Thermal Comfort in The Havelis of Jaisalmer

See
discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/33776485
Thermal comfort in havelis of Jaisalmer.

ARTICLE
Source: OAI
READS
533
1 AUTHOR:
Jane Matthews
Curtin University
31 PUBLICATIONS 54 CITATIONS
SEE PROFILE
Available from: Jane Matthews

Retrieved on: 21 February 2016
THERMAL COMFORT IN THE HAVELIS OF

JAISALMER
A STUDY OF THE THERMAL PERFORMANCE OF AN URBAN
COURTYARD HOUSE.
by
Jane Matthews
Thesis submitted in partial fulfilment of the requirements for the degree of

Doctor of Philosophy
School of Architecture, University of East London, June, 2000
ABSTRACT
This thesis provides a detailed case study of the havelis of Jaisalmer, Rajasthan. Jaisalmer
was selected as the study area not only because of the abundance and richness of the havelis
in this region, but also because of the extreme nature of the climate in and around the area. In
addition, the haveli has the added advantage of being a high-density urban form;
economically and ecologically significant in the context of India's rapidly expanding urban
population and associated energy and environmental crisis.
Environmental data was collected in and around two havelis within the fort, over a period of
twelve months, focusing on the larger of the two monitored buildings, Hotel Suraj. In
addition a short transverse comfort survey was conducted on the streets of Jaislamer, and
subsequently evaluated in the wider context of accepted adaptive model comfort prediction
equations. The results of the survey are applied to the assessment of the collected
environmental data, in terms of predicted occupant thermal comfort.
The relationship between indoor and outdoor temperature is examined in detail and an
attempt made to develop indoor temperature prediction equations for each of four long term
monitored zones in Hotel Suraj. With all results revealing mass as the governing factor for
the modification of climate in the havelis, a substantial portion of this thesis is subsequently
focused on developing methodologies for determining the decrement factor and time lag of
indoor zones in relation to outdoors. A new finding is the impossibility of a shift in daily
temperature cycle of more than 6 hours (or 0.25 of the cycle frequency).
Considerable effort has been expended on the visual presentation of data for this study. This
has involved the development of two stand-alone computer programs for the presentation of
thermal data, included on the CD at the end of this thesis.
ii
TABLE OF CONTENTS
Abstract
Acknowledgements
...
ii
...
viii
CHAPTER 1 - INTRODUCTION
1.1
1.2
1.3
1.4
1.5
1.6
Global concerns ...

India's history
...
India today
...
Traditional versus modern ..
Aims
...
Form of thesis ...
References
...
2
2
3
5
5
6
8
CHAPTER 2 - JAISALMER
2.1
2.2
2.3
2.4
Geography
Climate
History
Jaisalmer today
References
...
11
...
12
...
14
...
19
...
21
CHAPTER 3 - THE HAVELIS

3.1
3.2
Introduction
Building surveys
3.2.1
3.2.2
3.2.3
Hotel Suraj
The second haveli
Nohra
3.3
3.4
Discussion
...
Plans, sections and elevations .
References
...
...
23
...
24
...
25
31
33
...
...
33
34
41
CHAPTER 4 - ENVIRONMENTAL DATA COLLECTION

4.1
4.2
4.3
Introduction
...
Field study periods
Equipment
...
43
4.3.1
Temperature
44
...
43
44
iii
4.3.2
4.3.3
4.3.4
Humidity
...
...
Air movement
Surface temperature.
45
45
46
4.4
Data collected
...
47
4.4.1
Summer95 series
...
...
47
47
48
50
52
52
53
54
55
55
56
57
58
58
59
60
60
61
61
62
65
66
67
68
...
69
...
71
4.4.1.1 Logger locations

4.4.1.2 Results
4.4.1.3 Discussion
4.4.2
...
...
...
Suraj-W95-96 series

4.4.2.2 Results
4.4.2.3 Discussion
...
4.4.3
...
Suraj-S96 series

4.4.3.2 Results
4.4.3.2 Discussion
4.4.4
...
...
...
...
...
2ndhaveli-S96 series...

4.4.4.2 Results
4.4.4.3 Discussion
...
4.4.5
...
March96 series
4.4.5.1 Temperature results
4.4.6
4.4.6.1
4.4.6.2
4.4.7
...
...
Courtyard surface temperatures - HAV-HT series.

Results
Discussion
...
...
Air movement - Vyas-Air series..
4.4.7.1 Results
4.4.7.2 Discussion
4.5
...
Conclusions
References
...
CHAPTER 5 THERMAL COMFORT IN JAISALMER

5.1
5.2
5.3
5.4
5.5
5.6
5.7
Introduction
...
Heat exchange mechanisms....
Behavioural adaptations......
The science of thermal comfort......
Climate chambers...
The adaptive model......
The Jaislamer comfort survey......
74
5.7.1
5.7.2
5.7.3
Method
...
Calculating comfort temperature.....
The wider context ...
79
80
81
5.8
Thermal comfort in Hotel Suraj.....
82
5.8.1
Method used for calculating comfort levels in Hotel Suraj.......
82
74
75
76
77
77
79
iv
5.8.2
5.8.3
Comfort level results...

...
Discussion
83
86
5.9
Conclusions
References
...
87
...
89
CHAPTER 6 PRESENTATION OF RESULTS

6.1
6.2
6.3
6.4
6.5
6.6
6.7
6.8
Introduction
...
Temperatures represented as colours......
Initial temperature spectrum.....
Final temperature spectrum...
Automating the process......
...
Key Features
Thermamator an overview.....
Instructions for use......
92
6.8.1
6.8.2
6.8.3
6.8.4
6.8.5
6.8.6
6.8.7
Preparation of input files...

Loading picture files.....
Loading temperature data files....
Linking temperatures to zones.....
Changing the temperature range.....
Animating the data...
Quitting the application...
96
97
98
98
98
99
99
6.9
6.10
6.11
Further developments...
Comfator an overview.....
Instructions for use.......
99
92
92
93
95
95
96
96
101
101
6.11.1 Preparation of input files....

6.11.2 Loading files and linking data columns to building zones.....
6.11.3 Animating the data...
101
102
102
6.12
6.13
...
103
...
104
...
105
Discussion
Closure
References
CHAPTER 7 INDOOR TEMPERATURE TRENDS

7.1
7.2
Introduction
...
Indoor / outdoor temperature....
107
7.2.1
7.2.2
7.2.3
Method
Results
Discussion
107
108
110
7.3
Indoor / outdoor range.....
113
7.3.1
7.3.2
Method
Results
113
114
...
...
...
...
...
107
7.3.3
Discussion
7.4
Conclusions
References
...
115
...
116
...
119
CHAPTER 8 MASS AS A THERMAL REGULATOR

8.1
8.2
8.3
Introduction
...
The problem
...
Aims of chapter 7......
121
8.3.1
Form of chapter 7
...
123
8.4
1 approach Stepped correlation.....
123
8.4.1
8.4.2
Method
Results
...
8.4.2.a
8.4.2.b
8.4.2.c
8.4.2.d
8.4.2.e
Summer95.xls
SurajW95-96.xls
Suraj-S96.xls
2ndhaveli-S96.xls
Suraj12months.xls
...
8.4.3
Conclusions
...
124
124
125
127
128
130
131
132
8.5
2 approach Running mean
133
8.5.a
8.5.b
8.5.c
8.5.1
8.5.2
Method for conversion of exponential weight (W) to time lag.

Results of conversion
...
Discussion
...
Method
...
Results
8.5.2.a
8.5.2.b
8.5.2.c
8.5.2.d
8.5.2.e
Summer95.xls
SurajW95-96.xls
Suraj-S96.xls
2ndhaveli-S96.xls
Suraj12months.xls
8.5.3
8.5.4
Conclusions
...
Validation of exponential smoothing method
8.5.4.1
Discussion
...
135
138
139
139
139
140
142
143
144
145
146
148
151
8.6
Comparison of methodologies
152
8.6.1
Discussion
...
155
8.7
8.8
Decrement factor ...

Summary of conclusions..
References
...
156
st
...
...
...
...
...
nd
...
...
...
...
...
122
123
159
162
CHAPTER 9 CONCLUSIONS
9.1
9.2
Introduction
...
Indoor environment.
164
164
vi
9.3
9.4
9.5
9.6
9.7
Thermal comfort ...

Outdoor temperature as an indoor predictor..
Mass as a thermal regulator..
Presentation of results..
In conclusion
...
165
Appendix A - Tabulated air movement data.

Appendix B - Visual Basic code - Thermamator
Appendix C - Visual Basic code - Comfator
Appendix D Excel macro for automation of stepped correlation method.
171
Bibliography
213
...
165
166
168
168
183
198
211
vii
Acknowledgements
In undertaking and completing this work I have depended on the help and support of many
people.
Particular thanks to Fergus Nicol, who has been closely involved in the work from its
conception, and has supported it through to completion, Professor Susan Roaf, for invaluable
advice on form and structure, and Professor Krishan, not only for the initial ideas that set the
course of this study, but also for his support and guidance during several visits to India.
Professors Steven Szokolay, Max Fordam and David Spooner all kindly gave advice on the
analysis of the effects of mass, as did my brother Dr. David Matthews. Reverend Mike
Humphreys sacrificed much of his time to advise on statistical methodologies.
Many students from both the TVB School of Architecture in New Delhi and the University
of East London helped with the manual collection of data. This could not have been achieved
without the backing of Ashok Lall at the TVB School and Nick Weaver at UEL. The Vyas
family generously gave hospitality and time, and graciously suffered my intrusions into their
homes.
Many friends and family have patiently borne the side effects of thesis writing, and their
continued support has been enormously sustaining. In particular Ric Irvine who endured
several months of the hot, dry season in Jaisalmer, and took many of the enclosed
photographs. Audrey Jones and Jenny Norwood, who persisted with attempts at contact
despite my seeming disinterest, and Dr. Debra Bekarian, who not only offered enthusiasm
and support, but also spent several sleepless nights printing these final drafts. My father, Eric
Matthews deserves a special mention for his encouragement and kindness throughout.
Finally I would like to thank David Woodhouse of the Quality Assurance Department of the
University of East London for his relentless troubleshooting.
viii
CHAPTER 1
INTRODUCTION
Chapter 1 outlines a range of the global environmental problems facing

building designers in the new millennium, with particular reference to
designing for hot climates. It gives a brief outline of the history of India and a
number of it's modern day problems that are, in part, a legacy of this past. The
advantages and disadvantages of traditional architectural forms are discussed,
with particular reference to the havelis of Jaislamer, and the aims and form of
this thesis are then defined.
1.1 Global concerns

There is now a consensus amongst meteorologists and climate modellers that increasing
atmospheric concentrations of carbon dioxide and other "greenhouse gases" will lead to
significant changes in the climate of the world (Houghton et al., 1990). A large fraction of
the increase in the concentration of greenhouse gases is attributed to human activity, and an
increasing component of this fraction result from the expanding demand for electricity for
air-conditioning. Of the present total world energy consumption, 45% is used in heating,
cooling and lighting of buildings, and a further 5% in building construction (Rigg et al.,
1995), making building energy consumption the largest single global energy use sector. It is
also that with the greatest, and easiest, potential for conservation. It is possible to achieve
reductions of 50 70% by using known and tested concepts of passive, climatically adapted
building design for new buildings (Rigg et al., 1995). The savings achieved by retrofitting
and adapting existing buildings are much smaller.
The world population has grown from some 3.8 to 5.2 billion in the past 20 years, and it is
estimated that more than 50% of this population will be, by the end of this century, living in
cities (Tombazis, 1995). Within the span time of a generation we built the equivalent of what
has been built on this earth from the beginning of history (Tombazis, 1995). Clients are
increasingly seeking 'low energy' buildings, not only for economic, but moral reasons. This
was emphasised by Humphreys in 1992:
Now that it seems at least probable that the carbon dioxide we are generating is
producing a warming of the earth, it seems to me that the reduction of the burning of
fossil fuels is no longer just economically desirable, but perhaps a matter of survival
for substantial numbers of the world's population. We should leave the world in a
reasonable condition for the next generation, if it is within our power to do so.
And succinctly by Tombazis in 1995:
From the ethical point of view we have come to accept that we cannot continue
living and inheriting from our forefathers, whilst borrowing and mortgaging the
future of our children.
1.2. India's history.

India's location and geography is such that it has always been an invader's paradise. At the
same time its natural isolation, large size and indigenous religions has allowed it to adapt to
and absorb other cultures, resulting in a history in which no single empire has dominated for
long periods.
India's most ancient civilisation dates from 5000 years ago, and is known as the Indus Valley
civilisation. Since the first invasion by the Aryans of the north in about 1500BC (Sterlin, H.
et. al., 1990), India has been invaded by a succession of powers including the Persians,
Greeks, Turks and Mongols. In 1610AD the British invaded India, remaining rulers for over
three hundred years, and eventually dominating the entire sub-continent.
The British viewed India as a profitable venture and were loath to allow the Indian
population any real power. As the twentieth century dawned there were increasing
movements towards self-rule. Coupled with the desire for independence were the developing
tensions between the Hindus and Muslims. In 1915 Mohandas Karamchand Gandhi first
called for unity between the two religions and eventually led the country to independence
through a totally non-violent mass movement. The British left in 1948, but independence
came at a great cost. On leaving the British created the separate states of Pakistan and
Bangladesh as Muslim territories. Violence erupted and within a few weeks half a million
Hindus and Muslims had died. In 1948 on the eve of independence a Hindu fanatic
assassinated Gandhi. Jawarhalal Nehru became India's first Prime Minister. Although India's
political climate remains divisive, the country has attained apparent stability in recent years.
With a climate ranging from tropical monsoon in the south to temperate in the north, six
religions, and twenty-four main languages with numerous other, for the most part mutually
exclusive dialects, India continues to be a diverse and complex society, reflecting its
antiquity and history.
1.3. India today.

Today India is the second most populous country in the world with an estimated annual
growth rate of around 1.7% in 1996. Currently estimated at 950 million, Indias population is
expected to reach 1.15 billion by 2010 (Energy Information Administration, 1999). This
rapidly growing population, along with increased economic development, has placed a strain
on Indias infrastructure, and ultimately on the countrys environment. According to the
World Health Organisation, New Delhi is one of the top ten most polluted cities in the
world. Indias electricity is generated overwhelmingly by coal (70 percent). Increased coal
consumption over the past four decades has led to a nine-fold increase in energy related
carbon emissions.
ENERGY
COUNTRY
Argentina
Brazil
China
Russia
India
Indonesia
Mexico
Poland
South
Africa
South
Korea
Turkey
United
States
World
Total
POPULATION
(MILLIONS)
CONSUMPTION
(QUADRILLION
BTU)
CARBON
EMISSIONS
(MILLION
METRIC TONS)
NUMBER OF
PEOPLE PER
OF
HOUSHOLDS
WITH
CAR
ELECTRICITY
34.8
155.8
1211.7
148.1
935.7
193.8
94.9
38.6
2.5
6.8
35.7
26.8
10.5
3.1
5.6
3.8
34.5
67.0
807.5
428.7
213.2
52.2
92.6
84.0
7
14
290
21**
272
109
11
11
90*
91*
80
N.A.
88
39
95
N.A.
41.2
5.5
135.1
11
44
44.9
6.3
102.0
100
61.6
2.5
43.0
20
N.A.
263.4
88.3
1415.1
100
5724.4
362.2
6063.1
12
N.A.
Tbl. 1.1. India in a world context: 1995. (Energy Information Administration, 1999)
NOTES:-
*Urban population only

**For all of the former Soviet Union.
Whilst India per capita energy use and carbon emissions are lower than the world average,
because of the countrys large population and heavy reliance on coal they result in a
substantial percentage of the world energy use and carbon emissions. India faces great
challenges in energy and environmental issues as it enters the 21st Century. A rapidly
growing population and increased urbanisation will continue to increase demands for
electricity generation. Whilst only a small fraction of Indias energy budget is currently used
for air-conditioning, there is, nonetheless, an increasing tendency for the use of mechanical
cooling in offices and homes in urban areas. Even more serious is the increasing use of
building designs that will be comfortable only when air-conditioned, even in developments
where, due to economic and infrastructure restraints, air conditioning will not be readily
available.
1.4 Traditional versus modern.

It is nowadays widely recognised by architects and urbanists that traditional, regional
architecture is generally climatically well adapted (Bowen, 1981; Heschong, 1990 ), and that
the courtyard house, in particular, is well suited to the hot desert climate (Fathy, 1986;
Karaman et al., 1981; Moore, 1983;Oliver, 1987; Talib, 1984). The principles of good
thermal design used in such houses are equally valid today, and it should be possible for
modern designers to incorporate these design principles in buildings suitable for modern
living, thus conserving energy and providing better thermal comfort. However, most of the
writings on regional architecture are based on scientific intuition, describing simple ideas,
but inadequate as accurate guides for design. Per-Olof et al. (1985) confirm this view when
discussing vernacular architecture in hot climates:
"Local building traditions are a result of long periods of development which have
often led to good solutions. Through experience a certain amount of knowledge,
which cannot be understood or dealt with mathematically has been acquired. Because
of our ignorance of the knowledge thus acquired there is a risk that we disregard
factors that are of great importance"
More recently research has begun to question the role of the courtyard in desert climates,
and the high re-radiation levels recorded in smaller courtyards in very hot climates are
coming under particular investigation (Danby, 1986; Hanna, 1991; Roaf, 1990). That the
haveli form is climatically well adapted to this part of India is not doubted (Prasad 1988). In
addition the haveli has the added advantage of being a high-density urban building form.
Despite this, recent research has also shown that there is a current cultural trend for higher
income families in northern India to prefer the lifestyle afforded by the suburban villa type
of house (Prasad, 1987). The task of the modern architect is quite different from the
indigenous house builder, and the kind of behavioural change that have taken place over the
last fifty years are irreversible. However the principles of thermal design used in such
buildings are particularly relevant today, in light of current environmental and financial
pressures.
1.5 Aims
It is with this background of rapidly expanding urban population and associated energy and
environmental crisis, coupled with the currently unfashionable image of Indian vernacular
architecture (with the indigenous population) that the author has studied the elaborate
courtyard houses of Jaisalmer with the following aims.
To accurately survey two havelis within the fort of Jaisalmer, in order to provide an exact
record of mass, volume and construction.
To monitor temperature and humidity in several locations within the havelis over a
twelve month period.
To measure air movement and surface temperatures in several locations within the haveli
over a short period of time.
To relate recorded indoor temperatures to predicted human thermal comfort sensation.
To investigate and develop new approaches to the presentation of thermal building data
in a style useful to architects and students of architecture.
To analyse the data collected with the intention of promoting an understanding of the
way indoor temperature varies in relation to outdoor temperature and the attributes of the
building in the hot, dry, desert climate of Jaisalmer.
1.6 Form of thesis

The following chapter, chapter 2, sets the context for the havelis studied. The geography,
climate and history of the living fortress of Jaisalmer are described and illustrated. The case
studies, two medium sized havelis and a smaller courtyard house, are then introduced in
chapter 3. Plans, sections and elevations of both havelis are shown along with limited
drawings of the courtyard house.
Chapter 4 details the method and equipment used for the collection of environmental data in
and around both havelis. The recorded data is illustrated and basic statistics calculated.
Chapter 5 gives an introduction to the science of thermal comfort, describing climate
chamber and field study methods of prediction. A comfort survey conducted in Jaisalmer is
analysed, and evaluated in the wider context of accepted adaptive model comfort prediction
equations. The results are applied to the assessment of the data detailed in the previous
chapter in terms of predicted occupant comfort level.
Chapter 6 discusses the problems associated with presenting large amounts of data,
particularly if collected in many locations over long periods of time. The concept of
temperatures represented as colours is explored and a temperature spectrum is proposed. A
stand-alone computer program is developed and discussed. These principles are extended to
develop a second stand-alone computer program designed to predict and present thermal
comfort sensation levels.
Chapter 7 examines the indoor temperature trends over a twelve-month period in the largest
of the havelis. The relationship between indoor and outdoor recorded temperature, and
indoor and outdoor recorded daily range is investigated and indoor temperature prediction
equations for each of the monitored zones are proposed. Chapter 8 introduces the science of
heat transmission under non-steady state conditions. It details two methodologies developed
by the author with the intention of determining time lag between indoor and outdoor
monitored temperature. Following the review of both methodologies the decrement factors
of the internal spaces are examined in detail.
Chapter 9 draws together the various strands spun in earlier chapters and makes suggestions
for areas where further research is needed.
References - Chapter 1
BOWEN, A. (1981). Historical responses to cooling needs in shelter and settlement. In:
Proceeds of the International Passive and Hybrid Cooling Conference. Pp. 545-587.
Miami Beach, Florida.
DANBY, M. (1986). The international environmental aspects of the traditional Arab house
and their relevance to modern housing. In: The Arab House (Hyland and AlShahi
ed.). Pp. 83-90. Newcastle upon Tyne.
ENERGY INFORMATION ADMINISTRATION. (1999). http://www.eia.doe.gov/emeu/cabs/india
/indiach1.htm. Current on 16/11/99.
FATHY, H. (1986). Natural energy and vernacular architecture, University of Chicago Press,
London.
HANNA, R. (1991) The relationship between thermal performance, thermal comfort and
overall user satisfaction with the house form. Doctoral dissertation, School of
Architecture, Oxford Brookes University.
HESCHONG, L. (1990). Thermal delight in architecture. The MIT Pres, Massachusetts.
HOUGHTON, J. T., JENKINS, G. J. AND EPHRAUMS, J. J. (1990). Climate change: the IPCC
scientific assessment. Cambridge University Press.
HUMPREYS, M. A. (1992). Thermal comfort in the context of energy conservation. In:
Energy efficient building (Roaf, S.C. and Hancock, M. eds.). Blackwell Scientific
Pubications, Oxford.
KARAMAN, A. AND EGLI, H. G. (1981). Historical responses to cooling needs in shelter and
settlement. In: Proceeds of the International Passive and Hybrid Cooling Conference.
Pp. 3-7. Miami Beach, Florida.
MOORE, F. (1983). Learning from the past: Passive cooling strategies in traditional
contemporary architecture. In: Islamic Architecture and Urbanism (Germen, A. ed.).
Pp. 233-238, University of Dammam, Saudi Arabia.
OLIVER, P. (1987). Dwellings - the house across the world. Phaidon Press, Oxford.
PER-OLOF, C., PER-OLOF, N., BURTON, L. M. M (1985). Passive indoor climate regulation
for buildings in hot climates. Stokholm.
PRASAD, S. (1987). In the 21st century will the haveli have been completely destroyed? In:
Design, Oct.-Dec. pp.11-20.
PRASAD, S. (1988). The Havelis of North India. Unpublished PhD thesis, Royal College of
Art, London.
RIGG, A. AND LAHAV, R. (1995). Concept and reality. In: Climatically responsive energy
efficient architecture A design handbook (Krishan, A. ed.), Volume 1, pp. 1-12 - 120. SPA, Delhi.
ROAF, S. (1990). The traditional technology trap: Stereotypes of Middle Eastern building
types and technologies. Issue 25, TRIALOG.
8
STERLIN, H. & VOLWAHSEN, A. (1990)Architecture of the World - India. Germany.

TALIB, K. (1984). Shelter in Saudi Arabia. London Academy Editions.
TOMBAZIS, A. N. (1995). Architecture and bioclimatic design: Less is beautiful. In:
Climatically responsive energy efficient architecture A design handbook (Krishan,
A. ed.), Volume 1, pp. 1-1 - 1-11. SPA, Delhi.
CHAPTER 2
JAISALMER
Chapter 2 gives a comprehensive description of the living fortress town of

Jaislamer and it's surrounding region. Location, topography, climate and
history are detailed, illustrating the context in which the havelis of the fort
were built.
2.1. Jaisalmer - Geography

Ghoda kije kath ka, pind kije pashad
Bakhatar kije loh ka, tab dekho Jaisan
{A house of wood, legs of stone, a frame of iron will get you to Jaisalmer alone}
Agarawala, 1979
Jaisalmer is situated towards the western portion of the Thar or Indian desert in Rajasthan, at
latitude 26 54' N and longitude 70 55' E. The town is about 590 miles away from Delhi,
180 miles from Jodphur, and 50 miles from the Pakistan border.
Ill.2.2. Photograph of moving

dunes in the Western extremes
of Jaisalmer state.
Ill.2.1 Map of India showing location of Jaisalmer (utexas, 1997)
The surrounding region is rocky and barren, relieved by sparse, scrubby weeds, but no trees.
In the western extremes of the state, around the Sam area, the land is more desolate
consisting mainly of sand stabilised into sand dunes, but continuously changing shape.
11
There are no large hill ranges in the area and the rainfall is scant, making water a scarce and
valuable resource. The available water is generally slightly salty and underground giving rise
to the need for very deep wells. There are no perennial rivers in the district, but water
accumulated during the rainy season gives rise to small seasonal rivers. These do not last
long, usually drying up within weeks, allowing wheat and other crops to be grown on the
catchment area. Small tanks, locally known as Khadeens , for the accumulation of rain water
are common in this area and are used to irrigate grain crops (Somani, 1990).
2.2 Jaisalmer - Climate

Within India it is possible to define six regions with distinct climates as shown in ill. 2.3
(Krishan, 1995). The six climates are normally designated as hot/dry; warm/humid;
moderate; cold/cloudy; cold/sunny and composite. Jaisalmer falls at the more extreme end of
the hot/dry climate, within which can be identified four different seasons, pre-monsoon,
monsoon, post-monsoon and winter.
Ill. 2.3. Climatic zones of India (Krishan, 1995)
Ill. 2.4. Criteria for classification of climatic

zones (Krishan, 1995).
Pre-monsoon, which is the hottest season, extends from April to June, with the daytime
temperature often exceeding 45C, cooling down to a minimum of 26C at night. This is the
most stressful season. Between 11am and 6pm the air temperature is higher than that of the
12
body and so during this period air movement actually decreases thermal comfort. Coupled to
this extreme heat stress is the tendency, during this period, for very high wind velocities
causing severe dust storms.
Monsoon arrives late in Jaisalmer, around mid-July. In the eastern and south-eastern regions
of Rajasthan the monsoon arrives earlier, towards the end of June, as the highland areas of
the Aravalli and Hardoti channel the Arabian Sea branch of the south-west monsoon in this
direction. Although the daytime temperatures are cooler during this period, with a peak value
of around 36C, the night time conditions remain hot with a minimum temperature of around
28C. The humidity for this brief period is high with relative humidity reaching 80% at
night, and remaining above 65% during the day, making conditions unpleasantly oppressive.
The third season is the post-monsoon, starting early to mid September, with relatively high
temperatures and clear skies. Daily temperatures range from about 20C at night to over
35C in the daytime. Humidity is low, and for the indigenous poulation of Jaisalmer, this is a
relatively comfortable season.
The final season is winter, which extends from December to March, January being the
coldest month. The daytime temperatures for this period peak in the mid twenties, dropping
to less than 10C at night. Humidity is low. At this time of year Jaisalmer is inundated with
tourists enjoying the cool temperatures and clear skies.
Solar radiation is intense (800-950 W/m2) throughout the year and radiant heat emanating
from the ground and surrounding objects during afternoons and evenings can cause
considerable thermal discomfort. Apart from the very short monsoon season relative
humidity is very low (25-40%), with precipitation generally less than 500 mm/year. The
prevailing wind direction is south-westerly, but for four months of the year, November
through to February, the wind blows from the north east. Table 2.1 gives a summary of
monthly air temperature (Ta in C), and rainfall statistics (in mm) for Jaisalmer (reproduced
from Krishan, 1995).
Month av. max Ta

Month av. min Ta
Rainfall
Jan
Feb
Mar
Apr
May
Jun
Jul
Aug
Sep
Oct
Nov
Dec
21.9
10.4
2.1
26.5
14.1
1.2
32
21
2.6
36.5
25.8
1.5
40.2
29.3
5.2
40.3
30.4
6.8
35.9
28.5
89.5
34.4
27.2
85.8
34.7
26.9
13.9
34.3
25.1
1.3
28.8
19.4
4.9
23.6
12.3
2.2
Tbl. 2.1 Monthly air temperature and rainfall statistics for Jaisalmer
13
These figures describe one of the hottest, driest and consequently one of the least populated
areas of India. Jeffrey Cook (1980) stresses the danger in considering the climates in arid
regions as predictable. As well as the diurnal and seasonal changes and ranges, the calmness
of the desert weather cycles is occasionally punctuated by exceptionally violent storms.
During one of the author's visits to Jaisalmer in August 1995 a young boy from the town was
drowned during the floods at the beginning of the monsoon. It is not as extreme a situation
as parts of the Sahara or Gobi, because of its relatively small size and the short distance to
the sea and Himalayas modify its characteristics. As Spate(1984) describes it 'the desert is
not total - but it is bad enough'. Thus designing for the extremes of climate in this arid region
is particularly demanding.
2.3 Jaisalmer - History
Ill. 2.5. The living fortress of Jaisalmer.
The early history of Jaisalmer is not well documented. It is believed (Mishra,1995) that in
the remote past the major portion of Western Rajasthan, including the district of Jaisalmer,
was submerged under water during Jurassic, Cretaceous and Eocene ages. The land began to
emerge during Pleistocene and the last glacial periods. It is also believed (Mishra, 1995 ) that
the western arid region of India was known to man at least thirty to forty thousand years ago,
but there is no direct evidence of settlement in the Jaisalmer area during the early Stone Age.
The Bhatis of Jaisalmer claim descent from Lord Krishna. After Krishna died the power of
his people declined and many families moved away from the Indus settling in Afghanistan.
However they suffered constant attack from the nomadic tribes of this area and could not
retain their kingdom. They migrated towards India, settling first in the district of Punjab, but
constant marauding by the Muslims of Afghanistan and Sind compelled the Bhatis to move
further south-west.
14
From the tenth to the twelfth centuries AD the Bhati chiefs fought constant battles with the
Muslim forces. At the beginning of the twelfth century Bhati Dusad Lanja was leader of the
Bhattis in their capital, Lodurva, in the Thar Desert. He had two sons named Jaissal and
Vijayraj. Dusad named Vijayraj as his successor and Jaissal went towards Tanot and Sind
and established his power there. Relations with his brother were not good. Jaissal was
jealous of his brother's position in the capital and he plotted the seizure of his power. Jaissal,
along with many other disgruntled chiefs, challenged his brother in 1158, but was defeated.
Vijayraj became a powerful leader fighting off many Muslim invasions. When Vijayraj died
his son Bhoj succeeded to the throne.
In 1178 AD Muhammad Ghori and his large Muslim army left Multan to raid the important
towns of north-western India. It is said (Tod, 1950) that Jaissal contacted Muhammad Ghori
and sought his assistance in gaining the throne of Lodurva.
Lodurva was one of the first towns to be invaded. It was completely destroyed, and all
important temples were dismantled. Bhoj was slain in its defence. The Sultan continued his
march and was eventually defeated by the combined Rajput armies near Abu.
Jaisal still wanted the throne of the Bhati State. He plundered the treasures seized by the
Turks and returned to Lodurva. The town was ruined and open to invasion and so Jaisal
sought a spot better adapted for defence. Ten miles away on the summit of a rocky ridge he
discovered a Brahmin hermit. The hermit related the history of the triple peaked hill they
stood on. He said that Hare Krishna had come to the spot to attend a great sacrifice. Krishna
foretold that a descendent of his would erect a town by the rivulet and a castle on Trikuta,
the triple-peaked mount. It was pointed out that the water was bad. Krishna hit the rock and
the water bubbled up as a sweet stream (Tod, 1950).
The traditional date of the foundation of Jaisalmer fort is 1156 AD (Tod, 1950), but this is
disputed by Somani (1990) and said to be 1178AD, after the Muslim invasion of Mohammed
Ghori. All dates in this text correspond with the later theory. Construction of the fort
together with a well and the Laxminath temple was started by Jaisal, but he did not complete
the task before dying. He had two sons, the eldest was expelled and so it was the youngest,
Salivahan who was named successor. Salivahan completed the fort some 18 years later along
with Laxminath temple and the Jessal Kuwa well (Somani, 1990). The fort stand 30 feet
above the rock, which is itself some 250 feet above the plain, and, at it's widest point,
measures 1500 feet across. These were not peaceful times. During this period the Bhatti
Rajputs major opponents were the powerful Rathor clans of Jodhpur and Bikaner and
endless battles were waged for the possession of small forts and water holes. Cattle stealing
15
was a major pass-time, along with falconry and the hunt. The Bhattis gained a reputation for
being both ferocious and brave in battle and often treacherous as allies (Parwal, 1997).
Jaisalmer was at this time one of the safer places in western India, and many affluent
families moved there to escape invaders.
Towards the end of the thirteenth century the nature of the conflicts changed. Alauddin
Khilji, the then leader of the Muslims, had the kingdom of Delhi, Gujarat and the eastern
regions of Rajasthan in his command and was intent on founding an ever-increasing empire.
In 1308AD he began an attack on Jaisalmer. It is said (Goetz, 1978A) that by this time
Jaisalmer was the only remaining independent power in Northern India. The people of
Jaisalmer prepared rigorously for the
invasion. Large quantities of corn and
food were collected in the fort. Stone
boulders were placed on the ramparts
(the likes of which can still be seen
today. (ill. 2.6.)). The countryside
around the city walls was laid waste in
order to prevent the enemy from
growing food and fodder (Tod, 1950).
The fighting continued for several years
Ill. 2.6 Boulders on the ramparts of Jaisalmer fort.
until in 1315AD the Muslim army
breached the ramparts of Jaisalmer fort. The Bhattis knew that defeat was inevitable and so
proclaimed the rite of Jauhar. Thousands of women and children surrendered their lives
either by sword or fire. All valuables were burnt with them. The men, clad in ceremonial
saffron and opium intoxicated, opened the gates and rushed out to meet a heroic death.
The Bhatis reclaimed Jaisalmer fort some time before 1321AD (Somani, 1990). Fighting
continued with the Muslim forces, with the people of Jaisalmer successfully capturing many
of their horses and livestock. The Sultan in Delhi decided to send a strong force against
Jaisalmer. It was a long battle, but the Bhatis were betrayed by one of their own. The leaders
and their kinsmen met heroic deaths at battle and the women and children again committed
Jauhar. The exact date is not known.
After the fall of Jaisalmer the territory was in the possession of the Sultan. Several religious
shrines were dismantled and the surrounding countryside was badly desolated. It was a
turbulent period and Jaisalmer faced great hardship. Many of the wealthy migrated towards
Sind and Gujarat.
16
The Muslim power began to wane after 1342AD and wealthy Jains began to return to
Jaisalmer to settle. Soon after the town became a centre for religious and cultural activities.
Jaisalmer's location, deep in the desert, made it a difficult area for the Muslim army, based in
Delhi, to control. The Sultan, conscious of the slow demise of his power, was anxious to
have Jaisalmer under a friendly power, and so it was agreed that the Rawals of Jaisalmer
would pay an annual tribute to the Delhi Sultans in order to preserve a circumscribed
independence.
There followed a relatively peaceful era, allowing trade
and commerce in the area to flourish. During the 16th and
early 17th century there was a second extensive stage of
construction in the fort, and it was during this period that
the first of the case study havelis, Hotel Suraj, and the
small courtyard house, Nohra, were built. Three gateways
into the fort, and seven bastions were built. A second wall,
15 feet high, was also constructed, providing additional
defence. Literary activities thrived and a good number of
businessmen from northern India passed through Jaisalmer
on their way to the western states. By the time the British
invaded India, Jaisalmer was a very prosperous state, and a
cluster of Jain and Brahminical temples had sprung up
Ill. 2.7. Suraj Pol (sun gate). The
inside the fort. The Maharawal at the time, Amarsingh, was
third, outermost gate, built during
a lover of the arts and sponsored many cultural activities in
the second stage of connstruction.
Jaisalmer. He desired more money for public works and the
building of palaces and gardens, and so, in the early 1660s, imposed a tax on the rich. When
the minister of state protested he was immediately put to death. The people of Jaisalmer paid
the money. An additional 92 bastions were added beside the earlier 7 making 99 in total.
As the opium trade with Afghanistan and China increased, so did the wealth amassed in
Jaisalmer. Goetz (1978B) states that during the reign of Amarsingh Jaisalmer became one of
the leading Rajput states, controlling the whole southern Thar Desert. But this prosperity was
not to last. In the mid eighteenth century relationships with adjoining rulers took a turn for
the worse and fighting broke out on the border territories of Jaisalmer. A fourth, outer,
gateway to the fort was constructed with a connecting wall adjoining the main fort wall. The
population began to move out of the fort and build homes on the slopes below. As a security
measure, towards the end of the 18th century, a strong boundary wall, 30 feet high, was built
encircling the downtown population.
17
In addition to the border troubles, Mulraj II, the then Maharawal, had to face the hostilities
of his powerful nobles. He selected an ambitious man named Sarup Singh as his minister and
together they managed to suppress the activities of the chiefs. Naturally the chiefs felt angry
at their treatment and began to plot the downfall of the Maharawal and his minister. Sarup
Singh reduced Raisingh's (Mulraj's son, and heir apparent) subsistence allowance and the
already hostile atmosphere became considerably worse.
In 1782AD Raisingh stabbed Sarup Singh to death. Mulraj rushed back to the palace, but he
was too late. Raisingh had assumed power, and put his father under surveillance. However
he only remained on the throne for three months before Mulraj was reinstated by his chiefs.
Sarup Singh's son, Salim Singh, then only twelve years old was nominated as minister by
Mulraj. Salim Singh would not forgive the murder of his father and as his power increased
he diverted more energy into crushing those involved with his fathers death. He murdered
dozens of chiefs as well as Raisingh and all his closest relatives. Mulraj stood by and
watched the murder of his relatives, comparatively powerless against the cunning and
ruthless mind of Salim Singh. For all his cruelty Salim Singh was a very clever administrator
and the ruling class flourished under his harsh taxation regime. Many imposing havelis were
built in the downtown area of Jaisalmer, most notably Salim Singh's own haveli and the five
Patwon ki havelis. Jaisalmer had become a centre for merchants indulging in the caravantrade with the neighbouring states and the prosperity was reflected in the fronts of the houses
built at the time. The second case study haveli was built during Salim Singh's administration.
Ill.2.8. Faade of Patwon ki haveli* Ill. 2.9. Salim Singh haveli*
Ill. 2.10. Downtown Jaisalmer*
* reproduced from Kamiya (1999)
In 1818AD almost all Rajput chiefs entered into friendly relations with the British and
Jaisalmer followed suit (Somani, 1990). Salim Singh's power was still growing and Mulraj's
18
successor, Gajsingh, was anxious to change this, but by this time this was a near impossible
task. Attempts were made on Salim Singh's life, but in the end it was one of his own wives
that successfully poisoned him in 1823. During his life Salim Singh was so powerful that no
one dared to speak out against him. He collected immense wealth through bribery and
murder, but his heavy taxes drove many families to emigrate from Jasialmer. Agarawala
(1979) states that Salim Singh was particularly cruel to the Paliwal Brahmins, famous as
enterprising cultivators and land owners, eventually driving them southward out of the Thar
Desert shortly before 1820. Their empty villages can still be seen, partially buried in the
sand. The State never fully recovered from the loss of talent and trade resulting from Salim
Singh's high-handedness.
The population of the city continued to fall gradually throughout the late 19th and early 20th
centuries, from 115,00 in 1891 to just 67,000 in 1931 (Ward, 1989). In 1938 electricity was
introduced to Jaisalmer and a year later stone roads were constructed from Jaisalmer to
Pokoran and Bikaner, but still the population declined. By the time Jaisalmer State was
absorbed into Rajasthan in 1948 the city of Jaisalmer was almost deserted. The rise of
shipping trade and the port of Bombay, coupled with partition and the cutting of trade routes
through Pakistan sealed the city's fate. However the 1965 and 1971 Indo-Pakistan war
revealed Jaisalmer's strategic importance. In addition the Indira Gandhi Canal, built in the
70's and passing about forty miles to the north-west of Jaisalmer transformed the desert
economy. By this time many of the buildings in Jaisalmer, all in need of constant
maintenance, had been empty for some time. After an unscheduled visit to Jaisalmer in
January 1975, Indira Gandhi expressed her displeasure over the neglect of the fort town and
in a letter to the Chief Minister of Rajasthan she suggested that measures should be taken
immediately for the conservation of the town (Gosh, 1975).
2.4. Jaislamer today

