UCL ENERGY INSTITUTE Assessment of Shipping’s Efficiency Using Satellite AIS data Tristan Smith, Eoin O’Keeffe, Lucy Aldous and Paolo Agnolucci Prepared for the International Council on Clean Transportation March 2013 CONTENTS CONTENTS ................................................................................................................................................................. i LIST OF FIGURES ................................................................................................................................................... iii LIST OF TABLES ...................................................................................................................................................... v ACKNOWLEDGEMENTS .................................................................................................................................... vi EXECUTIVE SUMMARY ....................................................................................................................................... 1 OBJECTIVE ........................................................................................................................................................... 1 METHOD .............................................................................................................................................................. 1 FURTHER WORK .................................................................................................................................................. 2 INTRODUCTION ..................................................................................................................................................... 4 1. METHODS AND DATA FOR ESTIMATING ENERGY EFFICIENCY ......................................... 8 1.1. METHOD FOR TECHNICAL EFFICIENCY CALCULATION ...................................................................... 9 1.2. DATA SOURCES AND MISSING DATA APPROACH .................................................................................. 9 1.2.1. Main engine power ................................................................................................................... 10 1.2.2. Boiler fuel consumption .......................................................................................................... 10 1.2.3. Main engine % MCR in the design condition and design speed ....................................... 11 1.2.4. Main engine specific fuel consumption ................................................................................. 11 1.2.5. Fuel type of main engine and main engine type................................................................... 12 1.2.6. Auxiliary engine assumptions .................................................................................................. 15 1.3. METHOD FOR OPERATIONAL EFFICIENCY CALCULATIONS............................................................. 15 1.4. DATA SOURCES AND MISSING DATA APPROACH ................................................................................ 17 1.4.1. Identification of vessel state using AIS data ......................................................................... 18 1.4.2. Derivation of operational data ................................................................................................ 20 1.4.3. Extrapolation of operational parameters .............................................................................. 23 1.4.4. Generating the region to region flows ................................................................................... 24 1.4.5. Loading condition ..................................................................................................................... 24 1.4.6. Days at sea .................................................................................................................................. 25 1.4.7. Operating speed ........................................................................................................................ 25 1.4.8. Reliability indicators.................................................................................................................. 25 1.4.9. Speed impact parameter ........................................................................................................... 26 1.4.10. Loading condition impact parameter ..................................................................................... 28 1.4.11. Condition impact parameter ................................................................................................... 28 1.4.12. Main engine specific fuel consumption impact parameter ................................................. 30 1.4.13. Weather impact parameter ...................................................................................................... 31 1.5. DATA AND METHOD APPLICATION IN THIS STUDY ........................................................................... 32 1.5.1. Payload capacity ........................................................................................................................ 32 1.5.2. Coverage of the global fleet..................................................................................................... 33 1.5.3. Missing dwt ................................................................................................................................ 33 1.5.4. Missing speed ............................................................................................................................. 35 1.5.5. Missing design draught ............................................................................................................. 35 1.5.6. Missing data, all variables, with respect to gross tonnage .................................................. 37 2. CALCULATIONS OF WORLD FLEET ENERGY EFFICIENCY .................................................... 