Professional Documents
Culture Documents
Applied Energy
journal homepage: www.elsevier.com/locate/apenergy
h i g h l i g h t s
We model hybrid and conventional ship fuel consumption for crane operation.
Hybrid ship model based on data from ship owner.
Battery, diesel generator and control systems modelled.
A conventional and two hybrid control systems has been modelled.
The novel hybrid system has lower capital costs and saves 30% fuel and emissions.
a r t i c l e i n f o a b s t r a c t
Article history: We have investigated a hybrid power train for ship crane operations, using a lithium-ion battery in
Received 25 March 2014 conjunction with diesel gensets for auxiliary power generation, as an alternative to a conventional power
Received in revised form 22 October 2014 train using only gensets. We have simulated crane operations in port using both solutions, in order to
Accepted 17 January 2015
quantify the potential economic gains of using hybrid power generation. This study is based on a real
Available online 5 March 2015
open-hatch dry bulk vessel of 50,000 dwt, which is compared with a corresponding newbuilding ship
with hybrid auxiliary power generation. We have modelled the complete auxiliary power system, includ-
Keywords:
ing diesel generators, lithium-ion batteries, cranes and ship hotel consumers. We have developed a novel
Hybrid ship
Lithium-ion
hybrid control strategy that has the potential to reduce the minimal size and thereby cost of batteries for
Battery hybrid ships. Our results indicate that the hybrid solution will lead to about 30% reduced fuel consump-
Marine tion and CO2 emissions while operating cranes, which amounts to annual savings of $110,000, with
Simulation $450,000 savings over three years of operation, as well as reduced capital costs compared to the conven-
Economic analysis tional power generation system.
Ó 2015 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.apenergy.2015.01.066
0306-2619/Ó 2015 Elsevier Ltd. All rights reserved.
E. Ovrum, T.F. Bergh / Applied Energy 152 (2015) 162–172 163
Nomenclature
Hybrid a ship with energy storage, e.g. battery C cinst conventional installation costs, $
q, SOC state of charge C hop hybrid annual operational costs, $
C-rate normalized measure of battery current C cop conventional annual operational costs, $
Genset diesel engine with generator cb battery price, $/kWh
SFOC specific fuel oil consumption, g/kWh cg diesel genset cost, $/kWh
B, Capacity max. energy content of battery, kWh cm maintenance and repair costs, $/h
Pbmax maximum simulated (dis) charge power of the battery, cf fuel price, $/t
kW cp power electronics costs, $/kW
dwt deadweight tonnage, used for ship sizes fh hybrid fuel consumption, t/ year
PID a proportional-integral-derivative controller fc conventional fuel consumption, t/ year
E diesel genset engine size, kW th genset running time hybrid system, h
C max maximum C-rate tc genset running time conventional system, h
C min minimum C-rate x years before payback
Ng number of installed generators y years until operational savings equal battery cost
C hinst hybrid installation costs, $
is used for both hotel consumers (lighting etc.) and onboard cranes. energy to the ship’s power consumers in the same way the battery
The simulations rely on randomly generated auxiliary power in the Toyota Prius hybrid car does. In shipping, the term hybrid
demand profiles, based on data from the ship owner. The battery ship can have several alternative meanings, but in our case we will
model is based on a real battery pack produced by Corvus [6] use it in the same sense as in the automotive industry.
which has been installed on the Viking Lady [2]; and on To estimate the total cost difference between a hybrid ship and
Norway’s first battery-only ferry [4], which will start operating in a conventional ship, there are two main cost aspects that must be
2015. Propulsion power is provided by a main engine not connect- addressed:
ed to the auxiliary grid, and has not been included in the
modelling. Capital costs – Installation costs of the battery and auxiliary
By simulating the power generation versus a given power engines.
demand in time, we can test different hybrid power generation Operational costs – How can the battery contribute to reduced
control strategies. By adding a battery to the power generation sys- fuel consumption and maintenance during operation.
tem, the control strategy shifts from being just a question of having
the optimal number of gensets running, to a more complex issue. There are several compelling arguments to be made for hybrid
To illustrate, in a hybrid system one can run a different number power generation:
of gensets, or none at all; one can run the gensets at chosen loads,
letting the batteries take the responsibility for meeting the power Optimal load – The energy efficiency and emissions of diesel and
demand; the batteries can be charged or discharged at given rates, gas engines depend on the engine load, being far worse for low
and the gensets have to take responsibility for meeting the power loads than high loads (low load means low power output). In a
demand; one can have rules stating the minimum or maximum hybrid ship, the battery can be used to handle deviations from
capacity left in the battery; and more. We have developed two dif- average demand and allows the conventional engines to operate
ferent hybrid ship power generation control strategies, both of closer to the optimal load point.
which meet the power demand and result in a significant fuel con- Transients – The energy efficiency of diesel and gas engines is
sumption reduction compared to the conventional system. The lower when the engine is changing its load, while the emissions
first strategy, called the basic control strategy, slowly changes of SOx, NOx and particulate matter increase. In a hybrid system,
the genset loads to maintain the battery’s SOC, this has been used the battery can handle the transient demand while the conven-
by other groups simulating hybrid ships, e.g. [7]. Our second strat- tional engines deliver the average power.
egy, called the advanced control strategy, employs a novel mode Power redundancy – In some cases, a battery pack can presum-
where the genset will increase its load when the ship’s power ably count as a power source, allowing ship owners to install
demand is high and vice versa, requiring a substantially smaller fewer engines while still satisfying class rules for redundancy.
battery. Facilitate energy harvesting – Energy storage is necessary for effi-
We have used the simulations to determine a minimal battery cient onboard harvesting from renewable sources, such as solar
size for the conditions of our case study ship, and used this to esti- cells.
mate the costs, savings and payback time of a hybrid system com- Harbor mode – If the battery installation is large enough (which
pared with the conventional one. is not the case in this study) the battery can provide all the pow-
In Section 2 we discuss the modelling of lithium-ion batteries; er while in port, eliminating local emissions.
in Section 3 we discuss the modelling of the hybrid crane system;
in Section 4 we discuss the case study; in Section 5 we discuss the
1.2. Hybrid cranes case study
results; in Section 6 we give our conclusions.
To accurately quantify the benefits of hybrid auxiliary power
1.1. Hybrid ships generation we need detailed operational profiles. This is because
the potential for savings lie mostly in time-varying power
A hybrid ship in this context is a ship that has an energy storage demands [8]. Vessel types that have mostly constant power
device as part of its power generation system, e.g. flywheel, com- demands, such as long-range freight vessels are expected to gain
pressed air or electrochemical batteries. The energy storage unit less from hybrid power trains, while a ship operating cranes or
accumulates energy from the ship’s power generators or releases using dynamic positioning will have greatly varying power
164 E. Ovrum, T.F. Bergh / Applied Energy 152 (2015) 162–172
demand and therefore greater potential for reduced fuel consump- Table 1
tion with a hybrid system. Crane operations are especially suited The conventional and hybrid power generation system configurations. The conven-
tional system has three gensets installed, while the hybrid system replaces one genset
for hybrid power generation, because cranes can be run by electric with a battery pack with a capacity of B kWh. The second engine in the hybrid system
motors that can use regenerative braking. is not used in simulations, and is only retained for redundancy purposes.
The ship we have based our models on is an open hatch dry bulk
Conventional Hybrid
vessel of about 50,000 dwt. The ship uses three slewing type cargo
cranes1 powered by electric motors. We have obtained typical power 960 kW Diesel genset 960 kW Diesel genset
960 kW Diesel genset (960 kW Diesel genset)
demand profiles in port for this vessel type from the ship owner, see 960 kW Diesel genset B kWh Battery pack
Section 3.2 for the details. As the cranes are fully electric the descent
of the cranes’ load is slowed by using the electric motors as
generators, thus generating power. This is commonly referred to as
demand an electrochemical or a phenomenological model, as well
regenerative braking. In the actual vessel the generated energy from
as access to significant amounts of experimental data. Our primary
the crane braking is wasted, while in our hybrid model it is returned
objective was to investigate on a high level the benefits of hybrid
to the grid, and potentially used to charge the battery, if the total
power generation on ships, and in light of that we decided to
power demand of hotel consumers and cranes is negative.
implement an equivalent circuit model of a lithium-ion battery
In Table 1 we see the power generation system (auxiliary only,
with a state of charge dependent voltage.
not propulsion) configuration of the existing conventional ships,
with three gensets, and of the proposed hybrid ship, with a
lithium-ion battery pack replacing one of the gensets. 2.2. Model description
Lithium-ion batteries have several advantages over competing
energy storage technologies, in particular, higher energy and pow- The simplest equivalent circuit models of batteries consist of a
er densities (both to-weight and to-volume ratios), which are voltage source and a resistance in series. The most common
important design parameters for ships. In addition, lithium-ion elaboration of this simple model is to include one or more
batteries have already been chosen to be used in Norway’s first RC-loops in series. An RC-loop consists of a resistance and a
battery-powered ferry, and in the hybrid offshore supply vessel capacitance in parallel and is used to represent relaxation time
Viking Lady, which has been retrofitted with fuel-cells and batter- constants. In [10] a comparative study of equivalent circuit models
ies in connection with the project FellowSHIP III, sponsored by The for lithium-ion batteries, comparing lab tests with several different
Research Council of Norway. models, concluded that ‘‘the first-order RC model is preferred for
LiNMC cells’’. We tried running simulations with a single RC-loop
included in the battery model with a wide range of values of the
2. Battery modelling associated time constant (s ¼ RC), but found that this had no
significant impact on the fuel consumption of the hybrid system.
There are two major approaches to modelling batteries: We therefore chose to include only two components in the equiva-
Equivalent-circuit models, and electrochemical models. An lent circuit model: a voltage source and a resistance.
overview of various modelling strategies and the main battery In our model the voltage of the battery depends only on the
characteristics is given in Jongerden et al. [9]. Circuit models are state of charge (SOC), q, and the current through it, I (positive when
inherently dynamic, providing a relationship between the voltage discharging), and is given by
and current over time, see [10] for a comparative study of circuit
models. In addition, the models can include auxiliary models to U ¼ U 0 ðqÞ RI; ð1Þ
represent the batteries’ capacity, temperature dependency (e.g.
where R is the internal resistance of the battery and U 0 is the
[11]), state of charge dependency, as well as performance degrada-
open-circuit voltage (OCV).
tion. In [12] an equivalent circuit model with a temperature part is
We assume that U 0 only depends on the SOC, q, and disregard
calibrated to measurement data from an 18,650 cell. The model
the temperature-dependence, assuming a fixed temperature of
yields good results for transient response simulations and for ther-
25 °C. As mentioned earlier, we base our model on battery packs
mal behaviour, although deviating for low temperatures and high
manufactured by Corvus [6], which in turn consist of cells made
discharge rates. The same circuit model is used in [13], where
by Dow-Kokam [16]. We have fitted U 0 to data published by the
the cell’s complex impedance is simulated, adding another
cell manufacturer of experimentally measured voltage versus state
important tool for calibrating battery models. In [14] three differ-
of charge at 25 °C and 0.5 C, which is shown in Fig. 1 along with
ent circuit models are compared, while in [15] a circuit model is
corresponding curves for other C-rates. C-rate is a measure of the
used to simulate plug-in hybrid electric vehicles.
current through a battery, normalized by the capacity of the bat-
tery such that a C-rate of 1 (or 1 C) corresponds to a current that
2.1. Model specification and selection discharges the battery in one hour. In the published data we note
that there is not much difference between the cell voltage at
To model a battery in a hybrid power system one can in 25 °C and 60 °C in the SOC range between 10% and 100%, which
principle use both an electrochemical model as well as an equiva- we argue justifies the omission of the temperature dependence.
lent circuit model. Both these classes of models can provide the Since U 0 is fitted to a voltage curve for a finite discharge current
voltage, current and state of charge, which are the most important it is not exactly equal to the true open-circuit voltage, but since
parameters in this context. Equivalent circuit models are usually the voltage curve for 1 C is almost the same as for 0.5 C, we con-
simpler and somewhat less accurate than electrochemical models, clude that an extrapolation to 0 C will not be substantially different
but are far easier to use, especially in connection with other com- from 0.5 °C. The modelled open-circuit voltage, U 0 , is seen in Fig. 2.
ponents. Accurately modelling the lifetime of the battery however, To determine the value of the internal resistance, R, we have fit-
which is needed to estimate the time to replacement, cannot be ted the modelled voltage U, given by Eq. (1), to the voltage curves
done directly using an equivalent circuit model, and will typically for various C-rates (and by extension, for different currents) shown
in Fig. 1, using the least squares method. This procedure yields a
1
The ship has four cranes, but due to issues with manning three cranes are value for R ¼ 565:5 mX. Running simulations with 50% increa-
normally used. sed/decreased value of R results in a change of fuel consumption
E. Ovrum, T.F. Bergh / Applied Energy 152 (2015) 162–172 165
Table 2
Sensitivity analysis of the AC/DC converter efficiency. In column g we give the
converter efficiency, in column f we give the difference in fuel consumption of the
hybrid system (relative to the baseline), and in column s we give the difference in
payback time of the hybrid system, see Eq. (10).
X
3
PðWith regenerationÞ ¼ P Hotel þ Pi ; ð3Þ
i¼1
X
3
PðWithout regenerationÞ ¼ PHotel þ max ðPi ; 0Þ; ð4Þ
i¼1
The diesel genset models are lookup tables based on data from
genset manufacturer brochures [27]. For a given load,
Fig. 4. The conventional control system. A PI-controller maintains the genset loads
l 2 ½10%; 100%, the fuel consumption is linearly interpolated from
(which are the same for all three) such that the power delivered is equal to the
a table, and in Fig. 3 the resulting SFOC as a function of load is power demand from the crane operations and the hotel load.
shown. The electric power output is a function of the load times
the maximal power output, P ¼ lP max , with Pmax ¼ 960 kW. The gen- In the existing system, one or more gensets can be shut off, but the
set lookup table model is a steady state model that does not take running gensets are always at equal load. The load is determined
into account increased fuel consumption at transient conditions. by a PID controller which assures that the total power generated
is equal to the total power demand. The conventional system con-
3.4. Conventional control system sists of three equal-sized diesel gensets, each with nominal electri-
cal power output of 960 kW. The existing control system uses two
We designed the control strategy for the conventional system gensets when two cranes are in use and three gensets when three
(see Fig. 4) where three gensets constitute the conventional power or four cranes are in use, this means that in our simulations all
generation system, to be as close as possible to the existing system. three gensets are being used.
E. Ovrum, T.F. Bergh / Applied Energy 152 (2015) 162–172 167
The advanced control strategy is the same as the basic one, except
that it has an additional control mode, which we call ‘‘limit
C-rate’’ mode (we will refer to the only mode of the basic control
strategy as the ‘‘normal’’ mode). The advanced control mode is illus-
trated in Fig. 6. The motivation for developing the advanced hybrid
control strategy is to enable a smaller (hence, cheaper) battery pack,
at the expense of greater transients in the engine load. Greater tran-
sient loads may mean less efficient power generation, although the
Fig. 6. The ‘‘limit C-rate’’ control mode. As in the normal mode, the battery
transients are still greater for the conventional system, an effect provides/receives the difference in power between what is required by the
which we have not modelled. This control mode is engaged when consumers and what the genset generates. However, the genset load is determined
the normal mode battery C-rate would exceed the battery safety by a PI-controller which aims to keep the battery C-rate at either C min or C max , when
the corresponding limit would be exceeded in normal mode. The ‘‘limit C-rate’’
limits, denoted by C min /C max . In this mode the generator load will
mode will then increase the genset load to assist the battery, if needed, when the
be increased or decreased in order to keep the battery C-rate within battery C-rate is already at C max ; and decrease the genset load, if needed, when the
the safe range. While in this mode the average load and battery SOC battery C-rate is already at C min .
is ignored (though the normal mode will be forced if the SOC safe
limits are reached, i.e. the battery is in danger of running out of
charge or being overcharged). The safe C-rate range for the battery
costs of the conventional and hybrid systems. First we describe the
is based on discussions with the manufacturer [16], namely 2C/4C,
conventional system and then describe how this system could be
which means that the maximal charge current corresponds to 2C
redesigned as a hybrid system. We then use the fuel savings per
and the maximal discharge current corresponds to 4C. The C-rate
year of operation to calculate the payback time of the alternative
may exceed these limits without causing significant harm to the
design, as a function of capital costs and operational costs. We do
battery for short intervals of 1–10 s. However, as we can see from
this for both the basic and advanced hybrid systems described in
the results in Fig. 8, even though the maximum charging value is
Section 3.5.
2C, the average C-rate is closer to 1C/1C.
The conventional system consists of three diesel gensets provid-
In Fig. 7 we show 12 min of simulated crane operation using the
ing all the electric power for the hotel and cranes on the ship, see
basic hybrid strategy.
Table 1. The hybrid system has a battery installed to assist with the
power load when the power demand is high, and to be charged
4. Case study when the power demand is low. In the hybrid system there is no
need to install three diesel gensets; in fact the system only requires
To analyze the economic benefits of installing a battery on the one diesel genset as the average load in the test case is less than
case study ship, we consider both the capital costs and operational 960 kW. Due to redundancy requirements, however, we install
168 E. Ovrum, T.F. Bergh / Applied Energy 152 (2015) 162–172
C cop ¼ cf f c þ cm t c ; ð8Þ
C hop ¼ cf f h þ cm t h : ð9Þ
We define the payback time, x, as the years before the costs of the
hybrid and conventional systems are equal,
yDC op ¼ cb B; ð12Þ
4.1. Cost analysis
assuming no change in battery prices. This tells us that as long as
There are two main operational costs we consider for the con- the battery life is longer than y, the hybrid system will be more
ventional and hybrid systems, fuel consumption and diesel engine profitable than the conventional.
maintenance and repairs. Diesel engine maintenance and repair
contribute significantly to the operational costs of ships, and the 4.2. Determining battery size
ship owner has provided us with an estimated cost of maintenance
and repair for the diesel engines per hour of operation. The hybrid For an application with a given maximum power demand, such
system only runs one diesel engine at a time, while the conven- as we have in our case study of the hybrid cranes, we determine
tional system uses three engines at a time (see Section 3.4), which the necessary size or capacity, in kWh, of the battery pack from
means that for every hour the ship is in port loading the hybrid the C-rate limitations of the battery and the maximum C-rate
system will have one hour of engine operation, while the conven- encountered in simulated operations. As we discussed in
tional has three hours of operation. Section 3.5, we set the battery C-rate limits at 2C/4C, meaning
In Section 5 we present a cost analysis based on ranges of prices maximally sustained2 2C charging and 4C discharging of the battery.
for fuel and batteries. Today, the fuel price is about 1000 $/t (dol- The C-rate limits are recommended by cell manufacturers in order to
lars per ton) for marine diesel oil (MDO), and about 650 $/t for limit degradation of the cells, as excessive currents will lead to dra-
heavy fuel oil (HFO). In [25] the prices for typical auxiliary diesel matically shorter useful lifespans. Different battery chemistries have
engine generator sets is set to 350 $/kW. The battery price today different current tolerances, and some manufacturers also provide
is roughly 1000 $/kWh, while an estimated price for the power so-called ‘‘power cells’’ that focus on being able to deliver high
electronics, in $/kW, was obtained through private communica- C-rates, rather than high capacities. This means that deciding upon
tions with a power electronics manufacturer. We find the annual which battery to use for different hybrid ship applications requires
fuel consumption by multiplying our simulated fuel consumption a good understanding of the details of the ship’s operation.
for one hour of crane operations with the shipowner’s estimates The strategy of the hybrid control system greatly impacts the
that the ship is loading or unloading for 1470 h a year. required battery size. Using the basic hybrid control strategy, the
In calculating the profitability of the hybrid system, we look at maximal charge current and the maximal discharge current of
the payback time of the hybrid system compared to the conven- the battery are about the same. In this case the necessary battery
tional one. The payback time is the time to break even, i.e. when size is determined by the maximal discharge C-rate of 2. Using
the installation and operational costs are equal for the hybrid sys- the advanced strategy allows for a battery of smaller capacity,
tem and the conventional systems. and in this case the battery size is limited by the maximal dis-
The capital costs for conventional system (Section 3.4) and the charge C-rate of 4.
hybrid system (Section 3.5) are given by Our strategy for determining the battery size yields results that
are highly dependent on the actual operation of the hybrid ship,
C cinst ¼ Ng cg E; ð6Þ with the maxima and minima in the ship’s power demand being
C hinst ¼ Ng cg E þ cb B þ cp Pbmax : ð7Þ the main input parameters (yielding maximal power into and out
of the battery). The profitability of the hybrid system, illustrated
The maximum simulated (dis) charge power of the battery, Pbmax ,
and the battery size, B, will be different for the basic and the 2
More than 10 s.
E. Ovrum, T.F. Bergh / Applied Energy 152 (2015) 162–172 169
by our calculations of the payback time, depends very much on the savings as the basic one, but requires only half the battery size in
required capacity of the battery pack, which was the motivation for this case, significantly reducing the installation costs of the hybrid
developing the advanced control strategy. system. This result indicates that there needs to be done further
research to find different control strategies that can be suitable
for different types of hybrid ship operations.
5. Results
5.2. Lifetime considerations
In this section, we show the results from modelling the opera-
tion of three cranes on a hybrid ship. We model the operation of
In order to perform an accurate analysis of the savings of the
the cranes for one hour using ten different randomly generated
battery throughout its life, one must use a battery model that
power demand profiles as described in Section 3.2. We average
includes performance degradation due to age. The annual savings
the results for the ten different profiles to get a representative hour
predicted by a model with performance degradation would
of operation. For this hour of operation we calculate the fuel con-
decrease as the battery aged. We have not included performance
sumption of the hybrid and conventional systems, and we multiply
degradation in our models, but instead focused on a shorter time
this by the hours of crane operation per year (see Section 4.1) to get
frame, aiming to find the payback time of the investment.
the annual fuel consumption. We have not modelled emissions of
There are several factors influencing the useful lifetime of a bat-
NOx ; SOx and particulate matter, but as seen in [5] we expect similar
tery. Cell temperature is a very important aspect; a ship operating
or greater reduction of these emissions as well. We then proceed to
in the Persian Gulf might have significantly shorter battery life-
discuss the validity of our model and how reliable our results are.
times than a ship operating in the North Sea due to the high ambi-
Below we show the annual savings of the hybrid systems, the pay-
ent temperatures. In this paper we do not consider temperature or
back time of the hybrid systems, and time until the savings equal
the energy consumption of a temperature control system for the
the price of a new battery (see Section 4.1). Fig. 8 shows results from
battery installation.
a 1 h simulation of the two hybrid control strategies.
The other important factor for battery lifetime is cycling. Cycles
with a high depth of discharge are more harmful to the battery than
5.1. Economic results shallow cycles. We have not modelled degradation effects in the
battery, but we do model the cycling of the battery and can there-
In Table 4 we show the results from the comparison between fore give an indication of whether or not the battery will have a long
the hybrid and conventional systems, for both the basic and the life before needing to be replaced. In Fig. 8 we show the engine load
advanced control strategies. The fuel consumption for the two and C-rate of the battery operating for one hour. From the figure
hybrid strategies and the relative fuel savings compared with the one can see the highly fluctuating C-rate as the battery meets the
conventional setup are almost identical, but the battery size for varying power demand of the system. At the same time the SOC
the basic hybrid strategy is 312 kWh while only 156 kWh for the of the battery hardly changes, fluctuating between 0.68 and 0.72,
advanced hybrid. The reason that the model predicts only a small for about 20 small cycles per hour. For the case of 20 cycles an hour,
difference in fuel consumption is that the diesel genset model the battery would have 20 1470 30; 000 micro cycles a year at
(see Section 3.3) does not take into account the increased fuel con- about 3% DoD, which according to the manufacturer would allow
sumption of diesel engines at transient loads, see 5.5. The advanced for hundreds of thousands of cycles. Without giving any definitive
control strategy simulations have diesel genset loads that are fluc- answer to the question of lifetimes, we conclude on the note that
tuating more than for the basic control strategy, but still less than the cycles are shallow and this will probably mean that the battery
for the conventional system, see Figs. 7 and 8. Therefore the lifetime will be determined by calendar fade rather than cycling. On
increased fuel consumption due to varying loads is greater for the other hand, the battery is constantly working at high C-rates
the conventional system than for the advanced control strategy and could reach high temperatures, which would accelerate the cal-
hybrid system. The annual operational savings of the hybrid com- endar aging and decrease its lifetime if not controlled.
pared with the conventional is shown in Fig. 9. As we cannot presently say what the lifetimes of the batteries
In Fig. 10 we show the contour map of x, the payback time for will be, we instead show how long it will take for the ship owner
the basic hybrid control strategy: the time until the installation to have saved enough to be able to buy another battery, see
and operational costs of the hybrid and conventional systems are Fig. 12. In the cases we have investigated this time is far lower than
equal. When we turn to using the advanced control strategy, how- what we expect the lifetime of the battery to be, and we therefore
ever, the battery can be smaller and the payback time is shorter, do not perform a full fledged life cycle analysis of the hybrid ship.
negative in fact, meaning that the installation costs are lower for The expected battery lifetime is a very important design para-
the hybrid than the conventional system, see Fig. 11. meter as the designer must take into account:
In Fig. 12 we show the contour map of y for the advanced con-
trol strategy, the time it takes for the operational savings of the Battery replacement: The battery must be replaced once its per-
hybrid system to equal the price of buying a new battery. As long formance has fallen below a certain level.
as the battery has a lifetime greater than y, the investment of Capacity fade: The capacity, the maximum of stored energy, of
installing the hybrid system will pay off. For the basic hybrid y is the battery is decreased both from cycling and from the age of
close to three years at today’s prices, while for the advanced hybrid the battery itself. The capacity fade must be taken into account
it is about a year and a half at today’s prices. when designing the battery so that it can deliver the required
In Fig. 13 we show the total savings of the two different hybrid power and energy throughout its life. Therefore the battery size
installations compared to the conventional system, as a function of must be increased from the ideal size to compensate for capa-
time in years. We see that the basic hybrid installation is more city fade.
expensive than the conventional system, and the total savings at Voltage drop: As the battery is cycled and ages, the internal
the beginning, year zero, is negative. After three years, however, resistance increases. The increased resistance decreases the bat-
the total savings are more than $250 k. For the advanced strategy tery’s maximum voltage and power, its charge/ discharge effi-
the hybrid installation costs $120 k less than the conventional, ciency, and increases its heat generation. The increased
and after three years the total savings are higher than $400 k. resistance can lead to lower maximum power of the battery,
This novel control strategy results in approximately the same and higher fuel consumption in a hybrid system.
170 E. Ovrum, T.F. Bergh / Applied Energy 152 (2015) 162–172
Fig. 8. Engine load (top frame) and battery C-rate (bottom frame) for one hour of operation using the basic and the advanced control strategy. For the basic control strategy
the engine load varies only slightly, and the battery C-rate is mostly confined to the range [2, 2]. Using the advanced control strategy allows for a battery of only half the size
of the previous case. In this case the engine load varies more as a consequence of keeping the battery within the safe C-rate range [2, 4].
Fig. 9. The annual savings of the hybrid system compared to the conventional
system, taking into account fuel savings and reduced maintenance and repair. The Fig. 10. Contour plot of the payback time for the basic control strategy as a function
annual savings are the same for both the basic and the advanced control strategies. of fuel cost and battery cost (see Eq. (10)), with a battery size of 312 kWh. At current
At current fuel prices of about 1000 $/t the hybrid system has annual operational battery prices of about 1000 $/kWh and fuel prices of 1000 $/t the basic hybrid
savings of about $110,000. system payback time is less than one year.
E. Ovrum, T.F. Bergh / Applied Energy 152 (2015) 162–172 171
Fig. 12. Contour plot of the number of years before the savings from the hybrid
solution are large enough to buy another battery pack, for the advanced control
strategy. As long as the lifetime of the battery is greater than this, the hybrid
solution will be more profitable than the conventional solution. At today’s prices, Fig. 13. Total savings of the hybrid system compared to the conventional system,
the advanced hybrid system has saved up enough for another battery after about for both the basic control strategy (the lower dashed line) and the advanced
one and a half years. strategy (the upper line).
172 E. Ovrum, T.F. Bergh / Applied Energy 152 (2015) 162–172
Table 5 requires a battery size of 312 kWh that results in a payback time
This table shows the isolated benefits of the battery and the regenerative braking.
of less than a year; while the advanced hybrid control strategy,
No regeneration (t) Regeneration (t) D (t) with the novel ‘‘limit C-rate mode’’, requires a battery size of
Conventional 311 287 24 156 kWh yielding lower installation costs than the conventional
Hybrid 245 217 28 system and savings from day one for the advanced control strategy.
D 66 70 94
Acknowledgments
Diesel engine and power electronics technologies are highly
The authors would like to thank Jan Øivind Svardal from Grieg
matured and the prices will probably not change as much in the
Star for his essential contributions in terms of providing the vessel
coming years, we have therefore chosen to use fixed prices for
particulars and operational information.
genset and power electronics while using a range of prices for
Part of this work was done in connection with the project Low
batteries and fuel.
Carbon Shipping and the financial support by the Research Council
The ship we have modelled has diesel generator sets providing
of Norway is gratefully acknowledged.
electric power for the hotel, cranes and thrusters. The ship is not an
all-electric ship and uses mechanical power propulsion from its
variable speed diesel main engines. The mechanical propulsion References
means that when the ship is sailing, the battery cannot contribute
to propulsion, but the battery can be used for hybrid electric power [1] Corbett JJ, Winebrake JJ, Green EH, Kasibhatla P, Eyring V, Lauer A. Environ Sci
Technol 2007;41:8512–8.
generation. We have only modelled the crane operation, and have [2] Eidesvik offshore supply vessel Viking Lady. last visited 08.2013, <http://
not taken into account the benefits of hybrid power generation for vikinglady.no/>; 2013.
the hotel while the ship is sailing, maneuvering, or in harbour and [3] Østensjø platform supply vessel Edda Ferd. last visited 08.2013, <http://
ostensjo.no/?fleet=eddaferd>; 2013.
not loading. By not simulating all operational modes of the ship,
[4] Norled battery ferry, to start sailing in 2015. last visited 08.2013, <http://
our results understate the benefits of installing a battery and www.tu.no/industri/2012/11/09/her-er-verdens-storste-batteridrevne-ferge>;
hybrid system. 2013.
[5] Carolyn dorothy hybrid tugboat online report. last visited 08.2013, <http://
Overall, we have omitted details that can both increase and
www.arb.ca.gov/ports/marinevess/harborcraft/documents/hybridreport1010.
decrease the benefits of hybrid power generation for crane pdf>; 2010.
operations. Even though this is an early result, we believe our [6] Corvus energy. last visited 08.2013, <http://www.corvus-energy.com/>; 2013.
results might be too conservative, that they probably understate [7] Zahedi B, Norum LE, Ludvigsen KB. J Power Sources 2014;255:341–54.
[8] Vartdal B-J, Chryssakis C. In: Les Rencontres Scientifiques d’IFP Energies
the benefits of the hybrid installation on our case study ship. nouvelles – Int. Scient. Conf. on hybrid and electric vehicles – RHEVE; 2011.
[9] M. Jongerden, B. Haverkort, Battery Modeling, Technical Report, Centre for
Telematics and Information Technology, University of Twente; 2008.
6. Conclusions and recommendations
[10] Hu X, Li S, Peng H. J Power Sources 2012;198:359–67.
[11] Erdinc O, Vural B, Uzunoglu M. In: 2009 international conference on clean
We have described a simulation-based method of quantifying electrical power, ICCEP 2009, June 9, 2009–June 11, 2009. IEEE Computer
Society; 2009. p. 383–6.
the benefits of using hybrid power with batteries for energy
[12] Gao L, Liu S, Dougal RA. In: IEEE transactions on components and packaging
storage on ships, for the special case of crane operations. We have technologies, vol. 25, pp. 495–505.
modelled diesel generator sets, a battery pack, the cranes and the [13] Liaw BY, Nagasubramanian G, Jungst RG, Doughty DH. Fourteenth
ship hotel load, as well as developed two different control international conference on solid state ionics, June 22, 2003–June 27, 2003,
vol. 175. Elsevier; 2003. p. 835–9.
strategies for the hybrid power system in order to simulate power [14] Chen M, Rincon-Mora GA. IEEE Trans Energy Convers 2006;21:504–11.
generation of the ship. The battery model has been calibrated to [15] Kroeze RC, Krein PT. In: PESC ’08 – 39th IEEE annual power electronics
data from the Dow Kokam cells used by Corvus, the battery specialists conference, June 15, 2008–June 19, 2008. Institute of Electrical and
Electronics Engineers Inc.; 2008. p. 1336–42.
manufacturer for both the Viking Lady and Norway’s first battery [16] Dow Kokam data sheets, model slpb125255255h, must be ordered online. last
ferry. The presented method has been used in a case study of a visited 08.2013, <http://www.dowkokam.com/cell-products-technology/cells/
Grieg Star open hatch vessel with four cranes driven by electric kokam-cells/>; 2013.
[17] Liu W. Introduction to hybrid vehicle system modeling and control. Wiley;
motors, capable of regenerative braking. We have demonstrated 2013.
how one can use simulations to provide essential decision support [18] Hu X, Murgovski N, Johannesson L, Egardt B. Appl Energy 2013;111:1001–9.
for ship owners by quantifying the benefits of hybridization for a [19] Sorrentino M, Pianese C, Maiorino M. J Power Sources 2013;221:308–17.
[20] Zhang X, Mi CC, Yin C. J Power Sources 2014;245:292–300.
ship type that has not yet been hybridized.
[21] Wang L, Lee D-J, Lee W-J, Chen Z. J Power Sources 2008;185:1284–92.
We have estimated the optimal battery package size and the [22] Gully BH, Webber ME, Seepersad CC, Thompson RC. ASME 3rd international
annual fuel savings for the hybrid system, compared with the con- conference on energy sustainability, ES2009, July 19, 2009–July 23, 2009, vol.
1. American Society of Mechanical Engineers; 2009. p. 771–9.
ventional system. With given genset sizes and ship power demand,
[23] De Breucker S, Peeters E, Driesen J. Possible applications of plug-in hybrid
we used the manufacturer’s C-rate limitations to determine the electric ships. Inst. of Elec. and Elec. Eng. Computer Society; 2009. p. 310–317.
minimal battery size for the given hybrid control strategy. Our eco- [24] Rozine V, Ockerman J. J Ship Prod 2012;28:82–6.
nomic analysis takes into account the installation and fuel costs of [25] Dedes EK, Hudson DA, Turnock SR. Energy Policy 2012;40:204–18.
[26] Del Pizzo A, Polito R, Rizzo R, Tricoli P. In: XIX international conference on
both the conventional and hybrid systems. Due to the benefits of electrical machines. p. 1–6.
hybrid power generation, as well as the gains from regenerative [27] Caterpillar genset specs, must be ordered online. last visited 08.2013, <http://
braking of the cranes, we find that hybrid power generation for www.cat.com/power-generation>; 2013.
[28] Sciarretta A, Guzzella L. Control systems. IEEE 2007;27:60–70.
crane operation is highly profitable. The hybrid system with regen- [29] Lindgren M. Engine exhaust gas emissions from non-road mobile machinery.
erative braking saves $110,000 per year from crane operations on Swedish University of Agricultural Sciences; 2004.
the ship alone. The annual savings of the hybrid solution amount [30] Mckinsey analysis. last visited 08.2013, <http://www.greencarcongress.com/
2012/07/mckinsey-20120712.html>; 2013.
to a third of the battery costs. The basic hybrid control strategy