Tb

Ma

b

c

d

a

ARRA

KTIVN

1

otleivatccetaEhtd

0d

Energy and Buildings 42 (2010) 1610–1618

Contents lists available at ScienceDirect

Energy and Buildings

journa l homepage: www.e lsev ier .com/ locate /enbui ld

hermal monitoring and indoor temperature modeling in vernacularuildings of North-East India

anoj Kumar Singha,∗, Sadhan Mahapatrab, S.K. Atreyaa, Baruch Givonic,d

Instrument Design and Development Centre, Indian Institute of Technology Delhi, New Delhi 110016, IndiaDepartment of Energy, Tezpur University, Tezpur 784028, Assam, IndiaDepartment of Architecture, UCLA, Los Angeles, CA, USABen Gurion University, Beer Sheva, Israel

r t i c l e i n f o

rticle history:eceived 2 March 2010eceived in revised form 8 April 2010ccepted 10 April 2010

eywords:

a b s t r a c t

Vernacular architecture is still very popular and constructed widely in North-East India. In this paper,the result of long-term monitoring of two vernacular houses selected one in Tezpur (warm and humidclimate) and other one in Cherrapunjee (cold and cloudy climate) are presented. Long-term monitoringwork includes the measurements of temperature (inside and outside house), relative humidity (insideand outside house) and illumination level (inside and outside house) for 25 days in all the seasons (Jan-

hermal monitoringndoor temperature modelingernacular architectureorth-East India

uary: winter, April: spring/pre-summer, July: summer/monsoon and October: autumn/pre-winter) of theyear 2008. Temperatures profile across all the seasons represents strong daily and seasonal fluctuations.Formulae have been developed based on part of the monitoring data to predict the indoor maximum,average and minimum temperatures inside the same house occupied by the same family. The predictedformulae were developed based on the measured data for the month of January and July and were val-idated with the measured data of April and October months. It is found that the correlation coefficient

r all

(R2 value) is above 0.96 fo

. Introduction

Vernacular architecture is a term used to categorize methodsf construction which uses locally available resources to addresshe local needs [1]. These kinds of buildings are constructed usingocally available materials and shows a greater respect to thexisting environment and also takes into account the constraintsmposed by the climate [2]. It is revealed from different studies onernacular architecture that bioclimatism is a critical parameter forchieving sustainability of modern architecture and this conceptakes into account the solar passive techniques and micro-climaticonditions in building design; which improves the building artifi-ial energy efficiency and thermal comfort conditions in the builtnvironment [3–5]. Vernacular houses of North-East India acrosshe different bioclimatic zones are widely varied in their built formsnd functionality. Entire region has more than 50 ethnic groups.ach ethnic group has distinct cultural and social setup. Vernacular

ouses constructed by these ethnic groups are in direct response tohe local climate, social and cultural setup. This type of houses andesign layout are still very popular and are widely constructed [6].

∗ Corresponding author. Tel.: +91 11 26591430; fax: +91 11 26582037.E-mail address: mksinghtu@gmail.com (M.K. Singh).

378-7788/$ – see front matter © 2010 Elsevier B.V. All rights reserved.oi:10.1016/j.enbuild.2010.04.003

the six formulae for the entire monitoring period.© 2010 Elsevier B.V. All rights reserved.

The existing thermal comfort standards is based on heat bal-ance model of human body and derived from extensive laboratoryexperiments in different climatic chambers [7]. However, the con-ditions in a building are much more dynamic in terms of boththermal environment and occupants activities. This leads to devia-tion in the results when these thermal comfort models are appliedto existing buildings [7]. So, there was a need of better understand-ing that takes into account of both dynamic thermal environmentand occupants activities. Research on ‘adaptive’ theory of thermalcomfort first began in mid-1970s in response to the first oil shockwith a prime concern to find out the human impact on the globalclimate environment. Adaptive theory primarily uses outdoor envi-ronmental variables to predict the indoor thermal environment [7].This automatically takes into accounts of climatic conditions, socialconditioning and other contextual factors.

Givoni presented formulas predicting daily maximum, averageand minimum indoor temperature of two unoccupied buildingsin Pala, California, with very scant climate data [8]. Kruger andGivoni reported long-term monitoring work of outdoor temper-ature measurement at seven houses in Curitiba, Brazil and they

have generated such formulas for the occupied houses with cor-relation coefficients up to 0.9894 [9,10]. In this study, it has beendemonstrated that the management of the houses by the occupantshad larger impact on the indoor temperatures than the physi-cal properties of the buildings. Kruger and Givoni have reported

d Buildings 42 (2010) 1610–1618 1611

lmplimtit

ioahavhmntp

tmtflimztitcdttamz

2I

wvfalitocfewsMlhoad

M.K. Singh et al. / Energy an

ong-term thermal monitoring of an occupied passive solar apart-ent in an arid environment also with mathematical formulas

redicting its indoor temperature [11]. Validation of the formu-as showed a good agreement between measured and calculatedndoor maximum, average and minimum temperature. These for-

ulas are applicable, and can be used, only for the specific buildinghat were monitored, occupied by the same family, or by a fam-ly with similar lifestyle and procedures for controlling the indooremperatures.

No serious study has been done so far related to thermal mon-toring or indoor temperature prediction of the vernacular housesf northeastern India. In this study, naturally ventilated housesre considered for detailed monitoring. Selection of vernacularouses for long-term monitoring is based on common building plannd functionality [6]. In this study, two fully operational naturallyentilated vernacular houses are considered. One is in warm andumid climate (Tezpur) and the other one is in cold and cloudy cli-ate (Cherrapunjee) [12]. Houses considered for the study have

umber of solar passive as well as climate oriented design fea-ures. All these features are elaborated in later sections of thisaper.