38 2.1. AGGREGATIONS INTO FLEET STATISTICS ........................................................................................... 39 2.2. FILTERING ................................................................................................................................................. 39 2.3. COMPARISON WITH NUMBERS OF SHIPS REPORTED IN IMO 2ND GHG ........................................ 40 2.4. VESSEL SPEEDS AND SLOW STEAMING................................................................................................. 42 2.5. FUEL CONSUMPTION ............................................................................................................................... 50 2.6. CAPACITY UTILISATION AND TRANSPORT WORK .............................................................................. 54 2.7. FLEET ENERGY EFFICIENCY DATA AND COMPARISON WITH IMO 2ND GHG STUDY ................ 58 2.8. SHIPPING ROUTES AND ENERGY EFFICIENCY .................................................................................... 68 2.9. UNCERTAINTY .......................................................................................................................................... 69 2.9.1. Uncertainty in an individual ship‘s technical efficiency estimate ...................................... 70 2.9.2. Uncertainty in an individual ship‘s operational and normalised operational efficiency estimate ..................................................................................................................................................... 70 2.9.3. Implications of individual ship‘s estimated uncertainty to the study of fleet aggregate statistics .................................................................................................................................................... 71 3. SHIP EFFICIENCY AND PRICES ............................................................................................................. 72 3.1. IMPORTANCE OF EFFICIENCY AND ESTABLISHMENT OF A TWO-TIER MARKET .......................... 74 3.2. RATIONALE OF THIS STUDY ................................................................................................................... 75 3.3. METHODOLOGICAL APPROACH AND GUIDE FOR INTERPRETING TABULAR RESULTS............... 78 3.3.1. Mean comparison...................................................................................................................... 78 3.3.2. Regression analysis .................................................................................................................... 79 3.4. IMPACT OF ENERGY EFFICIENCY ON THE TIME CHARTER MARKET .............................................. 80 3.4.1. Evidence from the industry press .......................................................................................... 80 3.4.2. Results from mean comparison and regression analysis ..................................................... 81 3.5. IMPACT OF EFFICIENCY ON THE NEW AND SECOND HAND MARKETS .......................................... 84 3.5.1. Evidence from the industry press .......................................................................................... 84 3.5.2. Results from mean comparison and regression analysis ..................................................... 84 3.6. IMPACT OF ENERGY EFFICIENCY ON THE DEMOLITION MARKET ................................................. 92 3.6.1. Evidence from the industry press .......................................................................................... 92 3.6.2. Results from mean comparison .............................................................................................. 92 3.7. SUMMARY OF FINDINGS .......................................................................................................................... 97 4. CONCLUDING REMARKS ......................................................................................................................... 99 4.1. SUMMARY OF FINDINGS ........................................................................................................................ 100 4.2. OBSERVATIONS ON THE ESTIMATED TECHNICAL AND OPERATIONAL EFFICIENCY ............... 100 4.3. OBSERVATIONS ON THE ESTIMATED RELATIONSHIP BETWEEN EFFICIENCY AND PRICES..... 105 4.4. INFERENCES FROM THE ANALYSIS OF EFFICIENCY ......................................................................... 106 4.4.1. Energy Efficiency Design Index .......................................................................................... 106 4.4.2. The need to understand the operational efficiency and the drivers of operational efficiency ................................................................................................................................................ 107 4.4.3. Market based measures discussion and techno-economic analysis ................................. 108 4.5. FURTHER WORK FOR IMPROVING ESTIMATES .................................................................................. 108 4.5.1. Estimating shipping activity and operational parameters using AIS .............................. 108 4.5.2. Estimating fuel consumption for technical and operational parameters ....................... 109 4.5.3. Longitudinal analysis .............................................................................................................. 109 REFERENCES........................................................................................................................................................ 111 ANNEX 1 – ASSUMPTIONS BY SHIP TYPE ............................................................................................... 115 ANNEX 2 – FLEET DISAGGREGATION CATEGORIES ...................................................................... 116 ANNEX 3 – COUNTRY-TO-REGION MAPPING ..................................................................................... 