Professional Documents
Culture Documents
Abstract—This paper investigates load control and demand re- supply from renewable energy systems (RES) and microgrid in-
sponse in a smart grid environment where a bidirectional commu- stallations, it is important to develop scheduling and control al-
nication link between the operator and the smart flexible devices
supports command and data flow. Two control schemes are inves- gorithms to balance the operation of the smart grid and minimize
tigated that can provide energy management, taking into account costs from imported energy or even minimize blackout events.
user’s comfort, via binary on-off policies of the smart flexible de- In order to guarantee a reciprocal fair management scheme, the
vices. A dynamic control algorithm is introduced that considers
real time network characteristics and initiates command flow when control and scheduling algorithms need to take into account
critical parameters exceed predefined thresholds. To sustain fair- users’ comfort and profits as well as operator’s gains.
ness in the system, priority based and round robin scheduling al- Most of the papers found in the literature deal with optimiza-
gorithms are proposed. A continuous control algorithm is also ex- tion or scheduling algorithms that meet certain criteria of energy
plored to define the higher bounds of energy savings. To quan-
tify the discomfort of users that participate in this type of ser- management but are targeted only to the user or the operator
vices, a heuristic consumer utility metric is proposed and mea- side, neglecting fairness and comfort issues. In [3] scheduling
surements with a flexible device (air conditioning unit) are per- algorithms for the offline and online problem of non-preemp-
formed to model empirically possible time intervals of the control
scheme. Reciprocal fair energy management schemes are investi- tive and preemptive scenarios are investigated. The goal is to
gated being both operator and user centric. It is shown that great guarantee cost minimization from power tasks with user ori-
energy and cost savings can be achieved providing the required de- ented time thresholds. A satisfiability investigation of elastic
grees of freedom to the smart grid to self-adapt during peak hours.
demand in the smart grid is presented in [4]. The authors study
Index Terms—Demand response, home energy management,
system behavior under uncertainties. In [5] the demand response
load control, scheduling, smart grids, smart sensors.
problem is addressed to provide a fault tolerant operation of a
microgrid using multi-agent algorithms. User disruption under
I. INTRODUCTION load control is addressed in [6]. Scheduling algorithms that face
the demand response problem and are based on forecasted elec-
Fig. 3. The average user activity curve over a summer period [21]. Fig. 4. Measurements of a typical flexible load. An air-condition unit was mea-
sured using a plug wireless sensor.
TABLE I
STATES OF OPERATION OF SMART DEVICES (4b)
The cost and the load assumed to present a convex relation [3]
described by (5).
(5)
(1a)
KOUTITAS: CONTROL OF FLEXIBLE SMART DEVICES IN THE SMART GRID 1337
(6c)
thus minimizing data exchange between the agent and the
supervisor. On many occasions the CCA algorithm needs to
be performed at different time periods during the day, thus the The decision variables of the problem are the binary commands
external control scheme is more preferable. A characteristic (states of operation) transmitted from the smart grid to the agent
example is the reduction of power load during high electricity (6b). The constraints incorporate comfort issues, which are de-
price time periods. During CCA the supervisor continuously noted in the first and the last line of (6c) and the active duration
sends commands to switch on/off the flexible loads at every of the user in the system which is described by the second line
time interval and . Taking into account queuing of (6c).
