This article was downloaded by: [University of Waikato]

On: 14 August 2015, At: 19:23

Publisher: Routledge
Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: 5 Howick Place,
London, SW1P 1WG

Applied Economics
Publication details, including instructions for authors and subscription information:
http://www.tandfonline.com/loi/raec20

Joint optimization of strategic fleet planning and

contract analysis in tramp shipping
a b
Jørgen Laake & Abraham Zhang
a
Kristian Gerhard Jebsen Skipsrederi AS, Bergen, Norway
b
Department of Management Systems, Waikato Management School, The University of
Waikato, Hamilton, New Zealand
Published online: 14 Aug 2015.

Click for updates

To cite this article: Jørgen Laake & Abraham Zhang (2015): Joint optimization of strategic fleet planning and contract
analysis in tramp shipping, Applied Economics, DOI: 10.1080/00036846.2015.1076151

To link to this article: http://dx.doi.org/10.1080/00036846.2015.1076151

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained
in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no
representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the
Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and
are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and
should be independently verified with primary sources of information. Taylor and Francis shall not be liable for
any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever
or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of
the Content.

This article may be used for research, teaching, and private study purposes. Any substantial or systematic
reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any
form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http://
www.tandfonline.com/page/terms-and-conditions
 APPLIED ECONOMICS, 2015
http://dx.doi.org/10.1080/00036846.2015.1076151

Joint optimization of strategic fleet planning and contract analysis in tramp

shipping
Jørgen Laakea and Abraham Zhangb
a
Kristian Gerhard Jebsen Skipsrederi AS, Bergen, Norway; bDepartment of Management Systems, Waikato Management School,
The University of Waikato, Hamilton, New Zealand

ABSTRACT KEYWORDS
Maritime transportation is one of the most capital-intensive industries. Fleet planning is vital but Fleet planning; fleet size and
challenging to shipowners because the industry is extremely volatile. Relatively few papers have mix; contract analysis; tramp
studied strategic fleet planning in tramp shipping, which is intertwined with contract analysis and shipping; optimization;
different from that in industrial or liner shipping. This article develops a mixed-integer program- mixed integer programming
ming model, and it is the first of its kind that jointly optimizes strategic fleet planning and the
JEL CLASSIFICATION
selections of long-term and spot contracts in tramp shipping. The model can be used to C61; L99
Downloaded by [University of Waikato] at 19:23 14 August 2015

determine the best mix of long-term and spot contracts for a given fleet and/or to find the
optimal fleet size and mix for a set of contracts. It can be used as a basis for a fleet renewal
programme, informing decisions on when to sell, whether to buy old or new ships, and when to
charter in or out vessels. A numerical example is given to illustrate how to use the model to
evaluate different operations strategies.

I. Introduction profit. Liner shipping operates in accordance with

pre-published itineraries and schedules. It is similar
Maritime transportation is one of the most capital-
to a bus service (Fagerholt et al. 2010).
intensive industries. A ship usually costs millions of
Strategic fleet planning in tramp shipping is inter-
US dollars and has a lifetime of approximately
twined with contract analysis and different from that
20–25 years. Strategic fleet planning deals with ship
in industrial or liner shipping. Contract analysis is
acquisition, sale, chartering and scrapping. It is vital
unique to tramp shipping as shipowners need to
for shipowners and has long-term implications for
analyse the best mix of long-term and spot contracts
the profitability of a shipping business. It is challen-
for a given fleet and/or find the optimal fleet size and
ging because freight markets and ship prices are
mix for a set of contracts. Although many papers
highly volatile due to the cyclical nature of the ship-
(Jaramillo and Perakis 1991; Perakis and Jaramillo
ping industry (Stopford 2009). Great skill and also
1991; Cho and Perakis 1996; Powell and Perkins
luck are needed to choose the right time to order
1997; Imai and Rivera 2001; Fagerholt, Johnsen,
new or second-hand ships, charter ships in or out,
and Lindstad 2009; Meng and Wang 2010, 2011)
and scrap old ones. To succeed in the long run, it is
have studied strategic fleet planning in liner ship-
essential for shipping companies to employ sound
ping, very few works have been undertaken in tramp
tools for strategic fleet planning.
shipping (Fagerholt and Lindstad 2000; Xie, Wang,
There are three modes of shipping operations:
and Chen 2000; Fagerholt et al. 2010; Alvarez et al.
industrial, tramp and liner shipping (Lawrence
2011). Among these works, unfortunately, none of
1972). Industrial operators own the cargo and try
them has considered both long-term and spot con-
to minimize the cost of transporting it between
tracts in conjunction with strategic fleet planning.
ports. Tramp shipping can be compared to a taxi
To narrow the research gap, this article develops a
service, as the ships follow available cargo. A tramp
new model for joint optimization of strategic fleet
shipping company often has some long-term con-
planning and contract analysis in tramp shipping.
tracts and takes on optional spot cargo to maximize
The optimization model informs the selection of both