117 ANNEX 4 – RESULTS BY SHIP TYPE ........................................................................................................... 123 ANNEX 5 – ROUTE EFFICIENCY ................................................................................................................. 204 LIST OF FIGURES Figure 1: Sample set of 5000 tankers (1770 data points, 35%), 1200 LPG carriers (451 data points, 38%) and 2500 container ships (1465 data points, 59%), Source: Clarksons, 2012 ......................................... 12 Figure 2: Power Distribution of Containerships/LPG carriers / Oil tankers (sample size: 8725 ships, 2394 undefined), source Clarksons, 2012 ..................................................................................................... 14 Figure 3: Coverage of container ships of 0-7,000 GT in August and September 2011 (top) and container ships greater than 65,000 GT in August 2011 (bottom). Black dots indicate messages received from vessel at the reported locations ....................................................................................................................... 19 Figure 4: Overview of approach for generating operational data ...................................................................... 20 Figure 5: Graphic showing vessel states and generated edges. The yellow ―comet‖ (original code provided by Dr Austwick, UCL CASA) indicates the current location of the vessel (previous locations are shown in its stream with transparency increasing over time). The red Bezier curves (steeper curve indicates origin) indicate the voyages generated (linking origin to destination, not showing the route taken), with each node indicating a port stop. This example is for a VLCC vessel in March 2011. ... 22 Figure 6: Indicative representation of the addition of the network generated using both the modelled approach (red network) and the message 5 (blue). When there is a match between the modelled port time and the Message 5 ETA, the Message 5 edge is overlaid (in the graphic there was a match for Shanghai and Rotterdam but not Singapore and Felixstowe). The origin port for the message 5 edge is set as the origin port for the same edge in the modelled network. The reason it is not set as the previous port indicated in the message part (i.e. an edge from Rotterdam to Shanghai) is that there are gaps in the message 5. ................................................................................................................................ 22 Figure 7: Scatter plot matrix of proportion of days at sea, the sum of the maximum amount of time missing at each edge or node in the modelled network and the proportion of modelled edges not matched to edges generated from the S-AIS message 5 for the VLCC fleet represented in the model. For example the proportion of modelled edges not matched in the network in Figure 6 would be 0.5. ....................................................................................................................................................................... 23 Figure 8: Exemplar vessel speed power relationship ........................................................................................... 27 Figure 9: Speed dependency in the propulsion coefficient ................................................................................. 27 Figure 10: The variation in resistance components as a function of draught .................................................. 28 Figure 11: Draught – Power relationship example, draught = draught/design draught ................................ 28 Figure 12: Changes in condition efficiency and shaft power with time out of dry dock ............................... 30 Figure 13: sfc – Engine load curve (MAN 6S70ME-C8.2) ................................................................................. 31 Figure 14: Distribution of ships with missing deadweight according to built year ......................................... 34 Figure 15: Distribution of ships missing speed data, by year built .................................................................... 35 Figure 16: Distribution of ships missing draught data, by year built ................................................................. 36 Figure 17: number of ships (x 104) missing dwt, speed or draught data, by Gross Tonnage ........................ 37 Figure 18: The global regions used in the model: Africa, Australasia, Brazil, Canada, Central America, China, Europe, India, Indian Subcontinent, Japan, Middle East, North East Asia, Russia, South America, South East Asia, USA. .................................................................................................................... 39 Figure 19: Ratio of average operating to design speed, crude oil tankers all ages ........................................... 44 Figure 20: The fleet average ratio of design speed to operating speed for each of the size and age categories, crude oil tankers ............................................................................................................................ 45 Figure 21: Loaded to ballast average speed ratios for the crude oil fleet .......................................................... 46 Figure 22: Ratio of average operating to design speed, LPG tankers all ages .................................................. 47 Figure 23: Average operating speed in the loaded and ballast condition, LNG tankers ................................ 48 Figure 24: Ratio of average operating to design speed, LNG tankers .............................................................. 48 Figure 25: Ratio of average operating to design speed, container ships ........................................................... 49 Figure 26: The fleet average ratio of design speed to operating speed for each of the size and age categories, container ships ............................................................................................................................... 