
creating more equations yields more unknowns. In order to find a solution, the
Boussinesq approximation (Autodesk 2012), shown below, is used to define eddy
viscosity and eddy conductivity.

𝜇 𝑡 =
− 𝑢𝑢
2
𝑈
𝑥
=
− 𝑢𝑣
𝑈
𝑦
+
𝑉
𝑥
=
− 𝑣𝑤
𝑉
𝑧
+
𝑊
𝑦
= ⋯

𝑘 𝑡 =
− 𝐶 𝑝 𝑢𝑡
𝑇
𝑥
=
− 𝐶 𝑝 𝑣𝑡
𝑇
𝑦
=
− 𝐶 𝑝 𝑤𝑡
𝑇
𝑧

When the Boussinesq approximation is substituted in, the partial differential equations

32

become those listed below (Autodesk 2012).

Continuity equation

𝑡
+
𝑈 𝑥
+
𝑉 𝑦
+
𝑊 𝑧
= 0

X-Momentum equation

𝑈
𝑡
+ 𝑈
𝑈
𝑥
+ 𝑉
𝑈
𝑦
+ 𝑊
𝑈
𝑧
= 𝑔 𝑥 −
𝑃
𝑥
+ 𝑆 𝐷𝑅
+ 𝑆 𝜔 +

𝑥
[2(𝜇 + 𝜇 𝑡 )
𝑈
𝑥
]
+

𝑦
[(𝜇 + 𝜇 𝑡 )(
𝑈
𝑦
+
𝑉 𝑥
)]+

𝑧
[(𝜇 + 𝜇 𝑡 )(
𝑈
𝑧
+
𝑊
𝑥
)]

Y-Momentum equation

𝑉
𝑡
+ 𝑈
𝑉
𝑥
+ 𝑉
𝑉
𝑦
+ 𝑊
𝑉
𝑧
= 𝑔 𝑦 −
𝑃
𝑦
+ 𝑆 𝐷𝑅
+ 𝑆 𝜔 +

𝑥
[(𝜇 + 𝜇 𝑡 )(
𝑈
𝑦
+
𝑉
𝑥
)]
+

𝑦
[2(𝜇 + 𝜇 𝑡 )
𝑉
𝑦
]+

𝑧
[(𝜇 + 𝜇 𝑡 )(
𝑉
𝑧
+
𝑊
𝑦
)]

Z-Momentum equation

𝑊
𝑡
+ 𝑈
𝑊
𝑥
+ 𝑉
𝑊
𝑦
+ 𝑊
𝑊
𝑧
= 𝑔 𝑧 −
𝑃
𝑧
+ 𝑆 𝐷𝑅
+ 𝑆 𝜔 +

𝑥
[(𝜇 + 𝜇 𝑡 )(
𝑈
𝑧
+
𝑊
𝑥
)]
+

𝑦
[(𝜇 + 𝜇 𝑡 )(
𝑉
𝑧
+
𝑊
𝑦
)]+

𝑧
[2(𝜇 + 𝜇 𝑡 )
𝑊
𝑧
]

33

Energy equation
𝐶 𝑝 (
𝑇
𝑡
) + 𝐶 𝑝 𝑈 (
𝑇
𝑥
) + 𝐶 𝑝 𝑉 (
𝑇
𝑦
) + 𝐶 𝑝 𝑊 (
𝑇
𝑧
)
=

𝑥
[(𝑘 + 𝑘 𝑡 )
𝑇
𝑥
]+

𝑦
[(𝑘 + 𝑘 𝑡 )
𝑇
𝑦
]+

𝑧
[(𝑘 + 𝑘 𝑡 )
𝑇
𝑧
]+ 𝑞 𝑉
Where C
p
constant pressure specific heat
g
x
, g
y
, g
z
gravitational acceleration in x, y, z-directions
k thermal conductivity
q
V
volumetric heat source
T temperature
t time
u, v, w velocity component in x, y, z-direction
U, V , W mean velocity in x, y, z-direction
u', v', w' fluctuation about the mean velocity in x, y, z-direction
μ viscosity
ρ density

The energy equation used by Simulation CFD is based on the traditional energy balance
equation given as
𝑄 𝑥 + 𝑄 𝑦 + 𝑄 𝑧 + 𝑄 𝑆 =𝑄 𝑥 :𝑑𝑥
+ 𝑄 𝑦 :𝑑𝑦
+ 𝑄 𝑧 :𝑑𝑧
+ 𝑑𝑈

Where Q
x
is the heat transferred across area dy dz in time dt (Incropera 1981).

𝑄 𝑥 =−𝑑𝑦𝑑𝑥 𝑘 𝑇
𝑥
𝑑𝑡

And the heat transfer at x + dx is
𝑄 𝑥 :𝑑𝑥
=−𝑑𝑦𝑑𝑧 [𝑘 𝑇
𝑥
+

𝑥
(𝑘 𝑇
𝑥
)𝑑𝑥 ]𝑑𝑡

And similarly for the y and z directions. Q
S
is the heat transferred from the energy
generated by the source 𝑞 ̇ 𝑆 ′′′
.

34

𝑄 𝑆 =𝑞 ̇ 𝑆 ′′′
(𝑑𝑥𝑑𝑦𝑑𝑧 )𝑑𝑦

Given that humidity is considered in the simulations, the energy equation becomes

𝐶 𝑃 𝑇 𝑡
+ 𝑢
𝐶 𝑃 𝑇 𝑥
+ 𝑣
𝐶 𝑃 𝑇 𝑦
+ 𝑤
𝐶 𝑃 𝑇 𝑧
=

𝑥
[𝑘 𝑇
𝑥
]+

𝑦
[𝑘 𝑇
𝑦
]+

𝑧
[𝑘 𝑇
𝑧
]+ 𝑞 𝑉

Tracking the moisture in the fluid air requires another equation that represents the mass
fraction of the moisture.

𝑓
𝑡
+ 𝑢
𝑓
𝑥
+ 𝑣
𝑓
𝑦
+ 𝑤
𝑓
𝑧
=

𝑥
[𝐷 𝑓
𝑥
]+

𝑦
[𝐷 𝑓
𝑦
]+

𝑧
[𝐷 𝑓
𝑧
]

The previous two equations with the steam tables are used to calculate the moisture of the
air (Autodesk
®
2012).

2.4 APPLICATIONS: CFD with fabric membranes
CFD analyses are already popular in the building industry and are typically performed by
mechanical engineers. It “is increasingly being used to predict the effects of wind on
buildings and on the people in and around them” (Palmer 2003, 995). Additionally, upon
verification studies with physical modeling in wind tunnels, CFD analysis “is well suited
to studying the effects of wind speed on pedestrian comfort within and around buildings”
(Palmer 2003, 995). CFD analyses are also more flexible than creating physical models,
in that it is easy to change the model, change the direction and velocity of the air, and
repeat the analysis. Additionally, CFD analyses provide cost savings, as shorter
development times are needed because fewer experiments are needed (Reif 2011, 79).

35

One published example of conducting a CFD analysis on a fabric membrane structure
was found. The project was in Ashford, Kent, UK and was designed and analyzed by
Buro Happold. As of 2003, it was the largest continuous tensile canopy in the world at
30,000 m
2
. The goal of the simulation was to study the food court and “simulate the area
in order to engineer a solution to problems that were occurring with draughts in occupied
zones” (Palmer 2003, 996).

Figure 2.8: 3D view of the fabric membrane analysis model (Image Palmer 2003, 996).

In the food court area, the fabric membrane was modeled in great detail. However, less
detail was given to the portion of the membrane that was not near the food court. This can
be seen in Figure 2.8, with the food court located in the lower right of the model. The
mesh of the model was executed in a similar way. The food court area had a more refined
mesh, while other areas had larger mesh elements. An image from the existing scenario
analysis is seen in Figure 2.9.

36

Figure 2.9: 3D view of existing structure above the food court (Image Palmer 2003, 997).

To find a solution to prevent the draughts that were making the space too cold for
occupants, the engineers followed a parametric process of studying the effects glass
walls, membrane structures, and hedges had on limiting the draught. Given the climate
data, several wind conditions with varying directions and speeds were analyzed. A set of
these solutions are seen in Figure 2.10.

Figure 2.10: Plot of air speed on a horizontal plane at a height of 1.2 m in the model without hedges
present (left) and with low permeability hedges present (right) (Image Palmer 2003, 997-8).

The conclusions from this study were that hedges with a height of 2.25 m were capable of
reducing wind speeds by 57.55%, a desirable result.

37

CFD analysis for fabric membrane structures with considerations for airflow and thermal
comfort are entirely feasible. Based on the literature research, there have been a limited
number published, but the research also supports that through proper analyses and set up
such simulations are possible.

2.5 CONCLUSIONS: Background research contributing to the study
The preceding sections describe the fundamentals of fabric membranes, thermal comfort,
and CFD analyses. These fundamentals have contributed to the following chapters in
preparing the remainder of the study and serve as a deeper understanding for the reader.

38

CHAPTER 3 : METHOD & PROCESS OF STUDYING AIRFLOW FOR FABRIC
MEMBRANE FORMS THROUGH CFD

Airflow and thermal comfort is dependent on the fabric membrane’s geometry. This
chapter outlines the methodology and process used to pursue the following questions:
 What is the general airflow pattern for a particular fabric membrane form?
 How does the form influence the air velocity and movement?
 What are the minimum, maximum, and average air velocities?
 How do these overall patterns contribute to thermal comfort, including
temperature and humidity?

3.1 INTRODUCTION: Overview of Method & Process Development
The process developed for simulating and analyzing the airflow of fabric membranes
went through many revision and refinement phases. The first step was to develop the
process for simulation and verify that it worked. Once this was done, a sensitivity
analysis was conducted that determined the extent of influence for the variables to be
tested. Based on the knowledge gained from the sensitivity analysis and additional
research conducted to determine the scope and scale, the simulation scenarios were
designed. These included climate conditions, model geometry, and desired output data.

3.2 SOFTWARE WORKFLOW: Using software interoperability
Prior to the simulations, a development phase was used to plan the process and workflow.

39

The most important aspect of this phase was the creation of the software workflow
because it would dictate the use and collection of the results and analysis.

Significant research in computational fluid dynamic software was conducted. This
research concluded that Autodesk
®
Simulation CFD (Autodesk
®
Simulation CFD 360
2012) was a viable option for the desired analyses. Autodesk
®
Simulation CFD has
interoperability with Autodesk
®
Revit
®
(Autodesk
®
Revit
®
2012), a software that can be
used to create the site geometries for the analysis models. In addition, Simulation CFD
has well supported learning materials through the featured Fundamentals Training and
example tutorial models available. The extensive wikiHelp web site featured
fundamental, theoretical, and practical application articles that ensure the user has high
level and detailed knowledge of the software and its methods. Additional software that
was researched but not used in this research includes Cradle’s scSTREAM CFD (Cradle
scSTREAM 2011), CFDmax, Integrated Environmental Solutions (IES) (Integrated
Environmental Solutions Limited 2011), and STAR-CCM+. These programs were not
used for reasons that include lack of support for architectural study purposes, unclear
documentation and difficulty to learn, and reluctance of providers to issue a student
license.

After selecting the CFD program to use, a software workflow was developed that would
dictate the majority of the process. Based on the afore mentioned research and decision
making, the software workflow below was developed. Further description of each

40

software can be found in the subsequent sections.

Figure 3.1: Software workflow developed for geometry creation, simulation, and analysis of results.

3.2.1 Formfinder
The interoperability of the CFD software with the site geometry software was beneficial.
However the creation of the unique anticlastic curvature of fabric membrane geometries
was not available within the selected software suite (Autodesk
®
Building Design Suite).
Therefore, a software program that was not only capable of creating the anticlastic
geometries but also capable of exporting the geometries into the CFD software was
necessary. It was determined that Formfinder (Wehdorn-Roithmayr 2012), a software for
architects and engineers to design flexible non-rigid structures (namely fabric), would
serve as a suitable software to create the geometries. The most recent version 3.5 of
Formfinder was used.

The first step of the workflow involves the creation of the membrane geometries in
Formfinder. The software allows users to create realistic anticlastic curvatures using

41

form-finding algorithms. Once users create a form, Formfinder allows designers to
compare the created form to existing projects in a database of form active structures. This
database includes projects, typologies, building materials, details, and cost estimates.

Figure 3.2: Screenshot from Formfinder. The saddle shape on the left was built within the program. The
catalogue on the right finds similar geometries based on the plan shape, surface support, border support,
and border conditions.

After the geometries are created, they are exported as a .dwg file that exists as the 3
dimensional lines that make up the surface curvatures. A .dwg is a file format that saves
2D and 3D design data. This format allowed the Formfinder geometry to be transferred to
other software tools within the software workflow.

3.2.2 Autodesk
®
AutoCAD
®
Civil 3D
®
and Autodesk
®
Inventor
®

The purpose of AutoCAD
®
(Autodesk
®
AutoCAD
®
Civil 3D
®
2011) and Inventor
®

(Autodesk
®
Inventor
®
2011) in the workflow was to create a surface that can be analyzed
in the CFD simulation software. In order to conduct an accurate CFD simulation on the
Formfinder geometry, the object must be a 3D surface or a .sat file. Opening the
exported .dwg in AutoCAD
®
or Inventor
®
, the line curvatures were created into a surface

42

using the “Loft” command. After ensuring the surface was properly formed without
geometry inaccuracies or gaps, the surface geometry was exported as a .sat file.

Figure 3.3: The screenshot on the left shows the .dwg file as exported from Formfinder into AutoCAD®.
The screenshot on the right shows the .dwg created into a surface, or .sat file.

Inventor
®
has the additional capability of being able to launch an active 3D model into
Simulation CFD. In test and practice scenarios where only the anticlastic geometry was
needed, Inventor
®
was used instead of AutoCAD
®
because it made the launch into
Simulation CFD easier. Occasionally the models destined for CFD simulation would be
too complex to launch from Revit
®
, so they would be exported from Revit
®
as a .dwg,
opened in Inventor
®
, and launched into Simulation CFD from Inventor
®
. The developers
of Simulation CFD are aware of this workaround.

