65
THERMAL COMFORT
Contents of this chapter, extensively deals with the Thermal Comfort problem
formulation and its assessment along with the respective literature survey:
Introduction to thermal comfort.
Fuzzy logic introduction and thermal comfort problem solving using fuzzy logic.
Quantitative analysis and optimization of thermal comfort in office buildings with
results.
Quantitative analysis and optimization of thermal comfort in resident buildings with
results.
3.1. INTRODUCTION.
Thermal comfort is highly subjective, not only is it subject to personal preference but
also to varying temperatures. Both internal and external temperatures sensing is integrated in
such a way that the resulting effect would either move towards restoring deep body
temperature or move away from it. A cold sensation will be pleasing when the body is
overheated, but unpleasant when the core is already cold. At the same time, the temperature
of the skin is by no means uniform. Besides variations caused by vasoregulation, there are
variations in different parts of the body, which reflect the differences in vasculation and
subcutaneous fat. The wearing of clothes also has a marked effect on the level and
distribution of skin temperature.
Thermal comfort for human is one of the major problems at present. Providing
thermal comfort for occupants in buildings is really a challenging task because thermal
comfort is not only influenced by temperature but also factors like relative humidity, air
velocity, environment radiation, activity level and cloths insulation. These entire six variables
play a major role in providing thermal comfort.
Thermal comfort can be calculated by an equation called Fanger‘s ‗Predicted Mean
Vote‘ (PMV) as given by Fanger. This equation gives the optimal thermal comfort for any
activity level, clothing insulation and for all combinations of the environmental variables
such as air temperature, air humidity, mean radiant temperature and relative air velocity.
66
Human thermal comfort is defined by ASHRAE as the state of mind that expresses
satisfaction with the surrounding environment (ASHRAE Standard 55). Maintaining thermal
comfort for occupants of buildings or other enclosures is one of the important goals of design
engineers.
Thermal comfort is maintained when the heat generated by human metabolism is
allowed to dissipate, thus maintaining thermal equilibrium with the surroundings. Any heat
gain or loss beyond this, generates a sensation of discomfort. It has long been recognized that
the sensation of feeling hot or cold is not just dependent on air temperature alone.
The problem that we are going to deal with here is the thermal comfort of offices and
homes which use natural ventilation only.
3.1.1. IMPORTANCE OF THERMAL COMFORT
Thermal comfort is very important to many work-related factors. It can affect the
distraction levels of the workers, and in turn affect their performance and productivity of their
work. Besides, thermal discomfort has been known to lead to Sick Building Syndrome
symptoms. The US Environmental Protection Agency's Building Assessment Survey and
Evaluation Study found that higher indoor temperatures, even within the recommended
thermal comfort range, increased worker symptoms. The occurrence of symptoms increased
much more with raised indoor temperatures in the winter than in the summer due to the larger
difference created between indoor and outdoor temperatures.
3.1.2. FACTORS DETERMINING THERMAL COMFORT
Metabolism
Clothing Insulation
Relative Humidity
Air temperature
Mean radiant temperature
Air velocity
67
3.1.2.1. Metabolism.
While measuring metabolism rates, many factors have to be taken into account. Each
person has a different metabolism rate, and these rates can fluctuate when a person is
performing certain activities, or when he is under certain environmental conditions. Even
people who are in the same room can feel significant temperature differences due to their
differing metabolic rates, which makes it very hard to find an optimal temperature for
everyone in a given location. (Khodakarami, 2009); (Smolander, 2002); (Toftum, 2005).
3.1.2.2. Clothing Insulation.
During cold weather, layers of insulating clothing can help keep a person warm. At
the same time, if the person is doing a large amount of physical activity, many layers of
clothing can prevent heat loss and consequently lead to overheating. Generally, the thicker
the garment is, the greater insulating abilities it has. Depending on the type of material the
clothing is made out of, air movement and relative humidity can decrease the insulating
ability of the material.
The amount of clothing is measured against a standard amount that is roughly
equivalent to a typical business suit, shirt, and undergarments. Activity level is compared to
being seated quietly, as in a classroom. Clo units can be converted to R-value in SI units
(m²·K/W) or RSI) by multiplying Clo by 0.155 (1 Clo = 0.155 RSI). (In English units 1 clo
corresponds to an R-value of 0.88 °F·ft²·h/Btu.)
3.1.2.3. Relative Humidity.
The human body has sensors that are fairly efficient in sensing heat and cold, but they
are not very effective in detecting relative humidity. Relative humidity creates the perception
of an extremely dry or extremely damp indoor environment. This can then play a part in the
perceived temperature and their thermal comfort. The recommended level of indoor humidity
by ASHRAE is in the range of 30-60%.
A way to measure the amount of relative humidity in the air is to use a system of dry-
bulb and wet-bulb thermometers. A dry-bulb thermometer measures the temperature not
relative to moisture. This is generally the temperature reading that is used in weather reports.
68
In contrast, a wet-bulb thermometer has a small wet cloth wrapped around the bulb at its
base, so the reading on that thermometer takes into account water evaporation in the air. The
wet-bulb reading will thus always be at least slightly lower than the dry bulb reading. The
difference between these two temperatures can be used to calculate the relative humidity. The
larger the temperature differences between the two thermometers, the lower the level of
relative humidity.
The wetness of skin in different areas also affects perceived thermal comfort.
Humidity can increase wetness on different areas of the body, leading to a perception of
discomfort. This is usually localized in different parts of the body. The local thermal comfort
limits for local skin wetness differ between different skin locations of the body. The
extremities are much more sensitive to thermal discomfort from wetness than the trunk of the
body. Although local thermal discomfort can be caused from wetness, the thermal comfort of
the whole body will not be affected by the wetness of certain parts.
Recently, the effects of low relative humidity and high air velocity were tested on
humans after bathing. Researchers found that low relative humidity engendered thermal
discomfort as well as the sensation of dryness and itching. It is recommended to keep relative
humidity levels higher in a bathroom than other rooms in the house for optimal conditions.
3.1.3. THERMAL STRESS
The concept of thermal comfort is closely related to thermal stress. This attempts to
predict the impact of air movement, and humidity for military personnel undergoing training
exercises or athletes during competitive events. Values are expressed as the Wet Bulb Globe
Temperature or Discomfort Index. Generally humans do not perform well under thermal
stress. People‘s performances under thermal stress are about 11% lower than their
performance at normal thermal conditions (Hancock, Ross, & Szalma, 2007)l; (Leon, 2008).
Also, human performance in relation to thermal stress varies greatly by the type of task a
person is completing. Some of the physiological effects of thermal heat stress include
increased blood flow to the skin, sweating, and increased ventilation.
69
3.1.4. EFFECTS OF NATURAL VENTILATION.
Many buildings use a HVAC (Heating Ventilation Air Conditioning) unit to control
their thermal environment. Recently, with the current energy and financial situation, new
methods for indoor temperature control are being used. One of these is natural ventilation.
This process can make the controlled indoor air temperature more susceptible to the outdoor
weather, and during the seasonal months, the temperatures inside can become too extreme.
During the summer months, the temperature inside can rise too high and cause the need for
open windows and fans to be used. In contrast, the winter months could call for more
insulation and layered clothing to deal with the less than ideal temperatures.
3.1.5. OPERATIVE TEMPERATURE
The ideal standard for thermal comfort can be defined by the operative temperature.
This is the average of the air dry-bulb temperature and of the mean radiant temperature at the
given place in a room. In addition, there should be low air velocities and no 'drafts,' little
variation in the radiant temperatures from different directions in the room, and humidity
within a comfortable range.
The operative temperature intervals varied by the type of indoor location. ASHRAE
has listings for suggested temperatures and air flow rates in different types of buildings and
different environmental circumstances.
3.1.6 .THERMAL SENSITIVITY OF INDIVIDUALS
The thermal sensitivity of an individual is quantified by the descriptor FS, which takes
on higher values for individuals with lower tolerance to non-ideal thermal conditions. This
group includes pregnant women, the disabled, as well as individuals whose age is above 14 or
below 60, which is considered the adult range. Existing literature provides consistent
evidence that sensitivity to hot and cold surface declines with age and that there is also a
gradual reduction in the effectiveness of the body in thermoregulation after the age of 60.
This is mainly due to a more sluggish response of the counteraction mechanisms in the body
that are used to maintain the core temperature of the body at ideal values.
70
Situational factors include the health, psychological, sociological and vocational
activities of the persons. Restaurant employees often have the air-conditioner temperature to
suit themselves, rather than the resting clients or incoming new customers from the
temperature outside the building.
3.1.7. GENDER DIFFERENCES
While thermal comfort preferences between genders seem to be small, there are some
differences. Females are much more likely to be sensitive to thermal conditions. Females are
also more likely to be uncomfortable with the room temperature, and will find the
temperature too hot or too cold before many men would. Many times, females will prefer
higher temperatures. But while females were more sensitive to temperatures, males tend to be
more sensitive to relative humidity levels.
3.2. PREDICTED MEAN VOTE
A large number of thermal comfort indices have been set up for the analysis of indoor
climates and the design of HVAC systems. But only a few of them have been used to
evaluate the ability of an existing room climate to create satisfactory thermal conditions for
occupants. The most common and best understood one is Fanger‘s ‗Predicted Mean Vote‘
(PMV).
3.2.1. THERMAL SENSATION INDEX
For many years, it has been desirable to determine directly human‘s thermal sensation
in a given environment condition and for a specified activity level and clothing insulation.
Until the 60‘s, thermal comfort calculation was limited by the lack of a well-defined unit to
represent the degree of the thermal sensation. Such a unit appeared in 1970 when Fanger
defined the PMV ‗Predicted Mean Vote‘ as the index that gives the expected degree of
thermal comfort in relation to all the above-mentioned six thermal parameters. Besides that,
Fanger presented a general comfort equation which describes the conditions under which the
average sensation of a large group of people will feel thermal neutrality. He defined the
thermal neutrality of a person as the condition of mind in which the subject would prefer
neither warmer nor cooler surroundings Eq. (3.1) represents the comfort equation proposed
by Fanger
71
(3.1)
(3.2)
hc=
The parameters are defined as follows:
PMV: Predicted mean vote.
M: Metabolism (W/m2).
W: External work, equal to zero for most activity (W/ m2).
Icl: Thermal resistance of clothing (Clo).
Fcl: Ratio of body‘s surface area when fully clothed to body‘s surface area when nude.
Ta: Air temperature (0C).
Tmrt: Mean radiant temperature (0C).
Vair: Relative air velocity (m/s).
Pa: Partial water vapour pressure (Pa).
Hc: Convection heat transfer coefficient (W/m 2 k)
Tcl: Surface temperature of clothing (0C).
72
3.3. FUZZY THERMAL SENSATION INDEX
3.3.1. LITERATURE SURVEY
The occupants‘ thermal comfort sensation is addressed here by the well-known
comfort index (PMV - Predicted Mean Vote). In this context, different strategies for the
control algorithms are proposed by using only-one-actuator system that can be associated
with a cooling and/or heating system. The first strategy is related to the thermal comfort
optimization and the second one includes energy consumption minimization while
maintaining the indoor thermal comfort criterion in an adequate level (Freire, Roberto,
H.C.Oliveira, & Nathan.Mendes, 2008). Temperature in an automobile cabin is an important
factor in the occurrence of traffic accidents . A better climate control system in an automobile
improves thermal comfort which results in increased driver caution and thereby improves
driving performance and safety in different driving conditions. Thermal loads while
minimizes energy consumption. The compressor used in a cooling system is driven by
automobile engine and it therefore increases the fuel consumption. Manual control requires
skill and experience about system. Automatic control frees the driver from this task. (Wang &
Mendel, 1992) (Xia, R.Y, & Zhao, 1999) designed two controllers for controlling the indoor
air temperature of a car, the general fuzzy controller and the state feedback with weighting
fuzzy controller. By comparing the results of the two experiments, they showed that the state
feedback with weighting fuzzy controller is more efficient than the other in controlling the
automobile indoor air temperature. Fuzzy PMV is used instead of Fanger‘s PMV. PMV index
is used to show thermal comfort and its variables are simplified. With this simplification, if
the inside cabin temperature and air velocity are known, a good prediction of comfort can be
obtained. Two fuzzy controllers with temperature feedback and PMV feedback are designed.
An index for energy consumption is also suggested. It is shown that controller with PMV
feedback is more effective than controller with temperature feedback. The PMV controller
also better minimizes the energy consumption.
A new approach based on fuzzy logic to estimate the thermal comfort level
depending on the state of the following six variables: the air temperature, the mean radiant
temperature, the relative humidity, the air velocity, the activity level of occupants and their
clothing insulation. The new fuzzy thermal sensation index is calculated implicitly as the
consequence of linguistic rules that describe human‘s comfort level as the result of the
interaction of the environmental variables with the occupant‘s personal parameters. The fuzzy
73
comfort model is deduced on the basis of learning Fanger‘s ‗Predicted Mean Vote‘, PMV
equation. Unlike Fanger‘s PMV, the new fuzzy PMV calculation does not require an iterative
solution and can be easily adjusted depending on the specific thermal sensation of users.
These characteristics make it an attractive index for feedback control of HVAC systems.
(Hamdi, Lachiver, & Michaud, 1999). Two fuzzy controllers one with temperature as its
feedback and the other PMV index as its feedback are designed. Results show that the PMV
feedback controller better controls the thermal comfort and energy consumption than the
system with temperature feedback (YadollahFarzanth & Tootoonchi, 2008). The occupants‘
thermal comfort sensation is addressed here by the well-known comfort index (PMV -
Predicted Mean Vote). In this context, different strategies for the control algorithms are
proposed by using only-one-actuator system that can be associated with a cooling and/or
heating system. The first strategy is related to the thermal comfort optimization and the
second one includes energy consumption minimization while maintaining the indoor thermal
comfort criterion in an adequate level (Freire, Roberto, H.C.Oliveira, & Nathan.Mendes,
2008).
(Soyguder & Ali, 2009)The optimal values of PID parameters were obtained by using
Fuzzy sets. Fuzzy adaptive control has been performed to maximize the performance of the
system. Efficiency of fuzzy adaptive control (FAC) developed method was successfully
obtained (Soyguder & Ali, 2009). Fuzzy logic offers a promising solution to this conceptual
design through fuzzy modelling. Numerous fuzzy logic studies are available in the non-
mechanical engineering field and allied areas such as diagnostics, energy consumption
analysis, maintenance, operation and its control. Relatively little exists in using fuzzy logic
based systems for mechanical engineering and very little for HVAC conceptual design and
control (Soyguder & Ali, 2009). During the past several years, fuzzy control has emerged as
one of the most active and fruitful areas for research in the applications of fuzzy set theory,
especially in the realm of industrial processes, which do not lend themselves to control by
conventional methods because of a lack of quantitative data regarding the input-output
relations. Fuzzy control is based on fuzzy logic—a logical system which is much closer in
spirit to human thinking and natural language than traditional logical systems. The fuzzy
logic controller (FIX) based on fuzzy logic, provides a means of converting a linguistic
control strategy, based on expert knowledge, into an automatic control strategy (MinNing &
Zaheeruddin, 2010). (Hamdi, Lachiver, & Michaud, 1999). The thermal comfort of the
occupants of a building depends on many factors including metabolic rates, clothing, air
74
temperature, mean radiant temperature, and air velocity and humidity. In most buildings,
however, only temperature and humidity can be controlled. Indeed, in many European
buildings, over a wide range of humidity, only zone temperature is controlled. In such cases,
the control objective is to maintain the zone temperature within a pre-defined range
(Thompson & Dexter, 2009).
Low operational efficiency especially under partial load conditions and poor control
are some reasons for high energy consumption of heating, ventilation, air conditioning and
refrigeration (HVAC&R) systems. In order to improve energy efficiency, HVAC&R systems
should be efficiently operated to maintain a desired indoor environment under dynamic
ambient and indoor conditions (Ning & Zaheeruddin, 2010).
A new approach based on fuzzy logic is introduced to estimate the thermal comfort
level, depending on the state of the following six variables: the air temperature, the mean
radiant temperature, the relative humidity, the air velocity, the activity level of occupants and
their clothing insulation.
New fuzzy thermal sensation index is calculated implicitly as the consequence of
linguistic rules that describe human‘s comfort level as the result of the interaction of the
environmental variables with the occupant‘s personal parameters. The fuzzy comfort model is
deduced on the basis of learning Fanger‘s ‗Predicted Mean Vote‘ (PMV) equation. The new
fuzzy PMV calculation does not require an iterative solution like Fanger‘s PMV and can be
easily adjusted depending on the specific thermal sensation of users. These characteristics
make it an attractive index for feedback control of HVAC systems.
Since the involved heat transfer processes are relatively complicated, the
mathematical expression derived for the calculation of the PMV is complicated and not
suitable for feedback control systems (Fanger.P.O, 1970)
; (Federspiel, 1882) (Int-Hout, 1990)
(Sherman, 1985). In order to overcome these problems, Fanger and ISO proposed to use
Tables and diagrams to simplify the determination of PMV in practical applications. Other
researchers proposed to use simplified models of PMV to avoid the iterative process. Such
thermal sensation indexes have been proposed in Refs. (Auliciems.A, 1984) (Coome, Gan,G,
& Awbi,H.B, 1992) (Culp, Rhodes, Krafthefer, & Listvan, 1993) (Federspiel, 1882)
(Fountain, Brager, Arens, Bauman, & Benton, 1994) (Gan & Croome, 1994) (Sherman,
75
1985) and they are deduced after significant modifications of Fanger‘s comfort equation.
Sherman (Sherman, 1985) proposed a simplified model of thermal comfort, based on the
original work of Fanger. In order to reach his objective, and to calculate the value of PMV
without any iteration, he linearized the radiation exchange terms to remove the T4
dependence
on temperature. Then, he simplified the convection coefficient to eliminate the iterative
solutions and finally, he used the dew point temperature instead of relative humidity to avoid
its dependence on air temperature. Sherman (Sherman, 1985) indicated that these
simplifications should not affect the precision of the PMV calculation only when the
occupants are near the comfort zone.
Based on the above mentioned assumptions, Sherman concluded that the resulting
simplified index could be computed explicitly in a compact form. However, Sherman‘s
thermal sensation index was not linearly parameterized (Federspiel, 1882) and therefore, not
suitable for on-line calibration and could not be used in a control algorithm of HVAC
systems. Federspiel proposed another thermal sensation index (V), that is, a modification of
Fanger‘s PMV index (Federspiel, 1882). To simplify the derivation, he supposed that the
radioactive exchange and the heat transfer coefficient are linear. In addition, it was assumed
that the bodily heat production and the clothing insulation are constant. In addition,
Federspiel supposed that the occupants are in a thermal neutrality condition. All these
assumptions were applied to the derivation of a thermal sensation index, that is an explicit
and linearly parameterized function of the four environmental variables. However, it was
outlined that (V) becomes non-linear as soon as the activity level or the clothing insulation
are changed. This problem limits the use of the thermal sensation index (V) in zones, where
the two above-mentioned factors are changing in time.
Other researchers recognized that the above-cited assumptions are difficult to reach
and the simplification of the original PMV model results in a significant error when they are
not respected (ASHRAE, 1989) (Federspiel, 1882). At present, the challenge is to derive a
thermal sensation index, based on the original work of Fanger without any simplification and
which can be used in feedback control applications with an on-line calibration. With this
goal in mind, the derivation in this project results in an index that does not require any
iteration solutions because it is implicitly dependent on the state of the air temperature, the
mean radiant temperature, the air velocity, the relative humidity, the activity level occupants
and the thermal resistance of their clothing. Since Fanger‘s thermal comfort model itself is
76
proposed to use, it was not necessary—at the expense of the accuracy—to simplify the
comfort model that makes calculations easier. It is difficult to justify control schemes, based
on a simplified model of thermal comfort, where it is necessary to verify the credibility of the
model simplifications. However, the proposed thermal comfort model can be used as a new
indoor climate high-level performance control variable of the HVAC systems without any
simplification of the original work of Fanger, and it does not require iterative solutions. At
present, such a thermal sensation index has not yet been developed.
3.3.2. PROBLEM FORMULATION
The aim is to derive a thermal sensation index, based on the original work of Fanger
that can be used in feedback HVAC control applications with an on-line calibration and
without requiring any simplification. This work presents a new strategy for the design of an
accurate thermal sensation index that does not require any iteration solutions and that can be
used as a high level performance variable in the control of HVAC systems. According to the
state of the six parameters that affect human‘s thermal comfort, the proposed thermal
sensation index can be calculated directly on the basis of knowledge gleaned from the
original work of Fanger. In order to reach this objective, fuzzy logic modelling, which is
defined as a method of describing characteristics of systems, using fuzzy reasoning, is used to
approximate the human‘s comfort/discomfort level in a given indoor climate. By analysing
the influence of the individual variables on the thermal sensation index, it becomes possible
to evaluate linguistically, how each variable influences the thermal sensation.
It was shown that it is impossible to consider the effect of the six variables on the
human thermal sensation independently, as the effect of each of them depends on the level
and the state of the other variables; the thermal comfort level is a complicated result of the
interaction of the six variables (Fanger.P.O, 1970). On the basis of this analysis, a new fuzzy
PMV is designed and a general fuzzy rule base is derived to describe the state of human‘s
thermal sensation. For any activity level and any clothing, the fuzzy rule base is able to
calculate all combinations of the four environmental variables which will create optimal
thermal comfort. The newly designed thermal sensation index can then be easily used in
feedback control of HVAC systems.
77
3.3.3. FUZZY PMV
Ensuring thermal comfort of occupants and reducing energy consumption are the
control challenges in the development of modern industrial HVAC systems. For thermal
comfort and energy efficiency, it is desirable to design a HVAC control system that can
guarantee high performance and good robustness with regard to the variation of the
environmental variables as well as the activity level of occupants and their clothing
insulation. Research has shown that it is possible to reach these objectives if HVAC control
strategies are based on the thermal sensation index instead of air temperature alone
(Auliciems.A, 1984) ; (Fanger.P.O, 1970) ; (Federspiel, 1882) ; (Fountain, Brager, Arens,
Bauman, & Benton, 1994) ; (MacArthur, 1986) ; (Sherman, 1985). Presently, the non-linear
behaviour of human‘s thermal sensation and the unavailability of a direct quantitative PMV
regarding the inputs–output relations make it very difficult or impossible to design a direct
control strategy of HVAC systems that regulate thermal comfort levels.
Fig.3.1. PMV and thermal sensation.
To overcome this problem, the thermal sensation index as shown in fig.3.1 should be
calculated as an implicit result of the six previously mentioned variables influencing human‘s
thermal sensation. Fuzzy logic theory was proposed to make quick and direct calculation of
the thermal comfort level in a given indoor climate. The new fuzzy thermal sensation index
(fuzzy PMV) can be designed by extracting knowledge from Fanger‘s comfort equation and
by transforming it into rules and membership functions. The basic design idea is to transform
all possible combinations of the variables that affect thermal comfort into linguistic fuzzy
78
implications to describe the thermal sensation index. This is done that, the input–output
relationships are transformed into a set of fuzzy rules and the human thermal sensation is
evaluated as a result of a fuzzy evaluation of the state of the six input variables that affect
thermal comfort. Instead of using Eq. (3.1) to calculate a PMV value, it becomes possible to
calculate it directly by using some linguistic rules such as:
IF the air temperature (Ta) is High,
AND relative air velocity (Vair) is Very small,
AND radiant mean temperature (Tmrt) is Close to air temperature,
AND the activity level (MADu) is Low,
AND the clothing (Ic1) is Very light,
THEN PMV is near zero (the indoor climate is comfortable).
While the six input variables are described by a set of fuzzy terms, the above-
presented design strategy requires a high number of fuzzy rules and thus, a large amount of
time calculation. This number of rules can be reduced significantly by considering that the
fuzzy PMV model is composed of two subsystems: the personal-dependent model and the
environmental model.
Their interconnection is shown in Fig.3.2. On one side, the personal-dependent model
evaluates the air temperature range (∆Ta), in which the predicted mean vote is found to be
close to zero. ∆Ta is evaluated, depending on the state of the occupants‘ activity level and
their clothing insulation. On the other side, the environmental model calculates the PMV
value according to the state of T and the four environmental variables.
79
Fig. 3.2 Architecture of the fuzzy thermal sensation index.
3.3.3.1 THE FUZZY PMV CALCULATION
The design strategy of the fuzzy thermal sensation index is achieved in four steps
process. First, the input and output variables are chosen. As shown in Fig. 3.2, the personal-
dependent subsystem input variables are the activity level of occupants and their clothing
insulation. Its output variable is the ambient temperature range in which the predicted mean
vote is close to zero. The environmental-dependent subsystem input variables are air
temperature (Ta), air velocity (Vair), mean radiant temperature (Tmrt) and relative air humidity
(RH). The output variable is the value of the predicted mean vote (fuzzy PMV).
The second design step is to derive the fuzzy rule base that should be used to evaluate
the PMV, depending on the state of the input variables. The general method developed by
(Wang & Mendel, 1992) is used to generate an accurate fuzzy rule base by extracting
knowledge from Fanger‘s thermal sensation vote equation. To generate all fuzzy rules that
represent all possible combinations of the six variables, each of the input and the output
spaces are divided into symmetric triangular membership functions as shown in Fig. 3.3.
80
The activity level and the clothing insulation are described by three and four
triangular membership functions, respectively for accuracy purposes. On the other side, the
air velocity, the air temperature and the predicted mean vote are transferred into fuzzy subsets
by using seven triangular membership functions to describe each of them. For instance, the
relative humidity is supposed to be 50% and the mean radiant temperature is supposed to be
close to the ambient air temperature. Since thermal comfort can be obtained by many
different combinations of the six above-mentioned variables, a conflict among the generated
rules appears. This is due to the interdependence between thermal comfort influencing factors
as the effect of each of them depends on the level and the conditions of the other factors. In
order to resolve this conflict, we assigned a degree to each of the generated rules to keep
only the rule from a conflict group that has maximum degree. In this way, not only the
conflict problem is resolved, but also the number of rules is greatly reduced (Wang &
Mendel, 1992)..
Fig. 3.3 Initial membership functions.
81
3.3.3.2 THE PERSONAL-DEPENDENT MODEL RULES.
Studies of the influence of the input variables on the thermal sensation index [15, 19]
demonstrated the powerful dependence between the personal-dependent variables and the
environmental variables. Since the variation of Icl and MADu affects the body heat
production and consequently, the mean temperature of the outer surface of the clothed body
(Tcl), the relative influence of the activity level and the clothing insulation on the thermal
comfort is assembled in the personal-dependent model. This is done to evaluate the air
temperature range that should ensure thermal comfort. This is realized by a fuzzy reasoning
that uses three and five membership functions to describe the activity level and the clothing
insulation states respectively.
Table 3.1 shows the 15 fuzzy rules used to evaluate the temperature range in which the
predicted mean vote is close to zero.
This rule base can be expressed linguistically as:
IF the occupant has Light clothing AND he or she is sedentary
THEN the ambient temperature should be Very high (in [28.280–31.58
0C]
range)
IF the occupant has Medium clothing AND his or her activity level is Medium
THEN the ambient temperature should be normal (in [19.50C –23.5
0C] range)
IF the occupant has Very heavy clothing AND his or her activity level is Medium
THEN the ambient temperature should be low (in [10.80C –14
0C] range)
Table 3.1
Fuzzy evaluation of the temperature range in which the thermal sensation is neutral
The clothing The activity level
insulation Low Medium High
Light [28.2 0C –31.5
0C] [24
0C –28.5
0C] [19.5
0C –25.5
0C]
Medium [25.9 0C –28.0
0C] [19.5
0C –23.5
0C] [13
0C –18.4
0C]
Heavy [23 0C –28.0
0C] [15
0C –24
0C] [7.0
0C –12
0C]
Very heavy [19.5 0C –23.5
0C] [10.8
0C –14
0C] [0.0
0C –0.6
0C]
82
3.3.3.3 THE ENVIRONMENTAL MODEL RULES.
From a practical point of view, the personal-dependent model is used to adjust the
environmental model which starts with the evaluation of the air velocity to determine the
operative temperature that ensure a predicted mean vote equal to zero. To this end, seven
membership functions are used to describe the state of both the air velocity Vair and the air
operative temperature To. These variables generate a maximum of seven fuzzy rules such as
the following examples:
IF the air velocity is V1 THEN the operative temperature is T 1
IF the air velocity is V2 THEN the operative temperature is T 3
IF the air velocity is V7 THEN the operative temperature is T7 etc.
