Energy and Buildings 39 (2007) 893–905

www.elsevier.com/locate/enbuild

On the relation between the energy and social characteristics

of the residential sector
M. Santamouris *, K. Kapsis, D. Korres, I. Livada, C. Pavlou, M.N. Assimakopoulos
Group Building Environmental Studies, Physics Department, University of Athens, 157 84, Panepistimioupolis, Athens, Greece
Received 5 November 2006; accepted 8 November 2006

Abstract
Social, financial, energy and technical data from about 1110 households have been collected during 2004 in the major Athens area. The sample
has been divided in seven income groups and a detailed analysis has been performed. Important conclusions have been drawn regarding the quality
of households, the operational conditions and the energy spent per income group. Low income people are more likely to be living in old buildings
with poor envelope conditions. The cost per person and unit area is much higher for the low income group for both heating and electricity. Fuel
poverty is quite high, especially when the actual oil prices are considered.
# 2006 Elsevier B.V. All rights reserved.

Keywords: Energy poverty; Energy consumption; Buildings; Income

1. Introduction consumption, income and the social and technical character-

Residential buildings are among the more important energy
consumption sectors. The annual energy consumption of the
sector in OECD countries is close to the 20% of their total
energy consumption [1]. It also presents a very strong
increasing trend. In the period 2000–2005 the increase was
close to 6.6%. Residential energy consumption in non-OECD
countries is much lower. In 2003, the residential energy
consumption per capita was almost six times higher in the
OECD countries than in the non-OECD countries, averaging
24.9 million and 4.1 million Btu per person, respectively [2].
Energy in the residential sector is used for heating and
cooling, lighting, hot water and in electric appliances. The
physical size of the residence, the climatic zone, the design
characteristics of the building, the income level of the
household and the available energy infrastructure are key
factors in determining the energy consumption in residences.
Knowledge of the energy and social characteristics of the
households is a key element for design policies for the
residential sector. Some relevant studies have been recently
carried out aiming to identify the relations between energy
* Corresponding author. Tel.: +30 2107276847.
E-mail address: msantam@phys.uoa.gr (M. Santamouris).
0378-7788/$ – see front matter # 2006 Elsevier B.V. All rights reserved.
doi:10.1016/j.enbuild.2006.11.001
istics of the residential sector [3–5].
Energy consumption of households is strongly related to the
family income. Earlier research has shown that the higher the
income, the higher the energy consumption of the household
[6–9]. The specific relation between the household income and
the energy consumption of homes varies from country to
country as a function of their economic and technical
characteristics [10–13].
Important research has been carried out regarding energy
expenditure in low income households. Fuel poverty is defined
by the UK government as the number of households who need to
spend in excess of 10% of income on energy services in order to
achieve a specified heating standard. When expenditure exceeds
20% of the income, the family is defined as suffering ‘severe
energy poverty’. Statistical data around Europe has shown that
fuel poverty reaches high levels in the South of Europe and
according to moderate calculations accounts for about 12% of the
households in Italy, 30% in Greece, 26% in Spain and 44% in
Portugal [14]. In parallel, fuel poverty in England alone totals
between 2.8 million and 3.9 million households [15], In Ireland,
estimations show that 17.6% of the households are energy poor
[16], around 226,000 houses. About 27% of the fuel poor houses,
around 4.7% of the total housing stock, are suffering from
chronic fuel poverty. Also, 12.7% of the households suffer from
intermittent levels of fuel poverty, i.e. occupants are occasionally
unable to heat their homes.
894 M. Santamouris et al. / Energy and Buildings 39 (2007) 893–905
Fuel poverty has important effects, mainly on human health
[17,18]. There is increasing evidence that lack of heating and
poor indoor environmental conditions in low income house-
holds cause serious health problems and increase mortality.
Low outdoor temperatures during the winter season are
associated with an annual average of 40,000 excess winter
deaths in the UK. This country, together with Ireland, presents
the highest rates of excess deaths in Northern Europe during
winter [16]. Some research has attributed this excess mortality
to poor standards of energy efficiency and resultant cold
housing in these countries [17,19].
The residential sector in Greece accounts for about 24.5% of
the total final consumption in the country and presents a high
increasing trend. In the period 1990–2002, the energy
consumption of the residential sector increased by 54%, while
the total energy consumption increased by 35% in the same
period [20]. In parallel, the energy consumption for space
heating, electrical appliances and lighting increased by 65%
since 1990, while space heating accounts for almost 72% of the
total energy consumption (3.43 Mtoe) [20]. Given that the
number of residential buildings has considerably increased
during the same period, it is calculated that the unit
consumption per dwelling has increased by 33%, with an
annual growth rate of 2.6%, and has risen from 0.98 toe/
dwelling in 1990 to almost 1.32 toe/dwelling in 2002. The
corresponding increase for space heating per unit, has been
30% while the corresponding consumption of appliances and
lighting has almost doubled, and has risen from 0.37 Mtoe in
1990 to 0.80 Mtoe in 2002, presenting an annual increase rate
close to 6.6% [20].
Table 1
Specific information and data collected through the questionnaire
Type of information
Group 1: general
Group 2: income
Group 3: energy system and operational conditions
Group 4: energy consumption
Group 5: quality of the envelope
The structure of residential energy demand has been studied
by many researchers [20–23]. Most of the studies focus mainly
on the analysis of the macroeconomic factors and the forecast
for the future energy consumption.
The present study aims to analyse the main characteristics of
the residential sector in Athens, Greece, and investigate the
inter-relations between the energy consumption, income of
households and the technical and social structure of the housing
sector. Data from about 1110 households have been collected
during 2004 in the greater Athens area and analysed. The
methodology followed as well as the main results of the study,
are discussed in the following section.
2. Data collection and organisation of the study
Information and data have been collected through interviews
with the members of the selected families and corresponding
inspections of each building. The information collected has
been organized into five main groups (Table 1):
 Group 1. General Information, involving data about the
location of the building, its floor area, the number of
occupants, the year of construction, the type of residence, the
number of storeys, and some other information about the
neighboring buildings.
 Group 2. Information on the annual income of each family.
 Group 3. Information on the operational schedules of the
building. In particular, on the set point temperature, the
heated and non heated part of the building, and on the mean
daily hours of heating.
Location of the building
Surface of the building
Year of construction
Type of residence Single storey
Two/three storey
Maisonette
Apartment
Other
Number of storey
Number of inhabitants per household
Number of other buildings or apartments in contact with
Existence of a pilotis
Monthly income of the family
Set point temperature (8C)
Percentage of spaces non heated
Mean daily hours of heating
Type of heating system
Monthly energy consumption for heating fuel
Monthly electricity consumption
Type of fuel used for heating
Frequency of the heating system maintenance (years)
Number and power of a/c systems
Type of glazing (single, double, other)
Type of insulation
 M. Santamouris et al. / Energy and Buildings 39 (2007) 893–905
 Group 4. Information on the energy consumption and the type
of fuels used; in particular, on the energy consumption for
heating and electricity uses, the type of fuel used for heating,
the heating system, the number and type of air conditioners
installed in the house and, finally, information on the
maintenance of the heating system. Electricity consumption
has been obtained directly from the bi-monthly electricity bills,
while oil consumption for the apartment buildings has been
obtained from the monthly invoices issued for each apartment
by the managers of the buildings. Finally, for the detached
buildings, the oil purchase invoices have been checked.
 Group 5. Information on the quality of the envelope and in
particular: on the type of glazing and the existence and type of
insulation.
All information is considered as fully confidential. A
database has been created and the necessary quality control has
been performed, while all extreme values have been excluded.
At the end, data from 945 buildings have been used. Based on
the distribution of the annual income, seven income classes
have been defined (Fig. 1), and specific analysis has been
performed for each income class in a comparative way. Income
classes have been defined using intelligent clustering analysis.
Clustering is a mathematical technique to classify numerical
data. It is based on the identification of sub-groups on a data set,
called ‘clusters’, where all objects are described by similar
characteristics. Every cluster contains a number of member
objects represented by given locations in the space and a center
defined as the point in the cluster space where the sum of the
specific distances of every member object belonging in the
cluster is minimized. Clustering techniques have been
extensively used to create natural groupings of data in energy
sciences. Fuzzy clustering is a quite new, ‘intelligent’ technique
that considers that each individual element in the data set
belongs to a cluster to some degree that is defined by a
membership function.
Fig. 1. Cumulative frequency distribution of the annual income for the whole
sample.
895
3. Results of the study
Energy consumption of households depends mainly on the
characteristics and the quality of the building and in particular:
the area of the building, the quality of the envelope, the number
of inhabitants, the type and the efficiency of the heating and
cooling system, the installed electrical appliances, etc. In
parallel, the operational conditions of the energy system and in
particular the set point temperature, the hours of operation of
the heating and cooling system and the conditioned area,
determine the energy consumption for heating and cooling.
Therefore, it is very important to analyse all the above
parameters and investigate their relation to the specific energy
consumption of the households.
3.1. Analysis of the general characteristics of the
households
3.1.1. Floor area of households per income group
The distribution of the floor area per household is shown in
Fig. 2. As observed the higher the income the higher the
household area. The median values of the occupied areas for the
seven income zones are 60 m2, 80 m2, 85 m2, 96 m2, 110 m2,
116 m2 and 128 m2, respectively. It is characteristic that the
mean occupied area for the richest income group is about 115%
higher than the corresponding area for the poorest group.
The mean occupied floor area per person is close to 37 m2.
The number of inhabitants per household increases as a
function of income. It is found that there is a statistical
significant correlation between the average number of persons
per household and the logarithm of the average income value
for each group. The average value of persons per household as
well as the median value of the occupied space per person is
given in Table 2 for all income groups. As shown the median
value of the occupied area per person does not show important
differences between the various income groups and only for the
richest group the occupied surface per person is much higher.
However, it is found that the median value of the occupied area
Fig. 2. Surface of buildings by income class.
896 M. Santamouris et al. / Energy and Buildings 39 (2007) 893–905