3.2.3 Autodesk
®
Revit
®

Additional modeling software was required for instances where surrounding site context
was necessary as Formfinder only creates the anticlastic geometric curvatures seen in
fabric forms. When designing the software workflow, a conscious effort was made to
select programs that were of the same software suite for ease of interoperability. Revit
®

43

was selected as a result, for creating the context and site surroundings as it can directly
launch the active model into Simulation CFD.

Figure 3.4: The ability to launch the geometry into Simulation CFD from Revit® is located in the Add-Ins
ribbon.

The .sat file from AutoCAD
®
was imported into Revit
®
and properly scaled to fit into the
site model. The context and site conditions of the simulation models do not need to
include details such as windows and doors, so a simple representative mass was suitable
and appropriate. Additionally, the flow geometry box was also created in Revit
®
as a
mass encompassing the model (Figure 3.5).

3.2.4 Autodesk
®
Simulation CFD
The final piece of the software workflow in which the actual analysis and simulation took
place is Autodesk
®
Simulation CFD. Previously known as CFDesign, Simulation CFD is
an analysis tool that is interoperable with the Autodesk
®
building design suite,
specifically Revit
®
. It creates an interactive workspace using a 3D CAD model. A unique
feature of the program is that it allows changes to the geometry and reanalyzes results
instantly (Autodesk
®
2012).

44

Simulation CFD is available as two different types of downloads. The first is a
conventional software download where the analysis and simulation is powered by the
user’s computer. This download is known as Simulation CFD. The second download is
hosted in the cloud, or in other words, the model is analyzed and simulated remotely
using powerful servers accessed through the internet. This download is known as
Simulation CFD 360. Both downloads yield the same results. For all of the simulations,
Simulation CFD 360 was used because it was faster than conducting the analysis on the
author’s personal computer. Since both versions of the software yield identical results, the
term “Simulation CFD” throughout the remainder of this thesis will refer to both
Simulation CFD and Simulation CFD 360.

3.2.4.1 Solution Strategy for Simulation CFD
The overall design process for conducting viable analyses in Simulation CFD was based
on the method given in the training course. It outlines the following questions as
considerations for what parameters can be modified and how to build upon simulations.

 What is the purpose of analyzing the design?
o What are the operating conditions?
o What are the materials used?
 What information will be output?
o What are the goals for the design?
o Are there criteria for success or failure?
 What can be changed about the design to meet the goals?

45

o Can the operating conditions change?
o Can the materials change?
o What parts of the design can be changed?

These questions were considered when designing the simulation scenarios and
determining how to use the variety of parameters available to change.

3.2.4.2 Workflow for Simulation CFD
While each phase had various settings and inputs, the overall process of working in
Simulation CFD was consistent. It works within the previously described software
workflow.

1. Assemble 3D CAD geometry (in Revit
®
) and launch into Simulation CFD
2. Repair and modify the geometry to eliminate small, insignificant elements
3. Assign materials
4. Assign boundary conditions
5. Assign and modify the mesh
6. Specify solving parameters and start the simulation
7. Visualize the results
8. Re-launch simulation with different geometry
9. Compare results in Microsoft Excel

46

3.2.4.3 Best Practices for Working within Simulation CFD
The following guidelines were followed as best practice for working within the software.
These practices were mentioned in the Autodesk
®
Self-Paced Fundamentals Training
course and were discovered again during the workflow process.

Simplify the Model
Structural connections and components of fabric membranes are not essential to the
airflow analysis especially since many of the elements are too small to affect the results.
As the guidelines for CFD analysis state, it is encouraged to eliminate features that are
too small to impact the outcome of the analysis. As a result, with the exception of the
fabric membranes themselves, the context and surrounding elements were represented as
masses with no further detail.

Another important aspect of the geometry is to avoid interferences, where one volume
intersects another. The model is meshed by volumes, so intersecting volumes lead to
intersecting meshes that cause an error in the analysis. Additionally, small gaps which
cause problems in the meshing process should be eliminated. Most of the geometry
refinement should be done in the CAD program. However Simulation CFD allows for
some geometry refinement upon start up, including merging edges with small degrees,
removing small objects, and filling small voids.

47

Building Flow Geometry
The flow geometry is the shape in which the analysis occurs. When the analysis occurs
within a building, or an internal space, the flow geometry is defined by the shape of the
room. For outdoor analyses, a separate flow geometry must be constructed. In order to
ensure that the flow geometry does not affect the analyses, the sizing guidelines in Figure
3.5 are suggested. The arrow denotes the direction of flow. For example, if a flow
geometry is too small, the boundaries can “bounce back” the airflow and create
inaccurate results.

Figure 3.5: Suggested flow geometry dimensions for external airflow.

The flow geometry can be created in any part of the software workflow, from AutoCAD
®

to Simulation CFD. In this thesis, the flow geometry was created in Revit
®
before
launching into Simulation CFD and modeled as a mass that encompassed the site model.

Refine the Mesh by Regions
When the model is meshed, its volumes are broken up into elements. The elements are
defined by nodes. The nodes are where the calculations are done. The finer the mesh, the

48

more accurate and precise the results will be since there are more calculation nodes.
However, a finer mesh also yields a longer calculation time. Given a large volume, it is
valuable to refine the mesh in regions where more precise results are needed, and use a
coarse mesh in regions of lesser importance. This focus can yield accurate results without
overburdening the analysis with too many calculation nodes. A visual check of the
surfaces after they are meshed can determine the fit of the assigned mesh. If the surfaces
look as they were designed, then the mesh is a good fit. If the surfaces are jagged and
rigid compared to the original geometry, then the mesh is too coarse.

Figure 3.6: Refinement region (brown translucent box) used to specify a finer mesh in the pedestrian.

Check Convergence
When solving an analysis, Simulation CFD runs the simulation numerous times. Each
simulation is known as an iteration, and the number of iterations varies with each type of
analysis. In the beginning of an analysis the iterations will vary greatly between each

49

other. As the solution normalizes and becomes more accurate, the plot lines will stop
changing because it has reached a steady state solution. This is known as convergence.

Simulation CFD will run the analysis until it detects convergence, or the number of
iterations the user has set is reached, whichever comes first. As a result, best practice is to
set the number of iterations high and run the analysis until the solution converges itself
and Simulation CFD stops the iterations. The image below shows a solution that has
reached convergence after 356 iterations, represented by the horizontal lines.

Figure 3.7: Convergence plot after 356 iterations achieving a steady state solution.

3.3 WORKFLOW TESTING: Preventative measures
Numerous simulations were conducted to test the robustness of the software workflow in
various scenarios. The process for testing the workflow started with the most basic
scenarios and built upon the complexity. Figure 3.8 depicts the first basic scenario in
which only the saddle shape fabric membrane geometry was modeled. The 2D view is
from the side of the flow geometry, where the wind would flow from left to right.

50

Figure 3.8: Saddle shape within the flow geometry in Simulation CFD.

The intent of this simple test was to ensure that the anticlastic geometries were capable of
being brought through the workflow and that Simulation CFD could mesh and analyze
the geometry. For this scenario, only flow was enabled, no heat transfer, radiation, or
humidity. The only boundary condition was an inlet velocity and outlet pressure. The
results yielded velocity values and flow patterns. The conclusion of this test was that the
software workflow works, and the geometry can be meshed and analyzed in Simulation
CFD.

More scenarios were tested, with each building upon each other with increasing
complexity. The purpose was to make sure the final simulation, which has many
boundary conditions and solving parameters, would work properly.
Table 3.1 outlines the progression of twelve test scenarios.

51

Test
Number
Geometry
Velocity
Flow
Heat
Transfer
Solar
Radiation
Humidity
1 Membrane geometry •
2 Membrane geometry • •
3 Membrane geometry •
4 Membrane geometry • • •
5 Membrane geometry • • • •
6 Context geometry •
7 Context geometry • • •
8 Context geometry • • • •
9 Membrane + context geometry •
10 Membrane + context geometry • •
11 Membrane + context geometry • • •
12 Membrane + context geometry • • • •

Table 3.1: Test scenarios for different geometry and solution parameters.

The results from these tests are not included as they do not contribute to the final
conclusion of airflow for fabric membranes. The accuracy of the results were measured
from the convergence plots, which denoted when the solution normalized. Additionally,
the velocity flow results were compared against those from other airflow simulation
software programs, Autodesk
®
Project Vasari (Autodesk
®
Project Vasari 2012) and
Autodesk
®
Project Falcon (Autodesk
®
Project Falcon 2012), to ensure consistency. These
tests serve as an exploration of the software workflow and can be seen as pre-emptive
testing to avoid software problems in later tests.

3.4 SENSITIVITY ANALYSIS: Testing variable influences
After verifying that the software workflow worked, a sensitivity analysis of the CFD

52

software was conducted. The sensitivity analysis was conducted to determine which
variables were the most influential on the results and therefore important to focus on in
later simulations. The following variables were tested and are ranked from most
influential to least:

 Input Air Velocity
A range of low and high air velocities resulted in variation in airflow
patterns, air temperatures, relative humidity, and thermal comfort.
 Fabric Membrane Shape
The fabric membrane shape varied between the saddle, wave, arch, and
point shapes. These shapes influenced airflow patterns, air velocities, air
temperatures, relative humidity, and thermal comfort.
 Input Air Direction
The direction of the airflow heavily influenced the airflow patterns and air
velocities.
 Ambient Air Temperature
Variations in ambient air temperature concluded that the air temperature
most significantly affects the thermal comfort measurements in addition to
mean radiant temperature.
 Flow Geometry Size
Flow geometry sized above the recommended dimensions did not
significantly influence any of the results.

53

 Fabric Membrane Material Properties
The material assigned to the fabric membrane shades included concrete,
steel, wood, and PVC. Changing these did not affect the measured results
of velocity, flow patterns, temperature or humidity.

For each analysis, all settings remained constant, with the only changing the value of
each of the above variables. Basic settings were used to quickly assess differences
between the different tests.

From the sensitivity analysis the conclusions in Table 3.2 were drawn. These conclusions
were used to design the more advanced simulations.

Parameter Conclusion Advanced Analysis
Geometry Influence Variable
Air Velocity Influence Variable
Air Direction Influence Variable
Air Temperature Influence Variable
Material No Influence Consistent
Flow Geometry No Influence Consistent

Table 3.2: Results for sensitivity analysis. Influence denotes an observed change in air velocity, flow
pattern, or air temperature results. Variable means the parameter will be included as a changing variable
in the advanced simulations.

3.5 CLIMATE CONDITIONS: Different scenarios
Three climate conditions were applied to the model geometries. These climate conditions

54

were selected to include variance in wind speed, wind direction, and temperature. The
conditions were modeled after three sites: Kuala Lumpur, Malaysia; Munich, Germany;
and Abu Dhabi, United Arab Emirates. The conditions were not meant to replicate the
sites exactly, rather the conditions were set to test different inputs based on sites with
similar climates.

These sites were selected because they either currently feature tensile membrane
architecture or are proposed sites for it. They also represent a variety in climate
classifications: hot and humid, cool and temperate, and hot and arid, respectively.

Table 3.3 describes the climate settings input into Simulation CFD. The data was
gathered from Climate Consultant 5.3 (Liggett 2012), which uses climate data from the
Department of Energy in the EPW format. More detailed climate data can be found in
Appendix B.

Location
Airflow
Direction
Air Velocity
[mph]
Air
Temperature
[°F]
Humidity
[%]
Ground
Temperature
[°F]
Kuala Lumpur N 3 82 78% 80
Munich WSW 15 46 72% 47
Abu Dhabi NW 10 81 59% 81

Table 3.3: Description of site conditions used for CFD simulations.

55

3.5.1 Kuala Lumpur: Hot and humid
Kuala Lumpur is located near the equator, at 31.2°N, 101.55°E with an elevation of 72
feet above sea level. The climate is classified as hot and humid and has an average annual
rainfall of 94.21 inches. The air temperature throughout the year stays consistent, and the
range between minimum and maximum design temperatures is only 17°F. The annual
average temperature is 82°F.

Due to its location, Kuala Lumpur experiences consistent solar radiation throughout the
year, although it is the lowest of the scenarios. Humidity varies throughout the day with a
range of 60% to 95%. Humidity is the lowest in the early afternoon. The primary annual
wind direction is from the North at 3 mph.

When the adaptive thermal comfort model for 80% acceptability is applied to the climate
data, 75.8% of the hours of the year, or 6,641 out of 8,760, are considered comfortable if
natural ventilation is used. The remainder of the hours are not comfortable because they
are too hot.

3.5.2 Munich: Cool and temperate
Munich is the furthest north climate at 48.13°N, 11.7°E and an elevation of 1,735 feet
above sea level. The climate is classified as temperate and is the coldest climate
simulated. It has an average annual rainfall of 31.90 inches. The air temperature
throughout the year varies greatly, and the range between minimum and maximum design

56

temperatures is 90°F. The annual average temperature is 46°F.

Munich’s solar radiation varies throughout the year. The summer time receives more solar
radiation than the winter. Humidity varies throughout the year with a range of 75% to
90% in the winter and 50% to 59% in the summer. The primary annual wind direction is
from west south west at 7 mph. However, to create variance amongst the climate
scenarios, a velocity of 15 mph was used for the simulations. This air velocity is not
unheard of in Munich. In fact, gusts over 45 mph have been recorded.

When the adaptive thermal comfort model is applied to the climate data, only 7.5% of the
hours of the year, or 656 out of 8,760, are considered comfortable if natural ventilation is
used. A small number of the hours are too hot, but the remainder are too cold.

3.5.3 Abu Dhabi: Hot and arid
Abu Dhabi is located at 24.43°N, 54.65°E and an elevation of 88 feet above sea level.
The climate is classified as hot and arid and has a low average annual rainfall of 2.3
inches. The air temperature throughout the year varies a moderate amount, and the range
between minimum and maximum design temperatures is 76°F. The annual average
temperature is 81°F.

Abu Dhabi’s solar radiation is consistent throughout the year and the highest of the
scenarios. The summer time receives more solar radiation than the winter. Humidity

57

varies throughout the year with a range of 45% to 80% in the winter and 20% to 70% in
the summer. The primary annual wind direction is from North West at 10 mph

When the adaptive thermal comfort model is applied to the climate data, 34.6% of the
hours of the year, or 3,029 out of 8,760, are considered comfortable if natural ventilation
is used. The remaining uncomfortable hours are either too cold or to hot.