CONTACT: Abraham Zhang abraham.zhang@gmail.com

© 2015 Taylor & Francis
 2 J. LAAKE AND A. ZHANG
long-term and spot contracts over multiple planning
periods and maximizes total profit. It suggests when to
buy and sell new or second-hand vessels and when to
charter them in or out. It is also flexible enough to be
used to evaluate whether it is more profitable to termi-
nate freight operations and engage in vessel trading
only in a given set of market conditions.
The article is organized as follows: Section II
defines the research problem in detail, Section III
presents the optimization model, Section IV gives a
numerical example, Section V presents results and
analysis and Section VI concludes the research.
II. Problem description
Three types of contracts or charter-parties are most
Downloaded by [University of Waikato] at 19:23 14 August 2015
relevant to strategic fleet planning in tramp ship-
ping. They are (1) spot or voyage charter, (2) con-
tract of affreightment (COA) and (3) time charter
(TC) (Stopford 2009). In a spot contract, a ship-
owner is paid a freight rate for every unit of cargo
he transports from one point to another. In a COA,
a shipowner agrees to transport a specified amount
of cargo belonging to one or many owners for a
fixed price per tonne. The terms of a COA usually
span a number of years. In a TC, a charterer pays a
shipowner a fixed daily or monthly amount. The
charterer pays for voyage-related costs such as
those for fuel, harbour, canal fees and cargo hand-
ling. The shipowner pays the operational expendi-
tures (OPEX) and takes the operational risk, that is,
he has to pay if the ship breaks down.
COA is the most common long-term contract type
in tramp shipping (Fagerholt et al. 2010). It is a rela-
tively stable income source for shipowners because its
freight rate is fixed during the contract period. As
mentioned previously, tramp shipping uniquely
requires contract analysis for strategic fleet planning.
Decisions about fleets are closely interwoven with
choice of contracts. Their interplay allows two different
approaches to strategic fleet planning. Either the COAs
to which a shipowner has committed himself deter-
mine the size of the fleet or the fleet size determines the
COAs into which he can enter.
Strategic fleet planning in tramp shipping is com-
plicated for several reasons. First, more often than
not, a fleet is heterogeneous, that is, it consists of
several different vessel types and sizes. Cargo capa-
city, speed and operational cost vary with vessel type
and size. It is often a challenge to identify the
optimal mix of vessel types and sizes. It is compli-
cated by the fact that some routes, such as those
through the Suez and Panama canals, have size con-
straints, and some vessel types are thus more suitable
than others on specific routes.
Second, shipowners need to weigh up the risk and
cost associated with buying/chartering a vessel.
Chartering avoids the risk of buying a ship, the prices
of which are unpredictable and highly volatile. However,
chartering may be more costly over the long term (Meng
and Wang 2011). A shipowner has to optimize the
timing of ship acquisition and/or time chartering.
Third, there are important trade-offs to evaluate
when choosing whether to buy a second-hand vessel
or a new one. The advantage of buying a second-
hand vessel is that it is ready to be put into service
fairly quickly in comparison to a newly built one,
which usually takes a couple of years from the order
is placed to the vessel is finished. The disadvantage is
that older vessels can be less fuel efficient and cost
more to operate (Stopford 2009).
Last but not least, strategic fleet planning is not just
about choosing when to increase fleet capacity, but also
about choosing when to reduce it. This can be done by
either selling or scrapping vessels. The best time for
this depends not only on the rates at the time but also
on the duration of any current COAs. A shipowner will
typically go for as many COAs as possible if he thinks
there will be low spot rates and the opposite if he
believes spot rates will be high. How much of the
fleet capacity that should be covered by COAs and
how much that should be open for spot cargoes is
consequently an important strategic decision.
This can all be distilled to the following basic
questions of the strategic fleet planning and contract
analysis problem in tramp shipping, questions of
● Which COA contracts and how many spot
cargoes to take on
● When to buy/sell new or old ships of various
types/sizes
● When to charter in/out ships of various types/
sizes
Note that it is seldom that a whole new fleet must be
determined. Often, adjustments to an existing fleet
are sufficient. Needs for adjustment can arise
because vessels have to be sold or scrapped or
because new COAs have been taken on. To simplify
the problem, we consider scrapping a ship the same
as selling it because what matters to the shipowner
in both cases is the salvage value.
 APPLIED ECONOMICS 3