Although in the late seventies Jaisalmer was considered to be a ghost town (Goetz, 1978A),
today it is accessible by plane, train and regular bus services and has become a very popular
tourist destination, offering camel safaris into the desert. This has, however, proven to be a
double-edged sword. Previously the city's water was collected manually from a reservoir
lake. The introduction of pumped water combined with the demands of the tourist has
resulted in larger amounts of water being pumped into the city than the open drains can
handle. The excess water is seeping into the foundations of Jaisalmer rendering many
structures unstable. Carpenter (1994) has most recently highlighted this dilemma through the
creation of the international campaign called 'Jaisalmer in Jeopardy' (Carpenter, 1995). The
funds from this charity go directly towards specific restoration projects in the city, but the
19
spectacularly carved havelis are being demolished and 'repaired' into banality at an alarming
rate. The commercial pressures and high maintenance costs are proving too much for even
the most committed conservationist. In the course of this study five new hotels opened
within the fort alone, and one of the properties surveyed was been altered beyond
recognition. Gradually, through the work of Sue Carpenter and Indian bodies such as
INTACH, the Indian National Trust for Art and Cultural Heritage, the historical significance
of Jaisalmer is slowly being realised, and the city is once again blooming.
Chapter 3 provides a record of two existing havelis and a small, adjacent courtyard house
within the fort of Jaisalmer. It documents volume, mass and construction of the havelis, and
discusses some of the patterns of life for which these mansions were designed.
20
AGARAWALA, R. A. (1979). History, Art and Architecture of Jaisalmer. Delhi.
CARPENTER, S. (1994). Collapse of the golden city. In: New Scientist. Dec 17th.
CARPENTER, S. (1995). Return to the golden city. In: New Scientist. Dec. 23rd. pp.36-37.
COOK, J. (1980). Landscaping for microclimatic advantage in arid-zone housing. In:
Housing in Arid Lands (G. Golany, ed.), Chap. 19. Architectural Press.
GHOSH, B. (1975). A conservation plan for Jaisalmer. In Urban and Rural Planning Thought.
April, SPA, Delhi.
GOETZ, H. (1978A). The Architecture of Jaisalmer. In: Rajput Art and Architecture (J. Jain,
ed.), pp. 161-162. Wiesbaden.
GOETZ, H. (1978B). Jaisalmer: Treasure of the desert. In: Rajput Art and Architecture (J.
Jain, ed.), pp. 161-162. Wiesbaden.
KAMIYA, T. (1999). http://www.asahi-net.or.jp/~wu3t-kmy/engl.htm. Current on 09/09/99.
KRISHAN, A. (1995). Climatically responsive energy efficient architecture - A
design
handbook, Volume 2. SPA, Delhi.
MISHRA, P. (1995). Researches in Indian Archaeology, Art, Arch., Culture and Religion.
PARWAL, N. K. (1997). http://www.rajasthanweb.com/history/jaisal/index.htm Current on
10/10/97.
SOMANI, R. V. (1990). History of Jaisalmer. Panchsheel Prakashan, Jaipur
SPATE, O. H. K. (1984). India and Pakistan: A general and regional geography.
TOD, LIEUT. -COL. J. (1950). Annals and Antiquities of Rajasthan, 2 Vols., London.
WARD, P. (1989). Rajasthan, Agra, Delhi: A travel guide. The Oleander Press, England.
UTEXAS (1997). http://www.lib.utexas.edu/libs/pcl/map_collection/middle_east_and_asia/
india_poli.jpg. Current on 24/10/97.
21
CHAPTER 3
THE HAVELIS
This chapter gives a detailed, illustrated account of two medium sized havelis,
and a third, smaller, courtyard house, located within the fort of Jaisalmer.
Plans, sections and elevations of both havelis are shown along with limited
drawings of the courtyard house.
3.1. Introduction
The term 'haveli' is of uncertain origin1, with various dictionary definitions, but amongst the
alleys of north Indian towns its use suggests a distinguishable type of inward looking, urban
courtyard house, and signifies a way of life, based on traditions, and patterns of etiquette and
behaviour, that was more than just a form of architecture (Tillotson, 1994).
Havelis are found in a belt that extends west to east over the modern Indian states of
Rajasthan, Uttar Pradesh, Bihar and into Bengal, and to the north over Haryana and Punjab
and to the south over Madhya Pradesh and Gujerat (Prasad, 1988) (see illustration 3.1.).
.
Ill. 3.1 India, showing the haveli region (original utexas, 1997, edited by author).
This area is characterised by a predominantly hot and dry climate, although there are many
other areas with the same climate. The area is also broadly influenced by Islam.
Possibly Persian - "a surrounded or enclosed place" (Wacziarg et al., 1982)

23
The havelis are the building blocks of old Indian towns. The largest were town houses of
noblemen and feudal landlords, more like small palaces than houses. The smallest were often
homes and offices of respectable merchants, with courtyards just big enough to provide a
lobby, light-well and ventilation shaft. In Mughal times a haveli normally accommodated
several families who lived together as an economic, civic and social unit, sharing many
common amenities, but today there are fewer inhabitants. The original function of a haveli,
apart from providing a residence, was to wall in the domestic life of a family (Wacziarg et
al., 1982). Secluded from the outside world, the haveli sets its own pace of life.
3.2. Building Surveys

It was decided that detailed measurements of two medium sized havelis would be attempted
in preference to limited measurements in many different buildings. Jaisalmer is one of the
more traditional areas of India and privacy for women in the home is very important.
Prolonged and thorough intrusions into family homes by outsiders are not readily accepted
and so, consequently, none of the buildings in this study were occupied as family homes.
Ill.3.1.Jaisalmer fort showing location of havelis, and courtyard house.
24
Hotel Suraj was surveyed during the first visit to Jaisalmer in June/July 1995. It was
measured in sufficient detail for plans, sections and elevations to be accurately drawn at 1:50
scale. During a second and shorter visit in April 96 a smaller courtyard house, Nohra,
opposite and belonging to the same family as Hotel Suraj, was surveyed2. During a third and
final visit in March/April 97 a second haveli was surveyed in similar detail to Hotel Suraj.
Plans, sections and elevations of both havelis are shown along with limited drawings of the
courtyard house in section 3.4, towards the end of this chapter.
3.2.1. Hotel Suraj

This medium sized haveli is located on the Southeast face of the fort wall, not far from the
Jain temple with the front, street elevation facing Northwest.
Ill. 3.2 Street view of Hotel Suraj.
The haveli is three stories high. In addition there is

a basement which covers about a third of the floor
area towards the back of the building. The present
owners, Gokul and Namneed Vyas, state that the
haveli was built some five hundred years ago over
a period of about twenty-seven years. As is
common in India the haveli has been owned by the
same family for the entirety of this period, passed
from eldest son to eldest son. Although there is
plenty of evidence internally to support the present
owners claims, it is very unlikely that the main
street faade is the original, as many of it's features
(in particular the cusped arches and Bangaldar
roof, see footnote 3) are characteristic of Mogul
fashion not found in Jaisalmer before the early
sixteenth century (Goetz,1978a).
The haveli is constructed of immensely thick blocks of the local honey-yellow sandstone,
with thinner limestone panels in the more intricately carved areas. Both masonry and carved
areas are composed of prefabricated panels assembled with stone keys or iron cramps no
mortar was used. This method of construction makes the changing of building elements a
relatively easy task, the replacing of one prefabricated unit with another, and it is common
2
Also during the same visit, students from the TVB School of Architecture in Delhi surveyed fifty meters of
streetscape running outside hotel Suraj in order to facilitate the investigation of air temperature and surface
temperature around the buildings. This has not been included in this study.
25
practice today to see whole balconies salvaged from old buildings being incorporated into
the new. This would support the notion that Hotel Suraj's present faade is not the original.
In all but one room the stone is exposed with no render, making the grain and structure
clearly visible. The masonry panels are in excess of half a meter thick whilst the limestone
areas are much thinner- in the region of six centimetres. The ceilings are of timber
construction, a material not common in the area, the use of which reinforces the belief that
the building is considerably older than its elevation would suggest. The beams span typically
three to four metres, supported on masonry columns. Spans are increased with the use of
masonry brackets, which swing out from the columns in an S-curve.
Ill. 3.4. Detail of brackets, first floor front room.
Ill.3.5. Detail of ceiling, first floor, back room.
The floors of the upper stories are lined with a cow dung and mud plaster, whilst the ground
floor is of sandstone. (The roofs were also mud plaster until June, 95 when they were
replaced with six inches of concrete, requiring little maintenance, but considerably changing
the thermal performance!)
The haveli stands on a plinth just over a metre above the street level. This is common to all
buildings in Jaisalmer helping to keep the dust and sand out. In addition, the plinth provides
a semi-public area that, although separate from the street, still allows interaction with it. The
entrance, an ornate wooden door hung in a stone frame, is recessed behind a series of five
cusped arches carved with biomorphic designs (again suggesting Mogul influence). Inside
the progression from public to private continues gradually from the street inwards with the
26
most private rooms at the back of the haveli, a pattern recurring on all floors. Immediately on
the left as you enter the haveli are the male and female wash rooms with the family temple
opposite to the right. These are all relatively recent additions.
Light floods into the haveli through an almost
central courtyard measuring approximately five
meters by three with the longest axis running northeast/south-west. The courtyard is an extremely
important element of any haveli, and considerable
amounts of effort and money are spent to make the
internal facades as elaborate as the outside ones
(Jain, 1998). The courtyard of Hotel Suraj is no
exception; it is the heart of the house, connecting
other spaces, both horizontally and vertically.
Scattered around the interior are small, carved niches
designed to house oil lamps. In addition small metal
hoops are embedded into the walls, columns and
ceilings at regular intervals. The owner of the haveli
Ill. 3.6. Light floods in to the central
told how, before electricity, tapestries and heavy
court-yard on the ground floor.
cloths were hung from these during the winter
months to remedy cold draughts. In the summer grass blinds were hung in their place and
regularly dampened so as to cool any incoming breeze. This is substantiated by Mrs Ali's
description of zenanas (female quarters) in the havelis of Lucknow (1973):
As they have neither doors nor windows to the halls, warmth or privacy is
secured by means of thick wadded curtains, made to fit each opening between
the pillarsThe wadded curtains are called purdahs; these are sometimes
made of woollen cloth, but more generally of coarse calico.Besides the
purdahs, the openings between the pillars have blinds neatly made of bamboo
strips, wove together with coloured cords.These blinds constitute a real
comfort to every one in India, as they admit air when let down, and at the
same time shut out flies and other annoying insects; besides which the
extreme glare is shaded by them.
And further in Sarah Tillotson's work on traditional life in the havelis of northern India
(1988):
Above ground, rooms and verandas were kept cool by tattis (screens) made
up of fragrant roots of khas grass. Servants kept these constantly wet, and the
27
evaporation of the water cooled and freshened the breeze as it passed through
the tattis into the room which they shaded.
A stone staircase leading up to the first and second
floors and the kitchen open onto the north-west
side of the courtyard. The thin wall of the kitchen
is perforated with star shaped holes to enhance
ventilation. Two semi-basement rooms are located
at the far end of the haveli, one considerably larger
than the other. The largest has a single window,
about half a meter square, opening onto the
ramparts. Other than the door, this is the only
opening on the ground floor. The basement is
accessed through a trap door towards the rear of
the haveli.
On the first floor at the front of the haveli is the

mahal, the grandest room in the haveli,
Ill. 3.6. Stone staircase of Hotel Suraj
traditionally used as a reception room, a place for
the men of the house to conduct business. The walls of this room are plastered
(approximately
15mm
thick),
and
beautifully painted. Tillotson (1994)
describes one such plaster as ". a
Composition, made of freestone beaten
small, lime, gum and sugar, which makes a
dazzling white and is smooth as glass.".
Even though the plasterwork is severely
damaged, the owner has resisted the
temptation to renovate for fear of loosing
the original character. It is customary for
the mahal of a Rajasthani haveli to be
Ill.3.7. The mahal, the first floor front room of
Hotel Suraj
located, as in this case, above the entrance.
Between the mahal and the courtyard is a small room, now converted into an on-suite
bathroom, but originally intended as a storeroom. In fact all the large rooms in the haveli
have an associated smaller storeroom. Traditionally only a few rooms- bathroom, kitchen,
stores - in a haveli were designed for a specific function. Larger rooms were used instead as
season, need and mood dictated (Prasad, 1998). A room used to conduct business in the
28
morning could well be used for entertaining dinner guests in the evening. Furniture and
accessories were kept in the storeroom and brought out as and when required.
At the front of the mahal, and overhanging the street is a semi-enclosed balcony or jharokha,
providing a sitting platform ideally placed for views and breezes. In all havelis the jharokha
is made with extra care; it ia a showpiece, an expression of wealth. These jharokhas are ideal
niches for relaxing of an evening, allowing discrete participation with the outside world.
Agarawala (1979) quotes the verse:
Galiyo sir gaddal bichhe,
Amal bate apraman.
Mahal changi drav mansa,
Samajhtiya Jaisani.
Translated as:
In the projected balconies bedrolls are spread,
One can share opium without discrimination.
Prosperous people live in beautiful houses,
Such is the city of Jaisalmer with beauty and taste.
Ill.3.8. First floor middle back room, drawing its

light and air from the main courtyard.
At the back of the mahal is what now seems

to be little more than a cupboard, but was
originally a passageway linking the public
front mahal with the more private back
rooms. This passageway could be used by
the women of the house, without the risk of
being seen by male guests. At the end of the
passageway is a large room opening onto the
courtyard and leading to a smaller room
opening onto the fort battlements. The
windows of the smaller room are glazed - the
only glass found in the entire building. To
the east of these rooms is a small wing
overhanging the battlements. This is now an
on-suite bathroom, but would originally have
contained a latrine - a hole designed so waste
fell outside into a gandi gali (dirty alley)
where sweepers would clean it away. Today
the battlements are still used as a gandi gali,
but no one clears it away.
Ill.3.9. Second floor roof terrace.

29
The staircase to the second floor brings you

out onto a roof terrace, secluded form the
main street by a large, elegant room referred
to by the owners as the winter room. This
room, more than any other, is influenced by
Mogul fashion. Both the street and roof
terrace elevations are decorated with
bangaldar roofs3, a feature described by
Goetz (1978B) as post Mulraj II, dating this
Ill.3.10. Street elevation of winter room,
particular section somewhere around the mid
showing jharokha with bangaldar roof
eighteenth century. It was not uncommon for
havelis to grow up over many centuries. In her semi-autobiographical novel Rama Mehta
(1977) describes her families haveli in Udaipur as 'like a banyan tree; once it had taken root
it spread'. The winter room again has the characteristic attached smaller room, now
converted into an en-suite shower-room.
Towards the back of the haveli are, to the
author's mind, two of the most pleasant
spaces in the building. The first, orientated
towards the courtyard, is an enclosed private
terrace, a roofless room given the Rajasthani
name chandni, after the moon (chand). The
second is located over the battlements, open
to the elements on two sides, but covered by
a roof. At the far end of the second space is a
Ill.3.11. View form the second floor back room,
simple balcony, offering panoramic views of
through to the chandni.
the Thar Desert towards Gadi Sagar Tank.
There are two smaller rooms adjacent to these outdoor spaces which were used as private
family rooms. The second of these again overhangs the battlements, slightly larger than the
floor below to facilitate the use of a latrine as described earlier. This has once more been
converted into a shower room.
3
TILLOTSON (1987) Curved roof or eave, derived from a Bengali prototype, particularly in a modified form.
The original form, as it is found in Bengal, consists of a shallow dome covering a rectangular area and curved
more deeply along its long axis than across its short one, resulting in curved eaves along the long sides, each
bent like a bow. It is this form which is imitated faithfully on some pavilions of Shah Jahan's palace in the Agra
fort, and on Jess Mandir in the Palace of Amber. In more common Rajput usage, the form is smaller, the curved
eave is made by an exaggerated bent chajja, (sloping projection from a faade, in he place of an eve or stringcourse, more exaggeratedly protruding, and more deeply curved, even to a semi-circle) and a rather shapeless
blocky dome sits uncomfortably on top. This is the form found for example in the Gul Mahal on Jag Mandir, in
udaipur. However in the desert palaces (and havelis) the form is used less to crown pavilions than to cover
shallow projecting balconies; the domes consequently became redundant, and they are either merely
represented in relief sculpture or dispensed with altogether, leaving only the bent chajja.
30
Continuing up the staircase brings us to the roof terrace. The party walls to the north-east
and south-west are extended just over a meter above the floor level so as to make these areas
private places to sleep at night. Above and to the south-west of the chandni is a brick
constructed water tank, a recent addition. A further stair, leading to a throne-like platform,
elevated to catch every breeze, surmounts the roof of the staircase.
3.2.2. The second haveli

This is again a medium sized haveli, with a footprint about three quarters the size of that of
hotel Suraj. It is located on the south-east face of the fort wall, with the front street elevation
facing south-west.
Ill. 3.12 Street faade of second haveli photo montage
One half (which used to be the public or male

half) belongs to the Hotel Suraj family, the other
belongs to their cousin. Over time ownership of
havelis inevitably becomes fragmented in this
way, making maintenance and upkeep of the
building difficult to co-ordinate. On the whole this
building, although presently unoccupied, is in
remarkably good condition. The owners date the
haveli as being two hundred years old, and the
building is in keeping with the fashion in
Jaisalmer at this time. It was commissioned by the
then owners of Hotel Suraj with wealth amassed
during travels.
Although there are palpable similarities with Hotel Suraj, the external influences experienced
by the owners during their travels are manifest in the planning of this haveli. In contrast to
the airy impression created by Suraj's open planning, this building is compartmentalised,
probably influenced by the British need for privacy within the home. The boundary between
male/public areas and female/private areas also seems more pronounced, but this may just be
a product of the existing division of ownership.
The haveli is two stories high, with a small basement covering about an eighth of the floor
area. Unusually for a residence of this size it has three courtyards. It is situated on a corner
plot, with its long axis running parallel to the fort wall. The construction is similar to that of
31
Hotel Suraj except that stone beams have been used in place of wood. The carving, although
using similar motifs to Hotel Suraj, is not as intricate or as deeply set.
There are two entrances into the building. The first, on the left, leads into the largest of the
three courtyards. The staircase is situated in the northern corner of this space, and next to
this, towards the back of the courtyard, is a trap-door leading down to the basement. This
was the public side of the haveli, the male domain or mardana. On the first floor the
staircase arrives at a narrow balcony/corridor overhanging the main courtyard. This leads to
the mahal, which extends across the entire front elevation. Off to the right is a small room,
still used as a store and so full that internal measurements were impossible. The mahal has
two Jharokas overlooking the street, both capped with bangaldar roofs.
The second entrance, on the right of the front faade leads into the female quarters or
zenana. A small second courtyard provides some light to this long, thin plan, but on the
whole this section is rather dark. To the left of the courtyard is the kitchen, with a second
staircase opposite. The chamfered room off to the right was used as a family temple and so
access to survey was not appropriate. Off the staircase on the first floor is a long thin room
overhanging the battlements formerly used as a latrine. A larger room is adjacent with a
smaller storeroom leading off towards the rear.
Towards the centre of the haveli on the first
floor is the third courtyard. Another
chamfered room sits above the temple.
Unlike Hotel Suraj, no recent alteration
have been made to this building and stone
washing troughs are still embedded in the
mud floors of both this room and the
mahal. The Indian method of bathing (in
which a small pot is used to scoop water
Ill. 3.13. Mahal of 2nd haveli, showing washing
from a container and pour it over the body)
trough in bottom left hand corner.
meant that no special fittings other than
some form of drainage were required. In this case drainage is out of a small hole in the wall
onto the battlements below.
Both staircases continue up a further level onto the roof, again made private by the extension
of the parapet to a height of just over a meter. Adjacent to the staircase servicing the zenana
is a second latrine overhanging the battlements.
32
3.2.3. Nohra
Nohra (big gate) was a beautiful little courtyard house directly opposite Hotel Suraj and
owned by the same family. It is not considered to be a haveli, but has been included by the
author, because it no longer exists in its
original state. No environmental data has been
collected for this building and the survey was
carried out on the request of the owners to
facilitate the conversion into a low budget
hotel. The accommodation, as it was, consisted
of two separate wings, joined on the first floor
by a roof terrace. The plan is triangular in
shape the longest side of which adjoins the
Jain temple. The wing to the west and facing
Hotel Suraj was the Mardana. This has not
been altered during the conversion, and so has
not been surveyed. This portion of the building
is illustrated in Agarawala (1979). The wing to
the north-west was the zenana and now no
longer exists as documented.
Ill. 3.13 . Street faade of the Mardana of
Nohra (Agarawala, 1979).
The Main entrance is through a huge gateway

under the mardana. Carts could be driven
straight into the courtyard through this entrance and two such carts are still in the owner's
possession and kept in the basement.
3.3.Discussion.
Today the havelis of Jaisalmer are well-documented (Agarawala, 1979; Tillotson, 1987) and
noted for their outstanding beauty and design. Nath et al. (1997) claim "Civil domestic
architecture in stone cannot be seen in more intrinsic beauty than in Jaisalmer", whilst Oliver
(1997) states "the richly embellished havelis of the merchants mark a pinnacle of vernacular
achievement". However, it is not sufficient to have a qualitative understanding of such
buildings. To develop confidence in the natural heating and cooling methods used in these
havelis we need to know the kind of thermal environment that prevails within. Chapter 4
details the method and equipment used for the collection of environmental data in and
around both Hotel Suraj and the second haveli. As Hotel Suraj was the most accessible, and
33
also the most elaborate, of the case studies, the majority of the field work for this thesis was
centred in this haveli. Despite the intricately carved facades, Hotel Suraj is an exceptionally
high mass building, particularly on the lower floors, and this is likely to significantly effect
the climate indoors. Previous research (Gupta, 1984) carried out in similar buildings in the
downtown area of Jaisalmer suggested the high mass construction shifted the indoor peak
temperature by twenty-four hours. This resulted in a smaller indoor than outdoor temperature
range, but still gave the peak temperature indoors in the day when higher temperatures can
be better tolerated. The possibility of achieving a 24 hour shift in indoor temperature will be
discussed in chapters 8 and 9.
3.4. Plans, sections and elevations.

The following pages show the survey drawings for Hotel Suraj, the second haveli and Nohra
courtyard house.
34
HOTEL SURAJ
35
HOTEL SURAJ
36
HOTEL SURAJ
37
2ND HAVELI
38
2ND HAVELI
39
NOHRA
40
References - chapter 3
ALI, MRS M. H. (1973). Observations of the Mussulmauns of India, Karachi.
GOETZ, H. (1978A). The architecture of Jaisalmer. In: Rajput Art and Architecture (J. Jain,
GUPTA, V. K.(1984). A study of the natural cooling systems of Jaisalmer. Unpublished PhD
thesis, Indian Institute of Technology, Delhi.
JAIN, K. (1998). Spatial organisation and aesthetic expression in the traditional architecture
of Rajasthan. In:Paradigms of Indian architecture. Space and time in representation
and design (G.H.R. Tillotson, ed.), pp. 159-175. Curzon Press, Surrey.
MEHTA, R. (1977). Inside the Haveli. Delhi.
NATH, A. & WACZIARG, F. ( 1997). Arts and Crafts of Rajasthan. Thames and
Hudson.
OLIVER, P. (1997). Encyclopaedia of vernacular architecture of the world. Volume 2:
Cultures and habitats. Cambridge University Press.
PRASAD, S. (1988). The havelis of north India. Unpublished PhD thesis, Royal College of
Art, London.
PRASAD, S. (1998). A tale of two cities: house and town in India today. In:Paradigms of
Indian architecture. Space and time in representation and design (G.H.R. Tillotson,
ed.), pp. 159-175. Curzon Press, Surrey.
TILLOTSON, G. H. R. (1987). The Rajput Palaces: The Development of an Architectural
Style 1450 - 1750. New Haven.
TILLOTSON, S. (1994). Indian Mansions: A Social History of the Haveli. The Oleander
Press, Cambridge.
UTEXAS, (1997). http://www.lib.utexas.edu/libs/pcl/map_collection/middle_east_and_asia/
india_geo.jpg. Current on 24/10/97.
WACZIARG, F. AND AMAN, N. (1982). Rajasthan: The painted walls of Shekhavati. Delhi.
41
42
CHAPTER 4.
ENVIRONMENTAL DATA COLLECTION
This chapter details the method and equipment used for the collection of
environmental data in and around both Hotel Suraj and the second haveli. The
recorded data is illustrated and basic descriptive statistics calculated. This
data is further analysed in chapters 7 and 8.
4.1. Introduction
This chapter describes the collection of data over a 16 month period, in and around both of
the havelis detailed in the previous chapter. This data was collected in order to accurately
assess the thermal performance of the havelis, in particular hotel Suraj. In subsequent
chapters these results are used to evaluate the havelis in terms of occupant thermal comfort.
In addition the author hopes to build on existing data collection methodologies (Nicol, 1993)
to establish which environmental parameters, over what period of time, are fundamental to
the understanding of the thermal comfort performance of a building in this hot, dry desert
climate.
4.2. Field study periods

Three field trips, of varying lengths, were conducted over an 18-month period. In addition,
temperature and humidity loggers were left in-situ between visits.
Summer, 95. The author stayed in Hotel Suraj for a 6-week period from the beginning of
June to the end of July. This is the hottest time of year in this area. On occasions outdoor
recorded temperature exceeded 45C in the late afternoon, and rarely dropped below
30C during the night. Few tourists visit Jaisalmer during this period of extreme climate,
and so the author had access to all areas of Hotel Suraj allowing a thorough survey (see
previous chapter), and comprehensive logging of temperature and humidity throughout
the building.
March, 96. In March 1996 the author returned to Jaisalmer for 4 weeks in conjunction
with a shorter visit by students from the University of East London and TVB (Delhi)
schools of architecture. During this period sole access was granted to the haveli and a
series of manual measurements were recorded throughout the building. In addition a
small thermal comfort survey was conducted in and around the streets of Jaisalmer.
Access to the second haveli was secured, and a detailed survey and temperature and
humidity logging begun.
March, 97. A return visit was made to Jaislamer in order to retrieve the data loggers
from both buildings and complete the survey of the second haveli.
43
4.3. Equipment
4.3.1. Temperature
The temperature of a zone is defined in terms of both air temperature and the temperature of
the surfaces within the space (the radiant temperature). Radiant temperature is a more
complex entity than air temperature. It varies not only with the position in the room, but also
with the direction in which it is measured. This is usually addressed by calculating the mean
radiant temperature (MRT) of the environment. This is defined as the uniform surface
temperature of an imaginary black enclosure with which a person (also assumed as a black
body) exchanges the same heat by radiation as in the actual environment (Balaras, 1994).
However, the measuring of all surface temperatures in an environment and calculation of
MRT is very time consuming. Nicol (1993) recommends the use of globe temperature,
measured using a globe thermometer, for the estimation of the effect of radiant temperature.
The globe thermometer consists of a hollow sphere, generally painted black, measuring
40mm in diameter (roughly that of a table tennis ball), with a thermocouple or thermometer
located at the centre. The temperature assumed by the globe at equilibrium is the result of a
balance between heat gained or lost by radiation (from the surfaces) and by convection (from
the air).
In this study indoor and outdoor temperature was recorded using TinyTalk miniature, singlechannel dataloggers. These units are housed in 35mm opaque, flip-top film canisters, making
them both small and lightweight. They are accurate to 0.2C and can store up to 1800
readings. They are sold as air temperature loggers, but since their arrival on the market in the
early 90's there has been some dispute as to whether the temperature they record is actually
closer to radiant temperature (due to the film canister housing). Nigel Oseland of the BRE
conducted a simple experiment on a windowsill for two days. The results were as follows:
(private communication).
There can be up to 1.5C difference between air temperature and the temperature
recorded by a standard TinyTalk logger, housed in a 35mm film pod, during hot periods.
The standard TinyTalk logger appears to read temperature somewhere between air and
globe most, but not all of the time. In general, when the recorded temperature is not
between air and globe, the difference is within the tolerance of the logger (0.2C), but
occasionally the difference can be larger (up to 0.5C).
He concluded that the TinyTalk loggers are more akin to globe thermometers than air
temperature thermometers.
44
4.3.2. Humidity
Relative humidity was recorded using TinyTalk (RH) miniature single channel dataloggers.
The specification is as above, with a range of 0 - 90%, and an accuracy of 4% at 20C. As
relative humidity is dependent on air temperature, all recorded humidity levels have been
converted to water vapour pressure values (Pa) as recommended by Nicol (1993) using the
formula:
Pa = RH exp(18.956 - 4030.18 / (Ta + 235))
(milibars)
Where:
Pa = water vapour pressure
RH = recorded relative humidity
Ta = recorded air temperature1
(from McIntyre, 1980)
4.3.3. Air movement

Indoor air velocity was measured using medium range kata thermometers (kindly provided
by the BRE). Nicol (1993) describes this traditional method of measurement:
"This (the kata thermometer) is essentially an alcohol-in-glass thermometer with a
large bulb. The stem of the thermometer is marked with only two graduations, 5F
apart. The method is to heat the thermometer to a temperature above the upper
graduation, and then to measure the time the alcohol column takes to drop between
the marks. This is called the Kata Cooling Time and is dependent on air temperature
and air velocity in a way that allows air velocity to be calculated."
The indoor air speeds (Va) were calculated using the formula:
Va = {(2.82K / (t(53-Ta))) -0.6}2

Where:
K = Kata Factor of the thermometer
1
Recorded tinytalk temperature used, see section 4.3.1.

45
t = cooling time
Ta = air temperature2
The kata thermometer overcomes any problem of directionality, but, like all manual
methods, is time consuming, demanding both patience and persistence on the part of the
experimenter. The horizontal direction of the air movement was estimated using the smoke
from burning joss sticks. This method was not used to estimate vertical direction as the heat
from the joss stick would have made this unreliable.
Outdoor air velocity was measured using an Airflow vane anemometer, with a maximum
accuracy of 3%.
4.3.4. Surface temperature

Surface temperature was measured using a Digitron digital handheld infrared thermometer.
This made it possible to measure surface temperatures of building elements not easily
accessible to contact thermometers. The Digitron thermometer has an accuracy of 3% of
the reading. The measurement area is shown in Illustration 4.1
Ill. 4.1. Measurement area of Digitron handheld infrared thermometer.
Recorded tinytalk temperature used, see section 4.3.1.