3.6 MODEL GEOMETRY: Realistic external scenario
With the workflow functionality confirmed and the sensitivity analysis conducted, a
testing scenario was designed. The purpose of this testing scenario was to create a
convincing environment in which the fabric membrane shades could be analyzed in.

3.6.1 Context Geometry
The site geometry was modeled to imitate an open environment with surrounding
buildings. Comparable to an outdoor market or shopping center, the site features a
pedestrian walkway passing through the center of the site. The fabric membrane forms
are placed over this pedestrian walkway. Examples of this scenario can be seen in Figure
3.9. The site is small scale so impacts at the human level are detectable. Previous
preliminary studies conducted as practice showed that a large scale model produced
inconclusive results at the human level. The results from these tests are not included as
they do not contribute to the final conclusions of airflow for fabric membranes.

58

Figure 3.9: Swindon Parade in Swindon, UK (left) and Paseo Acoxpa in Mexico City, Mexico (right). Both
are examples of outdoor areas with fabric membrane shades covering the pedestrian walkways (Images
Architen Landrell Associates Ltd. & BirdAir).

Figure 3.10 describes the site context. The surrounding buildings were modeled as simple
box masses in Revit
®
, as the simulations and solutions run better when insignificant
details are left out.

Figure 3.10: Plan, section, and axon view of site context.

59

3.6.2 Fabric Membrane Shade Geometry
Following the workflow, the fabric membrane shade geometries were created in
Formfinder. The geometries were converted into a surface in AutoCAD
®
before imported
into Revit
®
.

The anticlastic geometries that were used as the fabric membrane shades were the four
basic fabric membrane shapes discussed in Chapter 2 (saddle, wave, arch, and point).
These four shapes were selected because they are fundamental anticlastic curvatures and
can have many variations. These pure forms and variations of them are commonly seen in
fabric membrane architecture.

The shapes were generated in Formfinder using the shape catalogue (Figure 3.2) to
define the geometries. An example of the shade placement in the context model is given
in Figure 3.11. All of the sizes of all the fabric membrane shapes in context can be seen in
Appendix C.

60

Figure 3.11: Saddle, wave, arch, and point shades over the pedestrian walkway.

3.6.3 Human Models
Human models are volumes that represent occupants in form and heat transfer properties.
Three human models were included in the model geometry. The purpose of the human
model was to provide a volume that would easily read temperature, PPD, and PMV data.
The model is freely available in the Simulation CFD wikiHelp website as a Revit
®
2013
compatible family. The placement of each human model was intended to represent logical
locations of actual occupants in a real scenario. Additionally, the location of the human
models was intended to spread over a range to achieve different data outputs. The
placement of the human models varied in the x and y directions to account for variation
of the results measured within the pedestrian area. More human models would increase
the number of elements in the model and the calculation time. The location of each
human was the same in every model.

61

Figure 3.12: Human model family from Revit® 2013. Red circles denoting the location of the human
models in plan view.

3.6.4 Flow Geometry and Ground Geometry
In addition to the site geometry and shade shapes that were assembled in Revit
®
, the flow
geometry and ground geometry were also constructed in Revit
®
. The flow geometry
defines the physical boundaries of the simulation. Using the recommended size settings
(Figure 3.5), the flow geometry was built as a mass encompassing the model. In all three
cases, it was a rectangular prism. The flow geometry varies with the climate scenarios
(refer to
Table 3.3). For example, the flow geometry for Kuala Lumpur was not suitable for
Munich. Therefore, each of the three climate scenarios had a unique flow geometry box
that aligned with the cardinal directions of the inlet air velocities. Figure 3.13 compares
the flow geometries for each climate scenario with the arrows describing the flow
direction in each climate.

62

Figure 3.13: North is up. Shown are the flow directions and flow geometries for Kuala Lumpur (left),
Munich (middle), and Abu Dhabi (right).

Additionally, a ground geometry was built that was external to the flow geometry. The
purpose of this ground geometry was to appropriately simulate ground heat transfer.
Figure 3.14 depicts a model complete with flow and ground geometries.

Figure 3.14: Complete model geometry that was launched into Simulation CFD from Revit®. It is
important to note that the flow geometry is dependent upon the direction of the input air velocity. This
particular flow geometry is appropriate for an air velocity coming from the North.

63

3.7 CFD SIMULATION SETTINGS: Building & assigning each simulation model
After the model was built and the climate settings were chosen, the next step was to
prepare the simulation in the CFD software and solve the model for air velocity, air
temperature, and relative humidity. The process and settings described were reviewed by
an engineer on the Autodesk
®
Simulation CFD team. The reviewer offered comments for
improvement that were taken as well as added confidence that the simulations were being
performed best to the ability of the software. Comments included advice for refining the
mesh through regions and enabling solar radiation as a solution parameter.

As mentioned in the previous software workflow section 3.2, each model was built in
Revit
®
then launched directly into Simulation CFD. This section describes the process
once the “Launch Active Model” button was pressed in Revit
®
.

The first prompt from Simulation CFD is to save the model as a new design scenario, or
as part of an existing one. For each simulation, a new design scenario was created. In
previous test simulations, several different models were saved as part of an existing
design scenario. This was beneficial in that several models could be run in sequence
without interruption. However, at the realistic level of simulation that this section
describes, the file sizes were already quite large with just one design scenario (500 to 900
MB); adding to the design scenarios made the file large and cumbersome and as a result
slow.

64

Figure 3.15: Screenshot from Simulation CFD. The numbers denote the order in which the settings were
applied.

The settings described below were repeated for each of the four basic membrane
geometries in addition to two base case scenarios. Each simulation was set up identically,
with only variance in the geometries and climate based values that were input. Figure
3.16 describes all the simulations that were performed.

Figure 3.16: All six shade conditions were simulated in the three climate scenarios.

3.7.1 Geometry Tools (1)
Upon launching the geometry and saving as a new design scenario, Simulation CFD tries
to detect geometry that will be problematic in meshing. This dialogue box is known as

65

Geometry Tools and is the first to appear after saving the study. It allows users to repair
and modify geometries to eliminate small and insignificant elements. Figure 3.17 shows
the dialogue box for merging edges. Merging the edges will join and edges with an angle
of less than 5 degrees, and in this instance will merge 66 edges. For all of the simulations,
edges less than 5 degrees were merged. The other tabs of the dialogue include eliminating
small objects, filling voids, and creating external volumes. None of these other tabs were
pertinent to the simulations so they were not used.

Figure 3.17: Geometry Tools dialogue box with merging edges less than 5 degrees is done.

3.7.2 Assign Materials (2)
After a visual inspection of the geometry to insure it imported from Revit
®
correctly,
materials were assigned to the model. The selection of materials was influenced by the
results of the sensitivity analysis in that they were to be consistent for all simulations.
Table 3.4 breaks down the material assignment by geometry. A description of why each
material was chosen is given in the subsequent sections. From the sensitivity analysis it
was concluded that the material assignments were not influential within the context of the
study and scale of the model. Therefore, it is important that the materials are accurately

66

representative, but similar materials are also acceptable.

Geometry Component Material Environment
Flow Geometry Moist Air Variable
Ground Plane Concrete Fixed
Context Buildings Concrete Fixed
Fabric Shades Nylon Fixed
Human Models Human Fixed

Table 3.4: Material assignment for each simulation by geometry component.

After all the materials had been assigned, the model was explored to ensure that all
materials had been properly assigned. A quick check at the material assignment legend
can check that all geometry components have a material assigned to them.

Figure 3.18: Model during the material selection process. The legend in the lower left denotes the material
assignments per color.

3.7.2.1 Moist Air
In comparison to “Air,” moist air was used to enable humidity. A fixed environment
means that the properties of the air will stay constant throughout the analysis. The
alternative is a variable environment where the properties vary, which is an appropriate
setting for natural convection scenarios. The latter was chosen for all simulations to

67

simulate airflow with the influence of temperature and humidity. Figure 3.19 shows
dialogue boxes to select the material, and then describe the environment. Figure 3.20
details the specific material properties of Moist Air.

Figure 3.19: Material selection dialogue box (left) and material environment selection dialogue box
(right).

Figure 3.20: Material properties of Moist Air.

68

3.7.2.2 Concrete
Concrete was used for the ground plane and context buildings. This material was used
because concrete is a common construction material in the defined scenarios. The
environment was set as fixed.

Figure 3.21: Material selection dialogue box (top) and material properties of Concrete (bottom).

69

3.7.2.3 Nylon (Polyester with PVC Coating)
“Nylon” was the name given to the material used for the fabric membrane shades. The
reality of the material represents the properties of a polyester structural fabric with a PVC
coating. The durability of this fabric is typically 15 to 20 years (Bechthold 2008, 83).
Additional properties of the material are explained in Figure 3.22. Since the geometry of
the fabric shades were created with no thickness, just as a surface, they were not
considered a volume. Therefore, the material had to be assigned as a surface with a
thickness of 0.0375 inches. The environment was set as fixed.

Figure 3.22: Material selection dialogue box (top) and material properties of Nylon (bottom).

70

3.7.2.4 Human
The built in material suggested for human models in Simulation CFD is a material called
“human.” This material imitates the thermo-physical properties of a human occupant. As
a volume in the simulation, the results are capable of being color mapped over the human
models. The environment was set as fixed.

Figure 3.23: Material selection dialogue box (top) and material properties of Human (bottom).

71

3.7.3 Boundary Conditions (3)
Following assigning the materials, the boundary conditions were constructed. The
boundary conditions defined the air velocity, temperature, and humidity inputs. The same
boundary conditions were applied to each simulation, with variation in regards to the
specific values. The climate scenarios (3.5) are the input values for the boundary
conditions. Table 3.5 describes how the boundary conditions were assigned to the model.

Boundary Condition Value Assigned to
Air Velocity climate dependent flow box inlet face
Pressure 0 psf flow box outlet face
Air Temperature climate dependent all external flow box faces
Ground Temperature climate dependent bottom ground plane
Humidity climate dependent flow box inlet face
Total Heat Generation 115 W human model

Table 3.5: Boundary conditions as assigned to the model. Note that “climate dependent” refers to the
values for each different climate scenario outlined in
Table 3.3.

Like assigning materials, boundary condition assignments can be checked via the legend.

Figure 3.24: Model during the boundary condition assignment process. The legend in the lower left
denotes the boundary conditions assigned to colored stripes on the model.

72

3.7.3.1 Air Velocity & Pressure
The air velocity and pressure boundary conditions describe the flow for the analysis. The
inlet of the flow is the face of the flow geometry that corresponds to the cardinal direction
of the air velocity. The outlet is the face on the opposite end. The inlet face has a velocity
boundary condition, while the outlet face has a pressure boundary condition. The velocity
varied with the climate scenario chosen (3.5), but the pressure was always gage pressure
at 0 psf. Gage pressure refers to relative pressure, and a value of 0 is used to define the
outlet. This inlet/outlet boundary condition was set under the guidance of airflow best
practices from Simulation CFD (Autodesk
®
2012).

Figure 3.25: Boundary condition dialogue boxes for defining the inlet velocity and outlet pressure. Velocity
Magnitude is the variable for each climate scenario.

73

Figure 3.26: Each of the climate scenarios with the inlet faces of the flow geometry denoted in red and the
outlet face marked in blue. Order from left to right is Kuala Lumpur, Munich, and Abu Dhabi with North at
the top.

3.7.3.2 Air & Ground Temperature
The air and ground temperature boundary conditions describe the ambient air
temperatures. The air temperature boundary condition was applied to all five exposed
faces of the flow geometry (four sides and one top). The sixth face that was not assigned
a boundary condition was the face adjacent to the ground geometry. Instead, the bottom
plane of the ground geometry was assigned a temperature boundary condition that
corresponded to the ground temperature. These boundary conditions were applied in the
same fashion for all the climate scenarios with the temperature being the variable that
changed.

74

Figure 3.27: Boundary condition dialogue box for temperature. The boundary condition was the same for
both air and ground assignment, with only the temperature being different.

Figure 3.28: Temperature boundary condition assignments to the flow geometry. Every face with a teal
stripe denotes a temperature assignment (six total). The view on the left is from the top Northwest corner.
The view on the right is from the bottom Southeast corner

3.7.3.3 Humidity
Humidity was assigned only to one face, the same face as the velocity inlet. The input
value varied with each climate simulation and was represented as a value between 0 and
1, with 1 equal to 100% humidity.

Figure 3.29: Humidity boundary condition dialogue box.

75

3.7.3.4 Total Heat Generation
The total heat generation boundary condition applies to the human models. It is the only
boundary condition that is not applied to the flow geometry and is assigned directly to a
volume. Total heat generation is representative of the metabolic rate of the occupants. For
all the simulations, a value of 115 W was assigned to each human, which is equivalent of
the occupants walking about slowly.

Figure 3.30: Total heat generation boundary condition dialogue box settings.

3.7.4 Assign and Modify the Mesh (4)
The mesh is a very important aspect of a valid simulation. It is a contributing factor to the
accuracy of the solution and the time taken to reach the solution. The mesh was assigned
in three different ways to ensure the areas of the model with greater importance had a
finer mesh while areas of lesser importance had a coarser mesh.

First the mesh was set to Automatic Sizing. This setting is the preferred choice to use as
the automatic sizing feature typically creates a manageable and accurate mesh.
Next, given the anticlastic geometries of the shades, additional refinement on the shades
was needed to ensure their geometry was represented properly. The automatic sizing

76

mesh assigned to the shades created a very high number of elements, around 11 million.
This number is too high for a reasonable simulation time so the mesh sizing was
increased, then refined for specific areas. However, care had to be taken that the mesh did
not become too coarse and represented the shades’ curvature as jagged. To change only
the shades, they were selected and the size adjustment was set to use uniform, but the size
increased to 12 in the mesh editing dialogue box.

Figure 3.31: Mesh editing dialogue box. For the shades, the size adjustment was set to 12.

Finally, additional mesh attention was given to the pedestrian area. Using a mesh
refinement region described in section 3.2.4.3, the mesh of the pedestrian area was
refined to 0.85. This was done to increase the accuracy of the results for this region since
the bulk of the results would be taken from the pedestrian area.

77

Figure 3.32: Refinement region of the pedestrian area.

When the automatic mesh sizing was assigned, the approximate number of elements was
typically around 11 million. After the adjustments were completed on the shades and
pedestrian area, the total number of elements was reduced closer to 2 million. This
method of decreasing the elements ensured that the accuracy of the simulation where it
mattered was maintained and the calculation time remained manageable. The final mesh
can be seen in Figure 3.33.

