IET Renewable Power Generation

Research Article

Optimal operation of hybrid power systems ISSN 1752-1416

Received on 1st June 2016
Revised 24th November 2016
including renewable sources in the sugar Accepted on 17th January 2017
doi: 10.1049/iet-rpg.2016.0443
cane industry www.ietdl.org

Marcelo Menezes Morato1 , Paulo Renato da Costa Mendes1, Julio Elias Normey-Rico1, Carlos
Bordons2
1Departmento de Automação e Sistemas (DAS), Universidade Federal de Santa Catarina, Brazil
2Departamento de Ingeniería de Sistemas y Automática, Universidad de Sevilla, Spain
E-mail: marcelomnzm@gmail.com

Abstract: This study presents a control structure, based on model predictive control, applied to energy management
optimisation in a sugar cane processing plant including renewable sources. The proposed energy plant is set upon a sugar cane
processing industry and has to produce and maintain an amount of electric power throughout the year, defined by contract. This
plant is, also, bound to produce flows of steam in different pressures, to comply to the demands of the production process of
ethanol and sugar, from the sugar cane. The renewable sources in the system include photovoltaic, wind power generation and
the use of biomass, from the remains of the sugar cane. The proposed control algorithm has the task of performing the
management of which energy system to use (combined heat and power generation systems, boilers or others), maximise the
use of renewable energy sources, maximise the gains of the boilers (that vary according to the biomass mixture used), manage
the use of energy storages and supply the defined amount of energy. Simulation results show the satisfactory operation of the
proposed control structure.

1 Contextualisation Therefore, given the importance of sugar-ethanol power plants

The use and generation of energy in efficient ways are key
elements for achieving more ambitious goals for sustainable and
eco-friendly development. One of the biggest problems that energy
systems face is that of having to provide power to large and highly
unstable demands. The current foundations on energy generation
are about to change in a profound way [1]: affordable fossil fuel
reserves are decreasing each year (due to future scarcity and
shortage) whereas, at the same time, energy demands grow in every
country. Furthermore, the present global development structure
aims to avoid greenhouse emissions and environmental impact,
hence the highlights are given to renewable energy sources, eco-
friendly and sustainable [2].
With the crescent of the smaller and more distributed energy
plant structure for generation of power and heat, renewable power
generation will play a role of great importance in the energy
scenario in the forthcoming years [3]. The biggest issue related
with this kind of system is the unreliability and inconstant quality
of the renewable energy sources, these being intermittent and
inducing unpredictable fluctuations in the energy output. A
solution to this matter is to include intermediate energy storage
units, as batteries, water bombs, super-capacitors, compressed air
structures, fly wheels and others. The storage of energy can help
renewable sources to became viable sources, solving the issue of
unpredictability and variability of the power output.
Notably, in the instance of this work, the Brazilian energy
scenario will be taken in to account. Brazil has an immensely
diversified energy matrix [4]. Solar energy can be considered as
one of the possible energy sources to enlarge the renewable energy
share, as the country has great potential for solar energy generation
in several regions. The investment in this sector has vastly risen in
the latter years, showing competitive costs [5].
The sugar cane processing plants are particularly significant to
this study [6], as the sugar cane distillation process has a great
amount of residue that can be treated as biomass. In sum, there are
three important renewable sources from the canebrakes: the
bagasse [7], the straw and the vinasse [8] – distillation residue from
the sugar cane juice (garapa), that can be transformed into biogas,
with the technology of anaerobic digestion [9].
in the Brazilian energy setting and knowing that these are mostly
established in high insolation sites, they become potential
candidates to be managed as distributed power plants of hybrid
sources, as seen in [10], considering biomass, biogas, solar and
wind power energy and, also, natural gas. Thus, the control of
hybrid generation and storage, including renewable and non-
renewable sources, is a significant issue to be studied in order to
allow the optimum management and operation, carrying out a
coordination between legal standards, minimal environmental
standards and state-of-the-art techniques [11, 12].
As a result of the great intricacy of such energy systems
including intermediate storage and numerous generators, new
approaches [13] have been and have to be developed in order to
debate these new systems. The terms multiple energy carrier
systems (shortly, multi-carrier system) and hybrid energy systems
have become accepted when referring to these systems [14, 15].
Distributed generation herein refers to a set of smaller power
plants, located close to the end user, which may be integrated into a
network or operate autonomously, that progressively (in the
literature) have been called microgrids.
With greater demands for improvements in production and
efficiency, the operational requisites of energy systems are
evermore restrict and, for that, the operation and automatic control
have to aim optimal efficiency of these energy plants. Knowing
that individual optimal conditions of subsystems do not imply the
optimal condition of a whole system, a coordination between every
subsystem of a whole energy system becomes necessary. This
coordination is typically done by an automatic supervisor, in a
higher level in the hierarchy of the control structure of these energy
plants, and, by this, the operational set-points of each subsystem
are brought so that the operation of the whole energy system is near
its optimal operational point. In the sense of efficiency, the best
control technique should take into consideration the efficiency of
every subsystem while the quality requirements can be maintained.
The most natural approach to optimal control of processes subject
to restrictions is model predictive control (MPC). The development
of new results in the modelling, identification and predictive
control of hybrid power generation including renewable sources