III. An optimization model Table 2. Parameters.

Parameter Description Unit
Model assumptions and notations Tvr Time for vessel type v to complete one roundtrip on route r Days
TvtTOT Total available time for vessel type v in period t Days
In this section, an optimization model is developed TvLT Maximum lifetime of vessel type v Periods
Qit Demand of COA i in period t Tonne
to help answer the questions presented above for the Qv Capacity of vessel type v Tonne
Sit Upper limit of demand for spot contract i in period t Tonne
strategic fleet planning and contract analysis pro- Aivr Binary parameter equal to 1 if it is feasible to assign contract –
blem in tramp shipping. The model assumes that i to vessel type v on route r
Fv0 Number of owned vessels of type v in the planning period –
the transport demand is sufficiently large on each RCOA
i
Revenue of servicing COA i USD
route. Each vessel takes full loads and does not mix RSPOT
it
Revenue per unit transported on spot trade i in period t USD/
tonne
cargo from different contracts (Xie, Wang, and Chen RSvt Revenue of selling vessel type v in period t USD
RTC Revenue of time chartering out vessel type v in period t USD
2000), which is standard practice in the coal/iron ore vt
Cvr Voyage cost of sailing route r with vessel type v USD
trade. The arrangement of backhaul cargoes, if any, CvtI Cost of buying vessel type v in period t USD
CvtN Cost of ordering new building of type v in period t USD
is predefined in the routes. All decisions are made at CvtTC Cost of time chartering vessel type v in period t USD
the start of each planning period and TC can only be CvtO OPEX for a vessel of type v in period t USD
CvtC Capital expenditures (CAPEX) for a vessel of type v in period t USD
arranged for one period at a time. All costs and
profits are in constant dollars.
Downloaded by [University of Waikato] at 19:23 14 August 2015

Before the presentation of the formulation of the

Table 3. Decision variables.
model, we define notations in Tables 1–3.
Decision
variable Description
yvtTOT Total number of vessels of type v operated in period t
Model formulation yvtOWN Number of vessels of type v owned in period t
yvtTCin Number of vessels of type v that are time chartered in
Objective function period t
yvtTCout Number of vessels of type v that are time chartered out in
X X X period t
max RCOA i δi þ RSPOT
it
zit yvtI Number of vessels of type v acquired in period t
COA SPOT t2T yvtN Number of new buildings of type v ordered in period t
i2N i2N yvtS Number of vessels of type v that are sold in period t
XX XX δi A binary variable that is 1 if COA i is selected and 0
þ TC TCout
Rvt yvt þ RSvt ySvt otherwise
v2T t2T v2T t2T xvrt Number of roundtrips made by vessel type v on route r in
XX XX period t
 CT TCin
Cvt yvt  I I
Cvt yvt zit Quantity (in tonnes) transported on spot trade i in period t
v2T t2T v2T t2T
XX XX
 CvtN yNvt  Cvt xvt
v2T t2T v2T t2T The objective function equation (1) maximizes the
XX 
 CvtO þ CvtC yOWN
vt
profit. The first term calculates the revenue from
v2T t2T servicing COAs and spot contracts. For available
(1) COAs, it considers their revenues (in USD) and
whether they will be serviced. For spot contracts,
the revenue is derived from multiplying the revenue
Table 1. Sets and indices.
Set Index Description
per unit (in USD/tonne) and the quantity (in
  tonnes) of spot cargo that will be serviced. The
T t Time periods, T ¼ 0; 1; 2; . . . ; T MAX , where period 0
corresponds to the time period when the planning is second term computes the revenue from time-char-
done and T MAX is the number of time periods considered
NCOA i Available contracts of affreightment tering out and selling vessels, based on the transac-
NSPOT i Available spot cargo contracts tion values (in USD) and the number of vessels.
V v Available vessel types. Each type v also includes
information of in which period a vessel was acquired, tv , Similarly, the third term gives the costs of chartering
and its maximum lifetime, TvLT . Two vessels acquired in and buying vessels. The fourth term calculates the
different years, but similar in all other ways will then have
different indices costs of ordering new vessels based on their prices
Tv tv Time periods
 for vessel
 type  (in USD) and order quantities, plus voyage costs (in
v; Tv ¼ tv ; . . . min tv þ TvLT ; T MAX , where TvLT is the
number of time periods that corresponds to the vessel’s USD) for the different routes and vessels. Finally, the
lifetime fifth term defines the OPEX (in USD) and CAPEX
Rv r Available sailing routes for vessel type v
(in USD) of owned vessels.
 4 J. LAAKE AND A. ZHANG