This paper presents a comparative study based on long-termhermal monitoring of these selected houses at two different cli-

ates. Long-term monitoring and extensive interaction work withhe occupants of the selected houses is carried out to record the dif-erent climatic parameters like temperature, humidity and indoorighting level and other behavioral patterns that affect the function-ng of the buildings. Based on the collected data during long-term

onitoring, mathematical formulae are developed for each climaticone to predict the indoor temperature of the buildings. Indooremperature profile is important to understand the thermal behav-or of building. This is also one of the prime factors which influencehe comfort status inside the buildings [8,9]. Since, it is practi-ally impossible to record indoor temperature in every house, theseeveloped predictive mathematical formulae can be used to predicthe indoor temperature of the house of similar kind and activi-ies with fair accuracy in the same climatic zone [9–11,13]. Finallycomparative study has done based on these predictive mathe-atical formulas on both the buildings at two different climatic

ones.

. Climatic conditions and vernacular houses in North-Eastndia

North-East India is classified into three bioclimatic zones;arm and humid, cool and humid and cold and cloudy [12]. Twoernacular buildings are selected based on common layout andunctionality [6]. One house is located at Tezpur, Assam in warmnd humid climate and other one is at Cherrapunjee, Megha-aya in cold and cloudy climate (Fig. 1). The latitude of Tezpurs 26◦37′N and longitude is 92◦47′E and for Cherrapunjee lati-ude is 25◦17′N and longitude is 91◦44′E [12]. Most of the housesf the region are constructed in direct response to the climaticonstrains. Houses of each climatic zone are distinct in its builtorm. Building materials and their processing is also different inach climatic zone. Baked bricks, processed mud, bamboo (sand-iched between two layers of processed mud), cane and wood,

tone chips and rock slabs are the main building materials [6].ud processing is done by adding beaten straw, chopped jute and

ime. All the building materials used to construct the vernacularouses in this region are available locally. This provides an edgen environmental front as less energy is involved in processingnd transportation and henceforth minimal environmental degra-ation [6].

Fig. 1. Tezpur and Cherrapunjee location in bioclimatic zone.

3. The monitored houses and the families in Tezpur and inCherrapunjee

In this section, the selected house for monitoring at Tezpur(warm and humid climate) and at Cherrapunjee (cold and cloudyclimate) is described in detail including construction materials andtheir thermo-physical properties. Since the occupants tend to mod-ify the indoor environments by their adaptive actions to restorecomfort, details of family members and list of their probable adap-tive actions taken are also reported.

3.1. The monitored house and the family in Tezpur

The vernacular house selected for long-term monitoring is basedon common building plan and functionality of this climatic zone.This is a single storey house and five family members are livingin this house; two males, two females and one female child. Twomales are of age 65 and 37 years and females are of age 60 and 32years. Female child is 9 years old.

The house is constructed in such a way that it has open space inall the four directions and the external walls are exposed to ambientair. Ventilators are constructed just above the windows but belowthe ceiling. The house has inclined roof and extended on both sides.Extended roofs protect the wall during heavy rainfall and also func-tion as overhang. Roof is constructed of galvanized tin sheet. Dueto tin roofing, all the houses have false ceiling made up of eitherwood or asbestos sheet. False ceiling separates the living space frombeing directly exposed to roof. Ceiling also plays important role inminimizing the heat gain during summer and loss during winter.

The building built area is 94 m2 and the house is oriented inNE–SW direction. Openings in the form of windows and ventilatorsare evenly distributed throughout the exterior wall of the house.Window to wall ratio is 0.216 of this house. Net opening in theform of windows, ventilators and doors is about 50% of the floorarea [14]. Windows and doors of all other rooms are made up ofwood without any glazing. All the rooms has ventilators and are

single glazed.

Building external and inter-room partition wall (plaster, brickand plaster) thickness is 13.4 cm. No insulation is present at theinside surface of the external walls. Over all heat loss coefficient

1612 M.K. Singh et al. / Energy and Build

Table 1Building envelop characteristics of Tezpur and Cherrapunjee.

Envelope Overall heat loss coefficient (U value) (W/m2 K)

Tezpur building Cherrapunjee buildings

External wall 3.004 3.654Internal wall 2.443 2.633

oifao

cdaadtrkwTmw

3

bmaoasmd

This section discussed in detail about the long-term mon-

TH

TH

Floor 3.504 0.134Roof 5.637 5.637False ceiling 2.838 2.812

f the building envelope is presented in Table 1 [15]. Data loggers placed on the inter-room partition wall and at a height of 1.5 mrom the ground. The data logger is placed at the centre of the housend it is assumed that it will represent the temperature distributionf the building with a fair accuracy.

The monitoring work related to adaptations like opening andlosing of windows, switching on and off of ceiling fans and otheretails were not possible to record on hourly basis due to limitedccessibility inside the house. However, from the extensive inter-ction with the residents, most probable adaptation actions takenuring each season of the year are listed in Table 2. Most of theime, during the process of monitoring, building kept under freeunning. In the winter season, windows in the SE direction wereept open to receive morning sun to maximize solar heat gain andere closed in the evening and night time to reduce the heat loss.

hough for heating in winter, portable radiator heating arrange-ent was there but was seldom used. In summer, only ceiling fanas used for cooling.

.2. The monitored house and the family in Cherrapunjee

The selected house is oriented in SE–NW direction with totaluilt area of 44 m2. This house is constructed on south slope of theountain. The house has two female members; age of 45 years

nd other is 14 years. House is elevated from the ground by about

ne meter. The entire external wall of the house is exposed toir. The NE side wall faces the mountain. Windows and doors aremall in dimension in comparison to that of warm and humid cli-ate. Net opening is about 30% of the floor area in the form of

oors, windows and ventilators. Windows to wall ratio for both

able 2ouse daily use description at Tezpur.

Seasons Descriptions

All the time The house is occupied mainly by female memberdaytime. Female child goes to school during dayt

Winter Windows are opened during daytime and closedoperation of ceiling fan.

Pre-summer Windows are opened during daytime and closedrequirement. Ceiling fan operated as per requirem

Summer Windows are opened during daytime and in nighoperated.

