Brief presentation fleet of ships optimized programming

Programa de Atualização Profissional

Fleet of Ships Optimized Programming

methodology summary

by Gláucio Bastos, M.B.A., Ch.E.


abstract

target: from true characteristics of various types of fleets of ships and service vessels, cargoes and ports, it is generated a large number of ships scheduling initial solutions to serve loads in their ports, which will be refined in several optimization steps until achievement of a viable solution for greater financial return; this solution is then refined to produce the best 'timing’ decision for shipments starting, considering constraints in ports, docks availability, and stock levels in case of bulk cargo


neeed

during the last decades there has been an increase in competition among shipping companies, squeezing profit margins to a minimum

this situation is caused and amplified by the consolidation of the manufacturing sector, giving greater market power for cargo owners and shippers

to reduce this imbalance, there have been many mergers among shipping companies in the pursuit of scale economies gains


neeed

this trend of mergers and groupings leads to larger fleets, distributed geographically in extensive areas which require more efficient management

this picture of high complexity and dynamism makes uncertainties and delays inevitable phenomena in the operation of shipping, caused by factors such as adverse weather conditions, mechanical breakdowns, strikes and insufficient infrastructure, both in port and on board


neeed

in this context and considering the long lead times involved, it is imperative to shipping companies ensure efficiency by generating quality schedules based on advanced methods for decision support in terms of reliability and punctuality, which currently are the shortest and most effective way for achieving cost savings and generating higher profit margins


fleet characteristics

in shipping can be distinguished 03 different modes of operations:

industrial - the cargo owner also owns a fleet of vessels to transport it, the load is a certain amount of product to be shipped from one port to another within time windows, all charges are mandatory and must be met by the fleet available, no optional spot cargoes, the goal is to minimize costs


fleet characteristics in shipping can be distinguished 03 different modes of

operations:

tramp – operations that are similar to a taxi service, where ships carry available loads, is normal to an operator having some contractual charges and other options that can be met if considered profitable, the goal is to maximize profit; and

liner – operations that are similar to regular shipping lines, where the boats sail by pre-booked and published itineraries seeking to maximize their profit


issue description

the problem of ship schedule as analyzed here, applies to any type of fleet and can be a maximization problem of cost or profit and may involve contracts of affreightment (COA), signed to transport certain amounts or types of cargo between specified ports or among specific operations within a certain period, for an agreed payment per tonne or per service

there are cases where the fleet does not have enough capacity to meet all COAs during the planning horizon, so some loads can be met by spot charters, chartered for just a trip


issue description

time windows are imposed for loading and sometimes also for unloading

depending on the operation, the ships can serve single or multiple loads simultaneously with a new charging port being played and some cargo remaining on board

operations are considered with existing vessels without changes in fleet size for the short term, out of such changes are not interesting


issue description fixed costs are disregarded, since do not influence the

planning of optimum routes and schedules

variable costs of travel are considered mainly from consumption of fuel oil and diesel

it is considered a heterogeneous fleet of ships with special features, including different cost structures and load capacities

fees are charged for use of the port and channel, depending on the size of the vessel


related data the planning horizon is 3,600 hours or 150 days freight value (US$ / knot / m3) general data:

buffer–times for trip and port (hr) – slack time in case of possible delays

IFO price – interm.(trip) and MDO (port) (US$ / ton) time window for standard operation at port (hr)

cargo data: identification, volum (m3), weight (ton) load port, unload port cargo time availability at loading port (hr), cargo time of

arrival at unloading port (hr)


related data ship data:

identification, volumetric capacity (m3), dead-weight tonnage - dwt (ton)

mean speed (knot / hr) IFO consumption (ton / day), MDO consumption idle at

port (ton / day)

port data: identification, distance between ports of trip (knot) produtivity (m3 / hr), port fee (US$), cargo operation fee

