See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/323268084

An Economic Customer-Oriented Demand Response Model in Electricity

Markets

Conference Paper · February 2018

DOI: 10.1109/ICIT.2018.8352340

CITATIONS READS

2 130

5 authors, including:

Amjad Anvari-Moghaddam Josep M. Guerrero

Aalborg University Aalborg University
122 PUBLICATIONS   1,036 CITATIONS    963 PUBLICATIONS   27,466 CITATIONS   

SEE PROFILE SEE PROFILE

Vahid Vahidinasab
Shahid Beheshti University
48 PUBLICATIONS   559 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Power Quality Issues In Microgrids View project

Flexible Electric Vehicle Charging Infrastructure View project

All content following this page was uploaded by Amjad Anvari-Moghaddam on 19 February 2018.

The user has requested enhancement of the downloaded file.

 An Economic Customer-Oriented Demand Response
Model in Electricity Markets
Reza Sharifi, S. Hamid Fathi Amjad Anvari-Moghaddam, Josep M. Guerrero
Electrical Engineering Department Department of Energy Technology
Amirkabir University of Technology Aalborg University
Tehran, Iran Aalborg, Denmark
{reza.sharifi, fathi}@aut.ac.ir {aam,joz}@et.aau.dk

Vahid Vahidinasab
Electrical Engineering Department
Shahid Beheshti University
Tehran, Iran
v_vahidinasab@sbu.ac.ir

Abstract— Consumer choice theory is a branch of customers to optimize their electric energy consumption.
microeconomics. This theory relates to adjusting consumption Numerous DSM techniques have been proposed in the
expenditures and consumer demand curve. Consumer choice literatures. One of these is Demand Response (DR) programs.
science is trying to realize the buyer's decision-making process. DR programs, as defined by U.S. Department of Energy
This science studies customer characteristics, such as behavioral (DOE), are the programs that improve the electricity
criteria, to understand the consumer’s need. The concept of price consumption patterns of industrial, commercial, and residential
elasticity of demand (PED) has also been derived from this customers so as to reduce peak loads and thereby achieve better
theory. In fact, the PED is the percentage of changes in the prices as well as improved network reliability [3]. To make the
amount of demand relative to the price changes. In consumer
power demand more manageable, a DR program can change
choice theory, for each consumer according to behavioral
criteria, a unique response to price changes is considered.
the pattern of electricity usage by reducing the peak load or
Therefore, the consumer demand curve is a unique curve versus shifting the consumption from the peak to the off-peak hours
price changes. In the concept of PED, the elasticity investigation [4]-[5].
is performed only in a single point or over a small interval of the Most existing DR models have constructed by the concept
curve instead of the whole curve which is not suitable. Since most of Price Elasticity of Demand (PED). This concept is extracted
demand response (DR) models have been developed based on this from consumer choice theory that is a branch of
concept, this will also be deemed as a disadvantage for them. In
microeconomics. This theory is related to adjusting
this paper, we propose an economic DR model based on economic
consumption expenditures and consumer demand curve. This
theories and mathematical methods. In addition to abate the
defects of price-elasticity based DR models, the proposed model
science combines cases such as psychology, sociology, social
has the ability to respond to the various consumers with distinct anthropology and economics. Consumer choice is trying to
responses to price changes and can also adjust the consumer's understand decision-making process of buyers, whether in the
demands according to the consumer's preferences during form of one by one or group [6]. PED demonstrates the
different periods of a day. relationship between price and quantity demanded and in fact
the percentage changes in quantity demanded with respect to
Keywords— Demand-side management (DSM), price price changes. Each consumer has a unique reaction to price
elasticity of demand (PED), economic DR model, consumer utility changes and this difference among consumers is identified
function, Cobb-Douglas utility function. according to many parameters such as consumer behavior,
income, etc. [7]. Therefore, the demand curve can be unique
for each consumer. Existing price-elasticity based DR models
I. INTRODUCTION have paid less attention about this matter. In general, the
With the advent of electricity markets, the supply-side behavior of consumers by observing costs of electrical energy
quickly adapted to the new environment but it took longer for can be divided into three main groups [1]. These categories are
the demand-side to become aware of this toughie due to shown in Figure (1). Each group of consumers with regard to
absence of enough information and confronting tools to their behavior can manage the electric energy consumption at
participate imposingly in the electricity markets [1].To tackle different times. The concept of consumption management
this issue, demand-side tried to find appropriate tools which refers to a specific class of electric loads, which have the ability
eventually led to the emergence of a new subject called to transfer between different hours [8]. It should be noted that,
Demand Side Management (DSM) in electricity market. electrical loads categorized into two groups: non-flexible and
Implementation of DSM programs offers many benefits to a flexible loads [2]. The flexible loads are significantly
large number of stakeholders in deregulated energy markets controllable in terms of amount and time of use, such as a
[2]. DSM refers to initiatives and technologies that persuade washing machine, a dishwasher and some heating and cooling