Pre-winter Windows are opened during daytime and in nighupon requirement. Ceiling fan operated as per req

able 3ouse daily use description at Cherrapunjee.

Seasons Descriptions

All the time The house mostly remains unoccupied from mornschool.

Winter Windows are opened during daytime and closed iPre- summer Windows are opened during daytime and closed

glass which improves the illumination level insidSummer Windows are opened during daytime and in evenPre-winter Windows are opened during daytime and closed

ings 42 (2010) 1610–1618

the rooms is same and is about 0.108 [14]. Doors are completelywooden made but windows and ventilators are wooden framed sin-gle glazed. It is observed from the sun path that bedroom receivessun throughout the afternoon. Floor of living room is made up ofwooden plank and the floor of veranda, kitchen and bathroom isconcrete.

Building is constructed by using locally available material likewood, stone slabs, cane and bamboo. Roof of the house is inclinedon two sides and made up of galvanized tin sheet. False ceiling ofthe living room is made up of bamboo mat. Bamboo mat is porousin nature and increases the heat gain during the day time. Thick-ness of the external wall (plaster, rock slab, plaster) is 16 cm andthe thickness of the inter-room partition wall (plaster, woven bam-boo, plaster) is 4 cm. Table 1 presents the overall heat transfercoefficients of the building envelop [15]. Sensor with data loggeris placed on the inter-room partition wall of living room at theheight of around 1.5 m from the floor and assumed that temper-ature recorded at this point will be the average temperature ofdifferent points in the house.

An extensive interaction with the residents of the house hasbeen done to record the different adaptation processes which makethe living condition comfortable. Table 3 represents the most prob-able action taken by the people to control the indoor environmentduring each season of the year. Opening/closing of window andchanging clothing levels are the prominent adaptation actions. Dur-ing winter residents burned wood charcoal in a metal pan to keepthe indoor environment warm. This makes the indoor environmentwarm but results in headache and coughing due to smoke from un-burnt fuels. Throughout the monitoring work, this building is keptunder free running mode.

4. Monitoring conditions in the vernacular house in Tezpurand in Cherrapunjee

itoring work for 25 days in all the seasons (January: winter,April: spring/pre-summer, July: summer/monsoon and October:autumn/pre-winter) of the year 2008. It also discussed in detailsabout the measuring instrument precession and the details of mon-itoring parameters.

s from 9:00 to 19:00 h. As both the male members leave for work during theime.in evening and night time. But ventilators remain closed during entire season, no

in evening and night time. But ventilators are partially opened depending uponent.

t time subjected to requirement. Ventilators completely opened. Ceiling fan is

t time subjected to requirement. Ventilators are partially opened dependinguirement.

ing 7:00 to 14:00 h. One family member always went to work and other to

n evening and night time. But ventilators remain closed during the entire season.in evening and night time. Ventilators are never opened. They are fitted withe the built space.ing but closed in night time. Ventilators are completely closed.in evening and night time. Ventilators are completely closed.

d Buildings 42 (2010) 1610–1618 1613

4

cfyasa

pofJTRissfalTpgamma

4C

pFFmwlptoitarpd

worwir

5a

tRpz

M.K. Singh et al. / Energy an

.1. Long-term monitoring in the vernacular house in Tezpur

Fully functional vernacular house at Tezpur (warm and humidlimate) was selected for long-term monitoring. The selected houseor this study was kept under normal operation throughout theear. No restrictions were imposed on the residents and weredvised to carryout their as usual living conditions. This was neces-ary to get the monitoring results close to real situation. The indoornd outdoor data are used for thermal monitoring study.

Long-term monitoring work includes the measurements of tem-erature (inside and outside house), relative humidity (inside andutside house) and illumination level (inside and outside house)or 25 days in all the seasons (January: winter, April: pre-summer,uly: summer/rainy and October: pre-winter) of the year 2008.hese all parameters are measured through data loggers (HOBOH/Temp/Light/External Data Logger, USA). Monitoring is done

n winter from 8th January to 3rd February 2008, in spring/pre-ummer from 5th April to 29th April 2008, in summer/monsooneason from 7th July to 31st July 2008 and in autumn/pre-winterrom 13th October to 7th November 2008. The temperature sensorccuracy is ±0.7 ◦C, the humidity sensor accuracy is ±5% RH and theight intensity measurement accuracy is ±2 lumen/ft2, respectively.his work is also followed by extensive interaction with the occu-ants in all the seasons to know about the various adaptations theyo through to make themselves comfortable. The measured datare used to evaluate the thermal performance of the houses. Theseeasured data are used to develop the predictive mathematical for-ula. The indoor minimum, average and maximum temperatures

re predicted based on this mathematical formula.

.2. Long-term monitoring in the vernacular house inherrapunjee

Functional vernacular architecture based on common buildinglan and functionality was selected for long-term monitoring work.or the winter, the monitoring was done from 12th January to 5thebruary, in spring/pre-summer from 6th April to 30th April, inonsoon/summer from 8th July to 1st August and in autumn/pre-inter from 14th October to 8th November of the year 2008. Data

ogger is installed both inside and outside of the house for the aboveeriod. Data is recorded at the interval of 30 min. Other parametershat affect the thermal performance of the house are measuredr recorded like external wall thickness, internal wall thickness,nter-room partition wall thickness, false ceiling height, construc-ion materials, color of envelope, dimensions of doors, windowsnd ventilators as well as numbers. From the interaction with theesidents, various adaptations processes were recorded. Thermalerformance of the house is evaluated by using these measuredata.

In the month of October, due to illness of a family member, thereas a restriction in opening and closing of the windows. Windows

pened for very small time and that too in the mid of the day. Andest of the time it was remained closed. But in normal time, theindows remained opened throughout the day. Burning of wood

n a metal pan inside the house is a very common practice in Cher-apunjee for room heating in winter days.

. Thermal monitoring of the vernacular building in Tezpurnd in Cherrapunjee

In this section, the measured temperature data is plotted againstime to study the variation of outdoor and indoor temperatures.elationship between outdoor temperature swing and indoor tem-erature swing is discussed for all the 4 months for both the climaticones.