theory and by modeling as the “wanted” task the switch off Threshold Algorithm (TA): The TA algorithm is a dynamic
request , the CCA algorithm can be considered similar demand load control algorithm that takes into account real time
to a service system. The first parameter models the load, and capacity, as shown in the decision vari-
arrival time distribution (for our case is memoryless and is ables of Fig. 6. The aim is to try and keep the system’s load, at
written as ), the second parameter models the service time any time instance , under a given capacity threshold , taking
distribution (which is again memoryless- ) and the third pa- into account comfort and fairness issues. It must be stated that
rameter models the number of servers in the system that serve the TA algorithm can also be used for the time-of-use price sce-
the requests (they are infinite to model the case of continuously nario. For that case a price threshold can be defined by the su-
served requests ). In other words, the server of the system is pervisor or the agent of the system and the algorithm will try to
modeled to be always free to satisfy the tasks . The term keep costs under the predefined threshold. On/off actions are
server is used to describe a physical CPU unit or application performed only when the thresholds are exceeded. In a sim-
software in the controller (supervisor or the agent) and decides ilar to the CCA approach, the system can be considered as a
which requests to be served. The infinite number of servers is or service
hypothetical and one could also consider the case of one server system. The forth parameter models the service discipline, the
with zero service time of the tasks. For the CCA case, the states fifth parameter the waiting space of the queue and the sixth pa-
of operation of the flexible devices are modeled as rameter the population size. In this case we have one server with
zero service time, infinite waiting space, and population and ser-
vice discipline that can be either general independent or based
on priority criteria similar to shortest job next (SJN). The pri-
ority criteria reflect the fairness issues that are discussed in the
The block diagram of the control/scheduling algorithm
next paragraph of the paper. The server is assumed to become
is shown in Fig. 6. There are active users which are
free only when the condition of (7a) is not met. The smartgrid
pushed in the controller. The controller, according to the
sends to the controller jobs that are satisfied according to
time counter of each user and the decision variables
the priority criteria. When request is served, it means that
and , decides which users will be switched on or off.
flexible device of user is switched off and thus .
users are switched off if
For the TA case, the flexible devices can perform transitions be-
and - users are on if where
tween all possible states and these are defined as
and is the
identifier of the switch off command. Therefore, for the CCA
algorithm it is .
The objective of the CCA algorithm is to minimize energy
consumption according to the following set of equations: The objective of the TA algorithm is described according to
the following set of equations:
(6a) (7a)
(6b) (7b)
KOUTITAS: CONTROL OF FLEXIBLE SMART DEVICES IN THE SMART GRID 1339
responsible for the high power loads of the system and need
to participate more in the on/off strategy, compared to others.
In this paper, fairness is introduced in three different approx-
imations: in a Round Robin scheduling scheme, in a Highest
Power Next (HPN) priority scheme, and in a Reciprocal Fair
Management (RFM) approximation. Both the Round Robin and
the HPN scheme incorporate sorting algorithms and thus they
present a computational complexity of , where
the number of active users.
Round Robin: This is the case where the multi level queue of
is given priority according to a Round Robin fashion. The
scheduler switches off the first available user and then places
the user back to the queue until all users that have not been
(7c)
switched off are “served.” In this way, all flexible devices are
expected to be switched on/off approximately equal number of
The decision variables for this case are the same with (6b).
times meaning that (Algorithm 1).
The constraints have changed and now take into account switch
Highest Power Next (HPN): This is the case where the multi
off or maintenance states of the flexible devices, which are de-
level queue of is given priority similar to the SJN. For
scribed by the first line of (7c). The TA algorithm might become
the HPN the flexible devices that operate within users’ premises
similar to the CCA control policy if and only if .
that present high background power loads, are pushed
first in the queue. This algorithm can be considered similar to
Algrothim 1: ROUND ROBIN AND HPN FAIRNESS
a load balancing technique where the goal is to try to equalize
commands for Round Robin algorithm only the power loads of all users during the time period where the
create , (ST-Switch off Time) total load of the system exceeds the threshold. With the HPN
algorithm the number of received on/off commands follow a
commands for HPN algorithm only linear relationship with the average load of the user
create , (MP-Mean Power) (see Algorithm 1). A reward priority algorithm
common for both Round Robin and HPN algorithms for load shifting applications can be found in [23].
while Reciprocal Fair Management (RFM): The RFM algorithm is
Loop not a scheduling algorithm but mainly a constraint that guaran-
tees that the user that participates in the management procedure
create
will benefit compared to the NC case. In this way, both the op-
erator and the user sides present energy savings and thus cost
savings. This means or .
