1 Demand Side Management of Photovoltaic-Battery

2 Hybrid System
3 Zhou Wu, Henerica Tazvinga, and Xiaohua Xia
4 Department of Electrical Electronic and Computer Engineering, University of Pretoria,
5 Pretoria, South Africa e-mail: wuzhsky@gmail.com, henerica.tazvinga, xxia@up.ac.za

6 Abstract
In the electricity market, customers have many choices to reduce electricity
cost if they can economically schedule their power consumption. Renew-
able hybrid system, which can explore solar or wind sources at low cost, is a
popular choice for this purpose nowadays. In this paper optimal energy man-
agement for a grid-connected photovoltaic-battery hybrid system is proposed
to sufficiently explore solar energy and to benefit customers at demand side.
The management of power flow aims to minimize electricity cost subject
to a number of constraints, such as power balance, solar output and bat-
tery capacity. With respect to demand side management, an optimal control
method (open loop) is developed to schedule the power flow of hybrid system
over 24 hours, and model predictive control is used as a closed-loop method
to dispatch the power flow in real-time when uncertain disturbances occur.
In these two kinds of applications, optimal energy management solutions can
be obtained with great cost savings and robust control performance.
7 Keywords: solar energy, renewable hybrid system, distributed generation,
8 demand side management, optimal control

9 1. Introduction
10 Renewable energy (RE) sources, including wind, solar and their hybrid
11 systems, have become attractive options of providing energy globally for rea-
12 sons such as low cost, no pollutant emission, energy security, easy accessabil-
13 ity and reduction fossil fuel consumption [1, 2, 3, 4]. Photovoltaic (PV) array,
14 which is the main technology to convert solar energy into electric power, can
15 be stand-alone installed for providing electricity in some remote areas or be

Preprint submitted to Applied Energy February 16, 2015

16 connected to the grid for selling power generated. Because of instantaneous
17 and unstable nature of solar energy, PV usually works with battery storage
18 to provide continuous and stable power, i.e., the PV-battery hybrid system.
19 Battery storage can reduce the risk of PV’s intermittent power supply, and
20 always ensure demand satisfaction. Generally, grid-connected PV systems
21 without battery storage do not require sophisticated management strategies.
22 Prioritizing use of PV power is the only rule when the PV power is less
23 than the load demand. In contrast, battery storage brings more challenges
24 to energy management, as more complicated scenarios must be considered,
25 such as charging the battery from the grid or PV and discharging when nec-
26 essary. As a result, controllers are required for hybrid PV-battery systems,
27 such that the performance of solar usage can be significantly enhanced and
28 the grid regulation can be improved in terms of safety and efficiency.
29 For grid-connected hybrid PV-battery systems, the changing electricity
30 price, the timing of power transaction, and the mismatch between solar power
31 generation and load demand are main challenges in application [5, 6]. From
32 the perspective of demand side management (DSM), solar energy or grid
33 power may be stored when the PV can generate surplus power or when
34 the grid electricity is inexpensive. The stored energy can be managed for
35 economic usage in future when the electricity price is high over peak load
36 periods, or when the PV power is unavailable [7]. The grid-connected hybrid
37 system with DSM can help customers to reduce electricity cost, and also can
38 help utility to regulate the grid in terms of security and efficiency issues, such
39 as peak shaving, direct load control (DLC), and capacity market programs
40 [8]. Therefore, at both sides of electricity market hybrid systems may intro-
41 duce new opportunities to smart grid but also cause many challenges in the
42 following DSM programs.
43 (1) Peak shaving: it is necessary to decide when and how much to charge
44 the battery from the grid or PV before peak hours, so that power consumed
45 from the grid at peak hours can be reduced to satisfy the requirement of
46 shaving.
47 (2) Direct load control (in which a utility operator remotely shuts down or
48 cycles a customer’s electrical equipment at short notice to address system or
49 local reliability): customers have to control the operation of hybrid system to
50 ensure their demand is satisfied at the shutting time based on the frequency
51 and time of shutting at DLC.
52 (3) Capacity market programs (in which customers commit to respond
53 pre-specified load reduction when system contingencies arise and are subject

2
54 to penalties if they do not curtail power consumption when directed): such
55 programs involve issues such as how to decide on the amount of power stored
56 in the battery, how to use the instantaneous and stored power cooperatively
57 to complete the pre-specified load reduction, and how to minimize the penalty
58 if the customer demand exceeds the pre-specified demand.
59 (4) Time-of-use (TOU, where the electricity price is high in the peak
60 load time and low in the off-peak time): scheduling problems arise, such as
61 determining how to optimally operate the hybrid system in peak and off-peak
62 periods for minimizing electricity cost and satisfying the customer demand
63 as well.
64 It must be noted that the hybrid systems with battery storage may have
65 potential to take part in every DSM program or combined programs, which
66 can help the utility to regulate the grid and help customers to reduce en-
67 ergy cost. For simplicity, this paper will mainly focus on evaluating a grid-
68 connected PV-battery system under the TOU program with contracted sell-
69 ing as an example. It will be answered how customers optimally schedule the
70 hybrid system to earn cost savings with varying prices in the TOU program,
71 and how they manage their consumption to sell surplus power to the grid
72 over peak period.
73 Although storage systems are not common in large generation farms, for
74 residential and small-scale power producers many storage systems (battery,
75 ultra-capacitor and so on) have been incorporated in energy supply systems.
76 Nair and Garimella [9] argued that battery storage systems will have a sig-
77 nificant impact on the small-scale integration of renewable sources into the
78 commercial and residential sectors. For hybrid systems with battery storage,
79 energy management is a vital and difficult issue that has attracted great in-
80 terest among researchers [7, 10]. Many energy management systems (EMS)
81 have been developed for the utility to regulate microgrids and reduce gener-
82 ation cost. Some rule-based strategies were designed for energy management
83 of hybrid systems [11, 12, 13], which can obtain promising but not optimal
84 solutions to ensure practical constraints are satisfied. In [14], a determin-
85 istic planning method was proposed to perform robustly day-ahead power
86 flow scheduling for conventional and renewable generators. To improve the
87 performance of EMS, optimal control is a useful method to schedule power
88 flows of hybrid systems with minimum cost and maximum benefit [15, 16].
89 In [17], an EMS for a virtual power plant was proposed to minimize the elec-
90 tricity generation cost and to utilize renewable energy sufficiently. Authors
91 in [18] presented a dynamic optimal power flow control for power and heat