Fig. 2. Measured indoor and outdoor temperatures in January in Tezpur.

5.1. Thermal monitoring of the vernacular building in Tezpur

Thermal performance study is one of critical aspects of vernac-ular houses [14,16]. Though these structures are evolved throughgenerations, addressing the climate constraints and needs of resi-dents, but still in-depth study of thermal behavior is a necessary.Thermal performance study is carried out based on the temperaturedata recorded both inside and outside of the selected house. Fig. 2represents the temperatures profile for indoor and outdoor temper-atures for the month of January of the year 2008. It can be concludedfrom this temperature profile, that there are two distinct behaviors.From the first day of monitoring to 14th days and other is after 14thday to 25th day of monitoring. In the first part, there is a gradualincrease in both outdoor and indoor temperatures. However, in thesecond part there is an abrupt fall in outdoor and indoor temper-atures. This happens due to rainfall and cloudy weather after 14thday of monitoring. This points also justified by the recorded lowillumination level. In the first part, the average outdoor tempera-ture swing is 10–11 ◦C and indoor temperature swing is 4–5 ◦C withoutdoor day average temperature of 18 ◦C and indoor day averagetemperature of 18.5 ◦C. In the second part, the outdoor tempera-ture swing is 7–8 ◦C and indoor temperature swing is 3–4 ◦C withoutdoor day average temperature of 15 ◦C and indoor day averagetemperature of 16 ◦C. From the temperature profile, it is observedthat the minimum temperature inside the house is always recordedin the early hours of subsequent day. It is interesting to note thatthe said minimum temperature is the effect of the previous day.Indoor maxima lie on decreasing outdoor temperature curve andindoor minima lie on the increasing outdoor temperature curve.This provides an extra time to smoothen the indoor temperatureswing.

Temperature monitoring work for pre-summer season for theyear 2008 is carried out from 5th April to 29th April. From 5th to11th April, there is an increasing trend in indoor temperature butthis not the same with outdoor temperature. On 9th and 11th April,it is observed a decrease in outdoor temperature and also increasein minimum temperature. This effect stores the heat in the build-ing and capacitive effect results in increase in indoor maximumand minimum temperature on 11th April despite of decrease inoutdoor temperature. The effect of cloudy weather is seen on 12th,13th, and 14th, April. This leads to drastic reduction in both indoor

and outdoor temperature. For both January and April month, theoutdoor minimum temperature is the deciding factor for the nextday indoor maximum and minimum. From 5th April to 12th April,the outdoor temperature swing were 10–15 ◦C with day average

1 d Build

o5t23

m2wepsmitsia

O2omtpwwtTta

5C

sosFawiii

F

614 M.K. Singh et al. / Energy an

f 27.7 ◦C but for the same period indoor temperature swing was–6 ◦C with day average of 27.4 ◦C. From 13th April to 29th April,he outdoor temperature variation was 8–10 ◦C with day average of4.2 ◦C and for the same period indoor temperature variation was–4 ◦C having day average 25 ◦C.

Indoor and outdoor temperature monitoring work for sum-er/rainy season is carried out from 7th to 31st July of the year

008. Since this is summer month and also monsoon season, theeather is cloudy and the place experiences heavy rainfall. The

ffect of cloudy weather is easily observed in the profile. From thelot, it can also be concluded that the indoor maximum is not thatensitive to outdoor minimum as compared to January and Aprilonth for the same place. For the monitored period outdoor max-

mum varies from 27.5 to 38.8 ◦C and minimum varies from 25.2o 29 ◦C with a swing of 9–10 ◦C and day average of 29 ◦C. For theame period, indoor maximum varies from 28 to 34 ◦C and min-mum varies from 26.7 to 31 ◦C with a swing of 4–5 ◦C and dayverage of 29.4 ◦C.

Monitoring work for pre-winter month is carried out from 13thctober to 7th November 2008. Outdoor maximum varies from0 to 32 ◦C and minimum varies from 18.3 to 21 ◦C with a swingf 9–10 ◦C and day average of 24.4 ◦C. For the same period indooraximum varies from 22.1 to 29 ◦C and minimum varies from 22

o 25.7 ◦C with a swing of 3–4 ◦C and day average of 25.4 ◦C. In thisrofile, it is observed that on 25th, 26th and 27th October, thereas a sudden fall in outdoor and indoor temperature due to cloudyeather and rainfall. Due to this, building looses stored heat and

he resultant effect can be seen on the subsequent monitored days.here is a shift of 2 ◦C (average) in outdoor and indoor tempera-ure. Table 4 represents the day average and swing of temperaturet Tezpur in the different months of the year 2008.

.2. Thermal monitoring of the vernacular building inherrapunjee

Fig. 3 presents the outdoor and indoor temperature profile of theelected house at Cherrapunjee from 12th January to 5th Februaryf the year 2008. First 9 days (from 12th January to 20th January)hows gradual increase in both outdoor and indoor temperature.or next 10 days (from 21st January to 30th January), there isgradual fall in outdoor and indoor temperature due to cloudy

eather and rainfall. Again from 31st January to 5th February,

t can be observed; there is an increasing trend in outdoor andndoor temperature. From the plot, it can also be concluded thatndoor minimum is sensitive to previous day outdoor minimum.

ig. 3. Measured indoor and outdoor temperatures in January in Cherrapunjee.

ings 42 (2010) 1610–1618

Monitored period is divided in two parts based on the temper-ature fluctuations observed in Fig. 3. From Table 4, it has beenfound that though the day average temperature of both parts forindoor and outdoor are 16.4, 13.7 ◦C and 15.2, 12.3 ◦C, respectivelybut have same temperature swing for indoor, i.e. 5–6 ◦C and foroutdoor 9–10 ◦C. Difference between outdoor day average tem-perature and indoor day average temperature shows the thermalstorage capacity of the building fabric. From the plot, it is alsoobserved that the outdoor maximum and indoor maximum tem-perature difference is less, which predicts that the building is usingmaximum available sunlight to increase heat gain inside the builtspace.