Fleet conservation constraints Other constraints

Constraint (2) ensures that the total number of Constraints (12)–(17) define the characteristics of
owned vessels is preserved. Constraint (3) ensures variables. It should be mentioned that constraint
the same for the total number of each vessel type (14) imposes integrality requirements and thereby
controlled and operated by the shipping company. only allows chartering in or out for whole periods.
Constraint (4) specifies that one does not charter out Constraint (16) does not impose integrality require-
vessels that one does not own. Constraint (5) speci- ments, making it possible to take into account a
fies that vessels are sold or scrapped before their roundtrip that continues from one period to another.
lifetimes expire.
δi 2 f0; 1g " i 2 N COA (12)
yOWN
vt ¼ yOWN
v;t1 þ yvt  yvt þ yvt
I S N
" v 2 V t 2 Tv
yOWN
vt ; ySvt  0 and integer; " v 2 V t 2 T
(2)
(13)
yTOT ¼ yOWN þ yTCin  yTCout " v 2 V t 2 Tv
vt ; yvt ; yvt  0 and integer
vt vt vt vt
yTOT TCout TCin
(3) (14)
" v 2 V t 2 Tv
Downloaded by [University of Waikato] at 19:23 14 August 2015

yOWN
vt  yTCout
vt " v 2 V t 2 Tv (4)
yIvtv ; yNvtv  0 and integer " v 2 V t 2 Tv (15)
þTvLT
tvX
yOWN
v;tv ¼ ySvt " v2V (5) xvrt  0 " v 2 V r 2 Rv t 2 Tv (16)
tv þ1

Constraints (6) and (7) define the initial fleet. zit  0 " i 2 N SPOT t 2 T (17)
Constraint (8) ensures that no newly built ships
can be acquired in the first period; this is to account
A feature of the model worth mentioning is the
for the lead time required for the order.
artificial period. Because of the definition of Tv (time
yOWN ¼ Fv0 " v2V (6) periods for vessel type v), the model forces vessels to
v;0
be sold before their lifetime ends or at the end of the
planning period, whichever point is reached in the
v;0 ¼ yv;0
yTOT þ yTCin
v;0  yv;0 " v2V
OWN TCout
(7)
fewest periods. However, this becomes a problem as
yNv;1 ¼ 0 " v 2 V (8) soon as the number of periods in the planning
interval is less than the number of periods left in a
vessel’s lifetime. A company that plans for, for exam-
Demand and capacity constraints ple, 5 years at a time should not necessarily sell a
Constraint (9) ensures that the fleet has sufficient new vessel after 5 years. It is for this reason that an
capacity to fulfil the demands of the COAs and spot artificial period is created and added to the original
cargo volumes selected for it. Constraint (10) sets an set of periods. In this artificial period, it is only
upper limit of cargo available for each spot trade. possible to sell vessels. This is a period during
Constraint (11) ensures that the total duration of all which it could be equally beneficial to keep vessels
roundtrips made on a route by each vessel type is in a fleet even if the model suggests selling them.
equal to or less than the total available time. The optimization model is a mixed-integer pro-
XX gramming model. It can be solved efficiently by any
Qv Aivr xvrt  Qit δi þ zit commercial optimization solver. We implemented
v2V r2Rv (9) the model using the optimization software Xpress
" i2N SPOT
[N COA
t2T on a laptop computer.

Sit  zit " i 2 N SPOT t2T (10)

IV. A numerical example
X
TvtTOT yTOT
vt  Tvr xvrt " v 2 V t 2 Tv (11) In this section, some areas of application for the
r2Rv optimization model are demonstrated. A realistic
 APPLIED ECONOMICS 5

but simplified scenario is posited in which two Case 2

bulk shipping operators have just merged into a
new company named WBC. The combined fleet
consists of 5 Supramax, 10 Panamax and 5
Capesize vessels. The management wants to inves-
tigate the profitability of different operations stra-
tegies and is interested in finding out if the
existing merged fleet is sufficient for the new
company or if they should consider replacing ves-
sel types and/or acquiring additional vessels.
Another option under evaluation is to terminate
all cargo transportation activity and focus on trad-
ing vessels and TC vessels.
Four long-term COAs and additional spot
trade on three routes are available. Detailed
information on these contracts is given in Table 4.
Downloaded by [University of Waikato] at 19:23 14 August 2015
WBC’s fleet is optimized so as to handle all the
proposed COAs. The fleet mix and size is deter-
mined based on maximal profit. The options of
buying second-hand vessels and newly built vessels
and of using TC are all to be considered together.
Case 3
WBC terminates all transporting activity and con-
tinues only with active vessel trading and TC.
Information about the vessels was collected from
Fairplay (2009). Where the specifications found for
the vessels were old, fuel consumption was updated
to fit 2013 specifications instead. A modification was
made of adjusting the sailing speed to 11 knots for
all vessels types, which is called slow steaming. This