Figure 3.33: Final mesh pattern of the model.

78

3.7.5 Solution Conditions (5)
The solution conditions were defined by what the desired outputs of the simulations were.
Information and data on the following metrics were solved for velocity flow patterns, air
velocity, air temperature, humidity, Predicted Percentage of Dissatisfied, Predicted Mean
V ote, and volume temperature.

The first tab of the Solve dialogue box (Figure 3.34) specifies if the solution is steady
state or transient. For all the simulations steady state was selected. The dialogue box also
sets the number of iterations to run. For all the cases, 500 iterations were specified. In
most cases, the solution converged before the 500 iterations had been completed. In some
cases, the solution did not converge, so additional iterations were added until the solution
converged. In order to include thermal comfort results, the Results Quantities button was
selected and Thermal comfort was selected under Miscellaneous.

Figure 3.34: Solve dialogue box with results quantities.

79

The Physics tab of the Solve dialogue box described what the simulation solves for. For
all the simulations flow, heat transfer, radiation, humidity, and solar heating were
included. In the solar heating dialogue box the location was set using the country and city
or through global coordinates. The time and date was set to a worst case scenario based
on information from the climate data represented in
Table 3.3.

Figure 3.35: Solve dialogue box with Advanced and Solar Heating options.

80

Once all the solution conditions were set, the SOLVE button was selected and the
simulation began. The calculation time for each of the models ranged from 20 minutes to
2 hours depending on the complexity of the mesh.

3.8 DATA COLLECTION: How the data was extracted
Once the solution finished and convergence was reached, the results were available to
view and export. Simulation CFD produced results in both visual form and tabular data.
Both types of results were used for comparison. Further explanation and example of the
results can be seen in Chapters 4 and 5. This section serves to describe the method of
extracting the results.

Simulation CFD allows all results, airflow patterns, air velocity, air temperature, relative
humidity, PMV , and PPD, to be output in a variety of units and scales. For the purpose of
consistency, the same units for each property were used, and reasonable scales for each
scenario were used for fair comparison. Velocity was measured in miles per hour,
temperature was measured in degrees Fahrenheit, and relative humidity was measured as
a ratio between 0 and 1. PMV was defined by the voting scale in Table 2.1 and PPD was
represented on a scale of 0 to 100%.

3.8.1 Airflow Patterns
The airflow patterns were collected by visualizing directional vectors on a 2D slice

81

within the model. The 2D slice can be taken on the xy, xz, or yz axis, at a specified level,
and at any inclination.

Figure 3.36: Example of an angled plane with velocity results.

The arrows show the direction of the air velocity magnitude for each node. Additionally,
color can be mapped on the arrows to add a second layer of data. This can be air velocity,
air temperature, relative humidity, or any other parameter the simulation solved for. This
becomes beneficial for observing the interaction between two parameters. Figure 3.37
depicts the directional arrows with the color reading velocity magnitude in mph.

Figure 3.37: Example of velocity flow vectors denoting airflow patterns. The color of the arrows
corresponds to the speed in mph.

82

3.8.2 Air Velocity
Using the same 2D slice to view airflow patterns, the calculated air velocity of each node
can be exported as tabular data. The data is broken down into the Cartesian coordinate of
the node, the x, y, and z-component of the velocity, and the velocity magnitude for the
node. Figure 3.38 below shows this data in Microsoft Excel.

Figure 3.38: Air velocity exported to Excel. The data represents the node number, coordinate location of
the node, x, y, and z-components of the velocity, and the velocity magnitude.

Air velocity can also be visualized through color mapping, similar to the directional
arrows. However, the export of the tabular data was the preferred method of data
extraction for the air velocity because the data would be more quantifiable than
comparing the color map to the legend to decipher the air velocity.

83

Figure 3.39: Example of air velocity data mapped on a 2D plane.

3.8.3 Temperature
Temperature data can be extracted the same way as air velocity data. It is possible to map
the temperature data as color on the 2D slice, but the preferred method was to export the
tabular data to Excel.

Figure 3.40: Air temperature exported to Excel. The data represents the node number, coordinate location
of the node, x, y, and z-components of the velocity, and the temperature magnitude.

84

3.8.4 Relative Humidity
The same process as air velocity and temperature was used to extract relative humidity
data. Due to the low variation of the data, color mapping was not appropriate even at a
small specified scale. Therefore the best option was to export the data to Excel.

Figure 3.41: Relative humidity exported to Excel. The data represents the node number, coordinate
location of the node, x, y, and z-components of the velocity, and the relative humidity.

3.8.5 Human Models
Unfortunately there was no known process to extract the data from the human models
(PPD and PMV) in tabular form. Therefore the human models were compared using the
visual color overlay on the same scale. To supplement this, tabular data was also created
for PMV and PPD as explained in Chapter 5.

85

Figure 3.42: PPD, PMV , and Temperature represented through color mapping over the human model
volume.

Additionally, the minimum, maximum, and average temperature for each human can be
calculated and output as numeric values.

Figure 3.43: Output of the temperatures for each human model volume.

3.9 CONCLUSIONS: Development of process to yield results
The previously discussed process was developed based on the available resources and
desired outcomes of the simulations. Within the main software workflow, there are

86

methods described of how to work within each portion of the software workflow. The
purpose of the detailed process description is for replication and evaluation of the validity
of the study.

87

CHAPTER 4 : RESULTS FROM AIRFLOW STUDY

This chapter documents the results from the simulations described in Chapter 3 with
Appendix D. The raw data presented includes airflow patterns, air velocity, air
temperature, relative humidity, PMV , and PPD, and serves as the basis for the analysis
and conclusion in subsequent chapters.

4.1 AIRFLOW PATTERNS
The most visual result from the CFD simulations were the airflow patterns. Figure 4.1
and Figure 4.2 describe where the 2D slices of airflow patterns of the 3D data were taken
for each simulation.

Figure 4.1: Cross section of data from the x-y plane. This level was chosen to observe patterns around the
human model.

88

Figure 4.2: Cross section of data from the x-z plane and from the from the y-z plane.

To decipher any similarities or differences between the patterning at each of the 2D slices
in the pedestrian area, they were all scaled to ensure a fair comparison. Additionally, the
color air velocity data was removed from these images for easier comparison. A set of
these images is seen in Figure 4.3 for the x-y plane given Kuala Lumpur, Munich, and
Abu Dhabi respectively. The remainder of these results can be seen in Appendix D.

89

Figure 4.3: Airflow patterns from the x-y plane of Kuala Lumpur, Munich, and Abu Dhabi.

4.2 AIR VELOCITY
The air velocities in the pedestrian area were compared between the different shade
geometries. The velocities were measured at the elevations of 1’ 6”, 3’, and 5’ 6” (Figure
4.4) for each simulation. These levels were individually measured, then averaged as per
best practice guidelines for thermal comfort.

𝑎𝑣𝑒𝑟𝑎𝑔𝑒 𝑣𝑒𝑙𝑜𝑐𝑖𝑡𝑦 =
𝑣 𝑙𝑒𝑣𝑒𝑙 1
+ 𝑣 𝑙𝑒𝑣𝑒𝑙 2
+ 𝑣 𝑙𝑒𝑣𝑒𝑙 3
3

90

Figure 4.4: Analyses levels where air velocity was measured.

Figure 4.5 graphically represents the velocity averages from the three elevations for each
shade geometry and climate scenario. For each shade geometry the average is denoted by
“x,” and the input value is represented by the red line. Table 4.1compares the averages.

Velocity Averages [mph]
Climate
Shade Condition
None Flat Saddle Wave Arch Point
Kuala Lumpur 2.83 2.29 2.49 1.67 2.03 2.51
Munich 8.34 7.85 9.47 6.32 4.74 8.37
Abu Dhabi 6.74 6.15 8.04 5.93 5.87 7.24

Table 4.1: Velocity averages for all three climate scenarios.

91

Figure 4.5: Velocity averages for Kuala Lumpur, Munich, and Abu Dhabi.
Kuala Lumpur Velocity Averages Munich Velocity Averages Abu Dhabi Velocity Averages

92

4.3 AIR TEMPERATURE

Similar to the air velocity, the air temperature at the elevations in Figure 4.4 was
extracted, averaged, and put into the graphs seen in Figure 4.6 for each simulation. For
each shade geometry the average is denoted by “x,” and the input value is represented by
the red line. Table 4.2 compares the averages.

𝑎𝑣𝑒𝑟𝑎𝑔𝑒 𝑡𝑒𝑚𝑝𝑒𝑟𝑎𝑡𝑢𝑟𝑒 =
𝑡 𝑙𝑒𝑣𝑒𝑙 1
+ 𝑡 𝑙𝑒𝑣𝑒𝑙 2
+ 𝑡 𝑙𝑒𝑣𝑒𝑙 3
3

Air Temperature Averages [°F]
Climate
Shade Condition
None Flat Saddle Wave Arch Point
Kuala Lumpur 82.21 83.24 82.40 82.10 82.03 82.33
Munich 50.28 51.90 53.37 50.00 50.93 52.86
Abu Dhabi 81.14 82.15 82.89 81.97 81.41 83.11

Table 4.2: Air temperature averages for all three climate scenarios.

93

Figure 4.6: Temperature averages for Kuala Lumpur, Munich, and Abu Dhabi.
Kuala Lumpur Temperature Averages Abu Dhabi Temperature Averages Munich Temperature Averages

94

4.4 RELATIVE HUMIDITY
Relative humidity was also measured at the elevations in Figure 4.4. The figures below
show the relative humidity measured. For each shade geometry the average is denoted by
“x,” and the input value is represented by the red line. Table 4.3 compares the averages.

𝑎𝑣𝑒𝑟𝑎𝑔𝑒 𝑟𝑒𝑙𝑎𝑡𝑖𝑣𝑒 𝑕 𝑢𝑚𝑖𝑑𝑖𝑡𝑦 =
𝑟 𝑕 𝑙𝑒𝑣𝑒𝑙 1
+ 𝑟 𝑕 𝑙𝑒𝑣𝑒𝑙 2
+ 𝑟 𝑕 𝑙𝑒𝑣𝑒𝑙 3
3

Relative Humidity Averages
Climate
Shade Condition
None Flat Saddle Wave Arch Point
Kuala Lumpur 0.767 0.778 0.769 0.776 0.742 0.769
Munich 0.583 0.614 0.588 0.623 0.581 0.567
Abu Dhabi 0.569 0.594 0.592 0.589 0.578 0.582

Table 4.3: Relative humidity averages for all three climate scenarios.

95

Figure 4.7: Relative humidity averages for Kuala Lumpur, Munich, and Abu Dhabi.
Munich Humidity Averages Abu Dhabi Humidity Averages Kuala Lumpur Humidity Averages

96

4.5 HUMAN MODEL DATA
The human models that were included in the simulation scenarios represent the metrics of
Predicted Percentage of Dissatisfied (PPD), Predicted Mean V ote (PMV), and
temperature. These results are represented as a color map over the human model volume
and as maximum, minimum, and average temperature readings for the volume. Figure
4.8 denotes the labels applied to the human models respective to their locations in the
context for future reference.

Figure 4.8: Locations and labels of human models in site context.

Figure 4.9 shows the PPD and PMV data for Person B for each geometry shade in the
three climate scenarios.

97

Figure 4.9: PPD (left) and PMV (right) data on human models for all climate and shade scenarios.

4.6 CONCLUSIONS: Raw data representation
The preceding sections display the initial raw data gathered from the simulations. The
PPD Data PMV Data

98

airflow patterns, PMV , and PPD are represented as visual data, while the air velocity, air
temperature, and relative humidity are reflected as tabular data. More discussion of the
relevance and variance amongst the results can be found in the following chapter.

99

CHAPTER 5 : THERMAL COMFORT ANALYSIS OF AIRFLOW
STUDY

This chapter describes and analyzes the previously documented results of Chapter 4 and
Appendices E and F. The purpose of this chapter is to portray and interpret the data in
order to draw conclusions in Chapter 6.

5.1 ANALYSIS BY VARIABLE: Velocity, temperature, and relative humidity
The velocity, temperature, and humidity results were averaged from the three different
analysis planes discussed in Chapter 4. The results measured from each simulation were
compared against three different base cases. The first was the scenario in which no shades
were modeled, the second featured flat shades, and the third was the input values. The
percent differences between the base cases and the fabric membrane shade models were
calculated and compared between climates. Observations were made as to which forms
were closer to the base case and which had higher percent differences in addition to how
the forms increased or decreased the measured variable.

Table 5.1 shows the analysis results in order from low to high compared to each other for
velocity, temperature, and relative humidity.

100

Kuala Lumpur Munich Abu Dhabi
Lower
Velocity

Higher
Velocity
Wave Arch Arch
Arch Wave Wave
Flat Flat Flat
Saddle None None
Point Point Point
None Saddle Saddle

Kuala Lumpur Munich Abu Dhabi
Lower
Temperature

Higher
Temperature
Arch Wave None
Wave None Arch
None Arch Wave
Point Flat Flat
Saddle Point Saddle
Flat Saddle Point

Kuala Lumpur Munich Abu Dhabi
Less
Humid

More
Humid
Arch Point None
None Arch Arch
Saddle None Point
Point Saddle Wave
Wave Flat Saddle
Flat Wave Flat

Table 5.1: Results ordered from lowest to highest for velocity, temperature, and relative humidity.

From these results several observations can be made for each variable.
5.1.1 Velocity
Comparing only the averaged results, the flat shade base case, and wave and arch fabric
membrane shades had slower velocities when compared to the no shade base case and
saddle and point fabric membrane shades. Introducing the percent difference analyses, the
flat shade base case and point fabric membrane shade behaved most similarly to the no
shade base case. In all the simulation scenarios, the average measured velocity in the

101

pedestrian area decreased compared to the input velocity. The average decrease for Kuala
Lumpur was 23.3%, Munich was 49.9%, and Abu Dhabi was 33.4%.

For the most part, the fabric membrane shades behaved consistently across the climate
scenarios. Table 5.2 details the percent difference values compared to the no shades
scenario and the input velocity values. A negative value represents where the velocity
decreased and a positive value represents where the velocity increased.