Indoor and outdoor temperature monitoring work for pre-summer season is carried out from 6th to 30th April 2008. Thevariation in the outdoor maximum and minimum temperature rep-resents the persisting weather. The monitored period is dividedinto two parts, from 6th April to 13th April and from 14th Aprilto 29th April, respectively. From Table 4, it has been observed thatthe indoor day average temperatures for both parts are higher thanthe day average outdoor temperature. Also for both parts of indoorand outdoor temperature swing are 5–6 and 10–12 ◦C, respec-tively. From the graph, it can be concluded that building is showingthermal capacitive effect. Indoor temperature is not that sensi-tive to outdoor temperature because difference between indoormaximum to that of outdoor maximum temperature and indoorminimum to that of outdoor minimum temperature has increased.

In summer/rainy season monitoring work is carried out from8th July to 1st August 2008. During this month the area has cloudyweather and receives heavy rainfall. Due to overcast, the outdoormaximum temperature remains below indoor maximum. Duringthis month the monsoon wind is also there and this leads to fur-ther decrease in outdoor temperature. During this period, sincewindows and doors were remained closed and cooking was doneon chullah by burning fire wood, internal heat gains increases. Foranalysis, the collected data are divided into two parts, from 8th Julyto 24th July (part I) and from 25th July to 1st August (part II), respec-tively. From 8th July to 24th July indoor and outdoor day averagetemperatures are 23.4, 22.5 ◦C, and temperature swing is 2–3 ◦C,respectively. From 25th July to 1st August (clear days) indoor andoutdoor day average temperatures are 25.8, 24.9 ◦C and tempera-ture swings are 4–5, 10–11 ◦C, respectively.

Temperature monitoring work for pre-winter season is carriedout from 14th October to 8th November 2008. The recorded periodcan be divided into two periods. One before 24–27th October andother after 27th October. First period ranges from 14th October to23rd October. During this period, the indoor maximum is alwayshigher than the outdoor maximum. So it can be concluded, thatthe heat gain during the day time in this period is much higherthan the loss at night time. As explained earlier, that during thesame period the normal functioning of the house is affected. On24–27th October, it has been observed a drastic fall in outdoor andindoor temperature due to cloudy weather and rainfall. Again from28th October to 8th November, building shows higher heat gain.But this gain is less in comparison to the first period. During thisperiod building is showing thermal capacitive effect. From the plot,it can conclude that the indoor minimum is strongly related tooutdoor minimum. From 14th October to 23rd October, the out-door temperature swing is 6–7 ◦C with day average temperature of23.4 ◦C but for the same period indoor temperature swing is 6–7 ◦Cwith day average temperature of 26.1 ◦C. Again from 24th Octo-ber to 8th November, the outdoor temperature variation is 7–8 ◦C

with day average temperature of 20.3 ◦C and for the same periodindoor temperature variation is 7–8 ◦C and the day average tem-perature is 21.9 ◦C. Table 4 represents the day average and swingof temperature at Cherrapunjee in the different months of the year2008.

M.K. Singh et al. / Energy and Buildings 42 (2010) 1610–1618 1615

Table 4Day average temperature and swing at Tezpur and Cherrapunjee.

Place Month Day average temperature (◦C) Temperature swing (◦C)

Indoor Outdoor Indoor Outdoor

Part I Part II Part I Part II Part I Part II Part I Part II

Tezpur

January 18.5 16.0 18.0 15.0 4–5 3–4 10–11 7–8April 27.4 25.0 27.7 24.2 5–6 3–4 10–15 8–10July 30.1 28.8 30.2 29.5 4–5 3–4 9–10 9–10October 26.6 24.0 26.3 23.4 3–4 3–4 9–10 9–10

6t

efowitttbtmtvgtmspast

sptswm

Fi

Cherrapunjee

January 16.4 13.7 15.2April 24.9 25.2 24.0July 23.4 25.8 22.5October 26.1 21.9 23.4

. Procedure for mathematical modeling of the indooremperatures

Indoor temperature prediction for particular building in differ-nt climatic zones, based on outdoor climatic parameters is possibleor occupied free running building. These predictions are devel-ped through simple formulae by multiple regressions techniques,hich relate indoor daily maximum, average, minimum and other

ndependent variables. In developing these formulas for predictinghe indoor daily maximum, average, and minimum temperatures,he first issue, with respect to each one of these indoor parame-ers, is to find out which parameter of the outdoor climate couldest serve as a basis for prediction. This involves the analysis ofhe patterns of the relationship between the daily outdoor maxi-

um, average and minimum temperatures, as a set, and the indooremperature parameter of interest. This analysis can be performedisually by plotting the indoor parameter of interest over the back-round of the outdoor daily maximum, average and minimumemperatures. Once this pattern is observed, it is a relatively simple

atter to express it in a formula. The constants of the formulae arepecific to particular building incorporating the thermo-physicalroperties. Since the houses considered for the study are functionalnd occupied so the constants are also specific to family, as per-onal management of the house has a significant effect on indooremperatures.

Analysis of the plot of the indoor maximum in Tezpur (Fig. 4)hows that the relationship between the indoor maximum tem-erature and the outdoor maximum is affected also the changes in

he average of the outdoor conditions (avg). This factor may repre-ent changes in the “management” of the building by the occupantsith the changes in the weather. In dealing with the indoor mini-um temperature, it was found, by a similar plot, that the indoor

ig. 4. Indoor maximum and outdoor climatic conditions in Tezpur during the mon-toring and the ‘Periods’ average temperature.

12.3 5–6 5–6 9–10 9–1024.4 5–6 5–6 10–12 10–1224.9 2–3 4–5 2–3 10–1120.3 6–7 7–8 6–7 7–8

minimum temperature of the given day is affected by the drop inprevious day’s maximum temperature to the present day minimumtemperature and is presented by ‘Tdrop’.