The value of a COA is based on the whole contract;
that is, demand for all periods must be met in order
to commit to it. The amount to be transported is
given in tonnes and specified for each period for the
COAs. For the spot contracts, Table 4 gives the
revenue per tonne transported, and the upper limit
of cargo volume.
The management wants to evaluate the following
three different cases.
Case 1
WBC continues to operate in the coal/iron ore mar-
ket without making any changes to the combined
fleet. If possible, all COAs are serviced. Otherwise,
the best COAs are chosen based on a maximal profit
evaluation. The possibilities of doing this with and
without using TC should be investigated.
has become more common in today’s market since
fuel prices are now much higher than those in the
1980s and 1990s while rates continue to stay low.
The relevant vessel types are Supramax (55 000
deadweight tonnage (DWT)), Panamax (75 000
DWT) and two types of Capesize (150 000 DWT
and 170 000 DWT, respectively).
The planning interval is of five periods (1–5),
where each period represents 1 year. The input
data have been collected from various industrial
companies, meaning that the data for the first period
are representative of the industry. No effort has been
put into obtaining forecast data as it is highly expen-
sive. This, in turn, is due to the various analyses that
must be done; each input is dependent on several
macroeconomic factors. In a real case, such data
should of course be obtained as outputs are never

Table 4. Contract data.

Contract Loading Unloading Period 1 Period 2 Period 3 Period 4 Period 5
number Type Commodity port port $/tonne Value ($) (tonne) (tonne) (tonne) (tonne) (tonne)
1 COA Iron ore Tubarao, Rotterdam, 255 000 000 10 000 000 10 000 000 10 000 000 0 0
Brazil the
Netherlands
2 COA Iron ore Tubarao, Beilun, 476 000 000 20 000 000 20 000 000 20 000 000 20 000 000 0
Brazil China
3 COA Coal US East Rotterdam, 448 000 000 0 20 000 000 20 000 000 0 0
Coast the
Netherlands
4 COA Iron ore Dampier, Beilun, 336 000 000 0 15 000 000 15 000 000 15 000 000 15 000 000
Western China
Australia
5 Spot Coal US East Rotterdam, 16 5 000 000 5 000 000 5 000 000 5 000 000 5 000 000
Coast the
Netherlands
6 Spot Iron ore Tubarao, Rotterdam, 8.5 5 000 000 5 000 000 5 000 000 5 000 000 5 000 000
Brazil the
Netherlands
7 Spot Iron ore Dampier, Beilun, 8 5 000 000 5 000 000 5 000 000 5 000 000 5 000 000
Western China
Australia
 6 J. LAAKE AND A. ZHANG

more reliable than inputs. As this is purely an illus- Table 6. Result from case 1 with TC.
trative case to show the applicability of the model, Profit $ 979 952 898
Contracts
the authors have used the same value for the data for served COA1 COA2 COA3 COA4 SPOT1 SPOT2 SPOT3
all periods, so that this is a case of a steady market Period 1 1 0 1 1 5 000 000 5 000 000 4 872 727
Period 2 1 0 1 1 5 000 000 5 000 000 5 000 000
with zero development. The rates/cost for time char- Period 3 1 0 1 1 5 000 000 5 000 000 5 000 000
tering in vessels are set at 5% higher than the rates/ Period 4 1 0 1 1 5 000 000 5 000 000 5 000 000
Period 5 1 0 1 1 5 000 000 5 000 000 4 781 818
revenue of time chartering them out. The rate dif-
ference is meant to account for broker and manage-
ment fees. It also prevents the model from Table 7. Fleet changes in case 1 with TC.
suggesting chartering in a vessel only to charter Operated New TC TC
it out. Period vessels Owned Acquired building in out Sold
CAPEX are calculated based on the assumption 1 10 20 0 0 10 20 0
2 25 20 0 0 22 17 0
that all vessels are acquired with a loan of 80% of the 3 25 20 0 0 22 17 0
4 13 20 0 0 12 19 0
acquisition price with an interest rate of 4% per year. 5 12 20 0 0 12 20 0
Ship prices are calculated by subtracting ships’ 6 – – – – – – 20
annual depreciation from their acquisition value.
Downloaded by [University of Waikato] at 19:23 14 August 2015