46
4.4 Data collected

4.4.1. Summer95 series
This data was collected in Hotel Suraj while the author was present and so for this period
some information is available regarding weather outdoors and human activity indoors.
Twelve tiny talk temperature data loggers and three humidity loggers were used. The loggers
were started at 11.30am on the 23rd June 1995 and set to record at 36-minute intervals,
giving 40 readings per day. Three runs were downloaded, giving a recorded period of just
less than five weeks, ending on the 21st July at 9am.
4.4.1.1. Logger locations

The air temperature in a room can vary substantially from place to place. In particular there
can be marked vertical layering of air temperature in a room. Humphreys (1974) found that
the temperature measured near ceiling level in a school classroom with warm air convector
heating was as much as 10C higher than the temperature experienced by the children sitting
on the floor. Ideally temperature sensors should be placed at a vertical height representative
of the occupants' experience (Nicol, 1993). However, as tourists were using Hotel Suraj
during the recording periods, the loggers needed to be both unobtrusive and secure. The high
ventilation rate in the majority of rooms would considerably reduce the vertical temperature
gradient, and so, for reasons of security all loggers were tied from existing hooks at a
distance of approximately 25cm from the ceiling.
Four of the temperature loggers were located externally, on the ground floor of the central
courtyard, on the porch, in the second floor courtyard and under the balcony eaves of the
mahal on the first floor. The other nine temperature loggers were located internally, in the
major spaces of the haveli on each of the four floors. Relative humidity recordings were
taken in the basement, on the porch and in the courtyard on the second floor. The exact
locations of all loggers are shown in ill. 4.2.
47
KEY
Humidity
Temperature
1TopFront
21stFlBack
3Outdoors
4TopMid
51stFlMid
6TopFlBack
71stFlFront
8GrFlBack
9GrFlCourt
10.Basement
11.GrFlPorch
12.TopFlCourt
RH1..PaPorch
RH2..PaTopCourt
RH3..PaBasement
Ill. 4.2. Logger locations for Summer95 data series.3
4.4.1.2. Results
Graph 4.1 shows the results of the summer95 temperature series. The logger located in the
top front room (1) went missing after the second run and so data for the period 7th to the 21st
July for this location is missing.
The recording period for the summer95 series accounts for the tail end of the hot dry season
and the very beginning of the warm wet season. On the 22nd June, the day before logging
began, Jaislamer recorded the highest rainfall in 100 years. The following seven days show a
gradual increase in daily maximum temperature, whilst daily minimum temperatures remain
almost constant. On the 28th and 29th June the outdoors temperature logger (no.3) recorded
maximum temperatures of over 47C at 7.00pm. The recorded temperatures for the other
monitored outdoor zones - the top floor courtyard (no.12), the ground floor porch (no.11)
and the ground floor courtyard (no.9) - for the same period are 41C, 41.5C and 39.5C
respectively. This suggests that the external spaces in and around Hotel Suraj temper the
ventilation air before it enters the haveli.
Logger no.4 (TopMid) actually located in a room on the near side of the section AA
48
50
45
TopFront
1stFlBack
outdoor
TopMid
40
temp
1stFlMid
TopFlBack
1stFlFront
GrFlBack
35
GrFlCourt
Basement
GrFlPorch
TopFlCourt
30
7/21/95 12:00
7/19/95 12:00
7/17/95 12:00
7/15/95 12:00
7/13/95 12:00
7/11/95 12:00
7/9/95 12:00
7/7/95 12:00
7/5/95 12:00
7/3/95 12:00
7/1/95 12:00
6/29/95 12:00
6/27/95 12:00
6/25/95 12:00
6/23/95 12:00
25
time
Graph 4.1 Temperature data recorded for the summer95 series in Hotel Suraj
The 10 day period from the 1st to the 11th July shows considerably lower maximum and
minimum temperatures. Intermittent strong, dust-laden, warm winds were noted throughout
this period, bringing some relief from the heat, but making night ventilation impossible. The
next 7 days (11th to 18th July) were overcast, resulting in uncharacteristically small daily
ranges. This period was particularly thermally stressful during the evenings and nights as
outdoor minimum temperatures were often as high as 31C. The last three days of the
recording period marked the beginning of the warm, wet season, with hard bursts of rain
(accompanied by children dancing in the streets). It is interesting to note that the temperature
recorded in the basement remains constant, at around 30C for the duration of the recording
period.
Graph 4.2 shows the water vapour pressure data for the summer95 period. For the first seven
days of the logging period the recorded Pa in the porch and topCourt zones follow a similar
pattern, with maximum values around 7am and minimum values around 7pm. However as
the weather becomes more windy and overcast the pattern in the topCourt zone noticeably
changes with daily values often peaking mid afternoon. The daily pattern in the porch zone
remains constant, although the daily variation decreases. It is also noticeable that the water
vapour pressure levels recorded in the basement show more variation than the temperature
data recorded in the same location, particularly during the first seven days of the logging
period.
49
water vapour pressure (Pa) (Millibars)
40
35
30
25
PaPorch
PaTopCourt
PaBasement
20
15
10
7/21/95 12:00
7/19/95 12:00
7/17/95 12:00
7/15/95 12:00
7/13/95 12:00
7/11/95 12:00
7/9/95 12:00
7/7/95 12:00
7/5/95 12:00
7/3/95 12:00
7/1/95 12:00
6/29/95 12:00
6/27/95 12:00
6/25/95 12:00
6/23/95 12:00
date
Graph 4.2. Water vapour pressure (measured in millibars) recorded for the summer 95 series in Hotel Suraj
4.4.1.3. Discussion.
Table 4.1 gives a summary of the summer95 temperature and humidity series statistics.
14
10
11
12
Pa1
Pa2
Pa3
AV.
34.1
33.3
33.6
35.0
33.1
33.7
33.9
32.0
33.2
30.2
33.4
33.1
25.5
23.3
25.1
MAX.
42.2
38.8
47.7
40.1
37.6
38.8
39.3
35.7
40.6
33.7
42.2
41.8
37.1
31.4
29.0
MIN.
29.5
29.5
27.0
30.7
28.5
28.8
28.5
28.5
29.2
29.9
27.7
27.4
8.3
12.7
22.6
STDEV.
2.76
1.67
3.81
2.00
1.79
1.94
1.95
1.36
2.17
0.19
3.15
3.25
4.7
2.9
1.0
Tbl.4.1. Descriptive statistics for summer95 series in Hotel Suraj.
In general table 4.1 shows most rooms in Hotel Suraj have similar average temperatures as
that recorded outdoors (no.3). However the top middle (no.4), top front (no.1) and first floor
front (no.7) rooms have notably higher average temperatures than outdoors over this period.
Of the three mentioned the top middle room performs the worst with an average temperature
1.4C hotter than outdoors. This is a small room with little opportunity for ventilation,
This logger went missing after the second run and so data for the period 7th to the 21st July for this location is
missing.
50
orientated so as to receive both east and west solar radiation. It is likely to perform
considerably better in the winter period. Although the results for the top front room give a
high average temperature this is almost certainly due to the short recording period for this
location. The poor result for the first floor front room is the most surprising of the results.
This is the most important room in the haveli (and the most expensive!). The exposed roof
and thin front elevation are possible contributory factors.
It is interesting to note that both the monitored courtyards have significantly lower average
temperatures than outdoors, with the top floor courtyard (no.12) performing slightly better
than the ground floor courtyard (no.9). The first floor back room (no.5) also performs well
during this period, giving an average temperature 0.5C below outdoors. This suggests that
pre-cooled courtyard air ventilates this zone. The basement (no.10) and the ground floor
back room (no.8) are both considerably cooler than outdoors at this time of year. Both these
rooms, but particularly the basement, are probably coupled to the ground temperature,
making annual cycles very significant when monitoring their performance.
It is notable that the two outdoor monitored spaces have quite different water vapour
pressure readings at this time of year, with the top courtyard (Pa2) giving significantly lower
average values than the porch (Pa1). The average value recorded in the basement (Pa3) is
higher than that recorded in the top courtyard, but slightly lower than that recorded in the
porch. This does not support the theory, highlighted by Carpenter in 1994, that excess water
from the open drains is seeping into the foundations of the fort. The increased sensation of
humidity in the basement is a product of the lower temperatures in this zone at this time of
year.
Although most of the monitored zones show similar average temperatures over this period
there is a wide variety of maximum temperatures. This indicates that at any one time there is
a range of available temperatures in the different zones of Hotel Suraj. The basement shows
the lowest recorded maximum temperature, being 14C lower than the recorded outdoor
maximum temperature over the same period.
The average temperature results have been transposed onto the Hotel Suraj section and are
shown in illustration 4.3. This highlights the existence of vertical average temperature
gradation, with the coolest rooms at the bottom of the haveli and the hottest rooms at the top.
There also seems to be a less significant, but none the less evident, horizontal gradation in
temperature from the back to the front of the haveli. The more massive back of the haveli,
although orientated south-westwards, performs better thermally than the lighter front of the
haveli. This suggests that available mass is more significant than orientation in tempering
indoor thermal environments.
51
Ill. 4.3. Average temperature results from the summer95 series in Hotel Suraj.
4.4.2 Suraj-W95-96 series

At midday on the 17th October 1995 five temperature data loggers were placed in Hotel
Suraj, one on each of the four floors and one on the ground floor of the main courtyard. They
were set to record at 2hr 24min intervals giving ten readings per day. In all 1642 readings
were collected over a six-month period which ended on the 29th March 1996.

The author was not present for this logging period and with dozens of tourists passing
through the hotel every week security was a primary concern when choosing logger
locations. The four rooms chosen are not for tourist use. The basement is seldom used.
Occasionally the owner takes an electric fan down to give the room a bit of an airing, but
otherwise the room is unventilated. The ground floor room is used infrequently and only by
the owners. It has no windows, and so the only ventilation is through the door, which opens
onto the courtyard. The external wall is buried to a height of 1.5m above floor level. The first
floor location is a small room, formerly a corridor/balcony and now used only for storage. It
has windows opening into the courtyard, but the wooden shutters, although ill-fitting, are
constantly closed. On the second floor, the room is again used only for storage. It has two
small windows, one facing the fort bastions, the other facing the chandni. Again both have
ill-fitting shutters, which are constantly closed. Air changes in all the internal locations are
52
therefore relatively small and constant. The external west wall of the first and second floor
rooms was originally a party wall with the neighbouring haveli. Unfortunately this haveli is
now only one storey high, and so both these rooms are exposed to solar radiation on the west
elevation. External temperature was recorded on the ground floor of the main courtyard as
this was considered the most secure outdoor location.
The loggers were all fixed into position as for summer95.xls. The exact locations are shown
in ill. 4.4.
KEY
1.Basement
2.GrFlBack
3.1stFlBack
4.2ndFlBack
5.Outdoors
Ill. 4.4. Logger locations for Suraj-W95-96 series in Hotel Suraj
4.4.2.2. Results
Graph 4.3. shows the results for the Suraj-W95-96 temperature series. Results indicate that,
for this period, all monitored rooms appear to perform well thermally, in general providing a
warmer environment indoors than out. Although the basement shows no daily variation, it
does show considerable annual variation. Results around mid January show a difference of
up to 13C between minimum outdoor and equivalent basement recorded temperature.
Results are appreciably smaller for the rest of the monitored period.
53
40
35
30
Outside
Basement
temp
25
GrFl-back
20
1stFl-back
2ndFl-back
15
10
3/19/96
3/5/96
2/20/96
2/6/96
1/23/96
1/9/96
12/26/95
12/12/95
11/28/95
11/14/95
10/31/95
10/17/95
time
Graph 4.3. Temperature data for Suraj-W95-96 series in Hotel Suraj.
4.4.2.3. Discussion
Table 4.2 gives a summary of the Suraj-W95-96 temperature series statistics for all data in
each monitored zone. In addition the statistics for the temperatures recorded in January (the
coldest month) are also shown.
BASEMENT
GRFLBACK
1STFLBACK
2NDFLBACK
OUTDOORS
ALL
ALL
ALL
ALL
ALL
JAN.
JAN.
JAN.
JAN.
JAN.
MAXIMUM
25.51 23.81 23.89 19.79 24.95 20.22 24.92 20.22 22.04 17.09
23.79 36
24.89 36.41 22.03
30.27 24.51 31.4 21.71 36
MINIMUM
23.09 23.09 17.83 17.83 15.38 15.38 14.69 14.69 9.34
ST. DEV.
1.95
AVERAGE
0.45 3.58
0.77
4.24
1.64
4.39
1.97 4.91
9.34
2.59
Tbl. 4.2. Summary temperature statistics for Suraj-W95-96
The results from table 4.2 show that all indoor average temperatures are higher than outdoors
during the winter period. As both the basement and ground floor back room are most likely
coupled with the ground temperature, this is an expected result for these two zones.
However, both the first and second floor rooms have average temperatures of almost 3C
higher than the outdoors over this period, with the average statistics for the month of January
also showing the same pattern. This can only be a product of considerable passive solar gain
through the external walls and shuttered windows of the rooms. Both range and standard
54
deviation show a steep increase from the bottom to the top of the haveli, indicating an
inverse relationship with the available mass in each monitored zone. The average
temperature results have been transposed onto the Hotel Suraj section and are shown in
illustration 4.5.
Ill.4.5. Average temperatures for the SurajW95-96 series in Hotel Suraj
4.4.3. Suraj-S96 series

At 15:25 on the 3rd April 1996 the five temperature data loggers were replaced in hotel Suraj.
They were reset to record at 2hr. 24min. intervals again giving ten readings per day. In all
1800 readings were collected over a six month period with the run ending on the 30th
September 1996 at 13:00.

All temperature loggers were located as for the Suraj-W95-96 series (shown in Illustration
4.4). Unfortunately the data logger on the second floor was stolen sometime during this
period and so data for the entire six months at this location is missing.
55
4.4.3.2. Results
Graph 4.4. shows the results for the Suraj-S96 temperature series.
45
40
Outside
temp
35
basement
Ground-back
30
First-back
25
9/18/96
9/4/96
8/21/96
8/7/96
7/24/96
7/10/96
6/26/96
6/12/96
5/29/96
5/15/96
5/1/96
4/17/96
4/3/96
20
time
Graph 4.4. Temperature data for the Suraj-S96 series.
It is immediately noticeable that the daily average outdoor temperature is much more stable
over this six month period than in the previous data set, although the daily range seems to
show more variation. This is illustrated in graph 4.5, which shows the calculated daily
average temperatures for both the SurajW95-96 and Suraj-S96 data sets.
40
SurajW95-96
35
30
OpenCourt
temp
Basement
GrndBack
25
1stFloor
2ndfloor
20
SurajS96
15
9/17/96
8/20/96
7/23/96
6/25/96
5/28/96
4/30/96
4/2/96
3/5/96
2/6/96
1/9/96
12/12/95
11/14/95
10/17/95
10
date
Graph 4.5. Calculated daily average temperatures in Hotel Suraj over a 12 month period.
56
The relationship between the basement and outdoors is inverted in the summer with the
basement giving consistently lower temperature readings than outdoors. The summer
temperatures in the ground floor back room seem to be slightly cooler than outdoors, whilst
those of the first floor room are higher.
4.4.3.3. Discussion
Table 4.3 gives a summary of the Suraj-S96 temperature series statistics for all data in each
monitored zone. In addition the statistics for the temperatures recorded in June (the hottest
month) are also shown.
AVERAGE
MAXIMUM
MINIMUM
ST. DEV.
BASEMENT
GRFLBACK
1STFLBACK
OUTDOORS
ALL
ALL
ALL
ALL
JUNE
JUNE
JUNE
JUNE
29.11
29.68 31.74
32.49 33.54
34.19 32.10
32.76
30.27
29.93 35.65
35.65 39.26
39.26 41.77
41.77
26.28
29.20 28.48
28.48 27.72
27.72 24.51
25.21
0.94
0.20 1.18
1.19 2.20
2.68 3.16
3.72
Tbl. 4.3. Summary of Suraj-S96 temperature series statistics.
The results in table 4.3 show that average temperature, range and variation in temperature all
increase from the bottom of the haveli to the top. The first floor back room again has an
average temperature higher than outdoors, although not to such a degree as in the previous
data set.
Both the Suraj-W95-96 and the Suraj-S96 series were combined into one twelve month
series - Suraj-12months. The annual average temperatures and annual range for each of the
monitored zones were calculated and transposed onto the Hotel Suraj section5. The results
are shown in illustration 4.6.
57
KEY
Annual Average Temperature
Annual Range
Ill. 4.6. Annual average temperature and annual range for Hotel Suraj
Results over the one year period show that both the basement and ground floor zones have
similar annual average temperatures to that of the outdoor monitored zone. This suggests
that, when considered over a twelve month period, direct ground coupling cannot be
accurately described as a passive cooling system. Ground coupling merely dampens the
annual temperature range6. This is in contradiction to the writings of many contemporary
researchers (Givoni, 1990; Labs, 1981).
4.4.4 2ndhaveli-S96 series

At half past five on the 3rd April1996 five data loggers were placed in the second haveli, one
in the basement, two on the ground floor and two on the first floor. They were set to record
at 2 hours and 40 minutes intervals giving ten readings per day. In all 1800 readings were
collected over a six-month period ending at four thirty on the 30th September.

The exact logger locations are shown in illustration 4.7. Unfortunately both data loggers on
the first floor were stolen sometime during this period and so data for the entire six months
at these locations is missing.
5
6
Data for the second floor back room was excluded as it was only recorded over a six month period.
It also shifts the maximum and minimum temperatures forward in time. This is discussed in chapter 8
58
KEY
1 GrFlMainCrt
2 GrFlFront
Basement and SurajOut
not shown.
Ill. 4.7 Logger locations for the 2ndhaveli-S96 series.
4.4.4.2. Results
Graph 4.5 shows the results for the 2ndHaveli-S96 series. Outdoor temperature recorded over
the same period in the courtyard of Hotel Suraj has been included for comparison.
45
40
SurajOut
35
temp
basement
GrFlMainCrt
30
GrFlFront
25
9/29/96 0:00
9/15/96 0:00
9/1/96 0:00
8/18/96 0:00
8/4/96 0:00
7/21/96 0:00
7/7/96 0:00
6/23/96 0:00
6/9/96 0:00
5/26/96 0:00
5/12/96 0:00
4/28/96 0:00
4/14/96 0:00
3/31/96 0:00
20
date
Graph 4.5. Temperature data recorded for 2ndHaveli-S96 series
59
It is immediately noticeable that the courtyard of the second haveli is performing quite
differently to the courtyard in Hotel Suraj, showing much less variation over this summer
period. This may be because this haveli was closed and unoccupied during the monitoring
period, therefore limiting the possibilities for cross ventilation (a marked characteristic of
hotel Suraj). The basement of the second haveli, however, seems to show more temperature
variation than that of Hotel Suraj, although still seems cooler than any of the other monitored
zones.
4.4.4.3. Discussion
Table 4.4 gives a summary of the 2ndhaveli-S96 temperature series statistics.
AVERAGE
MAXIMUM
MINIMUM
STDEV.
BASEMENT
GRFLMAINCRT GRFLFRONT
SURAJOUT
30.30
32.94
25.58
1.37
31.59
36.79
27.0
1.81
32.10
41.77
24.51
3.16
31.37
36.0
27.38
1.61
Tbl. 4.4 Summary of 2ndHaveli-S96 temperature statistics.
The results from table 4.4 show that, at this time of year, the second haveli basement is not
only over 1C warmer on average than it's counterpart in Hotel Suraj, but also shows
considerably more variation over the same period. The converse is true when comparing the
two courtyards. The courtyard of the second haveli is, on average, over 0.5C cooler in the
summer than that of Hotel Suraj, and displays much less variation for the entirety of this
period. Results for the ground floor front zone are very similar to those of the ground floor
back room in Hotel Suraj, despite the latter benefiting from a semi-buried external wall.
4.4.5. March96 series

In addition to the long term monitoring of the various zones in the two havelis, a series of
manual measurements were recorded in Hotel Suraj during the March, 1996 field study
period. The five temperature data loggers were reset to record at 16 minute intervals in the
locations shown in illustration 4.4 in order to support the manual recording. The recording
began at 9am on the 30th March and ended at 3pm on the 3rd April. An additional humidity
60
logger, located next to the ground floor back temperature logger, was set to record at the
same intervals over the same period.
4.4.5.1. Temperature results

Graph 4.6 shows the temperatures recorded over the March96 period.
38
36
34
32
temp
Basement
30
GrFlBack
1stBack
2ndBack
28
Courtyard
26
24
22
4/3/96 15:00
4/3/96 9:00
4/3/96 3:00
4/2/96 21:00
4/2/96 15:00
4/2/96 9:00
4/2/96 3:00
4/1/96 21:00
4/1/96 15:00
4/1/96 9:00
4/1/96 3:00
3/31/96 21:00
3/31/96 15:00
3/31/96 9:00
3/31/96 3:00
3/30/96 21:00
3/30/96 15:00
3/30/96 9:00
20
time
Graph 4.6. Recorded temperatures over the March96 period.
The outdoor temperature (measured in the courtyard) varies between a minimum of 24C at
around seven in the morning and a maximum of 36C at around three in the afternoon. The
shape of the outdoor temperature curve shows that the courtyard heats up considerably faster
than it cools down. Furthermore, on the night of the 31st March and 1st April, both the first
and second floor zones cool down as much as 2C further than the courtyard. This may be
attributable to internal gains in and around the courtyard at this period.
4.4.6. Courtyard surface temperatures - HAV-HT series.

On the 1st April surface temperature readings in the main courtyard were manually recorded
using the Digitron infra red thermometer. Readings were taken for the courtyard floor and
six locations on each of the four courtyard elevations. These readings were repeated 6 times
during the day, at 7:00am, 10:20am, 1:00pm, 4:00pm, 7:00pm and 10:30pm. In addition air
temperature was simultaneously recorded on the ground floor of the courtyard.
61
4.4.6.1 Results
Because of the problems associated with displaying many locations (26 in this series) on one
graph, results of the hav-ht data set have been presented using images generated by the
thermamator program, developed by the author and described in chapter 6. All surface
temperatures are shown in the appropriate zone of a bird's eye perspective of the Hotel Suraj
courtyard. The recorded air temperature is shown within the north-point. In addition solar
perspectives7 of the central courtyard area of Hotel Suraj at 10:20am, 1:00pm and 4:00pm on
the 1st April were graphically constructed, and are shown alongside the relevant results.
These diagrams show which areas of the courtyard were exposed to direct sunlight at the
time the results were recorded.
7:00am
A graphic technique used to draw buildings as seen by the sun (Los, 1981).
62
10:20am
1:00pm
63
4:00pm
7:00pm
64
10:30pm
4.4.6.2. Discussion.
Results from the hav-ht series show that surface temperatures on the ground floor of the
courtyard, and particularly the courtyard floor are cool throughout the day. Between 7:00am
and 10:20am the surface temperatures of the first and second floor elevations are similar to
the air temperature recorded on the ground floor of the courtyard. However, between 1:00pm
and 4:00pm these temperatures, particularly on the north and the east elevations, rise
dramatically reaching a maximum of 47C on the second floor at 4:00pm.
The Hav-ht series was recorded at the beginning of April, and at this time of year the 1pm
solar perspective shows that the sun does not quite penetrate to the ground floor of the
courtyard, and so this area remains cool throughout the day. However, in May and June,
when air temperatures are much hotter, and the sun higher in the sky, this is not likely to be
the case, and the author would expect surfaces, even on the ground floor, to be considerably
hotter than recorded air temperature during the afternoon. During this extreme hot, dry
period it is very unlikely that the courtyard would pre-cool Hotel Suraj's ventilation air.
65
4.4.7. Air movement - Vyas-Air series

Medium range kata thermometers were used to measure
indoor air speeds at a height of 600mm above floor level.
A grid was drawn on the floor plan and spot
measurements taken at each grid point. The data
collectors worked in 3 teams of 2, each pair being
responsible for one of three rows of locations (row a, b or
c). The cooling time of the Kata thermometer was
measured using a stopwatch, whilst the air temperature
was simultaneously recorded using a hand held digital
Ill. 4.8. Recording velocity of air
movement in Hotel Suraj.
Ill. 4.9. Recording direction of air

movement in Hotel Suraj.
thermometer. The horizontal direction of the air

movement was estimated using the smoke from burning
joss sticks. This method was not used to estimate vertical
direction as the heat from the joss stick would have made
this unreliable. An average wind speed and direction
measurement was recorded on the roof, using a vane
anemometer, for the duration of the exercise.
This process was repeated three times during the day,
nine times in total. The morning readings were taken, as
far as possible, between 9.00am and 10.00am, the
afternoon readings between 3.00pm and 4.00pm and the
evening readings between 10.00pm and 11.00pm. These
nine runs are as follows:
29-3-96-EVE. Readings taken from 22:00 to 22:40 on the first floor giving 18 locations
in total.
30-3-96-MORN. Readings taken from 9:35 to 10:27 on the ground, first and second
floors, giving 33 locations in total.
30-3-96-AFT. Readings taken from 15:10 to 16:05 on the ground first and second floors,
giving 29 locations in total.
30-3-96-EVE. Readings taken from 22:45 to 23:36 on the first floor giving 19 locations
in total.
31-3-96-MORN. Readings taken from 10:55 to 12:22 on the first and second floors
giving 24 locations in total.
66
31-3-96-AFT. Readings taken from 15:40 to 16:10 on the ground and first floors giving
23 locations in total.
1-4-96-MORN. Readings taken from 10:23 to 11:25 on the ground first and second
floors giving 28 locations in total.
1-4-96-EVE. Readings taken from 22:00 to 23:31 on the first floor giving 19 locations in
total.
2-4-96-AFT. Readings taken from 15:33 to 16:09 on the ground and first floors giving 23
locations in total.
4.4.7.1. Results
These recorded air speeds are shown on the following plans. The arrows represent the
location, direction and magnitude of the air movement (see Appendix A for data in tabulated
format).
67
4.4.7.2. Discussion
Illustrations 4.10. to 4.12. show estimated airflow patterns during the morning, afternoon and
evening in Hotel Suraj. The 10:20am and 10:30pm courtyard surface temperature recordings
discussed in section 4.4.6. are concurrent with the 1-4-96-MORN and 1-4-96-EVE air
movement series and these results have been used to inform the likely vertical air movement
within the main courtyard of Hotel Suraj.
Ill. 4.10. Estimated direction of air movement in Hotel Suraj between 10:00am and 11:00am at the beginning of
April 1996.
Ill. 4.11. Estimated direction of air movement in Hotel Suraj between 3:00pm and 4:00pm at the beginning of
April 1996.
68
Ill. 4.12. Estimated direction of air movement in Hotel Suraj between 10:00pm and 11:00pm at the beginning
of April 1996.
Results from all air velocity data recorded on the second floor suggest that, throughout the
day, the prevailing wind is the driving force behind indoor air movement on this level.
However, the results for the first floor show evidence of stack effect within the central
courtyard with air moving outwards from the courtyard in the hot afternoon, and inwards
during the cooler evening, despite the constant direction of the prevailing wind. In the hotter
months of May and June, with the sun penetrating to the floor of the courtyard the author
would expect this effect to be increased.
In general Hotel Suraj is well designed regarding air movement, providing ample
opportunity for occupant control. Openings are located to facilitate cross ventilation and
shutters are designed to enable varying degrees of air movement. In addition, the
measurements recorded in the haveli suggest that, during the hottest part of the year stack
effect within the courtyard would further facilitate ventilation, regardless of outdoor air
velocity.
4.5. Conclusions.
Results from the temperature data gathered in and around Hotel Suraj show that:
In the summer the majority of indoor spaces have cooler daily average temperatures than
outdoors.
69
In the winter the majority of indoor spaces have warmer daily average temperatures than
outdoors.
Most of the monitored indoor spaces have a similar, or slightly higher, annual average
temperature than outdoors.
Daily temperature range and annual temperature range increase from the bottom to the
top of the haveli at all times of year, giving a variety of maximum and minimum indoor
temperatures.
The central courtyard contributes, by means of stack effect, to the ventilation of the
indoor spaces from April through to July.
As discussed in chapter 3, only a few rooms in a haveli were traditionally designed for a
specific purpose. Instead, larger rooms were used as season, need and mood dictated. The
results recorded in Hotel Suraj indicate that, at any one time, there are a variety of indoor
temperatures available. The occupants migrate around the haveli at different times of day, in
the various seasons, taking advantage of the different climatic conditions in each place. Roaf
(1989) also noted this "intra-mural migration" in her study of traditional housing in Yazd. In
addition the results suggest that the available mass in Hotel Suraj is the largest factor
governing the climate indoors. In high mass zones, such as the basement, with little or no
daily fluctuation in temperature, long logging periods are essential, spanning, ideally, a
minimum of twelve months.
This chapter has detailed the temperatures achieved within Hotel Suraj over a twelve month
period. Chapter 5 converts these recorded temperatures into predicted occupant thermal
comfort sensation.
70
BALARAS, K. (1994). Thermal Comfort. In: Passive cooling of buildings. (Santamouris ed.)
Pp. 91-125. C.I.EN.E., Athens.
CARPENTER, S. (1994). Collapse of the Golden City. In: New Scientist, Dec. 17th.
London.
GIVONI, B. (1990). Performance and applicability of passive and low-energy cooling
systems. Energy and Building, vol.17, pp.177-199.
HUMPHREYS, M. (1974). Classroom temperature, clothing and thermal comfort - A study of
secondary school children in summer. Building Services Engineer , vol. 41, pp191202.
LABS, K. (1981). Direct-coupled ground cooling: Issues and opportunities. In: Proceeds of
the International Passive and Hybrid Cooling Conference. Pp. 131-135. Miami
Beach, Florida.
LOS, S. (1981). Solar perspective: A graphic techique to draw buildings as seen by the sun.
In: Proceeds of the International Passive and Hybrid Cooling Conference. Pp. 131135. Miami Beach, Florida.
NICOL, J. F. (1993). A handbook of field studies toward an adaptive model. University of
East London.
ROAF, S. (1989). The windcatchers of Yazd. PhD thesis. Department of Architecture,
Oxford Brookes University.
71
72
CHAPTER 5.
THERMAL COMFORT IN JAISALMER
This chapter gives an introduction to the science of thermal comfort, describing

both climate chamber and field study methods of comfort prediction. A comfort
survey conducted in Jaisalmer is analysed, and evaluated in the wider context
of accepted adaptive model comfort prediction equations. The results are
applied to the assessment of the data detailed in chapter 4 in terms of occupant
comfort level.
5.1. Introduction
"The thermal qualities - warm, cool humid, airy, radiant, cosy are an important part
of our experience of space; they not only influence what we choose to do, but also
how we feel about the space. " Heschong (1990)
In reality the temperatures recorded in a building mean little unless set in the context of
occupant comfort. Olygay (1992) states that man's energy and health depend in large
measure on the direct effects of his environment. In excessively hot or cold conditions both
health and productivity decline. Man, however, can survive in a wide range of environments.
The following sections detail mans adaptive abilities and limitations. Later in the chapter
upper and lower comfort limits for the population of Jaislamer will be estimated over a
twelve month period, and used to assess the performance of Hotel Suraj.
5.2. Heat Exchange mechanisms

Man has developed thermal strategies to cope with adverse climates. The body keeps the
temperature of the internal tissue more or less constant, regardless of the relatively wide
variations in the external environment.
The physical means of heat exchange between the human body and its surroundings
comprise the processes of radiation, conduction, convection and evaporation. Radiation is in
the form of electromagnetic waves, which pass from a hot object to a cooler one even
through a vacuum. The amount of heat loss by radiation from the body to the surroundings is
dependent upon the temperature difference between the body and the mean temperature of
all the surfaces surrounding it. Conduction refers to the flow of heat from one object to
another with which it is in contact. Little body heat exchange takes place via this route.
Convection indicates the exchange of warm and cold molecules between the surfaces of
objects separated by a fluid or gaseous medium. Heat loss or gain by convection is
dependent upon the difference in temperature between the body's surface and the air. A layer
of air in contact with the surface is heated up, and because air is a poor conductor, heat loss
relies primarily on the removal of this layer of warm air by air currents. Wind or cooling of
the skin with an electric fan can be considered as a forced convection, and it is quite
effective as a heat exchange route. Customarily convective heat exchange is assumed to be
proportional to the square root of the air velocity (Givoni, 1969). The humidity determines
the evaporative capacity of air and hence the cooling efficiency of sweating. Evaporation of
sweat is the chief protective mechanism against over heating when ambient temperature rises
above 32C to 34C (Samueloff, 1980).
74
In a hot climate context heat exchange by physiological means is accomplished in 2 steps:

1. Heat transfer by circulation from the body core to the surface. The body is able to dilate
or constrict blood vessels beneath the skin and in this way augment or restrict blood flow
in the subcutaneous layer. This controls the skin surface temperature and thereby the
heat lost or gained by convection or radiation.
2. Heat transfer from the skin accomplished by sweat evaporation.
The rate at which heat can be dissipated from the body is therefore dependent on four
environmental factors:
1. Air temperature
2. Humidity
3. Air movement
4. Radiant temperature
5.3. Behavioural adaptations

Man can also consciously control the amount of heat lost and gained by the body through
behavioural changes. The most obvious reaction is to change clothing level, but people can
also regulate their metabolic rate through a change in activity level, or maximise their
surface area by adopting a different posture.
However in an overheated climate there is a limit to how much clothing can be taken off.
Cook (1981) points out in hot climates were radiation is dominant (as in Jaisalmer), more
clothing rather than less may be desirable. Research by Berger (1988) indicates that in a hot
dry climate the main discomfort felt is due to the dryness of the air. When humidity is below
20% the skin withers and chaps. Sweat evaporation increases the humidity of the air close to
the skin and at temperatures above 37C this air layer will be cooled. Under these conditions
he suggests baggy white garments with small renewal rates, as most thermally suitable.
In addition human beings will migrate towards more favourable conditions. Roaf (1992)
describes the way the occupants of traditional houses in the Middle East move between the
various environments in different parts of the house to ensure their comfort throughout the
day. Similar diurnal and seasonal migration patterns can be observed in the people of
Jaisalmer.
Although food intake does not produce any significant variation of heat production
throughout the day (McIntyre, 1980) it has been noted (Cook, 1981) that in overheated
climates food is lighter, with less animal fats consumed. The author was surprised at the
amount of piping hot chai (spiced Indian tea) consumed in Jaisalmer. Research by Clark et
al. (1977) showed that in hot conditions, were the body is sweating, the intake of very hot
75
drink will produce a sharp increase in the sweat rate, which paradoxically leaves the skin
temperature lower than it was, and the person feeling cooler. A hot drink on a hot day is
more effective at cooling the body than a cool drink.
5.4. The science of thermal comfort

The advent of fire allowed human beings a mechanism, independent of the fluctuations of
climate and their own metabolism, for coping with adverse climates. Fascination with this
potential for control of the environment prompted the invention of mechanical systems that
made natural thermal strategies seem obsolete by comparison. First the problem of heating
was solved, and then Wilis Cartier invented air-conditioning. Once the technology to
completely control the thermal environment was developed people became curious about
what a truly optimal thermal environment might be. The science of thermal comfort was
born.
In 1966 the American Society of Heating, refrigeration and Air-conditioning Engineers
defined thermal comfort as "that state of mind which expresses satisfaction with the thermal
environment" (ASHRAE 1966). Three years later Givoni (1969) defined it as "the absence of
irritation and discomfort due to heat or cold". The science of thermal comfort has been
concerned with predicting what set of conditions corresponds most closely to this optimal
thermal environment, and how tolerant people are of deviations from it. As thermal sensation
cannot be measured in any direct way scales of warmth sensation were developed. In this
way subjective comfort votes were given numerical values and treated quantitatively so that
statistical analysis may be carried out. Yaglou (1927) was using the first of these scales over
seventy years ago, but the scale used throughout this project is the scale introduced by
Bedford (1936) during his survey of conditions in factories in the United Kingdom.
BEDFORD SCALE
Much too warm
Too warm
Comfortably warm
Comfortable (and neither cool nor warm)
Comfortably cool
Too cool
Much too cool
Tbl. 5.1. The Bedford thermal sensation scale.

76
5.5. Climate chambers

Many attempts have been made to express some or all four environmental variables in one
single figure or comfort index. Early work in this area was carried out using climate
chambers - specially constructed rooms in which the thermal environment could be carefully
controlled. The most widely recognised work done on climate temperatures is that done by
Ole Fanger (1970). He proposed a model in which the heat produced through metabolism
was equal to the heat lost to the environment from the surface of the body by convection,
evaporation or radiation. Using the laws of heat exchange he devised a formula for
predicting the comfort vote for a set of conditions - the Predicted Mean Vote or PMV.
Fanger's research was performed in steady-state conditions. Subjects remained in any one
particular set of condition for three hours and so time was effectively stopped. It was
Fanger's stated view that "in all case thermal comfort is the 'product' which is produced and
sold to the customer by the heating and air-conditioning industry" (Fanger 1970), and in such
an environment, conditions would be held more or less constant. However, in non-airconditioned buildings, the environment is constantly changing.
5.6. The Adaptive Model

The adaptive model of thermal comfort was first proposed in 1972 by Nicol et al.
Humphreys (1993) described the adaptive principle this model is based on: "If a change
occurs such as to produce discomfort, people react in ways which tend to restore their
comfort". An alternative technique was used. People were met in their normal environment,
whilst performing their normal activities, and asked their opinion of the thermal
environment. The environment is measured, but not altered as it is the reaction to their real
life context that is being explored.
Results from field surveys (Humphreys, 1976) indicated that in the long term people in
different places adapt to the average conditions they experience. About 90% of the variation
in the comfort temperatures can be explained by the variation of the average temperature.
Comfort temperature is related to indoor temperature, which, particularly in free-running
buildings, is related to outdoor temperature. These results are shown in graph 5.1. Each
circle represents the results of one survey.
77
Graph 5.1. The relationship between the comfort temperature Tc derived from a particular survey, and the
monthly mean outdoor temperature (from Humphreys, 1978).
Humphreys (1978) was the first to show the strength of the relationship between outdoor
temperature and comfort temperature indoors. Using regression analysis he showed that:
Tc = 12.1 + 0.53Toequation 5.1

Where:
Tc = Comfort temperature
To = Monthly mean outdoor temperature
To reinforce the notion that PMV was not telling the whole story Humphreys (1992) took
published research from 5 such field studies and compared PMV with actual mean vote. He
found that PMV was about 1.5 units too low at 20C and about 1.5 units too high at 33C
and unbiased at around 27C. This showed that people are comfortable in a much wider
range of indoor climates than would have been expected from the results of climate chamber
research.
The discrepancy with PMV could be explained by behavioural adaptations to local climates.
In Nicol's field survey work in Pakistan (Nicol et al., 1994) it was shown that changes in
clothing level accounts for about 3.5 - 4C of the change in comfort level, and changes in air
speed accounts for about 1-2C. In this environment it was shown that
Tc = 17.0 + 0.38To. equation 5.2
78
Where:
Tc = Comfort temperature
To = Monthly mean outdoor temperature
5.7. The Jaisalmer comfort survey.

