Electricity Consumption And Consumer Behaviour

Shubham Upadhyay
Omkar Wagh Rahul Shevde
Electrical Department
Electrical Department Electrical Department
Lokmanya Tilak College of
Lokmanya Tilak College of Lokmanya Tilak College of
Engineering
Engineering Engineering
Navi Mumbai, India
Navi Mumbai, India Navi Mumbai, India
shubhamupadhyay281@gmail.com
omkarwagh48@gmail.com rahul.shevde.07@gmail.com
Prof. Ninad Totre
Electrical Department
Lokmanya Tilak College of
Engineering
Navi Mumbai, India
ninadtotre@rediffmail.com

Abstract— Consumption as a whole is an essential term for

measuring or calculating the amount of a particular II. PREVIOUS WORK
commodity being utilized. As far as electricity is concerned, the Until now, there are three major modelling approaches for
consumers seldom pay attention to the electricity bills they household electricity consumption. These approaches are
receive unless the bills are more than usual. Hence it is a
limited due to less availability of data. They are:
necessity to be aware about the reasons in variation of electric
consumption. As consumer behavior is a subset of human 1) Top-Down approach: It focuses on the interaction
behavior, it is quite difficult to map and link the behavior between electricity consumption and economic factors at a
pattern of a particular individual with respect to electric higher level using aggregated socio-economic data[5].
consumption. In this paper, a survey has been conducted to 2) Bottom-up approach: It statistically analyses household
observe behavioral pattern of an average consumer by means data based on survey and electricity consumption
of a questionnaire. On the basis of data collected of survey, readings[5].
behavioral models of consumption are created to examine the 3) Physical model approach: It models physically measured
behavioral pattern of an average consumer. data on specified houses, their appliances and technologies.
All three approaches have their own pros and cons due to
Keywords—Electricity, Consumption, Behavior Pattern, difference in nature of their input data and assessment
Consumer, Statistical Analysis, Statistical Tools, Consumer capability. Top-down models are mostly high-level studies
Lifestyle. and analyse highly aggregated data at a national scale. The
majority of papers focus on analysing the socio-economic
I. INTRODUCTION impacts of the electricity sector, whereas Bottom-up
modelling uses segregated data to evaluate the impact of
Electricity, being one of the most basic necessities of human various factors on electricity consumption. Some bottom up
life, it has now become essential to monitor its consumption approaches use samples of houses’ building physics to
and to observe the behavioral pattern of a consumer. Hence, represent larger housing stock, combining building
study for the same is important so as to manage the growing electricity calculations with statistical methods[5].
energy demands & to gain control over unnecessary Number of international studies have focused on better
consumption determinants. For the same, observations are understanding of household electricity demand[5].
made with respect to energy profiles of individuals and their
interactions. In this paper, we provide a detailed statistical A. Top Down Approach
analysis by means of survey conducted on the basis of In Italy, a study used a top down approach to forecast annual
certain influential factors such as social, cultural, weather, residential electricity consumption using a multiple linear
economic, household demographics etc. on electric regression model. The study utilized a correlation to analyze
consumption and consumer behavior by means of statistical the sensitivities and relationships between aggregated power
tools. These statistical tools help us to provide an insight to demand and a set of demographic variables[7].
create models of consumer’s behavioral characteristics, their Another study in Italy considered ageing population’s
demands for electricity & variations with respect to impact on energy consumption. In the model presented in
influential and contributing factors. As a result, the this study, although changes in the age composition of the
interrelationship between different influential factors show population might generate shifts in consumption, it also
how consumer reacts and responds to various determinants results in slower economic growth. For Italy, population