The model applies the straight-line depreciation

method, which implies that the value of ships is
written off in equal portions over their expected
lifetimes (Stopford 2009). In addition, the loan
(80% of the acquisition price) is subtracted as it is
assumed that no repayment is made.
V. Results and analysis
Case 1 without TC
The summary in Table 5 shows result of calculation
of the optimal solution when we kept the model
from suggesting the purchase of second-hand or
new vessels or engagement in TC. Since the model
was then forced to ‘sell’ all vessels in period 6, the
income from that sale is not included in the result, as
they, in fact, stay in the owned fleet.
It can be seen here that only COA1 is served, this
being indicated by the value of 1 for all the periods
in which COA1 is valid. The numbers under the
different spot routes show how much of the available
spot cargo in tonnes is being transported in each
period.
Case 1 with TC
When we let the model choose whether to engage in
TC, it showed that the expected profit increased by
442% for the same periods (Table 6). All COAs
except COA2 were serviced, as was the majority of
the available SPOT cargo.
A shipowner is often interested in finding out the
way to deal with each type of vessel: whether to
operate it, keep it, time charter it in or out, etc.
Table 7 summarizes the fleet changes suggested
when TC is allowed. We observe a change in the
number of vessels operated in almost every period,
with a peak in periods 2 and 3. This corresponds to
the peak in these periods of the amount of cargo
lifted, as per Tables 4 and 6. Note that vessels that
are time chartered in and out are often of a variety of
types, as the model selects the optimal vessel type for
each contract. The model suggests frequent time
chartering in and out because optimal operational
performance calls for substantial changes in fleet size
and mix. When running the model, there was no
upper boundary (UB) set on how many vessels that
could be time chartered in each period. If required, a
shipowner may set such an upper boundary.

Table 5. Result from case 1 without TC. Case 2

Profit $180 940 215
In case 2, we removed the constraints on fleet
Contracts served COA1 COA2 COA3 COA4 SPOT1 SPOT2 SPOT3
Period 1 1 0 0 0 5 000 000 2 183 775 0
changes that we imposed in case 1. Table 8 shows
Period 2 1 0 0 0 5 000 000 2 183 775 0 that the COAs selected and SPOT cargo transported
Period 3 1 0 0 0 5 000 000 2 183 775 0
Period 4 1 0 0 0 5 000 000 2 183 775 5 000 000 is almost the same except for a slight difference for
Period 5 1 0 0 0 5 000 000 5 000 000 5 000 000
SPOT3 in period 4. Even with such small changes,
 APPLIED ECONOMICS 7

Table 8. Result from case 2. Sensitivity analysis

Profit $ 1 051 500 625
Contracts The optimization of the solution takes place directly
served COA1 COA2 COA3 COA4 SPOT1 SPOT2 SPOT3 through the input data, and these data may or may
Period 1 1 0 1 1 5 000 000 5 000 000 4 872 727
Period 2 1 0 1 1 5 000 000 5 000 000 5 000 000 not be of absolute certainty. Uncertainty to some
Period 3 1 0 1 1 5 000 000 5 000 000 5 000 000 extent is a feature of forecast data. Since the model
Period 4 1 0 1 1 5 000 000 5 000 000 4 781 818
Period 5 1 0 1 1 5 000 000 5 000 000 4 781 818 is to be used as a support for strategic decisions, it is
essential to know how sensitive it is to variations in
the input data. In order to do this, three forecast
Table 9. Fleet changes in case 2. scenarios are used. These are recovery (a scenario of
Operated New TC TC 15% market increase per period), through (0%
Period vessels Owned Acquired building in out Sold
increase per period) and collapse (15% decrease per
1 10 0 0 0 10 0 20
2 25 0 0 0 25 0 0 period). By market increase and decrease, we here
3 25 0 0 0 25 0 0 mean developments in terms of parameters such as
4 12 0 0 0 12 0 0
5 12 0 0 0 12 0 0 spot cargo rates, vessel prices and fuel price as illu-
6 – – – – – – 0
strated in Fig. 1 for fuel prices. For the above cases,
Downloaded by [University of Waikato] at 19:23 14 August 2015

we used the through scenario. It should be acknowl-

the expected profit is higher in case 2 than in case 1
with TC. It can also be seen, in Table 9, that the
whole initial fleet is sold in period 1, implying that,
given the input data, it is not beneficial to own
vessels over this set of periods. In real life, a ship-
owner may want to be conservative and set a limit
on the number of vessels that can be sold in one
period. This limit is easy to configure in the model.
Case 3
The purpose of including this case was to investigate
how profitable it would be to only engage in trading
vessels and not take on any contracts. This was done
by setting the revenue of the COAs and the available
SPOT cargo in the input file to 0. The options for the
use of the initial fleet would then be limited to time
chartering the vessels out or selling them.
The model suggests selling all the vessels in the
first period. The income in case 3 then consists only
of the revenue from selling the initial fleet. The
reason why this is so low is that RSvt is calculated by
subtracting the ship’s annual depreciation from its
acquisition value and the value of the initial, assum-
ing that only interest has been paid (including in
CAPEX). It may be argued that this is more accurate
for vessels bought within the optimization periods,
and that an initial fleet might have a much lower
loan-to-value ratio. However, this will be different
from case to case. The model can and should be
modified accordingly in real-life applications.
edged that the designed scenarios are moderate and
simplistic considering the volatile nature of the mar-
itime world. A more in-depth description of such
relationships, such as that between fuel prices and
freight rates, can be found in UNCTAD (2010).
We investigated how sensitive the model was to
changes in input data. This was done by running the
model for case 2 with input data according to the
three scenarios in turn. This turned out to give
markedly different results as shown in Table 10.
For the through and collapse scenarios, the model
suggests selling all the vessels in the first period (as
indicated by the OPEX being zero for both scenar-
ios), while, for the recovery scenario, it suggests put-
ting some of the fleet out on TC until the end of
period 5.
It may seem peculiar that it is the through sce-
nario, with 0% changes, that yields the lowest result,
even though the rates are lowest in the collapse
Figure 1. Historical fuel prices and scenario forecast.
 8 J. LAAKE AND A. ZHANG