Where (V1...V 7) and (T1...T7) are fuzzy terms that describe the air velocity and the
operative temperature respectively that correspond to an optimal sensation of thermal. Once
the desired ambient air temperature is calculated, it is compared to the measured air
temperature Ta to determine the state of the predicted mean vote (PMV)
3.3.4. MATLAB AND LabVIEW SIMULATION.
3.3.4.1. IMPLEMENTATION OF PMV IN LabVIEW.
On the basis of the analysis, a new fuzzy PMV is designed and a general fuzzy rule
base is derived to describe the state of human‘s thermal sensation. Fuzzy PMV code is
written using mfile in Matlab and it is interfaced with LabVIEW as shown in fig.3.4
83
Fig. 3.4 Block diagram for PMV calculation.
84
Fig. 3.5 Front panel for PMV Calculation (1)
Fig. 3.6 Membership function (1)
85
Fig. 3.7 Front panel for PMV Calculation (2)
Fig. 3.8 Membership function (2).
86
3.3.5. RESULT AND DISCUSSION
3.3.5.1. SIMULATION RESULT FOR FUZZY PMV
PMV index is found out for different values of activity level and clothing
insulation using fuzzy PMV technique as discussed earlier.
Table.3.2. Fuzzy PMV values.
Activity Clo-value Vair Room temperature(Ta) PMV
0.196
0.123 0.118 24.9 -0.33
0.359
0.231 0.356 23.2 -0.533
0.663
0.329 0.612 20.6 -0.586
1.13 0.75 0.432 26.02 0.207
0.364 0.999 0.66 25.67 0.129
2.4 1.4 0.315 27.484 1.523
2.8 1.44 0.214 31.5 2.99
0.216 0.0918 0.169 9.67 -1.804
0.44 0.287 0.28 0.314 -2.47
0.127 0.42 0.026 0.798 -2.63
The PMV index which is found using fuzzy PMV varies from -3 to +3 as Fanger‘s PMV
index without any iteration. It will be acceptable only when PMV lies between -0.5 and +0.5.
Therefore the optimum values of PMV using fuzzy logic is
Table 3.3 PMV values which are in acceptable range from table.3.2.
From the above table the minimum PMV value is 0.129, the corresponding
Percentage of people dissatisfied is 5.3447.
3.3.6. CONCLUSION
Poor interior conditions contribute to discomfort. Thermal comfort is one of the most
important comfort factors. Important parameters that affect thermal comfort are air
temperature, relative humidity, air velocity, environment radiation, activity level of
passengers and clothing insulation. PMV index which combines the above parameters is used
Activity clo-value Vair Room temperature(Ta) PMV
0.196 0.123 0.118 24.9 -0.33
1.13 0.75 0.432 26.02 0.207
0.364 0.999 0.66 25.67 0.129
87
to indicate thermal comfort. It will be acceptable only when PMV lies between -0.5 and +0.5
By Fuzzy logic method it was found that the minimum PMV value is 0.129, the
corresponding Percentage of people dissatisfied is 5.3447.
3.3.7. DEFINITIONS OF THERMAL COMFORT AND ITS
PARAMETERS:
a) Thermal comfort.
That condition of mind which expresses satisfaction with the thermal environment
and is assessed by subjective evaluation
b) Thermal sensation:
A Conscious feeing commonly graded into the categories cold, cool, slightly cool,
neutral, slightly warm, warm and hot; it requires subjective evaluation.
c) Predicted mean vote (PMV):
An index that predicts the mean value of the votes of a large group of persons on the
seven point thermal sensation scale is the PMV.
d) Percent dissatisfied.
Percentage of people predicted to be dissatisfied due to local discomfort.
e) Predicted percentage of Dissatisfied. (PPD):
An index that establishes a quantitative prediction of the percentage of thermally
dissatisfied people determined from PMV
f) Air Temperature:
The temperature of the air surrounding the occupant
g) Mean radiant temperature:
The uniform surface temperature of an imaginary black enclosure in which an
occupant would exchange the same amount of radiant heat as in the actual non uniform
space is known as mean radiant temperature. The MRT affects the rate of radiant heat loss
from the body. Since the surrounding surface temperatures may vary widely, the MRT is
a weighted average of all radiating surface temperatures within line of sight. In winter,
levels of wall, roof, and floor insulation together with window treatments such as double
glazing, blinds, and drapes contribute to Mean Radiant temperature.
h) Air speed.
The rate of air movement at a point, without regard to direction
88
i) Velocity of Air.
Air motion significantly affects body heat transfer by convection and
evaporation. Air Movement results from free convection from the occupants‘
body movements. The faster the motion, the greater the rate of heat flow by both
convection and evaporation. When ambient temperatures are within acceptable limits,
there is no minimum air movement that must be provided for thermal comfort. The
natural convection of air over the surface of the body allows for the continuous
dissipation of body heat. When ambient temperatures rise, however, natural air flow
velocity is no longer sufficient and must be artificially increased, such as by the use of
fans.
j) Clothing insulation/ensemble (Icl)
The resistance to sensible heat transfer provided by a clothing ensemble and is
expressed in Clo units. The definition of clothing insulation relates to heat transfer
from the whole body and thus also includes the uncovered parts of the body, such as
head and hands.
k) Clo.
A unit used to express the thermal insulation provided by garments and clothing
ensembles where 1 clo = 0.155 m2 0
C/ W (0.88 ft2 0
F/Btu)
l) Humidity ratio.
It is the ratio of the mass of water vapour to the mass of dry air in a given volume.
m) Relative humidity (RH)
It is the ratio of the partial pressure of the water vapour in the air to the saturation
pressure of water vapour at the same temperature and the same total pressure.
n) Metabolic rate (M)
The rate of transformation of chemical energy into heat and mechanical work by
metabolic activities within an organism, usually expressed in terms of unit area of the
total body surface. Here it is expressed in met units.
o) Met
A unit used to describe the energy generated inside the body due to metabolic activity.
This is also equal to 58.2 W/m2, which is equal to the energy produced per unit
surface area of an average person, seated at rest. The surface area of an average
person is 1.8 m2 (19ft).
89
3.4. THERMAL COMFORT IN AN OFFICE BUILDING.
3.4.1. LITERATURE SURVEY
Human perception of air movement depends on environmental factors such as air
velocity, air velocity fluctuations, air temperature, and personal factors such as overall
thermal sensation, clothing insulation and physical activity level (metabolic rate) (Toftum,
2004). Air velocity affects both convective and evaporative heat losses from the human body,
and thus determines thermal comfort conditions (Tanabe, 1988; Mallick, 1996). If we agree
that thermal environments that are slightly warmer than preferred or neutral, can be still
accepTable to building occupants as the adaptive comfort model suggests (deDear, Brager,
2002; Nicol, 2004), then the introduction of elevated air motion into such environments
should be universally regarded as desirable. This is because the effect will be to remove
sensible and latent heat from the body, so body temperatures will be restored to their comfort
set-points. This hypothesis can be deduced from the physiological principle of alliesthesia
(Cabanac, 1971).
In hot and humid climates, elevated indoor air velocity increases the indoor
temperature that building occupants find most comforTable. Nevertheless, the distribution of
air velocities measured during these field studies was skewed towards rather low values.
Many previous studies have attempted to define when and where air movement is either
desirable or not desirable (i.e. draft) (Mallick, 1996; Santamouris, 2004). Thermal comfort
research literature indicates that indoor air speed in hot climates should be set between 0.2 -
1.50 m/s, yet 0.2 m/s has been deemed in ASHRAE Standard 55 to be the threshold upper
limit of draft perception allowed inside air-conditioned buildings, where occupants have no
direct control over their environment (de Dear, 2004) The new standard 55 is based on
Fanger‘s (1988) draft risk formula, which has an even lower limit in practice than 0.2 m/s.
None of the previous research has explicitly addressed air movement acceptability. Instead it
has focused mostly on overall thermal sensation and comfort (Toftum, 2002).
3.4.2. Research methods.
3.4.2.1. Outdoor Climatic environment.
Under the Koppen climate classification, the Coimbatore city has a tropical wet and
dry climate. It has mild winters and moderate summers. Karunya University office buildings
90
lie in the latitude of 100 55‘ 51.73‖ N and longitude of 76
0 44‘ 40.60‖ E with elevation 1551
ft. The surveys in this study were performed in the May 2009 and September 2009
3.4.2.2. Subjects.
A Sample size of 220 subjects in 8 different office buildings in the Karunya
University was collected in survey and field measurements. The offices interviewed are
multi-story buildings. The volunteers participating in the study comprised both men and
women. The average age of all respondents was 33.2 years, ranging from 23 to 57 years. All
the participants were in good health. The questionnaire covered several areas including
personal factors (name, gender, age, etc.), years of living in their current cities and personnel
environmental control.
The questionnaire also included the traditional scales of thermal sensation and thermal
preferences, current clothing garment and metabolic activity. The thermal sensation scale was
the ASHRAE seven point scale ranging from cold (-3) to hot (3) with neutral (0) in the
middle. The three point thermal preference scale asked whether the respondents would like to
change their present thermal environment. Possible responses were ―want warmer‖, ―no
change‖, or ―want cooler‖. Clothing garment check list were compiled from the extensive
lists published in ASHRAE -55, 2004. Metabolic rates were assessed by a check of activities
databases published in ASHRAE-55, 2004. The summary of the background characteristics
of the subjects are presented.
Table.3.4 Summary of the sample of residents and personal thermal variables
Sample size 220
Age (year)
Mean 33.2
Maximum 23 years
Minimum 5 months
Metabolic rate
Clothing insulation
75(W/m2)
1.5 Clo
91
3.4.2.3. Data collection.
Both physical and subjective questionnaires were obtained simultaneously in the visit
of the field survey. This study investigates thermal environment and comfort of office
buildings in the Karunya University. A total of 220 subjects in naturally ventilated 11 office
buildings ( with occupant – operable windows) provided 220 sets of cross-sectional thermal
comfort data, first field campaign from Mar 15, 2010 to Mar24,2010 and second field
campaign from Sep10,2010 to Sep 19, 2010 in Karunya University, Coimbatore. In both the
set of data collections the same buildings were taken into account for data collection. Indoor
climatic data were collected using instruments, with accuracies and response times in
accordance the recommendations of ANSI/ASHRAE 55. All the measurements were carried
out between 10:00 hours and 16:00 hours.
A number of instruments were used to measure the thermal environment conditions,
while the respondents filled in the subjective thermal comfort questionnaire. The instruments
were standard thermometer for air temperature, whirling hygrometer for humidity, globe
thermometer for radiant heat, kata thermometer for air velocity. Metabolic rate can be
estimated using standard Table found in ISO 7730. Among the residential respondents, air
temperature readings were taken at a minimum of two locations in each space and at two
different levels corresponding to the body level and the ankle level corresponding to
approximately 0.1 m and 1.2 m above the floor level, respectively. Instruments used in this
study met the ASHRAE standards‘ requirements for accuracy.
During the survey period, there were no significant sources of radiant heat in
residents‘ apartments. Therefore the operative temperature is close to the air temperature. The
insulation of clothing ensembles was determined using the Olsen‘s 1985 summation formula:
Icl= ∑ I clu,i where Icl is the insulation of the entire ensemble and I clu,i represents the effective
insulation of the garment i. The garments values published in the ANSI/ASHRAE Stand card
55-2004 was the basis for the estimation of clothing ensemble insulation. The general mean
clothing-insulation value of 1.5 clo was recorded among all the respondents. The majority of
the respondents were seated on partly or fully upholstered chairs at the time of survey. This
appears to have been reflected in the generally higher mean value of 1.1 clo recorded among
the subjects at home.
92
The metabolic rates were determined from the activities filled by the subjects and as
observed at the time of the survey. Uniform value of 75 W/m2 was assumed for respondents
of the residential buildings. This assumption is based on the ISO 7730 Table of metabolic
rates for provisions for relaxed seating which was assumed to be the case with all subjects in
their homes.
3.4.2.4. Subjective questionnaire.
The subjective questionnaire consists of the following areas. All the surveys are
―right now‖ surveys. It takes 15 minutes in average for a participant to answer those
questions.
93
3.4.2.5. Indoor climate.
Table.3.5 Summary of indoor climatic conditions in the first session for office thermal comfort
ROOM date SAMPLE Size Ta(0c) Vair Tmrt Pa Tcl
S a
nd
H,M
CA
, B
.Ed
15-Mar-10 46 18.6 0.94 21 0.9 27.5 16-Mar-10 46 16.7 0.65 22.8 0.34 28.4 17-Mar-10 46 17.0 0.53 23.5 0.79 28.2 18-Mar-10 46 32.0 0.24 23.1 0.45 28.8 19-Mar-10 46 17.8 0.34 22.8 0.35 27.7 20-Mar-10 46 18.7 0.15 22.7 0.67 28.4 21-Mar-10 46 19.5 0.87 20.6 0.78 27.9 22-Mar-10 46 32.6 0.97 21.8 0.67 27.3 23-Mar-10 46 27.0 0.76 21.6 0.23 27.1 24-Mar-10 46 31.0 0.65 20.6 0.26 28.9
EC
E,M
ED
IA,B
IO
TE
CH
,BIO
IN
FO
,
FO
OD
15-Mar-10 59 28.0 0.79 21.6 0.57 27.6 16-Mar-10 59 24.6 0.45 22.7 0.39 27.9 17-Mar-10 59 33.5 0.35 21.6 0.92 27.5 18-Mar-10 59 27.5 0.67 22.5 0.93 27.9 19-Mar-10 59 29.4 0.78 22.6 0.48 27.0 20-Mar-10 59 16.7 0.67 23.5 0.38 28.5 21-Mar-10 59 16.7 0.23 22.5 0.62 27.1 22-Mar-10 59 17.4 0.45 19.5 0.47 28.9 23-Mar-10 59 16.2 0.34 23.5 0.99 27.1 24-Mar-10 59 18.5 0.08 22.5 0.23 28.9
EE
E,
EIE
15-Mar-10 22 19.5 0.03 23.5 1.00 27.9 16-Mar-10 22 20.4 0.67 20.6 0.26 27.5 17-Mar-10 22 21.5 0.99 21.8 0.57 27.9 18-Mar-10 22 27.3 0.23 22.7 0.39 28.4 19-Mar-10 22 28.3 0.03 22.3 0.92 27.5 20-Mar-10 22 31.5 0.30 21.6 0.93 27.9 21-Mar-10 22 32.6 0.34 21.9 0.48 27.8 22-Mar-10 22 32.6 0.34 20.6 0.01 27.5 23-Mar-10 22 27.4 0.45 20.5 0.79 27.9 24-Mar-10 22 27.4 0.34 21.5 0.45 27.1
CIV
IL
15-Mar-10 15 28.5 0.78 22.4 0.35 28.9 16-Mar-10 15 29.5 0.38 22.6 0.67 27.6 17-Mar-10 15 30.5 0.62 20.0 0.78 28.6 18-Mar-10 15 34.0 0.47 20.6 0.67 28.0 19-Mar-10 15 28.4 0.99 21.8 0.23 28.9 20-Mar-10 15 28.5 0.23 21.6 0.01 27.1 21-Mar-10 15 29.5 1.00 20.6 0.80 28.9 22-Mar-10 15 34.0 0.28 21.6 0.90 27.9 23-Mar-10 15 18.4 0.74 23.4 0.54 28.9 24-Mar-10 15 19.5 0.25 22.6 0.34 27.9
ME
CH
15-Mar-10 23 28.4 0.84 23.5 0.09 28.0 16-Mar-10 23 19.6 0.26 22.5 0.03 28.5 17-Mar-10 23 20.9 0.26 23.5 1.00 29.0 18-Mar-10 23 23.5 0.57 23.5 0.28 27.1 19-Mar-10 23 27.3 0.39 22.9 0.74 28.9 20-Mar-10 23 28.3 0.92 21.6 0.25 27.5 21-Mar-10 23 18.6 0.93 21.9 0.02 27.9 22-Mar-10 23 19.5 0.48 20.6 0.3 27.5 23-Mar-10 23 16.0 0.72 20.5 0.03 27.9 24-Mar-10 23 21.5 0.64 20.7 0.90 28.0
CS
T
15-Mar-10 34 33.7 0.73 20.3 0.38 28.5 16-Mar-10 34 27.3 0.73 21.5 0.62 27.1 17-Mar-10 34 28.3 0.65 22.6 0.47 27.9 18-Mar-10 34 23.5 0.45 23.5 0.99 27.5 19-Mar-10 34 28.4 0.37 22.4 0.23 27.5 20-Mar-10 34 28.5 0.37 20.4 1.000 27.9 21-Mar-10 34 33.5 0.47 20.6 0.99 27.6 22-Mar-10 34 31.4 0.26 21.8 0.97 27.8 23-Mar-10 34 29.8 0.39 21.6 0.13 28.5 24-Mar-10 34 17.5 0.01 20.6 0.26 27.1
MB
A
15-Mar-10 21 27.3 0.1 21.6 0.57 28.9 16-Mar-10 21 28.3 0.34 22.8 0.39 27.1 17-Mar-10 21 26.4 0.99 23.5 0.92 27.8 18-Mar-10 21 33.8 0.97 22.5 0.37 27.1 19-Mar-10 21 18.6 0.13 23.5 0.37 27.5 20-Mar-10 21 34.0 0.26 20.0 0.57 27.9 21-Mar-10 21 24.6 0.46 21.6 0.39 27.5 22-Mar-10 21 28.5 0.56 21.9 0.92 27.5 23-Mar-10 21 34.0 0.48 20.6 0.26 27.0 24-Mar-10 21 23.5 0.23 20.5 0.57 29.0
MEAN 25.617143 0.504714 21.88857 0.536 27.91
MAX 34 1 23.5 1 29
MIN 16 0.01 19.5 0.01 27
AVERAGE 5.7656899 0.274964 1.097632 0.298403481 0.609859
94
ROOM Date SAMPLE SIZE Ta(0c) Vair Tmrt Pa Tcl
S a
nd
H,M
CA
, B
.Ed
10-Sep-10 46 28.4 0.45 21.5 0.23 27.5 11-Sep-10 46 19.6 0.35 22.4 1.0 28.4 12-Sep-10 46 20.9 0.67 22.6 0.99 28.2 13-Sep-10 46 23.5 0.78 20.0 0.97 28.8 14-Sep-10 46 17.8 0.67 20.6 0.13 27.7 15-Sep-10 46 18.7 0.23 19.5 0.26 28.4 16-Sep-10 46 19.5 0.40 20.6 0.57 27.9 17-Sep-10 46 32.6 0.80 21.8 0.39 27.0 18-Sep-10 46 28.4 0.90 21.6 0.92 27.1 19-Sep-10 46 19.6 0.54 20.6 0.26 28.9
EC
E,M
ED
IA,B
IO
TE
CH
,BIO
IN
FO
,
FO
OD
10-Sep-10 59 20.9 0.34 21.6 0.57 27.6 11-Sep-10 59 23.5 0.10 22.7 0.39 27.7 12-Sep-10 59 33.5 0.10 21.6 0.92 27.5 13-Sep-10 59 27.5 0.67 22.5 0.93 27.9 14-Sep-10 59 29.4 0.78 22.6 0.48 27.0 15-Sep-10 59 16.7 0.67 22.3 0.38 28.5 16-Sep-10 59 16.7 0.23 22.5 0.62 27.1 17-Sep-10 59 17.4 0.45 21.4 0.47 28.9 18-Sep-10 59 16.2 0.34 21.0 0.99 28.9 19-Sep-10 59 18.5 0.20 22.8 0.23 28.9
EE
E,
EIE
10-Sep-10 22 19.5 0.10 22.4 1.00 27.9 11-Sep-10 22 20.4 0.67 22.4 0.26 27.5 12-Sep-10 22 21.5 0.99 22.9 0.57 28.9 13-Sep-10 22 27.3 0.23 19.5 0.39 28.4 14-Sep-10 22 28.3 0.10 23.0 0.92 27.5 15-Sep-10 22 31.5 0.10 21.6 0.93 27.9 16-Sep-10 22 32.6 0.34 23.2 0.48 27.8 17-Sep-10 22 32.6 0.34 20.6 0.56 27.5 18-Sep-10 22 27.4 0.45 20.5 0.79 29.0 19-Sep-10 22 27.4 0.34 21.5 0.45 27.1
CIV
IL
10-Sep-10 15 28.5 0.78 22.4 0.35 28.9 11-Sep-10 15 29.5 0.38 22.6 0.67 27.6 12-Sep-10 15 30.5 0.62 20.0 0.78 28.6 13-Sep-10 15 34.0 0.47 20.6 0.67 28.9 14-Sep-10 15 28.4 0.99 21.8 0.23 28.9 15-Sep-10 15 28.5 0.23 21.6 0.40 27.1 16-Sep-10 15 29.5 1.00 20.6 0.80 28.9 17-Sep-10 15 34.0 0.28 19.5 0.90 27.9 18-Sep-10 15 18.4 0.74 20.6 0.54 28.9 19-Sep-10 15 19.5 0.25 21.8 0.34 27.9
ME
CH
10-Sep-10 23 28.4 0.84 22.1 0.09 28.0 11-Sep-10 23 19.6 0.26 23.5 0.03 28.5 12-Sep-10 23 28.4 0.23 22.1 1.00 29.0 13-Sep-10 23 19.6 1.00 21.3 0.26 27.1 14-Sep-10 23 20.9 0.99 21.5 0.39 28.9 15-Sep-10 23 23.5 0.97 21.6 0.01 27.5 16-Sep-10 23 18.6 0.13 21.9 0.03 27.9 17-Sep-10 23 19.5 0.26 20.6 0.34 27.5 18-Sep-10 23 16.0 0.57 20.5 0.99 27.9 19-Sep-10 23 21.5 0.39 20.7 0.97 28.0
CS
T
10-Sep-10 34 33.7 0.92 20.3 0.13 28.5 11-Sep-10 34 27.3 0.73 21.5 0.26 27.1 12-Sep-10 34 28.3 0.65 23.5 0.47 27.9 13-Sep-10 34 23.5 0.45 22.1 0.99 27.5 14-Sep-10 34 28.4 0.37 22.4 0.23 27.5 15-Sep-10 34 28.5 0.37 20.4 1.00 27.9 16-Sep-10 34 33.5 1.00 20.6 0.99 27.6 17-Sep-10 34 31.4 0.26 21.8 0.97 27.8 18-Sep-10 34 28.4 0.20 21.6 0.13 28.5 19-Sep-10 34 19.6 0.23 20.6 0.26 27.1
MB
A
10-Sep-10 21 31.5 1.00 21.6 0.57 28.9 11-Sep-10 21 32.6 0.99 22.8 0.39 27.1 12-Sep-10 21 32.6 0.97 22.4 0.92 27.8 13-Sep-10 21 27.4 0.13 22.4 0.37 27.5 14-Sep-10 21 27.4 0.26 22.0 0.37 27.9 15-Sep-10 21 16.9 0.57 20.0 0.57 27.6 16-Sep-10 21 29.5 0.39 21.6 0.39 27.8 17-Sep-10 21 28.5 0.92 21.9 0.92 28.5 18-Sep-10 21 34.0 0.10 20.6 0.26 27.4 19-Sep-10 21 23.5 0.23 20.5 0.57 27.8
MEAN 25.44429 0.506429 21.53857 0.551429 27.98571
MAX 34 1 23.5 1 29
MIN 16 0.1 19.5 0.01 27
AVERAGE 5.617768 0.298921 0.9935 0.309611 0.624914
Table.3.6 . Summary of indoor climatic conditions in the second session for office thermal comfort
95
The following values are taken from the data collected from questionnaire and
measurements for further optimization using different non-traditional algorithms. The
minimum and maximum values of each of these parameters were taken as the lower and the
upper limits of the parameters. These values were taken from both the set put together which
are taken in March and September so as to take a generalized thermal comfort of the
university buildings. These values are used in the optimization techniques to optimize the
final value and also to find the optimum value of the PMV.
Table.3.7. Range of values
Fcl Ta Tmrt Vair Pa Tcl M(met) Icl(clo)
Min 0 16 19.5 0.1 0.01 27 75 1.5
Max 1.5 34 23 1 1 29 75 1.5
In an attempt to reduce the processing time and to improve the quality of solution, and
particularly, to avoid being trapped in local minima, the non- traditional optimization is used.
In this problem, to find the optimum thermal comfort, ten non- traditional optimization
techniques are used. Each one has its own characteristics. Twenty trial runs were performed
for the problem in each of the ten methods. The performance of the different algorithms was
compared. The characteristics led to different solutions and run times. The results were
examined finally based on different criteria.
Each algorithm with its own option set and stopping criteria was used. All the non-
traditional optimization was run using MATLAB2010 to get the global optimum value for
each of the parameter and also the final value of the thermal comfort.
Therefore the Problem is
To minimize PMV for office with the regression coefficients is:
96
Subject to the following constraints (bounds)
0 ≤ Fcl ≤ 1.5;
16 ≤ Ta ≤ 34 ;
19.5 ≤ Tmrt ≤ 23;
0.1 ≤ Vair ≤ 1;
0.01 ≤ Pa ≤ 1;
27 ≤ Tcl ≤ 29;
M = 75;
Icl = 1.5
97
3.4.3. Algorithms
3.4.3.1. Genetic algorithm
3.4.3.1.1. Options Set for the Algorithm:
Initial population: 20.
Elite count: 2.
Cross over fraction as 0.8.
Max Time Limit: ∞.
Max Generations: 100.
Fitness Limit: -∞.
Selection: Stochastic.
3.4.3.1.2. Stopping Criteria:
If the maximum generations is reached (100).
If maximum time is reached (∞).