Table 2
Average number of persons per households as well as the median value of the
occupied area per person for all the income groups
Income group Average number Median value of
of persons per the occupied area
household per person (m2)
A: <9000 Euros/year 1.8 30.0
B: 9000–13,000 Euros/year 2.4 33.5
C: 13,000–24,000 Euros/year 2.8 30.0
D: 24,000–36,000 Euros/year 3.4 29.0
E: 36,000–63,000 Euros/year 3.5 32.5
F: 63,000–100,000 Euros/year 3.4 32.5
G: >100,000 Euros/year 3.4 41.0

per person presents a statistical significant linear correlation in

relation to the income, however, the increasing trend is not
statistically significant.

3.1.2. Age of households
Most of the buildings in Athens are recently built as the main
construction boom has happened since the 1970s. The average
age of the households in the sample is close to 25 years. The
mean age of the households per income group is given in
Table 3. Although differences are not very high it is evident that
the higher the income the lower the age of the buildings. A
statistical significant logarithmic correlation between the two
parameters has been observed. In particular, for the low and
medium income groups a more important decrease of the age of
households is observed as the income increases.
3.1.3. Type of households
Almost 56% of the sample families live in apartments. This is
very close to the figure given by the National Statistics Service
for the region of Athens [24]. As the income increases the number
of households living in apartments decreases. Thus, almost 64%
of the families in the lowest income group live in apartments,
while the corresponding number for the richest group is 48%. In
parallel, the number of families living in ‘maisonettes’ increases
as a function of income. Thus, while for the poorest group there is
no family living in such a building, the corresponding percentage
for the upper income group is close to 11%. The same pattern is
also valid for single storey houses. Only 7.5% of the low income
group lives in single storey houses, while the figure increases to
18% for the richest group.
For families living in multistorey apartment buildings an
analysis has been attempted to investigate if there is a relation
Fig. 3. The probability density of the relative vertical position of the households
in multistory apartment buildings for the lower and upper income groups.
between the income of the family and the floor or the height the
corresponding apartment is in the building. The relative
position of each apartment in a multistorey building has been
calculated by defining a parameter, RPA, as the ratio between
the floor on which the apartment is located to the total number
of storeys in the building. Thus, if the building has 10 storeys
and the specific apartment is located on the first floor, the ratio
RPA is equal to 0.1. Then, for each income group the
probability density of RPA is calculated. The obtained results
for the poorest and richest income groups are given in Fig. 3. As
shown, low income families live mostly in the lower part of
multistorey buildings while high income households live
mainly in the higher part of the buildings. This is mainly due to
the serious increase of the price of upper floor apartments.
In parallel, the relation between the income of families and
the size of multistorey buildings is investigated. Fig. 4, shows
the probability density regarding the number of apartments per