IET Renew. Power Gener. 1

© The Institution of Engineering and Technology 2017
Fig. 1 Generalised plant

will imply significant improvements in the operation, efficiency This work proposes the optimisation and control of a hybrid
and security of this relevant class of systems. energy system, through a MPC structure, considering process
Recent works have brought to light MPC-based control demands and various constraints. Section 2 describes the proposed
structures used to master microgrids with renewable sources. A hybrid generation system and details the system modelling, Section
microgrid is a set of loads, generators and storage systems that can 3 describes the optimisation problems that have to be solved and
be managed in a coordinated way in order to operate isolated or details the hierarchical MPC control structure. Finally, Section 4
connected of an external grid [16]. Valverde et al. [17] shows a shows simulations of the proposed system. The paper ends with
MPC-controlled hydrogen-based domestic microgrid; the authors some conclusions.
in [18, 19] also refer to optimal generation for renewable
microgrid. Galus and Andersson [20] considers plug-in hybrid 2 Studied system
electric vehicles based on multi-carrier system and Da Costa
Mendes P. et al. [21] proposes a MPC structure for energy 2.1 Hybrid generation and storage system
management of a microgrid. In terms of renewable sources This work proposes a generic hybrid energy generation system,
applications, Galus et al. [22] is specific about wind balancing based on sugar cane power plant, that produces sugar, ethanol and
capabilities using heuristics and MPC, whereas Morais et al. [23] electric power. The proposed system is composed by the following
focuses on a renewable microgrid in an isolated load area. Generic subsystems: two boilers, with different efficiencies; two steam
MPC-based structures are seen in [24, 25]. Economic optimisation turbines, with different efficiencies; a combined heat and power
is exposed in [26, 27]. Cogeneration is discussed in [28, 29]. system, herein denoted as CHP; a water chiller; a hot water tank;
photovoltaic panels; water heating solar panels; a wind turbine; two
pressure reduction valves; one heat exchanger; stocks of bagasse,
Table 1 Nomenclature of manipulated variables straw and compressed biogas and a battery bank. The tendered
Nomenclature Manipulated variable Unit system, being a plausible adaptation to a real plant, is interesting in
SPCB lower efficiency boiler's set-point (Mg/h) an economic and sustainable point-of-view, as it proposes the use
of renewable sources and the recycling of the sugar cane residues.
B lower efficiency turbine's set-point (kW)
SPTU This influences directly on the increase of the total energy
A
SPTU better efficiency turbine's set-point (kW) generation capacity and, also, on the efficiency and the
sustainability of the system - that, herein, does not use any fossil
PotBat input (output) energy flow to the battery bank (kW) fuel or energy source external to the sugar-ethanol production
SPCHP CHP's set-point (kW) process. The system aims to use the best possible technology,
SPch water Chiller's set-point (m3 /h)
finding an equilibrium point between legal norms, minimum
environmental standards and the state-of-the-art techniques. Fig. 1
SPTC heat exchanger's set-point (m3 /h) shows the outline of the proposed system and Table 1 details the
QOut
V
high-middle pressure reduction valve's set- (Mg/h) used nomenclature; QEA and QEB represent the biomass mixture
point flows, measured in Mg/h .
QMB
V
middle-low pressure reduction valve's set- (Mg/h) The two boilers are here nominated ‘better efficiency’ and
point ‘lower efficiency’ and produce, respectively, higher (67 bar) and
QTank
Esc
hot water escape flow (m3 /h) middle pressure (21 bar) steam. They are both powered by the
M middle pressure steam escape flow (Mg/h) biomass mixture of straw and bagasse, as seen in Fig. 2. The
QEsc
turbines, likewise the boilers, are here nominated ‘better efficiency’
B
QEsc low pressure steam escape flow (Mg/h) and ‘lower efficiency’. The better efficiency turbine has a low
PotNet instant electric power available to the (kW) pressure steam residue (1.5 bar), while the lower efficiency
network turbine's steam residue meets the middle pressure steam demands.
Both turbines produce electric power that, with the CHP and the