During the March96 field study period a short transverse thermal comfort survey of the
general population in Jaisalmer was undertaken. Seventy-three people were interviewed in
and around the streets of the fort during the morning, afternoon and evening of the 29th
March and 2nd April.
5.7.1. Method
Illustration 5.1 shows the format of the questionnaire used for the Jaisalmer survey.
COMFORT SURVEY
1) Subject number (allocate)
2) Indoors/Outdoors
3) Date
4) Time
5) Where from
6) Comfort vote:
1.
2.
3.
4.
5.
6.
7.
Much too cold

Too cold
Comfortably cool
Neither warm nor cool
Comfortably warm
Too warm
Much too warm
1.
2.
3.
4.
5.
Much warmer
A bit warmer
No change
A bit cooler
Much cooler
7) Preference:
8) No. of layers of clothing (excluding under-garments)

9) Activity:
1.
2.
3.
4.
5.
6.
Relaxing
Sitting quiet
Sitting active
Standing
Walking
Housework (active)
10) Air temperature

11) Globe temperature
12) Wet bulb temperature
13) Finger tip temperature
Ill. 5.1. Jaisalmer survey questionnaire format
79
The surveys were undertaken by groups of 2 to 3 researchers. One researcher asked the
survey questions whilst the others simultaneously recorded the environmental measurements.
Air and wet bulb temperatures were measured using a whirling hygrometer. Wet bulb
temperature was converted into water vapour pressure using the formula:
Pa = exp(18.956-(4030/(Tw+235)))-0.667(Ta-Tw)
And into relative humidity using the formula:
RH = Pa/exp(18.956-(4030/Ta+235)))
where:
Pa = water vapour pressure
Tw = wet bulb temperature
Ta = dry bulb temperature
RH = relative humidity
5.7.2. Calculating comfort temperature.

As the sample size in the Jaisalmer study is relatively small the Griffiths method has been
used to estimate the comfort air temperature (Tc(air)) and the comfort globe temperature
(Tc(globe))1. Griffiths introduced this method in 1990 for assessing the mean comfort
temperature for a small sample of comfort votes. He was dealing with data from buildings
and groups of buildings in which he had small samples of subjects. The data was insufficient
to produce a reliable regression estimate of the comfort temperature. Griffiths made the
assumption that the increase in temperature for each scale point on the comfort scale was
3C for a 7 point scale. This is the value obtained from climate chamber estimates. For each
comfort vote he subtracted 3C times the number of scale points above neutral (4) to obtain
the temperature which might be expected to result in neutrality. So, for each subject, the
estimated neutral, or comfort air temperature is calculated using the following equation:
Tc(a) = Ta-(C-4)3
Where:
1
Globe temperature data was only available for 53 out of the 73 subjects.
80
Tc(a) = comfort air temperature

Ta = air temperature at time of comfort vote
C = comfort vote (on a seven point scale)
By taking the mean of these temperatures he obtained a mean comfort temperature for the
sample.
The results of the survey, including estimated globe and air comfort temperatures, are shown
in table 5.2.
C
MEAN 4.59
Tglobe
Tair
Twet
Pa
RH
TfingTip
Tc(air)
Tc(globe)
34.28
32.15
18.28
11.80
0.25
34.93
30.38
32.52
MIN.
29.2
27
16
7.17
0.15
31.9
23
24.6
MAX.
43
34.5
21
16.38
0.37
38.4
37
40.1
Tbl. 5.2. Results of the Jaislamer comfort survey.
5.7.3. The wider context

Table 5.2 shows that, at the time of the survey (end of March / beginning of April), the
comfort air temperature for the residents of Jaislmer is around 30C. In order to confirm the
suitability of both Humphrey's and Nicol's comfort prediction equations in the climate of
Jaisalmer, values for Tc for March and April have been calculated using both methods. Table
5.3 shows the results using both standardised monthly mean outdoor temperature (Krishan et
al., 1995), and real monthly mean temperature (recorded in the courtyard of Hotel Suraj).
MARCH
APRIL
Monthly Mean (To)
Tc - Humphrey's
Tc - Nicol's
Standard data
26.5
26.15
27.07
Real data
27.2
26.52
27.08
Standard data
31.5
28.80
27.94
Real data
32.17
29.15
28.08
Tbl. 5.3. Estimated comfort air temperature for March and April using Humphrey's and Nicol's equations.
Results from table 5.3 show that both Humphrey's and Nicol's comfort equations give
acceptable results for Jaisalmer at this time of year. The average comfort temperature using
81
real data for March and April is 27.84 using Humphreys equation and 27.58 using Nicol's
equation. Both these results are slightly lower than the value from the Jaisalmer survey.
However the survey was conducted only during the warmer times of day. If subjects had
been interviewed during the cooler night-time or during the early morning hours, the comfort
temperature may well have been lower. In addition the level of humidity in Jaislamer is very
low, making higher temperatures slightly more tolerable. The effects of clothing and activity
levels have been discussed in earlier sections.
As the comfort temperature result achieved using Humphrey's equation appears slightly
more representative of the survey results; it has been used for all further comfort analysis.
5.8. Thermal comfort in Hotel Suraj

More recent research has shown that the relationship between comfort and outdoor
temperature is improved by using the running mean2 of the daily average outdoor
temperature, instead of the monthly mean of the outdoor temperature (Nicol et al., 1995).
This research also showed that the relationship was most improved when the running mean
series was assigned a weight of 0.2. Results from analysis in Chapter 8, detailed in table
8.11, shows that this weight, assigned to daily average data, would give a cycle time shift of
around 6 days. This is in strong agreement with the conclusion drawn by Nicol in 1992 that
people adapt almost completely to a change of conditions in around a week.
However, there are limits to the range of indoor temperatures that people can adapt to. Nicol
et al.'s research (in press) on interior design temperatures in Pakistan, indicates that the
comfort temperature ranges from 19.7C to 31.3C, in a climate with a monthly mean
temperature range of 4.9C to 35.5C.
5.8.1. Method used for calculating comfort levels in Hotel Suraj.
The running mean series, with a weight of 0.2, of the Suraj12months, daily average
series was calculated.
The predicted daily comfort temperature (Tc) was calculated using equation 5.1,
replacing the monthly mean outdoor temperature (To) with the calculated daily value
from the running mean series.
For a detailed description of running mean see chapter 8, section 8.5.

82
Using the Bedford thermal sensation scale, and assuming a 3C increase for each scale
point, upper and lower limits for each thermal sensation were calculated for each day.
The lower limit for neutrality (4) was calculated as the predicted comfort temperature
minus 1.5C; the lower limit for comfortably cool (3) was calculated by subtracting 4.5
from the predicted daily comfort temperature, and so on.
Each division of the thermal sensation scale was assigned an appropriate colour3.
The raw data ( recorded at 2.4 hour intervals) for each of the monitored zones of the
Suraj12months data series was, in turn, superimposed onto the thermal sensation graph.
5.8.2. Comfort level results.

Graphs 5.2 to 5.6 show the predicted comfort results for the outdoor, and each of the four
indoor monitored zones of the Suraj12months series. Predicted daily comfort temperature
results ranged from a minimum of 20.22C in the winter to a maximum of 30.83C in the
summer, well within the comfort temperature limits proposed by Nicol et al. for Pakistan.
Graph 5.2. Predicted outdoor comfort levels for the Suraj12months recording period (data at 2.4 hour
intervals).
Colour associations are discussed in more detail in chapter 6, section 6.2.

83
Graph 5.3. Predicted comfort levels in the basement of Hotel Suraj from October, 1995 to September, 1996
(data at 2.4 hour intervals)
Graph 5.4 Predicted comfort levels in the ground floor back zone of Hotel Suraj from October 1995 to
September 1996 (data at 2.4-hour intervals).
84
Graph 5.5 Predicted comfort levels in the first floor back zone of Hotel Suraj from October 1995 to September
1996 (data at 2.4-hour intervals).
Graph 5.6. Predicted comfort levels in the second floor back zone of Hotel Suraj from October 1995 to
September 1996 (data at 2.4-hour intervals).4
Data for second floor back zone only available for six months.
85
5.8.3. Discussion
The proportion of time that occupants of each of the monitored areas would experience
thermal discomfort was calculated. Discomfort includes bands 7 (much too warm), 6 (too
warm), 2 (too cool) and 1 (much too cool). The results are shown in table 5.4.
%
OCTOBER
NOVEMBER
DECEMBER
JANUARY
FEBRUARY
MARCH
APRIL
MAY
JUNE
JULY
AUGUST
SEPTEMBER
OUTDOORS
BASEMENT
GROUND
FIRST
SECOND
COLD
HOT
COLD
HOT
COLD
HOT
COLD
HOT
COLD
HOT
2.07
5.52
0.69
34.48
40.0
7.0
3.33
4.67
34.84
0.32
45.48
0.65
2.9
23.45
0.34
0.69
2.41
1.75
13.64
23.42
27.27
35.77
4.01
57.66
N/A
N/A
39.68
2.26
60.97
N/A
N/A
32.67
1.33
51.33
N/A
N/A
30.0
42.26
N/A
N/A
22.90
28.39
N/A
N/A
19.93
43.92
N/A
N/A
Tbl. 5.4. Percentage of time each monitored zone (Suraj12months, raw data series) is thermally uncomfortable
for each month of the year.
Graph 5.2 shows that the outdoor temperature is very rarely comfortable in Jaisalmer, being
too cool in the winter and too warm in the summer. October, November and the end of
February / beginning of March are the most comfortable times of the year. Although table
5.4 shows January as having the highest occurrence of thermally uncomfortable outdoor
temperatures, graph 5.2 indicates that during May and the first half of June day time
temperatures are often in the 'much too warm' (7) band. Whilst the inhabitants of Jaisalmer
often experience discomfort in the winter months, it is the hot, dry summer months of May
and June that are the most thermally stressful time of year. This is reflected in the results of
table 5.4 which show that all monitored zones in Hotel Suraj perform well thermally during
the months November through to February.
Both the basement and ground floor back zone perform very well throughout the year. Even
during the hot, dry summer months the recorded temperatures in both zones remain, on the
86
whole, thermally comfortable. It seems ironic that, in the harsh climate of Jaisalmer, where
exposed mass is so effective in keeping indoor temperatures within the bounds of human
thermal comfort, basements are never inhabited, but rather kept as secure storage spaces for
precious belongings. This is in contrast to the findings of Prasad (1988) in Old Delhi, where,
during the hot season, inhabitants of the havelis often used the basements for a siesta midafternoon. In addition Roaf's research (1989) on the traditional houses of Yazd, showed that
the basements in this area were often ventilated, and consequently used for sleeping during
extremely hot periods.
Although the first floor back zone appears, in table 5.4, to perform considerably worse than
outdoors during the summer, particularly during the high stress months of April, May and
June, graph 5.5 shows that this zone does moderate the summer climate to some extent.
Whilst the temperature outdoors is often 'much too warm' (7) the temperature recorded in the
first floor zone very rarely rises above 'too warm' (6). However, the results do indicate that
both the first and second floor zones should be considered as winter rooms.
5.9. Conclusions
Results from this chapter indicate that:
The comfort temperature for the indigenous population of Jaisalmer, at the end of
March/beginning of April, is in the region of 30C.
This is slightly higher than accepted adaptive model comfort equations would predict.
Possible contributory factors are
a) The very low humidity levels characteristic of the region.
b) The survey was only conducted during the warmer periods of the day.
Of the comfort equations examined, Humphrey's equation proved more representative of

the population of Jaisalmer.
The basement and ground floor back zones in Hotel Suraj perform very well throughout
the year.
The first and second floor zones of Hotel Suraj perform well during the winter months.
However, the first floor zone performs poorly during the high stress months of April,
May and June. There is no second floor zone data available for this season.
This chapter detailed the predicted thermal comfort levels in the monitored zones of Hotel
Suraj over a twelve month period. The results again show that each zone performs
87
differently at different times of the year, some being more comfortable in the cold season
and others in the hot.
Although the author has put considerable effort into the visual presentation of data in both
this and the previous chapter, there still remain some problems with displaying large
amounts of data, particularly over long periods of time. Chapter 6 describes two programs
developed by the author for the visualisation of thermal data in terms of temperature and
predicted thermal comfort sensation. When used with the data discussed in chapter 4 these
programs allow the user to examine temperatures or predicted thermal sensation in any
monitored zone, at any time of day during the monitoring period.
88
ASHRAE, INC. (1966). ASHRAE Standard 55-56, Thermal Comfort Conditions, New York.
BEDFORD, T. (1936). The warmth factor in comfort at work: a physiological study of heating
and ventilation. Industrial Health Research Board Report No. 76. HMSO London.
BERGER, X. (1988). The pumping effect of clothing. Int. J. Ambient Energy, Vol. 9(1).
CLARK, R. P., GOFF, M. R AND MULLAN, B. J. (1977). Skin temperatures during sunbathing
and some observations on the effect of hot and cold drinks on these temperatures. J.
Physiol, London. Vol. 267, pp 8-9.
COOK, J. (1981). Cooling as the Absence of heat: Strategies for the prevention of thermal
gain. In: Proceeds of the International Passive and Hybrid Cooling Conference. Pp.
612-621. American section of the International Solar Energy Society, Inc.
FANGER, P. O. (1970). Thermal Comfort. Danish Techical Press, Copenhagen.
GIVONI, B, (1969). Man, Climate and Architecture. Elsevier publishing company ltd.
GRIFFITHS, I. (1990). Thermal comfort studies in buildings with passive solar features; field
studies: report to the Commission of the European Community, ENS35 090 UK.
HESCHONG, L. (1990). Thermal delight in Architecture. The MIT Press, London.
HUMPHREYS, M. A. (1976). Field studies of thermal comfort compared and applied.
Building Services Engineer (JIHVE). Vol. 41, pp. 191-202
HUMPHREYS, M. A. (1978). Outdoor temperatures and comfort indoors. Building research
and practice, Vol. 6(2), pp. 92-105.
HUMPHREYS, M. A. (1992). Thermal comfort requirements, climate and energy. World
Renewable Energy Congress, Reading, England.
HUMPHREYS, M. A. (1993). An adaptive model for thermal comfort. Notes for research
seminar, Oxford Brookes University.
KRISHAN, A., AGNIHOTRI, M. R., AND JAIN, K. (1995). Climatically responsive energy
efficient architecture: A design handbook. Volume II, Data base - Indian context,
SPA, New Delhi.
MCINTYRE, D. A. (1980). Indoor Climate. Applied Science Publishers Ltd., London.
NICOL, J. F. (1992). Time and thermal comfort: Proc Eorld Renewable Energy Congress,
Reading.
NICOL, J. F., AND HUMPHREYS M. A. (1972). Thermal comfort as part of a self-regulating
system. In: conference proceedings, Thermal Comfort and Moderate Heat Stress.
HMSO, London.
NICOL, J. F., HUMPHREYS, M. A. AND RAJA, I. A. (1995). Developing indoor temperature
standards for naturally ventilated buildings. CIBSE National Conference.
NICOL, F., JAMY, G. AND SYKES, O. (1994). A survey of thermal comfort in Pakistan,
toward new indoor temperature standards, Oxford Brookes University.
89
NICOL, F., RAJA, I. A., ALLAUDIN, A. AND JAMY, G. N. (IN PRESS). Climatic variations in
comfortable temperatures: The Pakistan projects.
OLYGAY, V. and OLYGAY, A. (1992). Design with climate. Princetown University Press.
PRASAD, S. (1988). The havelis of north India. The urban courtyard house. Unpublished PhD
thesis. Royal College of Art, London.
ROAF, S.C. (1989). The wind catchers of Yazd.PhD thesis, Depatment of Architecture,
ROAF, S. C. (1992). The wind catchers of the Middle East. Menas Press.
SAMUELOFF, S, (1980). Physiological adjustment under arid-zone climatic stress: lessons for
the planners. In: Housing in Arid Lands (G. Golany, ed.), pp235-248.Architectural
Press.
YAGLOU, C. P. (1927). The comfort zone for men at rest and stripped to the waist. ASHVE
Trans., Vol. 33, pp165-179
90
CHAPTER 6.
PRESENTATION OF RESULTS
This chapter discusses the problems associated with presenting large amounts
of data, particularly if collected in many locations over long periods of time.
The concept of temperatures represented as colours is explored and a
temperature spectrum is proposed. A stand-alone computer program, designed
to animate temperature data over time is developed and discussed. These
principles are extended to develop a second stand-alone computer program
designed to predict and present thermal comfort sensation levels.
6.1. Introduction
"findings, if potentially useful to the designer are often communicated in a language
she finds difficult to understand" Sir Hugh Casson, 1976.
During the course of this research large amounts of temperature data was collected in many
different locations over long periods of time. Whilst graphical representations of data are, in
many instances, appropriate and informative, it was felt, in the following circumstances,
there was a need to explore different methods of presentation:
Where data sets have many recording locations. It can be difficult to distinguish one data
line from another in a graph.
Where data is recorded over a long period of time. A graph will illustrate the long-term
trends, but much of the shorter-term, daily variation is lost.
The following sections describe the evolution of two presentation tools developed in order to
address these shortcomings.
6.2. Temperatures represented as colours

The sensation of warmth is not originally perceived by the eye, but even so we find it natural
to talk of warm and cold colours. Although the association of colour with thermal sensation
is open to individual interpretation, research has shown that certain general reactions are
common (Luscher, 1971; Sivik, 1974). Heschong (1990) suggests this originates from the
way we perceive temperature. Our thermal sense is a separate sense located in specialised
nerve endings. These thermal nerve endings are heat-flow sensors - they monitor how
quickly our bodies are losing or gaining heat, and thus judge how much colder or warmer
than body temperature an object is. Our thermal sensors therefore cannot perceive
temperature at a distance. Instead we have learnt to rely on other senses to give us advance
clues. Thus reds and browns remind us of the light from a warm fire, and blues feel cold, like
water and ice. These reflex associations were used to develop colour spectrums that would
immediately give an idea of relative temperature.
3.3. Initial temperature spectrum

The initial temperature spectrum, shown in illustration 6.1, had blues representing colder
temperatures, reds representing warmer temperatures and yellow in the centre as the neutral
colour.
92
Ill. 6.1. Initial temperature spectrum.
This colour set was used to display selected data from

the Summer95.xls temperature series, as part of an
exhibition of the author's research, in the Herron Gallery
at the University of East London. The data for three
consecutive hot days was combined to produce data for
an average hot day. The temperature range for the period
was divided into 9 two-degree bands, and a different
colour from the colour set allocated to each band. Each
recorded temperature was thus converted to a colour and
then transferred onto the associated logger zone on the
drawn section of Hotel Suraj. The process was repeated
for each of eight different times of day. Illustration 6.2
shows the colour-coded sections.
This gave an attractive result, but it was thought that the
middle tones were ambiguous. Red is definitely hot and
blue is definitely cold, but is yellow perceived as cooler
or warmer than green?
6.4 Final temperature spectrum.

Research carried out by Sivik in 1974 demonstrated that
orangy-red and greenish blue were most strongly
Ill.5.2. Manually coloured sequence
associated with hot and cold respectively. In order to
for Hotel Suraj.
avoid the previous ambiguity, two such hues, with 100%
saturation, were chosen to represent the hot and cold extremes of the new spectrum. All
middle tones were achieved by adding a varying percentage of whiteness to either of the
polar hues. The resulting spectrum is shown in illustration 6.3.
93
Ill. 6.3. Final temperature spectrum.
Due to the problems associated with colour authenticity and reproduction, the colour
spectrum is also shown, in illustrations 6.4 and 6.5, using the Natural Colour System
developed by Hard (1969). This is a systematic method of describing the relationship
between colours purely from their perceptual qualities. The system describes hue as the
degree of similarity of any colour to the four basic colour attributes: yellowness, redness,
blueness and greeness. Saturation is described in terms of Chromaticness which means the
degree of similarity to the most conceivable colour strength of a certain hue. Brightness has
no directly comparable dimension in the Natural Colour System; instead each colour is
described according to its visual similarity to whiteness and blackness. Black is signified by
the letter s, which represents the Swedish equivalent Svart.
Ill. 6.4. Hot and cold extremes of the

colour spectrum shown using the Natural
Colour System.
Ill.6.5. Middle tones of the colour spectrum shown using

the Natural Colour System.
The more time that was spent discussing and experimenting with this and the previous
colour spectrum, the more it became apparent that the human eye was very sensitive to
colour changes, being able to detect even the slightest alteration in hue. This was essential as
the temperature range for a hot-dry area such as Jaisalmer is very large (over 50C annually),
and so for this approach to be useful, a large number of different colours would be required.
This method of presentation proved especially effective in demonstrating indoor thermal
94
performance as relative to outdoor recorded temperature, but the procedure, by hand, was
very time consuming.
6.5. Automating the process.

With large data sets a method for automating the process was needed. Two options for
achieving this were investigated:
The first was to create a playable video file for each of the data sets. This could be
created in an application such as Director with each playable video being associated to a
specific data set.
The second option was to develop a stand-alone computer program that could import any
data file (preferably in Excel format) and any picture file (preferably in Bitmap format).
Both methods had advantages, but the latter option held most potential. It involved more
effort and time initially, but the principles of the program could later be extended to the
animation of other environmental variables with only minor changes. As the author had very
little programming experience, a language was needed that was relatively easy to learn, but
with good graphical results. Visual Basic, a language evolved from the BASIC (Beginer's
All-Purpose Symbolic Instruction Code) programming language was chosen. Like it's
predecessor, Visual Basic is easy to understand because it is close to English. In addition it
has the advantage of a simple visual method of creating the application's appearance and a
straightforward mechanism for responding to events (such as the click of a mouse button).
6.6. Key Features.

The aim throughout was to keep the program simple and convenient to use, whilst still being
visually striking. Several features were identified as being essential for the success of the
program:
1. The temperature data and building drawing files must be importable in a common and
convenient format. Excel 5 format was chosen for the temperature data. This is the
earliest version of Excel written for Windows 95 and all later versions have the capacity
to save in this format. Bitmap format was chosen for the picture files. This allows even
95
the most basic images created in Paint to be used in the program, as well as being
convenient for screen capture images and scanners.
2. The end user must have the facility to set the temperature range. By default this will be
set to suit Jaisalmer's climate, but if the program is to facilitate the use of other data from
other climatic zones it must allow the user to input alternative minimum and maximum
values.
3. The program must be able to cope with data sets of different sizes, not only different
numbers of columns, but also different lengths of file.
4. The linking of a data column to an area on the picture file must be simple and visual.
The end product, named Thermamator, is a distributable application contained on 3 HighDensity diskettes or CD for use with Windows 95 or Windows NT. It is installed using
Microsoft's Setup Wizard.
6.7. Thermamator. an overview.

Thermamator is a thermal-data presentation tool, designed as an alternative to the graph. It
enables the user to identify any number of monitored zones on a pictorial representation of a
building, or studied area, and link these zones to columns of data stored in an Excel
spreadsheet. Each cell, in each row of data in the spreadsheet is, in turn, converted into an
appropriate colour (see section 6.4). The colour is automatically applied to the linked zone
on the picture. The program moves through the data set, row by row, enabling instant and
effortless visual comparisons of the thermal characteristics of different building zones, and
their variation over time.
6.8. Instructions for use.

6.8.1. Preparation of input files.
Temperature data files. These must be imported in Excel v.5 format. Time/Date
information must be held in column A, with column headings in row 1, as shown in
Illustration 6.6. These column headings will be used to prompt the user to identify each
temperature zone on the picture file.
96
Ill.6.6. Temperature data layout - time in column 1, column headings in row A.
Picture files. These must be imported in bitmap format. The colouring of the zones is
achieved with a flood-fill method. By double clicking with the left mouse button the user
identifies a x, y co-ordinate. The program then flood-fills from this point until a black
boundary is reached. It is therefore necessary to surround each data logger zone with a black
boundary. The simplest method of achieving this is on a high magnification in an
application such as Microsoft Paint.
6.8.2. Loading picture files.

Select Load BMP File from the File menu. A dialogue box will appear allowing you to move
through the directory tree and select any available BMP file (illustration 6.7).
Ill. 6.7. Importing picture files - select Load BMP File from the File menu
97
6.8.3. Loading temperature data files.

Select Load Txt File from the File menu. A dialogue box will appear allowing you to move
through the directory tree and select any available Excel file. A dialogue box will appear
asking for the number of temperature channels in the selected data set. This is usually the
total number of columns minus the date/time column.
6.8.4. Linking temperatures to zones.

A message box appears asking for the first temperature zone to be identified with a doubleclick of the left mouse button on the appropriate area of the picture file (illustration 6.8). The
zone is then flood-filled with grey. If the selection is correct click Accept and the message
box will once again appear asking for the identification of the second temperature zone. If
however the area coloured grey is not correct select Cancel and the message box will
reappear asking once again for the identification of the first temperature zone.
Ill.6.8. Linking temperatures to zones
6.8.5. Changing the temperature range.

By default the temperature range is set from 9C to 47C. Forty colours are pre-programmed
into the application, each one indicating a temperature increment of 1C. To change the
temperature range select Change Temp. Range from the Options menu. A dialogue box
appears allowing the user to input alternative minimum and maximum temperature values.
98
The programme automatically adjusts the temperature increment for each colour ((Max
Temp - Min temp)/38)).
6.8.6. Animating the data.

Once all temperature zones are identified the Play button is enabled. Click on this and the
programme runs from the beginning to the end of your data set at a default interval of 0.25
seconds (illustration 6.9). To change this interval select Speed from the Options menu and
tick the preferred setting. To move manually through the data set use the Rwd and Fwd
buttons.
Ill.6.9. Animating the data.
6.8.7. Quitting the application.

The animation is halted using the Stop button, and selecting Exit form the File menu
terminates the application.
6.9. Further developments.

In the same way that temperatures can be represented as colours, so can the feeling of
warmth and coolness. The principles of Thermamator have been extended to the animation
of comfort level, based on Humphreys' comfort temperature equation (1978), which showed
99
comfort temperature indoors to be related to the temperature outdoors in the following

manner:1
Tc = 12.1 + 0.53 To ..equation 6.1

Where:
Tc = comfort temperature indoors
To = monthly mean outdoor temperature
Applying the same principles used for temperature/colour association in Thermamator, the
seven divisions of the Bedford Scale of thermal sensation (Bedford, 1936)2 were each
assigned an appropriate colour. Applying these colours to Humphreys' equation (equation
6.1) and assuming each thermal sensation division spans 3C3 gives illustration 6.10.
Ill.6.10.Thermal sensation spectrum.
More recent research (Nicol et al., 1995) has shown that the relationship between comfort
temperature indoors and outdoor temperature is improved by using the running mean4 of the
daily average outdoor temperature, instead of the monthly mean outdoor temperature. The
equation used for calculating the running mean of a series is:
ei = W Yi + (1-W) e(i-1)
1
The theory of thermal comfort prediction is discussed in depth in chapter 5, and so only a brief outline is
given here.
2
See chapter 5, section 5.4.
3
See chapter 5, section 5.7.2.
4
See chapter 8, section 8.5, for a detailed description of running mean
100
where:
ei
e(i-1)
Yi
W
= value of the exponentially smoothed series already computed in time period i

= value of the exponentially smoothed series already computed in time period (i-1)
= observed value of the time series in period i.
= subjectively assigned weight or smoothing coefficient (where 0<W<1).
This research also showed that the relationship was most improved when the running mean
series (calculated from daily average values, giving a time interval of 24 hours) was assigned
a weight of 0.2.5
6.10. Comfator..an overview.

Comfator is a thermal sensation calculation and presentation tool, based on the principles of
Thermamator. Like Thermamator, it enables the user to identify any number of monitored
zones on an image of a building or studied area, and link each of these zones to an individual
column of previously recorded data stored in an Excel spreadsheet. The program calculates
the predicted comfort level for each recorded temperature value using Humphreys' comfort
equation (1978). Each comfort level value is then converted into an appropriate colour (see
illustration 6.10) and automatically applied to the linked zone on the image. The program
moves, row by row, automatically or manually, depending on user preference, from the
beginning to the end of the data set, thus enabling instant visual comparisons of the predicted
comfort levels in different building zones, and their variation over time.
6.11. Instructions for use

6.11.1. Preparation of input files
Temperature data files. These must be imported in Excel v.5 format. Time/Date
information must be held in column A, and recorded outdoor temperature data must be held
in column B. Logger locations/column identities must be held in row 1. Illustration 6.11
shows the necessary input format.
See chapter 5, section 5.8.

101
Ill. 6.11. Comfator data input format of Excel spreadsheet.
Picture files. These must be imported in bitmap format with each logger zone enclosed by a
black boundary, as for Thermamator.
6.11.2. Loading files and linking data columns to building zones.

These steps are as for Thermamator. Once the user has input the number of temperature
channels, Comfator prompts for the time interval between successive data readings. This
must be entered in hours. For example, a logger recording interval of 2 hours and 30 minutes
would be entered as 2.5 hours. This interval is used to calculate the correct weight for the
exponential smoothing of the outdoor temperature.
6.11.3. Animating the data

Animating the data set is as for Thermamator. However, as the calculation of the comfort
temperature is dependent on an exponential smoothing of the outdoor temperature, the user
can only move forwards through the data set. The RWD button has therefor been disabled.
6.12. Discussion
Initial feedback for both Thermamator and Comfator has been good. Thermamator has been
used by MSc students in both the University of East London and the Architectural
Association schools of architecture, and seems particularly useful for the visualisation of
102
surface temperature data. Research staff in the Solar Energy department of the University of
Siegan have also used Thermamator, in particular for the visualisation of thermal data of
climatic regions, and the thermal behaviour of an earth heat exchanger.
Although both programs are very simple to use, the preparation of the input files does
require some effort, particularly the zoning of the picture file. Almost all problems
encountered when using the programs are attributable to either incorrect data format, or
incorrect zone borders on picture files.
There are several shortcomings in the programming of the two packages. They are as
follows:
Problems with data file handling. Both programs currently generate an error when the
last row of data in the Excel spreadsheet is reached. The error automatically terminates
the program, making it impossible to run through the data set a second time, without
repeating the whole procedure of starting up the program, loading an image file, linking
each column to a zone, etc. The author was also unable to set up a procedure for loading
a second data file. Both programs have to be restarted in order to load each different data
set.
Problems with printing. There are no in-built printing features in either Thermamator
or Comfator. In order to print an image generated by either of the programs the user must
perform a screen dump6 whilst the image is on the screen. This copies the entire contents
of the screen onto the computer systems clipboard, from where all other applications can
access it. The user must then create a new file in any image manipulation program
(Photoshop, for example) and paste the saved image into the file. From here the image
can be trimmed, printed, saved etc.
Further features. The author had hoped to include a search facility in Thermamator
linked to the date/time data column. This would allow the user to jump to any specified
date or time, rather than view the data in order. In Comfator it was hoped to offer the
user the choice of other accepted comfort equations in addition to Humphreys.
The basic idea behind both Thermamator and Comfator - that of linking columns of data
directly to zones on a building image - was conceived in order to stimulate interest in indoor
thermal environment amongst students of Architecture. However, it is possible that both
programs could be used by architects as presentation / selling tools to clients, particularly
over the Internet, as playable files embedded in web sites. The monitoring of the thermal
Accomplished by pressing the 'Print Screen' key on the keyboard

103
performance of existing buildings is now widespread, and it is essential that the valuable
information gathered from this research be conveyed appropriately.
6.13. Closure
This chapter has detailed the development of software for the presentation of thermal
building data. Both programs are available on the CD-ROM enclosed at the back of this
thesis. To install Thermamator, first insert the CD-ROM into your CD-ROM drive. Select
Run from the Start menu and type 4D:\software\thermamator\setup.exe in the Open:
window, and follow the on screen instructions. The procedure is the same for the installation
of Comfator, typing 4D:\software\Comfator\setup.exe in the Open: window
The visual basic code for Thermamator and Comfator can be found in appendix B and C
respectively. Prepared data sets, in the correct format, can be found in the Software\data
directory. Prepared drawings of Hotel Suraj can be found in the Software\images directory.
For details of logger locations and file sizes see the relevant section in chapter 4.
Presuming D is the chosen letter for your CD-ROM drive

104
CASSON SIR H. (1976). Prerface in: Colour for Architecture. Ed. Porter, T. and Mikellides,
B. Studio Vista.
HARD A. (1969). The NCS Colour Order and Scaling System. Swedish Colour Centre,
Stockholm.
HESCHONG L. (1990). Thermal Delight in Architecture, MIT Press, Massachusetts.
HUMPHREYS, M. A. (1978). "Outdoor temperatures and comfort indoors." Building research
BEDFORD T. (1936)."The warmth factor in comfort at work." MRC Industrial Health Board
Report. No 76. HMSO.
LUSCHER M, (1971). The Luscher Colour Test, Jonathon Cape, London.
NORTON H., DAVIS H., DAVIS P. (1995). Peter Norton's Guide to Visual Basic for Windows
95, Sam's Publishing, Indianapolis.
SIVIK L. (1974). Colour Meaning and Perceptual colour Dimensions: A Study of exterior
Colours, Goteborg Psychology Reports No. 11, University of Goteborg.
105
CHAPTER 7.
INDOOR TEMPERATURE TRENDS
This chapter examines the relationship between indoor and outdoor recorded
temperature in the Suraj12months.xls temperature series. Through the analysis
of average and range temperature values, an attempt has been made to develop
indoor temperature prediction equations for each of the four monitored zones,
based solely on outdoor temperature.
7.1. Introduction
Monitoring of the thermal performance of existing buildings is now widespread (Roaf, 1980,
Gupta, 1984, Prasad, 1988, Ali, 1998), and information on the prediction of indoor comfort
temperature is plentiful (Fanger, 1970, Humphreys, 1978, Auliciems et al., 1986, Sharma et
al., 1986). However, it is a constant source of frustration that knowledge of how to design a
non-air-conditioned building that achieves these temperatures is not available.
Recent research (Givoni, in press) has shown that the most important climatic parameter for
the prediction of daily average indoor temperature in hot climates is the daily average
outdoor temperature, and that adding daily average solar energy data does not greatly add to
the prediction accuracy. This conclusion is of particular interest in developing countries,
such as India, where climatic data may often be limited to temperatures. However, the daily
average temperature is not the main concern when predicting indoor thermal comfort in hot,
dry climates. The daily maximum indoor temperature (and hence the daily range),
particularly in the summer months, is often the most critical.
This chapter takes a closer look at the Suraj12months.xls temperature series, in particular the
relationship between indoor and outdoor recorded temperatures and indoor and outdoor
recorded daily range. The following section attempts to develop indoor temperature
prediction equations for each of the four monitored zones in Hotel Suraj, based solely on
outdoor temperature. Subsequent sections investigate the relationship between indoor and
outdoor recorded range.
7.2. Indoor/outdoor temperature