3
 92 generation scheduling while considering PV generations coupled with storage
93 systems. A flexible battery management system was developed to optimize
94 the duration (hours) of charging and discharging battery for optimal power
95 flow control in distribution networks [19].
96 Beyond existing work, more emphasis should be made on two important
97 issues of renewable hybrid systems. Firstly, most researchers have consid-
98 ered energy management and demand response for large-scale integration of
99 renewable energy at the utility side [20, 21]. There is lack of comprehensive
100 work in consideration of optimal planning and DSM for small-scale hybrid
101 systems at the demand side, because many customers install hybrid systems
102 for stand-alone or back-up usage without any participation of DSM program.
103 DSM can be studied more in appliance scheduling of household [22] than in
104 scheduling of small-scale hybrid system. Secondly, uncertainties within fore-
105 cast errors of renewable energy and demand have been studied for large-scale
106 integration of renewable energy [23], but uncertainties at the demand side
107 are not well evaluated. Most related optimal scheduling methods cannot
108 handle complicated cases when hybrid systems experience external distur-
109 bances; only a few closed-loop control methods have been designed [7, 24].
110 Therefor, it is necessary to model the small-scale hybrid system, to compre-
111 hensively study optimal schedule with DSM over different seasons, and to
112 analyze uncertainty and robustness for the closed-loop control. This paper
113 will be organized to respond to the above two issues.
114 Some remote areas, where customers used to rely on stand-alone hybrid
115 systems for generating power, are being connected to the grid as part of
116 network upgrade. Now a new problem is how to use such installed small-
117 scale system efficiently. Based on our previous work [16], we consider DSM,
118 scheduling and uncertainty handling of the grid-connected hybrid system in
119 this paper. The diesel generator is now excluded, as the power its power
120 generation is less green and more costly than the grid. DSM of the hybrid
121 system is expected to help customers earn some payback and reduce electric-
122 ity cost. Another by-product advantage of DSM is the reduction of emissions
123 by utilization of clean PV technologies.
124 The main contributions of this paper are listed below. Firstly, as an
125 example of DSM, the hybrid system under TOU with power selling is modeled
126 to minimize the electricity cost while matching the customer demand and the
127 PV output. Secondly, optimal control is developed as an open loop method
128 to dispatch power flows of the hybrid system stably and economically. A
129 comprehensive study has been conducted to evaluate different situations over

4
130 weekend and weekday of winter and summer. Thirdly, in case of uncertainties
131 in the PV output and the customer demand, model predictive control (MPC)
132 is applied as the closed-loop control to ensure economic, robust and safe
133 operation of the hybrid system. MPC is a feedback control strategy that uses
134 an explicit model of plant to predict the future response of the plant over a
135 finite horizon. Only “the first part” of the sequence is applied to control at
136 the next state [25, 26]. MPC has been widely used in the closed-loop control
137 for adaptively changing control variables according to external disturbances
138 [26, 27, 28]. MPC is applied in this work because of its capability to explicitly
139 handle constraints and to adjust the power flows when disturbances occur.
140 In this paper, an optimal power flow management algorithm of a grid-
141 connected PV-battery hybrid system is developed. The objective is to min-
142 imize the electricity cost within the DSM framework by optimal power flow
143 control. Literature review is conducted on energy management of stand-alone
144 and grid-connected systems in Section 2. The structure of the grid-connected
145 PV-battery system and its sub-models are described in Section 3. The math-
146 ematic DSM model of the hybrid system is given in Section 4. Some results
147 of the optimal control are discussed in Section 5. In Section 6 based on the
148 steady state model an MPC approach is proposed as the closed-loop control,
149 while the last section is the conclusion.

150 2. Literature review

151 Hybrid renewable energy systems (HRES) have been studied in recent
152 years on both bottom and up levels, such as system design, installation,
153 operation and maintenance. The related studies mainly include issues of
154 modeling, control and optimization at each level. In [29], various sectors
155 in designing and implementation of HRES were comprehensively reviewed,
156 including configurations, criteria selection, sizing methodologies and control
157 & energy management. For stand-alone and grid-connected applications,
158 many control systems have been designed into three main categories, i.e.,
159 centralized [16, 30], distributed [31], and hybrid control paradigms [32]. Due
160 to scope of this paper, energy management methods (planning, scheduling
161 and control) are introduced in this section. Energy management for smart
162 grids have received considerable attention to achieve several targets, such as
163 balancing of generation and load, minimizing the generation cost, minimizing
164 transmission and distribution losses, preventing grid congestion, provision of
165 ancillary services.

5
166 For the stand-alone application, a rule-based power management strat-
167 egy was designed to manage power flows among different energy sources and
168 storage units [33]. A renewable micro-grid including a wind turbine, a solar
169 panel, a fuel cell and a storage battery was studied on the issue of optimal
170 scheduling [34], in which mixed-integer linear programming is used to solve
171 their proposed minimization model of generation costs subject to all opera-
172 tion technical constraints. A priority local control algorithm was developed
173 to gain optimal energy management of system loads and battery storage,
174 and therefore provided better energy efficiency and guarantee energy supply
175 for critical loads [35]. In [16], daily energy consumption variations between
176 winter and summer was considered into scheduling stand-alone HRES. The
177 authors had evaluated operational efficiency of the hybrid system over a 24-
178 hour period and optimal solutions can be found to reduce the corresponding
179 fuel costs. Finally 73% to 77% fuel savings in winter and 80.5% to 82% fuel
180 savings in summer can be achieved by the optimal control method. In [30],
181 a switched MPC method was designed for energy dispatching of the same
182 HRES.
183 For the grid-connected application at the utility side, storage manage-
184 ment, economic load dispatch and operation optimization of distributed gen-
185 erations was simplified into a single-objective optimization problem to design
186 a smart energy management system of micro-grid [36], which was solved by
187 a matrix real-coded genetic algorithm. A methodology capable of evaluating
188 the impact of wind generation and load uncertainties, as well as unexpected
189 generation outages was developed [23], in which an EMS integration frame-
190 work was proposed for power system operation, dispatch, and unit commit-
191 ment. A hybrid power generation system consisting of PV arrays and fuel
192 cells was studied by a model-based optimal approach [24], in which the power
193 generation cost is minimized. A dynamic supervisory control was proposed
194 to regulate a grid-connected hybrid generation system with versatile power
195 transfer for flexible operation and improvement of power quality [37].
196 For the grid-connected application considering DSM, energy management
197 strategies from both the demand side and utility side were developed to meet
198 the electricity demand while minimizing the overall operating and environ-
199 mental costs [38]. By integrating DSM and active management schemes,
200 an EMS was developed for optimizing the smart grid’s operation to better
201 explore renewable energy sources and reduce the customer’s electricity cost
202 [39]. DSM of distributed generation and storage system was studied as a
203 day-ahead optimization problem by a game theory approach in which each

6
204 active user at the demand side selfishly pursues minimal monetary expense
205 for buying/producing electricity [40]. Different demand response programs
206 were studied in DSM of hybrid systems [41, 42]. Optimal power management
207 was studied on grid-connected PV-battery system for joining peak shaving
208 service base on the dynamic programming method [41]. A heuristic-based
209 Evolutionary Algorithm was developed in smart grid for finding a general-
210 ized DSM strategy based on load shifting [42]. Many other computational
211 methods were studied in relative applications. A multi-objective method was
212 applied on a hybrid renewable system for maximizing its contribution to the
213 peak load and minimizing its overall intermittence cost, in which large-scale
214 DSM and DR technologies are also considered [20]. Neural networks were
215 applied to schedule and coordinate distributed generations for active DSM
216 [43].
217 Especially, an interesting application of HRES and DSM is smart build-
218 ing. The impact of DSM strategies in the penetration of HRES is analyzed
219 at some regional buildings [44]. HRES was studied for DSM and an energy
220 production management strategy was designed for building automation [45].
221 HERS was considered in an optimal residential load management strategy for
222 real time pricing demand response programs [46]. A smart home controller
223 strategy was designed to enable consumer economic saving and automated
224 demand side management in domestic environment [47]. The event driven
225 controller was designed for optimally scheduling household appliances by bi-
226 nary linear programming. An autonomous appliance scheduling strategy was
227 designed for household energy management based on HRES and DSM [22].
228 An optimization framework was proposed for integrated analysis of demand
229 response programs with high penetration of plug-in hybrid electric vehicles
230 (PHEVs) and PV from residential customer’s perspective as well as utility
231 company’s perspective [48].