Tdrop = Tmax(n − 1) − Tmin (1)

It has been observed from the weather data that in both theclimates relative humidity always assumes high values [12]. Soto adjust indoor temperature to achieve comfort in indoors reg-ulation of windows and ventilators (‘Vents’) plays an importantrole besides running ceiling fans in pre-summer, summer and pre-winter months. While doing the analysis, it is found that ventilationhas measurable effect only in warm and humid climate. So the termis included in the developed formula for the house in Tezpur only.

6.1. Mathematical modeling of the indoor temperatures in thebuilding of Tezpur

The generated formulas for the house in Tezpur is based on themeasured data of January and July and validated with the measureddata of April and October. Variable included in the generation of thepredictive formulae are:

• Outdoor maximum temperature (Tmax)• Outdoor daily average temperature (Tavg)• Outdoor temperature drop (Tdrop)• Outdoor minimum temperature (Tmin)• Diurnal temperature swing (Tmax − Tmin) (Swg)• Period’s outdoor average (Period)• Ventilators (Vent; Closed = 0, Semi open = 1 and Fully open = 2)• Temperature drop from previous outdoor maximum; Tmax (n − 1)

to current minimum Tmin

The indoor temperature formulae for maximum, averageand minimum temperatures are generated for January (win-ter), April (spring/pre-summer), July (summer/monsoon), October(autumn/pre-winter) months. With respect to the indoor max-imum, the relevant parameters are found to be the outdoormaximum (Tmax), the outdoor average (Tavg), the period’s average(Period) and the operation of the fans (Vent). The predictive formulafor indoor maximum temperature is given by the Eq. (2).

maximum = 1.2 + 0.2042 × Tmax + 0.2802 × Tavg + 0.5193

× Period − 0.44 × Vent (2)

The correlation coefficients (CC) for the maximum temperature

in the various months are: January CC = 0.8823; April CC = 0.7672;July CC = 0.7266 and October CC = 0.9332, respectively. The correla-tion coefficient for the whole monitoring period is CC = 0.9748.

With respect to the indoor average temperature, the relevantparameters are the outdoor average (Tavg), the diurnal swing (Swg),

1616 M.K. Singh et al. / Energy and Buildings 42 (2010) 1610–1618

Ft

tTt

a

pCw

mmpTt

m

pCw

ia

6b

rttijw(mpmt

The correlation coefficients (CC) for the average temperature inthe various months are: January CC = 0.9807; April CC = 0.9530; JulyCC = 0.9687 and October CC = 0.9720 and for the whole monitoringperiod CC = 0.9712.

ig. 5. Agreement between the measured and the computed maximum tempera-ures in the house in Tezpur.

he periods’ average (Period) and the operation of the fans (Vent).he predictive formula for average indoor temperature is given byhe equation 3.

verage = 1.8 + 0.5388 × Tavg + 0.1021 × Swg + 0.4178

× Period − 0.10 × Vent (3)

The correlation coefficients (CC) for the indoor average tem-erature in the various months are: January CC = 0.9168; AprilC = 0.7217; July CC = 0.6509 and October CC = 0.9637 and for thehole monitoring period CC = 0.9741.

With respect to the indoor minimum temperature, another cli-atic feature has an impact, the drop from the previous days’aximum to the current minimum (Tdrop). The relevant input

arameters are the outdoor minimum (Tmin), Tdrop, Period and Vent.he predictive formula to calculate the indoor minimum tempera-ure is given by Eq. (4).

inimum = 1.9 + 0.7405 × Tmin + 0.1823 × Tdrop + 0.1811

× Period + 0.1392 × Vent (4)

The correlation coefficients (CC) for the indoor minimum tem-erature in the various months are: January CC = 0.9338; AprilC = 0.7731; July CC = 0.8325 and October CC = 0.9848 and for thehole monitoring period CC = 0.9892.

The agreement between the measured and the computed max-mum, average and minimum temperatures in the house in Tezpurre shown in Figs. 5–7, respectively.

.2. Mathematical modeling of the indoor temperatures in theuilding of Cherrapunjee

In the month of October one person of the family was ill. As aesult, the daytime temperatures are at higher temperatures duringhe illness period, relative to the outdoor temperatures in compareo the other months. This factor has been included in the modelingn terms of ‘Ill’. The generated formulas for the house in Cherrapun-ee is based on the measured data of January and July and validated

ith the measured data of April and October months. Following Eqs.

5)–(7) represents the predictive formula to predict the daily indoor

aximum, average and minimum temperatures for this case. In therocess of developing these formulae with respect to the indooraximum temperature, the relevant parameters are found to be

he outdoor maximum (Tmax), the outdoor averages (Tavg), outdoor

Fig. 6. Agreement between the measured and the computed average temperaturesin the house in Tezpur.

temperature swing (Swg) and the factor ‘Ill’. The indoor maximumtemperature can be calculated by using the equation 5.

maximum = 1.3 + 0.1294 × Tmax + 0.8518 × Tavg + 0.2462

× Swg + 3 × Ill (5)

The correlation coefficients (CC) for the maximum temperaturein the various months are: January CC = 0.9765; April CC = 0.9039;July CC = 0.9827 and October CC = 0.9472 and for the whole moni-toring period CC = 0.9848.

With respect to the indoor average temperature, the relevantparameters are the outdoor maximum (Tmax), outdoor average(Tavg), the diurnal swing (Swg) and the factor ‘Ill’. Eq. (6) representsthe equation using which the average indoor temperature can becalculated of the house at Cherrapunjee.

average = 1.7 + 0.0715 × Tmax + 0.8910 × Tavg − 0.0316

× Swg + 2 × Ill (6)

Fig. 7. Agreement between the measured and the computed minimum tempera-tures in the house in Tezpur.

M.K. Singh et al. / Energy and Buildings 42 (2010) 1610–1618 1617

Ft

fcrTm

m

iJt

mC

7

tEcmtw

Fi

ig. 8. Agreement between the measured and the computed maximum tempera-ures in the house in Cherrapunjee.