7.2.1. Method
a) Actual indoor recorded temperature in each of the four monitored zones was plotted
against actual outdoor recorded temperature for the Suraj12months series.
b) A linear trendline1 was fitted, using Microsoft Excel, such that:
Excel uses the least-squares method to fit a straight line such that the calculated value for the two coefficients
(m and c) result in the sum of squared differences between each observed value in the data and each predicted
value along the trendline being minimized.
107
y = mx + c..equation 7.1
where:
y = predicted indoor temperature
m = slope
x = outdoor temperature
c = intercept
c) Excel also outputs a value for r2 - the coefficient of determination. This measures the
proportion of variation that is explained by the independent variable, in this case outdoor
temperature, in the trendline. Some researchers suggest that an adjusted r2 should be
computed to reflect the sample size (Berenson et al., 1989). The adjusted r2 is denoted as:
r2adj = 1- ((1- r2) (n-1) (n-2)) .equation 7.2

where:
n = sample size
All coefficients of determination shown in this chapter have been adjusted in this way.
d) Steps b) and c) were repeated for calculated daily average and monthly average
temperatures.
7.2.2. Results
Graphs 7.1, 7.2 and 7.3 show the results for actual, daily average and monthly average
temperatures respectively.
108
45
40
indoor temperature
35
30
25
20
15
10
5
10
15
20
25
30
35
40
45
outdoor temperature
Graph 7.1. All data
40
indoor temp
35
30
25
20
15
10
15
20
25
30
35
40
outdoor temp
Graph 7.2. Daily average data
40
35
indoor
30
25
20
15
15
17
19
21
23
25
27
29
31
33
35
outdoor
Graph 7.3. Monthly average data
109
Table 7.1 gives a summary of the results from these graphs. The slope, intercept and adjusted
coefficient of determination (r2adj) for each indoor zone's trendline are shown.
BASEMENT
slope
intercept
r2adj
GROUND
slope
intercept
r2adj
FIRST
slope
intercept
r2adj
SECOND
slope
intercept
r2adj
ALL DATA
DAILY AV.
MONTHLY AV.
0.28
19.87
0.57
0.66
9.87
0.83
0.80
7.72
0.90
0.82
6.74
0.86
0.32
18.77
0.67
0.75
7.51
0.94
0.87
5.67
0.98
0.93
4.44
0.97
0.34
18.10
0.67
0.79
6.45
0.98
0.88
5.54
0.99
0.96
3.70
0.99
Tbl. 7.1. Summary of linear trendline results for indoor temperature plotted against outdoor temperature.
7.2.3. Discussion
It is immediately noticeable that in all zones the adjusted r2 value increase from the left of
the table to the right, with monthly average data giving the highest values. This is to be
expected, to some extent, as the process of averaging removes some variation from the data
set. However, the adjusted r2 value also increase from the bottom to the top of the haveli. On
both the second and first floors over 99% of the average monthly variation in indoor
temperature can be explained by the average monthly variation in the outdoor temperature.
The basement, however, gives relatively low values for r2 for all of the data time-scales, with
monthly average data giving a maximum value of 66%. A possible explanation for this
gradation in values for r2 is the coincident gradation in mass. As the available mass increases
from the top to the bottom of the haveli, the dependence of indoor temperature on immediate
outdoor temperature decreases. It is feasible that the available mass in the basement zone is
sufficient to 'shift' the indoor temperature relative to the outdoor temperature by as much as a
month. Graph 4.32 shows that the basement indoor minimum temperature does infact occur
around 28 days later than the outdoor minimum temperature. This suggests that the
Chapter 4, section 4.4.2.2, page . Graph of temperature data for Suraj-W95-96 series
110
temperature in the basement zone is more dependent on the previous month's outdoor
temperature than immediate outdoor temperature.
Graph 7.4 shows the monthly average basement temperature plotted against the previous
monthly average outdoor temperature.
32
30
indoor temperature
28
26
24
22
20
15
17
19
21
23
25
27
29
31
33
35
outdoor temperature
Graph 7.4. Recorded indoor basement temperature plotted against the previous months outdoor temperature
(monthly average data).
Table 7.2 gives a summary of the results from this graph. The slope, intercept and adjusted
coefficient of determination (r2adj) for the basement's trendline are shown.
MONTHLY AV.
(with tL)
BASEMENT
slope
intercept
r2adj
0.40
16.51
0.97
Tbl. 7.2. Summary of linear trendline results for basement monthly average indoor temperature plotted against
the monthly average outdoor temperature of one month previous.
The results in table 7.2 show a dramatic increase in value for the adjusted coefficient of
determination, from 66% with the monthly average data to 97% with the time lagged
monthly average data. This supports the theory that, from the top to the bottom of the haveli,
the decrease in dependence of the indoor temperature on the immediate outdoor temperature,
is, to some extent, due to the increase in available mass. This increase in mass may also
111
explain, to some extent, why the actual recorded data (at 2.4 hour intervals) gives the lowest
values for r2adj. The available mass, even in the two upper floor zones, is relatively high, and
all zones are likely to have a time lag between indoor and outdoor maximum, and minimum,
temperatures. It is likely that the indoor temperature of the top two floors is responding to
the outdoor temperature several hours earlier, and the semi-buried ground floor zone may be
responding to outdoor temperatures occurring days previously, hence the monthly average
data gives the strongest correlation
In addition to the considerable variation in values for r2adj between raw, daily average and
monthly average data, the values for the trendline slopes and intercepts also vary. In general,
the slopes become steeper and the intercepts smaller as the time interval increases. These
trends result in higher predicted indoor temperatures at the hottest time of year and lower
predicted indoor temperatures at the coldest time of year when using the trendlines derived
from monthly average data. Whilst one would expect the r2adj of the trendline to increase
when using average data, the slope should remain constant for each logger zone. This
increase in trendline gradient, from the left to the right of table 6.1, suggests that some 'error'
is being removed form the relationship between indoor and outdoor temperature by the
process of averaging. The mathematical explanation for such 'dilution' of the best fit,
resulting in the lowering of the slope below its true value is discussed in detail in Humphreys
et al. (2000). This reinforces the belief that indoor temperature is responding to previously
recorded outdoor temperature, with the time difference between the two increasing from the
top to the bottom of the haveli.
Predicted indoor temperatures were calculated for each of the four zones in Hotel Suraj,
using the trendline equations derived from each type of data, for the outdoor temperatures of
9C, 27C and 42C (corresponding to minimum, average and maximum outdoor
temperature respectively). Table 7.3 shows these results.
These results confirm that both of the linear trendlines derived from daily average and
monthly average data give very similar results, with a maximum discrepancy of 0.7C on the
second floor zone and a minimum discrepancy of 0.1C on the first floor zone. The monthly
time lagged data gives a maximum temperature 0.9C above, and a minimum temperature
1.1C below, that of the standard monthly average data. The raw data gives different results
in all zones at both extremes of the outdoor temperature range.
112
BASEMENT
GROUND
FIRST
SECOND
Tout
All data
Daily Av.
Monthly Av. M.A.(with tL)
9C
22.4
21.6
21.2
20.1
27C
27.3
27.3
27.3
27.3
42C
31.4
32.0
32.4
33.3
9C
15.8
14.3
13.6
27C
27.8
27.8
27.8
42C
37.7
39.0
39.6
9C
14.9
13.5
13.4
27C
29.2
29.2
29.2
42C
41.1
42.2
42.4
9C
14.2
12.8
12.4
27C
29.0
29.5
29.7
42C
41.4
43.5
44.2
Tbl. 7.3. Predicted indoor temperature results, using derived linear trendlines, for outdoor minimum, average
and maximum temperatures.
7.3. Indoor/outdoor range.

The following sections investigate the relationship between indoor and outdoor recorded
daily range for each of the monitored zones in Hotel Suraj over a twelve-month period.
7.3.1. Method
a) The daily range was calculated for each of the five monitored zones in the
Suraj12months temperature series, using the formula:
Range = Tlargest - Tsmallest .equation 7.3.

b) Daily indoor range was plotted against daily outdoor range for each of the four indoor
zones.
c) A linear trendline was fitted using Microsoft Excel.
d) The adjusted coefficient of determination was calculated using equation 7.2.
113
e) The monthly average of the daily averages was calculated3 and steps b) to d) repeated.
7.3.2 Results
Graphs 7.5 and 7.6 show the results for the daily and monthly-average-of-daily temperature
ranges respectively.
6
KEY
range indoors
Basement
Ground
First
Second
0
0
10
12
range outdoors
Graph 7.5.Indoor daily range plotted against outdoor daily range - Suraj12months temperature series
4.5
KEY
3.5
indoors
Basement
Ground
First
Second
2.5
2
1.5
1
0.5
0
5
5.5
6.5
7.5
8.5
outdoors
Graph 7.6. Monthly average of daily indoor range plotted against monthly average of daily outdoor range Suraj12months series.
This is not the same as the monthly range - i.e. the maximum temperature occurring in a calendar month
minus the minimum temperature occurring in a calendar month. It is the average of all the daily ranges
occurring in a single calendar month.
114
Table 7.4 gives a summary of results from these graphs. The slope, intercept and adjusted
coefficient of determination for each indoor zone's trendline are shown.
DAILY AV. MONTH AV.

BASEMENT
slope
intercept
Adj.R2
GROUND
slope
intercept
Adj.R2
FIRST
slope
intercept
Adj.R2
SECOND
slope
intercept
Adj.R2
0.01
0.06
0.00
0.16
-0.08
0.16
0.15
2.01
0.16
0.17
2.58
0.07
0.01
0.03
0.00
0.20
-0.34
0.50
0.06
2.58
-0.03
-0.08
4.11
-0.20
Tbl.7.4. Summary of linear trendline results for indoor temperature range plotted against outdoor temperature
range.
7.3.3. Discussion
The results for adjusted r2 are all surprisingly low. With the exception of the monthly data
results for the ground floor zone, there seems to be very little relationship between indoor
and outdoor range. The trendlines derived from daily range data for the ground and first floor
zones both show that just under 16% of the variation in daily temperature range indoors is
accounted for by the variation in daily range outdoors. The adjusted r2 value for the ground
floor zone increases to just under 50% when monthly averages of the daily ranges are used.
However, all other results show that the variation in outdoor range has very little bearing on
the variation in indoor range. A plot of the daily ranges over the twelve month period, shown
in graph 7.7. reinforces this conclusion. Although the outdoor daily range shows
considerable variation over this twelve month period, and increases dramatically during the
hot summer months of May and June, the daily range for each indoor zone remains relatively
stable throughout the year.
Table 7.5 shows the average and standard deviation of the daily ranges over this twelve
month period. These results once again show that whilst the outdoor daily range varies
substantially over the year the indoor daily ranges are comparatively constant.
115
12
10
range
0
8-Oct
5-Nov
3-Dec
31-Dec
28-Jan
25-Feb
24-Mar
21-Apr
19-May
16-Jun
14-Jul
11-Aug
8-Sep
6-Oct
date
Graph 7.7. Daily range for Suraj12month temperature series.
OUTDOORS
BASEMENT
GROUND FLOOR
FIRST FLOOR
SECOND FLOOR4
AVERAGE
STANDARD DEVIATION
6.72
1.70
0.10
0.17
0.97
0.68
3.01
0.62
3.62
0.71
Tbl. 7.5. Calculated average and standard deviation of the daily ranges for each monitored zone in the
Suraj12months temperature series.
7.4. Conclusion
This chapter has shown that:
The dependence of indoor temperature on immediate outdoor temperature decreases

inversely with available mass, from the top to the bottom of Hotel Suraj.
In all but the basement the daily average outdoor temperature explains between 94 and
98% of the variation in daily average indoor temperature.
Only six months data available for this logger location

116
The temperature in the basement shows the strongest relationship with monthly average
outdoor temperature of one month previous.
The indoor daily range in each zone remains relatively stable, throughout the year,
despite large fluctuations in outdoor daily range.
The best fit daily average indoor temperature prediction equations and the average
daily range for the ground, first and second floor zones are as follows:
TGr = 0.66To + 9.87
with a daily range of 0.97C
T1st = 0.8To + 7.72
with a daily range of 3.0C
T2nd = 0.82To + 6.74
with a daily range of 3.62C5
Where:
TGr
T1st
T2nd
To
= Daily average temperature of ground floor zone

= Daily average temperature of first floor zone
= Daily average temperature of second floor zone
= Daily average temperature outdoors
The best fit monthly average temperature prediction equation for the basement is as
follows:
TMb = 0.4T(M-1)o + 16.5
with no daily variation
Where:
TMb = Monthly average basement temperature in time period M
T(M-1)o = Monthly average outdoor temperature in time period M-1
The results from this chapter indicate that the available mass in Hotel Suraj plays an integral
role in determining the temperature indoors. In the basement the effects of the mass is such
5
Only six months data available for this location

117
that indoor peak temperature is shifted by as much as one month, with outdoor daily
variations in temperature having little or no effect on the climate indoors. Chapter 8 takes a
closer look at the effects of mass on the indoor temperature in Hotel Suraj, and attempts to
develop methodologies for accurately quantifying the time difference between indoor and
outdoor temperature for each monitored zone.
118
ALI, Z. F. (1998). Le Corbusier's Chandigarh from an environmental point of view.
Proceedings of PLEA '98, Lisbon, pp 183-186. James and James Science Publishers
Ltd.
AULICIEMS, A. & DEDEAR, R. (1986). Air conditioning in Australia I; Human thermal
factors. Arch. Science Review, volume 29, pp67-75.
BERENSON, M. L. AND LEVINE, D.M. (1989). Basic business statistics, concepts and
applications. Prentice-Hall Internarional (UK) Limited, London.
FANGER, P. O. (1970). Thermal Comfort. Danish Technical Press.
GIVONI, B. (IN PRESS). Minimum climatic information needed to predict performance of
passive buildings in hot climates.
GUPTA, V. K. (1984). A study of the natural cooling systems of Jaisalmer. Unpublished PhD
HUMPHREYS, M. A. (1978). Outdoor temperatures and comfort indoors. Building Research
and Practice, volume 6, issue 2.
HUMPHREYS, M. A. & NICOL, J. F. (2000, IN PRESS). Effects of measurement and
formulation error on thermal comfort indices in the ASHRAE database of field
studies. AHRAE transaction, volume 106, issue 2.
KRISHAN, A., AGNIHOTRI, M. R., JAIN, K., TEWARI, P., RAJAGOPALAN, M. (1995).
Climatically responsive energy efficient architecture - a design handbook. Volume II:
database - Indian context. CASA, School of Planning and Architecture, New Delhi.
PRASAD, S. (1988). The havelis of north India. Unpublished PhD thesis, Royal College of
Art, London.
ROAF, S. C. (1989). The Windcatchers of Yazd. PhD. Thesis, Department of Architecture,
SHARMA & ALI (1986). Tropical Summer Index - a study of thermal comfort in Indian
subjects. Building and Environment, volume 21, issue 1, pp 11-24.
119
CHAPTER 8.
MASS AS A THERMAL REGULATOR
This chapter starts with an introduction to the science of heat transmission

under non-steady state conditions, including the definition of time lag and
decrement factor. It goes on to detail two methodologies developed by the
author with the intention of determining time lag between indoor and outdoor
monitored temperature. Following the review of both methodologies the
decrement factors of the internal spaces are examined in detail.
8.1 Introduction
Chapter 7 showed that the mass of the havelis was critical in determining indoor
temperatures, and that the basement, in particular, was responding to outdoor temperatures
that occurred up to one month previous. Much is currently made of the value of utilising
building fabric as a thermal buffer to reduce summertime temperature swings. However, as
addressed by Shaw and Treadway (1994), the effects are often considered in a notional
fashion because of the time and degree of specialist knowledge required to comprehensively
model the impact of exposed building mass.
When building elements are subjected to fluctuations in temperature (as in the havelis of
Jaisalmer), thermal resistance usually measured by means of u-values, is not by itself an
accurate indication of the thermal performance of a building element (vanStraaten, 1967).
The heat storing ability of the element also plays a vital role, and is usually expressed in
terms of its specific heat capacity - the heat stored in a unit volume of material per degree of
temperature rise (Goulding et al., 1992). The amount of heat an element is capable of storing
also depends on the density of the material, the more dense the material, the more heat it can
store. In addition, the material must be a good thermal conductor. Materials with high
thermal conductivity accumulate heat relatively quickly, allowing the heat to penetrate
through all the material. These three properties, specific heat capacity, density and
coefficient of conductivity are grouped together to give the diffusivity1 of the element, the
key property affecting heat travelling through a building element when exposed to
fluctuations in temperature.
In addition the location of the thermal mass and the degree to which the mass is exposed are
of importance. It has been demonstrated (Sodha et al. 1992) that increased surface area of the
storage material is more effective in reducing temperature swings than a large thermal
capacity.
It is possible to estimate both time lag and dampening of indoor room temperature by
looking at a graph of indoor and outdoor temperature plotted against time. In fig. 8.1 time
lag (tL) is given by the difference in hours between the occurrence of the peak temperature
outdoors and the corresponding peak temperature indoors. The dampening of the series is
quantified by means of a decrement factor. This is the ratio of the amplitude of internal
temperature wave to the amplitude of external temperature wave.
Specific heat multiplied by density gives heat/m3. The ratio of heat/m3 to conductivity gives diffusivity
(Fordham, private communication)
121
40
TL
35
Time Lag =tL
30
Decrement Factor = AI /AO
temperature
AO
25
AI
20
15
10
5
0
0
10
15
20
25
30
35
40
45
50
t ime
(T)
(t)
t
Fig. 8.1. Time lagLand decrement factor shown on standardised, generated data.
8.2. The problem

Mass affects indoor temperature in two ways:
a. It shifts the maximum and minimum temperatures causing a time lag between indoor and
outdoor temperature
b. It regulates the magnitude of the indoor temperature swings .
It is difficult to predict the time lag between outdoor and indoor temperature.2 It does not
equate to the time lag of any one of the building elements, but is a product of all the
elements, coupled with both the ventilation rate of the space, and energy gain from the sun.
Preliminary analysis of the data collected for this study suggests that the indoor temperature
time lag is not constant from season to season or even from day to day, which further
compounds the problem. In his study of the natural cooling systems in the havelis of
Jaisalmer, Gupta (1984) suggested that at least one of the havelis he studied had a twentyfour hour time lag. Some of the indoor spaces in Hotel Suraj are estimated as having time
lags exceeding twenty four hours. In such cases it is very difficult to confidently estimate the
time lag by graphical means. Sections 8.3 to 8.6 detail two different statistical approaches
adopted in order to develop a method of conclusively calculating time lag from logged
temperature data. Decrement factor of the internal spaces is dealt with in sections 8.7 to 8.9.
There are now many computational methods available to predict the effect of mass on indoor air temperatures
(Balaras, 1996), but all require the user to input data that must be based on assumptions (Milbank, 1974).
122
8.3 Aims of chapter 8
To develop a method of accurately determining the time lag between indoor and outdoor
monitored temperature.
To determine whether time lag varies according to season (winter vs. summer).
To investigate the relationship between time lag and decrement factor.
To ascertain what magnitude of logger interval is appropriate for determining time lag.
8.3.1 Form of chapter 8

This chapter is organised as follows:
Section 8.4 details the first "stepped correlation" technique developed to determine time
lag. The method used is described and the technique is applied to each monitored indoor
zone in each of the five data sets described in chapter 4. The results are discussed.
Section 8.5 details the second "running mean" technique. The technique is again applied
to each monitored indoor zone in each of the five data sets described in chapter 4 and the
results discussed.
Section 8.6 compares the time lag results achieved by both techniques
Section 8.7 calculates the daily decrement factor for each of the four monitored indoor
zones in the Suraj12months data set, and relates the results to the previously calculated
time lag.
Section 8.8 gives a summary of the conclusions drawn from this chapter
8.4 1st approach - Stepped Correlation

The correlation coefficient is a common method of determining the strength of associations
between variables in a data set. The value of the coefficient ranges from -1 for perfect
negative correlation, up to +1 for perfect positive correlation.
In all the recorded data sets the indoor temperature lags behind the outdoor temperature.
That is, the respective maximum temperature occurs later indoors than outdoors. If the
indoor temperature series is theoretically "moved back in time", a point will be reached
where indoor and outdoor peaks and troughs will occur simultaneously. At this point the
correlation coefficient between the two variables will be at a maximum, close to perfect
positive correlation. The shift in time required to achieve the maximum correlation
coefficient should be equal to the time lag between indoor and outdoor temperature.
123
8.4.1 Method
The following method was developed, and used to estimate the indoor temperature time lag
in relation to outdoor temperature, by means of stepped correlations.
1.
2.
3.
4.
5.
Each indoor air temperature series was correlated with the outdoor temperature series
for the same period (ts to tf).
Each indoor air temperature series was moved back one 'step' in time, the magnitude
of the step being determined by the time interval between successive logging.
Each indoor air temperature series for the period ts+1 to tf was correlated with the
outdoor temperature for the period (ts to tf-1).
Steps 2 and 3 repeated until the strength of correlation reaches a maximum before
starting to fall off.
The time lag is calculated by multiplying the number of steps taken to reach
maximum correlation, by the time interval between successive logging3.
In order to speed the process up a macro was written to operate within an Excel spreadsheet
(see Appendix D). The macro automatically steps the indoor temperature backward in time,
performing a correlation each turn. The number of steps performed on each data set was
determined by the size of the data file, combined with knowledge of the likely range of time
lags, estimated from graphical printouts of the data.
8.4.2 Results
The stepped correlation method was applied to four air temperature data sets:
a. Summer95.xls, both for the raw data (36 minutes for each step) and for the calculated
daily average (24 hours for each step)
b. SurajW95-96.xls, both for the raw data (2 hours 24 minutes for each step) and for the
calculated daily average (24 hours for each step)
c. Suraj-S96.xls, both for the raw data (2 hours 24 minutes for each step) and for the
d. 2ndHaveli-S96.xls, both for the raw data (2 hours 24 minutes for each step) and for the
The time lag was more accurately estimated by drawing a graph of the maximum three consecutive
correlations against the number of steps. A smoothed curve was drawn to join the points on the graph and the
peak correlation value projected onto the x-axis to give a corresponding step value.
124
e. In addition data sets b and c above were combined and analysed as twelve months data
(raw data and daily average)
8.4.2a Summer95.xls
Results of the stepped correlation performed on the raw and daily average data are shown in
graphs 8.1 and 8.2 respectively. 199 steps were performed on the raw data (equating to a
time lag of just less than 5 days) and 6 steps were performed on the daily average data
(equating to a time lag of 6 days). Tables 8.1 and 8.2 show the strongest three consecutive
correlations along with the number of steps performed to achieve these results.
0.8
0.8
0.6
strength of correlation
0.6
0.4
0.2
0
0
40
80
120
160
200
-0.2
St
re 0.4
ng
th
of 0.2
co
rre 0
lat
io
-0.2
n
-0.4
-0.4
-0.6
-0.6
No of steps (time lag * 36mins)
1st FlMid
GrFlCourt
TopFront
Basement
1st FlBack
GrFlPorch
No. of
steps
Est. tL
No. of steps (time lag*24hrs)
TopMid
TopFlCourt
TopFlBack
Graph 8.1 Summer95.xls, stepped correlation
Max 3
corr.
-0.8
-0.8
GrFlBack
1st FlFront
TopFront
GrFlBack
1st FlBack
GrFlCourt
TopMid
Basement
1st FlMid
GrFlPorch
TopFlBack
TopFlCourt
1st FlFront
Graph 8.2 Summer95.xls(daily av.) stepped correlation
Bsmnt
GrFlBk
1stBck
1stMid
1stFrt
2ndBck
2ndMid
2ndFrt
GrCrt
GrPch
2ndCrt
0.181
0.181
0.181
139
140
141
0.558
0.562
0.555
81
82
83
0.766
0.773
0.767
41
42
43
0.688
0.692
0.682
41
42
43
0.702
0.706
0.697
41
42
43
0.734
0.736
0.726
4
5
6
0.683
0.688
0.683
2
3
4
0.852
0.854
0.840
2
3
4
0.851
0.853
0.837
0
1
2
0.854
0.874
longer
49hrs9
mins
25hrs
16min
25hrs
3mins
25hrs
5mins
2hrs
49min
1hr
52min
1hr
35min
18min
none
none
Tbl. 8.1. Summer95.xls (36 min. data). Summary of results of stepped correlation
125
Max 3
corr.
No. of
steps
Est. tL
Bsmnt
GrFlBk
1stBck
1stMid
1stFrt
2ndBck
2ndMid
2ndFrt
GrCrt
GrPch
2ndCrt
0.169
0.305
0.342
4
5
6
0.113
0.122
0.026
2
3
4
0.642
0.705
0.646
0
1
2
0.480
0.503
0.481
0
1
2
0.753
0.822
0.806
0
1
2
0.756
.730
.572
0
1
2
0.739
0.714
0.605
0
1
2
0.892
0.894
0.732
0
1
2
0.845
0.779
0.671
0
1
2
0.838
0.722
0.576
0
1
2
0.841
0.706
0.552
0
1
2
>
6days
2.6
days
1 day
1 day
1.3
days
0.3
days
0.5
days
Tbl. 8.2. Summer95.xls (36 min. data, daily average). Summary of results of stepped correlation
As expected each of the zones responds differently to the outdoor temperature. As

highlighted in chapter 3 the majority of rooms in havelis are multifunctional, their use
determined by both time of year and time of day. This is discussed as a general cooling
strategy by Balaras (1994)
"One of the most effective ways however, in dealing with the problem of high
temperatures during the day, has also been the behavioural response of people. Use
of different spaces during the day, in search of cooler spaces, or changes in the
building use altogether during the summer, were very common behavioural actions."
And particularly of traditional houses in Baghdad by Fethi and Roaf (1986)
"The choice of living space is determined by the climate in that area at any particular
time of day or year. There exists a wide variety of diurnal and seasonal climatic
gradients horizontally and vertically, so the majority of rooms in a traditional house
are not designed for a specific function, but for a variety of specific internal
environments."
Although there is evidence of both vertical and horizontal gradients in time delay, it is the
vertical gradient that is most pronounced. Horizontally the rooms towards the back of the
haveli (southwest elevation) respond more slowly to outdoor conditions than those at the
front of the haveli (northeast elevation). This is most likely an attribute of the higher mass
content at the back of the haveli, than a product of orientation.
The vertical gradient also produces predictable results, with longer time delays on the ground
floor than higher up the building. However the magnitude of the difference in response time
between floors is more interesting. A look at a graphical printout of the data would suggest
zones on the first and second floors respond in a similar manner to outdoor temperature.
However this analysis shows quite clearly that zones on the first floor are responding to the
previous day's temperature, whilst those on the second floor are responding to the current
days temperature. The ground floor zone is shown to be responding to the outdoor
126
temperature of two days previous. The logging period was not long enough to enable the
determination of the time lag for the basement.
Three outdoor areas were monitored during this logging period. Only one, the ground floor
courtyard, showed any response delay. It was expected that the ground floor porch area in
particular would have a time lag, but although this area is the more massive of the three, it
receives very little direct sunlight and so is not subjected to large swings in radiant
temperature. In addition the air change rate in this area is considerably higher than on the
ground floor of the main courtyard, so aiding heat transfer. This analysis showed there to be
no time lag in this area. The third outdoor zone, the second floor courtyard, has both high air
change rate and a lighter construction and so predictably shows no time lag.
Results from the calculated daily average data (table 8.2) reinforce those of the raw data.
Although not accurate enough to inform precisely what time of day the minimum and
maximum temperatures will occur in each area, they do give an indication of which day's
temperature range the area is responding to.
8.4.2b SurajW95-96.xls
Results of the stepped correlation performed on the raw data and daily average data are
shown in graphs 8.3 and 8.4 respectively. 399 steps were performed on the raw data
(equating to a time lag of just less than 40 days) and 49 steps were performed on the daily
average data (equating to a time lag of 49 days). Tables 8.3 and 8.4 show the strongest three
consecutive correlations along with the number of steps performed to achieve these results4.
1.2
0.8
0.8
0.4
0.2
0
0
50
100
150
200
250
300
350
400
0.6
0.6
0.4
0.2
0
0
-0.2
-0.4
15
20
25
30
35
40
45
50
-0.4
-0.6
-0.6
No. of steps (time lag *24hrs)
No. of steps(time lag * 2hr 24min)
Basement
GrFl-back
1st Fl-back
2ndFl-back
Graph 8.3. SurajW95-96.xls - stepped correlation

correlation.
10
-0.2
Basement
GrndBack
1st Floor
2ndfloor
Graph 8.4. SurajW95-96.xls (daily av.) - stepped
The bsmnt raw data (tbl. 8.3) also showed a high correlation at 176.8 , giving tL = 424.32 hours (~17.6 days)
127
Max 3
corr.
No. of
steps
Est. tL
Bsmnt
GrFlBk
1stBck
2ndBck
0.845
0.845
0.845
121
122
123
0.891
0.893
0.891
23
24
25
0.938
0.965
0.960
10
11
12
0.935
0.969
0.956
0
1
2
293hr
31min
(12.2d)
58hr
5min
(2.4d)
26hr
53min
(1..1d)
3hrs
7min
Tbl. 8.3 SurajW95-96.xls (2hr24min. data).

Summary of results of stepped correlation.
Max 3
corr.
No. of
steps
Est. tL
Bsmnt
GrFlBk
1stBck
2ndBck
0.893
0.893
0.893
22
23
24
0.951
0.958
0.954
1
2
3
0.982
0.984
0.955
0
1
2
0.983
0.981
0.942
0
1
2
23 days
2 days
0.6 days
none
Tbl. 8.4 SurajW95-96.xls (2hr24min. data, daily av.).

Results of the stepped correlation between basement raw data and daily average data seem,
in this data set, to give inconsistent results. Table 8.3 shows there to be a peak in correlation
between basement temperature and outdoor temperature at just over 12 days. The daily
average data in table 8.4 however shows a peak at around 23 days. A glance at graph 8.3
shows that the correlation values for all of the first 370 steps in the basement raw data are
high (over 0.8), and consequently the peak value is not significantly higher than any other
value in this group. Results of the daily average data for the basement give a clearer result,
with a definite peak at around 23 days.
All three of the other rooms give results consistent with those of summer95.xls, reinforcing
the evidence of a vertical gradient in time delay, with the 2.4 hourly data giving more
accurate results than the daily average data.
8.4.2c Suraj-S96.xls
shown in graphs 8.5 and 8.6 respectively. 549 steps were performed on the basement raw
data (equating to a time lag of just under 55 days), and 399 steps were performed on both the
ground floor back room and the first floor back room raw data (equating to a time lag of just
under 40 days). 59 steps were performed on the daily average data for all three locations
(equating to a time lag of 59 days). Tables 8.5 and 8.6 show the strongest three consecutive
correlations along with the number of steps performed to achieve these results.
128
129
0.8
0.6
0.4
0.2
0
0
50
100
150
200
250
300
350
400
450
500
550
-0.2
-0.4
-0.6
steps(2hr 40min)
basement
Ground-back
First -back
Graph 8.5. Suraj-S96.xls- stepped correlation
Max 3
corr.
No. of
steps
Est. tL
Bsmnt
GrFlBk
1stBck
0.352
0.360
0.355
495
496
497
0.638
0.652
0.598
10
11
12
0.737
0.855
0.827
0
1
2
1190hr
53min
(49.6d)
25hr 55min
3hr 22min
Tbl. 8.5 Suraj-S96.xls (2hr40min. data). Summary

of results of stepped correlation.
Graph 8.6. Suraj-S96.xls (daily av.) - stepped

correlation
Max 3
corr.
No. of
steps
Est. tL
Bsmnt
GrFlBk
1stBck
0.536
0.556
0.547
49
50
51
0.744
0.840
0.833
0
1
2
0.932
0.941
0.793
0
1
2
50 dys 5hrs
1 dy 10hrs
0.5 dys
Tbl. 8.6 Suraj-S96.xls (2hr40min. data, daily av.).

Summary of results of stepped correlation
The most noticeable characteristic of the results of the stepped correlation on this data set is
the occurrence of two separate and distinct peaks. One possible explanation may be the
existence of a repeat pattern in the outside temperature conditions. Summer temperatures in
Jaisalmer tend to be more stable than those in winter (when the climate is either heating up
or cooling down), and so there is a much higher chance of repeat patterns occurring during
this season. Graph 8.7 shows the outside temperature data gathered during this period and
there is evidence of the existence of a repeat pattern with a cycle of around 45 days. This is
in concurrence with the time between peaks in both graphs 8.5 and 8.6.
45
temperature
40
35
30
25
9/29/96 0:00
9/15/96 0:00
9/1/96 0:00
8/18/96 0:00
8/4/96 0:00
7/21/96 0:00
7/7/96 0:00
6/23/96 0:00
6/9/96 0:00
5/26/96 0:00
5/12/96 0:00
4/28/96 0:00
4/14/96 0:00
3/31/96 0:00
20
time
approx. 45 days
Graph 8.7. Repeat patterns in outdoor temperature
In addition, results shown in tables 8.5 and 8.6 differ considerably from the two previous
data sets, the basement having a considerably longer time lag, whilst both the ground floor
and first floor rooms have significantly shorter time lags. The vertical gradient is however
still in evidence. These differences could be attributed to seasonal variation in time lag.
However, because of the conformity of the two previous data sets (one recorded in summer,
the other in winter) these discrepancies are more likely attributable to other influences such
as the previously mentioned patterns in outdoor temperature.
8.4.2d 2ndhaveli-S96.xls
(equating to a time lag of just less than 40 days), and 59 steps were performed on the daily
average data (equating to a time lag of 59 days). Tables 8.7 and 8.8 show the strongest three
consecutive correlations along with the number of steps performed to achieve these results.
1
1.2
1
0.8
0.8
0.4
0.2
0
0
50
100
150
200
250
300
350
400
0.6
0.6
0.4
0.2
0
0
-0.2
10
20
30
40
-0.2
-0.4
-0.4
-0.6
-0.6
steps(24hrs)
steps(2hr 40min)
basement
GrFlMainCrt
GrFlFront
basement
GrFlMainCrt
GrFlFront
50
60
Graph 8.8. 2ndhaveli-S96.xls - stepped correlation
Max 3
corr.
No. of
steps
Est. tL
Bsmnt
GrFlCrt
GrFlFrt
0.392
0.403
0.402
30
31
32
0.830
0.795
0.654
0
1
2
0.702
0.783
0.753
9
10
11
75hours
22mins
(3.1 days)
none
24hr 43min
Graph 8.9. 2ndhaveli-S96 (daily av.) - stepped

correlation
Max 3
corr.
No. of
steps
Est. tL
Bsmnt
GrFlCrt
GrFlFrt
0.559
0.589
0.580
2
3
4
0.908
0.935
0.841
0
1
2
0.891
0.925
0.832
0
1
2
3.3 dys
0.7 dys
0.8 dys
Tbl. 8.8 2ndHaveli-S96.xls (2hr40min. data, daily

av.). Summary of results of stepped correlation.
Tbl. 8.7 2ndHaveli-S96.xls (2hr40min. data).

This data set again shows two peaks in correlation. It was collected at the same time as
Suraj-S96.xls, and so reinforces the notion of repeat patterns in weather. In this haveli,
however, the basement is considerably shallower than that of Hotel Suraj and consequently
the results suggest that this area has a much shorter time lag than in the previous data sets.
8.4.2e Suraj 12months.xls
0.8
0.8
0.6
0.6
0.4
0.2
0
0
50
100
150
200
250
300
350
400
450
500
550
600
(equating to a time lag of just less than 60 days) and 120 steps were performed on the daily
average data (equating to a time lag of 120 days). Tables 8.9 and 8.10 show the strongest
three consecutive correlations along with the number of steps performed to achieve these
results.
0.4
0.2
0
0
-0.2
-0.2
-0.4
-0.4
-0.6
20
40
60
80
-0.6
steps(2hrs 40mins)
Basement
GrFl-back
1st Fl-back
steps (24hrs)
2ndFl-back
Basement
GrndBack
1stFloor
2ndfloor
100
120
Graph 8.10. Suraj12months.xls - stepped correlation
Max 3
corr.
No. of
steps
Est. tL
Bsmnt
GrFlBk
1stBck
2ndBck
0.881
0.881
0.881
255
256
257
0.925
0.927
0.923
19
20
21
0.950
0.971
0.965
0
1
2
0.929
0.964
0.953
0
1
2
614hr
53min
(25.6)
47hr
31min
(1.9d)
3hr
7min
2hrs
30min
Tbl. 8.9 Suraj12months.xls (2hr24min. data).

