An Optimization Model for the Aggregation of End-user Energy

Management Systems in a Residential Setting

Andreia M. Carreiro

, Carlos Henggeler Antunes

1,2

and Humberto M. Jorge

1,2

INESC Coimbra, Rua Antero de Quental nº 199, 3000-033, Coimbra, Portugal

Department of Electrical Engineering and Computers, Polo II, University of Coimbra, 3030-290 Coimbra, Portugal

Keywords: Energy Management Systems, Energy Box Aggregator, End-user Flexibility, Genetic Algorithms.

Abstract: This paper proposes a model for an aggregator of energy management systems (energy box aggregator -

EBAg) to operate as an intermediary between individual energy management systems (local energy boxes)

and the System Operator / Energy Market capable of facilitating a “load follows supply” strategy in a Smart

Grid context. The EBAg is aimed at using the flexibility provided by end-users of demand-side resources to

respond to system service requirements, via contracts, involving lowering or increasing the energy

consumption in each time slot. The aim is contributing to the balance between load and supply, avoiding

peaks in the aggregate load diagram, and coping with the intermittency of renewable sources. For this

purpose an optimization model for the EBAg has been developed, which is tackled using a genetic

algorithm based approach to deal with the combinatorial characteristics of the model.

1 INTRODUCTION

The efforts to reduce greenhouse gases (GHG)

emissions related to electricity generation have been

leading to a fast increase in the deployment of

renewable generation (European Commission,

2010). However, renewable sources present

characteristics that differ from conventional energy

sources. The power output is driven by

environmental conditions, which are inherently

variable and outside the control of generators and

system operators. They cannot be reliably dispatched

or perfectly forecast, and exhibit significant

temporal variability. As a result, the proper

integration of renewables into the electric grid

presents a major challenge and new tools are

required to ensure the grid reliability.

At the same time, the energy consumption in

European Union (EU) households has been steadily

growing due to the widespread utilization of new

types of loads and the requirement of higher levels

of comfort and services. The electricity consumption

breakdown in EU households was recently

characterized (de Almeida et al., 2001), recognizing

that several end-use loads present some kind of

flexibility; therefore, if properly controlled these

loads can be used as a demand side resource capable

of offering a responsive behaviour (Kowli et al.,

2010).

As far as security of supply is concerned, the

most severe problems due to power intermittence

occur in peak load hours, since most system

resources are already in use and a sudden reduction

of power generation can have critical consequences

on the system reliability. Thus, instead of acting on

the supply side, Demand Response (DR) programs

and technologies have the potential to contribute to

optimize consumption and reduce peak loads, in

(near) real-time. In this way, DR is an enabling

strategy for the successful integration of renewable

energy sources in the electric system, in a

perspective of integrated energy resource

management, involving controlling flexible loads

according to (price and/or emergency) signals from

the grid and end-users’ preferences. In addition, DR

can become a new source of revenue for entities that

“aggregate” this load flexibility (Joo et al., 2007).

In a Smart Grid context, it is expected that the

traditional end-user will become a prosumer (i.e.,

simultaneously producer and consumer) and

dynamic (time-differentiated) electricity tariffs will

be offered (Livengood and Larson, 2009). Therefore,

an in-house demand responsive energy management

system (Local Energy Box - LEB) is required, based

on fully interactive Information and Communication

Technologies (ICT), to help the end-user optimizing

191

M. Carreiro A., Antunes C. and M. Jorge H..

An Optimization Model for the Aggregation of End-user Energy Management Systems in a Residential Setting.

DOI: 10.5220/0004960101910197

In Proceedings of the 3rd International Conference on Smart Grids and Green IT Systems (SMARTGREENS-2014), pages 191-197