50 Figure 27: Average daily fuel consumption for all loaded and ballast days, crude oil tankers ...................... 52 Figure 28: Average daily fuel consumption for all loaded and ballast days, LNG tankers ............................ 53 Figure 29: Average daily fuel consumption for all loaded and ballast days, container ships ......................... 54 Figure 30: average capacity utilisation and ratio of loaded days to days at sea, crude oil tanker .................. 56 Figure 31: average capacity utilisation and ratio of loaded to ballast days, LNG tanker ................................ 57 Figure 32: technical efficiency in the crude oil tanker fleet ................................................................................. 60 Figure 33: operational efficiency in the crude oil tanker fleet............................................................................. 61 Figure 34: operational efficiency in the product tanker fleet .............................................................................. 61 Figure 35: operational efficiency in the dry bulk fleet .......................................................................................... 62 Figure 36: technical efficiency in the general cargo fleet ..................................................................................... 63 Figure 37: operational efficiency in the general cargo fleet ................................................................................. 63 Figure 38: technical efficiency in the LNG tanker fleet ....................................................................................... 64 Figure 39: operational efficiency in the LNG tanker fleet .................................................................................. 65 Figure 40: technical efficiency in the container ship fleet ................................................................................... 67 Figure 41: technical efficiency in the container ship fleet by size and age category ........................................ 67 Figure 42: operational efficiency in the container ship fleet ............................................................................... 68 Figure 43: Route operational vessel efficiency for dry bulk, weighted by dwt on route. The numbering on the x-axis corresponds to region names on the vertical axis in top to bottom order. Zero values are on routes where no vessels have been recorded. ......................................................................................... 69 Figure 44 Conditions required for more efficient newbuilds to be paid a higher price. ................................ 77 Figure 45 Logic of the value chain discussed in this study. ................................................................................ 78 Figure 46: Ship transport supply at given normalised operational efficiency levels ...................................... 101 Figure 47: Ship transport supply at given operation-to-design speed ratios .................................................. 102 Figure 48: Operational, technical, 90th percentile, and 95th percentile energy efficiency ........................... 103 Figure 49: Container ship transport supply by 3000-8000 TEU container ships by ships average annual CO emission rate ........................................................................................................................................... 103 2 Figure 50: Comparison between the estimations of normalised operational efficiency in this report and the operational efficiency data in the IMO 2nd GHG Study .................................................................. 104 Figure 51: Illustration of annual average container ship normalised energy efficiency per route, where green is better energy efficiency, red is lower ............................................................................................. 105 Figure 52: Relationship between technical and operational efficiency of the dry bulk and container shipping fleets .................................................................................................................................................. 107 LIST OF TABLES Table 1: Some different definitions of energy efficiency ....................................................................................... 5 Table 2: Summary of variables for technical efficiency calculations.................................................................. 10 Table 3: Assumptions for the boiler daily fuel consumption ............................................................................. 11 Table 4: Engine test bed measured ranges for a variety of engine type, size and age ..................................... 12 Table 5: Carbon factors for different fuels ............................................................................................................ 12 Table 6: Fuel type allocation according to engine size and power type (excluding engines installed on LNG carriers) .................................................................................................................................................... 13 Table 7: Specific fuel consumption values as a function of auxiliary engine size ............................................ 15 Table 8: Summary of variables for operational efficiency calculations ............................................................. 17 Table 9: Summary of impact parameters................................................................................................................ 18 Table 10: Input variables and assumptions for the calculation of the coefficients k and b ........................... 26 Table 11: Propulsion coefficient assumptions ...................................................................................................... 27 Table 12: Typical annual hull roughness increments (Carlton, 2007) ............................................................... 