Graph 8.11. Suraj12months.xls (daily av.) - stepped

correlation.
Max 3
corr.
No. of
steps
Est. tL
Bsmnt
GrFlBk
1stBck
2ndBck
0.940
0.941
0.941
25
26
27
0.979
0.981
0.977
1
2
3
0.990
0.991
0.975
0
1
2
0.983
0.981
0.943
0
1
2
26.2
days
1.9
days
0.6
days
None
Tbl. 8.10 Suraj12months.xls (2hr24min. data, daily

av.). Summary of results of stepped correlation
As this data set is a combination of SurajW95-96.xls and Suraj-S96.xls it was expected that
the results would reflect this, being around the average of the summer and winter values.
With the basement and ground floor this does seem to be the case. However, the calculated
time lag for the first floor is much the same as for Suraj-S96.xls (tables 8.5 and 8.6).
8.4.3 Conclusions
As there is so much variation in results between the different data sets, it seems that this
method does not consistently accurately determine time lag. The problem seems to be that
the method is open to error where patterns in outside weather exist, and for this reason it
appears that Summer95.xls and SurajW95-96.xls give the most credible results. However the
same general pattern in time lag is shown consistently throughout all data sets. The results
for Hotel Suraj demonstrate both vertical and horizontal time lag gradients. The vertical
gradient is the steeper of the two, with time lag increasing from the top to the bottom of the
building. The horizontal gradient increases from the front of the building towards the back.
Illustration 8.1 shows a section of Hotel Suraj with the combined time lag results of
Summer95.xls and SurajW95-96.xls printed in the appropriate logger zone.
Ill. 8.1 Combined time lag results of Summer95.xls and SurajW95-96.xls
It is difficult to ascertain whether any of the difference in results between data sets is due to
seasonal variation, or errors in methodology, but evidence leans towards the latter. Certainly,
in all but the basement shorter logging intervals are preferable. In the case of the basement
daily average values yield the best results.
Recent research (Nicol et al. (1995)) has shown that indoor temperature is more closely
related to the running mean of the outdoor temperature. The second approach for
determining time lag is built from this research.
8.5 2nd approach - Running Mean

Running mean is a technique used to smooth a time series and provide us with an
exponentially weighted moving average. It was used by Humphreys (1972, 1978) to
represent the outdoor temperature over a few days. In this type of moving average the most
recent value of temperature is given the greatest weight, whilst each preceding value is given
a weight which reduces in geometric progression according to the time which has elapsed.
133
The moving average has two connected features. It smoothes the series of temperatures and
it lags behind any changes that occur. The running mean temperature for time i (iTRM) is
calculated from the series:
TRM = W{iTOD + (1-W) (i-1)TOD + (1-W)2 (i-2)TOD + }
where iTOD is the outdoor temperature for time i and W is a time constant. Thus we can see
that over time, as the time period i increases, the weights assigned to the earlier, older data in
the time series may become negligible. This series is simplified to give the formula:
TRM = WTOD + (1-W)(i -1)TRM ...Eq. 8.1

(Berenson et al., 1989)
Where:
i
TRM
W
TOD
= time period
= running mean outdoor temperature
= subjectively assigned weight or smoothing coefficient (where 0<W<1).
= outdoor temperature
The choice of weight (W) assigned to the time series is quite important. Unfortunately this
selection is subjective. The weight determines the half-life of the series. In the case of
thermal building data this is the time taken for the running mean temperature to change
0.5C / 1C step change in outdoor temperature. The weight is related to the half-life by the
following expression:
t0.5 = 0.69 tint W..Eq. 8.2

where:
t0.5
tint
W
= half-life of smoothed series

= time interval between successive logger recordings.
= weight
Generally, the more massive the building, the longer the half-life. However, when dealing
with real buildings and real temperatures the changes that occur are not step changes, but
gradual increases and decreases. Half-life is therefore very difficult to visualise in terms of
relating indoor temperature to outdoor temperature. If a temperature series (T) is plotted,
along with the running mean of that temperature series (TRM), against time, the relationship
between the weight (W) assigned to TRM, and the time lag between T and TRM becomes
apparent (see fig. 8.2).
38
tL
TL
37
36
35
time
Temp (T)
34
33
TRM (W=0.3)
t(RM)W
=0.3
32
31
30
29
28
2:24
8:24
14:24
20:24
2:24
8:24
14:24
20:24
2:24
8:24
tem p
Time (t)
Fig. 8.2. Relationship between W and time lag with real data.
The author developed the following method in order to convert W more accurately into a
value for time lag:
8.5.a. Method for conversion of exponential weight to time lag
A column of data was generated in Excel using the formula:
Tout A sin(2ft ) D
when plotted this gives a sine wave where:
Tout
A
f
t
D
= temperature at time t
= amplitude of sine wave
= frequency
= time
= average temperature
If this sine wave is used to approximate the variation in outdoor temperature over one day:
Tout
A
f
t
D
= temperature outside at time t

= daily range / 2
= 1/24
= time in hours from 1 to 24
= daily average temperature
Using values appropriate for Jaisalmer (A=8, D=25) the following graph (fig.8.3) can be
plotted.
temperature
35
25
Tout
15
0
12
18
24
time
Fig. 8.3. Generated sine curve approximating outdoor temperature for 24 hours in Jaisalmer5
As some of the indoor spaces in both of the havelis are estimated as having time lags of
greater than twenty four hours, the annual temperature cycle also needs to be accounted
for. This is achieved by representing D (the daily average) as another sine wave. The
equation now becomes:
Tout = Asin(2fd th) + (Bsin(2fyth)) +C

Where:
Tout
A
5
= temperature outside at time t

= daily temperature range / 2
There should, in reality, be a phase angle included in the equation which would shift the sine wave forward in
time, but for the purpose of converting w to tL this is not relevant.
fd
th
B
fy
C
= 1/24
= time in hours from 1 to 8760
= annual temperature range / 2
= 1/8760
= average annual temperature
Again using values appropriate for Jaisalmer (A=8, B=20, C=20) the following graph
(fig.8.4) can be plotted.
50
40
temperature
30
20
Outdoors
10
0
0
50
100
150
200
250
300
350
-10
time(days)
Fig. 8.4. Generated sine curve approximating outdoor temperature for 1 year in Jaisalmer6
The running mean of the outdoor temperature was calculated using equation 8.1 for
values of W = 0.0001 to 0.999.
The maximum (peak) temperature for each weighted mean series was found, and the
corresponding time noted (tmax).
Outdoor tmax was subtracted from tmax for each weighted mean series. This time interval
corresponds to the time between peaks or tL.
There should, in reality, be a phase angle included in the equation which would shift the sine wave forward in
time, but for the purpose of converting w to tL this is not relevant.
8.5.b. Results
Table 8.11 gives a summary of results obtained from the previous methodology.
Wst
.0001
tL(hrs)
tL(dd:hh:mm)
1877.9
1109.9
821.9
725.9
78:05:54 46:05:54 34:05:54 30:05:54
605.9
485.9
389.9
25:05:54 20:05:54 16:05:54
Wst
.003
.02
tL(hrs)
tL(dd:hh:mm)
341.9
221.9
101.8
77.7
14:05:54 09:05:54 04:05:48 03:05:42
53.7
53.5
29.4
02:05:42 02:05:30 01:05:24
Wst
.04
.06
tL(hrs)
tL(dd:hh:mm)
29.4
29.3
29.2
29.2
01:05:24 01:05:18 01:05:12 01:05:12
5.1
5
4.9
00:05:06 00:05:00 00:04:54
Wst
.1
.18
tL(hrs)
tL(dd:hh:mm)
4.5
3.9
3.8
3.7
00:04:30 00:03:54 00:03:48 00:03:42
3.6
3.5
3.4
00:03:36 00:03:30 00:03:24
Wst
.21
.25
tL(hrs)
tL(dd:hh:mm)
3.3
3.2
3.1
3.1
00:03:18 00:03:12 00:03:06 00:03:06
3
2.9
2.9
00:03:00 00:02:54 00:02:54
Wst
.28
.4
tL(hrs)
tL(dd:hh:mm)
2.8
2.7
2.7
2.4
00:02:48 00:02:42 00:02:42 00:02:24
2.1
1.9
1.8
00:02:06 00:01:54 00:01:48
Wst
.55
.8
tL(hrs)
tL(dd:hh:mm)
1.6
1.5
1.3
1.2
00:01:36 00:01:30 00:01:18 00:01:12
.0006
.005
.045
.15
.22
.29
.6
.001
.01
.05
.16
.23
.3
.7
.0012
.015
.055
.17
.24
.35
.75
.0015
.002
.03
.065
.19
.26
.45
.9
.0025
.035
.07
.2
.27
.5
.95
1.1
1
.9
00:01:06 00:01:00 00:00:54
Tbl.8.11. Summary of results for the conversion of standardised exponential weight (Wst) to time lag (tL).
Because W is dependent on the frequency of logger recordings, the table of results has been
compiled for values of W corresponding to 1hour logger interval (referred to as standardised
weight or Wst). To convert for use with other logger intervals simply multiply the tabulated
value for W by the logger interval (in hours). The exponential smoothing was initially
performed on the data generated for figure 8.3 consisting of 365 daily cycles and 1 annual
cycle, at 1-hour intervals. The division of hours was achieved by exponentially smoothing 10
daily cycles at 0.1-hour intervals. The results shown in table 8.11 are a combination of these
two processes.
8.5.c. Discussion
Assuming indoor temperatures in free-running buildings follow an exponential smoothing of
the outdoor temperature, the maximum possible time shift in daily cycle is shown to be 6
hours (equal to one quarter of the 24-hour, daily cycle). In addition, once the available mass
is sufficient to shift the indoors daily peak by approximately 5 hours, the annual cycle begins
to be affected. This results in the shifting of the daily average temperature peak, along with
the dampening of the annual range. It follows that the maximum shift in daily average
temperature would be equal to one quarter of the 365-day annual cycle, approximately 91
days. When the available mass is sufficient to give a time lag of at least two weeks (as seems
to be the case in the basement of Hotel Suraj), the results suggest that the indoor temperature
is responding to daily average temperatures from previous years, even decades.
8.5.1 Method
The following method was developed, and used to estimate the indoor temperature time lag
in relation to outdoor temperature, by means of exponential smoothing.
1.
3.
The running mean of the outdoor temperature was calculated using equation 8.1 for
values of W = 0.1 to 0.9
Each calculated running mean series was correlated with each indoor air temperature
series, and the value of W corresponding to the maximum correlation was noted for
each indoor temperature series.
Steps 1 and 2 were repeated with smaller increments of W for values of W = peak W
4.
0.05.
W was converted to time lag using table 8.11.
2.
8.5.2 Results
The running mean correlation method was applied to four air temperature data sets:
a. Summer95.xls, both for the raw data (36 minutes for each step) and for the calculated
daily average (24 hours for each step)
b. SurajW95-96.xls, both for the raw data (2 hours 24 minutes for each step) and for the
c. Suraj-S96.xls, both for the raw data (2 hours 24 minutes for each step) and for the
d. 2ndHaveli-S96.xls, both for the raw data (2 hours 24 minutes for each step) and for the
e. In addition data sets b and c above were combined and analysed as twelve months data
(raw data and daily average)
8.5.2a Summer95.xls
0.9
0.9
0.8
0.8
0.7
0.7
Results of the exponential smoothing performed on the raw and daily average data are shown
in graphs 8.12 and 8.13 respectively. Exponential weight (on the x-axis) has been shown on
a logarithmic scale. Tables 8.12 and 8.13 show the strongest three consecutive correlations,
along with the exponential weight assigned to the series to achieve these results. The
standardised value for the exponential weight is also shown, along with the estimated value
for time lag (calculated from table 8.11).
0.6
0.5
0.4
0.6
0.5
0.4
0.3
0.3
0.2
0.2
0.1
0.1
0.00001
0.0001
0.001
0.01
0.1
0.0001
0.001
0.01
TopFront
GrFlBack
1stFlBack
GrFlCourt
TopMid
Basement
1stFlMid
GrFlPorch
TopFlBack
TopFlCourt
1stFlFront
Graph 8.12. Summer95.xls, exponential smoothing.
0.1
exponential weight
exponential weight
TopFront
1stFlBack
TopMid
1stFlMid
TopFlBack
GrFlBack
GrFlCourt
Basement
GrFlPorch
TopFlCourt
1stFlFront
Graph 8.13. Summer95.xls (daily average),

exponential smoothing.
Bsmnt
GrFlBk
1stBck
1stMid
1stFrt
2ndBck
.4358
.4478
.7503
.6788
.6845
.7963
.7335
.4378
.4339
.001
.4481
.4474
.35
.7527
.7521
.25
.6788
.6772
.3
.6847
.6836
.25
.7974
.7917
.06
.7338
.7335
.05
.0012
.0015
.4
.45
.3
.35
.35
.4
.3
.35
.07
.1
.055
.06
.2
.25
.8
.9
345
1.035
1.38
1.183
1.38
5.914
7.527
2.070
.518
Wst
.002
.67
.5
.58
.5
.12
.09
.33
1.33
Est tL
(hrs)
485.9
1.4
1.8
1.65
1.8
4.3
4.7
2.5
Max 3
corr.
W
Est t0.5
(hrs)
2ndMid
2ndFrt
GrCrt
GrPch
2ndCrt
.8418
.8529
.8540
.8735
.8602
.858
.1
.8530
.8525
.75
Tbl. 8.12. Summer95.xls (36 min. data). Summary of results of exponential smoothing.
1stBck
1stMid
1stFrt
2ndBck
2ndMid
2ndFrt
GrCrt
GrPch
2ndCrt
.6637
.4996
.8154
.7601
.7443
.9236
.8457
0.838
.841
.6638
.6637
.66
.4997
.4996
.62
.8155
.8154
.47
.7601
.7601
.84
.7443
.7442
.79
.9236
.9235
.62
.8458
.8457
.92
.67
.68
.63
.64
.48
.49
.85
.86
.8
.81
.63
.64
.93
.94
24.716
26.29
34.5
19.482
20.7
26.29
17.806
Wst
.028
.026
.02
.035
.033
.026
.039
Est tL
(hrs)
53.6
53.6
53.7
29.4
29.4
53.6
29.4
Bsmnt
Max 3
corr.
W
Est t0.5
(hrs)
GrFlBk
Tbl. 8.13. Summer95.xls (36 min. data, daily average). Summary of results of exponential smoothing
The results of these data sets are in contradiction with previous results, using the stepped
correlation method, with the same data. In all cases the correlation is poor (less than those of
the stepped correlation), and there is no evidence of either a vertical or a horizontal time lag
gradient. It is possible that this data set was recorded over too short a period for exponential
smoothing to be an effective method of determining time lag. With data sets of this
magnitude, results suggest that the stepped correlation method gives more credible results,
with higher correlations.
8.5.2b SurajW95-96.xls
in graphs 8.14 and 8.15 respectively. Exponential weight (on the x-axis) has been shown on
a logarithmic scale. Tables 8.14 and 8.15 show the strongest three consecutive correlations,
1
0.9
0.9
0.8
0.8
0.7
0.7
0.6
0.5
0.4
0.6
0.5
0.4
0.3
0.3
0.2
0.2
0.1
0.1
0
0.0001
0.001
0.01
0.1
0.00001
0.0001
0.001
0.01
0.1
exponential weight
exponential weight
Basement
GrFl-back
1stFl-back
2ndFl-back
Graph 8.14. SurajW95-96.xls, exponential smoothing
Max 3
corr.
Bsmnt
GrFlBk
1stBck
2ndBck
.9944
.9897
.9826
.9820
Basement
GrndBack
1stFloor
2ndfloor
Graph 8.15. SurajW95-96.xls (daily average)
Max 3
corr.
Bsmnt
GrFlBk
1stBck
2ndBck
.9944
.9950
.9930
.9897
.9959
.9924
.03
.9951
.9950
.16
.9930
.9930
.52
.9897
.9897
.62
.035
.04
.17
.18
.53
.54
.63
.64
473.143
97.412
31.246
26.286
.9958
.9923
.004
.9903
.9901
.025
.9826
.9825
.27
.9821
.9820
.32
.0035
.003
.02
.019
.25
.23
.3
.285
473.143
82.8
6.624
5.52
Wst
.0015
.0083
.1
.125
Wst
.0015
.0071
.022
.026
Est tL
(hrs)
605.9
149.8
4.5
4.2
Est tL
(hrs)
605.9
173.9
53.7
53.6
W
Est t0.5
(hrs)
Tbl. 8.14 SurajW95-96.xls (2hr24min. data).

Summary of results of exponential smoothing.
W
Est t0.5
(hrs)
Tbl. 8.15 SurajW95-96.xls (2hr24min. data, daily

av.). Summary of results of exponential smoothing.
The correlation coefficients achieved in this data set are surprisingly high. Over this sixmonth period the basement daily average temperature has a coefficient of 0.996 with the
running mean of the daily average outdoor temperature, with a standardised weight equal to
.0015. Again the vertical time lag gradient is evident, with considerably longer time lags on
the lower floors than the upper. Both the raw data and daily average data give the same result
of 25 days and 5.9 hours for the time shift in the basement. This is of the same magnitude as
the results from the daily average stepped correlation for the same data set. Results for the
other three floors are longer than the relative results from the stepped correlation method, but
consistent in rank.
8.5.2c Suraj-S96.xls
0.9
0.9
0.8
0.8
0.7
0.7
in graphs 8.16 and 8.17 respectively. Exponential weight (on the x-axis) has been shown on a
logarithmic scale. Tables 8.16 and 8.17 show the strongest three consecutive correlations,
0.6
0.5
0.4
0.6
0.5
0.4
0.3
0.3
0.2
0.2
0.1
0.1
0
0.0000001
0.000001
0.00001
0.0001
0.001
0.01
0.1
0.00001
0.0001
basement
Ground-back
Bsmnt
GrFlBk
1stBck
.7833
.8589
.9624
.0000001
.8596
.8589
.045
.9625
.9623
.22
.04
.035
.21
.2
16560000
41.4
7.886
Wst
4.2 *10-8
.017
.088
Est tL
(hrs)
2190
77.7
4.7
W
Est t0.5
(hrs)
basement
First-back
Graph 8.16. Suraj-S96.xls, exponential smoothing
Max 3
corr.
0.001
0.01
0.1
exponential weight
exponential weight
Tbl. 8.16 Suraj-S96.xls (2hr24min. data).

Ground-back
First-back
Graph 8.17. Suraj-S96.xls (daily average),

exponential smoothing
Bsmnt
GrFlBk
1stBck
.7521
.9018
.9714
.000001
.9018
.9016
.25
.9715
.9714
.56
.26
.27
.57
.58
16560000
63.692
29.053
Wst
4.2 *10-8
.011
.024
Est tL
(hrs)
2190
101.8
53.6
Max 3
corr.
W
Est t0.5
(hrs)
Tbl. 8.17 Suraj-S96.xls (2hr24min. data, daily av.).

In both tables 8.16 and 8.17 the correlation coefficient for the basement is noticeably lower
than in the previous data set. In addition, the estimated time lag is significantly longer than
results from the previous data sets and the respective results using the stepped correlation
method, giving the maximum value possible of around 91 days. However, graphs 8.16 and
8.17 show that the correlation coefficient reaches a plateau at an exponential weight equal to
.0001 and .001 respectively. The peak value is not significantly different from any of these
plateau values, and so one could conclude that the estimated time lag is between 78 and 91
days. Even with these lower correlations and discrepancies in results, the vertical time lag
gradient is again in evidence.
8.5.2d 2ndhaveli-S96.xls
0.9
0.9
0.8
0.8
0.7
0.7
0.6
0.5
0.4
0.6
0.5
0.4
0.3
0.3
0.2
0.2
0.1
0.1
0
0.0001
0.001
0.01
0.1
0
1
0.000001
0.00001
exponential weight
basement
GrFlMainCrt
0.0001
0.001
0.01
0.1
exponential weight
GrFlFront
Graph 8.18 2ndhaveli-S96.xls, exponential smoothing
basement
GrFlMainCrt
GrFlFront
Graph 8.19 2ndhaveli-S96.xls (daily average),

exponential smoothing
Bsmnt
GrFlCrt
GrFlFrt
.5891
.8970
.8840
.5911
.5828
.025
.8980
.8980
.22
.8842
.8841
.14
.02
.015
.21
.20
.13
.12
82.8
7.886
12.8
Wst
.008
.088
.054
Est tL
(hrs)
149.8
4.7
29.2
Max 3
corr.
W
Est t0.5
(hrs)
Tbl. 8.18 2ndHaveli-S96.xls (2hr24min. data).

Bsmnt
GrFlCrt
GrFlFrt
.7620
.9662
.9487
.7624
.7622
.0001
.9662
.9661
.44
.9487
.9486
.47
.001
.0012
.45
.46
.48
.49
16560
36.8
34.5
Wst
.000042
.019
.02
Est tL
(hrs)
>1877.9;
<2190
53.7
53.7
Max 3
corr.
W
Est t0.5
(hrs)
Tbl. 8.19 2ndHaveli-S96.xls (2hr24min. data,

daily av.). Summary of results of expo smoothing.
The results for the raw data shown in table 8.18 are consistent with the respective results
using the stepped correlation method. However, the results for the daily average data are not.
The estimated time lag for the 2nd haveli basement using this method is between 78 and 91
days, compared to 3.3 days with the stepped correlation method. Again the correlation
coefficient reaches a plateau (shown in graph 8.19), where the peak correlation could be
achieved with a standardised exponential weight of anything above .00042, but this still
gives significantly longer estimated time lags then the stepped correlation method.
8.5.2e 12months.xls
0.9
0.8
0.8
0.7
0.7
0.9
0.6
0.5
0.4
0.6
0.5
0.4
0.3
0.3
0.2
0.2
0.1
0.1
0
0.00001
0.0001
0.001
0.01
0.1
0.00001
0.0001
GrFl-back
0.01
0.1
exponential weight
exponential weight
Basement
0.001
1stFl-back
Basement
2ndFl-back
Graph 8.20. 12months.xls, exponential smoothing.
GrndBack
1stFloor
2ndfloor
Graph 8.21. 12months.xls (daily average),

exponential smoothing.
The results for both the raw and daily average data in this series gave uniformly high
correlation coefficients. Particularly noticeable is the correlation coefficient of 0.985 for the
daily average basement data. The estimated time lags are of the same magnitude as the
results generated using the stepped correlation method, and the vertical time lag gradient is
again in evidence.
Bsmnt
GrFlBk
1stBck
.9828
.9857
.9734
.005
.9666
.9676
.9666
.115
.9836
.9871
.9836
.31
.0045
.004
.11
.105
.3
.29
368
15.05
5.52
Wst
.0019
.046
.125
Est tL
(hrs)
509.9
29.3
4.2
Max 3
corr.
W
Est t0.5
(hrs)
Tbl. 8.20 Suraj12months.xls (2hr24min. data).

Bsmnt
GrFlBk
1stBck
.9824
.9778
.9914
.9851
.9843
.03
.9778
.9778
.55
.9914
.9914
.83
.035
.04
.56
.57
.84
.85
473.14
29.571
19.714
Wst
.0015
.023
.035
Est tL
(hrs)
605.9
53.7
29.4
Max 3
corr.
W
Est t0.5
(hrs)
Tbl. 8.21 Suraj12months.xls (2hr24min. data, daily

av.). Summary of results of exponential smoothing.
8.5.3. Conclusions
The exponential smoothing method gives more consistent results with data collected during
periods of transient daily average temperature (markedly SurajW95-96). This suggests that
data collected during "heating-up" or "cooling-down" seasons of the annual temperature
cycle is invaluable when analysing case studies of the effect of mass on indoor temperature.
When daily average temperatures are stable over a period of time the effect of extensive
exposed mass is difficult to interpret, particularly regarding shifts in annual temperature
cycles (as is the case in both the basement and ground floor rooms of Hotel Suraj).
Results from the Summer95.xls series also suggest that the exponential smoothing method is
not suitable for small data sets, particularly those recorded over short periods of time. This
problem is further compounded when analysing daily average data. With the exception of
this data set (Summer95.xls), all results for Hotel Suraj demonstrate the existence of a steep
vertical time lag gradient, with longer lags on the lower floors. In addition, the results
indicate that, where the exposed mass is sufficient to shift the annual temperature cycle as
well as the daily temperature cycle, daily average data gives the most credible results.
However, if the exposed mass is only sufficient to shift the daily temperature cycle the raw
data, in this case recorded at 2.4 hour intervals, gives the more credible results.
INCREASING TIME LAG
Illustration 8.2 shows a section of Hotel Suraj with the estimated time lag results of the
exponential smoothing method printed in the appropriate logger zone. Where available
results from the Suraj12months series were used (this being the largest data set). Where not
available results from SurajW95-96 were used (recorded during a transient temperature
period). Results from daily average data were used for both the basement and ground floor
back zones, whilst the time lags for the top two floors was estimated from results generated
from the raw data.
Ill. 8.2. Combined time lag results of Suraj12months and SurajW95-96
Neither the ground floor back nor first floor back rooms show any evidence of seasonal
variation in time lag when using the exponential smoothing method. However the basement
gives significantly longer results in the summer than the winter (about 91 days in the summer
compared with 25 days in the winter). This may be a product of the higher solar radiation
levels and longer daylight hours in Jaisalmer at this time of year. It could also be an anomaly
of the summertime data series, particularly in light of the relatively stable daily average
outdoor temperature over this recording period.
8.5.4 Validation of exponential smoothing method

Throughout the results and conclusions of the previous methodology the assumption has
been made that the indoor temperatures in free-running buildings follow an exponential
smoothing of the outdoor temperature. The high correlation coefficients achieved in the
results indicate that there is a very strong relationship between the two. However a near
positive correlation coefficient does not necessarily imply a unit relationship. In order to
validate the connection, results from the Suraj12months daily average series were further
analysed. Initially indoor temperature was plotted against the highest correlated exponential
smoothing of the outdoor temperature. If a unit relationship exists the plot should give a
straight line with a slope of 1. The results are shown in graphs 8.22 to 8.24.
35
33
y = 1.5855x - 14.793
expo smoothing of outdoor temp (deg C)
31
29
27
winter
25
summer
winter
23
summer
21
19
y = 1.8502x - 24.422
BASEMENT
17
15
20
22
24
26
28
30
32
indoor temp (deg C)
Graph 8.22 Plot of expo. smoothed outdoor temp., w=0.035, against indoor basement temp.
38
36
34
y = 1.1932x - 5.7018
exposmoothing of outdoor temp (deg C)
32
30
28
winter
26
summer
24
winter
22
summer
20
18
16
y = 1.1886x - 6.3325
14
GROUND FLOOR BACK
12
10
15
17
19
21
23
25
27
29
31
33
35
indoor temp (deg C)
Graph 8.23 Plot of expo. smoothed outdoor temp., w=0.56, against indoor ground floor back zone temp.
38
36
y = 0.9376x + 0.6591
34
expo smoothing of outdoor temp (deg C)
32
30
28
winter
26
summer
24
winter
22
summer
20
18
16
14
y = 1.0852x - 4.9914
FIRST FLOOR BACK
12
10
15
17
19
21
23
25
27
29
31
33
35
37
39
indoor temp (deg C)
Graph 8.24 Plot of expo. smoothed outdoor temp., w=0.84, against indoor first floor back zone temp.
In each graph the data set has been divided into a winter and summer series, with the
beginning of April as the division between winter and summer, and the beginning of October
as the division between summer and winter. Both results for the first floor back zone show a
close to unit relationship between the indoor and exponentially weighted outdoor
temperature (a slope of 0.94 in the summer and 1.09 in the winter). However, as the
exponential weighting of the series gets smaller, the slope increases. Whilst the summer and
winter gradient results for the ground floor back zone are reasonably close to 1 (both 1.19),
gradient results for the basement are significantly higher (1.59 in the summer and 1.85 in the
winter). In order to further investigate the relationship between the actual daily average
temperature series recorded in each of the zones, and the relative highest correlated
exponentially smoothed outdoor temperature series, both were plotted against time. The
results are shown in graphs 8.25 to 8.27.
35
33
31
temp (deg C)
29
27
Basement
25
expo. w=0.035
23
21
BASEMENT
19
17
15
26-Aug
14-Oct
2-Dec
20-Jan
9-Mar
27-Apr
15-Jun
3-Aug
21-Sep
9-Nov
date
Graph 8.25 Daily average basement temp. and expo. smoothed outdoor temp. (w=0.035) against time
38
36
34
32
30
temp (deg C)
28
26
GrndBack
24
expo. w=0.56
22
20
18
GROUND FLOOR BACK
16
14
12
10
26-Aug
14-Oct
2-Dec
20-Jan
9-Mar
27-Apr
15-Jun
3-Aug
21-Sep
9-Nov
date
Graph 8.26 Daily average ground floor temp. and expo. smoothed outdoor temp. (w=0.56) against time
40
38
36
34
32
temp (deg C)
30
28
1stFloor
26
expo. w=.84
24
22
20
FIRST FLOOR BACK
18
16
14
12
10
26-Aug
14-Oct
2-Dec
20-Jan
9-Mar
27-Apr
15-Jun
3-Aug
21-Sep
9-Nov
date
Graph 8.27 Daily average first floor temp. and expo. smoothed outdoor temp. (w=0.84) against time
8.5.4.1 Discussion
The exponential smoothing methodology makes several assumptions. Firstly it assumes the
relationship between indoor and outdoor temperature to be exponentially weighted. This
seems possible as this time series weights recent temperatures more heavily than those
further back in time. Nonetheless, there may be other mathematical models that are more
suitable. In addition the methodology assumes that the indoor and outdoor series have equal
mean temperatures. When working with the raw data sets, both daily average and annual
average temperatures are assumed equal. When working with the daily average data sets (as
in graphs 8.25 to 8.27) only annual average temperatures are assumed equal. Chapter 4
showed that, for the Suraj12months data set, the annual average outdoor temperature was
equal to that of the basement. However, annual average values for the ground and first floor
zones both exceed the outdoor value by 0.6C and 2.1C respectively. Finally, the
methodology also assumed a sinusoidal outdoor (and indoor) temperature pattern, and that
the phase angle in this sine wave was not significant.
Graph 8.25 shows that the exponentially smoothed outdoor temperature series is up to 5C
cooler in the winter, and up to 3C warmer in the summer, than the recorded basement
temperature. Both series peak and trough simultaneously, but the amplitude of the smoothed
outdoor series is double that of the indoor series. In both graphs 8.26 and 8.27 a trend-line
has been fitted to the data (6th order polynomial), in order to clarify the relationship between
the two series. Graph 8.26 shows that the trend-line for the smoothed outdoor temperature
series is up to 3C cooler in the winter, and up to 1C warmer in the summer than the trend-
line for the ground floor temperatures. Accounting for the disparity in mean annual
temperatures gives a result of 2.4C cooler in the winter and 1.6C warmer in the summer.
Again the amplitude of the exponentially smoothed series is greater than that of the indoor
series, but to a lesser extent than with the basement results. Graph 8.27 shows that the trendline of the smoothed outdoor series is up to 3C cooler in the winter and 1.5C cooler in the
summer than the trend-line for the first floor data. Allowing for the discrepancy in mean
average temperatures gives a result of 0.9C cooler in winter and 0.6C warmer in summer.
These results suggest that exponential smoothing of the outdoor temperature is not an
accurate approximation for indoor temperature. Furthermore, the smaller the value for w
assigned to the outdoor series (hence the greater the 'smoothing'), the larger the margin of
error.
8.6
Comparison of methodologies
The following tables (8.22 - 8.25) are a comparison of results achieved on each of four floors
using both the stepped correlation and exponential smoothing methods for determining time
lag between indoor and outdoor temperature in Hotel Suraj.
Stepped correlation
BASEMENT
correlation
Summer95
SurajW95-96
Suraj-S96
Suraj12months
Raw data
Daily average
Raw data
Daily average
Raw data
Daily average
Raw data
Daily average
Est. time lag

(dd:hh:mm)
Expo. smoothing
correlation
Est. time lag

(dd:hh:mm)
NA
NA
NA
NA
.438
NA
20:05:54
NA
.845
.893
12:05:31
23:00:00
.996
.996
25:05:54
25:05:54
.360
.556
.881
.941
49:14:24
50:05:00
25:14:24
26:04:48
.783
.752
.986
.985
91:05:54
91:05:54
21:05:54
25:05:54
Tbl. 8.22. Comparison of results of both methodologies for all data series recorded in the basement of Hotel
Suraj.
Stepped correlation
GROUNDFLOORBACK
correlation
Summer95
SurajW95-96
Suraj-S96
Suraj12months
Raw data
Daily average
Raw data
Daily average
Raw data
Daily average
Raw data
Daily average
Est. time lag

(dd:hh:mm)
Expo. smoothing
correlation
Est. time lag

(dd:hh:mm)
.562
.122
02:01:09
02:14:24
.448
NA
00:01:24
NA
.893
.958
02:10:05
02:00:00
.990
.995
06:05:48
07:05:54
.652
.840
.927
.981
01:01:55
01:10:00
01:23:31
01:21:36
.860
.902
.968
.978
03:05:42
04:05:48
01:05:18
02:05:42
Tbl. 8.23. Comparison of results of both methodologies for all data series recorded in the ground floor back
room of Hotel Suraj.
Stepped correlation
FIRSTFLOORBACK
correlation
Summer95
SurajW95-96
Suraj-S96
Suraj12months
Raw data
Daily average
Raw data
Daily average
Raw data
Daily average
Raw data
Daily average
Est. time lag

(dd:hh:mm)
Expo. smoothing
correlation
Est. time lag

(dd:hh:mm)
.773
.705
01:01:16
01:00:00
.753
.664
00:01:48
02:05:36
.965
.984
01:02:53
00:14:24
.983
.993
00:04:30
02:05:42
.855
.941
.971
.991
00:03:22
00:12:00
00:03:07
00:14:24
.963
.972
.987
.991
00:04:42
02:05:36
00:04:12
01:05:24
Tbl. 8.24. Comparison of results of both methodologies for all data series recorded in the first floor back room
of Hotel Suraj.
Stepped correlation
SECONDFLOORBACK
correlation
Summer95
SurajW95-96
Suraj-S96
Suraj12months
Raw data
Daily average
Raw data
Daily average
Raw data
Daily average
Raw data
Daily average
Est. time lag

(dd:hh:mm)
Expo. smoothing
correlation
Est. time lag

(dd:hh:mm)
.736
.756
00:02:49
none
.797
.760
00:04:18
01:04:24
.969
.983
00:03:07
none
.982
.990
00:04:12
02:05:36
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
Tbl. 8.25. Comparison of results of both methodologies for all data series recorded in the second floor back
room of Hotel Suraj.
The results of both methodologies for the basement zone are noticeably similar, with the
winter and 12months data sets giving time lag results in the region of 25 days, whilst the
summer data gives considerably longer time lags.
The ground floor back zone results display more variation. As mentioned in earlier chapters
this, unlike the other three zones, is an inhabited room, with the associated internal gains,
and occasional use of an electric heater in the winter, and desert cooler in the summer. These
factors could all contribute to the variation in results. Of the two approaches, the stepped
correlation method shows less variability, with time lag results ranging from just over 1 day
to just over 2.5 days, compared with the results of the smoothing method which range from
1.25 to 7.25 days. However, most results of both methods demonstrate a 2 day shift in annual
cycle, whilst the shift in daily cycle is harder to determine.
The results from the exponential smoothing method for the first floor are almost all in
agreement, with an annual cycle shift of 2 days, and a daily cycle shift of around 4.5 hours.
However, results from the stepped correlation method are slightly more diverse, with daily
cycle shift results ranging from 1.25 hours to just under 3.5 hours. The results for the shift in
annual cycle are divided between 1 day shifts and no shift.
Unfortunately there are few results for the second floor zone, but the stepped correlation
results indicate a time lag in the region of 3 hours, whilst the exponential smoothing results
indicate a time lag in the region of 4 hours.
8.6.1 Discussion
Both methods consistently show the same general vertical pattern in results, with the
estimated time lag increasing from the top to the bottom of the haveli. The stepped
correlation method also displays some evidence of a horizontal pattern with estimated time
lags increasing from the front to the back of the haveli.
Both methods also reveal the existence of shifts in annual temperature in addition to the
expected shifts in daily cycle. Floor by floor analysis of the results for Hotel Suraj indicates
that, whilst the top floor is coupled with outdoor air temperature, and thus responds primarily
to daily variations in temperature, the basement seems to be coupled to the ground
temperature, and responds primarily to annual variations in temperature. The ground and first
floor zones are coupled with both to some extent, with the ground floor responding more to
variations in the annual cycle, and the first floor responding more to variations in the daily
cycle. Although interconnected both cycles can be discussed and estimated separately, with
daily average temperature data being more than adequate for calculating shifts in the annual
cycle.
Both methods give more consistent results with data recorded during periods of transient
daily average temperature. This is particularly true when studying indoor zones with large
areas of exposed mass (sufficient to shift annual temperature cycles). With hindsight this
seems logical, as there can be no evidence of a time lag between indoor and outdoor
temperature if the outdoor temperature remains constant. This revelation also leads the
author to believe that there is no significant seasonal variation in time lag in Hotel Suraj. The
apparent difference between summer and winter results for the basement can be attributed to
almost constant daily average outdoor temperatures during the summer recording period.
Of the two methodologies, the exponential smoothing method is by far the more complex
and time consuming. This is alleviated, to some extent, by the preparation of table 8.11 for
the conversion of values for W to estimated time lags. Nevertheless, the process is still
lengthy in comparison. It is possible, however, that the process could be automated with the
use of macros. The exponential smoothing method does seem slightly more reliable when
dealing with large data sets, but the stepped correlation method gives consistently credible
results with even the shortest data set, and on the whole seems the most useful of the two
methods.
8.7 Decrement Factor