ISBN: 978-989-758-025-3

 2014 SCITEPRESS (Science and Technology Publications, Lda.)

the energy use without compromising comfort,

achieving energy savings and satisfying constraints

on the quality of the energy services provided. The

LEB should also enable two-way communication,

between the house and the grid in order to improve

the global performance of the electric power system

(Livengood and Larson, 2009); (Verschueren et al.,

2010). Otherwise, in a scenario of a low price signal

from the grid, all LEB devices would attempt to

achieve benefits for the end-user engaging in similar

actions (e.g., by shedding the same type of loads),

eventually taking no notice of the instability that

could impair the operation of the system, since the

true impact of household consumption arises when it

is summed up over a large number of houses (EU

Comission Task Force for Smart Grid). In this

context, a few recent studies have addressed the

combination of demand and supply sides, to

implement DR programs for the provision of system

services, i.e. the balancing services that are provided

by system operators for ensuring reliable system

operations (EU Comission Task Force for Smart

Grid). These services have been traditionally

provided by generators, which are capable of

adjusting their output rapidly in response to

unanticipated imbalances between supply and

demand. In the Smart Grid context, the provision of

these services by aggregating electricity consumers

using DR programs is becoming an attractive

alternative (Agnetis et al., 2011).

In this context, an aggregator energy

management system (Energy Box Aggregator -

EBAg) is proposed, which is an intelligent mediator

between the end-users (via LEB) and the grid

(System Operator / Energy Market, SO/EM)

allowing the coordination of a large-scale

dissemination of in-house DR devices (Figure 1).

The main aim of the EBAg is to provide system

service requirements, increasing the grid efficiency,

ensuring the required levels of supply security,

reliability and quality of service, taking into account

inputs such as end-user flexibility of operation of

demand-side resources and both grid and load

technical constraints, thus contributing to the

balance between load and supply coping with the

intermittency of renewable sources and avoiding

peaks in the load diagram, while minimizing

overall electricity supply costs (Kowli et al., 2010).

The optimization problem faced by the EBAg

consists in receiving requests from the grid to lower

or increase electricity consumption and allocating

these requests among clusters of LEB while

satisfying technical and quality of service

constraints, in order to maximize its profits. This is a

combinatorial problem and an approach based on

genetic algorithms (GA) has been developed, which

combines features for handling discrete and

continuous variables.

This paper is structured as follows. Section 2

introduces the problem formulation and the related

key concepts, also referring to model and business

strategy and the potential benefits for the end-user

and the grid (SO/EM). Section 3 presents the

optimization model and the algorithmic approach,

which has been used in an experimental case study

presented in section 4. The results of the simulations

are briefly presented in section 5. Concluding

remarks and future work are outlined in Section 6.

Figure 1: EBAg mediating the relation between LEB and

the grid.

2 BUSINESS MODEL AND

STRATEGY

This section presents a framework for analysing the

EBAg role in the electric power system and the

players involved. This comprises the information

that is transmitted to/from the LEB, which interfaces

the end-user directly with the EBAg, and the relation

of the EBAg with the grid (SO/EM).

Nowadays the only direct relation between the

grid and the end-user is with the retailer company,

being associated through a contract for power supply

with specific tariffs (power and energy prices). In

such framework end-users are not generally aware of

the existence of the energy market.

The energy consumption of domestic consumers

presents some flexibility and end-use loads may be

broadly characterized into four categories:

 Shiftable: loads that can be used in another

period of time and therefore can have their

SMARTGREENS2014-3rdInternationalConferenceonSmartGridsandGreenITSystems

192

working cycle anticipated or postponed but not

interrupted (e.g, dishwashers or laundry

machines);

 Re-parameterizable: loads that can have their

control parameters re-set (i.e., air conditioners);

 Interruptible: loads whose operation can be

interrupted during a certain period of time (e.g.,

electric water heaters).

 Non-controllable: loads that cannot be the target

of any type of demand response actions (e.g.,

computers, entertainment);

Based on this end-use load categorization, it is

possible to exploit the load flexibility within each

house. The large-scale deployment of LEB imposes

an essential challenge concerning the coordination

of grid and end-user objectives. I.e., requests from

the grid should be weighed against the flexibility of

end-use loads to be shifted, re-parameterized or

interrupted in a certain period of time. The EBAg

will gather this flexibility from the end-users by

means of each LEB, asking them to adjust their load

profile by offering a remuneration scheme specified

in a contract, which is typically variable along the

day with some relation with market prices.

Thereby, the EBAg is able to sell the flexibility

gathered by making offers to the grid according to

its requests in each time slot (i.e., increase or

decrease a certain amount of load), with the aim of

optimizing its own profits and offering benefits to all

entities involved (increasing retail profits,

decreasing consumption costs). Figure 1 and figure 2

depict the architecture adopted.