(US$ / m3) available ETAs for arrivings (hr)


solution• using MSLS heuiístics *

solution steps: 1ª: large amount (no_init) of initial solution are

generated by a partially random insertion heuristic of cargo on ships schedules

2ª: a certain amount (no_quick) of the best (by profit or cost criterion) initial solutions are improved by a fast local search heuristic

3ª: it is selected a certain amount (no_extended) of the best quick local search solutions to be enhanced by a extended local search heuristic


results from data of:

13 ships 50 cargoes 9 ports located among southeast USA and west Europe

the following developments in profit fleet were obtained: to no_init = 100, no_quick = 6, no_extended = 1

init. sol.: worst = $ 628.638 / best = $ 1.127.639 fast sol.: worst = $ 1.109.300 / best = $ 1.168.316 extended sol.: $ 1.295.348

to no_init = 1000, no_quick = 14, no_extended = 4 init. sol.: worst = $ 576.334 / best = $ 1.334.752 fast sol.: worst = $ 1.183.747 / best = $ 1.334.752 extend.sol.: worst = $ 1.237.389 / best = $ 1.334.752


results

conclusions: to a smaller number of initial solutions, it is evident the

effective contribution of each optimization level to achieve the best final solution

for a large number of initial solutions, the optimization included in his generation algorithm is sufficient to reach the best solution, even before the next two optimization levels so as was hit the nearest optimal solution, it can

hardly be optimized and is better than that achieved with a smaller number of initial solutions, featuring the robustness of the results


resultsoptimal solution – cargoes data: volums


resultsoptimum solution – cargoes data: times

1.ETA load port2.trip time3.load port time4.unload port time5.reduced trip time


resultsoptimum solution – scheduling result: cargoes through ships and ports

1.ship2.load port3.unload port


results

optimum solution– scheduling result: routes of loaded vessels

Port 2

Port 3

Port 1

Port 9

Port 5

Port 6

Port 7

Port 4

Port 8

Ship 9

Ship 10

Ship 12

Ship 1

Ship 3

Cargo 3

Cargo 5

Cargo 14

Cargo 10

Cargo 15

Ship 11

Cargo 16

Ship 2

Cargo 22

Ship 7

Cargo 24

Ship 5

Cargo 25

Ship 13

Cargo 27

Cargo 29

Cargo 31

Ship 6

Cargo 32

Ship 8

Cargo 42

Ship 4

Cargo 46


issue extension

if the fleet is designed to bulk transport, e.g. LNG, between liquefaction port (load) and regasification port (unload), it is considered availability of docks and stock (between maximum and minimum levels) at liquefaction port

if production exceeds contractual obligations for LNG delivery spot cargoes are sold on the market, including through other shipping company’s ships, considered as a means of controlling the inventory level without profit associated


issue extension customers specify certain dates or periods - windows of

time - when may or may not accept deliveries from the time window for supplying is determined the time

window for loading delivery should occur within the time windows:

internal – periods of customer preference, and external – besides internal comprises additional periods

of a few days before and after the internal delivery must necessarily take place inside the external

window which comprises also the inner window if delivery occurs in periods between internal and external

window, there will be contractual penalties, as exemplified in the following figure


issue extension

example of penalty cost function proportional to the square of the number of days before or after the inner window on which delivery is made - inner window is 7 days and outer window also contains 7 more days before and after the inner window


issue extension results

considering the same data and the solution given for the 1st part of the problem, in the extension through application of MILP method, came out the solution for minimization of penalty cost, including not only the docks restrictions but also the LNG stock in liquefaction port

in next figure are presented the optimal time partitions defined for boarding each LNG cargo within a time window with the total of 21 continuous partitions: 07 partitions of the inner window with more 07 before and 07 after completing the external window


issue extension resultsoptimum solution: cargoes and their time partitions for loading start

time partitions:from 01 to 21 insideexternal timewindow

Technology

Brief presentation fleet of ships optimized programming