Consumer Utility: Consumer utility, in economic theory, rep-
resents the satisfaction of the user, relative to a given value, by
using a given service. For the purpose of our investigation, a
heuristic metric, the consumer disutility (DU) was used to model
the dissatisfaction of user that receives on/off commands from
if true the smart grid. It was modeled by taking into account the total
continue next amount of time the users were in off mode over the time that
else these users would be on for the NC case compared to the CCA
scheme. In a mathematical form it follows :
next Loop
end
end while
C. Fairness Issues
By letting external parties control and schedule flexible de-
vices in a user premise, creates fairness and comfort issues that (8)
need to be taken into account by operators to sustain the effec-
tive penetration of services. The CCA and TA policies reduce
the total energy consumption of the system, compared to the NC where denotes the set of users that have received at least one
policy but there are many occasions where specific users might off command .
be forced to consume extra energy compared to the NC case.
This can be easily understood from the power profile presented D. Analytical Approximation
in (1a). According to the parameters and the It is not always feasible to keep the total load of the system
number of received on/off commands the control policy below the maximum capacity, similar to condition in (7a). Ac-
might cause excess energy consumption. Furthermore, fairness cording to the simulation parameters and the
issues arise from the fact that users with high are more characteristics of the users’ devices there
1340 IEEE TRANSACTIONS ON SMART GRID, VOL. 3, NO. 3, SEPTEMBER 2012
Fig. 7. Switch off time interval and number of switch on/off activations
as a function of number of users in the network. Solid line with asterisks
represents the user centric feasible solution.
Fig. 8. User state of operation during the day. TA algorithm. Fig. 9. Load and cost savings of the CCA and the TA algorithms relative to the
NC algorithm for the HPN and the Round Robin fairness case.
off interval is higher than the allowed margins. Thus, dis-
comfort becomes critical. The following conclusions can be derived. First of all, the load
and cost savings are independent on the type of fairness used
IV. SIMULATION RESULTS at the scheduling scheme. This is expected since the smart grid
switches off the required number of flexible devices to achieve
The scenario under investigation comprises users the goals (demand response) under the given constraints. The
with average load (nonflexible devices) described by uniform TA algorithm starts to present savings after the time where the
randomly distributed values between and total power approaches the threshold. Load savings of approxi-
W and kW to model houses and of- mately 18% are observed, resulting to cost savings of approxi-
fice buildings in a typical urban environment). For the opera- mately 33%. On the other hand, the CCA algorithm initiates the
tion of the flexible devices each user is characterized by uni- on/off control scheme when the user is active for a period greater
form randomly distributed values hrs, than independently to the threshold margin . This has as
hrs with to a consequence of more load and cost savings to be achieved that
satisfy (9a), hrs, kW, and kW. are almost constant with time (21% and 36% respectively). De-
These values can model typical operation of air-conditioners, spite the fact that the CCA algorithm presents high savings for
ovens, or water heaters. the operator, users might become dissatisfied with the system
The users are served by a local generation system that has a since they are continuously pushed to their discomfort limits.
capacity 35% higher (safety margin) than the needs of nonflex- Fairness Issues: The HPN and Round Robin fairness algo-
ible power demands at peak hours meaning that rithms are presented in Fig. 10. The number of received on/off
commands is presented as a function of the users’ average
power . For the simulation results, it was assumed that
all users connect to the system simultaneously and it was set
and kW. These values model the case that
with indicating the maximum number of simultaneously ac- all users are activated simultaneously, resulting to peak hour
tive users at peak hour. All simulation results are averaged over characteristics and so they are all forced to participate in the
500 independent runs of the algorithms to capture a great diver- control scheme during their activation period. These settings are
sity of possible scenarios that can be met in real life. used to gain a clearer picture for the comparison of the two fair-
Change of Operational States During a Day: The process of ness algorithms.