Table 10. Results for case 2 with different scenarios. vessels are the same for the through and collapse
Through Recovery Collapse scenarios. This indicates that even if shipowners
Income 1 727 334 545 2 792 869 500 1 946 973 436
COA 1 039 000 000 1 039 000 000 1 515 000 000
plan for the through scenario and the collapse one
SPOT 621 054 545 1 249 454 545 383 363 636 occurs instead, the decisions taken could still be
Selling vessels 67 280 000 30 258 800 48 609 800
TC income – 474 156 155 – close to the optimal solution, as both involve TC.
Expenditures 675 833 920 1 470 036 758 737 531 187 This leaves the question of what the consequences
TC costs 314 112 330 569 434 155 246 775 978
Buying second-hand – – – would be if the recovery scenario occurs and a ship-
Buying new building – – – owner has sold all the vessels in the fleet. By setting
Voyage costs 361 721 590 543 014 198 490 755 209
CAPEX – 58 719 000 – yOWN
vt ¼ 0, we can run the model and see that the
OPEX – 298 869 405 –
Profit 1 051 500 625 1 322 832 742 1 209 442 249
expected profit would be $1 264 983 992, which is a
loss of $57 848 750 or 4% compared to the optimal
solution. This indicates that the optimization model
scenario. This is because it is not only the spot rates provides fairly robust solutions that are valid for a
that decrease by 15% yearly, but also the OPEX and range of scenarios. It must also be taken into con-
vessel and fuel prices. Furthermore, COAs become sideration that, for those owners with in-house man-
more profitable in the collapse scenario because their agement, selling all the vessels will incur additional
Downloaded by [University of Waikato] at 19:23 14 August 2015

rates are locked in while the costs of servicing them
are reduced. This also explains why the model
recommended taking up more COA cargo in the
collapse scenario. Conversely, in the recovery sce-
nario, it is profitable to take up more spot cargo.
Having observed that the economical results
shown are sensitive to changes in the input data, it
would be interesting to also investigate how changes
in input affect the strategic decisions which the
model suggests. If similar decisions have to be
made to obtain the optimal result for each scenario,
it can then be argued that the model gives a fairly
robust solution. Decisions on when to order, buy or
sell vessels and in what quantity are important and
have large effects on the results. Choices between
long-term COAs also have a significant impact.
Finally, decisions on spot cargo and TC are impor-
tant, but the shorter time horizons for these deci-
sions make it easier to manage decision changes in
real life.
Table 11 shows that the numbers of vessels oper-
ated in each period are almost the same for the
through and recovery scenarios but very different
for owned vessels, while the numbers of owned
Table 11. Comparison of operated and owned vessels in case 2
for the different scenarios.
Through Recovery
Operated Operated
Period vessels Owned vessels Owned
1 10 0 10
2 25 0 25
3 25 0 25
4 12 0 13
5 12 0 12
costs, depending on the size of the fleet. This will be
related to redundant personnel, restructuring of the
organization and office, and so on. If it becomes
profitable later to build up the owned fleet again,
the additional costs will be incurred again by the
reversed process or by outsourcing of the manage-
ment. An owner with third-party management will
not be exposed to such costs, though whether to opt
for in-house management or to outsource it is
another discussion which will not be covered here.
As for the three cases that we have discussed, the
model suggests that WBC choose the strategy pro-
posed in case 2. Even though this turns out to be the
best strategy for this setting, it is not given that it
would be the best in other settings. Note that
although the model grasps the main input para-
meters, it simplifies. There are some costs that the
calculations do not include, including brokerage
fees, port dues and changes of fuel to more expensive
diesel when sailing in emission-controlled areas.
VI. Conclusions
Strategic fleet planning is vital to shipowners because
the shipping industry is very capital intensive.
Existing studies on strategic fleet planning have
mainly been concerned with liner shipping. They
Collapse are not applicable to tramp shipping due to its
Operated special requirements of selecting long-term and
vessels Owned
15 19 0 spot contracts together with fleet decisions. To the
15 33 0 best of our knowledge, no research has tackled this
15 33 0
15 21 0 more complicated problem. This paper narrows the
15 12 0 gap by developing a mixed-integer programming