If average change in function value < 10¯⁶. Table.3.8. Results of GA in 20 trails for office thermal comfort
Trail
s
Fcl Ta Tmrt Vair Pa Tcl PMV PPD Time Iters
1 0.8568 25.188
1
20.717
6
0.4163 0.0103 28.034
3
-0.5 5 0.5287
16
51
2 0.8472 25.64 20.330
7
0.6048 0.1861 28.637
7
-0.5 5 0.5461
46
51
3 0.8571 24.552
9
19.500
3
0.4337 0.0101 27.873
4
-0.5 5 0.5525
99
51
4 0.6605 22.233
8
19.508
8
0.3322 0.6014 28.658
2
-0.5 5 0.5376
16
51
5 1.5 16 19.5 1 0.01 27 -0.5 5 0.5418
25
51
6 0.5884 22.333
2
21.692
2
0.5732 0.1797 28.855
2
-0.5 5 0.5306
98
51
7 0.3902 17.426
1
20.816
8
0.8219 0.7775 27.668
3
-0.5 5 0.5395
39
51
8 0.5843 16.076
2
19.785
6
0.2075 0.307 27.035
7
-0.5 5 0.5297
8
51
9 0.8079 22.062 19.5 0.3219 0.0817 27.008
1
-0.5 5 0.5488
09
51
10 0.7726 24.399 22.325
7
0.9572 0.4146 28.222 -0.5 5 0.5460
48
51
11 0.4625 18.087
6
22.490
7
0.5298 0.2691 28.285
3
-0.5 5 0.5251
95
51
12 0.6641 22.959
5
22.464
2
0.5219 0.2602 28.980
3
-0.5 5 0.5323
24
51
13 0.4865 19.433
39
22.349
4
0.544 0.2023 28.61 -0.5 5 0.5418
36
51
14 0.6368 23.889
1
19.500
2
0.6699 0.3043 28.578
7
-0.5 5 0.5378
07
51
15 0.7589 22.550
1
22.550
1
0.3427 0.532 28.893
7
-0.5 5 0.5475
64
51
16 0.5228 18.123
3
20.965
4
0.3694 0.0101 27.734
6
-0.5 5 0.5385
74
51
17 0.7037 20.679
3
20.607
3
0.2374 0.323 28.207
6
-0.5 5 0.5318
3
51
18 0.661 16.015
8
19.500
5
0.1001 0.1524 27.033
2
-0.5 5 0.5394
53
51
19 0.5299 19.692
6
20.09 0.3646 0.7578 28.899
1
-0.5 5 0.5346
02
51
20 0.7467 24.610
8
20.494
5
0.7986 0.0151 28.151
7
-0.5 5 0.5503
46
51
Avg 0.7018
95
21.097
64
20.734
5
0.5073
55
0.2702
35
28.118
36
-0.5 5 0.5390
6535
51
98
Fig.3.9 .Convergence of GA
3.4.3.2. Simulated annealing
3.4.3.2.1. Options Set:
Initial Temperature: 100.
Annealing Function: Fast Annealing.
Reannealing interval: 100.
Time Limit: ∞.
Max.function evaluation: 3000 No. of variables.
Max. Iterations: ∞.
Function Tolerance: 10¯⁶. Objective Limit: 10¯⁶
3.4.3.2.2. Stopping Criteria:
Max. Time reached.
The average change in value of the objective function is < 10¯⁶. Max. Iterations are reached.
If the number of function evaluations reached.
If the best objective function value is less than or equal to the value of Objective limit
it is stopped.
0 10 20 30 40 50 60 70 80 90 1005
5.1
5.2
5.3
5.4
5.5
5.6
5.7
5.8
Generation
Fitness v
alu
e
Best: 5 Mean: 5
Best f itness
Mean fitness
99
Table.3.9. Results of SA in 20 trails for office thermal comfort.
Trail
s
Fcl Ta Tmrt Vair Pa Tcl PMV PPD Time Iters
1 0.3723 16.826
2
22.30
1
0.7753 0.7548 28.280
2
-0.5 5 5.1652
38
3006
2 0.7477 17.177
4
22.58
7
0.1059 0.9133 28.489
6
-0.5 5 4.1023
65
3001
3 0.6201 22.019
9
20.16
83
0.8629 0.3263 27.130
8
-0.5 5 3.7101
75
3001
4 0.5951 16.972 20.37
8
0.1619 0.032 28.156
5
-0.5 5 3.4373
66
3001
5 1.1091 27.334
8
20.40
96
0.8993 0.2761 28.589
8
-0.5 5 6.1380
07
3001
6 0.7143 22.547
2
22.07
03
0.3882 0.5892 28.889
9
-0.5 5 4.3433
753
3001
7 0.4554 17.446
1
22.19
4
0.6605 0.9469 27.586
7
-0.5 5 4.1677
54
3002
8 0.5366 21.344
2
20.09
96
0.5283 0.1032 28.485
8
-0.5 5 4.2340
63
3001
9 0.7255 18.794
5
20.92
19
0.1864 0.5557 27.686
4
-0.5 5 3.6822
38
3001
10 0.9322 25.292
9
20.75
94
0.88 0.6343 27.890
3
-0.5 5 3.6805
18
3002
11 1.2129 25.972
9
23.06
43
0.3384 0.7721 28.963
1
-0.5 5 4.1376
36
3001
12 0.3967 17.174
4
19.52
64
0.6171 0.3943 27.732
4
-0.5 5 4.0646
44
3003
13 1.016 24.805
6
21.00
45
0.7406 0.8484 27.505 -0.5 5 3.6395
3
3001
14 0.4847 16.007
7
23.06
92
0.501 0.6394 27.094
9
-0.5 5 3.1211
78
3001
15 1.251 26.879
6
23.41
85
0.7251 0.355 28.844
9
-0.5 5 5.4008
93
3001
16 0.9353 20.482
3
23.35
07
0.1291 0.5949 28.684
6
-0.5 5 3.3472
13
3001
17 0.3621 16.282
4
20.54
64
0.6345 0.8534 28.606
5
-0.5 5 4.2705
51
3002
18 0.345 16.402
1
20.94
81
0.7779 0.5505 28.195
8
-0.5 5 4.7207
78
3001
19 0.3584 17.709
5
20.75
02
0.7888 0.7585 28.822
6
-0.5 5 3.2562
75
3001
20 1.0492 25.481
8
19.77
5
0.1365 0.995 28.990
8
-0.5 5 3.4446
22
3001
Avg 0.7109t
T8
20.647
68
21.36
717
0.54188
5
0.59466
5
28.231
33
-0.5 5 4.1032
2097
3001.5
Fig.3.10. Convergence of SA
0 500 1000 1500 2000 2500 3000 35005
5.2
5.4
5.6
5.8
6
6.2
6.4
Iteration
Function v
alu
e
Best Function Value: 5
1 2 3 4 5 60
5
10
15
20
25
30Best point
Number of variables (6)
Best
poin
t
0 10 20 30 40 50 60 70 80 90 100
Time
Iteration
f-count
% of criteria met
Stopping Criteria
0 500 1000 1500 2000 2500 3000 35005
6
7
8
9
10
11
Iteration
Function v
alu
e
Current Function Value: 5
100
3.4.3.3. Pattern search
3.4.3.3.1. Options Set:
Poll Method: GPS positive Basis 2N.
Initial Mesh size: 1.
Expansion Factor: 2.
Contraction Factor: 0.5.
Mesh Tolerance: 10¯⁶. Max. Iteration: 100 No. of Variables.
Max. Function Evaluation: 2000 No. of Variables.
Max. Time Limit: ∞.
Function Tolerance: 10¯⁶
3.4.3.3.2. Stopping Criteria:
Mesh Tolerance: 10¯⁶. Max. Iteration: 100 No. of Variables.
Max. Function Evaluation: 2000 No. of Variables.
Max. Time Limit: Inf.
Function Tolerance: 10¯⁶ Table 3.10. Results of PS in 20 trails for office thermal comfort
Trails Fcl Ta Tmrt Vair Pa Tcl PMV PPD Time Iters
1 0.75 25 19.5457 1 0.255 28 -0.5 5 0.825115 26
2 0.75 25 19.5457 1 0.255 28 -0.5 5 0.3777792 26
3 0.75 25 19.5457 1 0.255 28 -0.5 5 0.359182 26
4 0.75 25 19.5457 1 0.255 28 -0.5 5 0.365426 26
5 0.75 25 19.5457 1 0.255 28 -0.5 5 0.378375 26
6 0.75 25 19.5457 1 0.255 28 -0.5 5 0.371941 26
7 0.75 25 19.5457 1 0.255 28 -0.5 5 0.38865 26
8 0.75 25 19.5457 1 0.255 28 -0.5 5 0.378278 26
9 0.75 25 19.5457 1 0.255 28 -0.5 5 0.375717 26
10 0.75 25 19.5457 1 0.255 28 -0.5 5 0.37874 26
11 0.75 25 19.5457 1 0.255 28 -0.5 5 0.373859 26
12 0.75 25 19.5457 1 0.255 28 -0.5 5 0.373993 26
13 0.75 25 19.5457 1 0.255 28 -0.5 5 0.401204 26
14 0.75 25 19.5457 1 0.255 28 -0.5 5 0.376038 26
15 0.75 25 19.5457 1 0.255 28 -0.5 5 0.370777 26
16 0.75 25 19.5457 1 0.255 28 -0.5 5 0.379511 26
17 0.75 25 19.5457 1 0.255 28 -0.5 5 0.372559 26
18 0.75 25 19.5457 1 0.255 28 -0.5 5 0.373777 26
19 0.75 25 19.5457 1 0.255 28 -0.5 5 0.357336 26
20 0.75 25 19.5457 1 0.255 28 -0.5 5 0.370143 26
Avg 0.75 25 19.5457 1 0.255 28 -0.5 5 0.39742001 26
101
Fig.3.11. Convergence of Pattern search
3.4.3.4. Particle swarm optimization
3.4.3.4.1. Options Set:
Max.Generation = 200.
Max. Time Limit= ∞.
Average change in fitness value= 10-6
.
Time Limit = ∞.
Function Tolerance= 10-6.
Cognitive Attraction = 0.5.
Population Size = 40.
Social Attraction = 1.25.
3.4.3.4.2. Stopping Criteria:
Max.Generation = 200.
Max. Time Limit= ∞.
Average change in fitness value= 10-6
Time Limit = ∞.
Function Tolerance= 10-6
0 5 10 15 20 25 305
5.2
5.4
5.6
5.8
6
6.2
6.4
Iteration
Function v
alu
eBest Function Value: 5
0 5 10 15 20 25 300
0.5
1
1.5
2
Iteration
Mesh s
ize
Current Mesh Size: 9.5367e-007
102
Table 3.11. Results of PSO in 20 trails for office thermal comfort
Trail
s
Fcl Ta Tmrt Vair Pa Tcl PMV PPD Time Iter
ss 1 0.877 23.0831 22.8579 0.4892 0.8702 28.038
1
-0.5 5 0.09709
1
51
2 0.5958 21.1338 22.8217 0.714 0.2186 27.781
2
-0.5 5 0.09368
1
51
3 1.4631 25.0762 23.2995 0.4324 0.8067 27.600
6
-0.5 5 0.09827 51
4 0.6583 21.7398 21.1193 0.5704 0.2864 27.628
8
-0.5 5 0.09296
2
51
5 0.6525 21.4012 22.0685 0.3756 0.2417 28.431
5
-0.5 5 0.10045
2
51
6 1.0041 24.7647 20.8495 0.8217 0.7388 27.348
2
-0.5 5 0.09109
4
51
7 0.7453 21.8714 21.2717 0.5292 0.6058 27.337
3
-0.5 5 0.09119
9
51
8 1.1379 26.0463 20.084 0.8315 0.6619 27.623
6
-0.5 5 0.09634
4
51
9 0.6253 21.6684 21.0984 0.6851 0.5796 27.652
2
-0.5 5 0.08928
1
51
10 0.6715 20.4392 22.7847 0.5851 0.1956 27.012
3
-0.5 5 0.09378
2
51
11 0.6897 21.5875 20.5594 0.3421 0.9625 28.450
5
-0.5 5 0.09887
9
51
12 0.6741 22.1253 22.8683 0.6112 0.5266 28.250
8
-0.5 5 0.09444
8
51
13 0.9268 22.4799 21.9717 0.4605 0.8204 27.159
1
-0.5 5 0.09827
2
51
14 0.6827 17.378 22.2322 0.1385 0.1425 28.499
4
-0.5 5 0.09087
5
51
15 1.141 25.56 20.3809 0.7445 0.1572 27.280
6
-0.5 5 0.10192
4
51
16 0.8636 25.5004 22.1541 0.8741 0.1722 28.560
5
-0.5 5 0.09687
1
51
17 0.5778 18.4669 21.1368 0.368 0.4676 27.513
7
-0.5 5 0.09738
4
51
18 0.6144 18.6382 20.227 0.1903 0.6684 28.745 -0.5 5 0.10031
3
51
19 0.4588 18.9213 20.6147 0.5844 0.2693 28.042
5
-0.5 5 0.09442
2
51
20 1.0798 26.1033 21.291 0.6194 0.0834 28.206
6
-0.5 5 0.09158
8
51
Avg 0.80697
5
22.1992
5
21.5845
7
0.5483
6
0.4737
7
27.858
13
-0.5 5 0.09545
66
51
3.4.3.5. GODLIKE
3.4.3.5.1. Options Set & Stopping Criteria:
Max.FunEvals = 10-5
.
Max. Iterations = 20.
Min. Iterations = 2.
Total. Iterations = 15.
Function Tolerance = 10-4
103
Table.3.12. Results of GL in 20 trails for office thermal comfort
Trail
s
Fcl Ta Tmrt Vair Pa Tcl PMV PPD Time Iter
s 1 0.9215 22.1134 20.5649 0.2555 0.3869 27.172
9
-0.5 5 3.02900
2
4
2 0.7984 23.2892 22.9587 0.6954 0.5445 27.970
4
-0.5 5 3.22748 4
3 0.8846 19.7001 21.4248 0.1635 0.5456 27.256
8
-0.5 5 2.64651
7
4
4 0.6674 21.7629 20.513 0.4172 0.4623 28.089
2
-0.5 5 3.02244
2
4
5 0.7742 23.3477 19.8398 0.2137 0.1333 28.836
3
-0.5 5 2.56603
7
4
6 0.3609 17.511 20.0836 0.7025 0.5641 28.725
3
-0.5 5 3.41047
6
4
7 0.5766 19.8268 22.4365 0.4799 0.3472 28.049
1
-0.5 5 3.06569
4
4
8 1.2381 26.437 21.5555
4
0.8851 0.5777 27.981
1
-0.5 5 2.26867
5
4
9 0.7883 21.7847 20.8019 0.3325 0.4128 27.550
2
-0.5 5 3.43545
2
4
10 1.2641 24.9938 20.2441 0.2914 0.7044 27.260
6
-0.5 5 2.81039
1
4
11 1.3031 26.4947 20.3178 0.5386 0.5936 27.771
6
-0.5 5 2.4456 4
12 0.6165 19.9522 21.7871 0.4122 0.5395 27.946
8
-0.5 5 2.80838
4
4
13 0.6843 24.1963 20.2035 0.6888 0.4447 28.601
7
-0.5 5 3.49286
4
4
14 1.1984 24.6278 21.0094 0.2734 0.8951 27.663
6
-0.5 5 2.99139
8
4
15 0.9948 25.3736 19.7741 0.6646 0.7064 27.701
3
-0.5 5 2.70929 4
16 0.8568 22.5513 22.0694 0.2081 0.1576 28.679
3
-0.5 5 3.77622
8
4
17 0.806 23.1159 20.8273 0.3435 0.7404 28.367
3
-0.5 5 3.29453
5
4
18 0.5515 20.9478 19.7064 0.5269 0.6486 28.228
5
-0.5 5 3.14123
9
4
19 0.6032 21.6678 20.0397 0.4603 0.9328 28.632
8
-0.5 5 2.92928
1
4
20 1.0894 23.5846 23.2053 0.2545 0.7804 28.371
7
-0.5 5 2.98221
3
4
Avg 0.84890
5
22.6639
3
20.9681
4
0.4403
8
0.55589
5
28.042
83
-0.5 5 3.00265
99
4
3.4.3.6. Fmincon.
3.4.3.6.1. Options Set for ‘Fmincon’:
Max.Iterations:400.
Max.function Evaluations: 100 No. of Variables.
Max.Time:∞.
Max. Function Tolerance: 10-6
.
3.4.3.6.2. Stopping Criteria for Global Search:
Max.Time: Inf.
Max Wait cycle: 20
104
3.4.3.6.2. Stopping Criteria for Fmincon:
Max.Iterations > 400.
Function Tolerance: 10-6
Table.3.13. Results of Fmincon in 20 trails for office thermal comfort
Trails Fcl Ta Tmrt Vair Pa Tcl PM
V
PPD Time Iters
1 0.48
71
21.774 19.5 0.6415 0.242 28.95
69
-0.5 5 16.6448
31
2282
2 0.33
58
18.080
19.865
7
0.9994 0.997 28.78
99
-0.5 5 15.1560
49
2282
3 1.17
57
28.003 20.124
4
0.6074 0.265 28.95
29
-0.5 5 12.9088
19
2254
4 0.61
46
21.317 21.170
7
0.567 0.437 27.87
29
-0.5 5 14.3027
52
2296
5 1.30
53
27.480 22.252
2
0.9583 0.941 28.81
46
-0.5 5 12.1787
76
2324
6 1.32
35
26.389 23.497 0.4962 0.832
9
28.75
17
-0.5 5 22.8326
64
2303
7 0.69
35
21.806 21.039
6
0.5311 0.265
1
27.46
99
-0.5 5 12.4062
42
2282
8 1.19
25
26.496 22.644
44
0.5357 0.021
6
28.62
83
-0.5 5 13.1254
5
2275
9 1.13
43
26.287 19.793
9
0.4584 0.680
2
28.06
71
-0.5 5 14.5899
27
2289
10 0.97
52
22.443 19.532
6
0.1832 0.624
3
27.19
56
-0.5 5 14.2329
38
2310
11 0.58
55
20.225 19.964 0.5444 0.133
9
27.00
43
-0.5 5 15.4775
48
2268
12 0.63
31
22.760 23.409
2
0.8145 0.368 28.54
31
-0.5 5 14.8478
26
2261
13 0.75
97
24.188 21.183
7
0.6327 0.483
3
28.46
21
-0.5 5 16.0136
6
2310
14 0.67
37
21.041 22.019
3
0.4636 0.591
4
27.88
35
-0.5 5 14.2385
94
2282
15 0.7 24.033 21.115
2
0.5247 0.1132 28.90
58
-0.5 5 14.6281
95
2289
16 1.27
27
26.672 22.538
9
0.7674 0.994
8
28.51
29
-0.5 5 13.5420
12
2296
17 0.78
85
18.265 22.571
9
0.1614 0.450
1
27.76
77
-0.5 5 17.6474
16
2275
18 1.08
22
23.798 22.954
5
0.3627 0.627
7
27.90
79
-0.5 5 13.5885
38
2338
19 0.83
7
25.021 20.648
3
0.7431 0.827
2
28.35
17
-0.5 5 12.3144
53
2289
20 0.71
63
17.569 19.509
6
0.121 0.799 27.69
9
-0.5 5 12.9897
1
2268
Avg 0.86
431
T
23.182
58
21.266
76
0.5556
85
0.534
81
28.22
689
-0.5 5 14.6833
2
2288.65
3.4.3.7. Direct evolution
3.4.3.7.1. Options Set:
Min. Value to Reach = 10-6
.
Population Size = 10 D.
Max. Iterations = 200.
Step Size F = 0.8.
Cross Over Probability = 0.5.
Strategy= 7 (DE/rand/1/bin)
105
DE/x/y/z, where DE stands for DE, x represents a string denoting the vector to be perturbed,
y is the number of difference vectors considered for perturbation of x, and z stands for the
type of crossover being used (exp: exponential; bin: binomial).
3.4.3.7.2. Stopping Criteria:
Max.Value of function reached= 10-6
.
Max.Iterations=200 Table.3.14. Results of DE in 20 trails for office thermal comfort
Trai
ls
Fcl Ta Tmrt Vair Pa Tcl PMV PPD Time Func Iter
1 0.991
5
24.65
36
22.59
93
0.631
2
0.785
1
27.545
5
0.1317 5 0.5458
45
12000 60
2 1.000
1
26.13
3
20.42
38
0.928
2
0.829
3
27.980
5
0.3486 5 0.5622
79
12000 60
3 0.436
8
19.53
02
22.94
45
0.693
3
0.133
5
28.514
3
0.3758 5 0.5249
03
12000 60
4 0.340
5
19.33
68
19.82
42
0.844
5
0.361
6
28.964
1
0.3095 5 0.5752
98
12000 60
5 1.287
9
26.83
82
21.27
96
0.320
3
0.735
4
28.492
1
0.3896 5 0.5230
7
12000 60
6 1.466
3
27.95
62
22.33
76
0.904
7
0.295
8
28.594
8
0.798 5 0.6029
35
12000 60
7 0.466
1
22.26 20.13
91
0.931
8
0.239
9
28.647
7
0.4113 5 0.5822
6
12000 60
8 0.503 19.38
55
21.52
84
0.468
2
0.199
1
28.127
9
0.3861 5 0.5843
69
12000 60
9 1.326
3
27.57
69
21.62
77
0.920
2
0.121
9
28.461
2
0.425 5 0.5403
72
12000 60
10 0.502
9
20.73
7
19.99
45
0.727
6
0.063
2
27.471 0.4183 5 0.5442
69
12000 60
11 1.286
8
27.19
28
21.68
93
0.327
2
0.476 28.787
3
0.399 5 0.5628
64
12000 60
12 0.449
4
17.97
5
23.43
13
0.651
6
0.171
8
27.691
2
0.4174 5 0.5632
16
12000 60
13 0.576
1
18.17
09
21.16
78
0.197
2
0.060
9
28.370
9
0.3551 5 0.5594
04
12000 60
14 1.121
5
27.52
77
20.47
21
0.813
8
0.045
5
28.290
7
0.1632 5 0.5387
61
12000 60
15 0.525
4
21.08
09
23.29
61
0.856
9
0.848
9
28.492
8
0.4765 5 0.5301
71
12000 60
16 0.784 25.14
39
20.43
08
0.926
7
0.668
6
28.019
6
0.2925 5 0.5641
29
12000 60
17 0.666
6
17.31
83
21.22
37
0.143
9
0.815
3
28.351
5
0.4286 5 0.5537
67
12000 60
18 1.484
5
28.29
65
20.56
07
0.807
4
0.543
8
28.480
6
0.3521 5 0.5278
86
12000 60
19 1.201
8
26.35
22
20.26
97
0.422
5
0.724
5
27.547
1
0.1079 5 0.5743
58
12000 60
20 1.363
9
25.74
71
21.96
99
0.773
3
0.785
7
27.317
6
0.3738 5 0.5803
29
12000 60
Avg 0.889
07
23.46
064
21.36
051
0.664
525
0.445
29
28.207
42
0.368 5 0.5570
242
12000 60
3.4.3.8. LGO
3.4.3.8.1. Stopping Criteria:
If the current best solution did not improve for
Program execution time limits > 600 seconds.
106
3.4.3.8.2. Local search termination criterion parameter:
first local search phase ends, if the function difference is less than
If max. constrain violation exceeds
Table.3.15. Results of LGO in 20 trails for office thermal comfort
Trails Fcl Ta Tmrt Vair Pa Tcl PMV PPD Time Iters
1 1.2668 25.8503 20.8447 0.1 0.2887 28.5338 -0.5 5 0.956907 3883
2 1.2668 25.8503 20.8447 0.1 0.2887 28.5338 -0.5 5 1.132054 3883
3 1.2668 25.8503 20.8447 0.1 0.2887 28.5338 -0.5 5 1.326114 3883
4 1.2668 25.8503 20.8447 0.1 0.2887 28.5338 -0.5 5 1.255759 3883
5 1.2668 25.8503 20.8447 0.1 0.2887 28.5338 -0.5 5 1.080224 3883
6 1.2668 25.8503 20.8447 0.1 0.2887 28.5338 -0.5 5 1.138645 3883
7 1.2668 25.8503 20.8447 0.1 0.2887 28.5338 -0.5 5 1.069755 3883
8 1.2668 25.8503 20.8447 0.1 0.2887 28.5338 -0.5 5 1.01494 3883
9 1.2668 25.8503 20.8447 0.1 0.2887 28.5338 -0.5 5 1.039328 3883
10 1.2668 25.8503 20.8447 0.1 0.2887 28.5338 -0.5 5 1.033702 3883
11 1.2668 25.8503 20.8447 0.1 0.2887 28.5338 -0.5 5 1.069969 3883
12 1.2668 25.8503 20.8447 0.1 0.2887 28.5338 -0.5 5 1.026883 3883
13 1.2668 25.8503 20.8447 0.1 0.2887 28.5338 -0.5 5 1.024515 3883
14 1.2668 25.8503 20.8447 0.1 0.2887 28.5338 -0.5 5 1.073316 3883
15 1.2668 25.8503 20.8447 0.1 0.2887 28.5338 -0.5 5 1.038982 3883
16 1.2668 25.8503 20.8447 0.1 0.2887 28.5338 -0.5 5 0.986438 3883
17 1.2668 25.8503 20.8447 0.1 0.2887 28.5338 -0.5 5 0.954037 3883
18 1.2668 25.8503 20.8447 0.1 0.2887 28.5338 -0.5 5 0.97541 3883
19 1.2668 25.8503 20.8447 0.1 0.2887 28.5338 -0.5 5 1.071272 3883
20 1.2668 25.8503 20.8447 0.1 0.2887 28.5338 -0.5 5 1.055186 3883
Avg 1.2668 25.8503 20.8447 0.1 0.2887 28.5338 -0.5 5 1.0661718 3883
3.4.3.9. glcCluster
3.4.3.9.1. Stopping Criteria:
Maximum Iterations = 10000;
Maximum Function count = 10000;
Tolerance of Variables = 10-5
Function Tolerance =10-7
107
Table.3.16. Results of glcCluster in 20 trails for office thermal comfort
Trails Fcl Ta Tmrt Vair Pa Tcl PMV PPD Time Iters
1 0.76 24.9946 21.4981 0.852 0.3922 28.6711 -0.5 5 0.724438 1532/1
2 0.76 24.9946 21.4981 0.852 0.3922 28.6711 -0.5 5 0.542579 1532/1
3 0.76 24.9946 21.4981 0.852 0.3922 28.6711 -0.5 5 0.620507 1532/1
4 0.76 24.9946 21.4981 0.852 0.3922 28.6711 -0.5 5 0.522383 1532/1
5 0.76 24.9946 21.4981 0.852 0.3922 28.6711 -0.5 5 0.60312 1532/1
6 0.76 24.9946 21.4981 0.852 0.3922 28.6711 -0.5 5 0.662002 1532/1
7 0.76 24.9946 21.4981 0.852 0.3922 28.6711 -0.5 5 0.523577 1532/1
8 0.76 24.9946 21.4981 0.852 0.3922 28.6711 -0.5 5 0.670174 1532/1
9 0.76 24.9946 21.4981 0.852 0.3922 28.6711 -0.5 5 0.686226 1532/1
10 0.76 24.9946 21.4981 0.852 0.3922 28.6711 -0.5 5 0.556734 1532/1
11 0.76 24.9946 21.4981 0.852 0.3922 28.6711 -0.5 5 0.683852 1532/1
12 0.76 24.9946 21.4981 0.852 0.3922 28.6711 -0.5 5 0.586134 1532/1
13 0.76 24.9946 21.4981 0.852 0.3922 28.6711 -0.5 5 0.67979 1532/1
14 0.76 24.9946 21.4981 0.852 0.3922 28.6711 -0.5 5 0.578231 1532/1
15 0.76 24.9946 21.4981 0.852 0.3922 28.6711 -0.5 5 0.597836 1532/1
16 0.76 24.9946 21.4981 0.852 0.3922 28.6711 -0.5 5 0.567322 1532/1
17 0.76 24.9946 21.4981 0.852 0.3922 28.6711 -0.5 5 0.686948 1532/1
18 0.76 24.9946 21.4981 0.852 0.3922 28.6711 -0.5 5 0.586635 1532/1
19 0.76 24.9946 21.4981 0.852 0.3922 28.6711 -0.5 5 0.657122 1532/1
20 0.76 24.9946 21.4981 0.852 0.3922 28.6711 -0.5 5 0.558994 1532/1
Avg 0.76 24.9946 21.4981 0.852 0.3922 28.6711 -0.5 5 0.6147302 1532/1
3.4.3.10. glcSolve
3.4.3.10.1. Stopping Criteria:
Max.Iterations is exceeded > No. of variables 1000.
Max.function evaluations > No. of variables 2000.