Percent Difference Compared to No Shades

Kuala Lumpur Munich Abu Dhabi
Flat -18.8 -5.9 -8.9
Saddle -12.0 13.6 19.2
Wave -41.0 -24.2 -12.0
Arch -28.2 -43.1 -13.0
Point -11.2 0.3 7.3

Percent Difference Compared to Input Velocities

Kuala Lumpur Munich Abu Dhabi
None -5.8 -44.4 -32.6
Flat -23.5 -47.7 -38.6
Saddle -17.2 -36.8 -19.6
Wave -44.4 -57.9 -40.7
Arch -32.4 -68.4 -41.3
Point -16.3 -44.2 -27.6

Table 5.2: Percent difference charts for velocity. A negative value denotes a decrease in velocity where a
positive value denotes an increase.

The variation in percent difference scale between the climate scenarios lends itself to the
different input velocities. Munich had the highest input velocity (15 mph), and the

102

percent difference scale for this scenario is the highest.

Referring to the ASHRAE 55 thermal comfort guidelines for air velocities of interior
spaces, the standard states that the general thermal comfort model is appropriate for air
velocities under 0.2 m/s, or 0.447 mph. Table 4.1 lists the average air velocities, of which,
none of the climate scenarios meet the appropriate air velocities, because they are all too
high. However, air velocities that are greater the threshold can still expand the comfort
zone through the elevated air speed circumstance (ASHRAE 2010, 6).

Elevated air speed allows for the increase of the upper temperature threshold for comfort
given that occupants are in control of the air velocity. Figure 5.1 demonstrates the
temperature offset through a graph assuming the occupant has between 0.5 and 0.7 clo
and physical activity is between 1.0 and 1.3 met (ASHRAE 2010, 6).

Figure 5.1: Temperature offset graph for elevated air speeds (Image ASHRAE 2010, 6).

103

The maximum air speed from Figure 5.1 coverts to 3.579 mph, which includes Kuala
Lumpur’s velocity averages, but Munich and Abu Dhabi are still too high. As a result, the
elevated air speed model is not appropriate.

The adaptive model for thermal comfort is better suited for the velocity data because it
has no air speed limitations. The adaptive model allows for variation in thermal
experience, clothing, availability of control, and occupant expectations (ASHRAE 2010,
9).

5.1.2 Temperature
The temperature results were similar to the velocity results in some respects. Comparing
only the temperature data, the no shade base case and wave and arch fabric membrane
shades had lower temperatures when compared to the flat shade base case and saddle and
point fabric membrane shades. The results for velocity were the same, except the no
shade base case and flat shade base case were switched. With the percent difference
analysis, wave and arch fabric membrane shades behaved most similarly to the no shade
base case. Saddle and point fabric membrane shades had the most variance from the no
shade base case. In all the simulation scenarios, the average measured air temperature
increased compared to the input temperature. This can be attributed to the fact that the
measured temperature also included the effects of solar radiation, an attribute that is not
included in the input temperature. The average increase in temperature for Kuala Lumpur
was 0.5%, Munich was 12.1%, and Abu Dhabi was 1.4%.

104

Compared to the velocity results, the fabric membrane shades behaved less consistently
when compared across climate scenarios. Looking at the percent difference values in
Table 5.3, the scale of the percent difference is much smaller than that of the velocity
percent difference analysis. The values are much closer, and the percent difference
between each scenario is not as variable as the velocity values. As previously mentioned,
the air temperature was not only influenced by the input temperature value, but also the
solar radiation solution parameter, air velocity input, and relative humidity input. The
conclusion from the temperature results is that the order of the values is not as important
as the spread of the values. The range of the saddle and arch shades was smaller than the
temperature range for the wave and point shades.
Percent Difference Compared to No Shades

Kuala Lumpur Munich Abu Dhabi
Flat 1.3 3.2 1.2
Saddle 0.2 6.2 2.2
Wave -0.1 -0.6 1.0
Arch -0.2 1.3 0.3
Point 0.2 5.1 2.4

Percent Difference Compared to Input Temperatures

Kuala Lumpur Munich Abu Dhabi
None 0.3 9.3 0.2
Flat 1.5 12.8 1.4
Saddle 0.5 16.0 2.3
Wave 0.1 8.7 1.2
Arch 0.0 10.7 0.5
Point 0.4 14.9 2.6

Table 5.3: Percent difference charts for temperature. A negative value denotes a decrease in temperature
where a positive value denotes an increase.

105

Again, Munich had the most variation amongst its percent difference values. It also had
the greatest increase in temperature.

Using the adaptive model for thermal comfort, mean monthly outdoor air temperature
and indoor operative temperature are used to define the acceptable limits. The mean
monthly outdoor air temperature was taken from the average values for each climate
scenario. The indoor operative temperature was calculated using the operative
temperature equation from the 2009 ASHRAE Handbook—Fundamentals. The adaptive
model is suited for outdoor air temperatures between 50° and 92.3°F. Temperatures below
or above this range are not included and should not be extrapolated. All of the
temperature averages fit into this range.

5.1.3 Relative Humidity
Considering only the averages, the no shade base case and arch fabric membrane shade
had lower humidity levels compared to the flat shade base case and wave fabric
membrane shade, which measured higher humidity levels. These results mimicked those
of the temperature results with the exception of the wave fabric membrane shade moving
to the higher end. The percent difference analysis showed that the flat shade base case
had more variance from the no shade base case in comparison to the other fabric
membrane shades. For Kuala Lumpur and Munich, the average measured relative
humidity decreased from the input humidity. For Abu Dhabi, the no shade base case,

106

wave, arch, and point fabric membrane shades decreased the humidity from the input
where the flat shade base case and saddle fabric membrane increased the humidity levels.
The average decrease in relative humidity from the input value for Kuala Lumpur was
1.7% and 17.7% for Munich. Averaging all the percent difference compared against the
input value for Abu Dhabi resulted in a 1.0% decrease. The arch shade was the most
successful in decreasing the high relative humidities of Kuala Lumpur and Munich.

Compared to velocity and temperature, relative humidity was the least consistent between
climate scenarios. Factors that influence the relative humidity measurements include
velocity, temperature, and solar radiation. With each of these factors varying between
each climate scenario, the measured relative humidity results are also variable. Kuala
Lumpur and Munich had similar humidity inputs (0.78 and 0.72 respectively), and with
the exception of the point shape, the percent differences from the input values are
consistent. Abu Dhabi had the lowest humidity input (0.59) and the percent differences
from the input value varied, similar to Kuala Lumpur and Munich (
Table 5.4).

107

Percent Difference Compared to No Shades

Kuala Lumpur Munich Abu Dhabi
Flat 1.4 5.4 4.5
Saddle 0.2 0.9 4.0
Wave 1.1 6.9 3.6
Arch -3.3 -0.3 1.7
Point 0.3 -2.7 2.3

Percent Difference Compared to Input Humidity

Kuala Lumpur Munich Abu Dhabi
None -1.6 -19.1 -3.6
Flat -0.3 -14.8 0.7
Saddle -1.4 -18.4 0.3
Wave -0.5 -13.5 -0.1
Arch -4.9 -19.3 -2.0
Point -1.4 -21.3 -1.4

Table 5.4: Percent difference charts for relative humidity. A negative value denotes a decrease in relative
humidity where a positive value denotes an increase.

Similar to the temperature results, the spread of the percent difference values for relative
humidity was much smaller than those for velocity. Again, Munich featured the largest
spread in percent difference.

For the general thermal comfort guidelines from ASHRAE 55, humidity must maintain a
dew point temperature of 62.2°F. There is no lower limit specified for humidity. The dew
point temperature can be checked by comparing the air temperature and relative
humidity. If the relative humidity is above 50%, which it is for all scenarios, then a
simple approximation can be used to calculate the dew point temperature. Table 5.5
shows the calculated dew point temperatures based on the air temperature and relative
humidity averages.

108

𝑇 𝑑𝑝
= 𝑇 −
9
25
(100−𝑅𝐻 )

Dew Point Temperature Averages [°F]
Climate
Shade Condition
None Flat Saddle Wave Arch Point
Kuala Lumpur 73.822 75.248 74.084 74.036 72.742 74.014
Munich 35.268 38.004 38.538 36.428 35.846 37.272
Abu Dhabi 65.624 67.534 68.202 67.174 66.218 68.062

Table 5.5: Dew point temperature averages.

All of the dew point temperatures for Kuala Lumpur and Abu Dhabi are above the
maximum dew point temperature. Therefore the general thermal comfort model is not
appropriate for the relative humidity data.

The adaptive model for thermal comfort is better suited for the relative humidity data
because it has no humidity constraints.

5.2 PMV & PPD THERMAL COMFORT ANALYSIS
The human models in the simulations were valuable in analyzing the results. They served
as placeholders when analyzing the flow in the pedestrian area to examine how the air
moved around occupants. Additionally, with the temperature, relative humidity, and solar
radiation data, the human models represented what was occurring in terms of heat
transfer. The PMV and PPD values displayed on the human model were useful in giving

109

an idea of how occupants felt as a result of the velocity, temperature, relative humidity,
and metabolic rate; however it could not be used for further analysis.

Human comfort is dependent upon environmental and personal factors. The
environmental factors include temperature, thermal radiation, relative humidity, and air
velocity. The personal factors are activity level (metabolic rate) and clothing. The way in
which Simulation CFD runs the solutions, all the factors are included however clothing is
an assumed unknown value and not user controlled. If clothing is assumed, this will
make the thermal comfort results between a cold climate and warm climate very
different. For example, if the assumption of clothing is long pants and a short sleeved
shirt, this clothing level might be appropriate for a warm climate (like Kuala Lumpur),
but is certainly not appropriate for a colder climate (Munich). The assumed clothing level
does not make for a fair comparison between scenarios as the occupants are assumed to
layer themselves more heavily given the colder climate. This discrepancy is seen in the
PMV and PPD results for Munich. Figure 4.9 shows the PPD results for Munich, where
naturally, 100% is dissatisfied as it is too cold given the low clothing level.

Still, PMV and PPD can be compared to understand how the fabric membrane shades
influence thermal comfort. To accomplish this and to interpret PMV and PPD results for
the whole site, not just the human models, PMV and PPD were mapped over the site
using the tabular output data. Additionally, color mapping PMV and PPD values
compared to the air velocity, air temperature, and relative humidity led to conclusions

110

about which variables are more influential in calculating thermal comfort. The following
equations from Fanger’s thermal comfort model were used to calculate PMV and PPD
(Fanger 1970, Fanger 1982). Further explanation of the equations are in section 2.2.

𝑃𝑀𝑉 =,0.303𝑒 ;0.036𝑀 + 0.028-*(𝑀 −𝑊 )−3.96𝐸 ;8
𝑓 𝑐𝑙
,(𝑡 𝑐𝑙
+ 273)
4
−(𝑡 𝑟 + 273)
4
-
−𝑓 𝑐𝑙
𝑕 𝑐 (𝑡 𝑐𝑙
−𝑡 𝑎 )−3.05,5.73−0.007(𝑀 −𝑊 )−𝑝 𝑎 -−0.42,(𝑀 −𝑊 )
−58.15-−0.0173𝑀 (5.87−𝑝 𝑎 )−0.0014𝑀 (34−𝑡 𝑎 )+

𝑃𝑃𝐷 =100−95𝑒 ,;(0.3353𝑃𝑀𝑉 4
:0.2179𝑃𝑀𝑉 2
)-

Variables M, t
a
, t
r
, and V were taken from the simulation results and inputs. The vapor
pressure of the air, p
a
, was derived from t
a
and relative humidity using the psychrometric
chart and steam tables. To avoid the assumptions from the simulation results in regards to
clothing, unique I
cl
values were assigned to each scenario.

Kuala Lumpur Munich Abu Dhabi
f
cl
1.107 1.151 1.107
h
c
based on velocity results
I
cl
0.57 clo 1.01 clo 0.57 clo
M 115 W/m2
p
a
based on t
a
and RH results
R
cl
0.0883 0.1565 0.0883
t
a
based on temperature results
t
cl
based on M, R
cl
, p
a
, t
a

t
r
based on mean radiant temperature results
V based on velocity results
W 0

Table 5.6: Values and derivations of variables used to calculate PMV and PPD.
The resulting color maps are represented below in Figure 5.2. The placement of the color

111

maps is shown in the model plan in Figure 5.3.

Figure 5.2: Color maps for velocity, air temperature, relative humidity, PMV , and PPD for fabric
membrane wave shades in Munich.

Figure 5.3: Color map inserted in the model plan.

This specific example is of the fabric membrane wave shades in Munich. When the data
is presented in this manner, the relationship between the velocity and PMV and PPD can
be seen in the color maps. In regions where the velocity is lower (green), the translation
to PMV and PPD is a vote closer to 0 (green) and lower percentage persons dissatisfied
(green), respectively. Air temperature does not appear to have a strong influence on the

112

PMV and PPD values. In the Munich scenario, areas of higher temperature (white) would
assume to be more comfortable, but this correlation is not clear. This is due to the range
of temperatures being much smaller for temperature than the range of values for the
velocity. Relative humidity also does not have a strong correlation to the PMV and PPD.
However, areas of higher velocities (white) have a lower relative humidity (green). As
with temperature, relative humidity does not have a large range so the results seem less
influential.

The color maps also made it easy to compare the variance in the variables and PMV and
PPD across the same scenario. Figure 5.4 demonstrates this for PPD (for PMV see
Appendix E).

113

Figure 5.4: Color maps of PPD for all climate scenarios.

With a lower PPD being the goal (green), the comparison of color maps shows that the
models that have a fabric membrane shade have a lower PPD.

Table 5.7 compares the average PMV and PPD for the three scenarios. The PPD tells
what percentage of people are dissatisfied while the PMV describes how they are
uncomfortable (warm/cold).

114

Kuala Lumpur Munich Abu Dhabi

PMV PPD PMV PPD PMV PPD
None 1.31 40.77 -1.20 36.92 1.64 58.22
Flat 1.28 39.33 -1.01 27.97 1.61 56.92
Saddle 1.28 39.11 -0.90 23.22 1.83 68.51
Wave 1.14 32.42 -1.03 28.45 1.59 55.67
Arch 1.20 35.43 -0.81 20.82 1.53 52.72
Point 1.28 39.14 -0.96 26.31 1.79 66.20

Table 5.7: PMV and PPD averages for each scenario.

