Optimizing the Machinery System Selection in a Ship Design Process, Taking Environmental Regulations and Uncertainty into Consideration
MetadataVis full innførsel
- Institutt for marin teknikk 
Shipping is known for being the most efficient mode of transportation. The world's seaborne trade totalled to almost 9.6 billion tons in 2013, and the maritime shipping industry carries about 90% of globally transported goods (UNCTAD, 2014, IMO, 2014a). This implies significant environmental footprints, and a big issue is how to reduce the harmful emissions from ships.The objective in this thesis is developing an optimization model that optimizes the concurrent selection of machinery system, fuel type used and air emission controls installed on a ship. An emission control is any measure that can be installed on the ship with the intention of reducing emissions. The model is applied for a ship in the design phase, and an evaluation of uncertain elements to consider for the operation of the ship in the future, is performed. In this thesis, the regulations of CO2, SOx and NOx are considered and challenges related to the stricter regulations applying in special emission control areas (ECAs) is accounted for. The revised Annex VI of the International Convention for the Prevention of Pollution from Ships (MARPOL) is reference for the emission goals for the work presented.The optimization model developed is a stochastic two-stage recourse model. The stochastic aspect account for uncertain fuel prices in the future. The model s objective function is cost driven and summarizes the total installation and operational cost for the machinery system and emission controls selected. Interaction effects are taken into consideration. One of the main constraints are ensuring compliance with the regulations given by considering the reduction in emissions from the selected main fuel type and emission controls.The first-stage of the stochastic model assumes known values for all elements considered. The second-stage, evaluates a set of scenarios where possible fuel projections are represented. Each scenario is represented with a given probability. The goal of the model is to provide a balanced view of the future in order to make better decisions today. Scenarios are generated with different projections of the fuel types HFO, MGO and LNG from 2020 to 2030. Emission controls are ensuring compliance with the regulations together with the selection of fuel, and the number of emission controls included in the case study for the model totals to 42. In the scenario generation a dependence between the development of HFO and MGO price is taken into account while the price development of LNG is considered independent. This results in 18 scenarios to evaluate, where eight of these have been taken into further consideration.A splitting of the execution of the model is found necessary for obtaining a reasonable running time. The running of the model on 13 SOx and NOx controls only, reduces the running time to 7 seconds with 18 scenarios and 15 time periods. Similarly, when the model is running for the 29 additional CO2 controls, optimal solution is found in 930 seconds. The splitting of the execution of the model can be justified from the fact that the CO2 measures are independent of the scenarios, and there are not significant interaction effects present between the two parts.The stochastic model is executed with input data evaluated in a case study. The results from the model, concludes with running on HFO and installing an exhaust gas recirculation system and a seawater scrubber as emission controls, in order to comply with given regulations. The equivalent deterministic model applied for comparison, presents the same conclusion. The presented results include special attention to certain changes in the input data, and how these changes affect the optimal solution presented by the model. The stochastic model did not come to a different conclusion than the corresponding deterministic model when original input data was applied. The stochastic model is still considered a favourable approach, as it represents possible outcomes of the future in a better way. The results presented show that when several outcomes of the future are assessed, a more diverse evaluation can be performed. Especially positive, is the short running time of the presented stochastic model when executing the model with NOx and SOx reducing measures only. This model is promising for producing decision support that is valuable in a real-life setting.