If the difference of objective function is < 10-6
108
Table.3.17. Results of glcSolve in 20 trails for office thermal comfort
Trails Fcl Ta Tmrt Vair Pa Tcl PMV PPD Time Iters
1 0.75 19 22.83333 0.25 0.0528 27.3333 -0.5 5 0.720697 1771
2 0.75 19 22.83333 0.25 0.0528 27.3333 -0.5 5 0.899943 1771
3 0.75 19 22.83333 0.25 0.0528 27.3333 -0.5 5 0.718765 1771
4 0.75 19 22.83333 0.25 0.0528 27.3333 -0.5 5 0.692557 1771
5 0.75 19 22.83333 0.25 0.0528 27.3333 -0.5 5 0.898273 1771
6 0.75 19 22.83333 0.25 0.0528 27.3333 -0.5 5 0.897523 1771
7 0.75 19 22.83333 0.25 0.0528 27.3333 -0.5 5 0.747152 1771
8 0.75 19 22.83333 0.25 0.0528 27.3333 -0.5 5 0.902511 1771
9 0.75 19 22.83333 0.25 0.0528 27.3333 -0.5 5 0.728462 1771
10 0.75 19 22.83333 0.25 0.0528 27.3333 -0.5 5 0.821457 1771
11 0.75 19 22.83333 0.25 0.0528 27.3333 -0.5 5 0.912196 1771
12 0.75 19 22.83333 0.25 0.0528 27.3333 -0.5 5 0.760455 1771
13 0.75 19 22.83333 0.25 0.0528 27.3333 -0.5 5 0.844765 1771
14 0.75 19 22.83333 0.25 0.0528 27.3333 -0.5 5 0.742261 1771
15 0.75 19 22.83333 0.25 0.0528 27.3333 -0.5 5 0.82484 1771
16 0.75 19 22.83333 0.25 0.0528 27.3333 -0.5 5 0.832487 1771
17 0.75 19 22.83333 0.25 0.0528 27.3333 -0.5 5 0.700963 1771
18 0.75 19 22.83333 0.25 0.0528 27.3333 -0.5 5 0.655894 1771
19 0.75 19 22.83333 0.25 0.0528 27.3333 -0.5 5 0.8988886 1771
20 0.75 19 22.83333 0.25 0.0528 27.3333 -0.5 5 0.739011 1771
Avg 0.75 19 22.83333 0.25 0.0528 27.3333 -0.5 5 0.79695503 1771
3.4.4. Comparison Table.
Table.3.18. Comparative results of optimization methods for office thermal comfort.
PMV PPD -OFFICE
Methods Fcl Ta Tmrt Vair Pa Tcl PM
V
PP
D
TIME ITER
S Genetic
algorithm
0.7018 21.097 20.734 0.5073 0.2702 28.118 -0.5 5 0.539065 51
Simulated
annealing
0.7109 20.647 21.367 0.5418 0.5946 28.231 -0.5 5 4.103220 3001
PS 0.75 25 19.545 1 0.255 28 -0.5 5 0.397420 26
PSO 0.8069
7
22.199
25
21.584
57
0.5483
6
0.4737
7
27.858
13
-0.5 5 0.095456
6
51
Godlike 0.8489
0
22.663
93
20.968
14
0.4403
8
0.5558
95
28.042
83
-0.5 5 3.002659
9
4
Fmincon 0.8643
1
23.182
58
21.266
76
0.5556
85
0.5348
1
28.226
89
-0.5 5 14.68332 2288
DE
optimizati
on
SOLUTIO
N
0.8890
7
23.460
64
21.360
51
0.6645
25
0.4452
9
28.207
42
0.36
8
5 0.557024
25
12000
LGO 1.2668 25.850
3
20.844
7
0.1 0.2887 28.533
8
-0.5 5 1.066171
8
3883
glcCluster 0.76 24.994
6
21.498
1
0.852 0.3922 28.671
1
-0.5 5 0.614730
2
1532
glcSolve 0.75 19 22.833
33
0.25 0.0528 27.333
3
-0.5 5 0.796955
03
1771
109
-5
0
5
10
15
20
25
30
35
Fcl Ta Tmrt Vair Pa Tcl PMV PPD TIME
GENETIC ALGORITHM
SIMULATED ANNEALING
PATTERN SEARCH
PSO
GODLIKE
NON LINEAR
NUMERICAL optimizaion SOLUTION
LGO
glcCluster
glcSolve
Analytical
Fig.3.12. Comparative graph for office thermal comfort
The PMV and PPD have the same value as -0.5 and 5 for all the ten
optimization techniques except for DE, which has 0.36 as PMV. The elapsed time is
maximum for fmincon and minimum for PSO and PS. All the other parameter values
are more or less the same for all the ten optimization techniques. Now, the parameter
values are taken separately and the ten optimization techniques need to be compared
so as to find which method is the best method of optimization.
110
3.4.5. PARAMETERS.
3.4.5.1. Ratio of clothed body surface area to body area exposed
when undressed (Fcl):
The heat produced must be dissipated to the environment, or a change in body
temperature will occur. The deep body temperature is about 37°C, whilst the skin temperature
can vary between 31°C and 34°C under comfort conditions. Variations occur in time, but also
between parts of the body, depending on clothing cover and blood circulation. There is a
continuous transport of heat from deep tissues to the skin surface, from where it is dissipated
by radiation, convection or (possibly) conduction and evaporation.
Table.3.19. Fcl results in all 10 methods
Trail
s
GA SA PS PSO G-L fminco
n
DE LGO glcCluste
r
glcSolv
e 1 0.8568 0.3723 0.75 0.877 0.9215 0.4871 0.9915 1.266
8
0.76 0.75
2 0.8472 0.7477 0.75 0.5958 0.7984 0.3358 1.0001 1.266
8
0.76 0.75
3 0.8571 0.6201 0.75 1.4631 0.8846 1.1757 0.4368 1.266
8
0.76 0.75
4 0.6605 0.5951 0.75 0.6583 0.6674 0.6146 0.3405 1.266
8
0.76 0.75
5 1.5 1.1091 0.75 0.6525 0.7742 1.3053 1.2879 1.266
8
0.76 0.75
6 0.5884 0.7143 0.75 1.0041 0.3609 1.3235 1.4663 1.266
8
0.76 0.75
7 0.3902 0.4554 0.75 0.7453 0.5766 0.6935 0.4661 1.266
8
0.76 0.75
8 0.5843 0.5366 0.75 1.1379 1.2381 1.1925 0.503 1.266
8
0.76 0.75
9 0.8079 0.7255 0.75 0.6253 0.7883 1.1343 1.3263 1.266
8
0.76 0.75
10 0.7726 0.9322 0.75 0.6715 1.2641 0.9752 0.5029 1.266
8
0.76 0.75
11 0.4625 1.2129 0.75 0.6897 1.3031 0.5855 1.2868 1.266
8
0.76 0.75
12 0.6641 0.3967 0.75 0.6741 0.6165 0.6331 0.4494 1.266
8
0.76 0.75
13 0.4865 1.016 0.75 0.9268 0.6843 0.7597 0.5761 1.266
8
0.76 0.75
14 0.6368 0.4847 0.75 0.6827 1.1984 0.6737 1.1215 1.266
8
0.76 0.75
15 0.7589 1.251 0.75 1.141 0.9948 0.7 0.5254 1.266
8
0.76 0.75
16 0.5228 0.9353 0.75 0.8636 0.8568 1.2727 0.784 1.266
8
0.76 0.75
17 0.7037 0.3621 0.75 0.5778 0.806 0.7885 0.6666 1.266
8
0.76 0.75
18 0.661 0.345 0.75 0.6144 0.5515 1.0822 1.4845 1.266
8
0.76 0.75
19 0.5299 0.3584 0.75 0.4588 0.6032 0.837 1.2018 1.266
8
0.76 0.75
20 0.7467 1.0492 0.75 1.0798 1.0894 0.7163 1.3639 1.266
8
0.76 0.75
Avg 0.70189
5
0.7109
8
0.75 0.8069
75
0.84890
5
0.8643
1
0.8890
7
1.266
8
0.76 0.75
111
Fig.3.13 Graph for Fcl results in all 10 methods
0 2 4 6 8 10 12 14 16 18 20 22
0.51.01.5
Trails
GA
0.51.0
SA
01
PS
0.51.01.5
PS
O
0.51.0
GO
D-L
0.51.01.5
NL
P
0.51.01.5
DE
12
LG
O
01
glc
Clu
01
glc
So
l
112
3.4.5.2. Air Temperature-Ta
It is the temperature of the air surrounding the occupant. Operative
temperature is the uniform temperature of an imaginary enclosure in which the
occupant would exchange the same heat by radiation and convection as in the actual
environment. When air temperature is low, convective heat loss increases with air
motion associated with increased activity, thereby decreasing the heat load on the
body evaporative system and resulting in a wider range of activity before discomfort
is felt.
Table.3.20 Ta results in all 10 methods
Trials GA SA PS PSO G-L fmincon DE LGO glcClu glcSol
ve 1 25.18 16.82 25 23.08 22.11 21.77 24.65 25.85 24.99 19
2 25.64 17.17 25 21.13 23.28 18.08 26.13 25.85 24.99 19
3 24.55 22.01 25 25.07 19.70 28.00 19.53 25.85 24.99 19
4 22.23 16.97 25 21.73 21.76 21.31 19.33 25.85 24.99 19
5 16 27.33 25 21.40 23.34 27.48 26.83 25.85 24.99 19
6 22.33 22.54 25 24.76 17.51 26.38 27.95 25.85 24.99 19
7 17.42 17.44 25 21.87 19.82 21.80 22.26 25.85 24.99 19
8 16.07 21.34 25 26.04 26.43 26.49 19.38 25.85 24.99 19
9 22.06 18.79 25 21.66 21.78 26.28 27.57 25.85 24.99 19
10 24.39 25.29 25 20.43 24.99 22.44 20.73 25.85 24.99 19
11 18.08 25.97 25 21.58 26.49 20.22 27.19 25.85 24.99 19
12 22.95 17.17 25 22.12 19.95 22.76 17.97 25.85 24.99 19
13 19.43 24.80 25 22.47 24.19 24.18 18.17 25.85 24.99 19
14 23.88 16.00 25 17.37 24.62 21.04 27.52 25.85 24.99 19
15 22.55 26.87 25 25.56 25.37 24.03 21.08 25.85 24.99 19
16 18.12 20.48 25 25.50 22.55 26.67 25.14 25.85 24.99 19
17 20.67 16.28 25 18.46 23.11 18.26 17.31 25.85 24.99 19
18 16.01 16.40 25 18.63 20.94 23.79 28.29 25.85 24.99 19
19 19.69 17.70 25 18.92 21.66 25.02 26.35 25.85 24.99 19
20 24.61 25.48 25 26.10 23.58 17.56 25.74 25.85 24.99 19
avg 21.09 20.64 25 22.19 22.66 23.18 23.46 25.85 24.99 19
113
FIG.3.14. Graph for Ta results in all 10 methods
0 2 4 6 8 10 12 14 16 18 20 22
152025
Trails
GA
152025
SA
22242628
PS
2025
PS
O
2025
GO
D-L
2025
NL
P
202530
DE
242628
LG
O
22242628
glc
Clu
1820
glc
So
l
114
3.4.5.3. Mean radiant temperature-Tmrt
It is the uniform surface temperature of an imaginary black enclosure in which
an occupant would exchange the same amount of radiant heat as in the actual non
uniform space. The MRT affects the rate of radiant heat loss from the body. Since the
surrounding surface temperatures may vary widely, the MRT is a weighted average of
all radiating surface temperatures within line of sight. In winter, levels of wall, roof,
and floor insulation together with window treatments such as double glazing, blinds,
and drapes contribute to Mean Radiant Temperature.
Table.3.21. Tmrt results in all 10 methods
Trials GA SA PS PSO G-L fmincon DE LGO glcClu glcSol
1 20.71 22.30 19.54 22.85 20.56 19.5 22.59 20.84 21.49 22.83
2 20.33 22.58 19.54 22.82 22.95 19.86 20.42 20.84 21.49 22.83
3 19.50 20.16 19.54 23.29 21.42 20.12 22.94 20.84 21.49 22.83
4 19.50 20.37 19.54 21.11 20.51 21.17 19.82 20.84 21.49 22.83
5 19.5 20.40 19.54 22.06 19.83 22.25 21.27 20.84 21.49 22.83
6 21.69 22.07 19.54 20.84 20.08 23.49 22.33 20.84 21.49 22.83
7 20.81 22.19 19.54 21.27 22.43 21.03 20.13 20.84 21.49 22.83
8 19.78 20.09 19.54 20.08 21.55 22.64 21.52 20.84 21.49 22.83
9 19.5 20.92 19.54 21.09 20.80 19.79 21.62 20.84 21.49 22.83
10 22.32 20.75 19.54 22.78 20.24 19.53 19.99 20.84 21.49 22.83
11 22.49 23.06 19.54 20.55 20.31 19.96 21.68 20.84 21.49 22.83
12 22.46 19.52 19.54 22.86 21.78 23.40 23.43 20.84 21.49 22.83
13 22.34 21.00 19.54 21.97 20.20 21.18 21.16 20.84 21.49 22.83
14 19.50 23.06 19.54 22.23 21.00 22.01 20.47 20.84 21.49 22.83
15 22.55 23.41 19.54 20.38 19.77 21.11 23.29 20.84 21.49 22.83
16 20.96 23.35 19.54 22.15 22.06 22.53 20.43 20.84 21.49 22.83
17 20.60 20.54 19.54 21.13 20.82 22.57 21.22 20.84 21.49 22.83
18 19.50 20.94 19.54 20.22 19.70 22.95 20.56 20.84 21.49 22.83
19 20.09 20.75 19.54 20.61 20.03 20.64 20.26 20.84 21.49 22.83
20 20.49 19.77 19.54 21.29 23.20 19.50 21.96 20.84 21.49 22.83
avg 20.73 21.36 19.54 21.584 20.96 21.26 21.36 20.84 21.49 22.83
115
FIG.3.15. Graph for T mrt results in all 10 methods
0 2 4 6 8 10 12 14 16 18 20 22
20
22
Trails
GA
202224
SA
182022
PS
202224
PS
O
202224
GO
D-L
202224
NL
P
202224
DE
182022
LG
O
202224
glc
Clu
20222426
glc
So
l
116
3.4.5.4. Velocity of air-Vair
Air motion significantly affects body heat transfer by convection and
evaporation. Air Movement results from free convection from the occupants‘ body
movements. The faster the motion, the greater the rate of heat flow by both convection
and evaporation. When ambient temperatures are within acceptable limits, there is no
minimum air movement that must be provided for thermal comfort. The natural
convection of air over the surface of the body allows for the continuous dissipation of
body heat. When ambient temperatures rise, however, natural air flow velocity is no
longer sufficient and must be artificially increased, such as the use of fans.
Table.3.22. Vair results in all 10 methods
Trial
s
GA SA PS PSO G-L fmincon DE LGO glcClu glcSol
ve 1 0.4163 0.7753 1 0.4892 0.2555 0.6415 0.6312 0.1 0.852 0.25
2 0.6048 0.1059 1 0.714 0.6954 0.9994 0.9282 0.1 0.852 0.25
3 0.4337 0.8629 1 0.4324 0.1635 0.6074 0.6933 0.1 0.852 0.25
4 0.3322 0.1619 1 0.5704 0.4172 0.567 0.8445 0.1 0.852 0.25
5 1 0.8993 1 0.3756 0.2137 0.9583 0.3203 0.1 0.852 0.25
6 0.5732 0.3882 1 0.8217 0.7025 0.4962 0.9047 0.1 0.852 0.25
7 0.8219 0.6605 1 0.5292 0.4799 0.5311 0.9318 0.1 0.852 0.25
8 0.2075 0.5283 1 0.8315 0.8851 0.5357 0.4682 0.1 0.852 0.25
9 0.3219 0.1864 1 0.6851 0.3325 0.4584 0.9202 0.1 0.852 0.25
10 0.9572 0.88 1 0.5851 0.2914 0.1832 0.7276 0.1 0.852 0.25
11 0.5298 0.3384 1 0.3421 0.5386 0.5444 0.3272 0.1 0.852 0.25
12 0.5219 0.6171 1 0.6112 0.4122 0.8145 0.6516 0.1 0.852 0.25
13 0.544 0.7406 1 0.4605 0.6888 0.6327 0.1972 0.1 0.852 0.25
14 0.6699 0.501 1 0.1385 0.2734 0.4636 0.8138 0.1 0.852 0.25
15 0.3427 0.7251 1 0.7445 0.6646 0.5247 0.8569 0.1 0.852 0.25
16 0.3694 0.1291 1 0.8741 0.2081 0.7674 0.9267 0.1 0.852 0.25
17 0.2374 0.6345 1 0.368 0.3435 0.1614 0.1439 0.1 0.852 0.25
18 0.1001 0.7779 1 0.1903 0.5269 0.3627 0.8074 0.1 0.852 0.25
19 0.3646 0.7888 1 0.5844 0.4603 0.7431 0.4225 0.1 0.852 0.25
20 0.7986 0.1365 1 0.6194 0.2545 0.121 0.7733 0.1 0.852 0.25
avg 0.5073
55
0.5418
85
1 0.5483
6
0.4403
8
0.55568
5
0.6645
25
0.1 0.852 0.25
117
FIG.3.16. Graph for Vair results in all 10 methods
0 2 4 6 8 10 12 14 16 18 20 22
0.00.51.0
Trails
GA
0.0
0.5
1.0
SA
012
PS
0.20.40.60.8
PS
O
0.20.40.60.8
GO
D-L
0.00.51.0
NL
P
0.00.20.40.60.81.0
DE
0
1
LG
O
01
glc
Clu
0
1
glc
So
l
118
3.4.5.5. Partial water vapour pressure-Pa
The upper and lower humidity limits on the comfort envelope are based on
considerations of respiratory health, growth, and other moisture-related phenomena in
addition to comfort. Humidification in winter must be limited at times to prevent
condensation on cold building surfaces such as windows. The environmental
parameters of temperature, radiation, humidity, and air movement are necessary for
thermal comfort, depending upon the occupant‘s clothing and activity level.
Table.3.23. Pa results in all 10 methods
Trial
s
GA SA PS PSO G-L fmincon DE LGO glcClust
er
glcSol
ve 1 0.0103 0.7548 0.25
5
0.870
2
0.386
9
0.2421 0.785
1
0.288
7
0.3922 0.0528
2 0.1861 0.9133 0.25
5
0.218
6
0.544
5
0.9971 0.829
3
0.288
7
0.3922 0.0528
3 0.0101 0.3263 0.25
5
0.806
7
0.545
6
0.2652 0.133
5
0.288
7
0.3922 0.0528
4 0.6014 0.032 0.25
5
0.286
4
0.462
3
0.4379 0.361
6
0.288
7
0.3922 0.0528
5 0.01 0.2761 0.25
5
0.241
7
0.133
3
0.9412 0.735
4
0.288
7
0.3922 0.0528
6 0.1797 0.5892 0.25
5
0.738
8
0.564
1
0.8329 0.295
8
0.288
7
0.3922 0.0528
7 0.7775 0.9469 0.25
5
0.605
8
0.347
2
0.2651 0.239
9
0.288
7
0.3922 0.0528
8 0.307 0.1032 0.25
5
0.661
9
0.577
7
0.0216 0.199
1
0.288
7
0.3922 0.0528
9 0.0817 0.5557 0.25
5
0.579
6
0.412
8
0.6802 0.121
9
0.288
7
0.3922 0.0528
10 0.4146 0.6343 0.25
5
0.195
6
0.704
4
0.6243 0.063
2
0.288
7
0.3922 0.0528
11 0.2691 0.7721 0.25
5
0.962
5
0.593
6
0.1339 0.476 0.288
7
0.3922 0.0528
12 0.2602 0.3943 0.25
5
0.526
6
0.539
5
0.368 0.171
8
0.288
7
0.3922 0.0528
13 0.2023 0.8484 0.25
5
0.820
4
0.444
7
0.4833 0.060
9
0.288
7
0.3922 0.0528
14 0.3043 0.6394 0.25
5
0.142
5
0.895
1
0.5914 0.045
5
0.288
7
0.3922 0.0528
15 0.532 0.355 0.25
5
0.157
2
0.706
4
0.1132 0.848
9
0.288
7
0.3922 0.0528
16 0.0101 0.5949 0.25
5
0.172
2
0.157
6
0.9948 0.668
6
0.288
7
0.3922 0.0528
17 0.323 0.8534 0.25
5
0.467
6
0.740
4
0.4501 0.815
3
0.288
7
0.3922 0.0528
18 0.1524 0.5505 0.25
5
0.668
4
0.648
6
0.6277 0.543
8
0.288
7
0.3922 0.0528
19 0.7578 0.7585 0.25
5
0.269
3
0.932
8
0.8272 0.724
5
0.288
7
0.3922 0.0528
20 0.0151 0.995 0.25
5
0.083
4
0.780
4
0.799 0.785
7
0.288
7
0.3922 0.0528
avg 0.2702
35
0.5946
65
0.25
5
0.473
77
0.555
895
0.53481 0.445
29
0.288
7
0.3922 0.0528
119
FIG.3.17. Graph for Pa results in all 10 methods
0 2 4 6 8 10 12 14 16 18 20 22
0.0
0.5
Trails
GA
0.00.51.0
SA
0
1
PS
0.00.51.0
PS
O
0.0
0.5
1.0
GO
D-L
0.00.51.0
NL
P
0.00.51.0
DE
0
1
LG
O
0
1
glc
Clu
0
1g
lcS
ol
120
3.4.5.6. Surface temperature of clothing-Tcl
Clothing, through its insulation properties, is an important modifier of body
heat loss and comfort. The insulation properties of clothing are, a result of the small
air pockets separated from each other to pre air from migrating through the material.
When preferred amount of clothing worn by building occupants decreased, then
correspondingly the preferred temperatures increased. These seasonal clothing
variations of building occupants allow indoor temperature ranges to be higher in the
summer than in the winter and yet give the occupants comfort. During winter,
additional clothing lowers the ambient temperature necessary for comfort and for
thermal neutrality.
Table.3.24. Tcl results in all 10 methods
Trial GA SA PS PSO G-L fminco
nn
DE LGO glcClu glcSol
ve 1 28.03
43
28.28
02
28 28.038
1
27.1729 28.9569 27.54
55
28.53
38
28.6711 27.333
3 2 28.63
77
28.48
96
28 27.781
2
27.9704 28.7899 27.98
05
28.53
38
28.6711 27.333
3 3 27.87
34
27.13
08
28 27.600
6
27.2568 28.9529 28.51
43
28.53
38
28.6711 27.333
3 4 28.65
82
28.15
65
28 27.628
8
28.0892 27.8729 28.96
41
28.53
38
28.6711 27.333
3 5 27 28.58
98
28 28.431
5
28.8363 28.8146 28.49
21
28.53
38
28.6711 27.333
3 6 28.85
52
28.88
99
28 27.348
2
28.7253 28.7517 28.59
48
28.53
38
28.6711 27.333
3 7 27.66
83
27.58
67
28 27.337
3
28.0491 27.4699 28.64
77
28.53
38
28.6711 27.333
3 8 27.03
57
28.48
58
28 27.623
6
27.9811 28.6283 28.12
79
28.53
38
28.6711 27.333
3 9 27.00
81
27.68
64
28 27.652
2
27.5502 28.0671 28.46
12
28.53
38
28.6711 27.333
3 10 28.22
2
27.89
03
28 27.012
3
27.2606 27.1956 27.47
1
28.53
38
28.6711 27.333
3 11 28.28
53
28.96
31
28 28.450
5
27.7716 27.0043 28.78
73
28.53
38
28.6711 27.333
3 12 28.98
03
27.73
24
28 28.250
8
27.9468 28.5431 27.69
12
28.53
38
28.6711 27.333
3 13 28.61 27.50
5
28 27.159
1
28.6017 28.4621 28.37
09
28.53
38
28.6711 27.333
3 14 28.57
87
27.09
49
28 28.499
4
27.6636 27.8835 28.29
07
28.53
38
28.6711 27.333
3 15 28.89
37
28.84
49
28 27.280
6
27.7013 28.9058 28.49
28
28.53
38
28.6711 27.333
3 16 27.73
46
28.68
46
28 28.560
5
28.6793 28.5129 28.01
96
28.53
38
28.6711 27.333
3 17 28.20
76
28.60
65
28 27.513
7
28.3673 27.7677 28.35
15
28.53
38
28.6711 27.333
3 18 27.03
32
28.19
58
28 28.745 28.2285 27.9079 28.48
06
28.53
38
28.6711 27.333
3 19 28.89
91
28.82
26
28 28.042
5
28.6328 28.3517 27.54
71
28.53
38
28.6711 27.333
3 20 28.15
17
28.99
08
28 28.206
6
28.3717 27.699 27.31
76
28.53
38
28.6711 27.333
3 avg 28.11
836
28.23
133
28 27.858
13
28.0428
3
28.2268
9
28.20
742
28.53
38
28.6711 27.333
3
121
FIG.3.18.Graph for Tcl results in all 10 methods
0 2 4 6 8 10 12 14 16 18 20 22
26
28
30
Trails
GA
26
28
30
SA
262830
PS
26272829
PS
O
27
28
29
GO
D-L
26
28
30
NL
P
28
30
DE
25
30
LG
O
25
30
glc
Clu
25
30
glc
So
l
122
3.4.5.7. Predicted mean vote (PMV):
It is an index that predicts the mean value of the votes of a large group
of persons on the seven point thermal sensation scale. The existing conditions
may not be amendable to every occupant. Each person has a distinct perception
of too hot, too cold, and comfortable. The objective in designing a common
thermal environment is to satisfy a majority of occupants and to minimize the
number of people who will inevitably be dissatisfied.
Table.3.25. PMV results in all 10 methods
Trials GA SA PS PSO G-L fmincon DE LGO glcClu glcSol
1 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5 0.1317 -0.5 -0.5 -0.5
2 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5 0.3486 -0.5 -0.5 -0.5
3 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5 0.3758 -0.5 -0.5 -0.5
4 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5 0.3095 -0.5 -0.5 -0.5
5 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5 0.3896 -0.5 -0.5 -0.5
6 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5 0.798 -0.5 -0.5 -0.5
7 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5 0.4113 -0.5 -0.5 -0.5
8 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5 0.3861 -0.5 -0.5 -0.5
9 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5 0.425 -0.5 -0.5 -0.5
10 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5 0.4183 -0.5 -0.5 -0.5
11 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5 0.399 -0.5 -0.5 -0.5
12 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5 0.4174 -0.5 -0.5 -0.5
13 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5 0.3551 -0.5 -0.5 -0.5
14 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5 0.1632 -0.5 -0.5 -0.5
15 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5 0.4765 -0.5 -0.5 -0.5
16 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5 0.2925 -0.5 -0.5 -0.5
17 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5 0.4286 -0.5 -0.5 -0.5
18 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5 0.3521 -0.5 -0.5 -0.5
19 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5 0.1079 -0.5 -0.5 -0.5
20 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5 0.3738 -0.5 -0.5 -0.5
avg -0.5 -0.5 -0.5 -0.5 -0.5 -0.5 0.368 -0.5 -0.5 -0.5
123
FIG.3.19. Graph for PMV results in all 10 methods
0 2 4 6 8 10 12 14 16 18 20 22
-1
0
TRAILS
GA
-1
0
SA
-1
0
PS
-1
0
PS
O
-1
0
GO
D-L
-1
0
NL
P
0.20.40.60.8
DE
-1
0
LG
O
-1
0
glc
Clu
-1
0
glc
So
l
124
3.4.5.8. Predicted percentage of Dissatisfied. (PPD):
An index that establishes a quantitative prediction of the percentage of
thermally dissatisfied people determined from PMV. As PMV changes from zero in
either the positive or negative direction, PPD increases. Determination of the PMV
and PPD indices and Specification of the Conditions for Thermal Comfort uses, limits
on PMV as an explicit definition of the comfort zone.