Figure 5.5 is a series of charts that represent the percent of improvement for each shade
form in the climates. Each PPD measurement that was calculated was compared to the
measurement from the same location in the no shade base case. The result is data that
shows where the most improvement occurred. For Munich, both the wave and arch
shades had all of the PPD measurements improve.

115

Figure 5.5: Percent improvement for PPD measurements.

116

The table confirms that for Munich, the arch shape has the lowest PPD. For Kuala
Lumpur, the wave had the lowest PPD, and for Abu Dhabi the lowest PPD was the arch
shape. From the velocity results in Chapter 4, these two shapes were found to have the
lowest velocities, thus, it makes sense that they have the lowest PPD values as it was
previously concluded that the PPD values are most influenced by the velocity.

With the exception of Abu Dhabi, the base case scenario with no shades has the highest
PPD, as expected. The uncovered pedestrian area subjects occupants to solar radiation
and higher air velocities, which influences the temperature and humidity levels.

For Kuala Lumpur (hot and humid) there is not a lot of variance in the PPD or PMV
values. This is because the input velocity was low (3 mph). As previously stated, the
input velocity is the most influential in the PMV and PPD results. The low input velocity
decreases the variance in the measured velocities, and therefore decreases the PMV and
PPD variances. Munich has considerably more variance, because it has a higher input
velocity (15 mph). Abu Dhabi also had more variance due to the higher input velocity (10
mph).

For Kuala Lumpur, the PMV hovered above slightly warm, for Munich it was above and
below slightly cool, and for Abu Dhabi it was near warm. Munich had the lowest PPD
values, even though it had the highest input velocity and lowest input temperature. This
leads to the possible conclusion that fabric membrane shades might best improve thermal

117

comfort in colder and windier climates.

None of the climate scenarios meet the ASHRAE acceptable ranges for thermal comfort
for interior spaces. For PMV acceptable values are between -0.5 and +0.5 and for PPD
acceptable values are less than 10%.

5.3 ADAPTIVE THERMAL COMFORT ANALYSIS
The PMV and PPD thermal comfort model interpreted the results better than the thermal
comfort calculations for Simulation CFD. However, the model was used despite the data
exceeding the limitations. The velocity averages of the scenarios exceeded the base
acceptable velocity (0.447 mph) as well as the elevated air speed maximum (3.579 mph).
The relative humidity data also did not fit the PMV and PPD thermal comfort model
because the dew point temperatures exceeded the maximum value.

The data was fit the adaptive thermal comfort mode. There are no velocity or humidity
limitations for this model. To fit the data to the model, the mean monthly outdoor air
temperature and indoor operative temperature are required. Table 5.8 shows the values
used for these inputs. The mean monthly outdoor temperature was set equal to the
average air temperature from each simulation. The operative temperature was calculated
from the following equation and assumptions (ASHRAE 2009, 9.7).

118

𝑡 𝑜 =
(𝑕 𝑟 𝑡 𝑟 + 𝑕 𝑐 𝑡 𝑎 )
𝑕 𝑟 + 𝑕 𝑐

𝑕 𝑟 = 4𝜀𝜎
𝐴 𝑟 𝐴 𝐷 (273.2+
𝑡 𝑐𝑙
+ 𝑡 𝑟 2
)
3

𝑕 𝑐 = 8.3𝑉 0.6

Where ε average emissivity of clothing, typically 0.95
σ Stefan-Boltzmann constant, 5.67E-8 W/(m
2
•K
4
)
A
r
/A
D
effective radiation area of body to DuBois surface area, 0.73

The t
r
, t
a
, t
cl
, and V values were used from the simulation outputs. The metabolic rate (M)
and clothing insulation (I
cl
)

values were the same as those used when calculating the
PMV and PPD model.

Kuala Lumpur Munich Abu Dhabi

t
a
[°F] t
o
[°F] t
a
[°F] t
o
[°F] t
a
[°F] t
o
[°F]
None 82.21 82.62 68.27 68.56 81.14 81.88
Flat 83.24 83.64 69.93 70.14 82.15 82.54
Saddle 82.40 82.80 71.37 72.50 82.89 83.30
Wave 82.10 82.51 68.00 68.32 81.97 82.36
Arch 82.03 82.44 68.93 69.17 81.41 81.81
Point 82.33 82.74 70.86 71.20 83.11 83.52

Table 5.8: Mean monthly outdoor temperature and indoor operative temperatures.

Mapping the data to the chart yields Figure 5.6. Both Kuala Lumpur and Abu Dhabi are
within the 90% acceptability limits of operative temperature. Munich ranges between the
90% acceptability limit and just outside the 80% limit because some of the operative

119

temperatures are too low. In order to be within the acceptable limits, the operative
temperature needs to increase.

Figure 5.6: Temperatures mapped over adaptable thermal comfort model.

From the specifications of the adaptive thermal comfort model, the shaded region should
not be extrapolated for data that lies outside of the range. This excludes some of the
Munich data points from the adaptive comfort model.

The adaptive model was less rigid and therefore more of the data is considered
acceptable. Unlike the PMV and PPD model, the adaptive model does not allow for clear
comparison between the fabric membrane shades.

120

5.4 FORM ANALYSES
The final part of the analysis was to understand why the different fabric membrane forms
produced different velocity, temperature, relative humidity, and thermal comfort results.
Conclusions were first drawn within each climate scenario, as the shade form was the
only variable that changed within the scenarios, then across the three climate scenarios.

5.4.1 Saddle Shape
The saddle fabric membrane form was the only shade that had opening on the top in
addition to the sides due to the overlapping pattern used. The saddle shade did not
significantly reduce the measured velocity, however, it did increase the temperature more
than other shades. The relative humidity varied between the climate scenarios. In Kuala
Lumpur and Munich the humidity decreased, and in Abu Dhabi it increased.

Figure 5.7: Overlapping pattern of the saddle shades created opening on the side and top.

Since the saddle shade configuration had openings on the top and sides, the air was
allowed to flow through and in between the shades. In the other fabric membranes, the
shades acted as more of a barrier, so the air velocities were decreased and diverted. The

121

direction of the airflow differed in the three scenarios, so air was able to enter the
pedestrian space through the sides and top of the shades and influence the pedestrian
area. The opening in the saddle shades did not block the air, resulting in the measured
velocities being similar to input velocities. The varied relative humidity, while slight, was
caused by the opened top and sides. The openings do not create an enclosed space such as
the other shades, so as the air moves freely through the top of the pedestrian space and
does not become stagnant and humidity does not increase.

In Kuala Lumpur the PMV and PPD did not differ greatly from the base case of no
shades. Since the input velocity of Kuala Lumpur was low, and velocity turned out to be
the most influential variable, this is expected. However, in both Munich and Abu Dhabi
the PMV and PPD differed greatly from the no shades base case. In Munich the thermal
comfort improved, but in Abu Dhabi the thermal comfort got worse. While it was
concluded that air temperature does not have such a weighted influence on thermal
comfort as velocity because of the smaller range, it is still a factor. In Munich, a
temperature increase was desired because the initial temperature was far below comfort
levels. Since the temperature increase for saddle shades was significant, this helped
improve the thermal comfort. In Abu Dhabi, the air temperature was already high, so a
great increase, which is what the saddle shade did, was not desirable.

5.4.2 Wave Shape
The wave fabric membrane from was enclosed on the tops and featured openings on the

122

side. Overall, the wave shade consistently decreased the velocity compared to the no
shade base case. It also decreased the measured temperature and increased the relative
humidity.

The significant velocity decrease is attributed to the profile of the wave shade. Figure 5.8
outlines the profile in red. This profile creates a zigzag pattern, and as it was repeated
over the pedestrian area, it created a length of the pattern. As air moved within the
pedestrian space, it travelled the path of the ridges and valleys and was slowed down.

Figure 5.8: Zigzag profile of the wave shade outlined in red.

As previously mentioned, solar radiation was a meaningful influence on the temperature.
Compared to no shades the air temperature measured for the wave shades only increased
slightly. This is because the shades protected the pedestrian area from solar radiation. The
fact that there was an increase in temperature was because the solar radiation impacted
the surrounding air and context. There was a small decrease in the measured relative
humidity. This was observed for all the models with the exception of the saddle shade.
While the humidity decreased, it was only slight when compared to the no shade base
case. The form of the wave shades encouraged air movement to alleviate the humidity,

123

but since the form also slowed down the air velocity significantly, the decrease in relative
humidity was not as great as it was for other forms.

In all three climate scenarios thermal comfort improved. For Kuala Lumpur, the fabric
membrane wave shade improved the thermal comfort the most. The wave shade
addressed the desirable changes in each scenario that were needed to achieve improved
thermal comfort. For Munich it decreased velocity and for Abu Dhabi it decreased
velocity in addition to moderating the temperature.

5.4.3 Arch Shape
The fabric membrane arch shade was fully enclosed on the top and sides. It behaved very
similarly to the wave shade in all variables with the exception of relative humidity. It
decreased velocity most out of all the shades when compared to the input values and also
slightly increased the air temperature. However, in terms of relative humidity, it
decreased for the arch shade, but much more than the wave shade. The similarities
between the arch and wave shade can be attributed to their similar profiles. Figure 5.9
outlines the similar profile of the arch shade. When this shade was repeated over the
pedestrian area, it created a zigzag over the length of the site.

124

Figure 5.9: Zigzag profile of the arch shade outlined in red.

The arch shade was the most influential membrane form in improving thermal comfort. It
was the second best fabric membrane shade for improving the PMV and PPD values for
Kuala Lumpur and the best for Munich and Abu Dhabi. Much like the wave shade, the
arch shade improved thermal comfort factors as needed per each climate.

5.4.4 Point Shape
Like the arch shade, the point shade was created a fully enclosed shade on the top and
sides. However, the point shade behaved more like the saddle shade. While it appears that
the profile of the point shape would be zigzagged like that of the wave and arch shades,
this is not the case. As the point becomes more slender near the top, the airflow to this
area diminishes. The reality of the point shape was that in the context geometry, the
airflow followed the profile outlined in red in Figure 5.10. A smoother, wave like profile
does not interact with airflow the same way as the rigid zigzag does.

125

Figure 5.10: Smooth profile of the point shade outlined in red.

The smoother profile did not decrease or divert the airflow, so the measured air velocities
were only slightly decreased from the input velocities. For all the climate scenarios, the
temperature increased a moderate amount. Also, like the saddle shade, the relative
humidity varied between climate scenarios. For all the cases though relative humidity
decreased compared to the input value.

In Kuala Lumpur the PMV and PPD did not differ greatly from the base case of no
shades. In Munich the thermal comfort improved, but in Abu Dhabi the thermal comfort
got worse. While it was concluded that air temperature does not have such a weighted
influence on thermal comfort as velocity because of the smaller range, it is still a factor.
In Munich, a temperature increase was desired because the initial temperature was far
below comfort levels. Since the temperature increase for point shades was moderate, this
helped to improve the thermal comfort. In Abu Dhabi, the air temperature was already
high, so a great increase, which is what the saddle shade did, was not desirable.

126

CHAPTER 6 : CONCLUSIONS FROM AIRFLOW CFD STUDY AND
THERMAL COMFORT ANALYSIS

In Chapter 1, it was hypothesized that:

i) Fabric membrane forms positively influence the airflow patterns, air velocity,
temperature, and relative humidity of the space they are covering.
ii) The positive influence will improve thermal comfort conditions as measured
by PMV and PPD defined by ASHRAE 55 and ISO 7730 when compared to
scenarios with no shades or flat shades.

The first hypothesis statement proved partially true, in regards to air velocity and
temperature specifically. The airflow patterns observed between the four fabric
membrane forms and the two base cases for each climate scenario had limited variance.
Also, the measured relative humidity remained inconclusive as to its behavior as the same
shade varied between increasing and decreasing the relative humidity measured between
different climate scenarios. Additionally, the range of relative humidity data was small as
was the air temperature data. However, clear trends in the measured velocity were
observed.

The second hypothesis statement proved to be true. The color mapped PMV and PPD
images in partnership with the averages demonstrated that with the exception of two
models, the remaining ten model scenarios improved the thermal comfort.

127

The conclusions should only be applied to the model and scenario detailed in the previous
chapters unless otherwise specified.

6.1 THERMAL COMFORT: Improved by fabric membrane shades
On average, the fabric membrane shades improved the thermal comfort conditions of
each scenario when compared to the no shade base case. The two exceptions to this were
the saddle and point shade scenarios in Abu Dhabi, which performed worse.

Comparing to the base case of no shades, 83.3% of all the fabric membrane shade
scenarios tested improved thermal comfort. Of the improved scenarios, 90% also had
improved thermal comfort when compared to the flat shade base case, or 75% of all the
fabric membrane shade scenarios tested had an improved thermal comfort compared to
flat shades. This demonstrates that not only do the fabric membrane shades improve
thermal comfort compared to no shades, but they also improve thermal comfort compared
to flat shades.

The best case thermal comfort improvement was for the arch shade in Munich which
decreased PPD by 44% from the no shade base case.

128

6.2 VELOCITY: Decreased by wave and arch shades
The measured velocity varied the most compared to temperature and relative humidity.
Due to the range of values, it also had the most control over the thermal comfort results.

The fabric membrane forms influencing thermal comfort most were the wave and arch
shades. These two shapes decreased in the input velocity for all the climate scenarios the
most. They also had the lowest velocities when compared to the no shade base case and
flat shade base case. These two shade forms were successful in reducing the input
velocities so that they were closer to an acceptable range for occupant comfort.
Therefore, these two forms had improved thermal comfort metrics. These two forms
warrant the most future study.

6.3 TEMPERATURE AND RELATIVE HUMIDITY: Unknown correlation to form
The temperature and relative humidity results were important in measuring how the
fabric membrane form shades influenced them. However the results were not as clear as
those drawn from the velocity and thermal comfort results. Therefore it is concluded that
temperature and relative humidity are less influential in determining the PMV and PPD
because their range was smaller than the velocity range.

Given the measured air temperature of the base case with no shades, the fabric membrane
shade simulations varied in whether the air temperature increased or decreased.
Unfortunately the increases and decreases were not consistent amongst the different

129

climate scenarios. The fabric membrane forms appeared to have no influence on the
measured air temperature.