Table 3
Mean age of households per income group
Income group Mean age of the building stock
(reference year: 2004)
A: <9000 Euros/year 29
B: 9000–13,000 Euros/year 27
C: 13,000–24,000 Euros/year 26
D: 24,000–36,000 Euros/year 23
E: 36,000–63,000 Euros/year 21
F: 63,000–100,000 Euros/year 19 Fig. 4. The probability density of the number of apartments per floor for
G: >100,000 Euros/year 19 households living in multistory apartment buildings that belong to the lower and
the upper income groups.
 M. Santamouris et al. / Energy and Buildings 39 (2007) 893–905
floor for households living in multistorey apartment buildings
that belong to the lower and the upper income groups. As
shown, high income households prefer to live in smaller size
multistorey buildings, while low income families live mainly in
buildings with a much higher number of apartments per floor.
3.2. Quality of the building envelope
The quality of the building envelope of dwellings is
determined by many parameters. However, two are the most
important ones that mainly determine heat flux through the
envelope: the insulation of the opaque elements and the type of
installed glazing. Insulation has become mandatory in Greece
after 1979 and thus all buildings built before this are not
insulated. In parallel, the rate of retrofitting of buildings to add
insulation is quite low. According to the National Statistical
Service of Greece, the number of refurbished dwellings per
year is close to 250. Installation of double glazing in residential
buildings has started in reality at the late 80s. There are many
statistics about the number of insulated residential buildings or
the number of buildings with single or double glazing but there
is no information linking the quality of the envelope to the
available income of households. As already mentioned
information on the use of insulation in the buildings as well
as on the installation of double glazing was collected during the
inspections of the buildings. The results of the analysis are
discussed below.
3.2.1. Use of insulation
Almost 52% of the households in the sample are living in
fully or partly insulated buildings. The study did not investigate
problems concerning the quality of insulation placing. Such a
percentage is far from the official figure given by Eurostat [25].
According to [25], the percentage of insulated dwellings in
Greece is between 12% and 16%. However, this figure refers to
the National Survey of 1991 and does not reflect the current
situation. According to [24], the number of dwellings
constructed after 1979, when insulation became mandatory,
represents almost 40% of the existing dwellings stock. Taking
into account retrofitting of non-insulated buildings and
demolition of old buildings, the figure found by the present
survey is quite logical.
The percentage of insulated buildings per income group is
given in Table 4. As shown there is a very clear relation between
Table 4
Percentage of households living in insulate buildings, buildings with double glazing and insulated buildings with double glazing
Income group Percentage of households
living in insulated
buildings
A: <9000 Euros/year 28.0
B: 9000–13,000 Euros/year 39.0
C: 13,000–24,000 Euros/year 43.0
D: 24,000–36,000 Euros/year 54.0
E: 36,000–63,000 Euros/year 68.0
F: 63,000–100,000 Euros/year 73.0
G: >100,000 Euros/year 70.0
897
the income level and the percentage of non-insulated dwellings.
The higher the income the higher the percentage of insulated
buildings. It is characteristic that only 28% of people of the
poorest group live in insulated buildings, while the correspond-
ing figure for the richest group is approximately 70%. A
statistical significant correlation between the percentage of
households living in insulated buildings and the logarithm of
the income is observed. The increasing trend is much higher for
the low and medium income groups.
3.2.2. Use of double glazing
The percentage of buildings with double glazing in the
sample is close to 45%. According to Eurostat [25], the
percentage of buildings with double glazing is 8%. For
the reasons mentioned above such a figure is far from reality.
The percentage of buildings with double glazing is given in
Table 4. As in the case of insulation, there is a very clear relation
between the percentage of buildings with double glazing and
the income levels. The higher the income the higher the
percentage of buildings with double glazing. The differences
between the income groups are quite high. For the poorest
group the percentage of double glazed buildings is 24% while
for the richest group the corresponding figure is 67%. It is found
that the correlation between the percentage of buildings with
double glazing and the logarithm of the income level is
statistically significant. This correlation presents a statistically
significant increasing trend for the lower and medium income
groups.
Finally, insulated buildings with double glazing are quite
rare for the lower income groups (8%), Table 4, while the
corresponding percentage increases to 60% for the highest
income group. As before, the higher the income the higher the
percentage of buildings with insulation and double glazing.
Therefore, it is found that the correlation between the
percentage of buildings with double glazing and insulation
and the logarithm of the income level is statistically
significant.
3.3. Heating–cooling system and operational conditions
Most of the households (88%), are using a central heating
system. The percentage is quite similar for all income groups.
Only, 2.4% of the households are using stoves for heating,
mainly in the low income groups. At the same time, 4.6% of the
Percentage of households Percentage of households
living in buildings living in insulated buildings
with double glazing with double glazing
24.0 8.0
33.0 23.2
41.0 27.3
50.0 37.6
62.0 51.3
65.0 63.2
67.0 60.0
898 M. Santamouris et al. / Energy and Buildings 39 (2007) 893–905