2 IET Renew. Power Gener.

© The Institution of Engineering and Technology 2017
 From a discrete time standpoint, internal storage state xs, at
sampling time k + 1, ct previous sample k and on the total
exchanged flow ⌣u Es (k) during the period ΔT , ranging from k to
k + 1, assuming ⌣
u Es (k) to remain constant during ΔT :

xs(k + 1) = Axs(k) + ⌣
u Es (k)ΔT (1)

As described in [33], the state space representation that models an

energy hub, considering z(k) is an array of external disturbances,
are, in this work:

Fig. 2 Biomass mixture x(k + 1) = Ax(k) + Bu(k) + Cz(k) (2)

renewable energy generation, are related to the electric demands. y(k) = Dx(k) + Eu(k) + Fz(k) (3)
The CHP is a gas engine that uses de energy present in the gas to
generate electric power. The heat released by the CHP is recovered 2.2.1 Manipulated variables: The system's manipulated variables
to warm water, making the CHP a highly efficient system. The are put in Table 1. The set-points will be treated by lower level
CHP, in this scenario, is boosted by the compressed biogas, made internal controls. The complete manipulated variables vector is:
from the vinasse. The use of intermediate storage units (battery
A B
bank and biomass stocks) allows the system to accumulate energy u = [SPTU SPTU PotNet SPCB QOut
V
M
QEsc …
(or biomass, that can be converted into energy) when the renewable B
(4)
generation is high and use-it when there is no renewable …QEsc SPCHP SPCh SPTC PotBat QVMB QTank
Esc ]
T

production.
The refrigeration demands are met with the water flow from the It is important to mention that QTank M B
Esc , QEsc and QEsc represent the
water chiller, that cools down the hot water from the water tank. escape flows of safety valves, to be used only in particular
This tank is an intermediate storage unit that stocks the hot water situations.
flow coming from the CHP (at constant temperature), from the heat
exchanger (converted from low pressure steam) and from the solar
2.2.2. Disturbance vector: The external disturbances of the
panels.
system are herein put as:
Several issues related to the studied system have to be taken
into account. First, the efficiencies of the boilers fluctuate T
according to the mixture of biomass that boosts them. The straw is z = Wndin Irrdin Bagin Strin Bgin (5)
a moist left-over from the sugar cane, so too much straw causes
problems with the ignition of the boilers. It is known that the best being Wndin the speed of the wind (measured in km/h) present in
possible mixture of bagasse and straw to boost boilers is 50% of the microgrid's area, used by the wind turbines to generate electric
each [30–32]. The static gains of the boilers vary according to the power, and Irrdin the amount of solar irradiation (measured in
mixture of biomass used; mixtures with greater amount of straw W/m2) on the microgrid's solar panels (photovoltaic and water
have bigger calorific power, but are wetter. This in an issue of heating). Bagin, Strin and Bgin represent the income (t/h) of
optimal control, once the amount of bagasse, available in a
canebrake, is always bigger than the amount of straw – in the bagasse, straw and compressed biogas to the respective stocks.
studied plant, there is an average income of 93 Mg/h of bagasse
and 8.7 Mg/h of straw. For this, this work takes into account 2.2.3 System output: The system output vector is defined as:
biomass blending factors, αa and αb, for each boiler, as seen in
T
Fig. 2. y = PProc QM
V QLV QCW PSale (6)
The proposed plant has four demands to satisfy: electric power
demand, due to ethanol and sugar production process; middle and being PProc the electric power produced due to the sugar cane
low pressure steam demands, defined by the process; refrigeration
processing demand (kW); QM
V the flow of middle pressure steam
(chilled water) demands, used to cool down generators, oil tanks
and water for fermentation units. It is important to mention that (Mg/h); QLV the flow of low pressure steam (Mg/h); QCW the flow
satisfying each demand alone is not adequate, as they are of cold water required by the distillery process (m3 /h); finally, PSale
inextricably linked. As an example: increasing the amount of represents the electric power made available for the external
power supplied by the turbines also increases the amount of steam network (kW).
produced.
The electric power that has to be produced by the microgrid
consists of two factors: the internal power demands, to maintain the 2.2.4 Stock flows: The system state vector is defined as:
sugar cane processing – in average 8000 kW, and the amount of T
energy that is sold to the local power distributer (this is defined by x = X Bat X Bag X Str X Bg XT (7)
a fixed contract, 11.52 GWh per month). This power generation
can be continuous, maintaining an average electric power, or time- where each entry represents the normalised percentage of each
varying, with different values of electric power made available at stock. By the principles of mass conservation, it can logically be
each hour of the day. deduced, for each stock, the following equations:

2.2 System modelling dX Bat

= ηbatPotBat (8)
dt
In this work, the system modelling approach used will be the
energy hub modelling methodology. An energy hub can be used to dX Bag
model the interface between energy producers, consumers and the = Bagin − Bagout = Bagin − (α AQEA + αBQEB) (9)
transmission line. From the system's outlook, an energy hub can be dt
identified as a black box unit that comprises the following
requirements: (i) input and output of electric power; (ii) energy dX Str
= Strin − Strout = Strin − [(1 − α A)QEA + (1 − αB)QEB] (10)
conversion; (iii) energy storage. dt

IET Renew. Power Gener. 3

© The Institution of Engineering and Technology 2017
Fig. 3 Proposed control structure
dX Bg
= Bgin − Bgout (11)
dt
dX T
= T in − T out = (QS + QCHP + QTC) − (QTank
Esc + QCh) (12)
dt
where the input and outflows, for each storage unit, are normalised.
ηbat represents the battery bank efficiency, that is different when
charging or discharging. Bagout and Strout, as show (9) and (10), are
a result of the biomass mixture flows, QEA and QEB. This is clear in
Fig. 2.Bgout is a result of the CHP's operating point, herein decided
by the variable SPCHP. T in is a consequence of the water heating
solar panels' operating point (QS), of the water heated by the CHP
(QCHP) and of the hot water coming from the heat exchanger (QTC),
as shows (12). This equation also shows that T out is a result of the
hot water destined to the water chiller (QCh, outcome of SPCh) and
of the hot water escape flow, herein decided by QTank
Esc .
3 Optimisation problems
It is important to address the optimal control of this studied system,
having the state space representation, as seen in (2) and (3). There
are two optimisation obstacles to be solved: the biomass mixture
optimisation and the demand optimisation.
3.1 Biomass mixture optimisation
The mixture optimisation consists in maximising the boiler gains
(conversion factor from amount of biomass mixture to amount of
produced steam (t/h)/(t/h)), that fluctuate according to the biomass
mixture composition. The more straw there is in the mixture, the
greater the gains (up to 50% of straw, due to moisture). Note that
this blending factors (statistical dummy variables, α A and αB) are,
respectively, related to the high efficiency boiler and the lower
efficiency boiler. QEA and QEB represent the biomass mixture that
runs into each boiler, in the respective order. The amount of
bagasse outflow and straw outflow from the stocks (Bagout and
Strout) are show in (9) and (10).
The optimisation here proposed is solved by the optimiser,
Optα, that has to maintain the blending factors as close to 0.5 as
possible, knowing the future straw income predictions. This is also
subject to the following constraints: (i) maintain the straw stock
inside the 10 and 100% limits; (ii) try to maintain the straw stock
near a reference of 20%, whenever possible; (iii) preserve the
blending factors as dummy variables (values from 0 to 1); (iv) limit
the outflow of straw.
4 IET Renew. Power Gener.
3.2 Demand optimisation
The other optimisation problem, solved by the MPC controller,
MPCmain, is the main purpose of this study: the microgrid has to
produce (sell) the contract-defined amount of electric energy, while
still meeting all the system demands: internal power demand,
steam demands and refrigeration demands.
As explained beforehand, the contract corresponds to the
generation of 11.52 GWh of electric energy, per month – herein
treated as a daily production of 384 MWh. The optimisation has to
define the manipulated vector (4) so that the daily production of
energy corresponds to the contract and the system state vector (7)
follows a reference state (e.g.: all stocks with 50%). This is subject
to the following restrictions: (i) The manipulated variables have to
stay within physical limits; (ii) The system output vector (6) has to
contemplate the system demands; (iii) The system state vector (7)
has to stay within bounded limits – every stock has its operational
band.
3.3 Control structure
The proposed control structure to solve the studied problem is
comprised of two levels. The first level is associated with
optimising the biomass blending factors. The second (and main)
level is associated with complying the process demands and
producing the contract-defined amount of electric energy, in the
most efficient and sustainable way possible.
Fig. 3 shows the proposed two-level hierarchical control
structure based on MPC. Both the optimiser Optα and the controller
MPCmain are operating together, on the same time scale (sampling
time ΔT = 1 h). The biomass mixture optimiser acts with
information of the biomass mixture flows (QEA(k) and QEB(k)) and
the straw stock percentual (X Str(k)) and transmits to the microgrid
and the controller the instant values of the blending factors (α A(k)
and αB(k)). The main controller acts with past blending factor
values (α A(k − 1) and αB(k − 1)), instant disturbance, system state
and system output vectors values ( z(k), x(k), y(k)) and handles the
manipulated variables (acting on the microgrid).
3.3.1 Main controller (MPCMain): The main MPC has the
following objective function:
Nc − 1 T
(Econtract − Esum)
J MPC = ∑ PotNetwork(k + i) −
ΔT
i=0
(Econtract − Esum)
QP PotNetwork(k + i) −
ΔT (13)
Nc − 1
Np − 1
+ ∑ quu^ (k + i) + ∑i = 0 x^ (k + i) − x^ ref (k + i) T
i=0
Qx x^ (k + i) − x^ ref (k + i) + qϵϵ
© The Institution of Engineering and Technology 2017
where Esum represents the electric energy that has already been
produced by the microgrid, at given iteration k; Econtract represents
the daily production of 384 MWh; the system state reference is put
as x^ ref ; ϵ represents a slack variable that is to be minimised, used in
the constraints, see (17); N p = 12h represents the prediction
horizon, while N c = 5 h represents the control horizon. As it can be
seen, (Econtract − Esum) represents how much electric energy the
microgrid still has to produce until the end of the day, due to
contract requirement. For this, when minimising
PotNetwork − (((Econtract − Esum))/ΔT) , the main controller forces
the production of energy at iteration k to approach the necessary
amount to end the contract requirement, hence, by the end of the
day, the amount of electric energy supplied to the network is the
one defined by contract. The objective function (13) is subject to
the following constraints:
x j ≤ x j(k + i + 1) ≤ x̄ j
^
(14)
u j ≤ u^ j(k + i) ≤ ūi (15)
y^ (k + i) = Demands(k) (16)
0 ≤ PotNetwork(k) ≤ χ + ϵ (17)
for i = 0, …, N p − 1, where qϵ and qu are positive defined vectors,
QP and Qx are positive definite weighting matrices and χ
represents the contract constant power (16 MW, if maintained at
every iteration k, the electric contract would certainly be
complied). The notation hat over variables (a^ ) is used to denote
variables over the prediction horizon, ai and āi denote minimum
and maximum allowed values respectively. The matrix QP is
adjusted so that the electric energy production is prioritised; Qx is
used to maintain the system state vector values near a referenced
region of 50% of all stocks. The matrix qu is used so that the
production of energy comes preferably from the most efficient and
sustainable energy sources and so that the use of the battery bank is
minimised.
Ramping rate constraints were not considered in this work, for
the energy generating subsystems are always considered as turned
on, even if at a minimum work rate, and the dynamics of the
energy generators are too fast if compared with the used sampling
time of T s = 1 h.
It is important to reason that although the daily energy
production set-point (Econtract) is set to 384 MWh (average of
monthly contract requirement), this value can be updated each day,
depending on the fulfilment of the monthly objective up to the
previous day. The use of prediction of the non-dispatchable
renewable sources to compute this monthly planning is an issue for
future works.
3.3.2. Optimiser (Optα): The biomass mixture optimiser was
implemented with the following objective function:
Np − 1
J Opt =
α
∑ α^ A(k + i) − 0.5 TQ A α^ A(k + i) − 0.5
i=0
Np − 1
+ ∑ α^ B(k + i) − 0.5 TQB α^ B(k + i) − 0.5
i=0
Np − 1
^ ref T ^ ref
∑
^ ^
+ X Str(k + i + 1) − X Str QStr X Str(k + i + 1) − X Str
l=0
(18)
subject to the following constraints:
^
X Str ≤ X Str(k + i + 1) ≤ X Str (19)
IET Renew. Power Gener.
© The Institution of Engineering and Technology 2017
α ≤ α^ A(k + i) ≤ ᾱ A (20)
αB ≤ α^ B(k + i) ≤ ᾱB (21)
0 ≤ [(1 − α A)QEA + (1 − αB)QEB)] ≤ Strout (22)
The first two terms of this objective function (18) represent the
approach of the blending factors to the optimal factors (0.5 each).
The third term of (18) is so that the predicted future straw stock
follows a reference of 20%, if possible. The weighting matrices
Q A, QB and QStr are tuned so that the optimiser's main goal is to
minimise α A and αB.
4 Results
The experimental results of the proposed control strategy, applied
to a simulated model of the studied microgrid, are presented in this
section. The control strategy was implemented using the software
Matlab [34] with Yalmip toolbox [35] and solver CPLEX [36]. The
experiments described in this section represent the simulation of
one day of electric energy production by the microgrid. The
microgrid control objectives are to maximise the use of renewable
energy sources, ensure the energy production defined by contract
and ensure the load demand at all periods of time. The use of the
renewable sources is visible. Table 3 depicts the simulation
parameters used and other important data to this study.
Fig. 4 shows a simulation scenario (full day) of the proposed
microgrid and details the use of each subsystem: the electric power
generated by the CHP, by the turbines and by the non-dispatchable
renewable sources (winds speed and solar irradiation). It can be
seen that the generation is enough to comply to the energy contract.
Fig. 5 shows the system's disturbances throughout this given
simulation scenario. It is important to remark that both Figs. 4 and
5 reflect the same day, but the presence of biomass and biogas
incomes does not directly imply into energy generation, for this
incomes pass through their respective stocks. The presence of the
CHP and the non-dispatchable renewable sources in the proposed
system implies reduction of the turbines energy production. Fig. 4
also shows how the external non-dispatchable renewable sources
predictions affect the use of the turbines, the battery bank and the
CHP. The prediction of greater peaks of wind speed and solar
irradiation impose the reduction of the turbines' usage. The
presence of valleys in the prediction curves impose in maximising
the use of the CHP and in using the battery bank to store energy –
the main goal of the battery bank is to mitigate short duration
fluctuations from the photovoltaic panels' and wind turbine's
energy output.
As can be seen in Figs. 4 and 5, the CHP's usage depends on the
biogas income behaviour. The lower efficiency turbine maintains
approximately 3 MW of electric power, throughout most part of the
day, increasing the production from 18 to 22 h, when there is an
electric demand peak and a reduction in the biogas income. The
better efficiency turbine contributes with around 15 MW, most
parts of the day, increasing the generation after 21 h, when there is
a demand peak and a reduction of the non-dispatchable renewable
sources. At the end of the day, the energetic contribution of the
CHP and the non-dispatchable renewable summed up to 126.37
MWh (23.27% of the total energy generation of 542.97 MWh).
Fig. 6 shows the system being controlled by the proposed structure,
comparing the process demands with the system outputs. As it can
be seen, the controlled plant is able to produce enough middle
pressure steam, low pressure steam, cold water flow and electric
power to fulfil the demands of ethanol and sugar production
process.
Fig. 7 shows the stocks behaviour of the proposed system. In
this simulation scenario, the battery bank acts to mitigate small
energy imbalances between the microgrid equipments, while
staying around 50%. At the end of the day, from 18 to 22 h, when a
high energy fluctuation happened, as a consequence of the electric
demand peak, the battery bank presented a SOC variation of +
15%, in order to absorb the energy generation excess. The biogas
stock starts with 80% and decreases down to around 50% (steady-
5
Fig. 4 Electric energy generation on the proposed system

Fig. 5 System disturbances

state), and is directly related with the CHP energy generation. The Fig. 8 shows the optimiser Optα performance when applied to
stock of bagasse is maintained continuously loaded, due to the the proposed system. This optimiser is able to maximise the gain of
great amount of income, as seen in Fig. 5. The straw stock, on the the boilers by reducing the biomass blending factors α A and αB to
other hand, gradually decreases, due to the biomass mixture 0.5, whenever possible. It can be noted that the optimiser cannot
optimisation, and the smaller amount of income. maintain the blending factors at the optimal condition (0.5)
throughout the whole day, because this forces the straw outcome
6 IET Renew. Power Gener.
© The Institution of Engineering and Technology 2017
Fig. 6 Fulfilment of the demands
Fig. 7 Behaviour of the stocks
Fig. 8 Behaviour of the blending factors
flow to be too voluminous and violate the minimum straw stock
constraint, as it is shown in Fig. 7. For this, the blending factor α A
gradually tends to 1. αB, in proposed scenario, can be maintained
around 0.5 as the lower efficiency boiler production is very low,
due to the low biomass mixture flow QEB.
The behaviour of the variables involved in the different steam
flow productions are shown in Fig. 9. As the high efficiency boiler
presents a great steam flow output to power up both turbines (for
IET Renew. Power Gener.
© The Institution of Engineering and Technology 2017
energy generation), the lower efficiency boiler is almost not used.
The middle pressure steam demand is attended (mainly) by the
high-middle pressure reduction valve and, also, by the lower
efficiency turbine's steam residue. As there is a residual middle
pressure steam flow, it is converted to low pressure steam flow.
The low pressure steam demand is attended by the production of
the higher efficiency turbine and the converted middle pressure
steam. The steam excess is discarded by the escape valves.
7
Fig. 9 Steam production

Fig. 10 Cold water production

Fig. 10 shows the variables involved to attend to the represents the energy generation by the microgrid's non-
refrigeration demands. As it can be seen, the CHP, the solar water dispatchable renewable sources.
heater and the heat exchanger are responsible for production the
hot water flow, stocked by the tank. The water chiller is responsible 5 Conclusion
for cooling the hot water to tend to the refrigeration demands. The
tank escape valve was not used because, as shown in Fig. 7, the hot This paper presented the issue of controlling a microgrid that
water tank level does not reach the maximum level (restriction). integrates renewable energy generation, biomass mixture
A series of daily simulation scenarios, considering different optimisation and hybrid storage technologies. The microgrid is an
prediction errors are detailed in Table 2. It is important to note that enhanced plant of a real sugar cane processing system, that
the total energy generation, in these simulation, have been set produces ethanol and sugar, with bagasse, straw and vinasse
within a limited percentage (5%) to the reference set-point (converted into biogas) tailings. A two-level hierarchical MPC
production of 384 MWh per day. Renewable energy generation control structure was proposed to perform the electric energy
production optimisation, management of storage and subsystems
and biomass blending optimisation. As showed by the simulation
Table 2 Different Daily Simulation Scenarios results, the proposed control structure presented satisfactory
Prediction error, % Renewable energy Total energy results.
generation, MWh generation, MWh For further works, an interesting theme is to study a higher level
1 42.573 395.60 management system, considering a fixed contract of electric energy
2 42.143 395.50 production to be diluted upon several microgrids structures and
5 40.853 395.10 greater prediction error upon the disturbances. It is also imperious
10 38.703 395.26 to study the proposed control strategy coupled with disturbance
estimation techniques, hence the predictive controller can consider
15 36.553 395.33
the long term renewable forecast for monthly planning.
20 34.400 394.96
25 32.252 394.66
35 30.108 394.86

8 IET Renew. Power Gener.

© The Institution of Engineering and Technology 2017
Table 3 Simulation parameters
Optimiser tuning parameters
N p = 12
Nc = 5
qu = [0.3 0.6 0 10 0.4 5 5 0.01 0 10 0.5 0.35 2 × 106]
Qx = diag 0.3 0 0 0 0
3
qϵ = 1 × 10
QP = 0.5
Q A = QB = 30
QStr = 5
Constraints parameters
x = diag 10 10 10 10 10
x̄ = diag 100 100 100 100 100
u = diag 11 × 103 3 × 103 0 0 0 0 0 800 0 0 −3 × 103 0 0
ū = diag 27 × 103 2 × 104 0 1 × 107 0 0 0 8 × 10 3
0 0 3 × 103 0 0
Battery bank ratings
Maximal energy flow= ± 3 MW
Total energy capacity= 3 MWh

6 References [20] Galus, M.D., Andersson, G. : ‘Power system considerations of plug-in hybrid
electric vehicles based on a multi energy carrier model’. ‘Power Energy
[1] Johansson, T.B.: ‘Renewable energy: sources for fuels and electricity’ (Island Society General Meeting. PES ’09 IEEE’, 2009, pp. 1–8
press, 1993) [21] Da Costa Mendes, P., Normey-Rico, J.E., Bordons, C., et al.: ‘Economic
[2] Shafiee, S., Topal, E.: ‘When will fossil fuel reserves be diminished?’ energy management of a microgrid including electric vehicles’. ‘Innovative
(Elsevier, 2009) Smart Grid Technologies Latin America (ISGT LATAM) IEEE PES’, 2015,
[3] Jiayi, H., Chuanwen, J., Rong, X., et al.: ‘A review on distributed energy pp. 869–874
resources and Microgrid’, Renew. Sustain. Energy Rev., 2008, 12(9), pp. [22] Galus, M.D., La Fauci, R., Andersson, G., et al.: ‘Investigating PHEV wind
2472–2483 balancing capabilities using heuristics and model predictive control’. Proc.
[4] Ministério de Minas e Energia, Governo Federal: ‘Resenha Energética Int. of Power and Energy Society General Meeting, IEEE, 2010, pp. 1–8
Brasileira: Exercício de 2014’, 2015, pp. 6–19 [23] Morais, H., Kadar, P., Faria, P., et al.: ‘Optimal scheduling of a renewable
[5] Tiba, C.: ‘Atlas Solarimétrico do Brasil’, 2000 micro-grid in an isolated load area using mixed-integer linear programming’,
[6] González, J.R.P.: ‘Libro Blanco de la Automatización y Control en la Renew. Energy, 2010, 35(1) pp. 151–156
Industria de la Caña de Azúcar’. Programa Iberoamericano de Ciencia y [24] Arnold, M., Negenborn, R.R., Andersson, G., et al.: ‘Model-based predictive
Tecnología para el Desarrollo (CYTED), 2011 control applied to multi-carrier energy systems’. Power Energy Society
[7] Alves, J.M.: ‘Paradigma técnico e co-geração de energia com bagaço de cana General Meeting, PES ’09, IEEE, 2009, pp. 1–8
de açúcar em Goiás’. Proc. Int. 6 Encontro de Energia no Meio Rural, Brazil, [25] Parisio, A., Rikos, E., Glielmo, L., et al.: ‘A model predictive control
2006 approach to microgrid operation optimization’. Proc. Int. of IEEE
[8] Rego, E.E., Hernandez, F., Del, M.: ‘ Eletricidade por digestão anaeróbia da Transactions on Control Systems Technology, 2014, 22(5) pp. 1813–1827
vinhaça de cana-de-açúcar: contornos técnicos, econômicos e ambientais de [26] Bozchalui, M.C., Hashmi, S.A., Hassen, H., et al.: ‘Optimal operation of
uma opção’. Proc. Int. 6 Encontro de Energia no Meio Rural, Brazil, SciELO residential energy hubs in smart grids’, IEEE Trans. Smart Grid, 2012, 3(4),
Brasil, 2006 pp. 1755–1766
[9] Pinto, C.P.: ‘Tecnologia da digestão anaeróbia da vinhaça e desenvolvimento [27] Parisio, A., Glielmo, L.: ‘A mixed integer linear formulation for microgrid
sustentável’,1999 economic scheduling’. Proc. Int. of IEEE Int. Conf. on Smart Grid
[10] Costa Filho, M.V.A.da: ‘Modelagem, controle e otimização de processos da Communications (SmartGridComm), 2011, pp. 505–510
indústria do etanol’, 2013 [28] Sánchez, M., Bernardo, J. : ‘Optimización de metodologías para la
[11] Greenweel, W., Vahidi, A.: ‘Predictive control of voltage and current in a fuel caracterización de biocombustibles sólidos procedentes de la industria del
cell-ultracapacitor hybrid’, IEEE Trans. Ind. Electron., 2010, 57(6), pp. 1954– olivar’ (Servicio de Publicaciones Universidad de Córdoba, 2015)
1963 [29] Schulze, M., Del Granado, P.C.: ‘Multi-period optimization of cogeneration
[12] Ferrari-Trecate, G., Gallestey, E., Letizia, P., et al.: ‘Modeling and control of systems: considering biomass energy for district heating’. Proc. Int. on 2nd
co-generation power plants: a hybrid system approach’, IEEE Trans. Control Power Systems Modeling Conf., 2009
Syst. Technol., 2004, 12(5), pp. 694–705 [30] Viana, L.F.: ‘Potencial energético do bagaço e palhiço de cana-de-açúcar, cv.
[13] Del Real, A., Galus, M.D., Bordons, C., et al.: ‘Optimal power dispatch of SP80-1842, em área de alambique artesanal’ (Universidade Federal de
energy networks including external power exchange’. Proc. Int. European Lavras, 2014)
Control Conf. (ECC), 2009, pp. 3616–3621 [31] Tolentino, G., Florentino, H.D.O., Sartori, M.M.P., et al.: ‘Modelagem
[14] Manwell, J.F.: ‘Hybrid energy systems’, Encycl. Energy, 2004, 3, pp. 215–229 matemática para o aproveitamento da biomassa residual de colheita da cana-
[15] Bakken, B.H., Haugstad, A., Hornnes, K.S., et al.: ‘Simulation and de-açúcar com menor custo’ (Bragantia, 2007), pp. 729–735
optimization of systems with multiple energy carriers’. Proc. Int. of [32] Júnior, R., Agudo, R.: ‘Análise da viabilidade do aproveitamento da palha da
Scandinavian Conf. on Simulation and Modeling, Sweden, Citeseer, 1999 cana de açúcar para cogeração de energia numa usina sucroalcooleira’
[16] Lasseter, R.H.: ‘Microgrid’. Proc. Int. of Power Engineering Society Winter (Universidade Estadual Paulista, 2009)
Meeting, IEEE, 2002, pp. 305–308 [33] Geidl, M.: ‘Integrated modeling and optimization of multi-carrier energy
[17] Valverde, L., Rosa, F., Del Real, A., et al.: ‘Modeling, simulation and systems’ (TU Graz, 2007)
experimental set-up of a renewable hydrogen-based domestic microgrid’, Int. [34] Mathworks: ‘Matlab’, 2009
J. Hydrog. Energy, 2012, 38(27), pp. 11672–11684 [35] Lofberg, J.: ‘YALMIP: a toolbox for modeling and optimization in
[18] Garcia-Torres, F., Bordons, C.: ‘Optimal Economical Schedule of Hydrogen- MATLAB’. IEEE Int. Symp. on Computer Aided Control Systems Design,
Based Microgrids With Hybrid Storage Using Model Predictive Control’, 2004, pp. 284–289
IEEE Trans. Ind. Electron., 2015, 62(8), pp. 5195–5207 [36] ILOG, IBM: ‘CPLEX’, 2007
[19] Petrollese, M.: ‘Optimal generation scheduling for renewable microgrids
using hydrogen storage systems’, 2015

IET Renew. Power Gener. 9

© The Institution of Engineering and Technology 2017