232 3. Description of PV-battery system

233 The hybrid system evaluated in this paper consists of PV arrays and
234 battery bank that are both connected to the grid. The output power of the
235 PV array feeds customers’ demand directly. If the demand is less than the
236 PV’s output, the surplus PV power will be charged into the battery bank.
237 If the demand is larger than the PV’s output, the deficient power will be
238 covered by the battery or the grid. The grid plays an important role in the
239 hybrid system for charging the battery and directly supplying customers with

7
240 electricity. The battery can be charged by the grid in the off-peak period, and
241 then discharged in the peak period to save electricity cost. The grid provides
242 electricity directly when the customer demand cannot be satisfied by the PV
243 and the battery. The schematic of this hybrid system is shown in Figure 1, in
244 which arrows represent directions of power flows in the system. P1 is the solar
245 generation for charging the battery; P2 is the discharging power of battery
246 for load demand; P3 is the grid power for charging the battery; P4 is the grid
247 power for load demand; P5 is the solar generation for load demand; P6 is the
248 battery discharge for selling power to the grid. In the hybrid system, several
249 converters such as direct current/alternating current (DC/AC) and DC/DC
are required for voltage and current matching.

Grid P3 P6 P4

P1 P2
PV Battery Load

P5

Figure 1: Schematic of the hybrid system

250

251 3.1. PV array

252 Each solar array consists of several solar cells to convert sunlight into DC
253 power. The hourly power output of a given area can be simply formulated
254 as:
Ppv = ηpv Ipv Ac , (1)
255 where Ppv is the hourly power output from the PV array; Ac is the size of
256 PV array; ηpv is the efficiency of power generation; Ipv is the hourly solar
257 irradiation incident on the PV array (kWh/m2 ).
258 The hourly solar irradiation incident on the PV array is closely related
259 to time of a day, season of the year, tilt, location, global irradiation, diffuse

8
260 fraction etc. In this study, the simplified isotropic diffuse formula [16, 49] is
261 used as
Ipv = (IB + ID )RB + ID , (2)
262 where IB is the beam component of the hourly global irradiation and ID is the
263 hourly diffuse irradiation respectively. RB is a geometric ratio of the actual
264 irradiation on the tilted plane to the standard irradiation on the horizontal
265 plane.
266 The efficiency of power generation can be modeled in a complicated for-
267 mula, which can be expressed as a function of the hourly irradiation Ipv and
268 the ambient temperature TA as
 
0.9βIpv (TC0 − TA0 )
ηpv = ηR 1 − − β(TA − TR ) , (3)
Ipv0
269 where ηR is the PV generator efficiency that is measured at the referenced
270 cell temperature TR (25◦ C); β is the temperature coefficient for cell efficiency
271 (typically 0.004-0.005 /◦ C); TC0 (typically 45◦ C) and TA0 (typically 20◦ C)
272 are the cell and ambient temperatures at nominal operating cell temperature
273 (NOCT) test conditions; Ipv0 is the average solar irradiation on the array at
274 the NOCT conditions.

275 3.2. Battery bank

276 Being constrained in battery capacity, the state of charge (SOC) changes
277 dynamically owing to possible charge by the PV and grid or possible discharge
278 for customer usage. For a given profile of power generation, customers’ de-
279 mand will mostly affect the SOC of battery. Let t denote time of day (hourly),
280 and Pb (t) denote the SOC of battery at the tth hour. Based on the SOC at
281 the previous hour, the dynamic change of SOC can be formulated as
1
S(t + 1) = S(t) + ηC [P1 (t) + P3 (t)] − [P2 (t) + P6 (t)], (4)
ηD
282 where S(t) is the SOC at the tth hour; S(t + 1) is the SOC at the next hour.
283 ηC ≤ 1 and ηD ≤ 1 are the coefficients of charging and discharging efficiency.
284 According to Eq. (4), the current SOC S(t) can be expressed by the initial
285 SOC S(0) of a day as
t−1 t−1
X 1 X
S(t) = S(0) + ηC [P1 (τ ) + P3 (τ )] − [P2 (τ ) + P6 (τ )]. (5)
τ =0
ηD τ =0

9
286 The SOC of a battery has several constraints, such as the maximal allow-
287 able capacity and the depth of discharge (DOD). The lower bound of SOC
288 S min can be expressed by the DOD as

S min = (1 − DOD)S max , (6)

289 where DOD is the depth of discharge; S max is the maximum capacity of the
290 battery; S min is the minimum allowable SOC of the battery. The SOC must
291 be bounded within the scale [S min , S max ].

292 4. DSM Model of PV-battery system

293 Optimal schedule of the evaluated hybrid system aims to minimize elec-
294 tricity cost within the framework of DSM. In this paper, the TOU program
295 is a typical program of DSM for consideration, in which the electricity price
296 changes over different periods according to the electricity supply cost, for ex-
297 ample a high price for peak load periods, medium price for standard periods
298 and low price for off-peak periods. In our study, the daily electricity price at
299 the target region can be given as
 S
 ρk , t ∈ Tk , Tk = [7, 10)S [18, 20)
ρ(t) = ρo , t ∈ To , To = [0, 6) S[22, 24) S , (7)
ρs , t ∈ Ts , Ts = [6, 7) [10, 18) [20, 22)


300 where rhok = 0.20538$/kWh is the price for the peak load period; ρo =
301 0.03558$/kWh is the price for the off-peak period; ρs = 0.05948$/kWh is the
302 price for the standard period.
303 The proposed DSM model includes three parts. The first part is the cost
304 of buying electricity from the grid, which is used to afford the load demand
305 and charge the battery. The second part is the income of selling electricity
306 to the grid. The third part is the wearing cost of hybrid system. The total
307 cost can be formulated as,
23
X X
J= ρ(t) [P3 (t) + P4 (t)] − rk ρk P6 (t) + Ch , (8)
t=0 t∈Tk

308 where rk = 0.65 is the contracted ratio of the peak price ρk for selling power
309 during the peak load period. Ch is he wearing cost of system during the

10
310 control period, which is formulated as
23
X
Ch = a [P2 (t) + P6 (t)] + 24b, (9)
t=0

311 where a is the coefficient of battery wearing cost and b is the hourly wear-
312 ing cost of other components (a = 0.001, b = 0.002 in this paper). In the
313 objective function, control variables Pi (t) (i = 1, 2, ..., 6, 0 ≤ t < 24) have to
314 satisfy several constraints:
315 (1) PV’s output constraint: The PV’s power for charging the battery and
316 for customers’ instantaneous usage must be less than the PV’s output power
317 generated, which is mainly related to irradiation and ambient temperature.

P1 (t) + P5 (t) ≤ Ppv (t). (10)

318 (2) Power balance constraint: The load demand of customers must be
319 exactly satisfied by the total power of PV array, the grid and the battery as

P2 (t) + P4 (t) + P5 (t) = PL (t), (11)

320 where PL (t) is the load demand over the period [t, t + 1).
321 (3) SOC boundary constraint: The SOC of the battery must be less than
322 the battery’s capacity S max and larger than the minimal allowable value S min
323 as
S min ≤ S(t) ≤ S max . (12)
324 (4) Power flow constraint: For safety and other physical reasons, power
325 flow from each source must be non-negative and less than the maximum
326 allowable value as

0 ≤ Pi (t) ≤ Pimax , (i = 1, 2, ..., 6), (13)

327 where Pimax is the defined maximum power delivered per hour.
328 (5) SOC terminate state constraint: For the convenience of dispatching
329 power over continuous days, the battery should not be used till the initial
330 SOC value is reached. In this model, the termination SOC of the battery
331 must be no less than the initial as SOC

S(0) ≤ S(24). (14)

11
332 It can be noticed that the installation cost is not considered in the model,
333 as the scope is restricted to a discussion of how to control the installed hybrid
334 system in the operational step. Therefore, some important issues related
335 to installation, such as economic analysis and optimal sizing of the hybrid
336 system [50, 51], are neglected in our model. The operational costs of the PV
337 and battery are taken as negligible values for the evaluating period, so they
338 are not incorporated in the model.