With respect to the indoor minimum temperature, the climaticeature Tdrop (the drop from the previous days’ maximum to theurrent day’s minimum) is found to have an important impact. Theelevant input parameters are the outdoor minimum (Tmin) anddrop only. Following Eq. (7) can be used to calculate the indoorinimum temperature.

inimum = 2.2 + 1.015 × Tmin + 0.10606 × Tdrop (7)

The correlation coefficients (CC) for the minimum temperaturen the various months are: January CC = 0.9496; April CC = 0.7908;uly CC = 0.8588 and October CC = 0.9695 and for the whole moni-oring period CC = 0.9822.

The agreement between the measured and the computedaximum, average and minimum temperatures in the house in

herrapunjee are shown in Figs. 8–10, respectively.

. Discussions

This study is carried out to understand the thermal behavior ofhe vernacular houses at two different bioclimatic zones of North-

ast India. Two vernacular houses, one each from warm and humidlimate and cold and cloudy climate were selected for long-termonitoring based on the functionality and repeatability. These

ypes of layout and structure are still very popular and are beingidely constructed. Monitoring work of both the houses at Tezpur

ig. 9. Agreement between the measured and the computed average temperaturesn the house in Cherrapunjee.

Fig. 10. Agreement between the measured and the computed minimum tempera-tures in the house in Cherrapunjee.

and Cherrapunjee has been carried out simultaneously in all thefour seasons of the year 2008. From the temperatures profiles, itcan be concluded that, for each month the outdoor temperaturedata can be divided into two parts. First part has higher averagetemperature than the later part. For winter and pre-winter monthsthe difference between average temperatures of two parts is 3 ◦Cbut for pre-summer and summer months it is about 1.5 ◦C. For allthe months, the indoor average is always higher than outdoor aver-age. Similarly from the outdoor and indoor temperature profile ofCherrapunjee house, it can be concluded that outdoor temperatureis very much fluctuating with sharp changes in outdoor conditions.For this house, for all the months the indoor temperature average isalways higher than outdoor average temperature. For the month ofOctober, it is observed that indoor temperature always higher thanthe outdoor temperature. This happens because one relative of theresidents was ill. So during the monitoring time, the movement aswell as the normal functioning of window was restricted and alsothe temperature of the room was kept intentionally higher thanoutdoor temperature. This is also reflected in the formulae gen-erated to predict indoor temperatures. The building constructionmaterial has high thermal mass and has restricted the indoor tem-perature swing by almost 50% to that of outdoor temperature. Dueto this, the indoor temperature fluctuation is smoothened. But theeffect of thermal mass is diluted because the buildings are naturallyventilated and are low on insulation level.

The predictive formulae are developed, based on the measureddaily indoor maximum, average and minimum temperatures. Theformulae are then validated against measurements taken indepen-dently in different time periods. In general, a fairly good agreementexisted between the onsite measurements and the formulae repre-senting the indoor maximum, average and minimum temperatures.In the predictive formulae for the house at Tezpur, the variable“Period” and “Vent” are important input parameters in all the cases.From these formulae, it can be concluded that building is workingas memory and its present day indoor environment is the com-bined effect of previous 2–3 days indoor environment. It can also beconcluded that occupants adaptive actions like opening/closing ofwindows, opening/closing of ventilators, opening/closing of doorsand switching on/off fans play important role in modifying indoorthermal environment. The other important input parameter “Tdrop”has measurable effect on the indoor daily average and maximum

temperatures representing mainly the rate of nocturnal radiantcooling. For all the three formulae, it is found that the correlationcoefficient (R2 value) is more than 0.96 for the entire monitoringperiod. This represents a strong positive correlation between vari-

1 d Build

aag

viwptafctmcpfd

iwppi‘itnuamT

8

ailaose

pvzcttrmhy

rtdcdr

mof

[

[

[

[

[

[

[

[

[

[

618 M.K. Singh et al. / Energy an

bles of the formula. From Figs. 5–7, it can conclude that fairly goodgreement exists between data of onsite measurements and dataenerated by using the developed predictive formulae.

In the predictive formulae for the house at Cherrapunjee, theariable “Swg” and variable “Ill” are very important input. The build-ngs of this zone uses the openings judiciously to keep the indoor

arm and remain closed most of the time, so ventilation is not asrominent as it is in Tezpur house (warm and humid climate). Buthe adaptive action is again taken care by the constant and the vari-ble “Ill” in the formulae. Indoor minimum temperature predictiveormula uses the variable “Tdrop”. From the temperature profile, itan be concluded that the present day minimum temperature ishe effect of drop in previous day maximum to present day mini-

um. Again in all the three formulae, it is found that the correlationoefficient (R2 value) is more than 0.97 for the entire monitoringeriod. This fact is well supported by Figs. 8–10 and represents aairly good agreement between data of onsite measurements andata generated by using the developed predictive formulae.

The use of predictive formulae, such as those presented here,s meant for indoor temperature predictions of a given building,

ith a given geometry and building materials, concerning differenteriods of the year or even exposed to different climatic conditions,rovided that the building can be operated in a similar way as dur-

ng its monitoring. These are also specific to a particular family, aspersonal management’ of the house has a significant effect on thendoor temperatures in unconditioned buildings. In the formulae,he thermal mass of the building are not used as the building areaturally ventilated with the provision of cross ventilation contin-ously throughout the day. It is also observed that there is a goodgreement exists between the measured and computed averageaximum, average and average minimum temperatures for both

ezpur and Cherrapunjee.

. Summary

Two vernacular houses, one each from warm and humid climatend cold and cloudy climate were selected for long-term monitor-ng. The selected houses are single storey and are constructed usingocally available materials. House at Tezpur has exposed wall tombient air from all the four directions. Net opening in the formf windows, ventilators and doors is about 50% of the floor area,tating the concern for ventilation to overcome high humidity bynhancing indoor air movement.