Table 5 increase in the installed a/c system per household is seen as a

Mean heating hours per day and mean set point temperature for all income
function of income. The mean value for the lower income group
groups
is approximately 0.6 units/house, while the corresponding
Income group Mean hours of Mean set point value for the upper income class is about 2.15. The percentage
heating per day temperature (8C)
of households with at least one installed air conditioner rises
A: <9000 Euros/year 7.5 17.9 from 48% to 69% for the lowest and highest income classes,
B: 9000–13,000 Euros/year 7.1 17.9 respectively. The observed increase per income group is
C: 13,000–24,000 Euros/year 6.7 18.3
statistically significant at a level of significance equal to 0.05.
D: 24,000–36,000 Euros/year 7.6 18.7
E: 36,000–63,000 Euros/year 7.9 18.5 Although it is clear that higher income groups use more air
F: 63,000–100,000 Euros/year 8.3 18.7 conditioning units, the ratio of installed air conditioners per
G: >100,000 Euros/year 8.5 18.9 square meter and person, is much higher for the lower income
people. As shown in Table 5, the density of installed air
conditioners per person is much higher for the lower income
households are using movable oil or electrical systems for group than for the middle one and is comparative to the figure
heating. of the richest group. However, the density of installed air
The mean percentage of heated space compared to the total conditioners per square meter is much higher for the lower
floor area of the households is close to 92% and similar figures income people than for all other groups. The same is
are found for all income groups. observed when the ratio of installed air conditioners per
The mean daily duration of heating is approximately 7.5 h/ square meter and person is calculated Table 5. Thus, although
day. The mean daily hours of heating for each income group is middle and high income people use more air conditioning, the
given in Table 5. There isn’t any important difference between relative cost of comfort during the summer period is much
the four poorest income groups, while the heating period higher for the lower income people. This can be explained by
increases considerably, up to 8.5 h/day, in the richest groups. the fact that low income people live in buildings with limited
However, there is a statistically significant correlation between thermal protection and also because low income housing is
the two parameters, but the trend in the lower and medium located in areas of Athens where the heat island is at a
income groups is not statistically significant. maximum.
The average set point temperature for the heating period was
close to 18.4 8C. A small but important upward trend in the set 3.4. Energy consumption characteristics
point temperature has been observed, as a function of the
income level (Table 5). The difference in the set point 3.4.1. Energy consumption for heating
temperature between the poorest and the richest groups is close The mean and median energy consumption per square metre
to 1 8C. As previously, the correlation between the two for heating purposes for each income group is given in Fig. 5.
parameters is statistically significant, but the increasing trend is The distribution has a U-type shape with high consumption at
not. the two limits. High energy consumption per square meter is
Concerning maintenance of the heating system, almost the observed for both the low and high income groups. Higher
82% of the households maintain the heating system on an energy consumption of the lower income group can be
annual basis, 11% every 2 years, and 6% every 3 years. Non explained by the poor quality of the envelope of the building. In
significant differences have been observed between the parallel, the high energy consumption of the upper income
different income classes. group can be explained by the longer hours of operation of the
During recent years, air conditioning (a/c) has achieved a heating system and the higher set point temperature. In parallel,
very high penetration in the residential sector. The average high income people live mainly in single detached houses that
number of air conditioners per households for all the income have a much higher energy consumption than apartments (see
groups is given in Table 6. A very high statistical significant below).