The measured relative humidity yielded the same results. Compared to the base case with
no shades, the fabric membrane shade simulations increased and decreased the relative
humidity with variation, but there was no consistency between climate scenarios. The
fabric membrane forms appeared to have no influence on the measured relative humidity
against what the hypothesis assumed.

6.4 APPLICATION OF CONCLUSIONS
The results of this research serve as an introduction into the ability to use fabric
membrane shades to influence occupant thermal comfort. In terms of suggestions for
designers, they are limited to the findings of the study and serve as a stepping stone for
further study and application. The suggestions are based on the thermal comfort
conclusions and influence of the forms on air velocity and air temperature. Suggestions
should not be made regarding relative humidity because the results were inconclusive.

The saddle shade had no strong influence on air velocity, however it did increase air
temperatures. Given a cool environment where winds are at an acceptable speed and do
not need to be changed, a saddle shade would be appropriate as it will slightly increase
the temperature but not alter the wind speeds.

130

The wave shade was one of the most successful forms for improving thermal comfort. It
decreased air velocity and temperature. In an environment with high wind speeds, a wave
shade would be appropriate because it will reduce the wind speeds to a more tolerable
level.

The arch shade was also quite influential in improving thermal comfort. It decreased air
velocity and increased temperature. Given this information, a cold climate with high wind
speeds will most likely have the most thermal comfort improvement if an arch shade is
used because it will decrease wind speeds and increase air temperature.

The point shade caused a slight decrease in air velocity and an increase in temperature. It
is suitable for cool environments medium wind speeds that only need to be reduced
slightly.

Finally it is important to note that the measures of thermal comfort in this research stem
from metrics used to evaluate indoor thermal comfort, while this research was conducted
for an outdoor environment. These metrics are more stringent than what is considered
tolerable for occupants of an outdoor space.

6.5 VALIDITY OF STUDY
Verification and validity studies for Simulation CFD are publically available on the
wikiHelp website (http://wikihelp.autodesk.com/Simulation_CFD/enu/2013/Help/0407-

131

Learning407/0807-Verifica807). For the solution specifications of the simulations
conducted in this research, the published verification studies project margins of error
between 0.092% and 4.25% against benchmarked values.

The remaining question is how realistic was the scenario that was modeled, and how
transferrable are the conclusions from the simulations to real scenarios. The setup was
designed to be a typical pedestrian area. Figure 3.9 shows actual examples that inspired
the scenario. Previous scenarios were simulated in which the shopping center was
modeled on a greater scale, with shades modeled at a height of 40 feet, and the results
were inconclusive due to lack of consistency. A stadium scenario was also modeled, but
since the scale of the site was so large in comparison to the scale of the desired results
(human model thermal comfort), the results were also inconclusive. Therefore, the site
model from which conclusions were made was sized as defined in Chapter 3.

The scenarios were modeled to be as accurate to real scenarios as possible. Best practices
for architectural and engineering purposes were followed in addition to subjecting the
simulation setup to review by a CFD software engineer. Additionally, the climate
scenarios were designed to explore a variety of input parameters.

The biggest hurdle in the replication of the study was defining the mesh. Using the
meshing regions and definitions defined in Chapter 3, it should be possible to replicate
the results. There might be a slight difference of the location of the nodes, however there

132

is the option to apply the mesh used in the scenarios to different models to assure that the
values measured at the nodes would be more replicable.

6.6 PROCESS OVERVIEW
The software workflow defined in Chapter 3 proved to be effective for the purpose of
this study. Minimal flaws were discovered during the process. However these issues did
not create a roadblock or affect the results of the study to the knowledge of the author.

Formfinder proved to be excellent for creating the anticlastic geometries, however
integrating the anticlastic curvature with the site geometries was not always effortless.
Formfinder states that site geometries could easily be imported and then the anticlastic
curvatures modeled in the site geometry, but this process proved to be difficult. Instead,
the site geometry and anticlastic geometry were modeled separately and assembled in
Revit
®
.

Occasionally the assembled models would not launch from Revit
®
into Simulation CFD.
The reason for this error was unknown, but it is assumed that the anticlastic curvatures
were too complex. The solution was to launch the model from Inventor
®
into Simulation
CFD. This workaround was suitable, but again, there is no explanation as to why a model
would not launch from Revit
®
, but would successfully launch from Inventor
®
.

It took numerous attempts and model versions in the beginning to get the models to

133

import, mesh, and solve. The reason many of the early models did not solve was because
of meshing errors. There was a balance of trying to simplify the model’s mesh in order
for it to be solvable without too many elements, but also without losing the integrity of
the anticlastic curvature. Because of these difficulties, more complex fabric membrane
forms were not tested.

Aside from these difficulties, the software workflow and the programs selected proved to
be suitable. While there were problems, the overall conclusion was that a better software
workflow could not be devised with what was currently available. It was acceptable to
work within the designed workflow constraints.

6.7 SUMMARY
The objective of this research was to determine whether or not fabric membranes forms
improved thermal comfort and which parameters thereof they influenced. With high
performance and large-scale fabric membranes becoming increasingly popular in the
modern building industry, this research served to inform designers and engineers how
their designs are impacting occupant comfort. From the research it was found that for an
open environment scenario where the fabric membranes act as shades, thermal comfort
can be influenced by the fabric membrane form. This influence comes from the shades’
ability to decrease velocity. Given an input velocity that is outside of the comfort zone,
the fabric membrane shade that decreases it the most will have the best thermal comfort
metrics of PMV and PPD.

134

This body of research can be used as a reference for predicted behavior when designing
fabric membrane forms. Based on the climate and requirements to improve thermal
comfort, design suggestions can be made as to which form will help to achieve the
thermal comfort needs. The suggestions of which form to use depends on the intentions
of how to influence the thermal comfort needs, defined through an increase or decrease in
velocity, air temperature, and relative humidity. It is also a stepping stone for further
research.

135

CHAPTER 7 : FUTURE WORK FOR FABRIC MEMBRANE
AIRFLOW AND THERMAL COMFORT ANALYSES

The scope of this thesis was controlled by constrains of resources and time. Given access
to indefinite and reliable resources in addition to significantly more time, the following
studies within similar scopes of this thesis should be pursued. The studies are organized
in to work that stems from the conclusions of this thesis, and work that expands the
conclusions.

7.1 CONTINUING WORK
This research explored the idea that fabric membrane forms have the potential to improve
thermal comfort. The conclusions that were formulated should be used as a launching
point for further studies that will build upon the data in this field. Based on conclusions
from this research, the following are possible research options to continue and expand
this study.

Form Analysis
The wave and arch forms were found to be the most influential in decreasing the velocity.
One could conduct a study that focuses on these two forms and explores how controlling
the geometry affects the results. This work could further inform the design community of
how membrane form decisions impact occupant comfort.

136

Site Analysis
One could explore how the site context and materials impact the results by changing the
site geometry and form, making the scale larger, introducing more pedestrian walkways,
and changing the material of the membranes. Additionally, one could change the program
of the space and, the associated site variables. This study would build upon the current
results and expand the design suggestions. It would make the conclusions more
applicable to different site scenarios. Finally one could use the results to better predict the
behavior of air movement through a pedestrian space.

Case Study
Using an existing site with fabric membrane shades, one could explore alternative design
options. The methodology would be to model the current scenario, then use the design
suggestions from this study to improve the thermal comfort. This study could produce
actionable results that can affect real architectural spaces.

7.2 OTHER WORK
The following potential studies include detailed analyses of fabric membranes and
airflow. Although beyond the scope of the initial investigation, useful information may be
determined. Conclusions concerning internal and external airflow relationships, CFD
engine verification, wind tunnel verification, dynamic and static reaction relationships,
and material properties may be made.

137

Internal & External Investigation
One could study how the air behaves on both sides of the fabric membrane and determine
if the behavior on the external side of the membrane influences the internal behavior of
the air. Further variables might be the material properties of fabric membranes such as
permeability, coatings, and tension of the fabric.

Comparative CFD Verification Study
Using different CFD software programs, one could conduct a verification study and
determine how results differ between software tools such as those previously mentioned
in section 3.2. This study will be useful in serving as a knowledge base for architectural
CFD studies.

Wind Tunnel Comparison
One could conduct wind tunnel tests on scaled fabric membrane models and study how
the results compare to CFD results. Additional value would be added by monitoring
actual sites and comparing the measurements to wind tunnel and CFD results.

Dynamic Behavior
The fabric membrane could be modeled so that it behaves dynamically. Proprietary
simulation tools exist that can accomplish this could be used or tools developed based on
existing or new algorithms. Then the results could be compared with the data about static
behavior to examine the influence dynamic reaction has on airflow results. This study can

138

determine the importance of dynamic reaction for fabric membranes.

Materiality
One could observe and quantify the differences between using different membrane
materials in a CFD analysis. Properties such as permeability, thickness, light
transmission, coatings, tensile strength, elasticity, and weight should be tested.

7.3 CONCLUSIONS
The airflow research that was conducted for fabric membrane structures should be used
as a stepping stone for further studies. Through the research, it was demonstrated that
measuring occupant thermal comfort within a pedestrian space covered by fabric
membrane shades was possible. It also demonstrated that fabric membrane shades can
improve occupant thermal comfort when compared to no shades and flat shades. The next
step in furthering the research would be to test more climate and site scenarios to
replicate and expand the results. Analyzing more site and climate conditions can confirm
the conclusions from this research and expand the knowledge base of how fabric
membranes impact occupant thermal comfort.

139

BIBLIOGRAPHY

Archi Expo. “Membrane-cable tensile structure (for zoo).” Photograph. 2009. Accessed March 4, 2013.
http://img.archiexpo.com/images_ae/photo-g/membrane-cable-tensile-structure-for-zoo-61063-
1906043.jpg

Architen Landrell Associates Ltd. “Swindon Parade.” Photograph. Accessed December 4, 2012.
http://www.architen.com/projects/swindon-parade

ASHRAE. 2001. Fundamentals: 2001 ASHRAE Handbook. American Society of Heating, Refrigerating
and Air-Conditioning Engineers, Inc.

ASHRAE. 2010. ANSI/ASHRAE Standard 55-2010. Atlanta: American Society of Heating, Refrigerating
and Air-Conditioning Engineers, Inc.

ASHRAE. 2009. Fundamentals: 2009 ASHRAE Handbook. American Society of Heating, Refrigerating
and Air-Conditioning Engineers, Inc.

Autodesk. 2012. “Autodesk WikiHelp: Simulation CFD 2013.” Accessed September 23.
http://wikihelp.autodesk.com/Simulation_CFD/enu/2013.

Autodesk
®
AutoCAD
®
Civil 3D
®
(Version 2012) [Software]. 2011. Retrieved from
http://students.autodesk.com/?nd=download_center

Autodesk
®
Inventor
®
Fusion (Version 2012) [Software]. 2011. Retrieved from
http://labs.autodesk.com/technologies/fusion

Autodesk
®
Project Falcon (Version Beta) [Software]. 2012. Retrieved from
http://labs.autodesk.com/utilities/falcon

Autodesk
®
Project Vasari (Version Beta 2) [Software]. 2012. Retrieved from
http://autodeskvasari.com/VasariBeta2

Autodesk
®
Revit
®
(Version 2013) [Software]. 2012. Retrieved from
http://students.autodesk.com/?nd=download_center

Autodesk
®
Simulation CFD 360 (Version 2013) [Software]. 2012. Retrieved from
http://students.autodesk.com/?nd=download_center

Autodesk Sustainability Workshop. 2011. “Stack Ventilation and Bernoulli’s Principle.” Accessed January
15, 2013. http://sustainabilityworkshop.autodesk.com/buildings/stack-ventilation-and-bernoullis-
principle.

Baskaran, A., and B. Murty, J. Wu. 2008. “Calculating roof membrane deformation under simulated
moderate wind uplift pressures.” Engineering Structures 31: 642-650. Accessed September 12, 2012.
doi: 10.1016/j.engstruct.2008.10.013.

Bechthold, Martin. 2008. Innovative Surface Structures: Technologies and Applications. Taylor & Francis.

BirdAir. 2008. “Fabric Roof for All Conditions.” Building Design & Construction.49:8.

140

BirdAir. 2011. “Tensotherm.” Accessed February 10, 2013.
http://tensothermroofing.com/explore/index.php#section1

BirdAir. 2012. “Sustainable Design.” Accessed October 16, 2012.
http://www.birdair.com/sustainableDesign/default.aspx

BirdAir. “Paseo Acoxpa.” Photograph. Accessed December 4, 2012.
http://www.birdair.com/projectGallery/paseo_acoxpa.aspx

BirdAir. “San Diego Convention Center.” Photograph. Accessed September 12, 2012.
http://www.birdair.com/projectGallery/san_diego_convention_center.aspx

Brakke, Ken. 2009. “Triply Periodic Minimal Surfaces.” Accessed January 28, 2013.
http://www.susqu.edu/brakke/evolver/examples/periodic/periodic.html

Cradle scSTREAM (Version 1304.2) [Software]. 2011. Retrieved from http://www.cradle-
cfd.com/products/stream/index.html

Croome, Derek J. 1985. “Acoustic design for flexible membrane structures.” Applied Acoustics. 18.

Devulder, T., and R. Wilson, J. Chilton. “The thermal behavior of buildings incorporating single skin tensile
membrane structures.” International Journal of Low Carbon Technologies 2:195-213.

Dutfield, Bill. “Germany Pavilion.” Photograph. 2008. Accessed January 15, 2013.
http://expo67.ncf.ca/expo_german_p14.html

Ebers, Georg. “Tent of Bedaween Arabs.” Illustration. 1878. Accessed January 25, 2013.
http://commons.wikimedia.org/wiki/File:Tent_of_Bedaween_Arabs_(1878)_-_TIMEA.jpg

Engineering Toolbox. “Fluid Mechanics.” Accessed January 28, 2013.
http://www.engineeringtoolbox.com/fluid-mechanics-t_21.html

Fabric Architecture. “Hideaway Park.” Photograph. Accessed September 12, 2012.
http://www.fabricarchitecture.com/project/hideaway-park-colorado-usa

Fanger, P.O. 1970. Analysis and Applications in Environmental Engineering. McGraw-Hill Book Company,
New York.

Fanger, P.O. 1982. Thermal Comfort. Krieger Publishing Company, Malabar, Florida.