29 Table 13: increments in total resistance over time ............................................................................................... 29 Table 14: Upper limits of sfc used in the estimate ................................................................................................. 31 Table 15: Ship types and their corresponding capacity measure ........................................................................ 33 Table 16: List of missing information for different ship types ........................................................................... 34 Table 17: Missing speed information by ship type ............................................................................................... 35 Table 18: Missing draught information by ship type ............................................................................................ 36 Table 19: Numbers of wet and dry bulk ships ...................................................................................................... 41 Table 20: Numbers of gas bulk ships ..................................................................................................................... 41 Table 21: Numbers of container ships and car carriers ....................................................................................... 42 Table 22: Speeds in the wet and dry bulk fleet ...................................................................................................... 43 Table 23: Speeds in the gas bulk fleet ..................................................................................................................... 46 Table 24: Speeds in the container ship and car carrier fleets .............................................................................. 49 Table 25: Main engine annual fuel consumption in the wet and dry bulk fleets (average per ship) ............. 51 Table 26: Main engine annual fuel consumption in the gas bulk fleets ............................................................ 52 Table 27: Main engine annual fuel consumption in the container ship and car carrier fleets ....................... 53 Table 28: Capacity utilisation and transport work in the wet and dry bulk fleets ........................................... 55 Table 29: Capacity utilisation and transport work in the gas bulk fleets ........................................................... 57 Table 30: Capacity utilisation and transport work in the container ship and car carrier fleets ..................... 58 Table 31: Technical and operational efficiency in the wet and dry bulk fleets ................................................ 59 Table 32: Technical and operational efficiency in the gas bulk fleets ............................................................... 64 Table 33: Technical and operational efficiency in the container ship and car carrier fleets .......................... 66 Table 34 Average time charter prices ($) paid in the in the dry bulk sector. .................................................... 82 Table 35 Average time charter prices paid in the container sector. ................................................................... 83 Table 36: Results from the regression analysis for time charters. ...................................................................... 84 Table 37 Average new build prices in the dry bulk sector. ................................................................................. 86 Table 38 Average new build prices in the wet bulk sector. ................................................................................. 87 Table 39 Average new build prices in the container sector. ............................................................................... 88 Table 40: Results from the regression analysis for new build prices. ................................................................ 89 Table 41 Average second hand prices in the dry bulk sector ............................................................................. 90 Table 42 Average second hand prices in the container sector ........................................................................... 91 Table 43: Results from the regression analysis for Second-Hand Prices. ......................................................... 92 Table 44: Selected characteristics of demolished ships and those operating in the dry bulk fleet ................ 94 Table 45: Selected characteristics of demolished ships and those operating in the wet bulk fleet. .............. 95 Table 46: Selected characteristics of demolished ships and those operating in the container fleet. ............. 96 Table 47: Summary of ship characteristics and average efficiency by ship type from this analysis ............ 101 ACKNOWLEDGEMENTS This report was funded by the International Council on Clean Transportation, as part of its World Shipping Efficiency Indices project. The authors thank Haifeng Wang, Nic Lutsey, and Sarah Chambliss for their technical contributions, and Fanta Kamakaté for critical review of this work. The authors also especially thank the participants of the study‘s Technical Advisory Committee for their reviews of the method and report. This committee consisted of Bryan Wood Thomas of the World Shipping Council; Angie Farrag of the Clean Cargo Working Group; Paul McStay and Dimitris Argyros of Lloyd‘s Register; Stewart McMahon of KPMG and Simon Davies. In addition, the authors thank Sam Kimmins and Will Dawson of Sustainable Shipping Initiative for their thoughtful comments. Whilst not a component of the project, the study builds on experience and knowledge gained during the Research Councils United Kingdom (RCUK) Energy programme and industry-funded Low Carbon Shipping – A Systems Approach project. EXECUTIVE SUMMARY OBJECTIVE The objective of this study is to generate new knowledge and insight on the subject of shipping‘s