Table 6
Number of air conditioners units per household, percentage of households with at least one air conditioner, number of air conditioners per person, number of air
conditioners per square meter of households and number of air conditioners per square meter and person
Income group Number of air Percentage of households Number of air Number of air Number of air
conditioning units with at least one air conditioners conditioners per conditioners per
per household conditioning unit (%) per person square meter of square meter and
households (100) person (1000)
A: <9000 Euros/year 0.60 48 0.47 9.8 7.4
B: 9000–13,000 Euros/year 0.84 53 0.40 1.1 6.1
C: 13,000–24,000 Euros/year 1.1 60 0.45 1.2 5.5
D: 24,000–36,000 Euros/year 1.2 63.2 0.40 1.3 4.4
E: 36,000–63,000 Euros/year 1.5 68.0 0.49 1.3 4.4
F: 63,000–100,000 Euros/year 1.70 67.0 0.53 1.3 4.5
G: >100,000 Euros/year 2.15 69.0 0.74 1.6 6.2
 M. Santamouris et al. / Energy and Buildings 39 (2007) 893–905
Fig. 5. Mean and median primary energy consumption for heating (litters oil/
m2/year), for all income groups.
In absolute values, the mean energy consumption for heating
varies between 154 kW h/m2/year and 200 kW h/m2/year,
while the corresponding median values vary between
107 kW h/m2/year and 130 kW h/m2/year. Such a level of
consumption is verified by a very detailed monitoring
performed by the University of Athens, and is quite high
given the climatic conditions of Athens. In particular, the level
of consumption per square meter is quite similar to mean
consumption of the residential sector in Austria [26]. Thus,
there is a very high potential for energy conservation.
A very similar pattern is found when the mean and median
heating energy consumption per person is calculated (Fig. 6).
At the same time when the mean heating energy consumption
per person and area unit is calculated (Fig. 7) it is evident that
the lower the income the higher the cost of heating per person
and unit of surface. As shown, the cost per person and square
meter for the lower income group is to about 127% higher than
the corresponding cost of heating for the richest group.
3.4.2. Energy consumption by type of dwelling
The median value of the heating energy consumption per
square meter for the apartments as well as for the detached
houses has been calculated for each income group. Fig. 8 shows
the specific heating energy consumption per unit area for the
detached houses and apartments as well as the median value of
Fig. 6. Mean and median primary energy consumption for heating (litters oil/
person/year), for all income groups.
899
Fig. 7. Mean and median primary energy consumption for heating (litters oil/
person/m2/year), for different income groups.
the whole building stock. As observed, the energy consumption
of detached houses is to about 50% higher than that of
apartments.
The distribution of the specific heating energy consumption
for the apartment buildings does not present an important
variation as a function of income, although there is a slight
falling trend. On the contrary the corresponding distribution for
the non apartment dwellings follows a U-shape pattern as
observed for the whole building stock. This can be explained by
examining closely the age distribution of the different types of
buildings.
Almost 90% of the non apartment buildings occupied by the
low income group have been constructed before 1979, the year
that insulation made mandatory, and as mentioned are of poor
energy quality. However, the corresponding value for the
apartment buildings is much lower at 55%. Thus, for the lower
income group, the stock includes many-older detached houses
and newer apartments. In the upper income group, only 30% of
the detached houses and 20% of the apartments were built
before 1979. Thus, the difference of quality between the income
classes is much higher in the case of the detached houses than is
the case in apartments.
Fig. 8. Median value of the primary energy consumption for heating (litters oil/
m2/year), for apartments, and detached dwellings, for all income groups.
900 M. Santamouris et al. / Energy and Buildings 39 (2007) 893–905
Fig. 9. Variation of the median energy consumption for heating for all house-
holds as a function of the year of construction.
3.4.3. Energy and age of households
New construction standards and energy efficient technolo-
gies used in recent years should have had a positive effect on the
energy consumption of recent buildings. The heating energy
consumption per square meter for the whole sample has been
calculated as a function of the age of the households (Fig. 9).
Buildings have been divided into six age groups (<1960, 1960–
1970, 1970–1980, 1980–1990, 1990–2000, >2000). As shown,
there is a decreasing trend as a function of time. The calculated
rate of decrease is close to 0.8 kW h/m2/year. Concerning the
temporal variation of the heating energy consumption for the
defined income classes, almost the same trend has been found
for all classes (Fig. 10).
3.4.4. The impact of insulation and double glazing on the
energy consumption for heating
To evaluate the impact of insulation and double glazing, the
database has been split into three groups:
(a) non-insulated buildings with single glazing
(b) insulated buildings with single glazing, and
Fig. 10. Variation of the median energy consumption for heating for income
classes A and F as a function of the construction year.
Fig. 11. Heating energy consumption of detached households, per unit of floor,
for: (a) non-insulated buildings with single glazing; (b) non-insulated buildings
with double glazing; (c) insulated buildings with single glazing and (d) insulated
buildings with double glazing.
(c) insulated buildings with double glazing.
The heating energy consumption per square meter has been
calculated for the three groups and for both the detached and
apartment dwellings. The results are given in Figs. 11 and 12 for
the detached and the apartment buildings, respectively.
For the detached houses the combined impact of insulation
and double glazing decreases the mean heating energy
consumption of the building stock to about 40 kW h/m2/year
compared to a single glazed non-insulated dwelling. This
corresponds to a reduction of approximately 20%. For
apartments the corresponding reduction is much lower,
8 kW h/m2/year, or 5%, as the envelope losses of apartments
are much lower than for the detached houses. Similar results are
obtained from detailed simulation studies.
3.4.5. Electricity energy consumption
The median value of the cost of electric energy per
household was calculated to be about 600 Euros/year. The
median value of the cost per household for all the income
groups is given in Fig. 13a. As shown, there is almost a linear
Fig. 12. Heating energy consumption of apartment dwellings, per unit of floor
in litters of oil per square meter per year, for: (a) non-insulated buildings with
single glazing; (b) insulated buildings with single glazing and (c) insulated
buildings with double glazing.
 M. Santamouris et al. / Energy and Buildings 39 (2007) 893–905 901

Fig. 13. (a) Variation of the median electric energy cost per household for all income groups; (b) variation of the median electric energy cost per square meter for all
income groups; (c) variation of the median electric energy cost per person for all income groups; (d) variation of the median electric energy cost per square meter and
person for all income groups.

relation between the annual expenses for electricity and the conditioner installed has been calculated. The annual cost
family income. Low income families pay almost 300 Euros/ per dwelling for each income group is given in Fig. 14. As
year while the corresponding cost for the highest income group shown, the use of air conditioning increases considerably the
is close to 775 Euros/year, i.e. 160% higher. A very similar annual electricity expenses especially in the low income
relation is found between the annual electricity cost per unit of groups. As a mean value, the use of air conditioning increases
area and the income (Fig. 13b). The low income group pays the annual energy expenditure by about 100 Euros/household,
almost 5 Euros/m2 and year while the corresponding value for or 0.6 Euros/m2, or 12.5 Euros/person. The increase is much
the highest income group is 6.9 Euros, i.e. 38% more. Increased higher for low income groups, where the increase of the energy
cost may be explained by the considerably higher installed cost caused by air conditioning use is close to 195 Euros/
power and use of electrical appliances and equipment in household, or 1.2 Euros/m2, or 87 Euros/person. Thus, the
households of the richest groups. findings of the previous chapter are verified: that although
However, when the annual electricity cost per person is middle and high income people use more air conditioning, the
calculated, a U-shaped figure is obtained (Fig. 13c). A similar relative cost of comfort during the summer period is much
shape has been obtained when the specific heating consumption higher for people on a lower income. As already mentioned,
per person and income group has been calculated (Fig. 6). In this may be explained by the fact that low income people live in
absolute values the difference between the various groups is not
so high. The maximum difference is approximately 11%, while
the difference between the poorest and richest group is close to
6%. Important results are obtained when the annual electricity
cost per unit of area and person is calculated (Fig. 13d). As
shown, the lower the income the higher the cost of electricity
per person and unit floor area. The correlation is found
statistical significant for the low and medium income groups. A
similar pattern has been found also for the heating energy
consumption per person and unit of area (Fig. 7). As calculated,
low income people pay almost 67% more per person and square
meter than high income people.
In order to evaluate the impact of air conditioning on the
electricity expenses of each income group, the specific Fig. 14. Variation of the annual median electric energy cost per household for
electricity cost for all families having at least one air families with and without air conditioning for all income groups.
902 M. Santamouris et al. / Energy and Buildings 39 (2007) 893–905