978-1-5386-4053-1/18/$31.00 ©2018 IEEE 1149

equipment [2]. Therefore, the consumer can manage of curve, this method is not an attractive way. For instance, in
consumption according to the price changes. [15], a procedure is proposed with the help of economic models
of demand to evaluate the effects of DR programs on the
electricity market. In this research, PED is intended a preset
constant, whereas, the PED in each period varies due to the
different prices of energy which leading to changes in
consumer behavior. Therefore, a robust DR model requires the
consideration of flexible PED which finite models have
provided this feature. References [16]-[17] provides a DR
model that provides two key features of an effective DR
program: adaptability to different consumers with different
disagreements over the DR program and adjustability to
consumer time preferences. In this paper, using economic
Fig.1. Consumers’ reactions to price changes [1]
theories and mathematical formulas, we are presenting a model
that can make these two features a good DR model.
This topic in economics refers to the concept of substituting
goods meaning that demand for one good increases when other The proposed DR model is implemented according to
goods prices rise, and vice versa. This concept is similar to the Time-of-use (TOU) pricing mechanism. TOU usually divides a
concept of power consumption shift from times with high day into different time periods and proposes different prices for
prices to time intervals with lower prices. So, consumption of each time period. In this model, it is assumed that the consumer
electricity with different prices can be considered as different does not intend to reduce his/her consumption. Instead, he/she
goods that due to the features of flexible loads have the ability can be incentivized to participate in the DR program to
to be substituted with one another. increase his/her profit. Therefore, according to the behavioral
criteria and considering the concept of substitute goods and the
As mentioned above, consumer’s reaction to price changes purpose of increasing profit, the consumer can shift a part of
is denoting the elasticity concept. PED is the percentage his/her energy consumption from peak times with high energy
change in quantity demanded to the price changes that could be price spikes to the off-peak or shoulder times with lower prices.
represented as follows [9]: As a whole, the contributions of this paper could be
summarized as follows:
ΔQ P
EP = . (1) • A new model for TOU-based DR program is proposed
ΔP Q
based on economic theories and mathematical
In this equation, ΔQ is the change in the quantity demanded methods for benefit maximization,
and ΔP is the change in price. It should be noted that the
inverse of the slope of the demand curve (ΔQ / ΔP) is only a • An efficient structure is presented for residential load
part of the equation and its value is negative. Because the price management by considering different levels of
and quantity are in opposite directions, when P increases Q participation in DR programs.
decreases and vice versa. The organization of this paper is as follows. The consumer
Significant researches have been done on consumer theory is introduced in Section II. Also, this section explains
responses to electricity prices [10]. Recently, much works have the mathematical formulation of the proposed DR model.
been dedicated to model the price elasticity in electricity Section III presents a sensitivity analysis to study the effect of
markets [11]-[15]. For example, in [11], a DR model is model parameters on consumer participation in DR programs.
introduced based on consumer behavior with the PED concept. Section IV provides numerical results and shows validity of the
The main idea of this model is economic and psychological proposed model. Finally, Section V concludes the paper by
analysis that accepts the difference between the effects of summarizing the main results and discussing future works.
rewards and punitive programs and argues that reward is a
superior method of creating habit compared to the punishment. II. CONSUMER THEORY
In [12], a responsive load model is designed based on the
concept of PED. The mathematical model is introduced to Consumer theory in the microeconomics examines how
calculate elasticity of any DR program based on the price of each consumer spends the money according to his/her priorities
electricity before and after the implementation of DR in term of the budget limits. Two important tools in this theory
programs. Reference [13] offers an optimized DR model that are utility functions and budget constraints. The interaction of
allows a consumer to adjust the hourly demand in response to these two tools enables a consumer to make an optimal
the hourly electricity price to maximize his/her profits. In [14], decision [18].
authors propose a model to discover the demand elasticity of To express the utility function of consumers, there are
consumers through demand-side bidding by electricity market many functions that can be selected. Cobb-Douglas utility
complex adaptive system and an agent-based model that function is one of the most famous utility functions that
simulates electricity markets. economists extensively use in macroeconomics [19].
As seen on most of DR models is constructed on the basis Maximize U (D1 , D 2 ) = D1a D 21−a 0 < a <1 (2)
of elasticity concept in which only a given point on the demand
curve will be investigated. Since PED varies in different points