of electric consumption. The behavior varies according to ageing has significant negative supply side effects which
specific situations in consumer’s lifestyle. reduce the overall energy use in the economy, and thus
Hence, steps must be taken on how different factors must be carbon emissions[8].
manipulated to improve efficient energy consumption in
future. B. Bottom Up Approach
In Canada, a study used a bottom up approach utilising the
Canadian Hybrid Residential End Use Energy and
Greenhouse Gas Emission Model (CHREM) – a statistical
method to analyse national residential energy consumption.
The model takes into consideration appliances, lighting, and average electricity bill (in terms of Rs.) is considered for
domestic hot water to forecast national annual energy plotting the model.
consumption. This study used the national Survey of
Household Energy Use (SHEU) undertaken in 1993[10]. IV. METHODOLOGY
The study initiated with exploring previous researches
C. Potential Strengths and Limitatons conducted in the same field. Several research papers were
The top-down approach provides a focused analysis which examined and necessary points for reference were selected.
considers the links between the economy and the energy Also, the national level surveys of government were
sector at a higher level, but generally lacks technological examined in order to know exactly how a nation-wide
detail and therefore cannot examine technological related analysis is performed. After completing the literature survey,
impacts. The bottom-up approach generally provides a Factors like social, cultural, regulation, climate are
considered as constraints. Factors such as age group, income,
better understanding of the technologically driven electric
basic electrical appliances, their frequency of usage,
consumption. However, it may require a larger sample size
consumption by means of electrical bills are selected that
and its reliability may depend on historical data which is not affected both consumption and consumer behaviour. These
always made available[5]. factors were studied in detail for better understanding of the
D. In India subject.
The household electricity sector module is a tool kit which
will allows users and modelers to examine, explore, study
and run a diverse number of situations. The approach which
is used in this model focuses mainly on the ownership and
the use of a house and its occupants’ electricity-consuming
electrical appliances. To show and exhibit its working two
ways are chosen. In first way which is based on the
maximum efficiency level which can be achieved on all the
currently in use appliances since 2008 in our country. The
second has higher energy efficiency and calculates the
decrease in the electrical consumption build exclusively on
engineering presumptions. The difference in the terminal
year of modelling is considerable. In reality however,
the reduction in expected gains from new technologies that
increase the efficiency of resource use, because of
behavioral or other systemic responses would drop down
some of the potential savings. It is interesting to note that, in
the first way of approach towards modelling, even though
the GDP increases at an annual average rate of 7.8 % the
modelling time period from 2005 to 2031, household
electricity usage grows by an average yearly rate of 5.8
percent, or as per 74 percent of GDP growth. As the
household size is decreasing, this means that income
flexibility is now even smaller when housing units are
considered as electrical energy consuming loads. From
2008, due to establishment, institution and launch of highly
efficient electrical equipment and appliances will obviously Figure.1:-Influention Factors taken in considertaion
help us to contain the growth of household unit's electrical Hence, it came to our attention that to represent the
power consumption. These findings need to be examined relationship between electric consumption and consumer
further by sensitivity analysis and with the help of other behaviour, a mathematical expression is essential[5]. The
studies and careful analysis. But there is a problem which is mathematical expression selected is as follows:
effect of popularity and technology. For most appliances, an Yi = Ai + Bi + Di + e (1)
analysis of different rounds of NSS data reveals increasing
ownership at the same household expenditure level[6]. Where Yi = Electricity Consumption