buildings with limited thermal protection and also because low
income housing is located in areas of Athens where the heat
island has maximum intensity. In parallel, the duration of
summer holidays for low income people is much lower or
negligible compared to that of the upper income groups. This
has an important impact as it increases the relative use of the air
conditioning in the low income groups.
3.5. Fuel poverty
Analysis of the data has shown that the mean income
fraction spent for heating and electricity is approximately 2.4%
and 3.1%, respectively (2004 values). Thus, almost 5.5% of the
annual income is spent on energy. The specific income fraction
spent by each income group for heating and electricity purposes
is given in Fig. 15a. As shown, annual heating expenditures
represent almost 6.2% of the total income of the poorest people.
The corresponding figure for the upper income group is close to
0.6%.
Annual expenditures for electricity represent almost the
5.9% of the income of the poorest group, while for the richest
group the corresponding value is about 1.1%. Thus, the total
annual expenses for energy represent almost the 12.1% of the
income of the poorest group, while for the richest group the
corresponding value is close to 1.4% (Fig. 15a).
As already mentioned, fuel poverty is defined in the UK as
the number of households needing to spend in excess of 10% of
income on energy services, and if expenditure exceeds 20% of
the income, the family is characterized as in ‘severe energy
poverty’. Analysis of the data has shown that 1.63% of the
households suffer from fuel poverty and 0.35% from severe fuel
poverty (2004 values). The corresponding values per income
group are given in Fig. 15c. As shown, fuel poverty in low
income groups, is approximately 16%. Fuel poverty conditions
are not found for incomes higher than 24,000 Euros/year.
Severe fuel poverty, in the low income group, is calculated
close to 4% (Fig. 15d). Severe fuel poverty is found only in
income groups with less than 13,000 Euros/year. All this data is
based on what households are actually spending for heating
purposes. Based on the UK definition of fuel poverty, data have
to be based on what households they need to spend in order to
achieve a specified comfortable and healthy heating standard.
Such an evaluation requires knowledge of the indoor
temperature during the winter period. During the study,
specific measurements of indoor temperature have been
performed in 50 dwellings belonging to the low income
groups. Measurements have been performed during the cold
period [27]. It has been found that almost all dwelling are
adequately heated. During the occupation period, average
indoor temperatures were found close to 18–19 8C. Thus, it is
considered that the data based on what households are actually
spending are very similar to the corresponding figures based on
what they need to spend to achieve a comfortable indoor
environment.