the algorithm for the TA algorithm is presented in Fig. 8. Users The following observations are derived. The HPN fairness al-
enter the system and become active (reduce number of users gorithm yield users with high power needs to be forced to switch
with and increase number of users with ). off their flexible devices more times than the users with lower
When the total power exceeds the threshold, the flexible devices power needs. Since the RFM case is always satisfied, anytime a
from certain users need to be switched off . This oc- user switches off the flexible loads it means that he/she reduces
curs during the peak hour traffic where the highest power load his/her overall power consumption. In that way, users with high
is met. When the time of operation of the flexible device ex- power loads are switched off more times, providing load bal-
ceeds the threshold then the device exits the system and thus, ance in the system. The fairness of this condition can be ex-
. plained by taking into account the fact that the users with high
Cost and Power Savings: The cost and load savings com- power needs are more responsible for the excess of the system’s
pared to the NC algorithm are presented in Fig. 9. The cost load thresholds. On the other hand, the Round Robin fairness
was assumed as a convex function of the instantaneous load de- scheme is independent to the average power of the users, as ex-
scribed by (5). The CCA and the TA algorithm for the HPN pected. For that case, all users are forced to switch on/off equal
and the Round Robin fairness are compared to the NC case. number of times providing another type of fairness policy in the
1342 IEEE TRANSACTIONS ON SMART GRID, VOL. 3, NO. 3, SEPTEMBER 2012
Fig. 11. Operator Centric graph) Probability to exceed given threshold and
percentage of time exceeding relative to NC, for variable capacity limits.
User Centric graph) User energy saving and consumer disutility. It was assumed
that satisfying RFM.
Fig. 10. Number of switch on/off commands as a function of users’ average-
power load for the HPN and the Round Robin fairness (TA algorithm).
extra load in the system to increase the number of switch on/off [10] H. Dagdougui, R. Minciardi, A. Quammi, M. Robba, and R. Sacile, “A
transitions. It can be seen as a load balancing algorithm that is dynamic decision model for the real-time control of hybrid renewable
energy production systems,” IEEE Syst. J., vol. 4, no. 3, pp. 323–333,
trying to keep “almost” equal power consumption between the Sep. 2010.
users in the network. The choice of the used fairness depends on [11] S. Tompros, N. Mouratidis, M. Draaijer, A. Foglar, and H. Hrasnica,
the operator’s policy opposite the users. Finally, the consumer “Enabling applicability of energy saving applications on the appliances
of the home environment,” IEEE Netw., vol. 23, no. 6, pp. 8–16, Dec.
disutility showed a decreasing trend with the operator’s capacity
2009.
and the ratio . To achieve an effective penetration of [12] Z. Md. Fadlullah et al., “Toward intelligent machine-to-machine com-
smart grid services in the future, it is important to maintain the munications in smart grid,” IEEE Commun. Mag., vol. 49, no. 4, pp.
ratio to high values and thus provide the required 60–65, Apr. 2011.
[13] C. Gomez et al., “Wireless home automation networks: A survey of
allowable discomfort margins to the control algorithms to adapt architectures and technologies,” IEEE Commun. Mag., vol. 48, no. 6,
to any load change, minimizing costs at both sides. pp. 92–101, Jun. 2010.
[14] A. Stavropoulos, A. Tsiolia, G. Koutitas, D. Vrakas, and I. Vlahavas,
“System architecture for a smart university,” in Proc. Int. Conf. Neural
Netw. (ICANN), Sep. 2010.
ACKNOWLEDGMENT [15] Th. Stavropoulos, D. Vrakas, A. Arvanitidis, and I. Vlahavas, “A
system for energy savings in an ambient intelligence environment,” in
The author would like to thank Mellon Energy and Kimatica Proc. ICT-GLOW 2011.
for providing important feedback and Prof. Leandros Tassiulas [16] A. P. Meliopoulos et al., “Smart grid technologies for autonomous op-
and Prof. Ioannis Vlahavas for the fruitful discussions on smart eration and control,” IEEE Trans. Smart Grid, vol. 2, no. 1, pp. 1–10,
Mar. 2011.
grids and algorithms. Finally, the author would like to thank Dr. [17] J. Lloret, M. Gilg, M. Garcia, and P. Lorez, “A group-based protocol
T. Dergiades for the discussions concerning economic aspects for improving energy distribution in smart grids,” in Proc. IEEE Int.
and Mr. Th. Stavropoulos for the measurements on the air con- Conf. Commun., 2011.