1150
 subject to By replacing (12) in (11):
(3)
B = P1D1 + P2 D 2 (1 − a).B
D off − peak = (13)
As mentioned, the reaction of customers to the price Poff − peak
changes, i.e., the degree of participation in DR programs is the
same as the concept of elasticity which is presented in (2) using and replacing (13) in (12):
the elasticity parameter "a". In fact, this parameter is a control a.B
parameter. D peak = (14)
Ppeak
Equation (3) is a constraint, indicating that the cost of
purchasing electrical energy must be equal to the allocated in this context, Dpeak and Doff-peak represent electric power
budget, where B is the total budget of the customer and P1 and consumed in peak, and off-peak period, respectively. To obtain
P2 are the prices of the goods (D1 and D2, respectively). the power consumption at each time interval of the study
period, one must use the following equations:
Here we are dealing with an optimization problem with a
utility function and several constraints. There are many D peak
methods for solving the optimization problem such as Dt peak = Dt ' peak × (15)
D 'peak
Lagrange method which could determine optimal points for
utility functions f(x) with one or more equality constraints hj(x). Doff − peak
Considering a constrained optimization and Lagrange Dt off − peak = Dt 'off − peak × ' (16)
Doff − peak
multiplier method, we can state that:
where D'peak and D'off-peak represent demand in peak, and off-
Maximize f (x ) peak period, respectively before implementation of DR
subject to (4) programs.
h j (x ) = 0 j = 1, 2,3,.....
*
III. SENSITIVITY ANALYSIS
If x is a local maximum, then there will be a fixed constant
such λ j (j = 1, 2,3,.....) that: To find the appropriate range for each parameter, there are
different methods that can be used to do so: (1) trial and error
l (2) sensitivity analysis. In this section, we use the sensitivity
∇f ( x * ) +  λ j ∇h j ( x * ) = 0 analysis to find the range of elasticity parameter "a".
j =1 (5) Sensitivity analysis refers to the impact study of the input
h j (x * ) ≤ 0 ; for all j = 1, 2,..., l variables of a statistical model to output variables [20]. As
To combine these conditions into an equation, we present an indicated in (3), the accepted general range for the elasticity
auxiliary function: parameter is 0< a <1. As shown in Fig.3, by increasing this
parameter in the mentioned range and observing the output, it
L (x , λ ) = f (x ) + λ h (x ) (6) is clear that the amount of consumption in the peak time
increases while opposite happens in off-peak time. So, a
and solve:
consumer can choose this parameter with regard to his/her
∇ x , λ L (x , λ ) = 0 (7) sensitivity toward the price changes. So, if a consumer selects
the small value for this parameter, the participation of that
So, in this way we will have: consumer in the DR program is more, meaning that this
consumer is deemed as a High Flexible type.
− peak + λ ⋅ {B-D peak .Ppeak - D off − peak .Poff − peak }
a 1− a
L = D peak .Doff (8)
dL
− peak − λ Ppeak = 0
a −1 1− a
= a.D peak .Doff
dD peak
a −1 1− a (9)
a.D peak .Doff − peak
 λ=
Ppeak
a −a
(1 − a).D peak .Doff − peak
λ= (10)
Poff − peak
dL
= B − D peak .Ppeak -Doff − peak .Poff − peak = 0
dλ (11) Fig.3. Changes in the elasticity parameter "a" and the amount of
 B = D peak .Ppeak + Doff − peak .Poff − peak consumption in each time period

By comparison of (9) and (10), one can write: In this section, we determine the proper value for this
parameter by making changes to the value of the elasticity
a.Poff − peak parameter "a" and observe the output. As shown in Fig. (4),
D peak = Doff − peak (12)
(1 − a).Ppeak with implementation of a DR program and selecting different
values for "a", the amount of consumption at peak time

1151
changes. Since the purpose of a DR program is to reduce Table.I Price of electricity in different time periods [22]
consumption at peak time, increasing "a" is to some extent Demand level Ppeak Poff-peak
justified that the amount of consumption at peak period is less Time Period 18:00-23:00 Other hours
than the amount of consumption when there is no DR program. Price (cents per kWh) 16 8
Therefore, according to the results depicted in Fig. (3), the
appropriate value is 0 < a < 0.8. Now if the sensitivity analysis The results from the implementation of the proposed model
for off-peak time is also done, we will find similar results as at peak time and off- peak time for various values of elasticity
shown in Fig. (5). parameter "a" according to the sensitivity analysis results are
shown in the Tables (II)- (III).
Table.II Changes in the power consumption during the peak time
Elasticity parameter
0.1 0.2 0.3 0.4 0.5 0.6 0.7
"a"
Dpeak (kWh)
6.54 7.06 7.63 8.30 9.04 9.86 10.84
(Proposed DR model)
Dpeak (kWh)
(Base Case-Without 11.71
DR)