339 5. Optimal control method

340 An open-loop optimal control method is used to dispatch the hourly power
341 Pi (i = 1, ..., 6) over a day to minimize the daily electricity cost, Eq. (8),
342 subject to constraints, Eq. (10-14). Because the objective function and
343 constraints are linear, this power flow control problem can be expressed as a
344 linear programming problem as

 Ax ≤ b
min f (x), s.t. Aeq x = beq , (15)
lb ≤ x ≤ ub


345 where f (x) represents the objective function; Aeq and be q are the coefficients
346 related with equality constraints; A and b are the coefficients related with
347 inequality constraints; lb and ub are the lower and upper bounds of variables.
348 These coefficients can be easily deduced according to the proposed model, so
349 the explicit details are omitted here.

350 5.1. Control system settings

351 The system evaluated is originally installed as the PV-diesel-battery hy-
352 brid system for off-grid consumers [16]. Because of enlarged coverage of the
353 grid, the target region is now connected to the grid. The diesel generator
354 has been excluded from the current system, because buying electricity from
355 the grid is cheaper and greener than using the diesel generator for power
356 generation. The sizing of PV and battery bank is based on a sizing model
357 in [2]. The parameters of this system are listed in Table 1. The maximum
358 power delivered on each flow is defined as 5 kW.
359 As customers’ daily demand changes between summer and winter as well
360 as between weekdays and weekends, four cases are evaluated, i.e., weekdays of
361 summer and winter and weekends of summer and winter, to find the optimal

12
 Table 1: Parameters of the PV-battery system
Nominal battery capacity 28.8 kWh
Battery charge efficiency 85%
Battery discharge efficiency 100%
Battery’s depth of discharge 50%
Initial state of charge 16 kWh
PV array’s capacity 7 kW

362 dispatch solution for each case. The load profiles of customers in the summer
363 and winter are calculated based on survey data, as given in Table 2 [16].
364 In this study, we have evaluated meteorological data, global irradiation,
365 diffuse irradiation and ambient temperature over the past few years in the
366 target region. For simplicity, the average output profiles in summer and
367 winter are predicted respectively, as plotted in Figure 2. It can be noticed
368 that power output in summer is larger than in winter over the daytime.
7
winter
summer
6

5
Power(kW)

0
0 5 10 15 20 25
Time(hour)

Figure 2: Profiles of hourly power output of the PV array

369 5.2. Results of optimal control

370 On a winter weekday, without the hybrid system the daily electricity
371 cost would be $4.27. When optimally operating the hybrid system the daily
372 electricity cost is reduced to $1.68. The income of selling electricity is $

13
 Table 2: Demand profiles of four cases
Winter Load (kW) Summer Load (kW)
Time
Weekend Weekday Weekend Weekday
00:30 1.5 1.5 1.5 1.5
01:30 1.5 1.5 1.5 1.5
02:30 1.5 1.5 1.85 1.85
03:30 1.5 1.5 1.95 1.95
04:30 1.5 1.5 1.85 1.85
05:30 1.95 1.65 1.5 1.5
06:30 1.95 1.65 1.65 1.15
07:30 1.65 1.35 1.65 1.25
08:30 1.35 1.35 1.7 1.3
09:30 3.25 3.0 1.75 1.32
10:30 3.25 3.0 1.75 1.35
11:30 2.15 1.95 1.75 1.32
12:30 2.15 1.95 1.25 1.25
13:30 2.15 1.95 1.32 1.32
14:30 2.15 1.95 1.35 1.35
15:30 2.15 1.95 1.35 1.35
16:30 2.15 1.65 1.45 1.45
17:30 1.8 1.65 2.1 2.15
18:30 2.31 3.25 2.4 2.31
19:30 3.81 3.25 3.8 3.25
20:30 2.31 2.31 3.8 3.25
21:30 2.31 2.15 2.0 2.0
22:30 2.31 2.15 1.95 1.95
23:30 1.35 1.35 1.65 1.65

373 3.06. In other words, customers can earn $1.38. In Figure 3(a), the optimal
374 power flows at the customer side are plotted for the winter weekday. The
375 customer side power flows include power flows from battery P2 , grid P4 and
376 PV P5 . During the off-peak period [0,6) and [22,24), only the grid provides
377 power owing to low electricity price. When the PV’s output is sufficient,
378 the highest priority is given to the usage of the PV power as shown in the
379 period [8,16). During the peak load period, the power stored in the battery
380 is used to satisfy the load demand. In Figure 3(b), the power flows at the

14
381 battery side are plotted. The battery side power flows include charging flows
382 from PV P1 and grid P3 , and discharging flows to load P2 and grid P6 . The
383 battery is mainly charged from grid during the off-peak periods. When the
384 PV’s output is larger than the load demand, the excessive power is stored
385 in the battery. Most battery storage is sold to the grid over the peak load
386 period, and the remaining is used to supply the demand over [18,20). It
387 can be noticed that the SOC increases during the off-peak period and the
388 high irradiation period, and decreases during the peak period. The SOC’s
389 boundary constraint and terminate constraint are satisfied.
390 On a winter weekend, without the hybrid system the daily electricity
391 cost is $4.47. When optimally operating the hybrid system, the daily cost of
392 electricity is reduced to $1.88, and the income of selling electricity is $3.11.
393 The net value earned is $1.23. For the winter weekend, the power flows at the
394 customer side are shown in Figure 3(c). It can be noticed that the result is
395 similar to that for the winter weekday. During the off-peak period, customers
396 use the grid power owing to the low electricity price. During certain standard
397 period, although the PV generation can fully satisfy the load demand, the
398 grid power has been used. To store enough power for sale, the battery is
399 not discharged during the standard and off-peak periods. In Figure 3(d),
400 the power flows at the battery side are plotted. Similarly, the battery is
401 charged sufficiently during the off-peak period and this stored power is used
402 during the peak period. During the sufficient irradiation period, the battery
403 is mainly charged by the PV. In this case, constraints related to the SOC are
404 satisfied during the charging and discharging processes.
405 On a summer weekday, without the hybrid system the daily electricity
406 cost is $3.49. When optimally operating the hybrid system, the electricity
407 cost is reduced to $1.32, but the income of selling electricity is $3.18. This
408 means customers can eventually earn $1.83. For the summer weekday, the
409 profiles of power flow at the demand side and the battery side have been
410 given in Figure 4(a) and (b). Because in this case the total demand is the
411 lowest among the evaluated four cases, during the standard period more PV
412 power can be stored in the battery and no grid power is required for charging
413 the battery. Most grid power is used during the off-peak period for satisfying
414 the load demand and charging the battery.
415 On a summer weekend, without the hybrid system the daily electricity
416 cost is $3.99. When optimally operating the hybrid system, the electricity
417 cost is reduced to $1.41. The income of selling electricity is $3.10, which
418 means customers can earn $1.69. As shown in Figure 4(c), the grid power

15
 5 5
P P1
2
4.5 P4 4.5 P
2
4 P 4 P3
5

3.5 3.5 P
6

Power (kW)

Power (kW)
3 3

2.5 2.5

2 2

1.5 1.5

1 1

0.5 0.5

0 0
0 5 10 15 20 0 5 10 15 20
Time (hour) Time (hour)
(a) (b)

5 5
P P
2 1
4.5 P4 4.5 P2
4 P 4 P
5 3

3.5 3.5 P6
Power (kW)

Power (kW)
3 3

2.5 2.5

2 2

1.5 1.5

1 1

0.5 0.5

0 0
0 5 10 15 20 0 5 10 15 20
Time (hour) Time (hour)
(c) (d)

Figure 3: Power flows during winter weekend and weekday: (a) customer side power flows,
i.e., from battery, grid and PV to load, for a winter weekday; (b) battery side power flows,
i.e., charging flows from grid and PV and discharging flows to load and grid, for a winter
weekday; (c) customer side power flows for a winter weekend; (d) battery side power flows
for a winter weekday.