technical and operational energy efficiency, both to assist shipping industry stakeholders in understanding the statistics and drivers of energy efficiency in the markets in which they are operating, and to contribute to the ongoing discussions in the policy sphere around energy efficiency and GHG emission reduction. This study, which is the first to use Satellite Automatic Identification System data to analyse the energy efficiency of the global fleet, is also intended to act as an evaluation of this new data source and a discussion of its benefits and any shortcomings when used for the estimation of energy efficiency. Studying the fleet in 2011, this study‘s particular focus is on the extent to which the recent phenomenon of ‗slow steaming‘ has influenced operational behaviour in different fleets (ship types and sizes), and the probable consequences to energy efficiency. As well as describing the technical and operational efficiency of different fleets of ships, analysis is also undertaken to assess whether energy efficiency appears to be represented in prices (time charter, newbuild, second-hand) in different markets and in the selection of ships for scrapping. METHOD There is no widely accepted method for quantifying the technical or operational efficiency of existing ships. Section 1 therefore outlines in detail the method, data and assumptions that have been applied here, primarily for the purposes of transparency. Both methods can be related back to the Energy Efficiency Design Index and Energy Efficiency Operational Indicator equations, a decision made to keep consistency with existing literature and to take advantage of the increasing familiarity in the sector with these quantifications. However, the details of the method in both cases have been tailored to the specifics of the input data used. In the case of operational efficiency estimation, Satellite Automatic Identification System (AIS) data is combined with established naval architecture and marine engineering analysis techniques in order to derive estimates of a ship‘s annual fuel consumption and CO emissions. 2 Operational efficiency is calculated using two different methods. One utilized Satellite AIS observed draughts to estimate the ship‘s loading condition and payload utilization. The second method (normalized operational efficiency) applied the IMO 2nd GHG Study estimate of capacity utilization to the ship‘s payload. For all energy efficiency calculations, an estimate of the quantification of uncertainty and a discussion of its significance to the findings is included in Section 2. A third analytical element of the study links the effects of ships‘ energy efficiency to wider market dynamics. To estimate the influence of ship energy efficiency on market prices, both regression analysis and mean comparison techniques, commonly used in economic analysis, were applied. DATA The data used as inputs to the method come from a variety of sources. Data characterizing ship‘s technical specifications are from Clarksons World Fleet Register. The dataset includes all ships active in 2011 (e.g. including ships built in previous years). The data characterizing ship‘s operational characteristics (speed, loading condition and voyage characteristics) are obtained from individual ship‘s AIS transponders, specifically Satellite AIS which means the AIS signal has been received by satellite, extending to global coverage previous AIS datasets which were obtained from shorebased receivers and only included ship movements in coastal waters. The Satellite AIS dataset is for the period 1st January 2011 to 31st December 2011 only. Where necessary (e.g. missing data), the technical and operational data is supplemented or compared with data and assumptions from the IMO 2nd GHG study, or in some instances other third party sources (e.g. describing the world fleet in 2007). Market data describing prices in different shipping markets are obtained from Clarksons Shipping Intelligence Network (2007-2012). KEY FINDINGS 1 The study‘s quantifications and discussion of energy efficiency are presented in Section 2, analysis and discussion of energy efficiency and prices in Section 3, and key findings in Section 4. The study finds that the proposed method and its novel application of Satellite AIS data can produce valuable insights into the energy efficiency of the existing fleet, particularly in characterizing the variability and heterogeneity of the fleet‘s operational parameters (e.g. average annual operating speed) as well as updating estimates of operational energy efficiency for the first time since their presentation in the IMO 2nd GHG Study. The study also presents a number of further uses of Satellite AIS, demonstrating its potential for analysing energy efficiency on routes between individual countries and regions and the geography of shipping‘s energy efficiency. The analysis estimates average operating speeds to be 10-15% lower for many of the bulk fleets (tankers, dry bulk), and approximately 25% lower for container ships, relative to the average annual operating speeds presented in the IMO 2nd GHG Study. The consequence of these observed differences in speed is significant reductions in fuel consumption (see Section 2 for details, but as much as 30 to 40% reduction for many of the bulk fleets and 50% and above for some container ship fleets relative to the estimates presented in the IMO 2nd GHG study). Ultimately, the speed reduction, which in turn reduces transport work, absorbs some of the impact of the main engine fuel consumption on energy efficiency, so that relative to the IMO 2nd GHG study estimates of overall efficiency, the improvement in operational efficiency (using IMO 2nd GHG capacity utilization data) is approximately 10% for many of the bulk fleets, rising to 30% for some of the container fleets. The study builds up the statistics of each fleet (ships of common size and type) using the technical and operational