The house in Cherrapunjee has almost 50% of built space com-ared to the Tezpur house. The house is compact (low surface toolume ratio) and net opening is about 30% of floor area. In thisone, opening/closing of window, opening/closing of doors andhanging clothing levels in summer and burning wood charcoalo keep indoor warm in winter are the prominent adaptive actionso restore comfort inside the houses. Ventilators are there but theyemained closed throughout the year. For better comparative study,onitoring work of both the houses at Tezpur and Cherrapunjee

as been carried out simultaneously in all the four seasons of theear.

During the monitoring period extensive interaction with theesidents has been done. From the interaction, it has been foundhat numbers of adaptive actions like opening and closing of win-ow, opening and closing of ventilators, switch on/off fans andhanging clothing level etc. are being in use by the occupants atifferent time of the year to modify the indoor environment for

estoring comfort.

Predictive formulae for indoor maximum, average and mini-um temperatures were developed, based on the measured daily

utdoor conditions during part of the monitoring period. Theormulae are then validated against measurements taken indepen-

[

[

ings 42 (2010) 1610–1618

dently in different time periods. In general, a fairly good agreementexisted between the onsite measurements and the formulae repre-senting the indoor maximum, average and minimum temperatures.

It should be emphasized that the methodology for generatingthe formulas used in this paper cannot be generalized. It is applica-ble, and can be used, only for the specific building, occupied by thesame family, or by a family with similar lifestyle and procedures forcontrolling the indoor temperatures in a passive building. Whensuch an occupied building is monitored for a given period, andthe data is analyzed by this methodology, it is possible to predictthe thermal performance of that building under different climaticconditions. Under these conditions it was successfully tested withmany families in different locations and with different types ofbuildings [8–13,17–21].

The paper, as well as the formulas and the methodology bywhich they were generated, did not intend to suggest ways forimproving indoor comfort. The application might be when a specifictype of building is developed for mass production and intended forconstruction in different climatic regions. It is then possible to testthe performance of this building type, when occupied, in one loca-tion and to be able to estimate its performance, occupied by similarfamilies, in other locations with different climatic conditions.

References

[1] C. Helena, Chapter 4: Bioclimatism in vernacular architecture, Renewable andSustainable Energy Reviews 2 (1/2) (1998) 67–87.

[2] C. Gallo, Bioclimatic architecture, Renewable Energy 5 (5–8) (1994) 1021–1027.[3] V. Olgyay, Design with Climate: A Bioclimatic Approach to Architectural

Regionalism, Van Nostrand Reinhold, New York, 1992.[4] M. Milne, B. Givoni, Architectural design based on climate, in: D. Watson (Ed.),

Energy Conservation through Building Design, McGraw Hill, New York, 1979,pp. 96–113.

[5] A. Sayigh, H. Marafia, Chapter 1: Thermal comfort and the development of bio-climatic concept in building design, Renewable and Sustainable Energy Reviews2 (1/2) (1998) 3–24.

[6] M.K. Singh, S. Mahapatra, S.K. Atreya, Bioclimatism and vernacular architectureof North-East India, Building and Environment 44 (5) (2009) 878–888.

[7] G.S. Brager, R.J. de Dear, Thermal adaptation in the built environment: a liter-ature review, Energy and Buildings 27 (1) (1998) 83–96.

[8] B. Givoni, Effectiveness of mass and night ventilation in lowering the indoordaytime temperatures. Part I: 1993 experimental periods, Energy and Buildings28 (1) (1998) 25–32.

[9] E. Kruger, B. Givoni, Predicting thermal performance in occupied dwellings,Energy and Buildings 36 (3) (2004) 301–307.

10] E. Kruger, B. Givoni, Outdoor measurements and temperature comparisons ofseven monitoring stations: preliminary studies in Curitiba, Brazil, Building andEnvironment 42 (4) (2007) 1685–1698.

11] E. Kruger, B. Givoni, Thermal monitoring and indoor temperature predictionsin passive solar building in an arid environment, Building and Environment 43(11) (2008) 1792–1804.

12] M.K. Singh, S. Mahapatra, S.K. Atreya, Development of bio-climatic zones innorth east India, Energy and Buildings 39 (12) (2007) 1250–1257.

13] B. Givoni, A.F. González, Measured and compounded indoor maximum aver-age and minimum temperatures of passive solar system, in: S.K. Dube, T.Muneer, G.N. Tiwari (Eds.), Third International Conference on Solar Radiationand Day Lighting. Solaris 2007, vol. I, Anamaya Publishers, New Delhi, 2007, pp.432–447.

14] M.K. Singh, S. Mahapatra, S.K. Atreya, Thermal performance study and eval-uation of comfort temperatures in vernacular buildings of North-East India,Building and Environment 45 (2) (2010) 320–329.

15] J.K. Nayak, J.A. Prajapati, Handbook on Energy Conscious Buildings, Solar EnergyCentre, Ministry of Non-conventional Energy Sources, Government of India,New Delhi, 2006.

16] F. Nicol, Comfort and energy use in buildings—getting them right, Energy andBuildings 39 (7) (2007) 737–739.

17] B. Givoni, F. Vecchia, Predicting thermal performance of occupied houses, in:Proceedings, PLEA 2001, Florianopolis, Brazil, November, 2001, pp. 701–706.

18] B. Givoni, E. Kruger, Temperature predictions: effect of thermo-physical prop-erties on the changes in the constants of the predictive formulas, in: PLEA’03,Santiago, Chile, 2003.

19] B. Givoni, E. Kruger, An attempt to base prediction of indoor temperatures of

occupied houses on their thermo-physical properties, in: PLEA’03, Santiago,Chile, November, 2003.

20] E. Gonzalez, B. Givoni, Testing and Modeling Evaporative Cooling Systems in aHot Humid Climate—Maracaibo, Int. Solar Energy Society, Orlando, FL, 2005.

21] B. Givoni, E. Kruger, Indoor temperature predictions in a passive solar buildingin an arid environment, in: ASES 2008, San Diego, CA, USA, May, 2008.