TABLE.3.26. PPD results in all 10 methods
Trials GA SA PS PSO G-L fmincon DE LGO glcClu glcSol
1 5 5 5 5 5 5 5 5 5 5
2 5 5 5 5 5 5 5 5 5 5
3 5 5 5 5 5 5 5 5 5 5
4 5 5 5 5 5 5 5 5 5 5
5 5 5 5 5 5 5 5 5 5 5
6 5 5 5 5 5 5 5 5 5 5
7 5 5 5 5 5 5 5 5 5 5
8 5 5 5 5 5 5 5 5 5 5
9 5 5 5 5 5 5 5 5 5 5
10 5 5 5 5 5 5 5 5 5 5
11 5 5 5 5 5 5 5 5 5 5
12 5 5 5 5 5 5 5 5 5 5
13 5 5 5 5 5 5 5 5 5 5
14 5 5 5 5 5 5 5 5 5 5
15 5 5 5 5 5 5 5 5 5 5
16 5 5 5 5 5 5 5 5 5 5
17 5 5 5 5 5 5 5 5 5 5
18 5 5 5 5 5 5 5 5 5 5
19 5 5 5 5 5 5 5 5 5 5
20 5 5 5 5 5 5 5 5 5 5
Avg 5 5 5 5 5 5 5 5 5 5
125
FIG.3.20. Graph for PPD results in all 10 methods
0 2 4 6 8 10 12 14 16 18 20 22
4
6
TRAILS
GA
4
6
SA
4
6
PS
4
6
PS
O
4
6
GO
D-L
4
6
NL
P
4
6
DE
4
6
LG
O
4
6
glc
Clu
4
6
glc
So
l
126
3.4.5.9 Elapsed Time
CPU time is the time for which the CPU was busy executing the task. It does
not take into account the time spent in waiting for I/O (disk IO or network IO). Since
I/O operations, such as reading files from disk, are performed by the OS, these
operations may involve noticeable amount of time in waiting for the I/O subsystems to
complete their operations. This waiting time will be included in the elapsed time, but
not in CPU time. Hence CPU time is usually less than the elapsed time.
TABLE.3.27. Elapsed time results in all 10 methods
Trial
llll
GA SA PS PSO G-L fmincon DE LGO glcClu
ster
glcSol
ve 1 0.52 5.16 0.825
115
0.097
091
3.029
002
16.6448
31
0.5458
45
0.956
907
0.7244
38
0.720
697 2 0.54 4.10 0.377
779
0.093
681
3.227
48
15.1560
49
0.5622
79
1.132
054
0.5425
79
0.899
943 3 0.55 3.71 0.359
182
0.098
27
2.646
517
12.9088
19
0.5249
03
1.326
114
0.6205
07
0.718
765 4 0.53 3.43 0.365
426
0.092
962
3.022
442
14.3027
52
0.5752
98
1.255
759
0.5223
83
0.692
557 5 0.54 6.138
007
0.378
375
0.100
452
2.566
037
12.1787
76
0.5230
7
1.080
224
0.6031
2
0.898
273 6 0.53 4.343
375
0.371
941
0.091
094
3.410
476
22.8326
64
0.6029
35
1.138
645
0.6620
02
0.897
523 7 0.53 4.167
754
0.388
65
0.091
199
3.065
694
12.4062
42
0.5822
6
1.069
755
0.5235
77
0.747
152 8 0.52 4.234
063
0.378
278
0.096
344
2.268
675
13.1254
5
0.5843
69
1.014
94
0.6701
74
0.902
511 9 0.54 3.682
238
0.375
717
0.089
281
3.435
452
14.5899
27
0.5403
72
1.039
328
0.6862
26
0.728
462 10 0.54 3.680
518
0.378
74
0.093
782
2.810
391
14.2329
38
0.5442
69
1.033
702
0.5567
34
0.821
457 11 0.52 4.137
636
0.373
859
0.098
879
2.445
6
15.4775
48
0.5628
64
1.069
969
0.6838
52
0.912
196 12 0.53 4.064
644
0.373
993
0.094
448
2.808
384
14.8478
26
0.5632
16
1.026
883
0.5861
34
0.760
455 13 0.54 3.639
53
0.401
204
0.098
272
3.492
864
16.0136
6
0.5594
04
1.024
515
0.6797
9
0.844
765 14 0.53 3.121
178
0.376
038
0.090
875
2.991
398
14.2385
94
0.5387
61
1.073
316
0.5782
31
0.742
261 15 0.54 5.400
893
0.370
777
0.101
924
2.709
29
14.6281
95
0.5301
71
1.038
982
0.5978
36
0.824
84 16 0.53 3.347
213
0.379
511
0.096
871
3.776
228
13.5420
12
0.5641
29
0.986
438
0.5673
22
0.832
487 17 0.53 4.270
551
0.372
559
0.097
384
3.294
535
17.6474
16
0.5537
67
0.954
037
0.6869
48
0.700
963 18 0.53 4.720
778
0.373
777
0.100
313
3.141
239
13.5885
38
0.5278
86
0.975
41
0.5866
35
0.655
894 19 0.53 3.256
275
0.357
336
0.094
422
2.929
281
12.3144
53
0.5743
58
1.071
272
0.6571
22
0.898
889 20 0.55 3.444
622
0.370
143
0.091
588
2.982
213
12.9897
1
0.5803
29
1.055
186
0.5589
94
0.739
011 Avg 0.53 4.103 0.397 0.095 3.002 14.6833 0.5570 1.066 0.6147 0.796
127
FIG.3.21. Graph for elapsed time in all 10 methods
0 2 4 6 8 10 12 14 16 18 20 22
0.530.540.55
Trails
GA
246
SA
0.40.60.8
PS
0.0900.0950.100
PS
O
2
3
4
GO
D-L
1520
NL
P
0.55
0.60
DE
1.01.2
LO
G
0.50.60.7
glc
Clu
0.70.80.9
glc
So
l
128
3.4.5.10. Iterations
Iteration is a computational procedure in which a cycle of operations is
repeated, often to approximate the desired result more closely. Iteration means the act
of repeating a process usually with the aim of approaching a desired goal or target or
result. Iteration in computing is the repetition of a process within a computer program.
It may also refer to the process of iterating a function i.e. applying a function
repeatedly, using the output from one iteration as the input to the next. Another use of
iteration in mathematics is in iterative methods which are used to produce
approximate numerical solutions to certain mathematical problems.
TABLE.3.28. Iterations results in all 10 methods
Trials GA SA PS PSO G-L fmincon DE LGO glcClu glcSol
1 51 3006 26 51 4 2282 60 3883 1532 1771
2 51 3001 26 51 4 2282 60 3883 1532 1771
3 51 3001 26 51 4 2254 60 3883 1532 1771
4 51 3001 26 51 4 2296 60 3883 1532 1771
5 51 3001 26 51 4 2324 60 3883 1532 1771
6 51 3001 26 51 4 2303 60 3883 1532 1771
7 51 3002 26 51 4 2282 60 3883 1532 1771
8 51 3001 26 51 4 2275 60 3883 1532 1771
9 51 3001 26 51 4 2289 60 3883 1532 1771
10 51 3002 26 51 4 2310 60 3883 1532 1771
11 51 3001 26 51 4 2268 60 3883 1532 1771
12 51 3003 26 51 4 2261 60 3883 1532 1771
13 51 3001 26 51 4 2310 60 3883 1532 1771
14 51 3001 26 51 4 2282 60 3883 1532 1771
15 51 3001 26 51 4 2289 60 3883 1532 1771
16 51 3001 26 51 4 2296 60 3883 1532 1771
17 51 3002 26 51 4 2275 60 3883 1532 1771
18 51 3001 26 51 4 2338 60 3883 1532 1771
19 51 3001 26 51 4 2289 60 3883 1532 1771
20 51 3001 26 51 4 2268 60 3883 1532 1771
avg 51 3001.5 26 51 4 2288.65 60 3883 1532 1771
129
FIG.3.22. Graph for Iterations results in all 10 methods
0 2 4 6 8 10 12 14 16 18 20 22
455055
Trails
GA
3000
3005
SA
242628
PS
455055
PS
O
345
GO
D-L
225023002350
NL
P
556065
DE
3500
4000
LG
O
140015001600
glc
Clu
1600170018001900
glc
So
l
130
The following Table exhibit the consistency of the methods for different parameters
and the corresponding values.
TABLE.3.29. Comparative Table for parameters in all 10 methods
Variable GA SA PS PSO GL Fmincon DE LGO Glc
Cluster
Glc
Solve
Fcl X X
0.75 X X X X
1.26
0.76
0.75
Ta X X
25 X X X
25.8
24.9
19 Tmrt X X
19.54 X X X
20.84
21.49
22.83
Vair X X
1 X X X
0.1
0.852
0.25
Pa √
0.255
0.28
0.39
0.05
Tcl X X 28
X X X 28.53
3
28.67
27.33
PMV -0.5
-0.5
-0.5
-0.5
-0.5
-0.5
-0.5
-0.5
-0.5
-0.5
PPD
5
5
5
5
5
5
5
5
5
5
Time 0.39 0.095
Iters X X 26 4 X X X X X
3.4.6. Result and Discussion
With the two extreme values of parameters from survey, the optimization is
carried out with different solvers. As they are of the stochastic type, their results may
vary from trial to trial and the problem is made to run for 20 trials (Elbeltagi, Tarek
Hegazy, & & Grierson, 2005) and an average of all trials is taken as the final value of
the parameter, by the solver. The solvers are compared with three different criteria.
1. Consistency
The consistency Table gives the parameters that remain constant for all
the trials. All the solvers give the same value of PMV& PPD for all the
runs except DE, which in turn indicate that the comfort requirements are
in the acceptable range.
131
Fcl - P.S & glcSolve (0.75), glcCluster (0.76), LGO (1.26)
Ta - P.S (25), glcSolve (19), glcCluster (24.9), LGO (25.8)
Tmrt - P.S (19.54), glcSolve (22.83), glcCluster (21.49), LGO
(20.84)
Vair - P.S (1), glcSolve (0.25), glcCluster (0.852), LGO (0.1)
Pa - P.S (0.255), glcSolve (0.05), glcCluster (0.39), LGO (0.28)
Tcl - P.S (28), glcSolve (27.33), glcCluster (28.67), LGO (28.53)
So we see that the solvers Pattern Search, glcSolve, glcCluster&
LGO remain constant throughout their runs.
2. Minimum Run Time
For a minimum run time of the problem, we got PSO (0.095 seconds),
Pattern Search (0.39 seconds).
3. Minimum Evaluation
This criterion will determine the effectiveness of the algorithm. From
the result table, we see that the Pattern Search and GODLIKE
algorithms have minimum evaluation of 26 and 4 respectively.
4. Simplicity of Algorithm
Of all the algorithms we have taken, the Pattern Search algorithm is
the most simplest followed by GA, PSO, DE, Simulated Annealing,
GODLIKE, Non-Linear, Direct algorithm.
5. Results according to Standards
This is the most important criterion that determines whether the solver is
practical or not. We got the standard values for a naturally ventilated
building from ASHRAE as:
Humidity: 30% to 60%
(http://www.epa.gov/iaq/largebldgs/i-beam/text/hvac.html)
132
This gives that the Pa should lie within the range of: 0.0765 to
0.501
(http://www.engineeringtoolbox.com/water-vapor-saturation-pressure-
air-d_689.html)
Operative Temperature: 17.75 to 28.5
Air velocity:0.2 to 0.8 ms-1
(1 ms-1
only at extreme conditions)
With the above standards the solvers which adhere to the standard are:
Air-Velocity: Fmincon, GA, SA, PSO, GL, DE, glcSolve.
Partial vapour pressure: GA, PS, PSO, DE, LGO, glcCluster,
glcSolve
Operative temperature: GA, SA,PS, PSO, Fmincon, DE, GL,
LGO, glcCluster, glcSolve
The following Table gives a summary of all the criteria for the solvers:
Table 3.30. Summary of all the criteria for the solvers
Criteria GA SA PS PSO Fmincon DE GL LGO glcClus glcSolve
Result
according to
ASHRAE
3/3
=100%
2/3
=67%
2/3
=67%
3/3
=100%
2/3
=67%
3/3
=100%
2/3
=67%
2/3
=67%
2/3
=67% 3/3
=100%
Consistency - - - - - -
Min-Run
Time - - - - - - - -
Min-
Evaluation - - - - - - - -
Simple
Algorithm - - - - - - - - -
133
Thus it is seen that the Pattern Search solver satisfies all the criteria and scores
67% for its practicality in giving result according to ASHRAE. So the
appropriate algorithm, for optimization of thermal comfort is suggested as
Direct search algorithm & the solver is PATTERN SEARCH
3.4.7. Conclusion.
This study investigates thermal environment and comfort of office buildings in
the Karunya University. A total of 220 subjects in naturally ventilated 8 office
buildings ( with occupant – operable windows) provided 220 sets of cross-sectional
thermal comfort data, first field campaign from Mar 15, 2010 to Mar24,2010 and
second field campaign from Sep10,2010 to Sep 19, 2010 in Karunya University,
Coimbatore. In both the set, the same buildings were taken into account for data
collection. Indoor climatic data were collected, using instruments with accuracies with
the recommendations of ANSI/ASHRAE 55. All the measurements were carried out
between 10:00 hours and 16:00 hours.
In the experiment conducted using ten non-traditional optimization techniques,
the thermal sensation takes the value -0.5, which is in the acceptable range , where
the acceptable range is -0.5 to +0.5 (ANSI/ASHRAE55-2004, 2004). From the
thermal comfort value, we can conclude that the thermal comfort of the office
buildings of the Karunya University is in the acceptable range and hence the thermal
comfort in this area is optimum.
Here, ten non-traditional optimization algorithms were presented. These
include: GA, SA, PS, PSO, GL, FMINCON, EA, LGO, glcCluster, glcSolve. A brief
description of each method is presented along with a Pseudo code to facilitate their
implementation. MATLAB programs were written to implement each algorithm. The
thermal comfort problem for the offices of the Karunya University was solved using
all algorithms, and the comparative results were presented.
134
3.5. THERMAL COMFORT IN A RESIDENTIAL BUILDING
3.5.1. INTRODUCTION.
Thermal comfort can be defined as that condition of mind which expresses
satisfaction with the thermal environment. The reference to ‗mind‘ indicates that it is
essentially a subjective term; however, there has been extensive research in residential
thermal comfort and a number of indices exists which can be used to assess
environments for thermal comfort.
Predicted mean vote (PMV) is a parameter for assessing thermal comfort in an
occupied zone based on the conditions of metabolic rate, clothing, air speed, besides
temperature and humidity. PMV values refer the ASHRAE thermal sensation scale
that ranges from -3 to 3 as follows: 3=hot, 2=warm, 1=slightly warm, 0= neutral, -
1=slightly cool, -2=cool,-3=cold.
Predicted Percentage Dissatisfied (PPD) is used to estimate the thermal comfort
satisfaction of the occupant. It is considered that satisfying 80 of occupant is good;
that is, PPD less than 20% is good (ANSI/ASHRAE55-2004, 2004).
3.5.2. LITERATURE SURVEY
This approach is based on field surveys of thermal comfort and it demonstrates
that people are more tolerant of temperature changes than laboratory studies suggest:
they consciously and unconsciously act to affect the heat balance of the body
(behavioural thermoregulation). These actions may change metabolic heat production
(changing activity or doing something more or less vigorously), the rate of heat loss
from the body (clothing, posture) or the thermal environment (windows, doors, blinds,
fans, thermostat adjustment) (Humphreys, 1994). Comfort may therefore be achieved
in a wider range of temperatures than predicted by ASHRAE when it is something that
individuals achieve for themselves. Adaptive variables are extremely important in
‗free running‘ buildings – those without active heating or cooling systems (Nicol, Raja
et al. 1999).
135
People in such buildings need to be able to control their immediate
environment by opening and closing windows, dressing in such a way as to maximise
comfort indoors and outdoors, and using shading as necessary. Research into the
comfort levels of sedentary individuals at home, at work and in a climate chamber,
shows that simply being ‗at home‘, in an environment that is familiar and under
control, is conducive to comfort and makes people less sensitive to temperature
(Oseland 1995). Advocates of the adaptive approach argue that the heat-balance
approach can become unduly normative. For example, when people in hot climates
say that they do not experience discomfort at temperatures classified as ‗severe‘
according to the heat-balance model, it can be attributed to their ‗low expectations‘ of
comfort (Fanger and Toftum, 2002). The possibility that these individuals may, in
fact, be comfortable is ignored. Taking this argument further, Stoops (1994) claims
that an element of thermal discomfort – thermal experience, beyond the normal
comfort boundaries contributes to overall well-being.
This is demonstrated by those who exercise vigorously, use saunas and take
holidays in the sun or the snow. It is not far-fetched to claim that variation is an
element of comfort and that people will choose to avoid thermal monotony. Adaptive
thermal comfort is a function of the possibilities for change as well as the actual
temperatures achieved (Nicol and Humphreys, 2002). In the face of evidence from
real-life conditions, the argument goes; the controlled PMV method of estimating
comfort levels can be seriously misleading and needs revising (Humphreys and Nicol,
2002). Advocates of the adaptive approach hold that it will eventually be possible to
produce thermal standards for buildings that do not resort to specifications of the
indoor climate, but use characteristics of a building such as materials, orientation,
moveable shading, heating system and controls (Nicol & Humphreys, 2002). If
buildings are designed and built to incorporate the right mix of these characteristics,
the occupants will be able to make themselves comfortable within them.
136
3.5.3. Research methods.
3.5.3.1. Outdoor Climatic environment.
Under the Koppen climate classification, the Coimbatore city has a tropical wet
and dry climate. It has mild winters and moderate summers. Karunya University
residential buildings lie in the latitude of 100 56‘ 18.89‖ N and longitude of 76
0 45‘
17.44‖ E with elevation 1510 ft. The surveys in this study were performed in the May
2009 and September 2009.
3.5.3.2. Subjects
A Sample size of 102 subjects in 11 different residences in the Karunya
University was collected in survey and field measurements. The dwellings
interviewed are multi-story apartments. The volunteers participating in the study
comprised both men and women. The average age of all respondents was 33.2 years,
ranging from 20 to 57 years and all the participants were in good health. The
questionnaire covered several areas including Personal factors (name, gender, age,
etc.) and personnel environmental control. The questionnaire also included the
traditional scales of thermal sensation and thermal preferences, current clothing
garment and metabolic activity. The thermal sensation scale was the ASHRAE seven
point scale ranging from cold (-3) to hot (3) with neutral (0) in the middle. The three
point thermal preference scale asked whether the respondents would like to change
their present thermal environment. Possible responses were ―want warmer‖, ―no
change‖, or ―want cooler‖. Clothing garment check list were compiled from the
extensive lists published in ASHRAE -55, 2004. Metabolic rates were assessed by a
check of activities databases published in ASHRAE-55, 2004. The summary of the
background characteristics of the subjects are presented.
137
Table.3.31 Summary of the sample of residents and personal thermal variables
Sample size 102
Age (year)
Mean 33.2
Maximum 20
Minimum 57
Metabolic rate 115(W/m2)
1.10Clo Clothing insulation
3.5.3.3. Data collection
Both physical and subjective questionnaires were obtained
simultaneously in the visit of the filed survey. This study investigates the thermal
environment and comfort of residences in the Karunya University, Coimbatore. A
total of 102 occupants in naturally ventilated 11 residences buildings ( with occupant
– operable windows) provided thermal perception data, first field campaign from
Mar 6, 2010 to Mar 15,2010 and second field campaign from Sep 1,2010 to Sep 10,
2010 in Karunya University, Coimbatore. In both the sets, the same buildings were
taken into account for data collection. Indoor climatic data were collected using
instruments with accuracies and response times in accordance with the
recommendations of ANSI/ASHRAE 55. All the measurements were carried out
between 06:00 hours and 20:00 hours. All the houses where survey is conducted are
non-air-conditioned residences, where natural ventilation is preferred. The result of
the filed survey and measurement study can be used to design a low energy
consumption system with consideration of occupant thermal comfort in Coimbatore,
Tamil Nadu.
A number of instruments were used to measure the thermal environment
conditions, while the respondents filled in the subjective thermal comfort
questionnaire. The instruments were Standard thermometer for Air temperature,
Whirling hygrometer for humidity, Globe thermometer for radiant heat, kata
138
thermometer for air velocity. Metabolic rate can be estimated using standard Table
found in ISO 7730. Among the residential respondents, air temperature readings were
taken at a minimum of two locations in each space and at two different levels
corresponding to the body level and the ankle level, corresponding to approximately
0.1 m and 1.2 m above the floor level, respectively. Instruments used in this study
met the ASHRAE 113-2006 standards‘ requirements for accuracy. The operative
temperature is found to be close to the air temperature. The insulation of clothing
ensembles was determined using the Olsen‘s 1985 summation formula: Icl= ∑ I clu,i
where Icl is the insulation of the entire ensemble and I clu,I represents the effective
insulation of the garment i. The garments values published in the ANSI/ASHRAE
Stand card 55-2004 was the basis for the estimation of clothing ensemble insulation.
The general mean clothing-insulation value of 1.10 Clo was recorded among all the
respondents. The great majority of the respondents were seated on partly or fully
upholstered chairs at the time of survey.
The metabolic rates were determined from the activities filled by the subjects
and as observed at the time of the survey. Uniform value of 115 W/m2 was assumed
for respondents of the residential buildings. This assumption is based on the ISO 7730
Table of metabolic rates for provisions for relaxed seating which was assumed to be
the case with all subjects in their houses.
3.5.3.4. Subjective questionnaire.
The subjective questionnaire consists of the following areas. All the surveys are
―right now‖ surveys. It takes 15 minutes in average for a participant to answer those
questions.
139
3.5.3.5. Indoor climate.
TABLE.3.32 (a) Summary of indoor climatic conditions in the first session for Resident thermal comfort
ROOM Date Sample size Ta(0c) Vair Tmrt Pa Tcl
Alp
ha
, B
eth
al
06-Mar-10 7 18.6 0.13 21 0.9 27.5 07-Mar-10 7 16.7 0.94 22.8 0.34 28.4 08-Mar-10 7 17 0.65 21.8 0.79 28.2 09-Mar-10 7 32 0.53 20.1 0.45 28.8 10-Mar-10 7 17.8 0.24 22.8 0.35 27.7 11-Mar-10 7 18.7 0.65 22.7 0.67 28.4 12-Mar-10 7 19.5 0.79 20.6 0.78 27.9 13-Mar-10 7 32.6 0.45 21.8 0.67 27.3 14-Mar-10 7 27 0.35 21.6 0.23 27.1 15-Mar-10 7 31 0.67 20.6 0.26 28.9
EL
IM,
CA
NN
AN
06-Mar-10 10 28 0.78 21.6 0.57 27.6 07-Mar-10 10 24.6 0.26 22.7 0.39 26 08-Mar-10 10 33.5 0.35 21.6 0.92 27.5 09-Mar-10 10 27.5 0.67 22.5 0.93 27.9 10-Mar-10 10 29.4 0.78 22.6 0.48 27 11-Mar-10 10 16.7 0.26 20.5 0.38 28.5 12-Mar-10 10 16.7 0.26 22.5 0.62 27.1 13-Mar-10 10 17.4 0.57 19.5 0.47 28.9 14-Mar-10 10 16.2 0.39 20.8 0.99 27.1 15-Mar-10 10 18.5 0.17 22.5 0.23 28.9
CA
RM
EL
, K
IDR
ON
06-Mar-10 22 19.5 0.15 20.4 1 27.9 07-Mar-10 22 20.4 0.67 20.6 0.26 27.5 08-Mar-10 22 21.5 0.99 21.8 0.57 27.9 09-Mar-10 22 27.3 0.23 22.7 0.39 28.4 10-Mar-10 22 28.3 0.1 22.3 0.92 26.5 11-Mar-10 22 31.5 0.94 21.6 0.93 27.9 12-Mar-10 22 32.6 0.65 21.9 0.48 27.8 13-Mar-10 22 32.6 0.53 20.6 0.01 27.5 14-Mar-10 22 27.4 0.24 20.5 0.79 27.9 15-Mar-10 22 27.4 0.57 21.5 0.45 26.9
SIN
AI,
TA
BO
R
06-Mar-10 18 28.5 0.39 22.4 0.35 28.9 07-Mar-10 18 29.5 0.13 22.6 0.67 27.6 08-Mar-10 18 30.5 0.13 20 0.78 28.6 09-Mar-10 18 34 1.1 20.6 0.67 30 10-Mar-10 18 28.4 0.99 21.8 0.23 28.9 11-Mar-10 18 28.5 0.23 21.6 0.01 26 12-Mar-10 18 29.5 1 20.6 0.8 28.9 13-Mar-10 18 34 0.47 21.6 0.9 27.9 14-Mar-10 18 18.4 0.26 22.1 0.54 28.9 15-Mar-10 18 19.5 0.39 22.6 0.34 27.9
PA
TR
OB
ER
TS
ON
, 06-Mar-10 23 28.4 0.1 22.4 0.09 28
07-Mar-10 23 19.6 0.26 22.5 0.03 28.5 08-Mar-10 23 20.9 0.26 22.6 1 29 09-Mar-10 23 23.5 0.57 22.8 0.28 27.1 10-Mar-10 23 27.3 0.39 22.9 0.74 28.9 11-Mar-10 23 28.3 0.92 21.6 0.25 27.5 12-Mar-10 23 18.6 0.93 21.9 0.02 27.9 13-Mar-10 23 19.5 0.48 20.6 0.3 27.5 14-Mar-10 23 16 0.26 20.5 0.03 27.9 15-Mar-10 23 21.5 0.26 20.7 0.9 28
HE
BR
ON
,
FR
AN
KIN
SO
N
06-Mar-10 22 33.7 0.57 23 0.38 28.5 07-Mar-10 22 27.3 1.1 21.5 0.62 27.1 08-Mar-10 22 28.3 0.65 22.6 0.47 27.9 09-Mar-10 22 23.5 0.45 23 0.99 27.5 10-Mar-10 22 28.4 0.37 22.4 0.23 27.5 11-Mar-10 22 28.5 0.37 20.4 1 27.9 12-Mar-10 22 33.5 0.47 20.6 0.99 27.6 13-Mar-10 22 31.4 0.26 21.8 0.97 27.8 14-Mar-10 22 29.8 0.39 21.6 0.13 28.5 15-Mar-10 22 17.5 0.1 20.6 0.26 27.1
MEAN 25.23667 0.486833 21.64 0.5365 27.9
MAX 34 1.1 23 1 30
MIN 16 0.1 19.5 0.01 26
AVERAGE 5.827869 0.282102 0.932501647 0.310051 0.76
140
Tabel3.32 (b) Summary of indoor climatic conditions in the Second session for Resident thermal comfort
ROOM Date
Sample
size Ta(0c) Vair Tmrt Pa Tcl
Alp
ha
, B
eth
al
01-Sep-10 7 28.4 0.45 22.1 0.23 27.5 02-Sep-10 7 19.6 0.35 22.4 1 28.4 03-Sep-10 7 20.9 0.67 20.4 0.99 28.2 04-Sep-10 7 23.5 0.78 20.6 0.97 28.8 05-Sep-10 7 17.8 0.67 21.8 0.13 27.7 06-Sep-10 7 18.7 0.23 21.6 0.26 26.9 07-Sep-10 7 19.5 0.4 20.6 0.57 27.5 08-Sep-10 7 32.6 0.8 21.8 0.39 29.5 09-Sep-10 7 28.4 0.9 21.6 0.92 27.1 10-Sep-10 7 19.6 0.54 20.6 0.26 28.9
EL
IM,
CA
NN
AN
01-Sep-10 10 20.9 0.34 21.6 0.57 27.6 02-Sep-10 10 23.5 0.27 22.7 0.39 28.6 03-Sep-10 10 33.5 0.34 21.6 0.92 27.5 04-Sep-10 10 27.5 0.67 22.5 0.93 27.9 05-Sep-10 10 29.4 0.78 22.6 0.48 27 06-Sep-10 10 16.7 0.67 22.3 0.38 28.5 07-Sep-10 10 16.7 0.23 22.5 0.62 27.1 08-Sep-10 10 17.4 0.45 21.4 0.47 28.9 09-Sep-10 10 16.2 0.34 22.4 0.99 30 10-Sep-10 10 18.5 0.35 22.5 0.23 26.9
CA
RM
EL
, K
IDR
ON
01-Sep-10 22 19.5 0.24 21.3 1 27.9 02-Sep-10 22 20.4 0.67 20.6 0.26 27.5 03-Sep-10 22 21.5 0.99 21.8 0.57 29.9 04-Sep-10 22 27.3 0.23 22.5 0.39 28.4 05-Sep-10 22 28.3 0.23 23.0 0.92 27.5 06-Sep-10 22 31.5 0.10 21.6 0.93 26.7 07-Sep-10 22 32.6 0.34 21.9 0.48 27.8 08-Sep-10 22 32.6 0.45 20.6 0.56 26.3 09-Sep-10 22 27.4 0.35 20.5 0.79 29.5 10-Sep-10 22 27.4 0.67 21.0.0 0.45 27.1
SIN
AI,
TA
BO
R
01-Sep-10 18 28.5 0.78 22.8 0.35 28.9 02-Sep-10 18 29.5 0.67 22.4 0.67 26 03-Sep-10 18 30.5 0.23 22.4 0.78 28.6 04-Sep-10 18 34 0.4 22 0.67 28.9 05-Sep-10 18 28.4 0.99 19.5 0.23 28.9 06-Sep-10 18 28.5 0.23 20.6 0.4 27.1 07-Sep-10 18 29.5 1.0 21.8 0.8 28.9 08-Sep-10 18 34 0.28 22.5 0.9 27.9 09-Sep-10 18 18.4 0.74 21.4 0.54 28.9 10-Sep-10 18 19.5 0.25 22.4 0.34 27.9
PA
TR
OB
ER
TS
ON
, 01-Sep-10 23 28.4 0.84 22.5 0.09 28
02-Sep-10 23 19.6 0.26 21.3 0.03 28.5 03-Sep-10 23 28.4 0.23 22.1 1 29 04-Sep-10 23 19.6 1.0 21.3 0.26 27.1 05-Sep-10 23 20.9 0.99 21.5 0.38 28.9 06-Sep-10 23 23.5 0.97 21.6 0.01 27.5 07-Sep-10 23 18.6 0.13 21.9 0.47 27.9 08-Sep-10 23 19.5 0.26 20.6 0.99 26 09-Sep-10 23 16 0.57 20.5 0.23 27.9 10-Sep-10 23 21.5 0.39 20.7 0.97 28
HE
BR
ON
,
FR
AN
KIN
SO
N
01-Sep-10 22 33.7 0.92 20.3 0.13 28.5 02-Sep-10 22 27.3 0.73 21.5 0.26 27.1 03-Sep-10 22 28.3 0.65 22.6 0.47 27.9 04-Sep-10 22 23.5 0.45 22.1 0.99 27.5 05-Sep-10 22 28.4 0.37 22.4 0.23 27.5 06-Sep-10 22 28.5 0.37 20.4 1 27.9 07-Sep-10 22 33.5 1.1 20.6 0.99 27.6 08-Sep-10 22 31.4 0.26 21.8 0.97 27.8 09-Sep-10 22 28.4 0.1 21.6 0.13 28.5 10-Sep-10 22 19.6 0.23 20.6 0.26 27.1
MEAN 24.95333 0.514833 21.60167 0.559833 27.955
MAX 34 1.1 23 1 30 MIN 16 0.1 19.5 0.01 26
AVERAGE 5.585437 0.279136 0.81334 0.31706 0.887087
141
The following values are taken from the data collected from questionnaire and
measurements for further optimization, using different non-traditional algorithms. The
minimum and maximum values of each of these parameters were taken as the lower
and the upper limits of the parameters. These values were taken from both the sets put
together which were taken in March and September so as to take a generalized thermal
comfort of the university buildings. These values are used in the optimization
techniques to optimize the final value and also to find the optimum value of the PMV
Table.3.33. Range of values
Fcl Ta Tmrt Vair Pa Tcl M(met) Icl(clo)
Min 0 16 19 0.1 0.01 26 115 1.1
Max 1.5 34 23 1.1 1 30 115 1.1
In an attempt to reduce the processing time and improve the quality of solution,
particularly to avoid being trapped in local minima, the non- traditional optimization
is used. In this problem to find the optimum thermal comfort, ten non- traditional
optimization techniques are used. Each one has its own characteristics. Twenty trial
runs were performed for the problem in each of the ten methods. The performance of
the different algorithms was compared. The characteristics led to different solutions
and run times. The results are finally examined based on different criteria.