data characterizing each individual ship. To present the results, a number of scatter plots have been used (individual ships as individual data points), as well as histograms (Section 2 and Section 4). These show that there is high heterogeneity in the efficiency of each fleet. One example that illustrates this is the very large crude carrier fleet, which has estimated technical efficiencies ranging between 2-3 gCO /t.nm, but estimated average annual normalized operational efficiency (i.e. each ship‘s efficiency 2 calculated to allow for variation in speed and fuel consumption specific to the operational parameters but with capacity utilization assumptions taken from the IMO 2nd GHG Study) of approximately 2- 9gCO /t.nm. This wide spread between the most efficient and least efficient ship in many of the fleets is 2 predominantly attributable to differences in average operating speed (even within fleets which have common design speeds), and shows a potential for further gains in average energy efficiency of many fleets from further take up of slow steaming. Analysis of prices and energy efficiency data in Section 3 shows that most markets display evidence that supports the expectations that more efficient ships command higher prices. Higher prices were observed in the time charter market, to a limited extent in the newbuild market and the second hand market. There is also evidence that energy efficiency is influencing the scrapping of ships. However in the majority of cases the higher prices do not fully represent the fuel cost saving implicit in the efficiency differential. Differences occur between the ship types studied with container ship fleets generally showing evidence of higher price premiums in recognition of technical efficiency. This is a finding with important implications to the ongoing discussion around market based measures. The observation of a price premium for energy efficiency implies that the shipping markets will convert price signals (e.g. a carbon price) into energy efficiency. However the observation that this premium does not appear to represent the full magnitude of the available fuel cost savings implies that much of the analysis that applies Marginal Abatement Cost Curves (MACCs) to estimate the magnitude of the CO emission reduction at a given carbon price may 2 be optimistic. The findings in this study support the idea that there may be market barriers (e.g. informational or split-incentive barriers) obstructing the adoption of energy efficiency in shipping, however there is further work required before this can be fully attributed or quantified. FURTHER WORK The finding referred to above, that there is significant heterogeneity in operational efficiency in a given fleet in a given year, points in turn to the need to understand the drivers of operational efficiency. This 2 study investigates whether there is a connection to ship age, but finds no evidence to support this idea. Further work could investigate whether there are other variables (e.g. type of charter, nature of the fixture, operator or customer preferences, technical constraints on machinery operation, crew details) which help to explain why the range of operational efficiency within a fleet is so high, and therefore whether anything could be done (e.g. incentive, training, funding) to enable a greater number of ships to operate at higher efficiency. In addition to the need for further cross-sectional analysis, longitudinal analysis can help to explain the influence that commercial pressures (e.g. fuel prices and freight rates/prices) have on fleet‘s operational efficiency. Relative to the IMO 2nd GHG Study (2007 data), a significant change in many of the parameters (speed, fuel consumption) and ultimately operational efficiency is observed in this study (2011 data). Whilst a difference in fuel price has occurred in that time frame, the larger difference between 2007 and 2011 is in the freight rates/prices, with the shipping industry seeing a widespread reduction in revenues. In addition to the change in commercial drivers, there has been increased discussion (e.g. in the media, conferences, policy arena) of measures that can improve operational efficiency (e.g. Virtual Arrival, slow steaming, voyage optimization, Ship Energy Efficiency Management Plans), some of which may have created change in behaviour. Recent history is therefore a rich, complex, but relatively untapped source deserving further work to extract information that could provide significant insights both for the commercial strategy of shipping‘s stakeholders and also the policy makers attempting to design regulation that will achieve GHG emission reductions in the most cost-effective manner. This study demonstrates there is a large potential for Satellite AIS data to be useful for understanding ship operational behaviour and efficiency. However, shortcomings include uncertainty on some of the data (particularly user-entered data such as ship draught), and difficulty with coverage (both sporadic coverage in the open ocean and poor coverage in coastal areas with high density of shipping). These shortcomings are manageable and this report details the processing steps for the raw data and how filtering can be applied to ensure that spurious data is not included in the fleet aggregate statistics. However, there is scope for further work to validate this processing to improve data quality and its application, such as further quantification of uncertainties through validation against other similar datasets (e.g. Long Range Identification and Tracking) and combination with global shorebased data to improve coverage. Whilst further work can be carried out on the theoretical naval architecture and marine engineering models that deploy the data to calculate fuel consumption, the most important next step is a detailed and transparent validation of such models against actual fuel consumption data from ship operators. 3