419 is consumed during the off-peak period on the summer weekend. The power
420 flows at the battery side are also shown in Figure 4(d), in which the battery is
421 discharged in the off-peak and standard periods. The results for the summer
422 weekend are almost the same as those for a summer weekday, and their
423 demand profiles are close to each other.
424 In sum, it is observed that by optimal control the hybrid system the
425 monthly income in winter is $40.20, which saves $170.60 for customers. The
426 monthly income in summer is $51.82, which saves $167.52 for the customers.

16
 5 5
P P
2 1
4.5 P 4.5 P
4 2
4 P5 4 P3

3.5 3.5 P
6

Power (kW)

Power (kW)
3 3

2.5 2.5

2 2

1.5 1.5

1 1

0.5 0.5

0 0
0 5 10 15 20 0 5 10 15 20
Time (hour) Time (hour)
(a) (b)

5 5
P P
2 1
4.5 P 4.5 P
4 2
4 P5 4 P3

3.5 3.5 P6
Power (kW)

Power (kW)
3 3

2.5 2.5

2 2

1.5 1.5

1 1

0.5 0.5

0 0
0 5 10 15 20 0 5 10 15 20
Time (hour) Time (hour)
(c) (d)

Figure 4: Power flows during summer weekend and weekday: (a) customer side power
flows, i.e., from battery, grid and PV to load, for a summer weekday; (b) battery side
power flows, i.e., charging flows from grid and PV and discharging flows to load and grid,
for a summer weekday; (c) customer side power flows for a summer weekend; (d) battery
side power flows for a summer weekday.

427 Because load demand in winter is larger than that in summer, the monthly
428 earn in winter is less than that in summer, but the total cost savings are
429 almost the same after optimally scheduling.
430 Over a year period study, load demand and PV output can be forecast
431 frequently on daily or weekly basis, as during a year they are varying largely
432 due to weather and human factors. Different load demand and PV output
433 have significant effects on operation hours and cost saving. Although we
434 have not given actual statistics for a whole year, based on the summer and

17
435 winter results the total income of selling electricity is approximately $552.12
436 and the total cost saving is approximately $2028.72. For a newly installed
437 PV-battery hybrid system, the capital cost of installation is $12500, and its
438 yearly operation and maintenance cost is $135. The payback period is about
439 6.6 years. For an existing PV-battery hybrid system, the capital cost of
440 installation is omitted and the payback period of modification must be less
441 than 1 year.

442 6. Model predictive control method

443 In the optimal control, customer demand loads during weekdays and
444 weekends are forecast as the average values for winter or summer; the profile
445 of PV output is also forecast as the average values for winter and summer.
446 In fact, the customer demand and the PV output may be disturbed from the
447 forecast values. In this section, such divergence is evaluated as system dis-
448 turbances on the demand and the PV output. The linear state-space model
449 will be deduced from the hybrid system model. From Eq. (11), the following
450 equation can be obtained

P4 (t) = PL (t) − P2 (t) − P5 (t) (16)

451 Then denote the control input as u(t) , [P1 (t), P2 (t), P3 (t), P5 (t), P6 (t)]T ,
452 the system state as x(t) , S(t) and the output as y(t) , P4 (t). From Eq.
453 (5), the linear state-space model with disturbances can be expressed as

x(t + 1) = Ax(t) + Bu(t) + Bw(t)
, (17)
y(t) = Cx(t) + Du(t) + Dw(t) + PL (t) + wL (t)

454 where A = I, C = 0, B = [ηC , −ηD , ηC , 0, −ηD ], D = [0, −1, 0, 0, −1, 0]; w(t)
455 and wL (t) are disturbances of input and output respectively. Although w(t)
456 is caused by differences of predicted and actual values in terms of the cus-
457 tomer load and the PV output, it is uneasy to determine w(t) according
458 to these differences. If the customer demand experiences disturbance wL (t)
459 and the PV output experiences disturbance wpv (t) respectively, the adjusted
460 disturbance w(t) is determined using the following proposed rules, in which
461 Pi (t), i = 1, 2, ..., 6 are pre-scheduled power flows without consideration of
462 disturbance.
463 (a) If wpv (t) is negative, reduce P1 (t) firstly to satisfy the PV’s output
464 constraint. If P1 (t) = 0, then reduce P5 (t) till the constraint is satisfied.

18
465 (b) For a positive wpv (t), increase P1 (t) as P1 (t) = P1 (t)+wpv (t) if wL (t) ≤
466 0; else increase P5 (t) as P5 (t) = P5 (t) + wpv (t).
467 (c) If P2 (t) + P5 (t) > PL (t) + wL (t), reduce P2 (t) firstly to satisfy the
468 power balance constraint. If P2 (t) = 0, then reduce P5 (t) till the power
469 balance constraint is satisfied.
470 (d) If P2 (t)+P5 (t) < PL (t)+wL (t), increase P2 (t) by α [wL (t) − ∆P2 (t) − ∆P5 (t)],
471 where ∆P2 (t) and ∆P5 (t) denote increments of P2 (t) and P5 (t) during steps
472 (a-c).
473 (e) Bound Pi (t)(i = 1, 2, 5) within [0, Pimax ]. If S(t − 1) + ηC P1 (t) −
474 P2 (t)/ηD − P6 (t)/ηD ≤ S max is not satisfied, reduce P1 (t) firstly to satisfy
475 this condition, then if P1 (t) = 0 continue to increase P2 (t) till the condition is
476 satisfied. If S(t − 1) + ηC P1 (t) + ηC P3max − P2 (t)/ηD − P6 (t)/ηD ≥ S min is not
477 satisfied, reduce P2 (t) to satisfy this condition. If the SOC cannot satisfy
478 the boundary constraint, reduce P3 (t) if S(t) > S max (t); increase P3 (t) if
479 S(t) < S min (t).
480 (f) The actual output is calculated according to Eq. (16).
481 Note that in (a) and (b), P5 (t) has higher priority of usage than P1 (t)
482 because the charging or discharging processes cause loss of energy. In (c), the
483 battery is seldom employed if P5 (t) is sufficient to provide power. In (d), when
484 the power balance at the customer side is broken, the battery takes some
485 responsibility for covering the disturbed load demand. The responsibility
486 rate α is an ad-hoc parameter. In this paper, α = 35% is used during the
487 standard period, α = 50% is used during the peak period, α = 20% is used
488 during the off-peak period. In (e), P1 (t), P2 (t), P3 (t) are re-adjusted when
489 the SOC boundary constraint is violated. At the last step, the output is
490 determined by the actual load demand and the system input. The rational
491 behind the rule is that PV power for customer usage has higher efficiency that
492 PV power for storage, and that load demand balance is mainly satisfied by
493 the grid power supply and then by the battery discharge. When the demand
494 increases, the battery will discharge more over the off-peak period than the
495 peak period for the price concern
496 MPC is developed for the closed-loop control, in which the objective func-
497 tion of DSM model is optimized over the receded prediction horizon. Com-
498 bining Eq. (8) and (16), the objective function over the prediction horizon