Figure 2: EBAg global architecture.

The EBAg sees its associated end-users grouped

into clusters, which contain end-users with similar

characteristics. The normal energy profile of the

cluster, i.e. in the absence of a flexibility request,

can be represented as a baseline load profile curve.

The EBAg knows the baseline load profile of

clusters as well as the possible load profile responses

associated with each request.

The time variable remuneration schemes are

known, i.e. the €/kW value paid by the grid to the

EBAg and by the EBAg to the clusters. The

incentive to remunerate the customer to participate

in DR programs is a significant issue for the success

of this type of programs (Quinn et al., 2010). An

adequate EBAg business model should be designed

to attract and maintain residential end-users under

demand management contracts, in which end-users

receive a reward to offer a certain amount of load

flexibility to the EBAg. Customers sign up for

programs depending on the benefits they derive in

the form of upfront payments, i.e. when demand

discounts and interruption payments exceed their

perceived cost of interruption or consumption

shifting (Fahrioglu and Alvarado, 2000).

The remuneration scheme presently implemented

is based on the electricity price at OMIP (MIBEL -

Iberian EM) (OMIP). The remuneration schemes are

in place when the EBAg requests for consumption

change are satisfied. The EBAg may be an Energy

Service Company (ESCO) that buys and manages

the end-user’s flexible load and sells it to provide

system services to the SO/EM.

Using the information gathered from the grid

(SO/EM) and the individual LEB aggregated in

clusters, a model to maximize the EBAg profits is

developed that is then tackled using a GA approach.

3 OPTIMIZATION MODEL

The model aims at determining the best matching of

the requests by the grid and the flexibility specified

by the LEB. Binary variables denote whether or not

the EBAg is able to gather load flexibility from the

clusters and offer it to the grid, in each time slot.

Continuous variables represent the corresponding

amount of energy managed.

Indices:

c - 1,2, …C. Identifies the cluster, where C is the

number of clusters associated with the EBAg. Each

cluster gathers a set of end-users (LEB).

f - 0,1,… F. Identifies the request that the EBAg

sends to cluster c; f=0 means that no request is sent

by the EBAg. The EBAg congregates the flexibility

offered by each cluster. A response function is

associated with request f.

t = 1,2,…T. Identifies the time slot. A time

resolution of 15 minutes is considered, thus having

T=96 time slots in one day.

Coefficients:

- Reward paid by the grid to the EBAg for the

load flexibility provided, in each time slot t.

AnOptimizationModelfortheAggregationofEnd-userEnergyManagementSystemsinaResidentialSetting

193

- Reward paid by the EBAg to the clusters for the

load flexibility provided, in each time slot t.

ctf

- Load flexibility (kW) response function (i.e.,

consumption reduction) of cluster c in each time slot

t, associated with request f.

Lmax

- Maximum value of flexible load (kW) in

each cluster c, in each time slot t.

Pmin

– Minimum amount of load (kW) the EBAg

can offer to the grid, in time slot t.

– Consumption reduction request from the grid

(kW), in each time slot.

Decision variables:



{0,1}. Represent the decision offers of the

EBAg to the grid, in each time slot t. If a

= 1 then

the EBAg is able to present an offer to the grid in

time slot t, according to the grid request R

Otherwise, a

= 0.

– Continuous variable representing the quantity of

power (kW) that the EBAg is capable to offer to the

grid, in each time slot t: P

> 0 when a

= 1.

ctf



{0,1}. Represents the decision of sending an

action/signal to each cluster c in time slot t

responding to request f. If b

ctf

=1 then the EBAg

obtains the load flexibility response function L

ctf

Otherwise, b

ctf

=0, f = 0 and the cluster is in the

baseline load profile.

– Continuous variable expressing the overall

flexibility for all clusters c associated with the

EBAg. It assumes only positive values since it

represents a consumption reduction, in each time

slot.

The objective function consists in maximizing

overall EBAg profits.





.









1



1







The first term accounts for the gains from selling the

load flexibility to the grid, while the second term

accounts for the cost that the EBAg has to pay to the

clusters for their participation.