Table.III Changes in the power consumption during the peak time

Elasticity
parameter 0.1 0.2 0.3 0.4 0.5 0.6 0.7
Fig.4. Consumption levels for different rates of elasticity parameter "a" "a"
(peak period)
Doff-peak (kWh)
(Proposed DR 26.67 24.14 25.50 24.91 24.17 23.29 22.32
model)
Doff-peak (kWh)
(Base Case- 24.43
Without DR)

As previously mentioned, a decrease in parameter "a"

would result in an increase in the willingness of consumers to
participate in the DR program. Since at peak time the prices are
higher than the other periods, with the increase of consumer’s
willingness, consumption in this period would decrease as
reported in the Table (II). On the other hand, at off-peak times
due to the lower prices, consumption increases with increasing
Fig.5. Consumption levels for different rates of elasticity parameter "a" consumer’s willingness which is clearly visible in the Table
(off-peak period) (III). As shown in the Table (IV), the electric energy
consumption with DR implementation during the study time
IV. RESULTS AND DISCUSSION period (Dall) is approximately constant for different values of
elasticity parameter "a", and this amount is the same as the
To evaluate the performance of the proposed model, a
demand in the absence of the DR program. Therefore, in the
typical load profile is adopted from [21]. In this load profile, as
presented model, only the levels of electrical energy
shown in the Fig. (6), 24 hours a day are divided into two
consumption have been displaced in response to prices at time
periods. The price of electrical energy for each period is shown
periods. On the other hand, as indicated in the same table, the
in Table (I).
amount of budget allocated to purchasing electrical energy has
decreased with increased willingness to participate to DR
programs.
Table.IV Changes in total demand and budget
Elasticity
0.1 0.2 0.3 0.4 0.5 0.6 0.7
parameter "a"
Dall
33.2 33.20 33.1 33.2 33.2 33.1 33.1
(kwh)
With DR
Budget
3.18 3.22 3.26 3.32 3.38 3.44 3.52
($)
Dall
33.1
Without (kwh)
DR Budget
Fig.6. Load profile used for assessment [21] 3.58
($)