Fig. 15. (a) 2004: Bar: percentage of annual expenditure for heating (o): percentage of annual expenditure for electricity line: percentage of annual expenditure for
heating and electricity. (b) 2006: Bar: calculated percentage of annual expenditure for heating (o): calculated percentage of annual expenditure for electricity line:
calculated percentage of annual expenditure for heating and electricity. (c) Bar: percentage of fuel poor (heating), per income group for 2004, line: calculated
percentage of fuel poor (heating), for 2006. (d) Bar: percentage of severe fuel poor (heating), per income group for 2004, line: calculated percentage of severe fuel
poor (heating), for 2006. (e) Bar: percentage of fuel poor (heating + electricity), per income group for 2004, line: calculated percentage of fuel poor
(heating + electricity), for 2006. (f) Bar: percentage of severe fuel poor (heating + electricity), per income group for 2004, line: calculated percentage of severe
fuel poor (heating + electricity), for 2006.
 M. Santamouris et al. / Energy and Buildings 39 (2007) 893–905
When both heating and electricity expenditures are
considered, the average percentage of households spending
more than 10% of their income for energy is increased to
11.3%, while 2% spends more than 20% (2004 values). The
percentage of families per income group spending more than
10% or 20% of their income for energy is given in Fig. 15e and
f, respectively. As shown almost 40% of the low income group
spends more than 10% of their income for energy (energy
poor), while almost one fifth of the poor households spends
more than 20% of their income for energy (severe energy
poor).
Given that this data was collected during 2004, and since the
price of fuel has almost doubled, figures on fuel and energy
poverty have been adjusted to the current situation, assuming
that the levels of energy consumption remain the same. The
actual price of oil [28], and electricity [29], as well as the
annual increase of GDP for 2004 and 2005 have been
considered [30]. New fuel prices have contributed to the
increase of the mean income fraction spent for heating from
2.4% to 4.5%. In parallel, the average percentage spent on
energy has increased from 5.5% to 7.4%. The mean income
fractions spent by each income group for heating, electricity
and energy are given in Fig. 15b. As shown, the average income
fraction spent for heating has increased from 6.2% to 11.6% for
the low income group. In parallel, the total expenses for energy
of the same group have increased from 12.1% to 17.6%. For the
upper income group, the fraction of the income spent for
heating increased from 0.6% to 1.1%, while total energy
expenses now represent 1.9% of the total income compared to
1.4% in 2004.
New fuel prices have increased the average percentage of
households suffering from fuel poverty from 1.6% to 8.4%. The
calculated fraction of fuel poor per income group is given in
Fig. 15c (line). For the lower income group, fuel poverty
increased from 16% to 36%, while fuel poverty is now observed
in the next richest income group (24,000–36,000 Euros).
Average severe fuel poverty increased from 0.35% to 1.5%. The
distribution of severe fuel poverty by income group is given in
Fig. 15d (line). As shown, the percentage of severe poor in the
lower income group increased from 4% to 12%, while severe
fuel poverty is observed in the three poorest income groups
instead of the two in 2004.
The average percentage of households suffering from energy
poverty, i.e. their total expenses for energy exceeds 10% of their
income, increased from 11.3% to 21.1%. The distribution of
energy poor per income group is given in Fig. 15e (line). It is
characteristic that energy poverty in the low income groups
increased from 40% to 60%. In parallel, the mean percentage of
severe energy poor, increased from 2% to 3.7%. The
distribution of severe energy poor per income group is given
in Fig. 15f (line). Severe energy poverty in low income groups
increased from 20% to 32%.
4. Main results and conclusions
Social, financial, energy and technical data from about 1110
households have been collected during 2004 in the greater
903
Athens area. The sample has been divided into seven income
groups and a detailed analysis has been performed.
The main conclusions are:
(a) There is an almost linear relation between income and
occupied space per household. The mean occupied area for
the richest income group is about 115% higher than the
corresponding area for the poorest group.
(b) The mean occupied surface per person is close to 37 m2.
The number of inhabitants per household increases as a
function of income. The median value of the occupied area
per person does not present important differences between
the various income groups and only for the richest group
the occupied surface per person is much higher.
(c) Concerning the distribution of the age of the households
per income group, it was found that the higher the income
the lower the age of the buildings. The mean age of
households of the lower income group is 29 years while for
the richest group is 19 years.
(d) Regarding the type of households per income group, it
was found that as the income increases the number of
people living in apartments decreases. Almost 64% of the
families in the lower income group live in apartments,
while the corresponding number for the richest group is
48%. In parallel, low income families live mostly in the
lower part of multistory buildings while high income
households live mainly in the higher part of the
buildings. Also, the analysis has shown that high income
households prefer to live in smaller size multistory
buildings, while low income families live mainly in
buildings with a much higher number of apartments per
floor.
(e) There is a very clear relation between the income level and
the percentage of non-insulated dwellings. The higher the
income the higher the percentage of insulated buildings.
Only 28% of people of the poorest group lives in insulated
buildings, while the corresponding figure for the richest
group is close to 70%. In parallel, the higher the income the
higher the percentage of buildings with double glazing. For
the poorest group the percentage of double glazed
buildings is 24% while for the richest group the
corresponding figure is 67%. Insulated buildings with
double glazing are quite rare for the lower income groups
(8%), while the corresponding percentage increases to 60%
for the high income group.
(f) The mean daily duration of heating is approximately 7.5 h/
day. The heating period increases considerably, up to 8.5 h/
day, in the richest groups. The average set point
temperature for the heating period is close to 18.4 8C.
The difference in the set point temperature between the
poorest and the richest groups is about 1 8C.
(g) A very high increase of the installed a/c units per household
is observed as a function of income. The mean value for the
lower income group is close to 0.6 air conditioners/
household, while the corresponding value for the upper
income class is about 2.15. However, the density of
installed air conditioners per square meter is much higher
904 M. Santamouris et al. / Energy and Buildings 39 (2007) 893–905
for the lower income people that for all other groups.
Although middle and high income people use more air
conditioning, the relative cost of comfort during the
summer period is much higher for the lower income
people.
(h) The distribution of the energy consumption for heating
presents a U-type shape with high consumption at the two
limits. High energy consumption per square meter is
observed for both the low and high income groups.
Concerning the mean heating energy consumption per
person and area it is found that the lower the income the
higher the cost of heating per person and unit of surface.
The cost per person and square meter for the lower income
group is to about 127% higher that the corresponding cost
of heating for the richest group.
(i) Regarding the heating energy consumption per type of
dwelling it is found that the energy consumption of
detached houses is to about 50% higher than that of
apartments.
(j) There is a considerable decreasing trend of the heating
energy consumption as a function of age of the buildings.
The calculated decreasing trend of the heating consump-
tion is close to 0.8 kW h/m2/year.
(k) For detached houses, the combined impact of insulation