Subject to:

1. Only one action/signal request f for each

cluster c, in each time slot t:





1,∀



,∀





0



2. In each time slot, it is assumed that the load

flexibility response function (L

ctf

) has an upper

bound of the maximum value of flexible load

(Lmax

) in each cluster c. The overall load

flexibility in each time period is obtained:











0





,∀





1



3. The amount of power offered to the grid is

bounded by a minimum offer and the overall load

flexibility response gathered from all clusters,

whenever the EBAg is able to present an offer to the

grid (a

=1):





















,∀





4. Whenever the EBAg receive a request from the

grid, R

, the offer to the grid, P

, cannot exceed the

requested flexible load in each time slot:









,∀





4 ALGORITHMIC APPROACH

A GA approach has been developed to tackle this

model, namely to cope with its combinatorial nature.

GAs efficiently exploit the search space using

genetic operators (selection, crossover and mutation)

to create new individuals (solutions) with expectedly

improved performance. The evolutionary search

process balances two main procedures: exploring the

whole search space and exploiting the

neighbourhoods of promising solutions. A GA for a

particular problem has generally the following

components: a representation (encoding) of potential

solutions to the problem; a procedure to create an

initial population of solutions, generally taking into

account model constraints; a fitness function to

evaluate the quality of solutions, possibly including

a penalty term associated with constraint violations;

genetic operators that make the solution population

evolve, trying to preserve its diversity; values for the

parameters (population size, probabilities of

applying genetic operators, etc.).

The model and the GA approach have been

implemented in Matlab. A problem-specific

repairing process of infeasible solutions obtained in

each generation has been designed. The size of the

population is 100 individuals. Binary tournament has

been used as the selection method. Elitism is

considered by retaining the 3 individuals with better

fitness, which go into the next generation population

without being subject to the operators in order to

guarantee that the best solutions are not lost. One-

point crossover operators have been designed taking

into account the binary/continuous nature of

SMARTGREENS2014-3rdInternationalConferenceonSmartGridsandGreenITSystems

194

decision variables. These operators ensure the

consistency relations to be satisfied between the

corresponding sets of binary and continuous

variables. Mutation is a genetic operator used to

maintain diversity in the population and promote the

exploration of the search space. The mutation

operator is also tailored to the binary and continuous

variables ensuring their mutual consistency after

mutation is carried out. For binary variables a bit flip

is done with a certain probability, and for continuous

variables a change is made within a given interval

centered in the present value using a uniform

distribution. After applying the mutation operator,

infeasible solutions are subject to a repairing

process. The evolutionary process ends after 100

generations, which has been set after

experimentation.

5 CASE STUDY

Experiments have been carried out, based on real

consumption data gathered through audits. These

data provided a realistic basis to the specification of

clusters, energy prices, baseline load profile, load

flexibility response functions, bounds for load

flexibility offered by each cluster, and bounds for

the offers the EBAg can make to the grid.

Figure 3: Maximum and minimum consumption of a

cluster, in each time slot.

The data for these experiments have been

obtained from a sample of 30 users of the Cloogy

technology (CLOOGY), which is an energy

management solution that allows monitoring and

controlling energy consumptions in households. The

energy consumption was monitored during one year,

January 2012 to January 2013, with a time resolution

of 15 minutes. An inquiry has then been made in

order to create the house energy profile of each end-

user (baseline load profile).

Clusters have been defined based on specific

features of each house, namely geographical

distribution, contracted power, number of

inhabitants in the house, number and type of

appliances and house typology (see table 1).

The experiments herein reported have been made

with a reduced data set in order to moderate the

computational effort. Each cluster is composed of

100 users (using the average of 30 end-users actually

audited taking into account their consumption

profile). Table 1 displays the daily consumption

(kWh), the average power, the maximum power and

the minimum power in each time slot for each

cluster.

Table 1: Examples of clusters.

The electrical signature of shiftable loads has

been monitored for different machine (laundry and

dishwashing) models and programs. These electrical

signatures have been used to derive the load

flexibility response functions associated with each

cluster.

The evolution of the GA can be seen in figure 4,

where the upper line displays the evolution of the

best solution in the population, and the lower line

displays the average performance of the population.

Figure 4: Representation of the GA performance, with 100

generations.

In general, after 60 generations the best solution

is achieved. The best solution obtained gives an

objective function value (EBAg profit) of

10.007€/day.