Hernandez, J. J. M. 2006. “Fabric Membranes as Daylighting Control Systems in Buildings.” PhD diss.,
University of Nottingham.

Huntington, Craig G. 2004. Tensioned Fabric Roof. Reston: American Society of Civil Engineers.

Incorpera, F.P., and D.P. DeWitt. 1981. Fundamentals of Heat Transfer. Wiley, University of Michigan.

Innovative Tensile. “Tensile Membrane Structures.” Photograph. Accessed September 12, 2012.
http://innovativetensile.in/images/mambrane/4.jpg

Integrated Environmental Solutions Limited (Version 6.0) [Software]. 2011. Retrieved from
http://www.iesve.com/software/download-centre

141

Jeong, W. Y., and S.K. An. 2001. “The transport properties of polymer membrane-fabric composites.”
Journal of Materials Science 36: 4797-4803.

Koch, Klaus-Michael. 2004. Membrane Structures. Prestel Publishing.

Light Weight Structures Architectural Solutions. “Yas Island.” Photograph. 2011. Accessed March 4, 2013.
http://ltwsas.files.wordpress.com/2013/01/yas-island-bbq-play-area-canopies-1-edit.jpg

Liggett, R. and M. Milne. 2012. Climate Consultant (Version 5.3) [Software]. Retrieved from
http://www.energy-design-tools.aud.ucla.edu/

M E+C. “A Functional Definition of Computational Fluid Dynamics.” Accessed August 31 2012.
http://www.me-c.it/index.php/cfd/wathis

Mark, Earl. 2011. “Optimizing Solar Insolation in Transformable Fabric Architecture: A Parameter Search
Design Process.” University of Virginia.

Mitchell Science. 2011. “Bernoulli Principle.” Accessed October 8, 2012.
http://mitchellscience.com/bernoulli_principle_animation

Miyauchi, H., and N. Katou, K. Tankha. 2010. “Behavior of a mechanically anchored waterproofing
membrane system under wind suction and uniform Pressure.” Building and Environment 46: 1047-
1055. Accessed September 12, 2012. doi: 10.1016/j.buildenv.2010.11.005.

Mojtahedi, Arad. “Munich Olympic stadium.” Photograph. 2006. Accessed January 28, 2013.
http://commons.wikimedia.org/wiki/File:Olympiastadion_Muenchen.jpg

Nazarboland, M.A., and X. Chen, J.W.S. Hearle, R. Lydon, M. Moss. 2007. “Effect of Different Particle
Shapes on the Modeling of Woven Fabric Filtration.” International Journal of Clothing Science and
Technology 20:2; 150-160. Accessed on September 17, 2012. doi: 10.1108/09556220810855201.

North Sails. 2013. “North Design Suite.” Accessed January 17, 2013.
http://www.na.northsails.com/TECHNOLOGY/NorthDesignSuite/tabid/23145/Default.aspx.

Palmer, G., and B. Vazquez, G. Knapp, N. Wright. 2003 “The Practical Application of CFD to Wind
Engineering Problems.” Paper presented at Eighth International IBPSA Conference, Eindhoven,
Netherlands, August 11-14.

Pieren, R. 2012. “Sound absorption modeling of thin woven fabrics backed by an air cavity.” Textile
Research Journal 82: 864-874. Accessed September 12, 2011. doi: 10.1177/0040517511429604.

Pombo, Juan Cruz Molina. 2008. “Mechanical Characterization of Fabrics for Inflatable Structures.”
Master thesis. West Virginia University.

Rief, S., and E. Glatt, E. Laourine, D. Aibibu, C. Cherif, A. Wiegmann. 2011. “Modeling and CFD-
Simulation of Woven Textiles to Determine Permeability and Retention Properties.” AUTEX Research
Journal 11.3: 78-83.

Roh, Jin-ho. 2008. “Thermoelastic Behaviors of Fabric Membrane Structures.” Advanced Composite
Material. 17: 319-328. Accessed January 16, 2013. doi: 10.1163/156855108X385285.

Scheuermann, Rudi, and Kieth Boxer. 1996. Tensile Architecture in the Urban Context. Butterworth
Architecture.

142

Schierle, G. G. 1968. Lightweight Tension Structures. UC Berkeley.

Schierle, G. G. 1969. “New Applications Of Industrial Fabrics In Architecture.” Industrial Fabric Products
Review.

Schierle, G. G. 2008. Structure and Design. Cognella.

Seidel, Michael. 2009. Tensile Surface Structures: A Practical Guide to Cable and Membrane
Construction. Ernst & Sohn.

Shi, Xing, and Eric Burnett. 2007. “Mechanics and test study of flexible membranes ballooning in three
dimensions.” Building and Environment 43: 1871-1881. Accessed September 12, 2012. doi:
10.1016/j.buildenv.2007.11.002.

Sivaprasad, Neha. 2006. “Wind design of fabric structures: determination of gust factors for fabric
structures.” Master thesis, University of Southern California.

Structureflex. “Tensile PVC Membranes.” Photograph. Accessed September 12, 2012.
http://www.structureflex.com.au/PVCMembranes.php

Susila, I Gede Adi. “Wind load predicting; how could CFD replace Wind Tunnel Test?” University of
Newcastle upon Tyne.

Urban Environmental Programme. 2010. “Greening Stadia: Green Point Stadium.” Department
Environmental Affairs and Tourism Republic of South Africa.

Wehdorn-Roithmayr, Robert. 2012. Formfinder Software GmbH (Version 3.5) [Software]. Retrieved from
http://www.formfinder.at/main/downloads/

Wikimedia Commons. “Bukit Nanas KL Monorail station.” Photograph. 2006. Accessed March 4, 2013.
http://commons.wikimedia.org/wiki/File:KLMonoRail-Top.JPG

Xiao, Xueliang, and Xuesen Zeng, Palitha Bandara, Andrew Long. 2012. “Experimental study of dynamic
air permeability for woven fabrics.” Textile Research Journal 82: 920-930. Accessed September 12,
2012. doi: 10.1177/0040517511429605.

143

APPENDIX A: ADDITIONAL TABLES FOR HUMAN COMFORT
METRICS

Table A.1
Metabolic rates for different activities according to Chapter 8 of the 2001 ASHRAE Handbook-
Fundamentals.

Activity Metabolic Rate [W/m
2
] Metabolic Rate [Btu/h•ft
2
]
Sleeping 40 13
Reclining 45 15
Sitting quietly 60 18
Standing 70 22
Walking slowly (2.0 mph, 0.9 m/s) 115 37
Walking moderately (2.7 mph, 1.2 m/s) 150 48
Walking briskly (4.2 mph, 1.8 m/s) 220 70
Seated, reading or writing 60 18
Typing 65 20
Filing (seated) 70 22
Filing (standing) 80 26
Walking about an office 100 31
Lifting/packing 120 39
Cooking 95-115 29-37
House cleaning 115-200 37-63
Dancing 140-255 44-81
Exercise 175-235 55-74
Playing tennis 210-270 66-74
Playing basketball 290-440 92-140

144

Table A.2
Clo values for clothing ensembles according to Chapter 8 of the 2001 ASHRAE Handbook-Fundamentals.

Clothing clo Value
Trousers and short-sleeve shirt 0.57
Trousers and long-sleeve shirt 0.61
Trousers, long-sleeve shirt, and suit jacket 0.96
Trousers, long-sleeve shirt, suit jacket, vest, and T-shirt 1.14
Trousers, long-sleeve shirt, long-sleeve sweater, and T-shirt 1.01
Trousers, long-sleeve shirt, long-sleeve sweater, T-shirt, suit jacket, and long underwear bottoms 1.3
Knee-length skirt and short-sleeve shirt, with sandals 0.54
Knee-length skirt, long-sleeve shirt, and full slip 0.67
Knee-length skirt, long-sleeve shirt, half slip, and long-sleeve sweater 1.1
Knee-length skirt, long-sleeve shirt, half slip, and suit jacket 1.04
Ankle-length skirt, long-sleeve shirt, and suit jacket 1.1
Walking shorts and short-sleeve shirt 0.36
Long-sleeve coveralls and T-shirt 0.72
Overalls, long-sleeve shirt, and T-shirt 0.89
Insulated coveralls, long-sleeve thermal underwear tops and bottoms 1.37
Sweat pants and long-sleeve sweatshirt 0.74
Long-sleeve pajama tops, long pajama trousers, short 3/4 length robe (slippers, no socks) 0.96

145

APPENDIX B: DETAILED CLIMATE DATA

The following charts are taken from Climate Consultant 5.3.

Kuala Lumpur Climate Data

Temperature Range

Hourly Temperature and Solar Radiation Averages

146

Average Wind Velocities

Annual Wind Rose

147

Daylit Hours and Solar Radiation Averages

Hourly Dry Bulb and Humidity Averages

148

Average Ground Temperature

Psychrometric Chart

149

Munich Climate Data

Temperature Range

Hourly Temperature and Solar Radiation Averages

150

Average Wind Velocities

Annual Wind Rose

151

Daylit Hours and Solar Radiation Averages

Hourly Dry Bulb and Humidity Averages

152

Average Ground Temperature

Psychrometric Chart

153

Abu Dhabi Climate Data

Temperature Range

Hourly Temperature and Solar Radiation Averages

154

Average Wind Velocities

Annual Wind Rose

155

Daylit Hours and Solar Radiation Averages

Hourly Dry Bulb and Humidity Averages

156

Average Ground Temperature

Psychrometric Chart

157

APPENDIX C: FABRIC SHADE DIMENSIONS

Figure C.1
Saddle Shade Dimensions

158

Figure C.2
Wave Shade Dimensions

159

Figure C.3
Arch Shade Dimensions

160

Figure C.4
Point Shade Dimensions

161

APPENDIX D: ADDITIONAL AIRFLOW PATTERNS

Figure D.1
Airflow patterns in the xz-plane.

162

Figure D.1
Airflow patterns in the yz-plane.

163

APPENDIX E: ADDITIONAL PMV AND PPD COLOR MAPS

Figure E.1
PMV Color Map

164

Figure E.2
PPD Percent Difference Color Maps

The following image represents the percent difference of the fabric membrane shades compared to the no
shade base case scenario. The scale ranges from -50% to +50%. A negative percent difference denotes an
improvement in the PPD value (PPD decrease) and is shown by green.

Abstract (if available)

Abstract In the current building industry, fabric membranes are often a choice construction when structural behavior, geometry, and cost are determining factors. Environmental performance is only considered an added benefit, and the research and analysis is commonly an afterthought. As fabric membranes become more present in the modern building industry, their environmental performance qualities need to be further evaluated and measured against energy standards. They offer unique performance qualities in the metrics of daylighting, solar radiation, and possibly acoustics. A limited amount of research has been conducted in how fabric membranes can enhance natural ventilation and occupant thermal comfort, but it can be proposed that when studying airflow patterns, fabric membrane structures will also behave differently from a static building material. Furthermore, they will improve human thermal comfort as a result of their unique anticlastic geometries. ❧ This thesis examines what airflow patterns occur within fabric membrane structures and how architects and engineers can use these forms to enhance occupant thermal comfort in their designs. The study assesses four basic fabric forms serving as shades over a pedestrian area through airflow patterns, air velocities, temperatures, humidity, Predicted Mean Vote (PMV), and Predicted Percentage of Dissatisfied (PPD) in three different climates: hot and humid, temperate, and dry and arid. The metrics were measured in terms of human comfort defined by ASHRAE Standard 55 and ISO 7730.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Night flushing and thermal mass: maximizing natural ventilation for energy conservation through architectural features

PDF

Ventilation shaft to increase effectiveness of natural ventilation

PDF

The form finding of tensile membranes - a tool for architects and designers

PDF

Environmentally responsive buildings: multi-objective optimization workflow for daylight and thermal quality

PDF

Improving thermal comfort in residential spaces in the wet tropical climate zones of India using passive cooling techniques: a study using computational design methods

PDF

Enhancing thermal comfort: data-driven approach to control air temperature based on facial skin temperature

PDF

Developing environmental controls using a data-driven approach for enhancing environmental comfort and energy performance

PDF

Exploration for the prediction of thermal comfort & sensation with application of building HVAC automation

PDF

Adaptive façade controls: A methodology based on occupant visual comfort preferences and cluster analysis

PDF

Multi-occupancy environmental control for smart connected communities

PDF

Enhancing thermal comfort: air temperature control based on human facial skin temperature

PDF

Thermal performance of a precast roof assembly: achieving comfort using dynamic insulation and photovoltaics in an extreme climate designed for a small residence at Joshua Tree National Park

PDF

A high-performance SuperWall: designed for a small residence at Joshua Tree National Park

PDF

Visualizing thermal data in a building information model

PDF

Impacts of indoor environmental quality on occupants environmental comfort: a post occupancy evaluation study

PDF

Human-building integration based on biometric signal analysis: investigation of the relationships between human comfort and IEQ in a multi-occupancy condition

PDF

Economizer performance and verification: effect of human behavior on economizer efficacy and thermal comfort in southern California

PDF

Tilted glazing: angle-dependence of direct solar heat gain and form-refining of complex facades

PDF

Exploring the influence of scheduling on daylighting performance evaluation in assisted living: testing two new metrics based on sDA and ASE

PDF

The intelligent control strategy of kinetic façades for daylight and energy performance: evaluating the daylight effect of adaptive systems based on parametric workflow

Asset Metadata

Creator Egger, Stephanie E. (author)

Core Title Airflow investigation of fabric membrane forms: a fluid dynamic analysis for thermal comfort

School School of Architecture

Degree Master of Building Science

Degree Program Building Science

Publication Date 05/31/2013

Defense Date 05/03/2013

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag fabric membrane,natural ventilation,OAI-PMH Harvest,Predicted Mean Vote,Predicted Percentage of Dissatisfied,thermal comfort

Format application/pdf (imt)

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Schierle, G. Goetz (committee chair), Chiu, Simon K. (committee member), Kensek, Karen M. (committee member), Simmonds, Peter (committee member)

Creator Email stephaniegger@gmail.com

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c3-279914

Unique identifier UC11295263

Identifier etd-EggerSteph-1705.pdf (filename),usctheses-c3-279914 (legacy record id)

Legacy Identifier etd-EggerSteph-1705.pdf

Dmrecord 279914

Document Type Thesis

Format application/pdf (imt)

Rights Egger, Stephanie E.

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the a...

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA