You are on page 1of 11

Ocean Engineering 89 (2014) 189–199

Contents lists available at ScienceDirect

Ocean Engineering
journal homepage: www.elsevier.com/locate/oceaneng

Ship synthesis model for the preliminary design of a fleet


of compressed natural gas carriers
Giuliano Vernengo a,n, Enrico Rizzuto b
a
University of Genoa, Department of Naval and Marine Engineering (DITEN), Italy
b
University of Genoa, Department of Civil, Chemical and Environmental Engineering (DICCA), Italy

art ic l e i nf o a b s t r a c t

Article history: The transportation by sea of compressed natural gas is a very recent subject, prompted by recent
Received 28 August 2013 changes in the perspectives of a possible economic exploitation of relatively small quantities of this
Accepted 11 August 2014 hydrocarbon, available as by-product of oil extraction in off-shore fields. An automatic and integrated
preliminary design procedure has been established to generate and evaluate feasible technical solutions
Keywords: for a trade of this kind. A first trial application has been carried out for a specific case, for which various
Ship synthesis model feasible solutions corresponding to fleets composed of a different number of equal ships are generated.
Compressed natural gas In the procedure, a parent hull shape is adopted in order to evaluate, through systematic variations of
Preliminary design the main dimensions, all the elements necessary to perform the classical steps of the design spiral (like
Integrated design
weight evaluation, buoyancy, trim and stability checks and motion resistance prediction) accounting for
Added resistance
the mutual interaction between the various items. When possible and necessary, parts of the procedure
have been calibrated with data from existing ships, in order to ensure realism in the predictions. This has
been done with reference e.g. to the evaluation of weight items and in particular to hull weight. On the
other hand, when setting boundaries to the range of variation in the ship dimensions, care has been
given not to constrain the investigation domain only to existing dimensional ratios, but to let the
procedure explore a wider range of solutions. A particular feature of the procedure is to include in the
evaluation process the seakeeping performances in terms of added resistance in waves. This aspect
interacts with the other part of the design process and influences the final outcome in terms of
performance of the solution. The main implications of accounting for the increase in motion resistance
are the capability of evaluating the reduction in the average speed, the corresponding reduction in the
annual cargo delivery and the increase in fuel consumption (all aspects evaluated in respect to nominal
conditions). The procedure, accordingly, is able to provide, in addition to several technical output
identifying the various solutions (such as main dimensions, weight, power installed, seakeeping
performances), indicators of the performances of the fleet in terms of annual cargo delivery, CapEx
and OpEx, that can be used to rank the various solutions.
& 2014 Elsevier Ltd. All rights reserved.

1. Introduction from off-shore fields to the market, including pipelines, liquefied


natural gas (LNG), natural gas hydrate (NGH) or compressed natural
In the continuous exploitation of new and diversified sources of gas (CNG, see Bortnowska, 2009). Each engineering solution, being
energy, a reconsideration of the role of natural gas has occurred different from the other ones for complexity, working pressures and
recently. Europe has been identified among world greatest gas import temperatures, gas states and gas containment system, is designed for a
markets as a rapidly growing one (Economides and Wood, 2009). specific market sector. In particular, CNG transported by sea is
Recently interest on natural gas has grown up, even if a great portion practicable for stranded fields or for oil fields that produce as by-
of world reserves are stranded, i.e. fields without an actual market due product a small amount of gas which cannot be re-injected or flared.
to their geographical location. Very often natural gas represents a by- Another possible application is in the early stage of an LNG trade, as a
product of oil fields. A variety of ways exist to transport natural gas preliminary start-up (Thomas and Dawe, 2003).
Good compression ratios of the gas, aimed at increasing the
energy stored per unit volume, are typically achieved with pressure
n
Corresponding author. of 230–250 bars. Cylindrical gas containment systems have been
E-mail address: giuliano.vernengo@unige.it (G. Vernengo). extensively studied both with static and fatigue failure analysis (see

http://dx.doi.org/10.1016/j.oceaneng.2014.08.012
0029-8018/& 2014 Elsevier Ltd. All rights reserved.
190 G. Vernengo, E. Rizzuto / Ocean Engineering 89 (2014) 189–199

Valsgård et al., 2004, 2010). In addition to the issues of the pressure


Input Data
vessels, others key arguments of the complex definition of a fleet
of ships devoted to such a specific trade are those dealing both
with market location (Khalilpour and Karimi, 2011) and with
Free Design Variables
logistic operations of the fleet (Nikolaou, 2010), which sum to those
linked with the structural and hydrodynamic behavior of the ship

Ship Generation and Check


itself (Malta et al., 2011). Ship Main Dimension Definition
From the point of view of naval architecture, a ship synthesis
model (SSM) approach has been used (Kara, 2010), (Papanikolaou,
Ship weight and weight breakdown

Weight Correction
2010), (Harries et al., 2011): fleets of engineering-feasible ships are
designed through an automatic procedure able to integrate pre-
liminary naval architecture analysis and simple but representative Hydrodynamic predictions
logistic concerns and to account for their mutual effects in the (calm water & added resistance)
overall costs evaluation. Whenever it has been possible, data from
existing similar ships have been used to check and/or integrate the
Trim, Stability and others checks
predictions of the proposed model.

Yes
2. Integrated design strategy Others Free Variable Combinations?
No
The design of a cargo ship, even at a preliminary stage, requires

Design Rank
elaboration of a great amount of data, most of which are strongly Measure of Merit (CapEx & OpEx)
interdependent, as, for example, the case of hull weight and
resistance. This implies that it is not possible to analyze and solve
Ranking
each problem separately (e.g. weight computation, calm water
resistance prediction, seakeeping performance analysis and
others) without establishing a time consuming iterative process.
Best Fleet Design
Accordingly, an automatic design method is highly beneficial to
explore easily a great number of different design combinations, Fig. 1. Flowchart of the integrated design process.
checking always the feasibility (i.e. maintaining the internal
coherence of the design) and fulfilling the trade requirements. In
the present case, the task has been accomplished integrating the
aspect peculiar to the specific CNG trade into a broad band ship
design method. Starting from a few input information on the
loading and the unloading terminals, on the number of ships of
the fleet and on the marine environment, fleets of feasible ships
are generated by a systematic variation of selected free design
parameters. Each solution is evaluated both from a technical and
an economic point of view and at the end of the process all the
feasible fleets are ranked, on the basis of the selected Measure of
Merit, for optimization purposes. The design procedure is sche-
matically shown in Fig. 1.

2.1. Geometrical characteristics of the ships

As a first step, the main dimensions of the current ship solution


are found. Since the essential requirement is the annual nominal
cargo capacity for the fleet, hull length, beam and height are
computed following the dimensions of the zone dedicated to the
cargo storage. The general arrangement of the cargo area is shown
in Fig. 2: the transported gas is stored into equal cylindrical
pressure vessels (PVs) with spherical ends, arranged with the axis Fig. 2. Ship cargo general arrangement.
in a vertical direction. The diameter, dPV, of the transversal section
of the vessels is considered in this context as a design (fixed) possible solutions. By varying the values of the free variables, the
datum. Accordingly, once the maximum number of pressure space is explored. The ship length is divided into five parts:
vessels in the longitudinal direction (number of rows), Nx, is extreme aft; engine room; cargo zone; zone devoted to house
selected, the length of the cargo zone is computed by means of the Submerged Turret Loading (STL), if present, and fore end. The
the following formula: STL is a device often present in similar types of ships, devoted to
moor the ship in the proximity of an open water loading/unload-
LCargo ¼ N x ðdPV þ t x Þ ð1Þ
ing terminal. The presence/absence of this device is considered a
where tx is the extra space in longitudinal direction needed to design datum to be provided in input to the procedure. If present,
house the pressure vessels support skirts and piping. Similarly, the the STL implies a dedicated portion of the ship length of about
beam of this zone, BCargo , is computed from the maximum number 12 m and an extra weight of 500 t. As initial guess, once the length
of pressure vessels in the transversal direction, Ny . Nx and N y are of the cargo zone is computed, the length of the other zones is set
‘free variables’, in the sense that they identify, together with the in proportion to the former one. Checks are carried out a later
other free variables, a specific solution in the space of all the stage of the procedure.
G. Vernengo, E. Rizzuto / Ocean Engineering 89 (2014) 189–199 191

The beam of the ship is directly computed from the beam of the the ship becomes:
cargo holds accounting for the double side hull; hence:
D ¼ hDB þhPV s þ Δh ð5Þ
BShip ¼ N Y ðdPV þ t Y Þ þ 2BDS ð2Þ where Δh was set equal to 0:16hPV s and hDB is defined according to
standard requirements (see e.g.: IACS, 2012):
where t y is the extra space due to pressure vessels support skirt
and piping in transversal direction and BDS is the breath of the hDB ¼ maxðminðBShip =15; 2 mÞ; 1 mÞ ð6Þ
double side, which is calculated accordingly with the following The minimum height of the ship is also checked by a statistical
requirement (see e.g. RINa, 2005): regression based on the ship length [see paragraph 2.4.4 below], in
BDS ¼ min ½Bship =15; 3 m BDS ≥1 m ð3Þ order to avoid unrealistically low ships in comparison to length
(which might be exposed e.g. to longitudinal strength problems).
The total number of pressure vessels stored inside the ship, nPV , is In case of failure of one of these checks, the current solution,
found as: which corresponds to a particular combination of input para-
meters, is discarded.
nPV ¼ C f ill N x Ny ð4Þ At this stage of the process, the main dimensions of each ship
of the fleet are defined, as well as its nominal gas capacity, the
where C f ill is a ‘filling coefficient’, pre-determined by means of an number of pressure vessels and their height.
offline analysis: for a few geometrical affine variations of the
model hulls, pressure vessels have been placed on the inner
2.2. Weight computation
bottom plane surface and their total number have been recorded;
the filling coefficient is then defined in order to correlate the total
The weight computation is the next step of the preliminary
number of vessels that are stored in the ship with their maximum
evaluation of the fleet of ships. Weight and barycenter position
longitudinal and transversal number. In Fig. 3, the trend of such
(longitudinal and vertical) are derived in order to obtain a
coefficient vs N x and N y is shown. As it can be noted, for a given
draught computation and proceed to the prediction of the other
length, larger ships ensure a higher filling coefficient due to a
performances.
better use of the inner space. For reasons of standardization in the
construction, the height of the pressure vessels, hPV , is the same
for all the PVs within a single ship. Such height is calculated from 2.2.1. Hull steel
the nominal gas capacity of a ship of the fleet, Q Ship . The latter The estimation of the weight of the hull, i.e. the weight of steel
quantity is in turn the outcome of a previous logistic analysis plates and stiffeners, has been carried out using the area of various
which accounts for the number of ships of the fleet, for the parts of the hull surface of the model hull. Three zones have been
nominal velocity, for the annual amount of cargo to be transported defined in height (bottom, side and deck) and four zones along the
and for the length of the route (see Giribone et al., 2013). length (aft, engine room, cargo and forward part); in addition, the
Checks on both the maximum and minimum height of the area of the double bottom and of the double sides have been
pressure vessels are performed, too. The upper limit is based on a considered. For all geometrical variations of the ship dimensions,
limitation of the maximum volume of gas carried in a transversal the extension of such areas have been recorded together with the
row: this requirement is enforced for safety reasons (if a problem positions of their geometrical centers. Weight coefficients for unit
of leaking arise, a row can be isolated from the plant, but the gas area have been determined for a number (17) of existing ships,
capacity of the row should not exceed the quantity that can be whose data have been made available from an anonymous con-
quickly discharged into air with acceptable risk). Such volume was fidential source. These ships show structural characteristics con-
in the present analysis set to 93,000 m3, based on a preliminary sistent with those under study, but do not transport liquid cargo in
evaluation of the evacuation plant. pressure vessels. Dimensions are reported in Table 1 without
The lower limit in the pressure vessels height is due to further details.
geometrical limitations in the shape of the vessels themselves: Each of these real ships have been divided into the same parts
solutions with pressure vessels lower in height than a diameter above considered and an average value of the ratio between steel
and a half (3 m, in the present analysis) are discarded (note that a weight and surface has been calculated from drawings and other
height of one diameter would correspond to a spherical container).
Defining hDB the height of the double bottom and Δh a Table 1
Global dimensions of the set of existing ships.
tolerance due to the presence of piping, the evaluated height of
LBP B D
1.00 (m) (m) (m)

0.95 63 13 7
81 14 7
Filling Coefficient

0.90 104 18 9
110 18 9
0.85 Ny = 2 137 17 7
Ny = 5 140 22 13
0.80 144 26 12
Ny = 7
156 35 25
Ny = 9 168 40 21
0.75 Ny = 12 174 32 19
176 32 18
0.70
184 32 17
219 32 20
0.65 219 32 21
10 20 30 40 50 60 70 80 90
237 42 21
Nx 239 44 21
264 48 23
Fig. 3. Filling coefficient for model hull affine variations.
192 G. Vernengo, E. Rizzuto / Ocean Engineering 89 (2014) 189–199

Bottom uniform reinforcements distribution, the position of the barycenter


0.50 of the hull may be approximated with the geometrical center of the
0.45 surface.
Weight Coeff. [t/m2]

0.40 The results in terms of lightship weight (LSW) obtained by


applying the mentioned procedure based on surfaces analysis to
0.35
AFT. the ships available for the calibration are presented in Figs. 7–9,
0.30
E.R. where the prediction is plotted vs. the actual values related to the
0.25 sample of existing ships. The small dispersions around the bisector
CARGO
0.20
FORW
lines indicate a good correlation between the two sets of values
0.15 and confirm the applicability of the approach.
0.10
0 50 100 150 200 250 300 350
Ship length [m]
Fig. 4. Weight coefficient for the bottom of the ship. 30000

25000
Side

LSW (Predicted) [t]


0.50 20000
0.45
Weight Coeff. [t/m2]

15000
0.40
0.35 10000
0.30 AFT.
E.R. 5000
0.25
CARGO
0.20 0
FORW 0 5000 10000 15000 20000 25000 30000
0.15
LSW (Existing ships) [t]
0.10
0 50 100 150 200 250 300 350 Fig. 7. Light Ship Weight—predicted vs. real.

Ship length [m]


Fig. 5. Weight coefficient for the side of the ship. 140

120
Deck
XCOG(Predicted) [m]

0.50 100

0.45 80
Weight Coeff. [t/m2]

0.40
60
0.35
AFT. 40
0.30
E.R.
0.25 20
CARGO
0.20 FORW 0
0 20 40 60 80 100 120 140
0.15
XCOG(Existing ships) [m]
0.10
0 50 100 150 200 250 300 350 Fig. 8. Longitudinal position XCOG of the center of gravity—predicted vs. real.

Ship length [m]


Fig. 6. Weight coefficient for the deck of the ship. 16

14
design documents. The trend of these weight per unit area coefficients
VCOG (Predicted) [m]

12
are shown in Figs. 4–6 vs the ship length: each plot corresponds to one
of the three vertical ship parts, while each curve in the figures refers to 10
a specific longitudinal zone. The ship length ranges from 50 m to more 8
than 300 m. As expected, the weight coefficients, at constant ship
length, are higher for the bottom zone and decrease for zones at a 6

higher location and, for the same vertical zone, they increase with ship 4
length (larger scantlings). Said C Eqi the equivalent weight coefficient of
2
the ith hull zone and Ai its area extent, the weight of the ship steel is
then computed by means of the following summation: 0
0 2 4 6 8 10 12 14 16
n1Zones
W Hull  Steel ¼ ∑ C Eqi Ai ð7Þ VCOG (Existing ships) [m]
i¼1
Fig. 9. Vertical position VCOG of the center of gravity from baseline—predicted
Under the simplifying hypothesis of homogeneous material and vs. real.
G. Vernengo, E. Rizzuto / Ocean Engineering 89 (2014) 189–199 193

2.2.2. Weight of cargo and pressure vessels The above formulation comes from regressions obtained on a set
The weight of the transported gas is directly computed from the of ships for which information were available. Uncertainties do
net gas capacity of each ship of the fleet under consideration. Such affect the prediction, but, considering the pressure vessels as a
capacity comes from logistic considerations involving the distance to separate item (hence excluding their weight from the light ship),
be covered in the round trip, the nominal velocity of the ship and the the weight of the outfitting results to be about 4–6% of the light
annual gas production of the offshore field. The number of available ship weight, while the superstructures weight is bounded in the
cruising days over a year, T Cruise , is computed as: range 5% to 10% of the light ship. As mentioned, the weight of the
steel pressure vessels is the major part of the overall weight: on
T Cruise ¼ 365  T Load  T Of f Load  T Down  T In=Out ð8Þ
average, it results to be 120–150% of the light ship, depending on
where T Load and T Of f Load are the days required for loading and the combination of the free design parameters. Approximations on
offloading operation respectively, T Down is the annual downtime due the estimations of outfitting and superstructure weights, there-
to maintenance and T In=Out represents the time spent to approach the fore, result to affect the total weight of the ship to a very small
service docks, which depend on the harbors characteristics. The days extent.
needed for each ship to travel a single round trip, T RT , are computed Once the maximum number of PVs which can be longitudinally
by dividing the distance to be covered during a round trip, dRT , by the placed inside a single compartment is determined, the number of
nominal speed of each ship, V s . Hence, the number of trips traveled by transversal bulkheads is found. The weight of transversal bulk-
each ship over a year, ntrips , can be found as the ratio between T Cruise heads W Bulkheads , comes from an analysis conducted on a set of the
and T RT . Once the gas capacity due to each ship over a year, Q ShipYear , is previously mentioned existing ships; it has been correlated with
computed, dividing the yearly gas field production, Q YFP , by the the dimensions of the ships via the product of the design height, D,
number of ships, ns , the nominal net gas capacity for each ship of a by the beam of the ship, BShip , as shown in Fig. 10.
fleet, Q ShipNet , is calculated as:
Q ShipNet ¼ Q ShipYear =ntrips ð9Þ 2.3. Hydrodynamic predictions and power computation

The nominal net gas capacity of each ship is computed including an The calm water resistance of each considered solution is
extra volume, corresponding to the portion of gas which remains predicted by means of the Holtrop–Mennen statistical method
inside the pressure vessels after the unloading operations (this (Holtrop and Mennen, 1982) which is widely used in the pre-
amount, therefore, is always present on board but not actually liminary design stage of ships. Although this method does not
delivered). require a large computational time, considering that many fleet of
One of the major contributions to the total displacement of the ships need to be evaluated in the process, this computations have
ship is represented by the steel pressure vessels. The thickness of the been performed off-line on several affine variation of the model
PVs is computed by means of classical formulations (ASME, 2010), as hull and results have been stored into a database structure. The
a function of the design pressure of the vessel, of the geometrical database is loaded once for all at the beginning of the process and
characteristics and the permissible stress for the material (ASTM/ interpolations are used to find the actual value of calm water
ASME A/SA 533 Type C steel is used). The volume of the walls of the resistance for each current case.
pressure vessels and, accordingly, their weight is easily determined. A Calm water propulsion power is used for the selection of the
support skirt realized in steel has been considered, too, in the weight power installed, in turn used to check the preliminary values of
for each pressure vessel. In addition, two further weight items linked the weight of both the main engine and of the inherent outfitting,
to PVs are considered: the transversal piping connecting the PVs of by use of suitable statistical regressions.
each row and the longitudinal main piping, which connects all the The calm water propulsion power is increased by a sea margin
transversal rows. The first item represents an extra weight for each percentage (see Parsons, 2003), Ms, (equal to 20% in the present
PV, in a range of 1–2.5% of the single PV weight itself, depending on study), which is to be regarded as a design parameter since it
the number of pressure vessels on a row; the second is an extra affects the performances of the examined solution, governing the
weight for each row, equal about to 0.5% to 2% of the total weight of performances in rough sea, but influencing also weight and
PVs in a ship, based on the number of pressure vessels along the consumption items. Hence the installed power becomes:
length of the hull. In the present analysis, the weight of a single PV
ranges from 50 t to 120 t, depending on the height (values adopted P install ¼ P calm ð1 þMsÞ ð12Þ
from about 10–25 m). Indeed, a key issue for the comparison of the different solutions
for a specific trade is represented by the behavior in the actual sea
2.2.3. Outfitting, superstructures and bulkheads weights conditions characteristic of the trade area, which are statistically
The weight of the hull outfitting, except those related to the represented by specific scatter diagrams. The average speed that
pressure vessels, is computed in function of beam and length of the ship is actually able to keep is evaluated by long term predictions
the portion of the ship under investigation. Named these local
beam and length BZone and LZone respectively, the following 0.16
equation was applied: 0.15
W Outf itting ¼ C Zone BZone LZone ð10Þ
W/(B*D) [t/m2]

0.14
CZone is a scalar coefficient, in ½tons=m2 , whose value depends on
0.13
the location of the zone considered (aft, cargo and fore zone).
The weight of the superstructures is estimated via another 0.12
simplified formula, based on the beam of the ship (see Eq. 11);
C Superstructure is again a scalar coefficient, in ½tons=m2 . The vertical 0.11
position of the center of gravity of this weight item is set
0.10
accounting for the number of levels in the superstructure and 0 200 400 600 800 1000
the inter-deck height. B*D [m2]
W Superstructures ¼ C Superstructure BShip 2 ð11Þ Fig. 10. Function used for the computation of the bulkheads weight.
194 G. Vernengo, E. Rizzuto / Ocean Engineering 89 (2014) 189–199

based on such scatter diagrams. The basis of this evaluation is 2.4.1. Trim
represented by the computation of the added resistance in waves, An even keel longitudinal condition for the ship is required; in
which has been investigated in the frequency domain by means of a case of failure of such requirement, the trim is corrected by means
strip theory based method (Bertram et al., 2006) for each selected of iteratively loaded weights on the baseline of the ship, in way of
wave length (in the typical range of sea waves, from very short to the aft or forward perpendicular (depending on the sign of the
very long waves). A sort of response amplitude operator (pseudo- trim). This type of trim correction is quite coarse, but the alter-
RAO, the term referring to the fact that response is quadratic with native strategy of a modification in the hull form implies a degree
wave amplitude) for the added resistance has been obtained for each of detail in the model that is beyond the scope of the present
solution over the set of wave frequencies (Gironi et al., 2012). Hence, study, whose aim is to evaluate the influence of the main
for a given hull geometry, an average added resistance is quantified dimensions on a series of parent ships.
for each short term period, identified by a single combination of
significant wave and average wave period for the sea state and by a
2.4.2. Stability
ship speed. The spectral method is applied to evaluate this short term
In order to ensure a minimum intact stability for the ship, if the
added resistance experienced by a given hull shape advancing in bow
metacentric height, GM T , results to be too less than a predeter-
waves. Accounting for a Bretschneider sea spectra, SW, for each of
mined value (in the present case:3 m)a new correction weight is
the different values of the average wave period, TM, and for unit
loaded on the base line of the ship, in a longitudinal position
significant wave height, H1/3, the added resistance response spectra,
which corresponds to the center of gravity (in order not to modify
SRAdd, is obtained by the following equation:
further the trim).
SRAdd ðL; B; T; V ship ; T m ; H 1=3 ; ωÞ ¼ SW ðT m ; H 1=3 ; ωÞ USAdd ðL; B; T; V ship ; ωÞ After both checks, due to the possible increases in weight
ð13Þ generated by the corrections, the hydrodynamic performances are
computed once again.
where the ship added resistance pseudo-RAO, SAdd is function of the
frequency and the parameters which identify the ship dimensions
and its operating conditions (the ship nominal speed). The value of 2.4.3. Dimensional ratios
the added resistance of the candidate hull is then computed by Since realistic ship designs are to be provided at the end of the
integration in the frequency domain, first for sea states of unit process, a few constraints dealing with characteristic geometric
significant wave height: ratios of the ship have been imposed.
Z Accordingly, the main dimension ratios found so far for the trial
1
R0Add ðL; B; T; V ship ; T m ; H1=3 ¼ 1; ωÞ ¼ SRAdd ðL; B; T; V ship ; T m ; H 1=3 ¼ 1; ωÞdω ð14Þ design under consideration are checked against predefined limit
0
values; in Table 2 limits on B=D, B=T, L=B are listed: because of the
then for all the significant wave height values: aim of exploring also unconventional ships, these boundaries are
intentionally wider than those for typical existing ships. The
RAdd ðL; B; T; V ship ; T m ; H1=3 ¼ 1; ωÞ ¼ RAdd' ðL; B; T; V ship ; T m ; H 1=3 ¼ 1; ωÞðH1=3 Þ2 ð15Þ computational time devoted to the analysis of one solution is of
Depending on the value of the computed added resistance two the order of a few seconds so enlarging the boundaries of the
scenarios are possible. When the required power exceeds the analysis has not dramatic consequences on the CPU time. On the
available one, the delivered power is considered as correspondent other hand, setting these boundaries eliminates from the analysis
to the installed power (100% of the Maximum Continuous Rating: the most unrealistic cases, saving anyway computational time. In
MCR of the engine) and a decrease in the speed is hypothesized on order to check the feasibility of the selected ranges for these ratios,
the basis of a quadratic relationship between resistance and speed. a set of existing bulk carriers (which ratios have been considered
8 suitable also for the actual kind of ships) has been accounted for:
< P required_tot ¼ ðRcalm ð1 þ MsÞÞV ship the values of the B=T and the L=B ratios of the ships of the sample,
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ð16Þ
: V~ ship ¼ RRcalm ð1þþRMsÞV ship together with the corresponding limit lines, are shown in Fig. 11
calm Add
and in Fig. 12 respectively. In the same figures, the limiting values
chosen for the analysis are also indicated by means of a black
On the contrary, if the total required power is less than the installed
dashed line. As it can be noted, all the ships of the sample are
power, the nominal speed is maintained and the total computed
comprised in the center of the selected field of investigation.
propulsive power is taken into account for the consumption evalua-
tion:
( 2.4.4. Minimum ship depth
P required_tot ¼ ðRcalm þ RAdd Þ U V ship
ð17Þ The following formula defines the minimum value for the
V~ ship¼V ship height of the ship D:

The above speeds and corresponding consumptions (each one Dmin ¼ 0:09LShip  2:6 ð18Þ
obtained for a specific sea state) are unconditioned by the probability
Once again, the formulation of Eq. (18) is derived by the analysis of
of occurrence of the sea states coming from the scatter diagram. The
an available set of existing ships (see Fig. 13). As it can be seen, in
annual average fuel consumption (which account for the increased
this case a conservative approach has been followed, imposing a
wave resistance), the annual average ship speed (decreased in
minimum value for the D/L ratio that not all the existing ships
respect to the nominal value for the same reason) and the inherent
satisfy. This is due to the fact that low values of D/L may imply
average annual quantity of delivered cargo (decreased with respect
to its nominal value due to the reduction in speed) represent results
Table 2
of this computation. Limits on main dimension ratios.

Lower limit Upper limit


2.4. Trim, intact stability and geometrical checks
B/D – 1.5
B/T 2 5
Once the ships of the fleet have been designed, simple checks L/B 2.5 10
on trim, intact stability and geometric ratios are performed.
G. Vernengo, E. Rizzuto / Ocean Engineering 89 (2014) 189–199 195

21 2.4.6. Freeboard
The difference between ship depth and draught is shown in
18 Fig. 14 for the same sample of existing ships: the poly-curve shown
in the figure is used as minimum allowed freeboard with respect
15
to the length of the ship; note that for ship length less than 100 m,
Draught [m]

12 freeboard takes almost constant value (green line), while for ship
B/T=5
lengths equal or greater than 108 m the freeboard is taken from
9 B/T=2 the “Table A” of the International Convention on Load Lines of
Actual Limit 1966, which is valid for tankers (red line); for ship lengths
6 Ship sample between 100 m and 108 m a linear interpolation has been adopted
(blue line). In conclusion, the minimum allowable freeboard is set
3
according to eq.19:
8
0 > L o 100 m
< 0:4
0 10 20 30 40 50 60 70 80 FBmin ¼ 0:105L  10:1 100 m o L o 108 m ð19Þ
>
:  2e  8 L3  2e  5 L2 þ 0:0283L  1:3456
Beam [m] L Z 108 m

Fig. 11. B/T boundaries.

2.4.7. Roll period


The last check deals with the roll period of the ship; an
80
evaluation of the roll period Troll for all ships is performed
70 according to the approximate formulation of Eq. (20), where the
radius of gyration K is derived as K ¼ 0:35Bship .
60
2πK
50 T Roll ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffi ð20Þ
Beam [m]

gGM T
L/B=2.5
40
L/B=10 A minimum value can be given as input parameter by the designer.
30 Actual Limit In this case, a failure in fulfilling the condition does not imply to
Ship sample discard the solution, but such failure is highlighted in results.
20

10
2.5. Evaluation of CapEx and OpEx
0
0 50 100 150 200 250 300 350 Since the aim of the present work is to understand the
Length [m] economic as well as technical feasibility of a specific sea trade
Fig. 12. L/B boundaries. comparing different possible engineering solutions, a measure of
merit which accounts for the economic value of the proposed
designs is needed. Two main cost items are taken into account: the
Capital Expenditure (CapEx) and the Operational Expenditure
30
(OpEx). The former item accounts for the design and construction
costs and in general for the initial investments, while all others
25
costs necessary to maintain and run the fleet during the operating
life, such as fuel and crew costs, are included in the latter term.
20
At this early stage of the design process, CapEx are computed as
Depth [m]

the sum of the costs of: hull (C Hull ), machinery (C Machinery ), pressure
15
D Min. vessels (C PV s , including dedicated piping and outfitting) and other
Depth ship outfitting; all these items are found as a function of their
10
weight, earlier determined, except for the cost of machinery which
5
Ships Sample L <= 100 m 100 m < L < 108 m L > 108 m
10
0
9
0 50 100 150 200 250 300 350
Ship length [m] 8

7
Fig. 13. Depth and minimum depth of the ship sample.
Freeboard [m]

longitudinal strength problems, and this aspect is not explicitly 5


treated elsewhere in the procedure. 4

2.4.5. Draught 1

Boundaries have been foreseen also for the draught of the ship. 0
The maximum value is given as a parameter in input to the 0 50 100 150 200 250 300 350

procedure, which might be relevant in case of shallow water ports Ship length [m]

(not used in the results reported), while the minimum allowed Fig. 14. Freeboard of the ship sample. (For interpretation of the references to color
value was arbitrarily set to 3 m to avoid propulsive problems. in this figure, the reader is referred to the web version of this article.)
196 G. Vernengo, E. Rizzuto / Ocean Engineering 89 (2014) 189–199

is function of the propulsive power: These information, too, are not likely to be available at a preliminary
stage of the design.
CapEx ¼ C Hull þ C Machinery þ C PVs þ C outf ð21Þ
Two indicators have been selected for the performance assess-
The following polynomial formulation, based on the total weight ment. The first one is the percentage of success in the annual cargo
of the hull, W H , has been adopted to compute the cost of hull delivery: this represents the amount of the cargo which is actually
construction: delivered over a year, in percentage of the nominal value; such
percentage strongly depends on the average velocity of the ships
C Hull ¼  7  10  6 W 2H þ 1:6891W H þ 4625:6 ð22Þ
of the fleet, which is lower than the design one, due to the effect of
The outfitting CapEx has been related to the weight of the systems rough sea.
inside and outside the engine room, W M , by the formula: The other indicator used for the fleet ranking is the Required
Freight Rate (RFR) t, (see Benford, 2003), that corresponds to the
C Outf ¼  1:7  10  3 W 2M þ 6:3671W M þ 304:89 ð23Þ
tariff (per unit ton of transported cargo) and that balances income
The machinery cost has been computed based on the installed and costs, producing a zero net profit. The computation is carried
power, P B , through the exponential formula: out over a selected period of 20 years, neglecting, however, the
interest rate effect (IRR is thus set equal to zero). As the yearly field
C Machinery ¼ 1:7808P 0:8984
B ð24Þ production (QYFP) of gas is known, and the CapEx and OpEx are
An algorithm is also implemented, accounting for a cost reduction computed, the tariff is calculated by means of the following
of 15% on all the items for the second ship in respect to the first simplified formula:
one of the fleet and of further 5% reduction for all the further sister n
CapEx þ∑20
n ¼ 1 ð1 þIRRÞ  OpEx
ships in respect to the second one. t¼ ð28Þ
20  Q YFP
As a default value, the same cost of the steel for the hull (Eq. 22)
has been used also for the cost of the steel pressure vessels, even
The simplification is justified by the fact that the final purpose of
though specific values can be input in the procedure. Further
the fleet generation process is not to predict the actual breakeven
developments of the model will include a refined cost of the
tariff required but to make relative comparisons among feasible
pressure vessels construction, accounting separately for the raw
design options.
material and the manufacturing cost and considering other mate-
According to the above, the fleet which is characterized by the
rials, in addition to steel.
lowest possible tariff, t, and ensures the highest possible percen-
As regards the contributions to the OpEx, they should be
tage of success in the annual cargo delivery may be considered as
divided into two main categories: Running costs, paid by the
the best one for the trade. These objectives are somehow conflict-
owner of the ship even if the ship is not trading, and the Operating
ing requirements (a ship which is reliable in delivery needs to have
costs in a strict sense, associated with the sailing ship, such as
a higher installed propulsion power and to use it more, thus
bunker or agency fees.
increasing CapEx and OpEx with respect to a less reliable solution).
The OpEx of the first operating year accounts for the consum-
According to the multi-objective optimization theory, the best
ables cost, Ocons (based on the average fuel consumption computed
solutions will lay on a Pareto boundary, whose shape is deter-
in the seakeeping calculation), for the maintenance cost, Omaint, for
mined by the relative weight of these two objectives. The single
the crew cost, Ocrew, and for other costs, Ooth, including insurance.
optimal solution, belonging to the Pareto boundary, can be
The functional relations used for crew and maintenance costs are
determined on the basis of these weights.
respectively expressed by the following equations:

Ocrew ¼ 236:34ðLBDÞ0:1347 ð25Þ

Oma int ¼ 0:015CapEx ð26Þ 4. Compressed natural gas (CNG) fleet design

The total OpEx are provided by the following sum: The integrated design procedure has been applied to a specific
test case. A CNG trade between a loading and an unloading
OpEx ¼ Ocons þ Oma int þ Ocrew þ Ooth ð27Þ
terminal along a route of about 900 nautical miles in the North
All the above formulations provide a simplified estimation of the Sea, characterized by the scatter diagram taken from (IACS, 2001),
inherent costs at a very preliminary stage of the design and are has been considered.
derived by proprietary data made available by courtesy of external The assessment of a fleet of ships is a design process which
sources (in anonymous form). involves a great number of variables. The choice of which variables
are to be let free to vary (free variables) possibly within a pre-set
range of variation and which ones, on the contrary, have to be kept
3. Ranking criteria to a fixed value (parameters) needs to be made at the beginning,
since the choice affects the outcomes of the process itself.
Since the aim of the presented procedure is to support the It is clear, on the other hand, that increasing the number of free
decision making at a preliminary phase, a ranking criterion to variables, while allowing the exploration of a wider design space,
allow the comparison of the proposed feasible solutions is to be results in a higher complexity of the computation, with longer
defined. The most effective measure of merit for a fleet of ships CPU times.
devoted to a specific sea trade would be the profit, i.e. the In the present application, for example, the sea-margin (see
difference between the discounted values of income and expenses Section 2.3) will be considered as a constant value parameter
distributed over time. The income values, however, may depend (equal for all solutions). In principle, however, it should be
very much on how the contract is formulated (which is not yet regarded as a free variable, because, as mentioned, it affects the
defined at the present stage of the analysis). performance of a given solution with respect to rough sea, in
Further, the discounting computation should be carried out with terms of possibility of keeping the schedule (percentage of
the techniques of actuarial mathematics, accounting for an interest success) but also in terms of CapEx (cost of the installed engine)
rate (Benford, 2003), and (if applicable) for a tariff escalation rate. and OpEx (cost of fuel).
G. Vernengo, E. Rizzuto / Ocean Engineering 89 (2014) 189–199 197

Table 3 Table 5
Loading/unloading terminals parameters. Free design parameters.

Parameter Value Unit Design Parameter Lower Bound Upper Bound Step

Route 900 Nm Number of ships 1 3 1


Loading prod. rate 50 MMscf/day Ship nominal speed [knots] 10 24 2
Loading rate 10 MMscf/h Long. number of PVs 12 72 1
Gas loading density 0.226 t/m3 Transv. number of PVs 4 22 1
Gas loading pressure 230 bar
Unloading prod. rate 50 MMscf/day
Unloading rate 10 MMscf/h
Gas unloading density 0.021 t/m3
Gas unloading pressure 30 bar 1 Ship 2 Ships 3 Ships
800

Nominal gas capacity per ship [MMscf ]


700

Table 4 600
Ship parameters.
500
Parameter Value Unit
400
Sea margin 20 %
Min. GMT 3 m 300
Min. roll period 8 sec
200

100
4.1. Input parameters for the scenario definition
0
Part of the input data for the problem definition is represented 10 12 14 16 18 20 22 24
by the characteristics of the terminals and of the transported gas. Ship speed [knots]
The production rate, as well as the loading and unloading rates
Fig. 15. Nominal gas capacity per ship vs ship speed (parameter: number of ships).
and the information linked with the gas, such as the loading and
unloading pressures and densities, are listed in Table 3.
As earlier mentioned, some constraints about ship character-
istics are also given in input to the process. These parameters, such 1 Ship Fleet 2 Ships Fleet 3 Ships Fleet

as the engine margin, the minimum allowed GMT or the minimum 100
Success in the annual cargo delivery [%]

allowed roll period, are listed in Table 4. 95

4.2. Bounds for variations of free variables 90

85
In the present test case, the number of ships composing a fleet,
nS, their nominal velocity, VS, the maximum number of pressure 80
vessels along the hull length, NX, and the maximum number of
pressure vessels along the beam of the ship, NY, are selected as 75
design variables. The selected ranges for the free variables adopted
70
for a first application of the model are listed in Table 5, together
with the step considered for the variations. Considering the 65
chosen bounds and steps, 27,816 possible fleets were considered 5000 25000 45000 65000 85000 105000 125000
in the process. Light Ship Weight of the fleet [tons]

Fig. 16. Percentage of success in the annual cargo delivery vs LSW of the fleet.
4.3. Numerical results and comparison of feasible solutions

Once the input parameters to the problem have been chosen, transport. The volume capacity required for each single ship (and for
several combinations of number of ships, nominal velocity and the fleet) has obviously an inverse dependence on the number of
nominal capacity, QS, are determined on the basis of a preliminary ships, but, according to what is noted above on the efficiency of the
analysis (Giribone et al., 2013). transport, such dependence tends to be very strong for a small
Results of this first step are shown in Fig. 15; each curve of the plot number of ships (Fig. 15). This effect depends only on the ‘bottleneck’
represents a given number of ships, and each point of a curve represented by the loading /unloading rate at terminals.
represents the nominal gas capacity, expressed in millions of standard For each of the mentioned triplets (nS; VS; QS) identified in the
cubic feet (MMscf), for each ship of the fleet, expressed in function of preliminary analysis, different solutions are generated, evaluated
nominal speed (in knots). Fig. 15 shows the strong effects that the and checked via the described design procedure by means of a
number of ships in the fleet has on the ship dimensions. Since the systematic variation of the two remaining free variables, NX and NY.
loading and offloading rates at terminals do not depend on ships The results of the whole design process are shown in Fig. 16 in
characteristics (in this specific example trade), the percentage of time terms of technical data: the percentage of success in the annual gas
devoted by each fleet to loading/unloading operations is very different delivery, which, as mentioned, accounts for the possible reduction in
in dependence on the number of ships: for instance, in a fleet made of speed due to the adverse sea conditions, is shown against the light
a single ship, when that ship is moored at terminal, the whole fleet is ship weight of each feasible designed fleet. Then, in order to compare
not sailing, while, for multiple ships, when one is charging or solutions with relative high performances, those featuring a percen-
discharging, the other ones are sailing, raising the efficiency of the tage of success higher than 85% are selected. In order to give an
198 G. Vernengo, E. Rizzuto / Ocean Engineering 89 (2014) 189–199

1 Ship Fleet 2 Ships Fleet 3 Ships Fleet 1 Ship Feelt 2 Ships Fleet 3 Ships Fleet
60 100

Success in the annual cargo delivery [%]


99
55 98
97
50 96
95
Ship Beam [m]

45 94
93
40
92
91
35
90
30 89
88
25 87
86
20 85
100 120 140 160 180 200 220 240 260 280 300 0 10 20 30 40 50 60 70 80 90 100
Ship Length [m] OpEx [%]

Fig. 17. Length and beam of the selected ship sample. Fig. 20. OpEx fleet vs annual percentage of success.

nominal values are computed with reference to nominal speed (in


1 Ship Fleet 2 Ships Fleet 3 Ships Fleet
calm water): due to the presence of waves, the actual speed is
100
Success in the annual cargo delivery [%]

99 always lower of the nominal one and this results in an annual


98 delivery of cargo smaller than the nominal. To overcome the
97 problem, a further design variable could be introduced, corre-
96
95 sponding to an increasing factor to be applied to the annual
94 volume of gas to be transported when defining the volume
93 capacity of the fleet (and of the single ship). A different approach,
92
91 simpler from a procedural viewpoint, is to use the actual percen-
90 tage of success obtained in the first run of the process as a
89 feedback to correct the input data: the procedure may be run for
88
87
a second time with an increased nominal gas capacity; acting this
86 way, a new percentage of success will be obtained which, applied
85 to a higher nominal value, may match the previous goal in terms of
55 60 65 70 75 80 85 90 95 100
cargo delivered.
Trade Tariff [%]
As regards the stripes of points which may be noted in
Fig. 18. Trade tariff vs annual percentage of success. Figs. 18–20 for the cases of single-ship fleets, they are caused by
the ship speed. In fact, to each stripe corresponds a velocity or at
least a range of two close ones which result in fleets of quite
1 Ship Fleet 2 Ships Fleet 3 Ships Fleet similar ships in terms of main dimensions and hence of overall
100
Success in the annual cargo delivery [%]

performances.
99
98
The lower trade tariff is experienced by fleets with two ships;
97 although, those fleets are not able to deliver more than about the
96 93% of the nominal capacity per year, due to the effects of the
95
94
adverse sea conditions. On the other hand, one-ship fleets seem to
93 be able to have better performances in terms of yearly delivered
92 gas but with higher costs. For the specific case under examination,
91
fleets composed by three ships do not seem to be competitive.
90
89 This result can be interpreted in terms of CapEx and OpEx.
88 Observing Figs. 19 and 20, it can be noted that the fleets composed
87
by a single ship have the higher investments costs due to the
86
85 greater dimensions of their ships, while there are not significant
40 50 60 70 80 90 100 differences between the solutions with two and three ships. On
CapEx [%] the other hand, the operational costs feature a different trend: the
Fig. 19. CapEx fleet vs annual percentage of success.
lower values are experienced by fleets made of a single ship and
the costs increase with the number of ships.
This kind of results are the basis for the non-trivial choice of
overview of the dimensions of the fleets belonging to this subset, they the best suitable configuration in terms of number of ships,
are represented in a length-beam plane in Fig. 17: as expected, the nominal velocity and main dimensions of each ship. The final
ships derived from fleets with a different number of vessels feature decision, however, should be based on a weighting of the two
quite different dimensions. aspects (success in delivery and total cost) that could possibly
The two ranking criteria of the fleet sample, i.e. the overall depend on the formulation of the specific transportation contract.
breakeven tariff and the annual success percentage, are shown
in Fig. 18, while annual success percentage relative to CapEx and
OpEx of the fleets are shown in Fig. 19 and in Fig. 20 respectively. 5. Conclusions and discussions
As regards the values of the percentage of success, it should be
noted that none of the fleet under consideration are able to supply An automatic and integrated preliminary design procedure has
100% of the nominal cargo. This is because at the beginning the been established to generate and evaluate feasible technical
G. Vernengo, E. Rizzuto / Ocean Engineering 89 (2014) 189–199 199

solutions for a trade of CNG and a first trial application has been future applications to the development of an optimization strategy
carried out for a specific case. The procedure covers the typical able to drive automatically the investigation towards the best
aspects of ship design: selection of main hull dimensions, identi- fleet configuration, without analyzing and storing all the possible
fication of weight (steel, outfitting, pressure vessels and cargo), combinations of solutions (as presently done). This, on the other
propulsion and fuel consumption at nominal speed, trim and hand, would imply the definition of a specific multi-objective
stability checks. It includes, also, more innovative subjects, like target function and the use of an optimization algorithm such as
the prediction of added resistance and consumption in waves, the one presented in Deb et al. (2002).
which makes the final economical comparison between the
various solutions more meaningful, by allowing a trade between
solutions with better seakeeping performances, with larger CAPEX Acknowledgments
(larger ships) and ships with worse hydrodynamic performances
but lower construction costs. The basis of this work has been laid down during the devel-
Given the novelty of the technical solution examined, the opment of a project developed for Registro Italiano Navale (RINa),
investigation is carried at a pre-feasibility stage, characterized by whose financial support is gratefully acknowledged.
estimations of technical as well as economical parameters with
quite wide margins of uncertainty. In this respect, the present
application of the ship synthesis model may be less accurate than References
analogous procedures focused at exploring more defined design
spaces, but, on the other hand, it covers new aspects related to the ASME, 2010. Boiler and Pressure Vessel Code, Section VIII, Division 1: Rules for
specific transportation mode that have a considerable impact on Construction of Pressure Vessels. American Society of Mechanical Engineers,
ISBN: 9780791832516.
the final ship design. Benford, H.B., 2003. Engineering Economy (Chapter 6). In: Lamb, T. (Ed.), Ship
An approach of this type is considered suitable to deal with Design and Construction, I. SNAME, Jersey City, NJ.
technical problems that have not yet a consolidated solution and Bertram, V., Veelo B., Soding H., Graf, K., 2006. Development of a freely available
strip method for seakeeping. In: Proceedings 5th International Conference on
therefore need to investigate large design spaces (much larger
Computer Applications and Information Technology in the Maritime Industries
than those already explored by existing solutions). In the present (COMPIT’06). Oegstgeest, The Netherlands. ISBN-10, 90-810065-3-3.
case, the peculiarity of the problem stands in the combination of Bortnowska, M., 2009. Development of new technologies for shipping natural gas
the properties of the transported cargo (i.e. the compressed by sea. Pol. Marit. Res. 3 16 (61), 70–78. http://dx.doi.org/10.2478/v10012-008-
0036-2.
natural gas, with very small specific gravity) with the character- Deb, K., Pratap, A., Agarwal, S., Meyarivan, T., 2002. A fast and elitist multi-objective
istics of the containment system (i.e. large pressure vessels in genetic algorithm: N.S.G.A.-II. IEEE Trans. Evol. Comput. 6 (2), 182–197. http:
steel, with significant weight and dimensions). This brought the //dx.doi.org/10.1109/4235.996017.
Economides, M.J., Wood, D.A., 2009. The state of natural gas. J. Nat. Gas Sci. Eng.
need for exploring, in a pre-screening phase, solutions beyond the 1875-51001 (1–2), 1–13. http://dx.doi.org/10.1016/j.jngse.2009.03.005.
classical variation ranges adopted for the characteristics of Giribone, P., Revetria, R., Testa, A., Vernengo, G., Rizzuto, E., Longo, R., Lo Nigro, A.,
existing ships. 2013. A simulation based methodology for supporting CNG ship design. In:
Proceedings of the 7th International Conference on Applied Mathematics,
A trial application of the procedure to a realistic case showed the
Simulation, Modeling (ASM ‘13), January 30–February 1, Cambridge, MA, USA.
capability of the methodology of providing useful results for a Gironi, C., Bruzzone, D., Rizzuto, E., 2012. A parametric prediction of added
preliminary evaluation of the feasibility of the general technical resistance in waves. In: Proceedings of the International Conference on Ship
solution (CNG carried by ship) and for a pre-screening of the more and Shipping Research (NAV12), 17–19th October, Naples, Italy. ISBN: 979-88-
904394-4-5.
promising base design choices: number of ships, speed, dimensions. Harries, S., Tilling, F., Wilken, M., Zaraphonitis, G., 2011. An integrated approach for
Refinements in the input data at a later stage of the design (in simulation in the early ship design of a tanker. In: Proceedings of the10th
particular as regards the weight and cost coefficients) are possible, International Conference on Computer and IT Applications in the Maritime
Industry. TU Hamburg, isbn:978-3-89220-649-1.
in order to obtain more accurate results. Holtrop, J., Mennen, G.G.J., 1982. An approximate power prediction method. Int.
Future developments of the procedure will be to consider Shipbuild. Progress 29, 166–170.
different solutions for the gas containment systems including the Khalilpour, R., Karimi, I.A., 2011. Investment portfolios under uncertainty for
utilizing natural gas resources. Comput. Chem. Eng.0098-135435 (9), 1827–1837.
use of innovative materials. This will allow an evaluation of such
http://dx.doi.org/10.1016/j.compchemeng.2011.04.005.
solutions in a cost/benefit perspective. Kara, Y.M., 2010. A tool for evaluating the early stage design of corvettes (Master's
Other developments in the use of the procedure could be thesis). Massachusetts Institute of Technology, Department of Ocean Engineer-
represented by the enlargement of the design space under investiga- ing, Cambridge, MA, USA.
Malta, E.B., Nogueira, M.C., Ramos, A.A., Sampaio, C.M., 2011. Hull shape optimiza-
tion. In the present application, a space of 4 design variables has been tion of a compressed natural gas ship. In: Proceedings of 22nd Pan American
investigated: number of ships, speed and maximum number of Conference of Naval Engineering, Maritime Transportation and Ports Engineer-
pressure vessels in the longitudinal and transversal direction, with ing, XII COPINAVAL-IPIN, Buenos Aires, Argentina, September 27–30, 2011.
Nikolaou, M., 2010. Optimizing the logistics of compressed natural gas transporta-
pre-defined ranges and steps of variation. Other free variables can
tion by marine vessels. J. Nat. Gas Sci. Eng.1875-51002 (1), 1–20. http://dx.doi.
easily be added in the definition of the investigation space, in org/10.1016/j.jngse.2010.02.001.
particular: the sea-margin coefficient adopted to select the installed Parsons, M.G., 2003. Parametric Design (Chapter 11). In: Lamb, T. (Ed.), Ship Design
and Construction, I. SNAME, Jersey City, NJ.
propulsion power from the calm water power prediction and an
Papanikolaou, A., 2010. Holistic ship design optimization (November 2010).
analogous increasing coefficient that could be defined to increase the Comput.-Aided Des. 0010-448542 (11), 1028–1044. http://dx.doi.org/10.1016/j.
volume capacity of the fleet with reference to the nominal volume, cad.2009.07.002.
computed with the nominal (calm water) speed. RINa, 2005. Rules for Ship Classification, Part B: Hull and Stability,
RegistroItalianoNavale.
The upgrading of these two quantities (already present in the IACS, 2001. Recommendation No. 34. Standard Wave Data, p. 2. The International
procedure, but treated as fixed parameters) to the role of design Association of Classification Societies (See IACS website at www.iacs.org.uk).
(free) variables would allow exploration of the trade-off between IACS, 2012. Common Structural Rules for Double Hull Oil Tankers. The International
Association of Classification Societies (See IACS website at www.iacs.org.uk).
the opposite strategies of increasing ship dimensions and increas- Thomas, S., Dawe, R.A., 2003. Review of ways to transport natural gas energy from
ing propulsive power. Both strategies can in fact be used to countries which do not need the gas for domestic use. Energy0360-544228
counterbalance the decrease of cargo delivery due to the speed (14), 1461–1477. http://dx.doi.org/10.1016/S0360-5442(03)00124-5.
Valsgård, S., Reepmeyer, O., Lothe, P., Strøm, N.K., Mørk, K., 2004. The development
reduction determined by adverse sea conditions.
of a compressed natural gas carrier. In: Proceedings of the 9th International
The possible increase in the number of design variables (those Symposium on Practical Design of Ships and Other Floating Structures. PRADS
above mentioned or other ones), however, should be coupled in 2004, September 12–17, 2004, Lubeck-Travemunde, Germany.

You might also like