and double glazing decreases the mean heating energy
consumption of the building stock to about 40 kW h/m2/
year compared to a single glazed non-insulated dwelling.
For apartments, the corresponding reduction is much
lower, 8 kW h/m2/year.
(l) There is an almost linear relation between the annual
expenses for electricity and the family income. High
income families pay almost 160% higher annual cost than
the low income ones. As it concerns the annual electricity
cost per unit of area and person it is found that the lower the
income the higher the cost of electricity per person and unit
of area. Low income people pay almost 67% higher
electricity cost per person and square meter than high
income people.
(m) The use of air conditioning increases considerably the
annual electricity expenses especially in the low income
groups. As a mean value, the use of air conditioning
increases the annual expenses to about 100 Euros/house-
hold, or 0.6 Euros/m2, or 12.5 Euros/person. The increase
is much higher for the low income groups, where the
relative increase of the cost because of the air conditioning
use is approximately 195 Euros/household, or 1.2 Euros/
m2, or 87 Euros/person.
(n) The mean income fraction spent for heating and electricity
is close to 2.4% and 3.1%, respectively (2004 values).
Annual heating expenditures represent almost 6.2% of the
total income of the poorest people. The corresponding
figure for the upper income group is about 0.6%.
(o) Almost 1.63% of the households suffer from fuel poverty
and 0.35% from severe fuel poverty (2004 values). Fuel
poverty in low income groups, is in the region of 16%.
Severe fuel poverty, in the low income group, is calculated
close to 4% (Fig. 15d).
(p) Concerning energy poverty, the average percentage of the
households spending more than 10% of their income for
energy is close to 11.3%, while 2% spends more than 20%
(2004 values). Almost 40% of the low income group
spends more than 10% of their income for energy (energy
poor), while almost one fifth of the poor households spends
more than 20% of their income for energy (severe energy
poor).
(q) Considering 2006 fuel prices and increase of GDP, the
mean income fraction spent for heating has increased from
2.4% to 4.5%, while the average percentage spent for
energy has increased from 5.5% to 7.4%. For the low
income group, the average income fraction spent for
heating increased from 6.2% to 11.6%. In parallel, the total
expenses for energy of the same group increased from
12.1% to 17.6%.
(r) New fuel prices have increased the average percentage of
households suffering from fuel poverty from 1.6% to 8.4%.
For the lower income group, fuel poverty has increased from
16% to 36%. Average severe fuel poverty has increased from
0.35% to 1.5%. The percentage of severe poor in the lower
income group has increased from 4% to 12%.
(s) The average percentage of households suffering from
energy poverty, has risen from 11.3% to 21.1%. Energy
poverty in the low income groups has increased from 40%
to 60%. The mean percentage of severe energy poor, has
also increased from 2% to 3.7%. Severe energy poverty in
low income groups has risen from 20% to 32%.
The whole analysis permits to conclude that energy
consumption spent for heating and electricity purposes is
quite high and there is a very high potential for energy
conservation. Energy consumption per person and square meter
both for heating and electricity, is much higher in the low
income groups mainly because of the poor quality of the
building envelope. Fuel and energy poverty reach quite high
levels in the low income groups and there is a dramatic raise
because of the recent increase of the fuel prices.
Energy policies addressing the residential sector should set
as a priority the improvement of the envelope quality of
dwellings where low income people is living. Further increase
of the fuel prices will aggravate seriously the economic
situation of these households and may create a serious social
problem. It is calculated that a further increase of oil prices by
10 dollars per barrel brings 2.5% of the population under fuel
poverty. The creation of a special fund to finance energy
retrofitting of low income households appears to be a necessity.
Acknowledgments
The authors wishes to thank Janet Rudge of the University of
East London, Fergus Nicol of the Metropolitan University of
London, Peter Wouters of the Belgium Building Research
Institute and K. Vasilakopoulou for reviewing this documents
and for the extremely useful comments they provided us. Also,
the authors wish to express their gratitude to all students of the
University of Athens they have participated in this study.
 M. Santamouris et al. / Energy and Buildings 39 (2007) 893–905
References
[1] International Energy Agency: Energy Information Administration,
Annual Energy Review, Paris, France, 2005.
[2] Department of Energy (DOE): International Energy Outlook 2006,
Report: DOE/EIA-0484(2006), Release Date: June 2006.
[3] Department of Energy (DOE): Residential Energy Consumption Survey,
USA, 2001.
[4] J.D. Healy, J. Peter, Clinch: fuel poverty, thermal comfort and occupancy:
results of a national household-survey in Ireland, Applied Energy 73
(2002) 329–343.
[5] English House Condition Survey: Follow Up Survey, DTI and DEFRA,
2001.
[6] L. Heslop, L. Moran, A. Cousineau, ‘‘Consciousness’’ in energy con-
servation behavior: an exploratory study, Journal of Consumer Research 8
(1981) 299–305.
[7] B. Morrison, P. Gladhart, J. Zuiches, J. Keith, D. Keefe, B. Long, Energy
and families: the crisis and the response, Journal of Home Economics
(1978) 18–21.
[8] D.K. Newman, D. Day, The American Energy Consumer, Cambridge,
MA, 1975.
[9] J.R. Ritchie, G.H. McDougall, J.D. Claxton, Complexities of household
energy consumption and conservation, Journal of Consumer Research 8
(1981) 233–242.
[10] G. Tso, K.W.Y. Kelvin, A study of domestic energy usage patterns in Hong
Kong, Energy 28 (2003) 1671–1682.
[11] B. Wouter, J.N. Klaas, Energy requirements of household consumption: a
case study of The Netherlands, Ecological Economics 28 (1999) 367–383.
[12] C.J. Lam, Climatic and economic influences on residential electricity
consumption, Energy Conversion Management 39 (7) (1998) 623–629.
[13] B.W. Ang, T.N. Goh, X.Q. Liu, Residential electricity demand in Singa-
pore, Energy 17 (1) (1992) 37–46.
[14] J.D. Healy, J. Peter Clinch, Fuel Poverty in Europe: A Cross Country
Analysis Using A New Composite Measurement, Environmental Studies
905
Research Series Working Papers, University College Dublin, Ireland,
2002.
[15] Fuel Poverty and Energy Efficiency in the UK Calouste Gulbenkian
Foundation, 2004.
[16] J.D. Healy, J. Peter, Clinch: quantifying the severity of fuel poverty, its
relationship with poor housing and reasons for non-investment in energy-
saving measures in Ireland, Energy Policy 32 (2004) 207–220.
[17] Clinch, J.D. Healy, Housing standards and excess winter mortality, Journal
of Epidemiology Community Health 54 (9) (2000) 719–720.
[18] J. Rudge, F. Nicol, Cutting the Cost of Cold: Affordable Warmth for
Healthier Homes, E & FN Spon, London, 2000.
[19] M. Curwen, Excess winter mortality: a British phenomenon? Health
Trends 22 (1991) 169–175.
[20] M. Iatridis, G. Zoidis, Energy Efficiency in Greece, 1990–2002, Monitor-
ing of energy efficiency in Europe A report based on ODYSSEE—MURE
data bases and Supported by the SAVE programme, Center Renewable
Energy Sources, October 2004.
[21] T.R. Vassilis, L.P. Michael, The structure of residential energy demand in
Greece, Energy Policy 34 (2006) 3137–3143.
[22] G.S. Donatos, G.J. Mergos, Energy demand in Greece: the impact of two
energy crises, Energy Economics 11 (2) (1989) 147–152.
[23] G. Hondroyiannis, Estimating residential demand for electricity in
Greece, Energy Economics 26 (2004) 319–334.
[24] National Statistics Service of Greece: Construction Report, 2003.
[25] Eurostat, Energy Consumption in Households, Office for Official Pub-
lications of the European Union, Luxembourg, 1999.
[26] E.V.A., the Austrian Energy Agency: Energy Efficiency and CO2 Emis-
sions in Austria, March 2002.
[27] K. Argiroudis, Measurement of the indoor environmental quality in 50
dwellings in Athens, M.Sc. report, University of Athens, 2005 (In
Greek).
[28] http://www.oil-price.net.
[29] Eurostat: Electricity Prices for the household sector, 2004, 2005, 2006.
[30] Information obtained from the National Bank of Greece, www.nbg.gr.