The load flexibility offered by each cluster to the

EBAg is illustrated in Figure 5, and Figure 6

represents the load diagram of the cluster before

AnOptimizationModelfortheAggregationofEnd-userEnergyManagementSystemsinaResidentialSetting

195

(baseline load profile) and after (flexible load

diagram) the AG execution.

Figure 5: Representation of amount of load flexibility

response function, in each time slot, provided by each

cluster (example of Cluster 2).

Figure 6: Representation of the baseline load profile and

the flexible load diagram of the Cluster 2.

The global load flexibility provided by all

clusters to the EBAg (Dt) is displayed in Figure 7.

Figure 7: Representation of the amount of load flexibility

offered by all clusters.

6 CONCLUSIONS

This paper presents the conceptual and operational

framework of a model for a new entity - EBAg, that

uses the load flexibility provided by each end-user

(via LEBs), responding to the grid requests to

facilitate a load follows supply strategy in a Smart

Grid context, with potential benefits for all

participants involved. The role of the EBAg is

twofold: it makes the most of demand responsive

loads according to in-house load flexibility and it

provides system services that contribute to improve

the system operation.

The optimization model from the EBAg

perspective presents combinatorial characteristics

and an approach based on GAs is proposed to deal

with it. Work is underway to deal with the dynamic

nature of the problem, uncertainty associated with

several parameters, and multiple objectives

(economic, quality of service, fairness).

ACKNOWLEDGEMENT

This work has been partially supported by

FCT under project grants PEst-

OE/EEI/UI0308/2014 and MIT/SET/0018/2009.

Also, it has been framed under the Energy for

Sustainability Initiative of the University of Coimbra

and supported by Energy and Mobility for

Sustainable Regions Project CENTRO-07-0224-

FEDER-002004.

REFERENCES

European Commission, “Energy 2020: A Strategy for

Competitive, Sustainable and Secure Energy”, SEC

(2010) 1346, Brussels, 2010.

de Almeida, A., Fonseca, P., Schlomann, B., Feilberg, N.,

(2001). Characterization of the Household Electricity

Consumption in the EU, Potential Energy Savings and

Specific Policy Recommendations. Energy &

Buildings, Vol. 43, Issue 8, pag. 1884-1894.

Kowli, A., Negrete-Pincetic, M., Gross, G., (2010). A

successful implementation with the Smart Grid:

Demand response resources, Power and Energy

Society General Meeting 2010, Minneapolis (USA).

Livengood, D. and Larson, R., (2009). The Energy Box:

Locally Automated Optimal Control of Residential

Electricity Usage, Service Science. 1 1-16.

Verschueren, T., Haerick, W., Mets, K., Develder, C., De

Turck, F., Pollet, T., (2010). Architectures for smart

end-user services in the power grid, Network

Operations and Management Symposium Workshops

2010 IEEE/IFIP, vol., no., pp.316-322.

EU Comission Task Force for Smart Grid (2010). Expert

Group 1: Functionalities of smart grids and smart

meters, December 2010, Brussels.

Joo, J. Y., Yoon, S. H. T., and Choi, J. W., (2007). Option

Valuation Applied to Implementing Demand Response

via Critical Peak Pricing, IEEE Power Engineering

Society General Meeting, 24-28.

Quinn, C., Zimmerle, D., Bradley, T. H., (2010). The

SMARTGREENS2014-3rdInternationalConferenceonSmartGridsandGreenITSystems

196

effect of communication architecture on the

availability, reliability, and economics of plug-in

hybrid electric vehicle-to-grid ancillary services,

Journal of Power Sources; 195: 1500-1509.

Fahrioglu, M., Alvarado, F. L., (2000) "Designing

incentive compatible contracts for effective demand

management, Power Systems, IEEE Transactions on,

vol.15, no.4, pp.1255,1260, Nov 2000.

Agnetis, A., Dellino, G., Pascale, G., Innocenti, G.,

Pranzo, M., Vicino, A., (2011) Optimization models

for end-user flexibility aggregation in smart grids: the

ADDRESS approach, 2011 IEEE 1

International

Workshop on Smart Grid Modeling and Simulation

(SGMS).

http://www.cloogy.pt/pt/

http://www.omip.pt/

AnOptimizationModelfortheAggregationofEnd-userEnergyManagementSystemsinaResidentialSetting

197