1152
 In Fig. (7), the consumption profile is drawn for all [5] Vahedipour-Dahraie, M., Najafi, H.R., Anvari-Moghaddam, A.,
scenarios in Tables II-III. As it is clear, the peak-time and off- Guerrero, J.M., “Study of the Effect of Time-Based Rate Demand
Response Programs on Stochastic Day-Ahead Energy and Reserve
peak time demand levels can be adjusted by changing the Scheduling in Islanded Residential Microgrids”, Appl. Sci., vol.7, no.4-
values of "a". It should also be noted that the growth of 378, pp.1-19, 2017.
residential electricity demand, along with the advance of smart [6] Mohajeryami, S, Moghaddam, I.N., Doostan, M., et al.: “A novel
grids, offers new horizons for using energy management economic model for price-based demand response”, Electric Power
systems in residential buildings (BEMS). In this regard, all the Systems Research, vol.135, pp. 1-9, 2016.
[7] Yang, W., Yu, R., and Nambiar, M.: “Quantifying the benefits to
equations and algorithms previously provided can be coded consumers for demand response with a statistical elasticity model,” IET
into a BEMS. The level of participation in DR programs (i.e., Generation, Transmission & Distribution, vol.8, no.3, pp. 503-515,
elasticity parameter "a") can be set by consumer in BEMS at March 2014.
any time. [8] Anvari Moghaddam, A., Mokhtari, G., Guerrero, J.M., “Coordinated
Demand Response and Distributed Generation Management in
Residential Smart Microgrids”, Energy Management of Distributed
Generation Systems, Dr. Eng. Lucian Mihet (Ed.), InTech, 2016. ISBN:
978-953-51-4708-4.
[9] Thimmapuram, P.R., Jinho, K., “Consumers' price elasticity of demand
modeling with economic effects on electricity markets using an agent-
based model,” IEEE Transactions on Smart Grid, vol.4., no.1, pp. 390-
397, 2013.
[10] Faruqui, Ahmad, Sanem Sergici, and Ahmed Sharif., “The impact of
informational feedback on energy consumption—A survey of the
experimental evidence,” Energy, vol. 35, no.4, pp. 1598-1608, 2010.
[11] Baboli, P.T., Eghbal, M., Moghaddam, M.P., et al.: “Consumer behavior
based demand response model”, Power and Energy Society General
Meeting, 2012, pp. 1-7.
[12] Moghaddam, M.P., Abdollahi, A. and Rashidinejad, M.: “Flexible
Fig.7. Residential load profile with and without DR programs demand response programs modeling in competitive electricity markets”,
Applied Energy, vol.88, no.9, pp.3257-3269, 2011.
[13] Conejo, A. J., Morales, J. M., Baringo, L.: “Real-Time Demand
V. CONCLUSION Response Mode”, IEEE Trans. Smart Grid, vol.1, no.3, pp. 236-242,
2010.
In this paper, the authors, with the help of mathematical and [14] Thimmapuram, Prakash R., and Jinho Kim.: “Consumers' price elasticity
economic models, proposed a new model for DR programs of demand modeling with economic effects on electricity markets using
based on the TOU method. An appropriate DR model can an agent-based model”, IEEE Trans. Smart Grid, vol. 4, no.1, pp.390-
make ability to the consumers to adjust the consumption levels 397, 2013.
in different time periods and it should also be a compatible [15] Faruqui, A., Hajos, A., Hledik, R.M., Newell, S.A., “Fostering economic
model with all types of consumers with any level of flexibility demand response in the Midwest ISO,” Energy, vol. 35, no. 4, pp.1544-
1552, April 2010.
versus price changes. As shown, in this model, with the help of [16] Sharifi, R., Anvari-Moghaddam, A., Fathi, S. H., Guerrero, J.M.,
the elasticity parameter "a", it is possible to easily adjust the Vahidinasab, V., “An Economic Demand Response Model in
consumption levels at different time periods for each type of Liberalized Electricity Markets with Respect to Flexibility of
consumers. The results proved that the proposed model could Consumers”, IET Gener. Trans. Dist., vol. 11, no.7, pp.4291 – 4298,
effectively enable DR action based on user’s preferences and 2017.
[17] R. Sharifi, A. Anvari-Moghaddam, S. H. Fathi, J.M. Guerrero, V.
help them to improve their utility function which in turn Vahidinasab, “Dynamic Pricing: An Efficient Solution for True Demand
demonstrated the significant advantages of the model over Response Enabling”, J. Renew. Sustain. Energy, vol. 9, no. 6, pp.1-14,
price-elasticity based DR models. As a step toward future 2017.
works, the proposed DR model in this paper can be extended to [18] Jonathan, L., Milgrom, P., “Consumer Theory”, Available online:
the models running based on other pricing methods such as http://www.stanford.edu/~jdlevin/Econ%20202/Consumer%20Theory.p
df (accessed on 12 Sep. 2016).
Real-Time Pricing (RTP) and Critical Peak Pricing (CPP). [19] Arnold, Z., Kmenta, J., Dreze, J., “Specification and estimation of Cobb-
Douglas production function models,” Econometrica: Journal of the
REFERENCES Econometric Society, pp. 784-795, 1966.
[20] Grégoire, A., Jouve, F., Toader, A.M., “Structural optimization using
[1] Sharifi, R., S. H. Fathi, V. Vahidinasab. “A review on Demand-side sensitivity analysis and a level-set method,” Journal of computational
tools in electricity market.” Renewable and Sustainable Energy Reviews physics, vol. 194, no. 1, pp. 363-393, 2004.
vol. 72, pp. 565-572, 2017. [21] Xcelenergy, Residential Daily Load Profiles, Available online:
[2] Sharifi, R., Fathi, S.H., Vahidinasab, V., “Customer baseline load https://www.xcelenergy.com/ (accessed on 09 Oct. 2016).
models for residential sector in a smart-grid environment”, Energy [22] U.S. Energy Information, Available online:
Reports, vol. 2, pp. 74-81, November 2016. https://www.eia.gov/electricity (accessed on 09 Oct. 2016).
[3] “Benefits of Demand Response in Electricity Markets and
Recommendations for Achieving Them - A Report to the United States
Congress Pursuant to Section 1252 of the Energy Policy Act of 2005”,
[available online]: https://eta.lbl.gov/sites/all/files/publications/report-
lbnl-1252d.pdf, Accessed on 09.08.2016.
[4] Anvari-Moghaddam, A. Vasquez, J.C. Guerrero, J.M., “Load Shifting
Control and Management of Domestic Microgeneration Systems for
Improved Energy Efficiency and Comfort”, 41st Annual Conference of
the IEEE Industrial Electronics Society (IECON’15), November 9-12,
Yokohama, Japan, 2015, pp. 96-101.

1153

View publication stats

Powered by TCPDF (www.tcpdf.org)