As discussed earlier in this chapter, exposed mass not only affects the indoor environment by
shifting the maximum and minimum temperatures, it also regulates the magnitude of the
indoor temperature swings. This section investigates the decrement factor for the four indoor
zones monitored for the Suraj12months data series. It was hoped that the exponential
smoothing method could be used to estimate decrement factor in conjunction with time lag.
However, results from section 8.5.4 indicate that this would generate inaccurate results.
Alternatively, daily decrement factors for the 12-month period were calculated by dividing
the internal daily temperature range for each zone by the external daily temperature range.
The results are shown in graph 8.28 and summarised in table 8.26.
1.2
decrement factor
0.8
DecFacB
DecFacG
0.6
DecFac1
DecFac2
0.4
0.2
6-Oct-96
22-Sep-96
8-Sep-96
25-Aug-96
28-Jul-96
11-Aug-96
14-Jul-96
30-Jun-96
2-Jun-96
16-Jun-96
5-May-96
19-May-96
21-Apr-96
7-Apr-96
24-Mar-96
25-Feb-96
10-Mar-96
11-Feb-96
28-Jan-96
14-Jan-96
31-Dec-95
3-Dec-95
17-Dec-95
5-Nov-95
19-Nov-95
8-Oct-95
22-Oct-95
date
Graph. 8.28. Daily decrement factor for each of four zones (Suraj12months.xls series), over a 12 month period.
SURAJ12MONTHS Basement
Daily
Decrement
Factor
GrndFloor
1stFloor
2ndFloor7
Av.
0.0155
0.1439
0.4689
0.6264
Min.
0.1690
0.3515
Max.
0.1238
0.4745
0.8172
1.0261
Tbl. 8.26. Summary of results of daily decrement factor calculations over a 12 month period.
7
Data for this zone only available for the six month winter period
Graph 8.28 indicates that all the zones show considerable variation in daily decrement factor
over the 12 month period. Both the decrement factor, and the variation in decrement factor,
increases from the bottom to the top of the haveli.
It is interesting to note that the maximum decrement factor for the second floor zone is
greater than one, indicating that the daily internal temperature range exceeds the daily
external temperature range. Graph 8.28 shows this happens on two separate occasions, both
during the month of January. Closer examination of the data for this period reveals that the
outdoor daily temperature range on both occasions was very low, suggesting a cloudy,
possibly wet day, with little solar radiation. At this time of year indoor temperatures for the
second floor zone are consistently higher than outdoor temperatures, due to passive solar
gains through the thinner, exposed walls. When conditions outdoors are cloudy, this zone
receives little passive gain, allowing the indoor temperature to fall lower than usual, thus
increasing the daily range, on both theses occasions just beyond that of outdoors.
Graph 8.29 shows the relationship between daily indoor decrement factor and daily outdoor
temperature range for each of the four zones monitored. As only winter data is available for
the second floor zone, the data for each zone has been separated into winter and summer
subsets in order to allow direct comparisons.
1.2
daily decrement factor indoors
R2 = 0.3779
R2 = 0.3464
0.8
DecFacB(W)
DecFacG(W)
DecFac1(W)
DecFac2(W)
DecFacB(S)
DecFacG(S)
DecFac1(S)
R2 = 0.3527
0.6
0.4
R2 = 5E-08
0.2
R2 = 0.0153
R2 = 0.0141
R2 = 0.0153
0
2
10
11
12
daily range outdoors
Graph 8.29 Relationship between daily indoor decrement factor and daily outdoor temperature range
(Suraj12months data set).
Graph 8.29 shows a relatively strong inverse relationship between first and second floor
zones, and daily outdoor temperature range, with r2 values of between 0.35 and 0.38. Whilst
summer and winter results for the first floor zone show a similar dependency on daily
outdoor temperature range, the winter data shows a steeper slope, that is, the decrement
factor for the first floor increases less per C gain in daily outdoor range in the winter than
the summer. The ground and basement zones, both show very little dependence on daily
outdoor temperature range with r2 values of less than 0.015 in all cases. Results from the
previous time lag methodologies suggest that the available mass in the first and second floor
zones is only sufficient to shift the daily temperature cycle; the annual cycle in both zones is
unaffected. In contrast, the ground and basement zones show substantial shifts in both cycles.
So, from the top to the bottom of the haveli, each monitored zone becomes progressively less
dependent on outdoor daily fluctuations in temperature, and more dependant on outdoor
daily average temperatures as part of an annual temperature cycle. This could, in part,
explain the lack of relationship between both the ground and basement zones and the daily
outdoor temperature range. If several years data was available, and a graph of annual
decrement factor plotted against outdoor annual temperature range, it is quite likely that the
strength of relationship would be reversed, with the basement showing the strongest
connection to outdoor annual temperature range.
Table 8.27 shows the relationship between average daily decrement factor and daily cycle
time lag, for the basement, ground and first floor zones of Hotel Suraj. The time lag used is
that calculated by the running mean method and presented in illustration 8.2 in section 8.5.3.
This method showed that the maximum shift in daily temperature cycle of a free-running
building is 6 hours, at which point there would be negligible daily variation in temperature,
and therefor a decrement factor approaching zero. At the other end of the scale, if indoor and
outdoor temperature range were similar, the decrement factor would be 1 with no daily time
lag. Both these data points have been added to the data set and shown in table 8.27 and graph
8.30.
AVERAGE DAILY DECREMENT
BASEMENT
GROUND FLOOR
FIRST FLOOR
FACTOR
DAILY CYCLE TIME LAG

(HRS:MINS)
0
0.0155
0.1439
0.4689
1
06:00
05:54
05:36
04:12
00:00
Tbl. 8.27. Relationship between average daily decrement factor and daily cycle time lag.
0.9
0.8
Daily decrement factor
0.7
0.6
Series1
Poly. (Series1)
0.5
0.4
0.3
0.2
y = -13.216x2 - 0.6864x + 0.9996

R2 = 0.9987
0.1
0
00:00
01:12
02:24
03:36
04:48
06:00
07:12
daily cycle time lag (hrs:mins)
Graph 8.30. Average daily decrement factor plotted against daily time lag for Suraj12months data set.
A trendline has been fitted to the data and the equation and r2 value displayed on the graph. It
shows a very strong quadratic relationship between the two variables, with 99.9% of the
variation in daily decrement factor explained by the variation in the daily cycle time lag.
8.8 Summary of conclusions for chapter 7

This chapter set out to develop methodologies for determining time lag in existing buildings.
Whilst there has been some variation in results, both within and between methodologies, a
consistent pattern has been apparent in all case. Both vertical and horizontal gradients in time
lag are evident in Hotel Suraj, with the vertical gradient being most pronounced. In all cases
the time lag gradient is directly proportional to available mass, increasing from the top to the
bottom, and from the front (street faade) to the back of the haveli.
The stepped correlation method, although open to error where repeat weather patterns exist,
gives the most credible results where data sets are short. Data sets need to be both large and
continuous for the exponential smoothing method to give plausible results. In both cases
higher correlations and more consistent time lag results are achieved with data sets recorded
over periods of transient daily average temperature, particularly when examining high mass
zones. With hindsight this seems obvious, as it is not possible to store heat in anything unless
it changes temperature8. However, thermal performance studies have, in the past, tended to
concentrate on periods of extreme climate, when the outdoor daily average temperatures are
relatively stable. In addition, both methodologies indicate that some zones within the haveli
have sufficient available mass to shift both daily and annual temperature cycles. Results from
the exponential smoothing method suggest that, whilst both annual and daily cycle time
shifts can be calculated and examined separately, they are both interrelated. It is not possible
to shift the annual temperature cycle without an associated shift in daily temperature cycle.
Results from this method also suggest that each cycle can not be shifted by more than a
quarter of the cycle frequency. So for a daily cycle of twenty-four hours, the maximum
possible shift, or time lag, would be six hours, and for an annual cycle of 365 days the
maximum time lag would be 91 days.
With regard to the optimum magnitude of logger interval, both methodologies give
comparable results. When considering building zones where it is anticipated that the
available mass is sufficient to substantially shift the annual temperature cycle, daily average
temperature data yields the best results. In all other cases it was found that 2.4 hour intervals
lacked sufficient detail, whilst 36 minute intervals were excessive. A recording interval in
the region of 1 hour would be preferable. In addition, outdoor temperature data for a
minimum of 1 month previous to the beginning of the monitoring period would be
advantageous, particularly when analysing high mass zones.
Although the time lag results for each monitored zone do vary from winter to summer, the
variation is not consistent. The basement of Hotel Suraj shows the most variation, giving
longer time lag results in summer than winter. However, during the summer correlations for
this zone are notably lower than in the winter, reaching a plateau rather than a peak. It is
therefor reasonable to assume that the variation in results is a product of the static daily
average outdoor temperature during the summer period, giving little opportunity for heat
storage.
All zones monitored over the twelve-month period in Hotel Suraj show considerable
variation in daily decrement factor, with both the magnitude, and variation in daily
decrement factor increasing from the bottom to the top of the haveli. Whilst the daily
decrement factor of the 1st and 2nd floor zones show a relatively strong relationship with
daily outdoor temperature range, the ground and basement show very little dependence. It is
predicted that this situation would be reversed if sufficient data was available to plot annual
8
There are phase change materials which can store heat without temperature changes (Fordham, private
communication)
decrement factor against annual outdoor temperature range, with the basement and ground
zones showing relatively high dependency. A plot of daily decrement factor against daily
cycle time lag9, for each monitored zone, shows a very strong, quadratic relationship exists
between the two variables. Whilst the three calculated, and two theoretical, data points used
in this analysis constitute an extremely limited bases for profound generalisation, this result
does indicate a need for further analysis on similar long term temperature data sets.
As calculated by the running mean method and presented in illustration 8.2, section 8.5.3.
References Chapter 8
BALARAS, C.A. (1996). The role of thermal mass on the cooling load of buildings. An
overview of computational methods. Energy and Buildings, Vol. 24, pp. 1-10.
BERGER, X. (1988). The pumping effect of clothing. Int. J. Ambient Energy, 9(1).
HUMPHREYS, M. A. (1992). Thermal comfort requirements, climate and energy. World
HMSO, London.
NICOL, F., JAMMY, G. & SYKES, O. (1994). A survey of thermal comfort in Pakistan, toward
new indoor temperature standards, Oxford Brookes University.
NICOL, J.F., HUMPHREYS, M.A.AND RAJA,I.A. (1995). Developing indoor temperature
Press.
SODHA, M.S., KAUR, J. AND SAWHNEY, R.L. (1992). Effect of storage on thermal
performance of a building. International Journal of Energy Research, Vol. 16, pp.
697-707
Trans., Vol. 33, pp.165-179
CHAPTER 9
CONCLUSIONS
This chapter draws together the various strands spun in earlier chapters,
making suggestions for areas where there is a need for further research.
9.1 Introduction
The aim of this thesis was to provide a detailed case study of the havelis of Jaisalmer.
Jaisalmer was selected as the study area not only because of the abundance and richness of
the havelis in this region, but also because of the extreme nature of the climate in and around
the area. In addition, the haveli has the added advantage of being a high-density urban form,
economically and ecologically significant in the context of India's rapidly expanding urban
population and associated energy and environmental crisis.
Throughout the previous chapters a consistent pattern of results has materialised, central to
the climatic success of the haveli. At any one period in time Hotel Suraj provides an
extensive choice of indoor climate to the inhabitant. Considerable differences in temperature
can be experienced by horizontal, and even more by vertical, migration through the building.
As discussed in chapter 3, only a few rooms in a haveli were traditionally designed for a
specific purpose. Instead larger rooms were used as season, need and mood dictated. These
findings are substantiated by Roaf's research (1989) into the lifestyle of the Yazdi and their
summer use of space in the house. She noted that "The family is in continual motion around
the house, both vertically and horizontally, in search of an optimum climatic environment."
These climatic differences are influenced by the orientation of the room, exposure to direct
and indirect radiation and ventilation rates; but the results of this study suggest that the range
of temperature is predominantly a product of the varying amounts, particularly floor to floor,
of available mass within the building.
9.2. Indoor environment

Chapter 4 detailed the method and equipment used for the collection of environmental data
in and around Hotel Suraj and the second haveli.
Results from the data collected in Hotel Suraj show that this haveli performs well thermally,
being generally warmer in winter, and cooler in summer, than the recorded outdoor
temperature. This response to the climate is confirmed by the preliminary analysis of data
collected in the second case study haveli. However, when analysing data collected over a
twelve month period, results showed that the majority of the monitored indoor spaces have
similar, or in some cases, slightly higher, annual average temperature than outdoors. This
suggests that there are no passive cooling mechanisms in Hotel Suraj. The difference
between indoor and outdoor temperature at any one time is achieved by modifying the daily,
and, in the higher mass zones, the annual, outdoor temperature cycles. These conclusions
164
highlight the importance, particularly in high mass buildings, of long term logging, with
monitoring periods spanning, ideally, a minimum of twelve months. Analysis of daily and
annual indoor temperature range showed a striking increase from the bottom to the top of the
haveli at all times of year, giving a variety of maximum and minimum temperatures.
Results from limited air movement recordings within Hotel Suraj show the haveli to be well
designed regarding air movement, providing ample opportunity for occupant control.
Openings are located to facilitate cross ventilation and shutters are designed to enable
varying degrees of air movement. In addition air movement and surface temperature
measurements recorded in and around the central courtyard imply that this feature
contributes, by means of stack effect, to the ventilation of the indoor spaces from April
through to July.
9.3. Thermal comfort

Results from a short, transverse, thermal comfort study (73 subjects) conducted in Jaisalmer
at the end of March / beginning of April, and analysed in chapter 5, showed a comfort
temperature in the region of 30C at this time of year. This is slightly higher than the
accepted adaptive model comfort equations of Humphreys (1978) and Nicol et al., (1994)
would predict, possibly in part due to the low humidity levels characteristic of the region.
The extension of the results of the comfort survey in combination with Humphrey's (1978)
comfort prediction equation indicate that the temperature outdoors in Jaisalmer is often too
cold during the months of December and January, and too hot during April through to July.
However, the same comfort prediction method used on the monitored indoor zones of Hotel
Suraj, shows that this haveli always has at least one comfortable zone, with the high mass
ground floor and basement zones performing particularly well during the thermally stressful
months of April, May and June.
These findings relate only to the climate as experienced and perceived by the indigenous
population of Jaisalmer. They bear little relationship to the climate as experienced by the
unacclimatized person, either living in, or visiting the area.
9.4. Outdoor temperature as an indoor predictor

Chapter 7 attempted to develop indoor temperature prediction equations for each of the four
monitored zones, derived from the Suraj12months series, and based solely on the recorded
outdoor temperature.
165
The results showed that the dependence of indoor temperature on immediate outdoor
temperature decreased with the available mass of the indoor zone, that is, from the top to the
bottom of the haveli. This resulted in the second, first and ground floor zones showing the
strongest relationship with the daily average of the recorded outdoor temperature, with
between 94% and 98% of the variation in daily average indoor temperature explained by the
variation in daily average outdoor temperature. The basement, however, showed the
strongest relationship with the monthly average of the outdoor temperature recorded one
month previous, which explained 97% of the variation in monthly average indoor
temperature.
Despite large fluctuations in outdoor daily temperature range over the twelve month
monitoring period, the indoor daily range for each zone remains relatively stable throughout
the year. Again the daily range decreases, with mass, from the top to the bottom of the
haveli, varying from 3.6C on the second floor to 0C in the basement.
9.5. Mass as a thermal regulator

Having established that the available mass was playing an integral role in determining the
temperature in each indoor zone, chapter 8 set out to explore, in detail, the effects of this
available mass on each of the indoor monitored zones in Hotel Suraj. Initially the time delay
(or lag) between indoor and outdoor maximum (and minimum) temperature was
investigated. Conventionally, time lag is discussed and calculated with regard to a single
building element, for example a particular wall or floor. Calculations are complex and time
consuming, and often require input based on assumptions (Matthews et al. 1991). This study
adopted an empirical approach to the determination of time lag, not of individual building
elements, but of indoor temperature in relation to outdoor temperature.
Two methodologies, the stepped correlation and the exponential smoothing methods, were
developed with the intention of determining indoor temperature time lag. Although there was
some variation in results, both within and between methodologies, a consistent pattern was
apparent in all cases. Both vertical and horizontal gradients in time lag were evident, with
the vertical gradient being most pronounced. In all cases the time lag gradient varied with
available mass, increasing from the top to the bottom and from the front (street faade) to the
back of the haveli.
The stepped correlation method, although open to error where repeat weather patterns exist,
gave the most credible results where data sets were short. Data sets need to be both large and
166
continuous for the exponential smoothing method to give plausible results. In both cases
higher correlations and more consistent time lag results were achieved with data sets
recorded over periods of transient daily average temperature, particularly when examining
high mass zones.
In addition, both methodologies indicated that some zones within the haveli have sufficient
available mass to shift both daily and annual temperature cycles. Results from the
exponential smoothing method suggested that, whilst both annual and daily cycle time shifts
can be calculated and examined separately, they are both interrelated. It is not possible to
shift the annual temperature cycle without an associated shift in daily temperature cycle.
This implies that the existence of a 24 hour time lag, as proposed by Gupta (1984) and
discussed in chapter 3, is highly unlikely. Results from this method also suggest that each
cycle can not be shifted by more than a quarter of the cycle frequency. So for a daily cycle of
twenty-four hours, the maximum possible shift, or time lag, would be six hours, and for an
annual cycle of 365 days the maximum time lag would be 91 days.
Although the time lag results for each monitored zone varied from winter to summer, the
variation was not consistent. The basement of Hotel Suraj showed the most variation, giving
longer time lag results in summer than winter. However, during the summer correlations for
this zone were notably lower than in the winter, reaching a plateau rather than a peak. It is
therefor reasonable to assume that the variation in results was a product of the static daily
average outdoor temperature during the summer period, giving little opportunity for heat
storage.
Finally the dampening effect of the mass on internal recorded temperature, measured by
means of the decrement factor, was investigated. All zones monitored over the twelve-month
period in Hotel Suraj showed considerable variation in daily decrement factor, with both the
magnitude, and variation in daily decrement factor increasing from the bottom to the top of
the haveli. Whilst the daily decrement factor of the 1st and 2nd floor zones showed a
relatively strong relationship with daily outdoor temperature range, the ground and basement
showed very little dependence. It is predicted that this situation would be reversed if
sufficient data was available to plot annual decrement factor against annual outdoor
temperature range, with the basement and ground zones showing relatively high dependency.
A plot of daily decrement factor against daily cycle time lag, for each monitored zone,
showed a very strong, quadratic relationship exists between the two variables. Whilst the
three calculated, and two theoretical, data points used in this analysis constitute an extremely
limited bases for profound generalisation, this result does indicate a need for further analysis
on similar long term temperature data sets.
167
9.6. Presentation of results

The problems associated with presenting large amounts of data, particularly if collected in
many locations over long periods of time, were discussed in chapter 6. Two stand-alone
computer programs were developed, as part of this study, in order to address these problems.
The first, Thermamator, was designed to animate temperature data over time, and the
second, Comfator, was designed to predict and present thermal comfort sensation levels,
based on the research outlined in chapter 5. Both pieces of software are available on the CD
enclosed at the back of this thesis.
As discussed in chapter 6 the initial feedback for both programs has been good.
Thermamator has been used by MSc. students in both the University of East London and the
Architectural Association schools of Architecture, and presented at PLEA 1998 (Matthews et
al. 1998), and seems particularly useful for the visualisation of surface temperature data.
Research staff in the Solar Energy department of the University of Siegan have also used
Thermamator, in particular for the visualisation of thermal climatic regions and the
behaviour of an earth heat exchanger.
The basic idea behind both Thermamator and Comfator - that of linking columns of data
directly to zones on a building image - was conceived in order to stimulate interest in indoor
thermal environment amongst students of Architecture. However, it is possible that both
programs could be used by architects as presentation / selling tools to clients, particularly
over the Internet, as payable files embedded in web sites.
9.7. In conclusion
This thesis has shown the importance of long term monitoring, particularly in case studies of
high mass buildings, such as the havelis of Jaisalmer. It has highlighted the many gaps in our
knowledge of the traditional architecture of this region, and the pressing need for field
studies on the subject. This would not only assist contemporary architects designing for such
climates, but also provide historical records of the fast dwindling traditional architecture of
the region.
Unfortunately, the havelis, like many traditional forms, are being replaced by modern,
energy consuming, mechanically cooled structures. They are fast disappearing before their
effectiveness has been properly assessed. The principles of good thermal design used in such
168
houses are equally valid today, and it should be possible for modern designers to incorporate
these design principles in buildings suitable for modern living, thus conserving energy whilst
still providing adequate thermal comfort. The following areas would particularly benefit
from further research:
Long term thermal monitoring, over a time period of at least 12 months, of high mass,
free-running buildings. Results from this study have consistently revealed the importance
of annual temperature cycles when analysing such building forms.
Further time series analysis of indoor temperature in relation to outdoor temperature,

with particular reference to time lag and decrement factor. An exponentially weighted
moving average method of analysis was adopted in chapter 8 of this study. However, it is
possible that other mathematical models would give an even stronger correlation
between indoor and outdoor temperature.
169
References Chapter 9
GUPTA, V. K. (1984). A study of the natural cooling systems of Jaisalmer. Unpublished
PhD. Thesis, Indian Institute of Technology, Delhi.
and practice, Vol. 6, No. 2, pp.92-105.
MATTHEWS, E. H., ROUSSEAU, P. G., RICHARDS, P. G. & LOMBARD, C. (1991). A
Procedure to estimate the effective heat storage capacity of a building. Building and
Environment, Vol. 26, No. 2, pp 179 - 188. Pergamon Press, UK.
MATTHEWS, J., NICOL, F., AND ROAF, S. (1998). Thermamtor - The thermal animation tool.
In: Environmentally friendly cities, proceedings of PLEA 98, Lisbon.
NICOL, F., JAMMY, G. AND SYKES, O. (1994). A survey of thermal comfort in Pakistan,
towards new indoor temperature standards, Oxford Brookes University.
ROAF, S. C. (1989). The windcatchers of Yazd, PhD thesis, Department of Architecture,
170
APPENDIX A.
TABULATED AIR MOVEMENT DATA.
90
180
270
Ill. A.1. Recording locations and horizontal movement direction key for Vyas-air data set.
172
29/03/96
am
F1
F2
F3
F4
F5
F6
F7
F8
F9
F10
F11
F12
F13
F14
F15
F16
F17
F18
G1
G2
S1
S2
S3
S4
S5
S6
S7
S8
S9
S10
aft
30/03/96
31/03/96
01/04/96
eve
am
aft
eve
am
aft
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
eve
am
aft
02/04/96
eve
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
am
aft
eve
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
Tbl. A.1. Status of openings in Hotel Suraj during air movement recording periods.
173
LOCATION TIME
KATA
COOLING AIR TEMP
29:3:96EVE
FACTOR (K) TIME (T)
(TA)
Outside
22:05
470
44
31.1
F1A
22:00
452
42
31.3
F1B
22:00
466
58
30.2
F1C
22:10
470
61
30.8
F2A
22:05
452
55
30.9
F2B
22:05
466
54
30.9
F2C
22:15
470
50
32.1
F3A
22:10
452
54
32.1
F3B
22:10
466
60
30.9
F3C
22:20
452
58
30.3
F4B
22:23
470
52
31.3
F5A
22:25
452
23
30.3
F5B
22:05
466
63
30.3
F5C
22:30
452
40
30.5
F6B
22:30
470
39
32.6
F7A
22:35
452
41
31.1
F7B
22:30
466
56
30.6
F7C
22:40
452
24
30.5
F8B
DIRECTION
VELOCITY
140
0
T
30
180
0
5
T
0
0
0
330
T
0
10
T
0
180
0
1.91
0.601349
0.637685
0.15503
0.143439
0.201291
0.251158
0.446659
0.280264
0.152913
0.13552
0.330146
2.581395
0.101698
0.666291
1.036738
0.67171
0.200353
2.387826
Tbl. A.2. Vyas-air results recorded in the evening on 29/3/96.
174
LOCATION
KATA
TIME
COOLING AIR TEMP
30:3:96MOR
FACTOR (K) TIME (T)
(TA)
Outside
9:35
466
35
26.3
F1A
9:32
470
37
26.3
F1B
9:35
452
40
24.6
F1C
9:38
466
32
26.2
F2A
9:35
470
20
26.3
F2B
9:40
452
45
26.6
F2C
9:40
466
33
26.6
F3A
9:40
470
49
26.6
F3B
9:45
452
46
26.6
F3C
9:43
470
36
27
F4B
9:45
466
58
27.7
F5A
9:45
470
16
27.7
F5B
9:50
452
48
26.6
F5C
9:47
470
32
27.4
F6B
9:51
466
33
27.7
F7A
9:50
470
20
27.7
F7B
9:50
452
57
27.6
F7C
10:03
466
35
29.6
F8A
9:55
470
17
27.6
F8B
9:55
452
24
27
G2
9.55
452
24
26.9
G3
9:57
452
30
26.7
G4
10:00
452
40
27
G5
10:10
470
36
26.7
S1B
10:15
452
33
26.6
S1C
10:15
466
43
28.6
S2A
10:12
470
40
26.6
S2B
10:20
452
33
26.8
S2C
10:15
470
23
26.9
S3B
10:25
452
35
26.8
S3C
10:20
466
60
37.9
S5B
10:20
470
26
27.8
S6B
10:27
466
32
32.1
S7B
DIRECTION
VELOCITY
200
300
T
10
130
180
180
100
180
90
0
4.0
0.650003
0.550021
0.272528
0.869226
2.681474
0.223662
0.825195
0.18027
0.202144
0.665898
0.087346
5.011705
0.164731
0.963898
0.948647
3.033683
0.078623
0.944067
4.340219
1.70075
1.685295
0.960316
0.391395
0.639797
0.744916
0.425753
0.429174
0.76432
2.043458
0.624117
0.723288
1.661789
1.550245
270
135
180
0
0
180
0
0
180
180
180
180
270
0
225
270
180
T
180
160
0
0
Tbl. A.3. Vyas-air results recorded in the morning on 30/3/96.
175
LOCATION
30:3:96AFT
Outside
F1C
F1B
F2C
F3C
F5C
F7C
G2
G3
G4
G5
S1C
S3C
F1A
F2B
F2A
F3B
F3A
F4B
F5B
F7A
F6B
F8A
F7B
F8B
S1B
S2B
S5B
S3B
S6B
TIME
3:10
3:15
3:15
3:18
3:25
3:28
3:33
3:35
3:35
3:40
3:50
3:51
15:13
15:18
15:19
15:24
15:28
15:30
15:35
15:38
15:39
15:42
15:42
15:43
15:57
16:00
16:05
16:10
16:50
KATA
FACTOR (K)
470
466
470
470
470
470
470
470
470
470
470
470
452
466
452
466
452
466
466
452
466
452
466
466
466
466
452
452
466
COOLING
TIME (T)
48
44
47
53
68
73
32
27
39
57
45
52
32
57
27
34
65
59
34
72
22
20
26
31
50
65
42
43
51
AIR TEMP
(TA)
29.5
29.4
29.5
29.5
30.1
30.9
30
30.1
29.9
29.7
30.9
30.6
30.5
30.2
30.1
30
30.2
30.5
30.6
30.8
31.1
30.2
31.1
31.3
31.4
30.9
30.4
30.1
30.9
DIRECTION
VELOCITY
180
0
0
290
90
10
5
180
180
180
0
30
180
350
315
T
60
140
315
210
T
180
270
180
180
300
180
150
260
T
3.0
0.330625
0.442922
0.36
0.215436
0.063073
0.049083
1.255691
1.906383
0.758982
0.158378
0.536893
0.289312
1.204384
0.169063
1.738228
1.059341
0.067642
0.152038
1.130822
0.038985
3.326325
3.51648
2.266191
1.528793
0.380415
0.099103
0.551837
0.482254
0.320277
Tbl. A.4. Vyas-air results recorded in the afternoon on 30/3/96.
176
LOCATION TIME
KATA
COOLING AIR TEMP DIRECTION
30:3:96EVE
FACTOR (K) TIME (T)
(TA)
210
Outside
22:45
452
41
27.3
180
F1B
22:52
452
49
27.5
180
F2B
22:55
452
52
27.9
180
F3B
22:55
470
45
27.6
45
F1C
22:58
470
45
28.7
160
F2C
23:00
470
55
29
90
F3C
23:03
470
57
29.7
180
F5C
23:04
452
39
27.7
T
F1A
23:05
470
59
29.9
180
F7C
23:08
452
36
28.1
180
F2A
23:14
470
62
29.6
330
F5A
23:15
452
46
28.4
180
F3A
23:21
470
47
29.1
0
F7A
23:23
470
37
29.1
20
F8A
23:26
452
41
29.3
0
F4B
23:27
470
25
28.6
0
F8A
23:29
470
44
27.8
330
F7B
23:30
452
50
29.1
0
F5B
23:36
470
37
30.2
190
F6B
VELOCITY
4.2
0.37171
0.176501
0.141817
0.31313
0.374631
0.163289
0.158378
0.478617
0.138745
0.675609
0.098323
0.277102
0.336303
0.807864
0.506606
1.967995
0.354437
0.217757
0.943081
177
LOCATION
KATA
TIME
31:3:96MOR
FACTOR (K) TIME (T)
(TA)
210
Outside
11:40
470
41
27.7
180
F1A
11:55
452
50
27.8
T
F1B
11:40
452
50
27.7
0
F1C
11:45
470
52
27.8
180
F2A
12:00
452
60
27.8
0
F2B
11:45
452
34
27.8
315
F2C
11:50
470
55
28.6
180
F3A
11:55
470
50
28.9
45
F3B
11:47
452
40
27.7
90
F3C
12:10
452
50
28.9
0
F4B
12:02
470
61
29
45
F5A
12:13
452
58
29
45
F5B
12:05
452
53
29
180
F5C
12:15
452
45
28.5
0
F6B
12:10
470
61
28.7
0
F7A
12:22
470
53
29.3
0
F7B
12:05
452
65
29
180
F7C
12:15
470
52
29.5
180
F8A
11:10
470
34
27.6
180
S1B
11:05
452
60
27.7
0
S1C
11:15
470
53
27.8
180
S2B
11:10
452
45
27.7
180
S3C
10:55
470
47
27.8
180
S6B
11:00
452
50
27.7
145
S7B
VELOCITY
2.2
0.459332
0.16943
0.166155
0.169289
0.059057
0.787967
0.150257
0.249917
0.434974
0.209574
0.093225
0.09966
0.161665
0.309287
0.086525
0.20718
0.047122
0.234852
0.873736
0.057448
0.153949
0.269962
0.26941
0.166155
178
LOCATION TIME
KATA
31:3:96AFT
FACTOR (K) TIME (T)
(TA)
205
Outdoors
15:40
452
44
29.6
0
F1C
15:42
452
36
29.9
0
F2C
15:45
452
53
30.3
0
F3C
15:48
452
52
30.2
0
F1B
15:50
452
57
30.8
30
F2B
15:54
452
58
30.6
45
F3B
15:58
452
47
30.5
0
F1A
16:02
452
52
30.6
180
F2A
16:04
452
62
30.6
0
F3A
16:05
452
63
30.5
T
F5C
16:08
452
51
30.8
T
F7C
16:10
452
58
30.8
0
F4B
16:14
452
59
30.6
180
F5B
16:12
470
70
30
T
F5A
15:45
470
34
29
100
G2
15:47
470
34
29
180
G3
15:50
470
34
29
0
G4
15:53
470
42
29
260
G5
15:56
470
48
29.7
T
F7A
16:00
470
61
30
T
F8A
16:05
470
39
30
0
F8B
16:07
470
47
29.9
0
F7B
16:09
470
47
29.7
180
F6B
VELOCITY
2.2
0.407038
0.870034
0.211106
0.225721
0.165895
0.145234
0.366428
0.244332
0.100997
0.089531
0.276474
0.152049
0.132836
0.049832
0.973376
0.973376
0.973376
0.511055
0.342325
0.118811
0.770168
0.385367
0.372467
179
LOCATION
1:4:96MOR
Outdoors
F1A
F2A
F3A
F1B
F2B
F3B
F1C
F2C
F3C
F5C
F7C
F4B
S1B
S2B
S3B
S6B
S7B
F7A
F8A
F8B
F7B
F6B
F5A
F5B
G2
G3
G4
G5
TIME
10:30
10:32
10:35
10:39
10:42
10:45
10:47
10:50
10:52
10:55
10:57
11:00
11:10
11:15
11:20
11:24
11:25
10:23
10:32
10:35
10:41
10:45
10:47
10:52
10:56
11:00
11:03
11:06
KATA
FACTOR (K)
452
452
452
452
452
452
452
452
452
452
452
452
452
452
452
452
452
470
470
470
470
470
470
470
470
470
470
470
COOLING
TIME (T)
53
44
63
51
60
62
55
52
58
46
61
56
56
50
57
48
43
60
59
29
36
44
61
64
34
53
48
56
AIR TEMP
(TA)
28.7
27.4
27.6
27.6
27.5
27.8
27.8
28
28.3
28.4
28.6
28.3
28.4
28.7
28.4
29.1
29.1
28.6
27.9
28.1
27.7
28
27.7
27.8
28.1
28
28
28
DIRECTION
VELOCITY
110
T
180
T
T
180
45
T
T
30
180
T
T
290
90
180
160
180
T
T
180
180
T
T
T
0
270
T
T
0.75
0.151869
0.282604
0.038632
0.147436
0.054335
0.046579
0.102178
0.144774
0.083949
0.277102
0.065732
0.103372
0.105795
0.201679
0.095499
0.261211
0.409964
0.093225
0.087023
1.315338
0.731374
0.365915
0.066982
0.049196
0.932299
0.160242
0.25452
0.120211
180
LOCATION
1:4:96EVE
F1A
F2A
F3A
F3B
F2B
F1B
F1C
F2C
F3C
F7C
F5C
F4B
F5B
F5A
F6B
F7B
F8B
F8A
F7A
TIME
22:37
22:40
22:45
22:47
22:50
22:55
22:58
22:00
22:05
22:48
23:00
23:04
23:04
23:10
23:16
23:17
23:22
23:26
23:31
KATA
FACTOR (K) TIME (T)
(TA)
470
60
29.9
0
470
55
30.1
0
470
60
30
150
470
65
30
0
470
52
30.1
180
470
58
30
180
470
65
30
180
470
57
30.2
180
470
56
30.2
180
452
60
29.9
0
452
64
30.1
T
452
51
30
90
452
53
30.1
0
452
43
30.3
T
452
43
29.7
0
452
45
29.5
0
452
19
29.5
90
452
54
29.6
90
452
36
30
T
VELOCITY
0.126933
0.204596
0.129913
0.082114
0.263203
0.154884
0.082114
0.176276
0.1919
0.102178
0.072741
0.236828
0.202689
0.498224
0.451883
0.366428
3.687785
0.167066
0.88251
181
LOCATION
2:4:96AFT
F1C
F2C
F3C
F3B
F2B
F1B
F1A
F2A
F3A
F4B
F5B
G2
G3
G4
G5
F7A
F8A
F8B
F7B
F6B
F7C
F5C
F5A
TIME
15:33
15:35
15:42
15:45
15:48
15:53
15:55
15:58
16:01
16:05
16:08
15:40
15:43
15:46
15:49
15:53
15:56
15:58
16:00
16:02
16:05
16:07
16:09
KATA
FACTOR (K) TIME (T)
(TA)
452
51
31.7
0
452
47
31.8
350
452
56
31.9
180
452
54
31.9
0
452
58
31.6
0
452
40
31.9
0
452
54
32.4
180
452
62
32.1
180
452
45
31.9
T
452
57
32.3
0
452
50
32.3
180
470
33
30.3
0
470
59
30.6
0
470
46
32.4
T
470
65
32.3
T
470
63
32.1
180
470
39
32.3
180
470
33
32.6
180
470
46
32.7
180
470
55
32.5
180
470
62
32
T
470
50
31.5
T
470
63
31.5
180
VELOCITY
0.328762
0.461374
0.229193
0.269044
0.182279
0.828531
0.297949
0.147203
0.551207
0.230683
0.398838
1.202702
0.162309
0.637908
0.148272
0.16533
0.999763
1.557649
0.671354
0.331224
0.174701
0.400601
0.143274
182
APPENDIX B.
VISUAL BASIC CODE - THERMAMATOR
FrmThermal
Private Sub cmdfwd_Click()

frmThermal.Data1.Database.Recordsets(0).MoveNext moves to the next row in the spreadsheet
SetFillColor Invokes public subroutine SetFillColor (see module section)
End Sub
Private Sub cmdplay_Click()

frmThermal.cmdstop.Enabled = True enables STOP button
frmThermal.cmdfwd.Enabled = False disables FWD button
frmThermal.cmdrwd.Enabled = False disables RWD button
cmdplay.Enabled = False disables PLAY button
Timer1.Enabled = True
enables timer
End Sub
184
Private Sub Form_Initialize()

maxTemp = 47
minTemp = 9
End Sub
Private Sub Form_Terminate()