Each algorithm with its own option set and stopping criteria was used. All the
non-traditional optimization was run using MATLab2011 to get the global optimum
value for each of the parameter and also the final value of the thermal comfort.
Hence the Problem is
To minimize PMV for office with the regression coefficients is:
142
Subject to the following constraints (bounds)
0 ≤ Fcl ≤ 1.5;
16 ≤ Ta ≤ 34 ;
19.5 ≤ Tmrt ≤ 23;
0.1 ≤ Vair ≤ 1.1;
0.01 ≤ Pa ≤ 1;
26 ≤ Tcl ≤ 30;
M = 75;
Icl = 1.5
143
3.5.4. Algorithms
3.5.4.1. Genetic algorithm
3.5.4.1.1. Stopping criteria:
The options set for the resident problem is same as that of the office problem
using genetic algorithm. The stopping criteria reached is same as in the office
problem. It is therefore the ―Minimum difference between two successive values of
objective function value is less than 10-6
‖ condition brings the stop of the iteration.
Hence we say that the solution is naturally converged to the global optimum point.
Table.3.34. Results of GA in 20 trails for Residence thermal comfort
Trails Fcl Ta Tmrt Vair Pa Tcl PMV PPD TIME Iters
1 0.686 19.4944 20.852
8
0.9459 0.5154 27 -0.5 5 0.5510
4
51
2 0.545
6
21.0165 20.000
2
0.9705 0.6392 29
.7
-0.5 5 0.2475
23
51
3 0.748
9
17.5037 21.746
7
0.6219 0.624 26
.4
-0.5 5 0.2527
98
51
4 0.856 16.9313 20.500
1
0.1002 0.0203 28
.8
-0.5 5 0.2599
75
51
5 0.473
7
16.124 19.625 0.592 0.1427 29 -0.5 5 0.2566
67
51
6 0.867
5
18.5659 22.300
7
0.4528 0.0923 26
.9
-0.5 5 0.2525
93
51
7 0.364
4
16.0234 19.630
8
1.0346 0.4752 29
.7
-0.5 5 0.2506
31
51
8 0.773
1
21.4679 22.526
1
0.7011 0.0748 28
.6
-0.5 5 0.2533
23
51
9 0.897
6
22.0884 20.685
8
0.5245 0.5398 28
.4
-0.5 5 0.2548
19
51
10 0.784
4
16.9396 20.482
7
0.2554 0.7357 28
.1
-0.5 5 0.2578
02
51
11 0.475
5
16.0002 21.051
2
0.8308 0.0666
6
27
.7
-0.5 5 0.2635
47
51
12 0.812
2
21.6438 19.874 0.2834 0.4901 30 -0.5 5 0.2509
9
51
13 1.073
5
16.0078 19.506
3
0.1002 0.0123 26
.1
-0.5 5 0.2683
59
51
14 0.900
7
23.9702 21.396
1
0.719 0.166 29
.1
-0.5 5 0.2599
1
51
15 1.001
8
16.0302 19.531
5
0.1587 0.0609 26 -0.5 5 0.2540
22
51
16 0.799 19.5761 19.633
9
0.471 0.5052 27
.4
-0.5 5 0.2515
63
51
17 0.991
1
19.4237 22.456
3
0.3849 0.9576 27
.4
-0.5 5 0.2589
12
51
18 0.908
5
17.7994 20.859 0.4229 0.6014 26 -0.5 5 0.2533
86
51
19 0.507
8
18.0889 21.322
2
0.9926 0.6103 28
.4
-0.5 5 0.2511
4
51
20 0.505
1
16.0001 19.501 0.3986 0.1023 29
.8
-0.5 5 0.2597
45
51
AVG 0.748
6
18.53478 20.674
1
0.5480
5
0.3716
0
28 -0.5 5 0.2704
37
51
144
Fig.3.23.Convergence of Genetic Algorithm
3.5.4.2. Simulated annealing
3.5.4.2.1. Convergence criteria met:
The stopping criteria and options set are same as that of the options for the
simulated Annealing for office problem. This case also reaches the final value with the
criteria ―Minimum difference between two successive values of objective function
value is less than 10-6
‖ stops the iteration. Hence we can say that the solution is global
optimum.
0 10 20 30 40 50 60 70 80 90 1005
5.05
5.1
5.15
5.2
5.25
5.3
5.35
5.4
5.45
5.5
Generation
Fitness v
alu
e
Best: 5 Mean: 5.0145
Best f itness
Mean fitness
145
Table.3.35. Results of SA in 20 trails for Residence thermal comfort
Trail
s
Fcl Ta Tmrt Vair Pa Tcl PM
V
PPD TIME Iters
1 0.754
5
16.7613 21.976
2
0.1787 0.070
7
29.
8
-0.5 5 3.11930
8
3000
2 0.943
3
17.03333 22.602
1
0.2307 0.219 27.
8
-0.5 5 6.31293
6
3000
3 1.253
3
22.7174 19.884
7
0.1307 0.139
9
28.
5
-0.5 5 3.97143
7
3000
4 0.685
7
17.2693 20.683
1
0.6501 0.131
2
26.
2
-0.5 5 3.24766
8
3000
5 1.064
5
19.1286 22.428
7
0.3345 0.236
8
26.
8
-0.5 5 3.92258
4
3000
6 1.363
7
21.5869 22.845
8
0.3663 0.950
3
27.
2
-0.5 5 4.56859
2
3000
7 0.811
4
23.3274 22.851
9
1.0785 0.926
8
29.
1
-0.5 5 3.90544
4
3000
8 0.666
4
17.3897 22.435
9
0.8208 0.145
7
26.
3
-0.5 5 3.31735 3000
9 1.248
4
25.5412 20.243
1
0.919 0.94 28.
2
-0.5 5 3.12108
5
3000
10 0.579
7
17.7054 22.877
2
0.6501 0.666
8
28.
9
-0.5 5 2.81164 3000
11 0.789
9
21.6412 21.474
7
1.0877 0.816
3
27.
6
-0.5 5 2.96249
2
3000
12 0.582
7
16.9418 22.565
7
0.5257 0.349
1
28.
9
-0.5 5 4.06173
1
3000
13 0.852 16.8994 21.102
1
0.1421 0.812
7
29.
8
-0.5 5 3.37426
1
3000
14 0.714
2
19.2517 21.618
7
0.4084 0.208
7
29 -0.5 5 3.39039
1
3000
15 0.417
5
16.182 20.374
5
1.0039 0.288
8
28.
4
-0.5 5 4.72661
1
3000
16 0.719 16.7953 21.368
7
0.5555 0.982 26.
6
-0.5 5 3.42990
3
3000
17 1.145
8
25.9543 19.545
2
1.0632 0.469
9
28.
5
-0.5 5 3.48569
7
3000
18 1.267
8
18.0648 22.915
5
0.8181 0.765 26.
8
-0.5 5 3.37842
1
3000
19 1.476
6
27.4851 19.996 0.1605 0.508
6
29.
9
-0.5 5 2.67733
7
3000
20 0.745
4
21.9933 22.626
4
0.8452 0.653
1
29.
1
-0.5 5 3.07847 3000
AVG 0.904
09
19.98347 21.620
81
0.5984
85
0.514
07
28.
2
-0.5 5 3.64316
79
3000
Fig.3.24.Convergence of SA:
0 500 1000 1500 2000 2500 30004
4.5
5
5.5
6
Iteration
Function v
alu
e
Best Function Value: 5
1 2 3 4 5 60
5
10
15
20
25
30Best point
Number of variables (6)
Best
poin
t
0 10 20 30 40 50 60 70 80 90 100
Time
Iteration
f-count
% of criteria met
Stopping Criteria
0 500 1000 1500 2000 2500 30005
6
7
8
9
10
Iteration
Function v
alu
e
Current Function Value: 5
146
3.5.4.3. Pattern search
3.5.4.3.1. Stopping Criteria Reached:
The solution is reached by the stopping condition, ―difference in function value less
than 10-6
‖ and also comparatively the iterations are of less in number, this indicates quick
convergence. The final value of the solution is naturally obtained.
Table.3.36. Results of PS in 20 trails for Residence thermal comfort.
Trails Fcl Ta Tmrt Vair Pa Tcl PMV PPD TIME Iters
1 1.25 25 21.5 0.875 0.0853 28 -0.5 5 0.293276 20
2 1.25 25 21.5 0.875 0.0853 28 -0.5 5 0.115934 20
3 1.25 25 21.5 0.875 0.0853 28 -0.5 5 0.128355 20
4 1.25 25 21.5 0.875 0.0853 28 -0.5 5 0.109838 20
5 1.25 25 21.5 0.875 0.0853 28 -0.5 5 0.109052 20
6 1.25 25 21.5 0.875 0.0853 28 -0.5 5 0.118687 20
7 1.25 25 21.5 0.875 0.0853 28 -0.5 5 0.116927 20
8 1.25 25 21.5 0.875 0.0853 28 -0.5 5 0.112212 20
9 1.25 25 21.5 0.875 0.0853 28 -0.5 5 0.105636 20
10 1.25 25 21.5 0.875 0.0853 28 -0.5 5 0.108933 20
11 1.25 25 21.5 0.875 0.0853 28 -0.5 5 0.113323 20
12 1.25 25 21.5 0.875 0.0853 28 -0.5 5 0.108616 20
13 1.25 25 21.5 0.875 0.0853 28 -0.5 5 0.112946 20
14 1.25 25 21.5 0.875 0.0853 28 -0.5 5 0.109518 20
15 1.25 25 21.5 0.875 0.0853 28 -0.5 5 0.122536 20
16 1.25 25 21.5 0.875 0.0853 28 -0.5 5 0.113252 20
17 1.25 25 21.5 0.875 0.0853 28 -0.5 5 0.109042 20
18 1.25 25 21.5 0.875 0.0853 28 -0.5 5 0.11886 20
19 1.25 25 21.5 0.875 0.0853 28 -0.5 5 0.120268 20
20 1.25 25 21.5 0.875 0.0853 28 -0.5 5 0.112034 20
AVG 1.25 25 21.5 0.875 0.0853 28 -0.5 5 0.122962
2
20
147
Fig.3.25.Convergence of Pattern Search
3.5.4.4. Particle swarm optimization
3.5.4.4.1. Stopping Criteria Reached:
The options and the stopping criteria which are set are same as that for PSO in the
IEQ Office Buildings problem. This case also the final solution reaches by the stopping
condition,‖ the change in the final value of the system is less than 10-6
‖ but the specialty is
the elapsed time which is less than other solvers. The global optimum solution is obtained
without any other stopping conditions.
0 2 4 6 8 10 12 14 16 18 204
4.5
5
5.5
6
Iteration
Functio
n valu
e
Best Function Value: 5
0 2 4 6 8 10 12 14 16 18 200
0.2
0.4
0.6
0.8
1
Iteration
Mesh siz
e
Current Mesh Size: 9.5367e-007
148
Table.3.37. Results of PSO in 20 trails for Residence thermal comfort
Trails Fcl Ta Tmrt Vair Pa Tcl PMV PPD TIME Iters
1 1.135
1
23.373
8
20.3311 0.410
9
0.23
45
28 -0.5 5 0.0830
84
51
2 1.215
5
27.201
7
20.0871 0.884
6
0.22
29
29.4 -0.5 5 0.0889
01
51
3 1.463
2
21.180
7
22.7409 0.329
3
0.64
41
26.6 -0.5 5 0.9933
23
51
4 1.273
7
21.878
9
21.3342 0.283
8
0.57
53
27.5 -0.5 5 0.0843
07
51
5 0.821 19.811 21.3711 0.230
5
0.40
21
29.8 -0.5 5 0.0937
17
51
6 0.645
2
17.441
9
22.2741 0.824
1
0.58
57
26.7 -0.5 5 0.0890
55
51
7 1.05 20.816 21.5616 0.323
7
0.11
24
27.6 -0.5 5 0.0904
19
51
8 1.257
3
25.339
1
21.5031 0.680
9
0.36
77
28.6 -0.5 5 0.0815
06
51
9 1.102
3
20.8411 22.2275 0.146
5
0.73
13
29.6 -0.5 5 0.0857
74
51
10 1.087
8
20.202
3
22.9389 0.364
8
0.33
86
27.3 -0.5 5 0.0867
11
51
11 1.258
9
20.251 20.8716 0.206
7
0.68
54
27 -0.5 5 0.0871
11
51
12 1.22 23.675 20.9342 0.481
6
0.60
79
27.9 -0.5 5 0.0789
93
51
13 0.865
4
20.202
8
21.8696 0.523
7
0.60
09
27.7 -0.5 5 0.0835
54
51
14 0.929
3
23.092
6
20.3715 0.623
6
0.56
36
28.4 -0.5 5 0.0909
56
51
15 0.656 21.150
7
20.7839 0.701
6
0.27
74
29.2 -0.5 5 0.0966
21
51
16 0.912
4
21.922
3
21.7725 0.614
6
0.80
25
28.2 -0.5 5 0.0844
75
51
17 1.215 23.384
5
21.7358 0.178
7
0.64
05
29.6 -0.5 5 0.0872
63
51
18 1.071
5
23.889
9
20.0722 0.253
7
0.68
76
29.5 -0.5 5 0.0847
49
51
19 1.191
2
21.378 22.2986 0.763
2
0.65
48
26.1 -0.5 5 0.0799
59
51
20 1.211
1
26.046
7
20.9604 0.883
7
0.05
96
28.7 -0.5 5 0.0836
63
51
AVG 1.079
1
22.154 21.402 0.485
5
0.48
97
28.2 -0.5 5 0.1317
07
51
3.5.4.5. GODLIKE
3.5.4.5.1. Stopping Criteria Reached:
The options and the stopping criteria which are set are same as that for GODLIKE in
the IEQ Office Buildings problem. This case also the final solution reaches by the stopping
condition,‖ the change in the final value of the system is less than 10-6
‖.The solver exchanges
the population among the solvers hence the iteration indicates number of times the population
is exchanged. The global optimum solution is obtained without any other stopping conditions.
149
Table.3.38. Results of GL in 20 trails for Residence thermal comfort
Trails Fcl Ta Tmrt Vair Pa Tc PMV PPD TIME Iters
1 1.1765 24.743
9
22.098
1
0.953
4
0.335
7
28.2 -0.5 5 2.0253
47
4
2 0.0931
4
20.158
1
21.183
4
0.453
5
0.451
7
27.3 -0.5 5 2.8866
51
4
3 0.6588 19.016 21.599
6
0.561
4
0.744
1
28.8 -0.5 5 2.4576
29
4
4 0.9286 22.162
3
21.734
3
0.634
6
0.539
8
28.1 -0.5 5 2.9460
23
4
5 0.9327 21.614
3
20.915
2
0.484
3
0.602
4
28.1 -0.5 5 2.5760
18
4
6 1.4841 25.448
4
21.483
5
0.193
3
0.446
8
29.2 -0.5 5 2.0798
02
4
7 0.9426 23.226
8
19.568
2
0.971
5
0.572
9
27.5 -0.5 5 2.8002
92
4
8 0.8473 24.468
8
20.402
1
0.892
1
0.927
3
29.5 -0.5 5 2.2756
23
4
9 0.5998 20.406
3
20.968
4
0.947
8
0.311
5
28.6 -0.5 5 2.4938
92
4
10 0.7063 19.2911 22.095
3
0.771 0.222
6
27.4 -0.5 5 2.6048
88
4
11 0.7225 19.489
4
21.137
7
1.054 0.774
2
26.5 -0.5 5 2.0752
49
4
12 0.8519 25.040
8
20.867 0.856
3
0.413
5
29.9 -0.5 5 2.4398
61
4
13 0.9494 23.454
7
21.108
7
0.526
8
0.755
1
29.2 -0.5 5 2.7876
42
4
14 1.004 23.994
9
21.184
7
0.568
5
0.754
9
29.1 -0.5 5 2.3716
22
4
15 1.3973 22.681
5
21.626 0.422
7
0.234
8
26.9 -0.5 5 2.1846
04
4
16 0.7105 17.875
7
20.698
3
0.814
9
0.851 26.1 -0.5 5 1.9704
65
4
17 1.1699 19.942
7
20.901
7
0.293
3
0.853
8
26.7 -0.5 5 2.1044
46
4
18 1.3068 19.945
9
21.978
9
0.206
1
0.509
8
27 -0.5 5 2.4025
94
4
19 1.2986 25.142
8
21.187
7
0.484 0.958
5
28.8 -0.5 5 1.9012
57
4
20 0.9073 17.740
7
22.414 0.468
9
0.823
7
26.3 -0.5 5 1.6231
68
4
AVG 0.9344
0
21.792
3
21.257
6
0.627
9
0.604
2
27.9 -0.5 5 2.3503
54
4
3.5.4.6. Fmincon.
3.5.4.6.1. Stopping Criteria Reached:
The options and the stopping criteria which are set are same as that for
Fmincon in the IEQ Office Buildings problem. This case also the final solution reaches by the
stopping condition,‖ the change in the final value of the system is less than 10-6
‖. The global
optimum solution is obtained without any other stopping conditions. The exception is that the
elapsed time is high comparatively; this is due to the traditional technique modified version
of using Lagrange‘s multipliers.
150
Table.3.39. Results of Fmincon in 20 trails for Residence thermal comfort
Trails Fcl Ta Tmrt Vair Pa Tcl PMV PPD TIME Iters
1 0.407 16 21.270
3
0.9032 0.291
2
29.
3
-0.5 5 13.7814
65
2338
2 0.879
2
20.877
2
22.877
5
0.4931 0.702 28.
6
-0.5 5 16.1614
32
2296
3 0.778
3
22.599
8
22.54 0.5856 0.378
1
30 -0.5 5 14.6616
01
2275
4 0.830
1
21.397
7
22.527
9
0.952 0.435
4
27.
5
-0.5 5 16.6104
66
2317
5 0.824
1
19.534
6
22.875 0.2386 0.139
9
30 -0.5 5 16.4216
74
2324
6 1.121
8
23.208
5
21.228
9
0.6998 0.732
4
27.
6
-0.5 5 15.0178
31
2310
7 1.486
9
28.5211 21.605
9
0.869 0.197
5
30 -0.5 5 16.7983
57
2296
8 0.989
2
25.4011 22.372
4
0.8043 0.919 30 -0.5 5 16.0812
53
2282
9 1.071
5
25.684
6
19.939
2
0.4713 0.279
5
29.
6
-0.5 5 14.7694
81
2289
10 1.282
6
21.477
6
21.257
1
0.19 0.208
4
27.
7
-0.5 5 16.3270
85
2268
11 0.670
3
21.171
5
20.541
9
0.5342 0.489
6
29.
9
-0.5 5 13.4919
01
2324
12 1.250
6
26.630
4
20.459
4
0.9163 0.872
9
29.
1
-0.5 5 16.5282
76
2387
13 1.488
8
26.588
1
21.203
2
0.5618 0.500
2
28.
9
-0.5 5 10.3041
11
2282
14 0.638
3
22.037
6
21.125
6
0.7943 0.631
8
30 -0.5 5 13.8142
12
2296
15 0.702
4
22.935
5
22.416
2
0.9204 0.901
4
30 -0.5 5 17.2800
53
2331
16 0.721
5
22.753
6
19.837
6
0.6266 0.848
4
29.
9
-0.5 5 16.2284
04
2289
17 1.437
5
20.097
1
21.9811 0.2312 0.611
3
26.
3
-0.5 5 12.5094
51
2324
18 0.971
3
23.615
9
21.584 0.5388 0.258
6
29 -0.5 5 13.1727
22
2289
19 1.273
2
21.610
4
20.191
7
0.5042 0.599
8
26 -0.5 5 14.1621
36
2324
20 1.036
1
22.904
8
20.132
4
0.9235 0.670
5
27 -0.5 5 13.1902
02
2324
AVG 0.993
0
22.752
4
21.398
4
0.6379
1
0.533
4
28.
8
-0.5 5 14.8656
06
2308
3.5.4.7. Direct evolution
3.5.4.7.1. Stopping Criteria:
The options and the stopping criteria which are set are same as that for DE in the IEQ
Office Buildings problem. This case also the final solution reaches by the stopping
condition,‖ the change in the final value of the system is less than 10-6
‖. It is seen from the
results that the final vectors (parameter values) is not consistent, this is because DE uses
different type of cross over method. The global optimum solution is obtained without any
other stopping conditions.
151
Table.3.40. Results of DE in 20 trails for Residence thermal comfort
Trails Fcl Ta Tmrt Vair Pa Tcl PMV PPD TIME Iters
1 0.385
1
16.459
3
20.929 1.0098 0.418
4
27.
9
0.113
1
5 0.5387
96
120000
2 1.229
6
26.238 19.507
7
0.4031 0.727
4
29.
3
0.484
3
5 0.5247
93
120000
3 0.987
4
16.628
4
22.2114 0.1628 0.551
6
26.
7
0.151
1
5 0.8904
05
120000
4 1.390
8
26.841
5
22.092
8
0.6045 0.974
6
29.
5
0.383
4
5 0.5416
12
120000
5 1.014 20.769
3
22.923
5
0.2386 0.974
3
28.
9
0.310
3
5 0.5693 120000
6 1.405
7
26.24 22.436
5
0.4426 0.093 29.
1
0.354
8
5 0.5909
35
120000
7 1.016
9
20.909
8
22.244
8
0.3214 0.211
6
27.
8
0.353
1
5 0.5615
17
120000
8 0.876
7
16.624
5
19.509
6
0.1327 0.681
5
28 0.396
1
5 0.5687
42
120000
9 1.044
5
24.985
6
21.8118 0.6006 0.211
4
29 0.315
2
5 0.5848
03
120000
10 1.275
5
24.260
4
21.451
2
0.8512 0.361
9
27.
5
0.492
3
5 0.5358
97
120000
11 1.447
7
27.105
9
21.774
3
0.3754 0.572
6
29.
9
0.495
7
5 0.5309
87
120000
12 1.257
5
21.700
1
22.760
2
0.4382 0.337
4
26.
8
0.371
2
5 0.5857
32
120000
13 1.317 22.042
4
22.912
2
0.2916 0.230
7
27.
1
0.161
9
5 0.5778
44
120000
14 1.094
6
24.671 20.958
6
0.5538 0.892
7
28.
8
0.377
2
5 0.5349
81
120000
15 1.254
9
23.621
5
21.471
9
0.1199 0.244
4
29.
1
0.254
5
5 0.5320
66
120000
16 0.917 18.008
7
20.201
7
0.2537 0.285
3
26.