19
499 TN = [k, k + Np ) (k ≥ 0 is an integer) is obtained as follows:
k+Np −1
X
Jc = ρ(t) [PL (t) + P3 (t) − P2 (t) − P5 (t)]
t=k
X (18)
− rk ρk P6 (t) + Ch .
t∈Tk ∩TN

500 where Np is hours over the prediction horizon, and the wearing cost of battery
Pk+N −1
501 over the prediction horizon is CB = t=k p a[P2 (t) + P6 (t)] + Np b. As in
502 Eq. (10-15), the constraints of MPC can be expressed as


 P1 (t) + P5 (t) ≤ Ppv (t)
 S min ≤ S(t) ≤ S max


0 ≤ Pi (t) ≤ Pimax , i = 1, 2, 3, 5, 6 , (19)
max
0 ≤ PL (t) − P2 (t) − P5 (t) ≤ P4




S(0) ≤ S(N)


503 where N is hours over the overall scheduling period. Note that S(0) ≤ S(N)
504 is only valid in the MPC approach when k ≥ N − Np + 1.
505 MPC is employed to solve this optimal control problem at each sampling
506 period. In the proposed MPC approach, an optimal control problem over
507 the prediction horizon is repeatedly solved (k = 0, . . . , N − Np ). The optimal
508 control problem, including the objective function and the set of constraints,
509 has been defined in Eq. (18) and (19). The optimization variable is the
510 power flow sequence at each sampling period. At the kth sample, an optimal
511 solution [u(k), u(k + 1), . . . , u(k + Np − 1)]T can be obtained after solving
512 the optimal problem. Only the first part of solution, i.e., u(k), is used in
513 the current period. According to our proposed rules, the disturbance of the
514 input w(k) applied to the system in the period [k, k + 1) can be determined.
515 When the planning horizon gets shorter than the prediction horizon Np , i.e.
516 k > N −Np +1, the prediction period will be decreased by 1 after each sample.
517 The procedure of the MPC approach can be illustrated as in Algorithm 1.
518 The schematic of MPC is illustrated in Figure 5. At the kth sampling instant,
519 the feedback state is firstly measured, and the control input u(k) is computed
520 as Algorithm 1. Then control input is implemented on the open loop plant
521 for system operation. Note that the main difference between the open loop
522 control and MPC is that the open loop control does not have the feedback
523 and the control input is pre-designed off-line, but MPC has the feedback and
524 real time control mechanisms.

20
 1 Set k = 0;
2 while k ≤ N − Np do
3 Measure the state value x(k);
4 Solve the optimal control problem Eq. (18) subject to Eq. (19);
5 For the solution [u(k), u(k + 1), . . . , u(k + Np − 1)]T , apply u(k) to
the system at the period [k, k + 1);
6 According to disturbances wL (k) and wpv (k), determine the input
disturbance w(k) based on the proposed rules;
7 k = k + 1;
8 end
9 Np = Np − 1;
10 while k < N do
11 Measure the state value x(k);
12 Solve the optimal control problem Eq. (18) subject to Eq. (19);
13 For the solution [u(k), u(k + 1), . . . , u(N − 1)]T , apply u(k) to the
system at the period [k, k + 1);
14 According to disturbances wL (k) and wpv (k), determine the input
disturbance w(k) based on the proposed rules;
15 k = k + 1; Np = Np − 1;
16 end
Algorithm 1: Pseudo-code of the proposed MPC approach

21
 Open loop plant
D

Z t 

u t  x  t  1 xt  y t 
MPC B z-1 C

xt  PL  t   ZL (t )
A

Figure 5: Schematic of the MPC approach

525 To verify the performance of MPC, three experiments with various dis-
526 turbances are evaluated. The sampling period is one hour. The scheduling
527 period is five workdays in winter, and the prediction period is 24 hours. This
528 closed-loop control is compared with the open loop control (as illustrated in
529 the previous section) regarding cost saving. For simplification, a distribu-
530 tion fitting method [23] is used as our uncertainty model of disturbance in
531 this paper. The distribution fitting method includes a hypothesis regarding
532 a standard probability distribution of the forecast error (solar generation or
533 load) and a fitting procedure used to find its parameters. Load and solar
534 forecast errors are assumed to follow truncated normal distribution (TND).
535 Probability density function (PDF) of TND is
1
σ
P DFN ( x−µ
σ
)
P DFTND (x) = , (20)
CDFN ( σ ) − CDFN ( a−µ
b−µ
σ
)

536 where µ is the mean value of non-truncated normal distribution; σ is stan-

537 dard deviation of non-truncated normal distribution; a and b are upper and
538 lower limits of TND, i.e., x ∈ (a, b), a < b; P DFN is the PDF of standard nor-
539 mal distribution; and CDFN is cumulative distribution function of standard
540 normal distribution.
541 (1) Positive disturbance on PV output: In this experiment, the solar
542 irradiation is assumed to be larger than the predicted irradiation. Then
543 the PV output experiences positive disturbance. For an hour, the positive
544 disturbance is assumed to follow TND with parameters µ = 0, σ = 0.4, a =
545 0, b = 1.5, which are obtained by analyzing historical data.

22
546 In the MPC approach, it is desired that the additional PV output distur-
547 bance can be used to charge the battery, so that the customer will use less
548 power from the grid. The resulted profiles of P2 and P4 are plotted in Figure
549 6(a) and (b). It can be noticed that more solar power is used to charge the
550 battery in the closed-loop than in the open loop control. It is also observed
551 that the customer needs less power from the grid in the closed-loop. The net
552 income for five days is $1.02 for open-loop control and $1.14 for MPC. The
553 earning has increased about 27 % after using MPC.

5 5
MPC MPC
4.5 open−loop 4.5 open−loop

4 4

3.5 3.5
Power(kWh)

Power(kWh)
3 3

2.5 2.5

2 2

1.5 1.5

1 1

0.5 0.5

0 0
0 20 40 60 80 100 120 0 20 40 60 80 100 120
Time(hour) Time(hour)
(a) (b)

5 5
MPC MPC
4.5 open−loop 4.5 open−loop

4 4

3.5 3.5
Power(kWh)

Power(kWh)

3 3

2.5 2.5

2 2

1.5 1.5

1 1

0.5 0.5

0 0
0 20 40 60 80 100 120 0 20 40 60 80 100 120
Time(hour) Time(hour)
(a) (b)

Figure 6: Power flows in Experiment 1 and 2: (a) P1 of Experiment 1; (b) P4 of Experiment

1; (c) P2 of Experiment 2; (d) P3 of Experiment 2

554 (2) Positive disturbance on customer demand:

555 In the second experiment, customers are assumed to demand more power

23
556 than the predicted value. Then the customer demand experiences positive
557 disturbance. For an hour, the positive disturbance is assumed follow TND
558 with parameters µ = 0, σ = 0.6, a = 0, b = 1.5, which are obtained by
559 analyzing historical data. For the open-loop control, the amount of additional
560 demand asks for the same amount of additional power from the grid, i.e.,
561 increasing P4 .
562 In the closed-loop control, the battery takes predefined responsibility to
563 provide power to satisfy the additional disturbance on demand. To achieve
564 this task, the battery must be charged sufficiently by the grid power over
565 the off-peak period to reduce the electricity cost. The power loss during the
566 charging process must be considered in the cost at the same time. Figure
567 6(c) and (d) show the profiles of P2 and P3 for both MPC and open-loop
568 control. An interesting observation is that the battery discharges more in
569 the peak and standard periods (as shown in (c)) and more grid power is used
570 to charge the battery over the off-peak time (22, 24] (as shown in (d)). As
571 a result, the net income for five days is $7.89 for MPC compared with $6.01
572 for open-loop control. It can be noticed that the earning increases around
573 31% after using MPC.
574 (3) Random disturbances on PV output and demand: In the third exper-
575 iment, the disturbances are random numbers, which can be positive or neg-
576 ative. The random disturbances follow TND with parameters µ1 = 0, σ1 =
577 0.4, a1 = −1.5, b1 = 1.5 for solar forecast error and µ2 = 0, σ2 = 0.6, a2 =
578 −1.5, b2 = 1.5 for load forecast error.
579 It is obvious that open-loop control cannot handle such a complicated
580 case because the PV output and the load balance constraints are not satis-
581 fied. However, by using the MPC approach, all constraints can be satisfied
582 and the electricity cost is also optimized. To compare effects of disturbances,
583 the MPC approach is also implemented on the hybrid system without con-
584 sideration of disturbance, in which the obtained results are called nominal
585 values. The profiles of SOC obtained by MPC during five days are given in
586 Figure 7, in which the profile under disturbances is compared with the nom-
587 inal profile under no disturbance. By using the MPC approach, the battery
588 SOC is maintained within a safe range, and the SOC under random distur-
589 bances converges to the nominal value. The robustness of MPC is achieved
590 owing to its closed-loop mechanism. The net income in this case is still
591 promising as $4.83.

24
 100
actual
nominal

95

90

85

80
SOC/Smax (%)

75

70

65

60

55

50
0 20 40 60 80 100 120
time (hour)

Figure 7: Profile of SOC for Experiment 3

592 7. Conclusion
593 Demand side management has been considered in the optimal scheduling
594 of small-scale PV-battery hybrid system on the behalf of customers. An
595 example of DR program, i.e., TOU with power selling over peak period, has
596 been studied for energy management in this paper. A model for reducing
597 electricity cost has been developed, which is much practical at the customer
598 side. The results have shown that the optimal solution to the operation of
599 hybrid system achieves the maximal use of solar energy and battery storage.
600 It can be observed that the battery plays a significant role in storing grid
601 power during off-peak periods and supplying power to customers during peak
602 periods. As a result, by scheduling the hybrid system, customers consume
603 minimal amount of power from the grid and reduce their monthly cost. It
604 has been shown that optimal control is a useful open-loop control method
605 for power flow control in DSM.
606 As that open loop control cannot handle the control task when the hybrid
607 system experiences disturbances in PV output and load demand, MPC has
608 been developed for controlling such a hybrid system when disturbances occur.

25
609 For this closed-loop control method, the linear state-space model has been
610 formulated. The proposed MPC approach has been employed to schedule the
611 disturbed hybrid system. Closed-loop control has been compared with open
612 loop control in the presented simulation study. Any disturbance experienced
613 in the system can be detected before the next control period, and the control
614 variable can be corrected accordingly for the next period by employing the
615 MPC approach. For a highly disturbed system that cannot be handled by
616 open loop control, MPC can achieve great control performance in terms of
617 accuracy and robustness. Furthermore, more cost savings can be obtained
618 by using the closed-loop control.
619 In this work, only TOU is evaluated in the small-scale hybrid system as
620 an example of DSM. Future work include considering other DSM programs
621 and extending the model to incorporate more renewable energy sources such
622 as wind power, biomass power, hydro power and so on.

623 References
624 [1] Shaahid S, El-Amin I. Techno-economic evaluation of off-grid hy-
625 brid photovoltaic-diesel-battery power systems for rural electrification
626 in saudi arabia — a way forward for sustainable development. Renew-
627 able and Sustainable Energy Reviews 2009;13(3):625 –33.

628 [2] Hove T, Tazvinga H. A techno-economic model for optimising compo-

629 nent sizing and energy dispatch strategy for PV-diesel-battery hybrid
630 power systems. Journal of Energy in Southern Africa 2012;23(4):18 –
631 28.

632 [3] Arun P, Banerjee R, Bandyopadhyay S. Optimum sizing of photo-

633 voltaic battery systems incorporating uncertainty through design space
634 approach. Solar Energy 2009;83(7):1013 –25.

635 [4] Giraud F, Salameh Z. Steady-state performance of a grid-connected

636 rooftop hybrid wind-photovoltaic power system with battery storage.
637 IEEE Trans on Energy Conversion 2001;16(1):1–7.

638 [5] Gabash A, Li P. Flexible optimal operation of battery storage sys-

639 tems for energy supply networks. IEEE Trans on Power Systems
640 2013;28(3):2788 –97.

26
641 [6] Kanchev H, Lu D, Colas F, Lazarov V, Francois B. Energy management
642 and operational planning of a microgrid with a PV-based active gener-
643 ator for smart grid applications. IEEE Trans on Industrial Electronics
644 2011;58(10):4583 –92.

645 [7] Palma-Behnke R, Benavides C, Lana F, Severino B, Reyes L, Llanos

646 J, et al. A microgrid energy management system based on the rolling
647 horizon strategy. IEEE Trans on Smart Grid 2013;4(2):996 – 1006.

648 [8] Aalami H, Moghaddam MP, Yousefi G. Demand response modeling con-
649 sidering interruptible/curtailable loads and capacity market programs.
650 Applied Energy 2010;87(1):243 –50.

651 [9] Nair NKC, Garimella N. Battery energy storage systems: Assessment
652 for small-scale renewable energy integration. Energy and Buildings
653 2010;42(11):2124 –30.

654 [10] Levron Y, Guerrero J, Beck Y. Optimal power flow in microgrids with
655 energy storage. IEEE Trans on Power Systems 2013;28(3):3226–34.

656 [11] Wang C, Nehrir M. Power management of a stand-alone

657 wind/photovoltaic/fuel cell energy system. IEEE Trans on Energy Con-
658 version 2008;23(3):957–67.

659 [12] Jain S, Agarwal V. An integrated hybrid power supply for distributed
660 generation applications fed by nonconventional energy sources. IEEE
661 Trans on Energy Conversion 2008;23(2):622–31.

662 [13] Teleke S, Baran M, Bhattacharya S, Huang A. Rule-based control of

663 battery energy storage for dispatching intermittent renewable sources.
664 IEEE Trans on Sustainable Energy 2010;1(3):117–24.

665 [14] Kanchev H, Lu D, Colas F, Lazarov V, Francois B. Energy management

666 and operational planning of a microgrid with a PV-based active gener-
667 ator for smart grid applications. IEEE Trans on Industrial Electronics
668 2011;58(10):4583–92.

669 [15] Riffonneau Y, Bacha S, Barruel F, Ploix S. Optimal power flow man-
670 agement for grid connected PV systems with batteries. IEEE Trans on
671 Sustainable Energy 2011;2(3):309–20.

27
672 [16] Tazvinga H, Xia X, Zhang J. Minimum cost solution to photovoltaic-
673 diesel-battery hybrid power systems for remote consumers. Solar Energy
674 2013;96:292 –9.

675 [17] Lombardi P, Powalko M, Rudion K. Optimal operation of a virtual

676 power plant. In: Power Energy Society General Meeting, 2009. PES
677 ’09. IEEE. 2009, p. 1–6.