model for joint optimization of strategic fleet plan-
ning and contract analysis in tramp shipping. The
model can be efficiently solved by commercial opti-
mization solvers such as Xpress.
The model makes several original contributions
as it is very flexible and can be employed for dif-
ferent fleet planning scenarios. First, it can be used
to evaluate contracts against each other and find
the best mix of long-term and spot contracts for a
given fleet, to find the optimal fleet size and mix for
a set of contracts, or to achieve both together.
Second, it can inform decisions on how to renew
fleets and when and whether to sell and buy old or
new ships. It takes into consideration the time
charter market, recommending when to charter in
vessels and when to charter them out. It also
Downloaded by [University of Waikato] at 19:23 14 August 2015
recommends vessel sizes/types for specific freight
contracts/routes so as to maximize profits. Last but
not the least, it is also useful for users that are only
engaged in active vessel trading and not transporta-
tion. A numerical example is given to illustrate the
application of the model.
This research has its limitations. The optimization
model presented is deterministic. It could be
extended as a stochastic model to incorporate uncer-
tainties in its formulation.
Disclosure statement
No potential conflict of interest was reported by the authors.
References
Cho, S.-C., and A. N. Perakis. 1996. “Optimal Liner Fleet
Routeing Strategies.” Maritime Policy & Management 23:
249–259. doi:10.1080/03088839600000087.
Fagerholt, K., M. Christiansen, L. M. Hvattum, T. A. V. Johnsen,
and T. J. Vabø. 2010. “A Decision Support Methodology for
Strategic Planning in Maritime Transportation.” Omega 38:
465–474. doi:10.1016/j.omega.2009.12.003.
Fagerholt, K., T. A. V. Johnsen, and H. Lindstad. 2009. “Fleet
Deployment in Liner Shipping: A Case Study.” Maritime
View publication stats
APPLIED ECONOMICS 9
Policy & Management 36: 397–409. doi:10.1080/
03088830903187143.
Fagerholt, K., and H. Lindstad. 2000. “Optimal Policies for
Maintaining a Supply Service in the Norwegian Sea.”
Omega 28: 269–275. doi:10.1016/S0305-0483(99)00054-7.
Fairplay, L. R. 2009. “Sea-Web.” http://sea-web.com.
Fernando, A. J., P. Tsilingiris, E. S. Engebrethsen, and N. M.
P. Kakalis. 2011. “Robust Fleet Sizing and Deployment for
Industrial and Independent Bulk Ocean Shipping
Companies.” Infor 49: 93–107.
Imai, A., and F. Rivera. 2001. “Strategic Fleet Size
Planning for Maritime Refrigerated Containers.”
Maritime Policy & Management 28: 361–374.
doi:10.1080/03088830010020629.
Jaramillo, D. I., and A. N. Perakis. 1991. “Fleet Deployment
Optimization for Liner Shipping Part 2. Implementation
and Results.” Maritime Policy & Management 18: 235–262.
doi:10.1080/03088839100000028.
Lawrence, S. A. 1972. International Sea Transport: The Years
Ahead. Lexington, MA: Lexington Books.
Meng, Q., and T. Wang. 2010. “A Chance Constrained
Programming Model for Short-Term Liner Ship Fleet
Planning Problems.” Maritime Policy & Management 37:
329–346. doi:10.1080/03088839.2010.486635.
Meng, Q., and T. Wang. 2011. “A Scenario-Based Dynamic
Programming Model for Multi-Period Liner Ship Fleet
Planning.” Transportation Research Part E: Logistics and
Transportation Review 47: 401–413. doi:10.1016/j.
tre.2010.12.005.
Perakis, A. N., and D. I. Jaramillo. 1991. “Fleet Deployment
Optimization for Liner Shipping Part 1. Background,
Problem Formulation and Solution Approaches.”
Maritime Policy & Management 18: 183–200.
doi:10.1080/03088839100000022.
Powell, B. J., and A. N. Perkins. 1997. “Fleet Deployment
Optimization for Liner Shipping: An Integer
Programming Model.” Maritime Policy & Management
24: 183–192. doi:10.1080/03088839700000069.
Stopford, M. 2009. Maritime Economics. 3rd ed. London:
Routledge.
UNCTAD. 2010. “Oil Prices and Maritime Freight Rates: An
Empirical Investigation.” Paper presented at the United
Nations Conference on Trade and Development.
Xie, X., T. Wang, and D. Chen. 2000. “A Dynamic
Model and Algorithm for Fleet Planning.” Maritime
Policy & Management 27: 53–63. doi:10.1080/
030888300286680.