III. DATA COLLECTION Ai = Household Appliances Effect

The data obtained for behavioral analysis came through a Bi = Behavioural Effect
survey consisting series of questions which had either Di = Household Demographics Effects
multiple choice or checkbox responses. The survey covered
over 180 households which are located all over Mumbai. e = Errors / Deviations
The questionnaire consisted questions regarding the number As it can be observed, electricity consumption is a function
of electrical appliances in an individual’s house, the of associated affecting factors which contain sub factors for
duration for utilizing the devices, the frequency of utilizing, each term individually. The terms are further explained in
the no. of occupants in the house, no. of employed persons, terms of equations as follows
annual income and availability of electric power. Also, an
 continious predictors we have number of all the appliance
Ai = γ1*AC + γ2*FG + γ3*EI + γ4*EG + γ5*LT + γ6*TV + sued like fan, lights, Air conditioners etc stated in the above
section as they are the most significance in this paper and in
γ7*WM (2) the place of categorial predictiors we put the various other
where γ1 to γ7 are coefficient vectors according to different important factors that cause effects on the above factors like
impact factors, variable AC stands for effects due to no. of Income group, number of occupants. And the regression plot
air conditioners, variable FG stands for effects due to no. of observed is shown below
refrigerators, variable EI for effects due to no. of Electric
Iron, variable EG for effects of no. of Geyser, LT for effects
due to no. of lights and lamps, TV for effects due to no. of
Television and WM for effects due to no. of Washing
Machine[5].
Bi = β1*AC + β2*FG + β3*EI + β4*EG + β5*LT + β6*TV +
β7*WM (3)
where β1 to β7 are coefficient vectors according to different
impact factors, variable AC stands for effects due to
frequency of usage of air conditioners, variable FG stands for
effects due to frequency of usage of refrigerators, variable EI
for effects due to frequency of usage of Electric Iron,
variable EG for effects of frequency of usage of Geyser, LT
for effects due to frequency of usage of Lights and Lamps, Figure 2:- Different statistical regression models based on
TV for effects due to frequency of usage of Television and household appliances effect
WM for effects due to frequency of usage of Washing From the regression plot observed we get the uneven
Machine[5]. variation in plot due to the variation of response received
and we get the values γ for the various different responses.
Similarly we can calculate the values of α,β by performing
Di = α1*I + α2*NumC0_10 + α3*NumC10_20 + regression for the different factors taken into consideration
α4*NumC20_30 + α5*NumC30_45 + α6*NumC45_60 for them respectively. The error function ‘e’ represents
deviations in electric consumption due to various irregular
+ α7*NumC>60 + α8*Nocc factors like Festivals, Festive rituals etc. which are of very
Where α1-α7 denotes the coefficient vector corresponding to short duration of time span like 2-3 days or a week
various impact factors. The variable I denotes the Income of maximum, they have less significane but cause deviation,
each household, NumC0-10 denotes how many occupants of hence it is considered in the above study.
ages 0-10 are in each household, NumC10-20 denotes how
VI. CONCLUSION
many occupants of ages 10-20 , NumC20-30 denotes how
many occupants of ages 20-30 are in each household, The ultimate aim of any energy distributor is to gain profit
NumC30-45 denotes how many occupants of ages 30-45, but also to ensure meeting load demands of a consumer. For
NumC45-60 denotes how many occupants of ages 45-60, this, it is essential that both, the distributor & consumer have
NumC>60 denotes how many occupants of ages more than good knowledge about various determinants that affect the
60 and NumOcc denotes the number of occupants in each electricity consumption. Hence, this paper makes an effort
household[5]. to understand what factors contribute towards the
These equations are represented in statistical models in the determinants. First, we made a discussion on the various
next section. factors in general and how they affect consumption. Then,
by means of conducted survey and collected database,
mapping of various models by means of statistical tools like
V. MODELLING AND ANALYSIS
Minitab is performed. Finally, by observing the models, we
For creation of statistical models, the mathematical hope to share the necessary information which would help
expression consisting various factors linking both to make the consumption determinant as best as possible &
consumption and consumers’ behaviour is considered. also make the nation, a very efficient energy consumer by
The expression is suitable for Linear Regression Models[7]. spreading awareness about avoiding any amount of
In this paper, we observe that the physical measure of unnecessary electric consumption.
consumption is the electric bill received by the individual.
So the response of the particular system will be nothing but VII. REFERENCES
the electricity bill. We perform the regression on every [1] [Suchismita Bhattacharjee, Ph.D.Georg Reichard, Ph.D.“Socio-
individaul paramterer on the above equation i.e. Ai,Bi,Di. Economic factors affecting Individual Household Energy
Consumption : A Systematic Review,” page no. 2-4
Here we show the regression modelling of Ai which
[2] K .V. S. Baba ,S. K. Soonee ,( 2013). Analysing the Electricity
depends on the appliances like AC,Washing Machine etc. in Demand Pattern - a field study, pp. 1-5
MINITAB. To perform the regression in Minitab, required [3] Matam Manjunath, Prashant Singh, Abhilash Mandal, Gaurav Singh
response as stated above that the physical measure of Parihar 2013. “Consumer Behaviour towards Electricity- a field
observing behaviour is by means of electric bill so the study. 4th International Conference on Advances in Energy Research
response in this case responsde will be Electric bill and in 2013”, ICAER 2013, page no. 543-546
[4] Pragyan Dash, “Consumption Pattern Of Electricity In Rural And [8] Giorgio Garau , Patrizio Lecca , Giovanni Mandras , “The impact of
Urban Areas: A Case Study Of Rourkela, Sundargargh District Of population ageing on energy use: Evidence from Italy”, 2013,pg.979-
Odisha” 978
[5] H. Fana, I.F. MacGillb, A.B. Sproula, ” Statistical analysis of driving [9] Mariam Gul ;Dr. Saad A. Qazi ,Waqar Ahmed Qureshi,
factors of residential energy demand in the greater Sydney region, “Incorporating economic and demographic variables for forwecasting
Australia” electricity consumption in Pakistan”.2nd International Conference on
[6] World bank ,“Residential Consumption Of Electricity In India Electric power and Energy conservation system,2011.pg.2-4
,Documentation of data and methodology”, July 2008 [10] Lukas G. Swana,∗, V. Ismet Ugursala , Ian Beausoleil-Morrisonb,
[7] Vincenzo Bianco, Oronzio Manca*, Sergio Nardini,“Electricity “Occupant related household energy consumption in Canada:
consumption forecasting in Italy using linear regression models”, 4 Estimation using a bottom-up neural-network technique”, 2010
June 2009,pg no;1414-1419.