End
End Sub
Private Sub mnuFileExit_Click()

Unload frmThermal
End
End Sub
Private Sub mnuFileLoadBmp_Click()

Dim response As Integer
declares 'response' as an integer variable
On Error GoTo fileerror

CommonDialog1.Filter = "Picture File (*.bmp)|*.bmp|" Sets common dialog box so that user can only enter
bitmap files
CommonDialog1.ShowOpen
TheBmpFile = CommonDialog1.filename Stores the entered filename in a public variable 'TheBmpFile'
frmThermal.picBuilding.Picture = LoadPicture(TheBmpFile) Loads 'TheBmpFile into the picture box on
frmThermal
mnuFileLoadTxt.Enabled = True Enables 'Load Excel v.5 File' option in the File menu
185
Exit Sub
fileerror:
response = MsgBox("Unable to load file. Check format before trying to load again", vbOKOnly +
vbInformation, "info")
End Sub
Private Sub mnuFileLoadTxt_Click()

CommonDialog1.Filter = "Excel database (*.xl*)|*.xl*|" Sets common dialog box so that user can only enter
Excel v.5 files
TheTxtFile = CommonDialog1.filename Stores the entered filename in a public variable 'TheTxtFile'
frmTempChannel.Show Opens frmTempChannel
End Sub
Private Sub mnuOptionsSpeed_Click()

frmSpeed.Show Opens frmSpeed
End Sub
Private Sub mnuOptionsTempRange_Click()

frmTempRange.Show
End Sub
Private Sub picBuilding_DblClick()
186
Const FLOODFILLBORDER = 0
Const FLOODFILLSURFACE = 1
frmThermal.picBuilding.FillColor = RGB(215, 217, 217) Sets floodfill colour to grey

crColor& = &H0 Sets border colour for flood fill to look for to black
wFillType% = FLOODFILLBORDER Sets floodfill type
x% = picBuilding.CurrentX Sets x co-ordinate for floodfill to start from
y% = picBuilding.CurrentY Sets y co-ordinate for floodfill to start from
tempX = picBuilding.CurrentX
tempY = picBuilding.CurrentY
Stores x co-ordinate in public variable tempX

Stores y co-ordinate in public variable tempY
suc% = ExtFloodFill(picBuilding.hdc, x%, y%, crColor&, wFillType%)
Invokes floodfill function
frmIDprompt.Hide
frmIDaccept.Show
End Sub
Private Sub picBuilding_MouseDown(Button As Integer, Shift As Integer, x As Single,

y As Single)
picBuilding.CurrentX = x Stores x co-ordinate of mouse click
picBuilding.CurrentY = y Stores y co-ordinate of mouse click
End Sub
Private Sub Slider1_Click()

frmThermal.txtTemp.Text = Slider1.Value
End Sub
Private Sub Timer1_Timer()

187
End Sub
FrmIDaccept
Private Sub cmdAccept_Click()

anchorX(c) = tempX
anchorY(c) = tempY
frmIDaccept.Hide
c=c+1
If c < Columns + 1 Then
frmIDprompt.Label1.Caption = "Please double click on " + frmThermal.Text1(c).Text + " zone"
frmIDprompt.Show
Else
frmIDaccept.Hide
frmTimeInt.Show
End If
End Sub
Private Sub cmdCancel_Click()

frmIDaccept.Hide
frmIDprompt.Show
End Sub
188
FrmSpeed
Private Sub cmdClose_Click()

frmSpeed.Hide
End Sub
Private Sub optspeed_Click(Index As Integer)

Select Case Index
Case 0
frmThermal.Timer1.Interval = 250 Sets timer interval to .25 seconds
Case 1
Case 2
frmThermal.Timer1.Interval = 1000 Sets timer interval to 1second
Case 3
frmThermal.Timer1.Interval = 2000 Sets timer interval to 2 seconds
Case Else
End Select
End Sub
189
FrmTempChannel

Unload cmdCancel.Parent
End Sub
Private Sub cmdEnter_Click()

Dim a As Integer

Columns = frmTempChannel.txtTempChannel.Text Stores the number of spreadsheet columns entered by the
user in a public variable 'Columns'
For a = 1 To Columns
Load frmThermal.Text1(a)
Creates columns no. of text boxes
Load frmThermal.Label1(a) Creates columns no. of labels

frmThermal.Text1(a).Visible = False makes text boxes invisible
frmThermal.Label1(a).Visible = False makes labels invisible
Next a
frmTempChannel.Hide
frmThermal.Data1.DatabaseName = TheTxtFile Links the Data Control to the userentered Excel database
frmThermal.Data1.Refresh Opens database
frmThermal.Data1.RecordSource = frmThermal.Data1.Database.TableDefs(0).Name Selects a table
containing selected field names from the data base
190
For b = 0 To Columns
frmThermal.Text1(b) = frmThermal.Data1.Database.TableDefs(0).Fields(b).Name Loads the first row of
data from the spreadsheet, cell by cell, into the previously created text boxes
Next b
frmThermal.Label1(a).DataField = frmThermal.Data1.Database.TableDefs(0).Fields(a).Name
Links
each column of data to a separate, previously created label.

Next a
frmIDprompt.Show
c=1
ReDim anchorX(Columns)
ReDim anchorY(Columns)
frmThermal.mnuFileLoadTxt.Enabled = False
Exit Sub
fileError:
End Sub
Private Sub txtTempChannel_KeyPress(KeyAscii As Integer)

Dim Key As String
Key = Chr(KeyAscii)
If (Key < "0" Or Key > "9") Then Checks for characters other than 0 through to 9
Beep
KeyAscii = 0
End If
191
End Sub
FrmTempRange

frmTempRange.Hide
End Sub
Private Sub cmdOK_Click()

On Error GoTo fileError
maxTemp = CInt(frmTempRange.txtMaxTemp.Text) rounds value to the nearest integer

minTemp = CInt(frmTempRange.txtMinTemp.Text)
frmThermal.Slider1.Max = maxTemp
frmThermal.Slider1.Min = minTemp
frmTempRange.Hide
Exit Sub
fileError:
MsgBox "Invalid value! Please try again", vbOKOnly + vbCritical, "STOP!"
End Sub
192
Private Sub txtMaxTemp_KeyPress(KeyAscii As Integer)

Dim Key As String
Key = Chr(KeyAscii)
If (Key < "0" Or Key > "9") Then
If Key <> "." Then
Beep
KeyAscii = 0
End If
End If
End Sub
Private Sub txtMinTemp_KeyPress(KeyAscii As Integer)

Dim Key As String
Key = Chr(KeyAscii)
If (Key < "0" Or Key > "9") Then
If Key <> "." Then
Beep
KeyAscii = 0
End If
End If
End Sub
Module
Public TheTxtFile As String
Public TheBmpFile As String
Public minTemp As Integer
Public maxTemp As Integer
193
Public Step As Long

Public c As Integer
Public anchorX() As Single
Public anchorY() As Single
Public tempX As Single
Public tempY As Single
Public Columns As Integer
Declare Function ExtFloodFill Lib "gdi32" (ByVal hdc As Long, ByVal X As Long, ByVal Y As Long, ByVal
crColor As Long, ByVal wFillType As Long) As Long
Public Function FixPath(InPath As String) As String

Dim T As String
T = InPath
If Right$(T, 1) <> "\" Then T = T + "\" Sticks a backslash on the end of Inpath if there is not one there
already
FixPath = T
End Function
Public Sub SetFillColor()

Dim a As Integer
On Error GoTo fileError
Step = (maxTemp - minTemp) / 38 divides temp scale into 38 equal bands
Sets colours for each temp band
If frmThermal.Label1(a).Caption = Null Then

frmThermal.picBuilding.FillColor = &HFF3939
ElseIf frmThermal.Label1(a).Caption < minTemp Then
frmThermal.picBuilding.FillColor = &HCA0202
ElseIf frmThermal.Label1(a).Caption < (minTemp + Step) Then
194
frmThermal.picBuilding.FillColor = &HEA0000
ElseIf frmThermal.Label1(a).Caption < (minTemp + (2 * Step)) Then
frmThermal.picBuilding.FillColor = &HFB0202
frmThermal.picBuilding.FillColor = &HFD0F0F
frmThermal.picBuilding.FillColor = &HFF1E1E
frmThermal.picBuilding.FillColor = &HFD3535
frmThermal.picBuilding.FillColor = &HFF4D4D
frmThermal.picBuilding.FillColor = &HFD6262
frmThermal.picBuilding.FillColor = &HFFA4A4
frmThermal.picBuilding.FillColor = &HFFB3B3
frmThermal.picBuilding.FillColor = &HFFBBBB
frmThermal.picBuilding.FillColor = &HFFC1C1
frmThermal.picBuilding.FillColor = &HFFDBDB
frmThermal.picBuilding.FillColor = &HFFE3E3
frmThermal.picBuilding.FillColor = &HE3E3FF
195

frmThermal.picBuilding.FillColor = &HDBDBFF
frmThermal.picBuilding.FillColor = &HC1C1FF
frmThermal.picBuilding.FillColor = &HBBBBFF
frmThermal.picBuilding.FillColor = &HB3B3FF
frmThermal.picBuilding.FillColor = &HA8A8FF
frmThermal.picBuilding.FillColor = &H9797FF
frmThermal.picBuilding.FillColor = &H8A8AFF
frmThermal.picBuilding.FillColor = &H4D4DFF
frmThermal.picBuilding.FillColor = &H1E1EFF
frmThermal.picBuilding.FillColor = &HD0DFF
frmThermal.picBuilding.FillColor = &HFF&
frmThermal.picBuilding.FillColor = &HFD&
frmThermal.picBuilding.FillColor = &HEA&
Else
frmThermal.picBuilding.FillColor = &HCC&
End If
196
crColor& = &H0
wFillType% = FLOODFILLBORDER
X% = anchorX(a)
Y% = anchorY(a)
suc% = ExtFloodFill(frmThermal.picBuilding.hdc, X%, Y%, crColor&, wFillType%)
Next a
Exit Sub
fileError:
MsgBox "You have reached the end of this data set.", vbOKOnly + vbInformation, "End of file"
frmThermal.Timer1.Enabled = False
End Sub
197
APPENDIX C.
VISUAL BASIC CODE - COMFATOR
FRMTHERMAL
Private Sub cmdfwd_Click()

frmThermal.Label1(Columns + 2).Caption = (((frmThermal.Label1(1).Caption) * Weight) +
((frmThermal.Label1(Columns + 1).Caption) * (1 - Weight))) calculates expo. smoothing of
label1(1), stores it in label1(columns +2)
frmThermal.Label1(Columns + 1).Caption = frmThermal.Label1(Columns + 2).Caption copies contents of
label1(columns +2) to label1(columns +1)
ComfTemp = (12.1 + ((frmThermal.Label1(Columns + 2).Caption) * 0.534)) Humphreys' comfort equation
End Sub
Private Sub cmdplay_Click()

frmThermal.cmdstop.Enabled = True enables STOP button
frmThermal.cmdfwd.Enabled = False disables FWD button
frmThermal.cmdrwd.Enabled = False disables RWD button
cmdplay.Enabled = False disables PLAY button
199

Timer1.Enabled = True
enables timer
End Sub
Private Sub cmdrwd_Click()

frmThermal.Data1.Database.Recordsets(0).MovePrevious moves to the previous row in the spreadsheet
End Sub
Private Sub cmdstop_Click()

frmThermal.Timer1.Enabled = False disables timer
frmThermal.cmdfwd.Enabled = True enables FWD button
frmThermal.cmdplay.Enabled = True enables PLAY button
frmThermal.cmdrwd.Enabled = True enables RWD button
frmThermal.cmdstop.Enabled = False disables STOP button
End Sub
Private Sub mnuFileExit_Click()

Unload frmThermal
End
End Sub
Private Sub mnuFileLoadBmp_Click()

declares 'response' as an integer variable
200

CommonDialog1.Filter = "Picture File (*.bmp)|*.bmp|" Sets common dialog box so that user can only enter
bitmap files
TheBmpFile = CommonDialog1.filename Stores the entered filename in a public variable 'TheBmpFile'
frmThermal.picBuilding.Picture = LoadPicture(TheBmpFile) Loads 'TheBmpFile into the picture box on
frmThermal
mnuFileLoadTxt.Enabled = True Enables 'Load Excel v.5 File' option in the File menu
Exit Sub
fileerror:
End Sub
Private Sub mnuFileLoadTxt_Click()

CommonDialog1.Filter = "Excel database (*.xl*)|*.xl*|" Sets common dialog box so that user can only enter
Excel v.5 files
TheTxtFile = CommonDialog1.filename Stores the entered filename in a public variable 'TheTxtFile'
frmTempChannel.Show Opens frmTempChannel
End Sub
Private Sub mnuOptionsSpeed_Click()

frmSpeed.Show Opens frmSpeed
End Sub
201
Private Sub picBuilding_DblClick()

frmThermal.picBuilding.FillColor = RGB(215, 217, 217) Sets floodfill colour to grey

crColor& = &H0 Sets border colour for flood fill to look for to black
wFillType% = FLOODFILLBORDER Sets floodfill type
x% = picBuilding.CurrentX Sets x co-ordinate for floodfill to start from
y% = picBuilding.CurrentY Sets y co-ordinate for floodfill to start from
tempX = picBuilding.CurrentX
tempY = picBuilding.CurrentY
Stores x co-ordinate in public variable tempX

Stores y co-ordinate in public variable tempY
suc% = ExtFloodFill(picBuilding.hdc, x%, y%, crColor&, wFillType%)
Invokes floodfill function
frmIDprompt.Hide
frmIDaccept.Show
End Sub
Private Sub picBuilding_MouseDown(Button As Integer, Shift As Integer, x As Single,

y As Single)
picBuilding.CurrentX = x Stores x co-ordinate of mouse click
picBuilding.CurrentY = y Stores y co-ordinate of mouse click
End Sub
Private Sub Slider1_Click()

frmThermal.txtTemp.Text = Slider1.Value
End Sub
202
Private Sub Timer1_Timer()

frmThermal.Label1(Columns + 2).Caption = (((frmThermal.Label1(1).Caption) * Weight) +
((frmThermal.Label1(Columns + 1).Caption) * (1 - Weight))) calculates expo. smoothing of
label1(1), stores it in label1(columns +2)
frmThermal.Label1(Columns + 1).Caption = frmThermal.Label1(Columns + 2).Caption copies contents of
label1(columns +2) to label1(columns +1)
ComfTemp = (12.1 + ((frmThermal.Label1(Columns + 2).Caption) * 0.534)) Humphreys' comfort equation
End Sub
FrmIDaccept
Private Sub cmdAccept_Click()

anchorX(c) = tempX
anchorY(c) = tempY
frmIDaccept.Hide
c=c+1
If c < Columns + 1 Then
frmIDprompt.Show
Else
frmIDaccept.Hide
frmTimeInt.Show
End If
End Sub
203

frmIDaccept.Hide
frmIDprompt.Show
End Sub
FrmSpeed
Private Sub cmdClose_Click()

frmSpeed.Hide
End Sub
Private Sub optspeed_Click(Index As Integer)

Select Case Index
Case 0
Case 1
Case 2
204
frmThermal.Timer1.Interval = 1000 Sets timer interval to 1second

Case 3
frmThermal.Timer1.Interval = 2000 Sets timer interval to 2 seconds
Case Else
End Select
End Sub
FrmTempChannel

End Sub

Dim a As Integer

Columns = frmTempChannel.txtTempChannel.Text Stores the number of spreadsheet columns entered by the
user in a public variable 'Columns'
For a = 1 To (Columns + 2)
Load frmThermal.Text1(a)
Creates columns+2 number of text boxes
Load frmThermal.Label1(a) Creates columns+2 number of labels

frmThermal.Text1(a).Visible = False Makes text boxes invisible
205
frmThermal.Label1(a).Visible = False Makes labels invisible

Next a
frmTempChannel.Hide
frmThermal.Data1.DatabaseName = TheTxtFile Links the Data Control to the userentered Excel database
frmThermal.Data1.RecordSource = frmThermal.Data1.Database.TableDefs(0).Name Selects a table
containing selected field names from the data base
For b = 0 To Columns
frmThermal.Text1(b) = frmThermal.Data1.Database.TableDefs(0).Fields(b).Name
Loads the first row of
data from the spreadsheet, cell by cell, into the previously created text boxes
Next b
frmThermal.Label1(a).DataField = frmThermal.Data1.Database.TableDefs(0).Fields(a).Name
Links
each column of data to a separate, previously created label.

Next a
frmThermal.Label1(Columns + 1).Caption = frmThermal.Label1(1).Caption Copies first outdoor temp. value

into label1(columns+1) (for calculating expo. smoothing)
frmThermal.Label1(Columns + 2).Caption = frmThermal.Label1(1).Caption
frmIDprompt.Show
c=1
ReDim anchorX(Columns)
ReDim anchorY(Columns)
frmThermal.mnuFileLoadTxt.Enabled = False
Exit Sub
fileerror:
End Sub
206
Private Sub txtTempChannel_KeyPress(KeyAscii As Integer)

Dim Key As String
Key = Chr(KeyAscii)
If (Key < "0" Or Key > "9") Then Checks for characters other than 0 through to 9
Beep
KeyAscii = 0
End If
End Sub
FrmTimeInt

End Sub

TimeInt = frmTimeInt.txtTimeInt.Text
Weight = 0.2 * TimeInt / 24 Sets exponential weight depending on time interval between logger recordings
207
frmThermal.cmdplay.Enabled = True
frmTimeInt.Hide
End Sub
Private Sub txtTimeInt_KeyPress(KeyAscii As Integer)
Dim Key As String
Key = Chr(KeyAscii)
If (Key < "0" Or Key > "9") Then Checks for characters other than 0 through to 9 and .
If Key <> "." Then
Beep
KeyAscii = 0
End If
End If
End Sub
Module, declarations
Public TheTxtFile As String
Public TheBmpFile As String
Public c As Integer
Public anchorX() As Single
Public anchorY() As Single
Public tempX As Single
Public tempY As Single
Public Columns As Integer
Public ComfTemp As Long
Public TimeInt As Long
Public HalfLife As Long
Public Weight As Long
Declare Function ExtFloodFill Lib "gdi32" (ByVal hdc As Long, ByVal x As Long, ByVal y As Long, ByVal
crColor As Long, ByVal wFillType As Long) As Long
208
Public Function FixPath(InPath As String) As String

Dim T As String
T = InPath
If Right$(T, 1) <> "\" Then T = T + "\" Sticks a backslash on the end of Inpath if there is not one there
already
FixPath = T
End Function
Public Sub SetFillColor()

Dim a As Integer
If frmThermal.Label1(a).Caption < (ComfTemp - 7.5) Then
frmThermal.picBuilding.FillColor = &HF23100
If temperature more than 7.5 deg C below ComfTemp
set floodfill colour to dark blue

ElseIf frmThermal.Label1(a).Caption < (ComfTemp - 4.5) Then
frmThermal.picBuilding.FillColor = &HFF8A8A
more than 4.5 deg C below comf temp - mid blue
ElseIf frmThermal.Label1(a).Caption < (ComfTemp - 1.5) Then

frmThermal.picBuilding.FillColor = &HFECDCD more than 1.5 deg C below comf temp - light blue
ElseIf frmThermal.Label1(a).Caption < (ComfTemp + 1.5) Then
frmThermal.picBuilding.FillColor = &HE6E6E6 less than 1.5 deg C above comf temp - light grey
frmThermal.picBuilding.FillColor = &HD0D0FF less than 4.5 deg C above comf temp - light red
frmThermal.picBuilding.FillColor = &H878AFE less than 7.5 deg C above comf temp - mid red
Else
frmThermal.picBuilding.FillColor = &H107FE more than 7.5 deg C above comf temp - dark red
End If
209
crColor& = &H0
wFillType% = FLOODFILLBORDER
x% = anchorX(a)
y% = anchorY(a)
suc% = ExtFloodFill(frmThermal.picBuilding.hdc, x%, y%, crColor&, wFillType%)
Next a
Exit Sub
fileerror:
MsgBox "You have reached the end of this data set.", vbOKOnly + vbInformation, "End of file"
frmThermal.Timer1.Enabled = False
End Sub
210
APPENDIX D
EXCEL MACRO FOR AUTOMATION
OF STEPPED CORRELATION
METHOD.
Sub MacroStepCorr()
'Outdoor temperature in column B, with heading in row 1'
'Indoor temperature in column C, with heading in row 1'
Dim i As Integer
For i = 1 To 400
'sets no. of stepped correlations performed'
Application.Run "ATPVBAEN.XLA!Mcorrel", ActiveSheet.Range("$B$1:$C$7154"), _
'Correlates column B1 to B7154 with column C1 to C7154'
ActiveSheet.Range("$E$1").Offset((3 * i), 0), "C", True
'Prints correlation result in column E, offsetting 3 rows per loop'
ActiveSheet.Range("$D$1").Offset((3 * i), 0).Value = I
'Prints step no. in column D, next to appropriate correlation result'
Range("C2").Select
'Selects columnC, row 2'
Selection.Delete Shift:=xlUp
'Deletes column C, row 2, shifts remaining column C cells up'
Next i
End Sub
212
BIBLIOGRAPHY
Bibliography
ALI, MRS M. H. (1973). Observations of the Mussulmauns of India, Karachi.
ALI, Z. F. (1998). Le Corbusier's Chandigarh from an environmental point of view.
Proceedings of PLEA '98, Lisbon, pp 183-186. James and James Science Publishers
Ltd.
ANAND, U. AND ANAND, V. (1982). Mansions of the sun. Al-Falak / Scorpion, London.
AULICIEMS, A. & DEDEAR, R. (1986). Air conditioning in Australia I; Human thermal
factors. Arch. Science Review, volume 29, pp67-75.
BALARAS, C.A. (1996). The role of thermal mass on the cooling load of buildings. An
overview of computational methods. Energy and Buildings, Vol. 24, pp. 1-10.
BALARAS, K. (1994). Thermal Comfort. In: Passive cooling of buildings. (Santamouris ed.)
Pp. 91-125. C.I.EN.E., Athens.
BERENSON, M. L. AND LEVINE, D.M. (1989). Basic business statistics, concepts and
applications. Prentice-Hall Internarional (UK) Limited, London.
BERGER, X. (1988). The pumping effect of clothing. Int. J. Ambient Energy, Vol. 9(1).
BOYLE, G. (1996). Renewable energy, power for a sustainable future. Oxford University
Press.
CARPENTER, S. (1994). Collapse of the golden city. In: New Scientist. Dec 17th.
CARPENTER, S. (1995). Return to the golden city. In: New Scientist. Dec. 23rd. pp.36-37.
CASSON SIR H. (1976). Prerface in: Colour for Architecture. Ed. Porter, T. and Mikellides,
B. Studio Vista.
COOK, J. (1980). Landscaping for microclimatic advantage in arid-zone housing. In:
Housing in Arid Lands (G. Golany, ed.), Chap. 19. Architectural Press.
DANBY, M. (1973). The design of buildings in hot dry climates and the internal environment.
Build International, vol. 6., no.1, pp55-76. London.
214
ELAWA, S. (1981). Housing design in hot arid zones with special reference to thermal
performance. Unpublished PhD. thesis, Department of Building Science, University
of Lund, Sweden.
ENERGY INFORMATION ADMINISTRATION. (1999). http://www.eia.doe.gov/emeu/cabs/india
/indiach1.htm. Current on 16/11/99.
FANGER, P. O. (1970). Thermal Comfort. Danish Techical Press, Copenhagen.
FATHY, H. (1972). The Arab house in the urban setting: Past, present and future. Longman,
University of Essex.
London.
GHOSH, B. (1975). A conservation plan for Jaisalmer. In Urban and Rural Planning Thought.
April, SPA, Delhi.
GIVONI, B. (1990). Performance and applicability of passive and low-energy cooling
systems. Energy and Building, vol.17, pp.177-199.
GIVONI, B. (IN PRESS). Minimum climatic information needed to predict performance of
passive buildings in hot climates.
GOETZ, H. (1978A). The Architecture of Jaisalmer. In: Rajput Art and Architecture (J. Jain,
GOLANY, G. (1980). Housing in arid lands. Architectural Press.
GRIFFITHS, I. (1990). Thermal comfort studies in buildings with passive solar features; field
studies: report to the Commission of the European Community, ENS35 090 UK.
GUPTA, V. K.(1984). A study of the natural cooling systems of Jaisalmer. Unpublished PhD
HARD A. (1969). The NCS Colour Order and Scaling System. Swedish Colour Centre,
Stockholm.
HOUGHTON, J. T., JENKINS, G. J. AND EPHRAUMS, J. J. (1990). Climate change: the IPCC
scientific assessment. Cambridge University Press.
HOUGHTON, J. T., CALLANDER, B. A. AND VARNEY, S.K. (1990). The supplementary report
to the IPCC scientific assessment. Cambridge University Press.
215
HUMPHREYS, M. (1974). Classroom temperature, clothing and thermal comfort - A study of

secondary school children in summer. Building Services Engineer , vol. 41, pp191202.
HUMPHREYS, M. A. (1992a). Thermal comfort requirements, climate and energy. World
.HUMPREYS, M. A. (1992b). Thermal comfort in the context of energy conservation. In:
Energy efficient building (Roaf, S.C. and Hancock, M. eds.). Blackwell Scientific
Pubications, Oxford.
HUMPHREYS, M. A. (1993). An adaptive model for thermal comfort. Notes for research
seminar, Oxford Brookes University.
HUMPHREYS, M. A. & NICOL, J. F. (2000, IN PRESS). Effects of measurement and
formulation error on thermal comfort indices in the ASHRAE database of field
studies. AHRAE transaction, volume 106, issue 2.
JAIN, K. (1998). Spatial organisation and aesthetic expression in the traditional architecture
of Rajasthan. In:Paradigms of Indian architecture. Space and time in representation
and design (G.H.R. Tillotson, ed.), pp. 159-175. Curzon Press, Surrey.
KAMIYA, T. (1999). http://www.asahi-net.or.jp/~wu3t-kmy/engl.htm. Current on 09/09/99.
KOSAMBI, D. (1975). An introduction to the study of Indian history. Popular Prakashan,
Bombay.
KRISHAN, A., AGNIHOTRI, M. R., JAIN, K., TEWARI, P., RAJAGOPALAN, M. (1995).
Climatically responsive energy efficient architecture - a design handbook. Volume II:
database - Indian context. CASA, School of Planning and Architecture, New Delhi.
LABS, K. (1981). Direct-coupled ground cooling: Issues and opportunities. In: Proceeds of
the International Passive and Hybrid Cooling Conference. Pp. 131-135. Miami
Beach, Florida.
LOS, S. (1981). Solar perspective: A graphic techique to draw buildings as seen by the sun.
In: Proceeds of the International Passive and Hybrid Cooling Conference. Pp. 131135. Miami Beach, Florida.
LUSCHER M, (1971). The Luscher Colour Test, Jonathon Cape, London.
216
MATTHEWS, E. H., ROUSSEAU, P. G., RICHARDS, P. G. & LOMBARD, C. (1991). A

Procedure to estimate the effective heat storage capacity of a building. Building and
Environment, Vol. 26, No. 2, pp 179 - 188. Pergamon Press, UK.
MATTHEWS, J., NICOL, F., AND ROAF, S. (1998). Thermamtor - The thermal animation tool.
In: Environmentally friendly cities, proceedings of PLEA 98, Lisbon.
MEHTA, R. (1977). Inside the Haveli. Delhi.
MISHRA, P. (1995). Researches in Indian Archaeology, Art, Arch., Culture and Religion.
NATH, A. & WACZIARG, F. ( 1997). Arts and Crafts of Rajasthan. Thames and Hudson.
NICOL, J. F. (1992). Time and thermal comfort: Proc Eorld Renewable Energy Congress,
Reading.
NICOL, J. F. (1993). A handbook of field studies toward an adaptive model. University of
East London.
HMSO, London.
NICOL, F., JAMY, G. AND SYKES, O. (1994). A survey of thermal comfort in Pakistan,
toward new indoor temperature standards, Oxford Brookes University.
NICOL, J. F., HUMPHREYS, M. A. AND RAJA, I. A. (1995). Developing indoor temperature
NICOL, F., RAJA, I. A., ALLAUDIN, A. AND JAMY, G. N. (IN PRESS). Climatic variations in
comfortable temperatures: The Pakistan projects.
NORTON H., DAVIS H., DAVIS P. (1995). Peter Norton's Guide to Visual Basic for Windows
95, Sam's Publishing, Indianapolis.
OLIVER, P. (1969). Shelter and Society. Barrie and Rockliff.
OLIVER, P. (1997). Encyclopaedia of vernacular architecture of the world. Volume 2:
Cultures and habitats. Cambridge University Press.
OLYGAY, V. and OLYGAY, A. (1992). Design with climate. Princetown University Press.
PARWAL, N. K. (1997). http://www.rajasthanweb.com/history/jaisal/index.htm Current on
10/10/97.
PER-OLOF, C., PER-OLOF, N., BURTON, L. M. M (1985). Passive indoor climate regulation
for buildings in hot climates. Stokholm.
PORTER, T. AND MIKELLIDES, B. (1976). Colour for Architecture. Studio Vista.
PRAMAR, V. S. (1989). Haveli: Wooden houses and mansions of Gujarat. Ahamedabad.
217
PRASAD, S. (1987). In the 21st century will the haveli have been completely destroyed? In:
Design, Oct.-Dec. pp.11-20.
PRASAD, S. (1988). The Havelis of North India. Unpublished PhD thesis, Royal College of
Art, London.
PRASAD, S. (1998). A tale of two cities: house and town in India today. In:Paradigms of
Indian architecture. Space and time in representation and design (G.H.R. Tillotson,
ed.), pp. 159-175. Curzon Press, Surrey.
RIGG, A. AND LAHAV, R. (1995). Concept and reality. In: Climatically responsive energy
efficient architecture A design handbook (Krishan, A. ed.), Volume 1, pp. 1-12 - 120. SPA, Delhi.
ROAF, S. (1989). The windcatchers of Yazd. PhD thesis, Department of Architecture,
ROAF, S. C. (1992). The wind catchers of the Middle East. Menas Press.
Press.
SHARMA & ALI (1986). Tropical Summer Index - a study of thermal comfort in Indian
subjects. Building and Environment, volume 21, issue 1, pp 11-24.
SHERRING, M. A. (1881). The tribes and castes of Rajasthan. London.
SHUKLA, D. C. (1978). Early history of Rajasthan. Bharatiya Vidya Prakashan, Delhi.
SIVIK L. (1974). Colour Meaning and Perceptual colour Dimensions: A Study of exterior
Colours, Goteborg Psychology Reports No. 11, University of Goteborg.
SMITH P. F. (1996). Options for a flexible Planet. The evidence, the policies and the possible
remedies. University of Sheffield
SODHA, M.S., KAUR, J. AND SAWHNEY, R.L. (1992). Effect of storage on thermal
performance of a building. International Journal of Energy Research, Vol. 16, pp.
697-707
SOMANI, R. V. (1990). History of Jaisalmer. Panchsheel Prakashan, Jaipur
SPATE, O. H. K. (1984). India and Pakistan: A general and regional geography.
STERLIN, H. & VOLWAHSEN, A. (1990)Architecture of the World - India. Germany.
TILLOTSON, G. H. R. (1987). The Rajput Palaces: The Development of an Architectural
Style 1450 - 1750. New Haven.
TILLOTSON, S. (1986). The tourist: Defender or destroyer. Preserving monumemts. The
India Magazine. March, Delhi.
TILLOTSON, S. (1994). Indian Mansions: A Social History of the Haveli. The Oleander
Press, Cambridge.
218
TOD, LIEUT. -COL. J. (1950). Annals and Antiquities of Rajasthan, 2 Vols., London.
TOMBAZIS, A. N. (1995). Architecture and bioclimatic design: Less is beautiful. In:
Climatically responsive energy efficient architecture A design handbook (Krishan,
A. ed.), Volume 1, pp. 1-1 - 1-11. SPA, Delhi.
UTEXAS (1997). http://www.lib.utexas.edu/libs/pcl/map_collection/middle_east_and_asia/
india_poli.jpg. Current on 24/10/97.
UTEXAS, (1997). http://www.lib.utexas.edu/libs/pcl/map_collection/middle_east_and_asia/
india_geo.jpg. Current on 24/10/97.
WACZIARG, F. AND AMAN, N. (1982). Rajasthan: The painted walls of Shekhavati. Delhi.
WARD, P. (1989). Rajasthan, Agra, Delhi: A travel guide. The Oleander Press, England.
Trans., Vol. 33, pp165-179
219

Thermal Comfort in The Havelis of Jaisalmer

Transféré par

Informations du document

Description originale:

Titre original

Copyright

Formats disponibles

Partager ce document

Partager ou intégrer le document

Options de partage

Avez-vous trouvé ce document utile ?

Ce contenu est-il inapproprié ?

Droits d'auteur :

Formats disponibles

Thermal Comfort in The Havelis of Jaisalmer

Transféré par

Droits d'auteur :

Formats disponibles

See

Thermal comfort in havelis of Jaisalmer.

Available from: Jane Matthews

THERMAL COMFORT IN THE HAVELIS OF

Thesis submitted in partial fulfilment of the requirements for the degree of

School of Architecture, University of East London, June, 2000

Global concerns ...

CHAPTER 3 - THE HAVELIS

CHAPTER 4 - ENVIRONMENTAL DATA COLLECTION

4.4.1.1 Logger locations

4.4.2.1 Logger locations

4.4.3.1 Logger locations

4.4.4.1 Logger locations

4.4.5.1 Temperature results

Courtyard surface temperatures - HAV-HT series.

Air movement - Vyas-Air series..

CHAPTER 5 THERMAL COMFORT IN JAISALMER

Thermal comfort in Hotel Suraj.....

Method used for calculating comfort levels in Hotel Suraj.......

Comfort level results...

CHAPTER 6 PRESENTATION OF RESULTS

Preparation of input files...

6.11.1 Preparation of input files....

CHAPTER 7 INDOOR TEMPERATURE TRENDS

Indoor / outdoor range.....

CHAPTER 8 MASS AS A THERMAL REGULATOR

1 approach Stepped correlation.....

2 approach Running mean

Method for conversion of exponential weight (W) to time lag.

Decrement factor ...

Thermal comfort ...

Appendix A - Tabulated air movement data.

Chapter 1 outlines a range of the global environmental problems facing

1.1 Global concerns

1.2. India's history.

1.3. India today.

*Urban population only

1.4 Traditional versus modern.

To relate recorded indoor temperatures to predicted human thermal comfort sensation.

1.6 Form of thesis

STERLIN, H. & VOLWAHSEN, A. (1990)Architecture of the World - India. Germany.

Chapter 2 gives a comprehensive description of the living fortress town of

2.1. Jaisalmer - Geography

Ill.2.2. Photograph of moving

Ill.2.1 Map of India showing location of Jaisalmer (utexas, 1997)

2.2 Jaisalmer - Climate

Ill. 2.3. Climatic zones of India (Krishan, 1995)

Ill. 2.4. Criteria for classification of climatic

Month av. max Ta

2.3 Jaisalmer - History

Ill. 2.5. The living fortress of Jaisalmer.

Ill.2.8. Faade of Patwon ki haveli* Ill. 2.9. Salim Singh haveli*

Ill. 2.10. Downtown Jaisalmer*

* reproduced from Kamiya (1999)

2.4. Jaislamer today

Possibly Persian - "a surrounded or enclosed place" (Wacziarg et al., 1982)

3.2. Building Surveys

Ill.3.1.Jaisalmer fort showing location of havelis, and courtyard house.

3.2.1. Hotel Suraj

Ill. 3.2 Street view of Hotel Suraj.

The haveli is three stories high. In addition there is

Ill. 3.4. Detail of brackets, first floor front room.