8
0.423
7
5 0.5884
87
120000
17 1.129
4
21.004
1
22.192
8
0.6023 0.795
6
26.
1
0.277
6
5 0.5435
63
120000
18 0.799
8
17.772
7
21.251
9
0.4192 0.704
2
27.
1
0.499 5 0.5482
53
120000
19 1.377 27.297
3
19.36 0.9973 0.671
8
28.
9
0.498
1
5 0.5470
68
120000
20 1.286
2
23.799
7
21.6118 0.9562 0.287
9
27 0.482
4
5 0.5532
43
120000
AVG 1.125
4
22.349
0
21.480
7
0.4887
5
0.511
4
28.
1
0.359
8
5 0.5724
51
120000
3.5.4.8. LGO
3.5.4.8.1. Stopping Criteria:
The options and the stopping criteria which are set are same as that for LGO in the
IEQ Office Buildings problem. The global solution reaches by the stopping condition,‖ the
change in the final value of the system did not improve‖. The elapsed time is close to that of
other Direct algorithm solvers but it does not use Lipchitz constant
152
Table.3.41. Results of LGO in 20 trails for Residence thermal comfort
Trails Fcl Ta Tmrt Vair Pa Tcl PMV PPD TIME Iters
1 0.514
4
16.484
6
21.412
7
0.595
2
1 29.
4
-0.5 5 0.9875
21
3851
2 0.514
4
16.484
6
21.412
7
0.595
2
1 29.
4
-0.5 5 0.8963
83
3851
3 0.514
4
16.484
6
21.412
7
0.595
2
1 29.
4
-0.5 5 0.9578
26
3851
4 0.514
4
16.484
6
21.412
7
0.595
2
1 29.
4
-0.5 5 0.8931
58
3851
5 0.514
4
16.484
6
21.412
7
0.595
2
1 29.
4
-0.5 5 1.1479
94
3851
6 0.514
4
16.484
6
21.412
7
0.595
2
1 29.
4
-0.5 5 0.7894
01
3851
7 0.514
4
16.484
6
21.412
7
0.595
2
1 29.
4
-0.5 5 0.8724
25
3851
8 0.514
4
16.484
6
21.412
7
0.595
2
1 29.
4
-0.5 5 0.9022
55
3851
9 0.514
4
16.484
6
21.412
7
0.595
2
1 29.
4
-0.5 5 0.9150
21
3851
10 0.514
4
16.484
6
21.412
7
0.595
2
1 29.
4
-0.5 5 0.9297
07
3851
11 0.514
4
16.484
6
21.412
7
0.595
2
1 29.
4
-0.5 5 0.9829
16
3851
12 0.514
4
16.484
6
21.412
7
0.595
2
1 29.
4
-0.5 5 1.1274
79
3851
13 0.514
4
16.484
6
21.412
7
0.595
2
1 29.
4
-0.5 5 0.9485
22
3851
14 0.514
4
16.484
6
21.412
7
0.595
2
1 29.
4
-0.5 5 0.8292
22
3851
15 0.514
4
16.484
6
21.412
7
0.595
2
1 29.
4
-0.5 5 0.8468
4
3851
16 0.514
4
16.484
6
21.412
7
0.595
2
1 29.
4
-0.5 5 0.9992
31
3851
17 0.514
4
16.484
6
21.412
7
0.595
2
1 29.
4
-0.5 5 0.8913
72
3851
18 0.514
4
16.484
6
21.412
7
0.595
2
1 29.
4
-0.5 5 1.1169
48
3851
19 0.514
4
16.484
6
21.412
7
0.595
2
1 29.
4
-0.5 5 1.0731
88
3851
20 0.514
4
16.484
6
21.412
7
0.595
2
1 29.
4
-0.5 5 0.9633
35
3279
AVG 0.514
4
16.484
6
21.412
7
0.595
2
1 29.
4
-0.5 5 0.9535
372
3822.
4
3.5.4.9 glcCluster.
3.5.4.9.1. Stopping Criteria:
The default options are taken from the solver from the previous run of the IEQ Office
Buildings problem. The global solution reaches by the stopping condition,‖ the change in the
final value of the system is less than 10-7
‖.Though glcCluster uses Clustering algorithm in
addition it has very less elapsed time.
153
Table.3.42. Results of glcCluster in 20 trails for Residence thermal comfort
TRI
AL
Fcl Ta Tmrt Vair Pa T
cl
PM
V
PP
D
TIME ITER
1 1.250
5
24.999
3
20.083 0.6009 0.064
9
2
8
-0.5 5 41.073852 6499
2 1.250
5
24.999
3
20.083 0.6009 0.064
9
2
8
-0.5 5 42.066377 6499
3 1.250
5
24.999
3
20.083 0.6009 0.064
9
2
8
-0.5 5 36.064554 6499
4 1.250
5
24.999
3
20.083 0.6009 0.064
9
2
8
-0.5 5 40.06594 6499
5 1.250
5
24.999
3
20.083 0.6009 0.064
9
2
8
-0.5 5 40.064215 6499
6 1.250
5
24.999
3
20.083 0.6009 0.064
9
2
8
-0.5 5 39.066121 6499
7 1.250
5
24.999
3
20.083 0.6009 0.064
9
2
8
-0.5 5 39.066596 6499
8 1.250
5
24.999
3
20.083 0.6009 0.064
9
2
8
-0.5 5 41.064278 6499
9 1.250
5
24.999
3
20.083 0.6009 0.064
9
2
8
-0.5 5 39.067789 6499
10 1.250
5
24.999
3
20.083 0.6009 0.064
9
2
8
-0.5 5 39.068794 6499
11 1.250
5
24.999
3
20.083 0.6009 0.064
9
2
8
-0.5 5 38.065819 6499
12 1.250
5
24.999
3
20.083 0.6009 0.064
9
2
8
-0.5 5 39.064496 6499
13 1.250
5
24.999
3
20.083 0.6009 0.064
9
2
8
-0.5 5 40.064734 6499
14 1.250
5
24.999
3
20.083 0.6009 0.064
9
2
8
-0.5 5 41.064495 6499
15 1.250
5
24.999
3
20.083 0.6009 0.064
9
2
8
-0.5 5 41.065159 6499
16 1.250
5
24.999
3
20.083 0.6009 0.064
9
2
8
-0.5 5 40.066629 6499
17 1.250
5
24.999
3
20.083 0.6009 0.064
9
2
8
-0.5 5 39.063954 6499
18 1.250
5
24.999
3
20.083 0.6009 0.064
9
2
8
-0.5 5 38.065024 6499
19 1.250
5
24.999
3
20.083 0.6009 0.064
9
2
8
-0.5 5 37.067345 6499
20 1.250
5
24.999
3
20.083 0.6009 0.064
9
2
8
-0.5 5 36.483809 6499
AVG 1.250
5
24.999
3
20.083 0.6009 0.064
9
2
8
-0.5 5 39.336999 6499
3.5.4.10. glcSolve
3.5.4.10.1. Stopping Criteria:
The options and the stopping criteria are taken from the previous run of IEQ Office
Building problem. The final solution reaches by the stopping condition,‖ the change in the
final value of the system is less than 10-6
‖. glcSolve uses one of the complex algorithm and
even after giving long range values for parameters (which is not recommended) it takes little
time to complete optimization.
154
Table.3.43. Results of glcSolve in 20 trails for Residence thermal comfort
Trail
s
Fcl Ta Tmrt Vair Pa Tcl PMV PPD TIME funeval ITER
S 1 1.2
5
25 20.081
7
0.6 0.05
1
28 -0.5 5 10.08039
4
15777 34
2 1.2
5
25 20.081
7
0.6 0.05
1
28 -0.5 5 9.927723 15777 34
3 1.2
5
25 20.081
7
0.6 0.05
1
28 -0.5 5 10.85973
9
15777 34
4 1.2
5
25 20.081
7
0.6 0.05
1
28 -0.5 5 11.44608
1
15777 34
5 1.2
5
25 20.081
7
0.6 0.05
1
28 -0.5 5 10.88563
6
15777 34
6 1.2
5
25 20.081
7
0.6 0.05
1
28 -0.5 5 9.828382 15777 34
7 1.2
5
25 20.081
7
0.6 0.05
1
28 -0.5 5 9.894192 15777 34
8 1.2
5
25 20.081
7
0.6 0.05
1
28 -0.5 5 10.66552
3
15777 34
9 1.2
5
25 20.081
7
0.6 0.05
1
28 -0.5 5 10.58143
6
15777 34
10 1.2
5
25 20.081
7
0.6 0.05
1
28 -0.5 5 11.31167
8
15777 34
11 1.2
5
25 20.081
7
0.6 0.05
1
28 -0.5 5 11.04851
8
15777 34
12 1.2
5
25 20.081
7
0.6 0.05
1
28 -0.5 5 10.39146
9
15777 34
13 1.2
5
25 20.081
7
0.6 0.05
1
28 -0.5 5 11.78036
2
15777 34
14 1.2
5
25 20.081
7
0.6 0.05
1
28 -0.5 5 11.52661
3
15777 34
15 1.2
5
25 20.081
7
0.6 0.05
1
28 -0.5 5 10.10532
1
15777 34
16 1.2
5
25 20.081
7
0.6 0.05
1
28 -0.5 5 11.00906
4
15777 34
17 1.2
5
25 20.081
7
0.6 0.05
1
28 -0.5 5 8.926088 15777 34
18 1.2
5
25 20.081
7
0.6 0.05
1
28 -0.5 5 9.97892 15777 34
19 1.2
5
25 20.081
7
0.6 0.05
1
28 -0.5 5 11.67263 15777 34
20 1.2
5
25 20.081
7
0.6 0.05
1
28 -0.5 5 12.09721
9
15777 34
AVG 1.2
5
25 20.081
7
0.6 0.05
1
28 -0.5 5 10.70084
94
15777 34
Table.3.44. Comparative results of optimization methods for resident thermal comfort
PMV PPD RESIDENCE
Methods Fcl Ta Tmrt Vair Pa Tcl PMV PPD TIME ITER
Genetic
algorithm
0.748
62
18.534
78
20.674
12
0.5480
5
0.3716
08
28 -0.5 5 0.2704
3725
51
Simulated
annealing
0.904
09
19.983
47
21.620
81
0.5984
85
0.5140
7
28.
2
-0.5 5 3.6431
679
3000
Pattern
search
1.25 25 21.5 0.875 0.0853 28 -0.5 5 0.1229
6225
20
PSO 1.079
095
22.154 21.402 0.4855
1
0.4897
4
28.
2
-0.5 5 0.1317
0705
51
G-L 0.934
402
21.792
26
21.257
64
0.6279
2
0.6042
05
27.
9
-0.5 5 2.3503
5365
4
fmincon 1.161
08
23.830
97
20.484
77
0.6089
98
0.1716
62
28 -0.5 5 7.6471
4423
1072
6 DE
optimizati
on
SOLUTI
ON
1.125
365
22.349
01
21.480
69
0.4887
45
0.5114
18
28.
1
0.359
75
5 0.5724
512
1200
00 LGO 0.514
4
16.484
6
21.412
7
0.5952 1 29.
4
-0.5 5 0.9535
372
3822
glcClus 1.250
5
24.999
3
20.083 0.6009 0.0649 28 -0.5 5 39.336
999
6499
glcSolve 1.25 25 20.081
7
0.6 0.051 28 -0.5 5 10.700
8494
1577
7
155
FIG.3.26. Comparative results of optimization methods for resident thermal comfort
From the above graph, we can observe that the PMV and PPD values are the same for
the all the ten optimization techniques as -0.5 and 5 except for DE which is 0.359 for PMV.
The elapsed time is maximum for glcCluster and minimum for Pattern search and particle
swarm optimization. All the other parameters, more or less, have the same values for all the
ten optimization techniques. Now individual parameters are taken into account separately to
find out which optimization method yields the best result.
156
3.5.5. PARAMETERS
3.5.5.1. Ratio of body’s surface area when fully clothed to body’s surface area
when nude-Fcl:
The heat produced must be dissipated to the environment, or a change in body
temperature will occur. The deep body temperature is about 37°C, whilst the skin
temperature can vary between 31°C and 34°C under comfort conditions. Variations
occur in time, but also between parts of the body, depending on clothing cover and
blood circulation. There is a continuous transport of heat from deep tissues to the skin
surface, from where it is dissipated by radiation, convection or (possibly) conduction
and evaporation.
TABLE.3.45. Fcl results in all 10 methods
Tria
ls
GENETI
C
ALGOR
ITHM
SA PS PSO G-L fminc
on
DE LG
O
glcClu
ster
glcSol
ve 1 0.686 0.754
5
1.25 1.1351 1.1765 0.407 0.3851 0.51
44
1.2505 1.25
2 0.5456 0.943
3
1.25 1.2155 0.0931
4
0.8792 1.2296 0.51
44
1.2505 1.25
3 0.7489 1.253
3
1.25 1.4632 0.6588 0.7783 0.9874 0.51
44
1.2505 1.25
4 0.856 0.685
7
1.25 1.2737 0.9286 0.8301 1.3908 0.51
44
1.2505 1.25
5 0.4737 1.064
5
1.25 0.821 0.9327 0.8241 1.014 0.51
44
1.2505 1.25
6 0.8675 1.363
7
1.25 0.6452 1.4841 1.1218 1.4057 0.51
44
1.2505 1.25
7 0.3644 0.811
4
1.25 1.05 0.9426 1.4869 1.0169 0.51
44
1.2505 1.25
8 0.7731 0.666
4
1.25 1.2573 0.8473 0.9892 0.8767 0.51
44
1.2505 1.25
9 0.8976 1.248
4
1.25 1.1023 0.5998 1.0715 1.0445 0.51
44
1.2505 1.25
10 0.7844 0.579
7
1.25 1.0878 0.7063 1.2826 1.2755 0.51
44
1.2505 1.25
11 0.4755 0.789
9
1.25 1.2589 0.7225 0.6703 1.4477 0.51
44
1.2505 1.25
12 0.8122 0.582
7
1.25 1.22 0.8519 1.2506 1.2575 0.51
44
1.2505 1.25
13 1.0735 0.852 1.25 0.8654 0.9494 1.4888 1.317 0.51
44
1.2505 1.25
14 0.9007 0.714
2
1.25 0.9293 1.004 0.6383 1.0946 0.51
44
1.2505 1.25
15 1.0018 0.417
5
1.25 0.656 1.3973 0.7024 1.2549 0.51
44
1.2505 1.25
16 0.799 0.719 1.25 0.9124 0.7105 0.7215 0.917 0.51
44
1.2505 1.25
17 0.9911 1.145
8
1.25 1.215 1.1699 1.4375 1.1294 0.51
44
1.2505 1.25
18 0.9085 1.267
8
1.25 1.0715 1.3068 0.9713 0.7998 0.51
44
1.2505 1.25
19 0.5078 1.476
6
1.25 1.1912 1.2986 1.2732 1.377 0.51
44
1.2505 1.25
20 0.5051 0.745
4
1.25 1.2111 0.9073 1.0361 1.2862 0.51
44
1.2505 1.25
avg 0.74862 0.904
09
1.25 1.0790
95
0.9344
02
0.9930
35
1.1253
65
0.51
44
1.2505 1.25
157
FIG.3.27. Graph for Fcl results in all 10 methods
158
3.5.5.2. Air Temperature -Ta
The temperature of the air surrounding the occupant, the operative temperature is the
uniform temperature of an imaginary enclosure in which the occupant would exchange the
same heat by radiation and convection as in the actual environment. When air temperature is
low, convective heat loss increases with air motion associated with increased activity, thereby
decreasing the heat load on the body evaporative system and resulting in a wider range of
activity before discomfort is felt.
Table.3.46. Ta results in all 10 methods
Trial
s
GENETIC
ALGORIT
HM
SA PS PSO G-L fminc
on
DE LGO glcClu
ster
glcSol
ve 1 19.4944 16.76
13
25 23.37
38
24.743
9
16 16.459
3
16.48
46
24.999
3
25
2 21.0165 17.03
333
25 27.20
17
20.158
1
20.877
2
26.238 16.48
46
24.999
3
25
3 17.5037 22.71
74
25 21.18
07
19.016 22.599
8
16.628
4
16.48
46
24.999
3
25
4 16.9313 17.26
93
25 21.87
89
22.162
3
21.397
7
26.841
5
16.48
46
24.999
3
25
5 16.124 19.12
86
25 19.81
1
21.614
3
19.534
6
20.769
3
16.48
46
24.999
3
25
6 18.5659 21.58
69
25 17.44
19
25.448
4
23.208
5
26.24 16.48
46
24.999
3
25
7 16.0234 23.32
74
25 20.81
6
23.226
8
28.521
1
20.909
8
16.48
46
24.999
3
25
8 21.4679 17.38
97
25 25.33
91
24.468
8
25.401
1
16.624
5
16.48
46
24.999
3
25
9 22.0884 25.54
12
25 20.84
11
20.406
3
25.684
6
24.985
6
16.48
46
24.999
3
25
10 16.9396 17.70
54
25 20.20
23
19.291
1
21.477
6
24.260
4
16.48
46
24.999
3
25
11 16.0002 21.64
12
25 20.25
1
19.489
4
21.171
5
27.105
9
16.48
46
24.999
3
25
12 21.6438 16.94
18
25 23.67
5
25.040
8
26.630
4
21.700
1
16.48
46
24.999
3
25
13 16.0078 16.89
94
25 20.20
28
23.454
7
26.588
1
22.042
4
16.48
46
24.999
3
25
14 23.9702 19.25
17
25 23.09
26
23.994
9
22.037
6
24.671 16.48
46
24.999
3
25
15 16.0302 16.18
2
25 21.15
07
22.681
5
22.935
5
23.621
5
16.48
46
24.999
3
25
16 19.5761 16.79
53
25 21.92
23
17.875
7
22.753
6
18.008
7
16.48
46
24.999
3
25
17 19.4237 25.95
43
25 23.38
45
19.942
7
20.097
1
21.004
1
16.48
46
24.999
3
25
18 17.7994 18.06
48
25 23.88
99
19.945
9
23.615
9
17.772
7
16.48
46
24.999
3
25
19 18.0889 27.48
51
25 21.37
8
25.142
8
21.610
4
27.297
3
16.48
46
24.999
3
25
20 16.0001 21.99
33
25 26.04
67
17.740
7
22.904
8
23.799
7
16.48
46
24.999
3
25
avg 18.53478 19.98
347
25 22.15
4
21.792
26
22.752
36
22.349
01
16.48
46
24.999
3
25
159
FIG.3.28. Graph for Ta results in all 10 methods
160
3.5.5.3. Mean radiant temperature-Tmrt
It is the uniform surface temperature of an imaginary black enclosure in which an
occupant would exchange the same amount of radiant heat as in the actual non uniform space.
The MRT affects the rate of radiant heat loss from the body. Since the surrounding surface
temperatures may vary widely, the MRT is a weighted average of all radiating surface
temperatures within a line of sight. In winter, levels of wall, roof, and floor insulation
together with window treatments such as double glazing, blinds, and drapes contribute to
Mean Radiant Temperature.
Table.3.47. T mrt results in all 10 methods
Tria
ls
GENETI
C
ALGORI
THM
SA PS PSO G-L fminco
n
DE LGO glcClu
ter
glcSol
ve 1 20.8528 21.976
2
21.5 20.33
11
22.098
1
21.270
3
20.929 21.41
27
20.083 20.08
17 2 20.0002 22.602
1
21.5 20.08
71
21.183
4
22.877
5
19.507
7
21.41
27
20.083 20.08
17 3 21.7467 19.884
7
21.5 22.74
09
21.599
6
22.54 22.211
4
21.41
27
20.083 20.08
17 4 20.5001 20.683
1
21.5 21.33
42
21.734
3
22.527
9
22.092
8
21.41
27
20.083 20.08
17 5 19.625 22.428
7
21.5 21.37
11
20.915
2
22.875 22.923
5
21.41
27
20.083 20.08
17 6 22.3007 22.845
8
21.5 22.27
41
21.483
5
21.228
9
22.436
5
21.41
27
20.083 20.08
17 7 19.6308 22.851
9
21.5 21.56
16
19.568
2
21.605
9
22.244
8
21.41
27
20.083 20.08
17 8 22.5261 22.435
9
21.5 21.50
31
20.402
1
22.372
4
19.509
6
21.41
27
20.083 20.08
17 9 20.6858 20.243
1
21.5 22.22
75
20.968
4
19.939
2
21.811
8
21.41
27
20.083 20.08
17 10 20.4827 22.877
2
21.5 22.93
89
22.095
3
21.257
1
21.451
2
21.41
27
20.083 20.08
17 11 21.0512 21.474
7
21.5 20.87
16
21.137
7
20.541
9
21.774
3
21.41
27
20.083 20.08
17 12 19.874 22.565
7
21.5 20.93
42
20.867 20.459
4
22.760
2
21.41
27
20.083 20.08
17 13 19.5063 21.102
1
21.5 21.86
96
21.108
7
21.203
2
22.912
2
21.41
27
20.083 20.08
17 14 21.3961 21.618
7
21.5 20.37
15
21.184
7
21.125
6
20.958
6
21.41
27
20.083 20.08
17 15 19.5315 20.374
5
21.5 20.78
39
21.626 22.416
2
21.471
9
21.41
27
20.083 20.08
17 16 19.6339 21.368
7
21.5 21.77
25
20.698
3
19.837
6
20.201
7
21.41
27
20.083 20.08
17 17 22.4563 19.545
2
21.5 21.73
58
20.901
7
21.981
1
22.192
8
21.41
27
20.083 20.08
17 18 20.859 22.915
5
21.5 20.07
22
21.978
9
21.584 21.251
9
21.41
27
20.083 20.08
17 19 21.32222 19.996 21.5 22.29
86
21.187
7
20.191
7
19.36 21.41
27
20.083 20.08
17 20 19.501 22.626
4
21.5 20.96
04
22.414 20.132
4
21.611
8
21.41
27
20.083 20.08
17 avg 20.67412 21.620
81
21.5 21.40
2
21.257
64
21.398
37
21.480
69
21.41
27
20.083 20.08
17
161
FIG.3.29.Graph for Tmrt results in all 10 methods
162
3.5.5.4. Relative air velocity-Vair
Air motion significantly affects body heat transfer by convection and evaporation. Air
Movement results from free convection from the occupants‘ bodily movements. The faster
the motion, the greater the rate of heat flow by both convection and evaporation. When
ambient temperatures are within acceptable limits, there is no minimum air movement that
must be provided for thermal comfort. The natural convection of air over the surface of the
body allows for the continuous dissipation of body heat. When ambient temperatures rise,
however, natural air flow velocity is no longer sufficient and must be artificially increased,
such as by the use of fans.
Table.3.48. Vair results in all 10 methods
Tria
ls
GENETIC
ALGORIT
HM
SA PS PSO G-L fminc
on
DE LGO glcClus
ter
glcSol
ve 1 0.9459 0.1787 0.8
75
0.410
9
0.953
4
0.903
2
1.0098 0.59
52
0.6009 0.6
2 0.9705 0.2307 0.8
75
0.884
6
0.453
5
0.493
1
0.4031 0.59
52
0.6009 0.6
3 0.6219 0.1307 0.8
75
0.329
3
0.561
4
0.585
6
0.1628 0.59
52
0.6009 0.6
4 0.1002 0.6501 0.8
75
0.283
8
0.634
6
0.952 0.6045 0.59
52
0.6009 0.6
5 0.592 0.3345 0.8
75
0.230
5
0.484
3
0.238
6
0.2386 0.59
52
0.6009 0.6
6 0.4528 0.3663 0.8
75
0.824
1
0.193
3
0.699
8
0.4426 0.59
52
0.6009 0.6
7 1.0346 1.0785 0.8
75
0.323
7
0.971
5
0.869 0.3214 0.59
52
0.6009 0.6
8 0.7011 0.8208 0.8
75
0.680
9
0.892
1
0.804
3
0.1327 0.59
52
0.6009 0.6
9 0.5245 0.919 0.8
75
0.146
5
0.947
8
0.471
3
0.6006 0.59
52
0.6009 0.6
10 0.2554 0.6501 0.8
75
0.364
8
0.771 0.19 0.8512 0.59
52
0.6009 0.6
11 0.8308 1.0877 0.8
75
0.206
7
1.054 0.534
2
0.3754 0.59
52
0.6009 0.6
12 0.2834 0.5257 0.8
75
0.481
6
0.856
3
0.916
3
0.4382 0.59
52
0.6009 0.6
13 0.1002 0.1421 0.8
75
0.523
7
0.526
8
0.561
8
0.2916 0.59
52
0.6009 0.6
14 0.719 0.4084 0.8
75
0.623
6
0.568
5
0.794
3
0.5538 0.59
52
0.6009 0.6
15 0.1587 1.0039 0.8
75
0.701
6
0.422
7
0.920
4
0.1199 0.59
52
0.6009 0.6
16 0.471 0.5555 0.8
75
0.614
6
0.814
9
0.626
6
0.2537 0.59
52
0.6009 0.6
17 0.3849 1.0632 0.8
75
0.178
7
0.293
3
0.231
2
0.6023 0.59
52
0.6009 0.6
18 0.4229 0.8181 0.8
75
0.253
7
0.206
1
0.538
8
0.4192 0.59
52
0.6009 0.6
19 0.9926 0.1605 0.8
75
0.763
2
0.484 0.504
2
0.9973 0.59
52
0.6009 0.6
20 0.3986 0.8452 0.8
75
0.883
7
0.468
9
0.923
5
0.9562 0.59
52
0.6009 0.6
avg 0.54805 0.5984
85
0.8
75
0.485
51
0.627
92
0.637
91
0.4887
45
0.59
52
0.6009 0.6
163
FIG.3.30. Graph for Vair results in all 10 methods
164
3.5.5.5. Partial water vapour pressure-Pa
The upper and lower humidity limits on the comfort envelope are based on
considerations of respiratory health, mould growth, and other moisture-related phenomena in
addition to comfort. Humidification in winter must be limited at times to prevent
condensation on cold building surfaces such as windows. The environmental parameters of
temperature, radiation, humidity, and air movement necessary for thermal comfort depend
upon the occupant‘s clothing and activity level.