[18] [Online]. Available: http://www.beywatch.eu/
ditioning unit. [19] Smart Grid Standards, IEEE P1701–P1705, 2011.
[20] [Online]. Available: http://www.zigbee.org/Standards/Zig-
BeeSmartEnergy/Overview.aspx
REFERENCES [21] [Online]. Available: http://www.smarthouse-smartgrid.eu/
[22] B. P. Rasmussen and A. G. Alleyne, “Gain scheduled control of an air
[1] International Telecommunication Union (ITU) Oct. 2008, pp. 15–64, conditioning system using the Youla parameterization,” IEEE Trans.
report on climate change. Control Syst., vol. 18, no. 5, pp. 1216–1225, Sep. 2010.
[2] L. Chen, N. Li, L. Jiang, and S. H. Low, “Optimal demand response: [23] A. Molderink, V. Bakker, M. G. C. Bosman, J. L. Hurink, and G. J. M.
Problem formulation and deterministic case,” in Control and Optimiza- Smit, “Management and control of domestic smart grid technology,”
tion Theory for Electric Smart Grids. New York: Springer, 2011. IEEE Trans. Smart Grid, vol. 1, no. 2, pp. 109–119, Sep. 2010.
[3] I. Koutsopoulos and L. Tassiulas, “Control and optimization meet the
smart power grid: Scheduling of power demands for optimal energy George Koutitas was born in Thessaloniki, Greece.
management,” in Proc. Int. Conf. Energy-Efficient Comput. Netw. He received the B.Sc. degree in physics from
(E-Energy), 2011. Aristotle University of Thessaloniki, Greece, 2002
[4] J.-Y. Le Boudec and D. C. Tomozei, “Satisfiability of elastic demand and the M.Sc. degree (with distinction) in mobile
in the smart grid,” in Proc. Int. Conf. Smart Grids, Green Commun., IT and satellite communications from the University of
Energy-Aware Technol. (ENERGY 2011), May 2011. Surrey, U.K., 2003.
[5] Y. Xu and W. Liu, “Novel multiagent based load restoration algorithms He defended his Ph.D. in radio channel modeling
for microgrids,” IEEE Trans. Smart Grid, vol. 2, no. 1, pp. 152–161, from the Centre for Communications Systems Re-
Mar. 2011. search (CCSR) of the University of Surrey in 2007
[6] B. Ramanathan and V. Vittal, “A framework for evaluation of ad- under a full scholarship. Currently, he is a member of
vanced direct load control with minimum disruption,” IEEE Trans. the academic and research staff at the School of Sci-
Power Syst., vol. 24, no. 4, pp. 1681–1688, Nov. 2008. ence and Technology of the International Hellenic University, Greece, where he
[7] M. Parvania and M. Fotuhi-Firuzabad, “Demand response scheduling also works at the Smart IHU project (rad.ihu.edu.gr). Finally, he is a Postdoc at
by stochastic SCUC,” IEEE Trans. Smart Grid, vol. 1, no. 1, pp. 89–98, the University of Thessaly (Dept. Computer Engineering and Telecommunica-
Jun. 2010. tions). His main research interests are in the area of wireless communications
[8] P. Du and N. Lu, “Appliance commitment for household load sched- (modeling and optimization), energy efficient networking, and smart grids. He is
uling,” IEEE Trans. Smart Grid, vol. 2, no. 2, pp. 441–419, Jun. 2011. involved in research activities concerning energy efficient network deployments
[9] T. Kim and H. V. Poor, “Scheduling power consumption with price and design, green IT, and sensor networks/actuators for smart grid applications.
uncertainty,” IEEE Trans. Smart Grid, vol. 2, no. 3, pp. 519–527, Sep. Dr. Koutitas, during his studies, received the Nokia Prize and Advisory Board
2011. Prize 2003 for the best overall performance and best M.Sc. Thesis.