678 [18] Derakhshandeh S, Masoum A, Deilami S, Masoum M, Hamedani Gol-

679 shan M. Coordination of generation scheduling with PEVs charging in
680 industrial microgrids. IEEE Trans on Power Systems 2013;28(3):3451–
681 61.

682 [19] Gabash A, Li P. Flexible optimal operation of battery storage sys-

683 tems for energy supply networks. IEEE Trans on Power Systems
684 2013;28(3):2788–97.

685 [20] Moura PS, de Almeida AT. Multi-objective optimization of a mixed

686 renewable system with demand-side management. Renewable and Sus-
687 tainable Energy Reviews 2010;14(5):1461 –8.

688 [21] Huang L, Walrand J, Ramchandran K. Optimal demand response with

689 energy storage management. In: 2012 IEEE Third International Con-
690 ference on Smart Grid Communications (SmartGridComm). 2012, p.
691 61–6.

692 [22] Adika CO, Wang L. Autonomous appliance scheduling for household
693 energy management. IEEE Trans on Smart Grid 2014;5(2):673 –82.

694 [23] Makarov Y, Etingov P, Ma J, Huang Z, Subbarao K. Incorporating

695 uncertainty of wind power generation forecast into power system oper-
696 ation, dispatch, and unit commitment procedures. IEEE Transactions
697 on Sustainable Energy 2011;2(4):433–42.

698 [24] Zervas P, Sarimveis H, Palyvos J, Markatos N. Model-based optimal

699 control of a hybrid power generation system consisting of photovoltaic
700 arrays and fuel cells. Journal of Power Sources 2008;181(2):327 –38.

701 [25] Garcı́a CE, Prett DM, Morari M. Model predictive control: Theory and
702 practicea survey. Automatica 1989;25(3):335 –48.

28
703 [26] Xia X, Zhang J, Elaiw A. An application of model predictive control to
704 the dynamic economic dispatch of power generation. Control Engineer-
705 ing Practice 2011;19(6):638 –48.

706 [27] Zhang J, Xia X. A model predictive control approach to the periodic im-
707 plementation of the solutions of the optimal dynamic resource allocation
708 problem. Automatica 2011;47(2):358 –62.

709 [28] Wu Z, Xia X. Model predictive control for improving operational effi-
710 ciency of overhead cranes. Nonlinear Dynamics 2014;:1 – 19.

711 [29] Upadhyay S, Sharma M. A review on configurations, control and sizing

712 methodologies of hybrid energy systems. Renewable and Sustainable
713 Energy Reviews 2014;38:47 – 63.

714 [30] Zhu B, Tazvinga H, Xia X. Switched model predictive con-

715 trol for energy dispatching of a photovoltaic-diesel-battery hy-
716 brid power system. IEEE Trans on Control Systems Technology
717 2014;doi:10.1109/TCST.2014.2361800.

718 [31] Lagorse J, Simoes M, Miraoui A. A multiagent fuzzy-logic-based en-

719 ergy management of hybrid systems. IEEE Transactions on Industry
720 Applications 2009;45(6):2123–9.

721 [32] Torreglosa J, Garca P, Fernndez L, Jurado F. Hierarchical energy man-

722 agement system for stand-alone hybrid system based on generation costs
723 and cascade control. Energy Conversion and Management 2014;77:514
724 –26.

725 [33] Wang C, Nehrir M. Power management of a stand-alone

726 wind/photovoltaic/fuel cell energy system. IEEE Transactions on En-
727 ergy Conversion 2008;23(3):957–67.

728 [34] Morais H, Kádár P, Faria P, Vale ZA, Khodr H. Optimal scheduling
729 of a renewable micro-grid in an isolated load area using mixed-integer
730 linear programming. Renewable Energy 2010;35(1):151 –6.

731 [35] Faxas-Guzmán J, Garcı́a-Valverde R, Serrano-Luján L, Urbina A. Prior-

732 ity load control algorithm for optimal energy management in stand-alone
733 photovoltaic systems. Renewable Energy 2014;68:156 –62.

29
734 [36] Chen C, Duan S, Cai T, Liu B, Hu G. Smart energy management
735 system for optimal microgrid economic operation. IET Renewable Power
736 Generation 2011;5(3):258–67.

737 [37] Kim SK, Jeon JH, Cho CH, Ahn JB, Kwon SH. Dynamic mod-
738 eling and control of a grid-connected hybrid generation system with
739 versatile power transfer. IEEE Transactions on Industrial Electronics
740 2008;55(4):1677–88.

741 [38] Wang X, Palazoglu A, El-Farra NH. Operational optimization and de-
742 mand response of hybrid renewable energy systems. Applied Energy
743 2015;143:324 –35.

744 [39] Cecati C, Citro C, Siano P. Combined operations of renewable energy

745 systems and responsive demand in a smart grid. IEEE Transactions on
746 Sustainable Energy 2011;2(4):468–76.

747 [40] Atzeni I, Ordonez L, Scutari G, Palomar D, Fonollosa J. Demand-side

748 management via distributed energy generation and storage optimization.
749 IEEE Transactions on Smart Grid 2013;4(2):866–76.

750 [41] Riffonneau Y, Bacha S, Barruel F, Ploix S. Optimal power flow manage-
751 ment for grid connected PV systems with batteries. IEEE Transactions
752 on Sustainable Energy 2011;2(3):309–20.

753 [42] Logenthiran T, Srinivasan D, Shun TZ. Demand side management in

754 smart grid using heuristic optimization. IEEE Transactions on Smart
755 Grid 2012;3(3):1244–52.

756 [43] Matallanas E, Castillo-Cagigal M, Gutiérrez A, Monasterio-Huelin F, no

757 Martı́n EC, Masa D, et al. Neural network controller for active demand-
758 side management with {PV} energy in the residential sector. Applied
759 Energy 2012;91(1):90 –7.

760 [44] Pina A, Silva C, ao PF. The impact of demand side management strate-
761 gies in the penetration of renewable electricity. Energy 2012;41(1):128
762 –37.

763 [45] Figueiredo J, Martins J. Energy production system management – re-

764 newable energy power supply integration with building automation sys-
765 tem. Energy Conversion and Management 2010;51(6):1120 –6.

30
766 [46] Lujano-Rojas JM, Monteiro C, Dufo-López R, Bernal-Agustı́n JL. Opti-
767 mum residential load management strategy for real time pricing (RTP)
768 demand response programs. Energy Policy 2012;45(0):671 –9.

769 [47] Giorgio AD, Pimpinella L. An event driven smart home controller en-
770 abling consumer economic saving and automated demand side manage-
771 ment. Applied Energy 2012;96:92 – 103.

772 [48] Zhao J, Kucuksari S, Mazhari E, Son YJ. Integrated analysis of high-
773 penetration PV and PHEV with energy storage and demand response.
774 Applied Energy 2013;112:35 – 51.

775 [49] Collares-Pereira M, Rabl A. The average distribution of solar radiation

776 correlation between diffuse and hemispherical and between daily and
777 hourly insolation values. Solar Energy 1979;22:155 –64.

778 [50] Yang H, Wei Z, Chengzhi L. Optimal design and techno-economic anal-
779 ysis of a hybrid solar–wind power generation system. Applied Energy
780 2009;86(2):163 –9.

781 [51] Roy A, Kedare SB, Bandyopadhyay S. Optimum sizing of wind-

782 battery systems incorporating resource uncertainty. Applied Energy
783 2010;87(8):2712 –27.

31