Table.3.49. Pa results in all 10 methods
Tria
ls
GENETIC
ALGORIT
HM
SA PS PSO G-L fminc
on
DE LG
O
glcClu
ster
glcSol
ve 1 0.5154 0.070
7
0.410
9
0.234
5
0.3357 0.2912 0.418
4
1 0.0649 0.051
2 0.6392 0.219 0.884
6
0.222
9
0.4517 0.702 0.727
4
1 0.0649 0.051
3 0.624 0.139
9
0.329
3
0.644
1
0.7441 0.3781 0.551
6
1 0.0649 0.051
4 0.0203 0.131
2
0.283
8
0.575
3
0.5398 0.4354 0.974
6
1 0.0649 0.051
5 0.1427 0.236
8
0.230
5
0.402
1
0.6024 0.1399 0.974
3
1 0.0649 0.051
6 0.0923 0.950
3
0.824
1
0.585
7
0.4468 0.7324 0.093 1 0.0649 0.051
7 0.4752 0.926
8
0.323
7
0.112
4
0.5729 0.1975 0.2116 1 0.0649 0.051
8 0.0748 0.145
7
0.680
9
0.367
7
0.9273 0.919 0.681
5
1 0.0649 0.051
9 0.5398 0.94 0.146
5
0.731
3
0.3115 0.2795 0.2114 1 0.0649 0.051
10 0.7357 0.666
8
0.364
8
0.338
6
0.2226 0.2084 0.361
9
1 0.0649 0.051
11 0.06666 0.816
3
0.206
7
0.685
4
0.7742 0.4896 0.572
6
1 0.0649 0.051
12 0.4901 0.349
1
0.481
6
0.607
9
0.4135 0.8729 0.337
4
1 0.0649 0.051
13 0.0123 0.812
7
0.523
7
0.600
9
0.7551 0.5002 0.230
7
1 0.0649 0.051
14 0.166 0.208
7
0.623
6
0.563
6
0.7549 0.6318 0.892
7
1 0.0649 0.051
15 0.0609 0.288
8
0.701
6
0.277
4
0.2348 0.9014 0.244
4
1 0.0649 0.051
16 0.5052 0.982 0.614
6
0.802
5
0.851 0.8484 0.285
3
1 0.0649 0.051
17 0.9576 0.469
9
0.178
7
0.640
5
0.8538 0.6113 0.795
6
1 0.0649 0.051
18 0.6014 0.765 0.253
7
0.687
6
0.5098 0.2586 0.704
2
1 0.0649 0.051
19 0.6103 0.508
6
0.763
2
0.654
8
0.9585 0.5998 0.671
8
1 0.0649 0.051
20 0.1023 0.653
1
0.883
7
0.059
6
0.8237 0.6705 0.287
96
1 0.0649 0.051
avg 0.371608 0.514
07
0.485
51
0.489
74
0.6042
05
0.5333
95
0.5114
18
1 0.0649 0.051
165
FIG.3.31. Graph for Pa results in all 10 methods
166
3.5.5.6. Surface temperature of clothing-Tcl
Clothing, through its insulation properties, is an important modifier of body heat loss
and comfort. The insulation properties of clothing are, a result of the small air pockets
separated from each other to pre air from migrating through the material. When preferred
amount of clothing worn by building occupants decreased, then correspondingly, the
preferred temperatures increased. These seasonal clothing variations of building occupants
allow indoor temperature ranges to be higher in the summer than in the winter and yet remain
comfortable. In winter, additional clothing lowers the ambient temperature necessary for
comfort and for thermal neutrality.
Table.3.50. Tcl results in all 10 methods.
Trial
s
GA SA PS PSO G-L fminc
on
DE LGO glcClu
ster
glcSo
lve 1 26.965
7
29.761
6
28 27.99
77
28.171
6
29.30
32
27.92
97
29.44
86
28.001 28
2 29.693
8
27.772
3
28 29.38
55
27.263
7
28.59
3
29.31
79
29.44
86
28.001 28
3 26.388
4
28.474
6
28 26.58
87
28.846
3
29.95
52
26.67
3
29.44
86
28.001 28
4 28.827
4
26.206 28 27.46
35
28.126
5
27.53
74
29.46
5
29.44
86
28.001 28
5 29.001
6
26.771
5
28 29.84
31
28.065
7
29.98
45
28.90
68
29.44
86
28.001 28
6 26.890
5
27.154
2
28 26.72
09
29.235
7
27.58
38
29.11
73
29.44
86
28.001 28
7 29.743
9
29.105
5
28 27.63
23
27.478
5
29.99
99
27.81
49
29.44
86
28.001 28
8 28.612
1
26.268
8
28 28.57
42
29.466
6
29.95
38
28.01
77
29.44
86
28.001 28
9 28.374
6
28.249
9
28 29.58
99
28.576
7
29.55
27
29.03
49
29.44
86
28.001 28
10 28.099
6
28.911
7
28 27.34
58
27.369
7
27.69
31
27.45
75
29.44
86
28.001 28
11 27.749 27.598
1
28 26.99
38
26.503 29.87
11
29.86
24
29.44
86
28.001 28
12 29.970
6
28.864 28 27.90
9
29.850
4
29.08
37
26.84
02
29.44
86
28.001 28
13 26.063
5
29.801
2
28 27.72
89
29.150
6
28.87
41
27.08
16
29.44
86
28.001 28
14 29.107
3
29.008
4
28 28.42
4
29.063
8
30 28.80
4
29.44
86
28.001 28
15 26.015
8
28.433
1
28 29.21
14
26.886
1
30 29.14
4
29.44
86
28.001 28
16 27.404
2
26.645
4
28 28.24
93
26.050
6
29.91
18
26.78
57
29.44
86
28.001 28
17 27.403
8
28.475
4
28 29.59
96
26.667
1
26.34
22
26.11
28
29.44
86
28.001 28
18 26.000
2
26.799
8
28 29.53
36
26.979
7
29.02
58
27.14
71
29.44
86
28.001 28
19 28.373
5
29.928 28 26.12
15
28.765
2
26.00
24
28.92
75
29.44
86
28.001 28
20 29.812
1
29.053
6
28 28.73
55
26.295 27.01
77
26.95
75
29.44
86
28.001 28
avg 28.024
88
28.164
16
28 28.18
241
27.940
63
28.81
427
28.06
988
29.44
86
28.001 28
167
FIG.3.32. Graph for Tcl results in all 10 methods
168
3.5.5.7. Predicted mean vote (PMV):
PMV is an index that predicts the mean value of the votes of a large group of persons
on the seven point thermal sensation scale. There is not even a single set of conditions that
will satisfy all occupants. Each person has a distinct perception of too hot, too cold, and
comfortable. The objective in designing a common thermal environment is to satisfy a
majority of occupants and to minimize the number of people who will inevitably be
dissatisfied.
Table.3.51. PMV results in all 10 methods
Trials GA SA PS PSO G- L fminc
on
DE LGO glc
Clu
glc
Sol
1 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5 0.113
1
-0.5 -0.5 -0.5
2 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5 0.484
3
-0.5 -0.5 -0.5
3 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5 0.151
1
-0.5 -0.5 -0.5
4 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5 0.383
4
-0.5 -0.5 -0.5
5 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5 0.310
3
-0.5 -0.5 -0.5
6 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5 0.354
8
-0.5 -0.5 -0.5
7 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5 0.353
1
-0.5 -0.5 -0.5
8 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5 0.396
1
-0.5 -0.5 -0.5
9 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5 0.315
2
-0.5 -0.5 -0.5
10 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5 0.492
3
-0.5 -0.5 -0.5
11 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5 0.495
7
-0.5 -0.5 -0.5
12 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5 0.371
2
-0.5 -0.5 -0.5
13 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5 0.161
9
-0.5 -0.5 -0.5
14 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5 0.377
2
-0.5 -0.5 -0.5
15 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5 0.254
5
-0.5 -0.5 -0.5
16 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5 0.423
7
-0.5 -0.5 -0.5
17 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5 0.277
6
-0.5 -0.5 -0.5
18 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5 0.499 -0.5 -0.5 -0.5
19 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5 0.498
1
-0.5 -0.5 -0.5
20 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5 0.482
4
-0.5 -0.5 -0.5
avg -0.5 -0.5 -0.5 -0.5 -0.5 -0.5 0.359
75
-0.5 -0.5 -0.5
169
FIG.3.33. Graph for PMV results in all 10 methods
170
3.5.5.8. Predicted percentage of Dissatisfied. (PPD):
An index that establishes a quantitative prediction of the percentage of thermally
dissatisfied people determined from PMV. As PMV changes away from zero in either the
positive or negative direction, PPD increases. Determination of the PMV and PPD Indices
and Specification of the Conditions for Thermal Comfort uses limits on PMV as an explicit
definition of the comfort zone.
Table.3.52. PPD results in all 10 methods
Trials GA SA PS PSO G-L fmincon DE LGO glcClu glcSol
1 5 5 5 5 5 5 5 5 5 5
2 5 5 5 5 5 5 5 5 5 5
3 5 5 5 5 5 5 5 5 5 5
4 5 5 5 5 5 5 5 5 5 5
5 5 5 5 5 5 5 5 5 5 5
6 5 5 5 5 5 5 5 5 5 5
7 5 5 5 5 5 5 5 5 5 5
8 5 5 5 5 5 5 5 5 5 5
9 5 5 5 5 5 5 5 5 5 5
10 5 5 5 5 5 5 5 5 5 5
11 5 5 5 5 5 5 5 5 5 5
12 5 5 5 5 5 5 5 5 5 5
13 5 5 5 5 5 5 5 5 5 5
14 5 5 5 5 5 5 5 5 5 5
15 5 5 5 5 5 5 5 5 5 5
16 5 5 5 5 5 5 5 5 5 5
17 5 5 5 5 5 5 5 5 5 5
18 5 5 5 5 5 5 5 5 5 5
19 5 5 5 5 5 5 5 5 5 5
20 5 5 5 5 5 5 5 5 5 5
avg 5 5 5 5 5 5 5 5 5 5
171
FIG.3.34. Graph for PPD results in all 10 methods
172
3.5.5.9 Elapsed Time.
CPU time is the time for which the CPU was busy executing the task. It does not take
into account the time spent in waiting for I/O (disk IO or network IO). Since I/O operations,
such as reading files from disk, are performed by the OS, these operations may involve
noticeable amount of time in waiting for I/O subsystems to complete their operations. This
waiting time will be included in the elapsed time, but not CPU time. Hence CPU time is
usually less than the elapsed time.
Table.3.53. Elapsed time results in all 10 methods
Trai
ls
GA SA PS PSO G-L fminc
on
DE LGO glcClu
ster
glcSol
ve 1 0.551
04
3.119
308
0.293
276
0.083
084
2.025
347
13.78
147
0.538
796
0.987
521
41.073
85
10.08
039 2 0.247
523
6.312
936
0.115
934
0.088
901
2.886
651
16.16
143
0.524
793
0.896
383
42.066
38
9.927
723 3 0.252
798
3.971
437
0.128
355
0.993
323
2.457
629
14.66
16
0.890
405
0.957
826
36.064
55
10.85
974 4 0.259
975
3.247
668
0.109
838
0.084
307
2.946
023
16.61
047
0.541
612
0.893
158
40.065
94
11.44
608 5 0.256
667
3.922
584
0.109
052
0.093
717
2.576
018
16.42
167
0.569
3
1.147
994
40.064
22
10.88
564 6 0.252
593
4.568
592
0.118
687
0.089
055
2.079
802
15.01
783
0.590
935
0.789
401
39.066
12
9.828
382 7 0.250
631
3.905
444
0.116
927
0.090
419
2.800
292
16.79
836
0.561
517
0.872
425
39.066
6
9.894
192 8 0.253
323
3.317
35
0.112
212
0.081
506
2.275
623
16.08
125
0.568
742
0.902
255
41.064
28
10.66
552 9 0.254
819
3.121
085
0.105
636
0.085
774
2.493
892
14.76
948
0.584
803
0.915
021
39.067
79
10.58
144 10 0.257
802
2.811
64
0.108
933
0.086
711
2.604
888
16.32
709
0.535
897
0.929
707
39.068
79
11.31
168 11 0.263
547
2.962
492
0.113
323
0.087
111
2.075
249
13.49
19
0.530
987
0.982
916
38.065
82
11.04
852 12 0.250
99
4.061
731
0.108
616
0.078
993
2.439
861
16.52
828
0.585
732
1.127
479
39.064
5
10.39
147 13 0.268
359
3.374
261
0.112
946
0.083
554
2.787
642
10.30
411
0.577
844
0.948
522
40.064
73
11.78
036 14 0.259
91
3.390
391
0.109
518
0.090
956
2.371
622
13.81
421
0.534
981
0.829
222
41.064
5
11.52
661 15 0.254
022
4.726
611
0.122
536
0.096
621
2.184
604
17.28
005
0.532
066
0.846
84
41.065
16
10.10
532 16 0.251
563
3.429
903
0.113
252
0.084
475
1.970
465
16.22
84
0.588
487
0.999
231
40.066
63
11.00
906 17 0.258
912
3.485
697
0.109
042
0.087
263
2.104
446
12.50
945
0.543
563
0.891
372
39.063
95
8.926
088 18 0.253
386
3.378
421
0.118
86
0.084
749
2.402
594
13.17
272
0.548
253
1.116
948
38.065
02
9.978
92 19 0.251
14
2.677
337
0.120
268
0.079
959
1.901
257
14.16
214
0.547
068
1.073
188
37.067
34
11.67
263 20 0.259
745
3.078
47
0.112
034
0.083
663
1.623
168
13.19
02
0.553
243
0.963
335
36.483
81
12.09
722 avg 0.270
437
3.643
168
0.122
962
0.131
707
2.350
354
14.86
561
0.572
451
0.953
537
39.337 10.70
085
173
FIG.3.35. Graph for Elapsed time results in all 10 methods
174
3.5.5.10. Iterations
Iteration is a computational procedure in which a cycle of operations is repeated, often
to approximate the desired result more closely. Iteration means the act of repeating a process
usually with the aim of approaching a desired goal or target or result. Iteration in computing
is the repetition of a process within a computer program. It may also refer to the process of
iterating a function i.e. applying a function repeatedly, using the output from one iteration as
the input to the next. Another use of iteration in mathematics is in iterative methods which
are used to produce approximate numerical solutions to certain mathematical problems.
Newton's method is an example of an iterative method.
Table.3.54. Iterations time results in all 10 methods
Trials GA SA PS PSO G-L fmincon DE LGO glcClu glcSol
1 51 3000 20 51 4 2338 60 3851 6499 34
2 51 3000 20 51 4 2296 60 3851 6499 34
3 51 3000 20 51 4 2275 60 3851 6499 34
4 51 3000 20 51 4 2317 60 3851 6499 34
5 51 3000 20 51 4 2324 60 3851 6499 34
6 51 3000 20 51 4 2310 60 3851 6499 34
7 51 3000 20 51 4 2296 60 3851 6499 34
8 51 3000 20 51 4 2282 60 3851 6499 34
9 51 3000 20 51 4 2289 60 3851 6499 34
10 51 3000 20 51 4 2268 60 3851 6499 34
11 51 3000 20 51 4 2324 60 3851 6499 34
12 51 3000 20 51 4 2387 60 3851 6499 34
13 51 3000 20 51 4 2282 60 3851 6499 34
14 51 3000 20 51 4 2296 60 3851 6499 34
15 51 3000 20 51 4 2331 60 3851 6499 34
16 51 3000 20 51 4 2289 60 3851 6499 34
17 51 3000 20 51 4 2324 60 3851 6499 34
18 51 3000 20 51 4 2289 60 3851 6499 34
19 51 3000 20 51 4 2324 60 3851 6499 34
20 51 3000 20 51 4 2324 60 3279 6499 34
Avg
TAB
LE.3.
51.
Elaps
ed
time
result
51 3000 20 51 4 2308 60 3822.
4
6499 34
175
FIG.3.36. Graph for Iterations results in all 10 methods.
176
Table.3.55. Comparative results of the parameters in all 10 methods.
Variables GA SA PS PSO G-L Fmincon DE LGO glcCluster glcSolve
Fcl X X 1.25
X X X X 0.51
1.25
1.25
Ta X X
25 X X X
16.48
24.9
25
Tmrt X X 21.5
X X X 21.4
20.08
20.08
Vair X X
0.875 X X X
0.6
0.6
0.6
Pa √
1
0.06
0.05
Tcl X X
28
X X X X
29.44
X
28.001
X
28 PMV
-0.5
-0.5
-0.5
-0.5
-0.5
-0.5
-0.5
-0.5
-0.5
-0.5
PPD 5
5
5
5
5
5
5
5
5
5
Time 0.12 0.13
Iter X X 20
4
X X X X X X
3.5.6. Result and Discussion
With the two extreme values of parameters from survey, the optimization is carried
out with different solvers. As they are of stochastic type their results may vary from trial to
trial so and the problem is made to run for 20 trials (Elbeltagi, Tarek Hegazy, & & Grierson,
2005) and an average of all trials is taken as the final value of the parameter, by the solver.
The solvers are compared with three different criteria
1. Consistency
The consistency Table gives the parameters that remain constant for all the
trials. All the solvers give the same value of PMV& PPD for all the runs
except DE, which in turn indicate that the comfort requirements are in the
acceptable range
Fcl - P.S (1.2), glcSolve (1.25), glcCluster (1.25), LGO (0.51)
Ta - P.S (25), glcSolve (25), glcCluster (25), LGO (16.48)
Tmrt - P.S (21.5), glcSolve (20.08), glcCluster (20.08), LGO (21.41)
177
Vair - P.S (0.8), glcSolve (0.6), glcCluster (0.6009), LGO (0.59)
Pa - P.S (0.485), glcSolve (0.05), glcCluster (0.064), LGO (1)
Tcl - P.S (28), glcSolve (28), glcCluster (28.001), LGO (29.44)
So we see that the solvers Pattern Search, glcSolve, glcCluster& LGO
remain constant throughout their runs.
2. Minimum Run Time
For minimum run time of the problem, we got PSO (0.131 seconds), Pattern
Search (0.122 seconds).
3. Minimum Evaluation
This criterion will determine the effectiveness of the algorithm. From the
result table, we see that the Pattern Search and GODLIKE algorithms have
minimum evaluation of 20 and 4 respectively
4. Simplicity of Algorithm
Of all the algorithms, we have taken the Pattern Search algorithm is the most
simplest followed by GA, PSO, DE, Simulated Annealing, GODLIKE, Non-
Linear, Direct algorithm.
5. Results according to Standards
This is the most important criterion that determines whether the solver is
practical or not. We got the standard values for a naturally ventilated building
from ASHRAE as:
Humidity: 30% to 60%
(http://www.epa.gov/iaq/largebldgs/i-beam/text/hvac.html)
This gives that the Pa should lie within the range of: 0.0765 to 0.501
(http://www.engineeringtoolbox.com/water-vapor-saturation-pressure-
air-d_689.html)
Operative Temperature: 17.75 to 28.5
Air velocity:0.2 to 0.8 ms-1
(1 ms-1
only at extreme conditions
With the above standards the solvers which adhere to the standard are:
Air-Velocity: Fmincon, GA, SA, PSO, GL, DE, glcSolve.
Partial vapour pressure: GA, PS, PSO, LGO, glcCluster, glcSolve
178
Operative temperature: GA, SA,PS, PSO, Fmincon, DE, GL, LGO,
glcCluster, glcSolve
The following Table gives a summary of all the criteria for the solvers:
Table 3.56. Summary of all the criteria for the solvers
Criteria GA SA PS PSO Fmincon DE GL LGO glcClus glcSolve
Result
according to
ASHRAE
3/3
=100%
2/3
=67%
2/3
=67%
3/3
=100%
2/3
=67%
3/3
=100%
2/3
=67%
2/3
=67%
2/3
=67% 3/3
=100%
Consistency - - - - - -
Min-Run
Time - - - - - - - -
Min-
Evaluation - - - - - - - -
Simple
Algorithm - - - - - - - - -
Thus, it is seen that the Pattern Search solver satisfies all the criteria and scores 67%
for its practicality in giving results according to ASHRAE. So the appropriate
algorithm, for optimization of thermal comfort is suggested as Direct search
algorithm & the solver is PATTERN SEARCH
3.5.7. Conclusion
This study investigates the thermal environment and comfort of residences in the
Karunya University, Coimbatore. A total of 102 occupants in naturally ventilated 11
residences buildings ( with occupant – operable windows) provided thermal perception
data, first field campaign from Mar6, 2010 to Mar15,2010 and second field campaign from
Sep1,2010 to Sep 10, 2010 in Karunya University, Coimbatore. In both the sets the same
buildings were taken into account for data collection. Indoor climatic data were collected
using instruments with accuracies and response times in accordance with the
recommendations of ANSI/ASHRAE 55. All the measurements were carried out between
06:00 hours and 20:00 hours. All the houses which were surveyed were non-air-conditioned
179
residences, where natural ventilation is preferred. The result of the filed survey and
measurement study can be used to design a low energy consumption system with
consideration of occupant thermal comfort in Coimbatore, Tamil Nadu.
In the experiment conducted, using ten non-traditional optimization techniques
the thermal sensation takes the value -0.5., which is in the acceptable range , where the
acceptable range is -0.5 to +0.5 (ANSI/ASHRAE55-2004, 2004). Therefore the thermal
comfort of the residential buildings of the Karunya University in Coimbatore is in the
acceptable range.
Here, ten non-traditional optimization algorithms were presented. These include: GA,
SA, PS, PSO, GL, FMINCON, EA, LGO, glcCluster, glcSolve. A brief description of each
method is presented along with a Pseudo code to facilitate their implementation. MATLab
programs were written to implement each algorithm. The thermal comfort problem for the
offices of the Karunya University was solved using all algorithms, and the comparative
results were presented and the pattern search method of optimization is the best method.
Bibliography
ANSI/ASHRAE55-2004. (2004). Thermal Environmental conditions for Human occupancy.
Atlanda, USA: American Society of Heating, Refrigerating and Air-Conditioning Engineers.
ASHRAE. (1989). Handbook-fundamentals, chapter 8,, Physiological Principles, Comfort
and Health.
ASHRAE, A. s.-2. (2007). Design for accepTable indoor air quality,. Atlanta: American
Society of Heating, Refrigerating and AIr-conditioning Engineers, Inc.,.
Auliciems.A. (1984). Thermobile controls for human comfort. Heating and ventlating
Engineeers , April/May 31-33.
Cabanac.M. (1971). Physiological role of pleasure. Science , V17, 1103 - 1107.
Coome, D., Gan,G, & Awbi,H.B. (1992). Evaluation of Thermal comfort and indoor air
quality. Proc.CIB'92 World Building Congress, (pp. 404-406). Montre'al.
Culp, C., Rhodes, M., Krafthefer, B., & Listvan, M. (1993). Silicon Infrared Sensors for
Thermal comfort and Control. ASHRAE Journal , April 38-42.
de Dear, R. a. (2002). Thermal comfort in naturally ventilated buildings:revisions to
ASHRAE Standard55. Energy and Buildings , 34 (6),pp 549-561.
180
de Dear, R. (2004). Thermal comfort in practise. Indoor Air , vol 14, 32-39.
Elbeltagi, E., Tarek Hegazy, 1., & & Grierson, D. (2005). . Comparison among five
evolutionary-based optimization algorithms. . Advanced Engineering Informatics , 19, 43-53.
Fanger, P., & Toftum, j. (2002). Extension of the PMV model to non-air-conditioned
buildings in warm climates. Energy and Buildings , 34,PP 533-536.
Fanger.P.O. (1970). Thermsl comfort: Analysis and Applications in Environmental
Engineering. New York: McGraw-Hill.
Fanger.P.O., T. (2002). Extension of the PMV model to non-air-conditioned buildings in
warm climates. Energy and Buildings , 34(6), 533-536.
Farzaneh, Y., & Tootoonchi, A. R. (2008). Controlling automobile thermal comfort using
optimizes fuzzy controller. Applied Thermal Engineering , Vol9 N04 1367-1376.
Federspiel, C. (1882). Used-adapTable and minimum-power thermal comfort Control. Ph.D
Thesiis,MIT,Department of Mechanicla Engineering.
Fountain, m., Brager, G., Arens, E., Bauman, F., & Benton, C. (1994). Comfort Control for
short-term occupancy. Energy and Buildings , 211-213.
Freire, Z., Roberto, H.C.Oliveira, G., & Nathan.Mendes. (2008). Predictive controllers for
thermal comfort optimization and energy savings. Energy and Buildings , 1353-1365.
Gan, G., & Croome, D. (1994). Thermal comfort models based on Field measurements.
ASHRAE Transactions , 782-704.
Hamdi, M., Lachiver, G., & Michaud, F. (1999). A new predictive thermal sensation index of
human response. Energy and Buildings , 167-178.
Hancock, P., Ross, J., & Szalma, J. (2007). A meta-analysis of performance response under
thermal stressors. Human factors , 49 (5),851.
Humphreys, M., & Nicol, J. (2002). The validity of ISO-PMV for predicting comfort votes in
every thermal environments. Energy and Buildings , 34, 667 - 684.
InternationalStandardISO7730. (1994). Moderate thermal environments- Determinatiob of
the PMV and PPD indices and specification of the conditions for thermal comfort.
Int-Hout, D. (1990). Thermal Comfort Calculations/A computer model. ASHRAE
Transactions , 96(1) 840-844.
Khodakarami, J. (2009). Achieving thermal comfort. VDM Verlag , ISBN 978-3-639-18292-
7.
Leon, L. (2008). Thermoregulatory responses to environmental toxicants: the interaction of
thermal stress and toxicant exposure. Toxicology and Applied pharmacology , 233 (1), 146.
181
MacArthur, J. (1986). Humidity and predicted-mean-vote-based PMV-Based Comfort
control. ASHRAE Transactions , 15-17.
Mallick, F. (1996). tHERMAL COMFORT AND bUILDING DESIGN IN THE TROPICAL
CLIMATES. eNERGY AND bUILDINGS , 23, 161 - 167.
MinNing, & Zaheeruddin, M. (2010). Neuro-optimal operation of a variable air volume
HVAC&R system. Applied Thermal Engineering , 385-399.
Nicol, F. (2004). Adaptive thermal comfort standards in the hot humid tropics. Energy and
Buildings , 36, 628 -637.
Nicol, J., & Humphreys.A'. (2001). Adaptive thermal comfort and sustainable thermal
standards for buildings. conference Moving thermal comfort Standards into the 21st century,
Windor,UK , 5 -8.
Nicol, J., Raja, I., ALLudin, A., & Jamy, G. (1999). Climatic variation in comforTable
temperatures: the Pakistan Projects. Energy and Buildings , 30, 261 - 279.
Ning, M., & Zaheeruddin, M. (2010). Neuro-optimal operation of a variable air volume
HVAC&R system. Applied Thermal Engineering , 385-399.
Oseland.N.A. (1995). Predicted and reported thermal sensation in climate chambers, offices
and homes. Energy and Buildings , 23(2) 105-115.
Santamouris, M. (2004). ADAPTIVE THERMAL COMFORT AND VENTILATION. aIR
INFILTRATION AND vENTILATION , 18-24.
Sherman, M. (1985). A Simplified Model of Thermal Comfort. Energy and Buildings , 8, 37-
50.
Smolander, J. (2002). Effect of cold exposure on older hmans. International Journal of
Sports Medicine , 23(2), 86.
Soyguder, S., & Ali, H. (2009). An expert system for the humidity and temperature control in
HVAC systems using ANFIS and optimization wiht fuzzy modelling approach. Energy and
buildings , 41, 814 - 822.
Stoops, J. (2004). A possible connection between thermal comfort and health. Retrieved
october 2010, from Lawrence Berkeley National Laboratory,Paper 55134:
http://repositories.cdlib.org/lbnl/LBNL-55134/
Tanabe, S. (1988). Thermal comfort requirements in Japan. Waseda Univeristy: Doctoral
Thesis.
Thompson, R., & Dexter, A. (2009). Thermal Comfort Control based on Fuzzy Decision-
making.
182
Toftem, J. (2002). Human response to combined indoor environment exposures. Energy and
Buildings , 34(6),601-606.
Toftum, J. (2004). Air movement - Good or bad? Indoor Air , 14,pp 40 -45.
Toftum, J. (2005). Thermal comfort Indices. Boca Raton: Handbook of Human Factors and
Ergonomics Methods,63,CRC Press.
Wang, L., & Mendel, J. (1992). Generating fuzzy rules by learning from examples, . IEEE
Trans. on Systems, Man, Cybernetics , 22 (6) 1414–1427.
Xia, Y., R.Y, & Zhao, Y. J. (1999). Thermal comfort in naturally ventilated houses in
Bejjing. HV&AC , 1-5.
YadollahFarzanth, & Tootoonchi, A. R. (2008). Controlling automobile thermla comfort
using optimizaed fuzzy controller. Applied Thermal Engineering , vol9,no.4,pg 1367-1376.
(n.d.). Retrieved from http://www.epa.gov/iaq/largebldgs/i-beam/text/hvac.html.
(n.d.). Retrieved from http://www.engineeringtoolbox.com/water-vapor-saturation-pressure-
air-d_689.html.