Estimating shipping s operational efficiency Tristan Smith, UCL Energy Institute With gratitude to my colleagues: Eoin O Keeffe, Lucy Aldous tristan.smith@ucl.ac.uk
http://www.theicct.org/sites/default/files/publications/ UCL_ship_efficiency_forICCT_2013.pdf
Global shipping emissions According to IMO 2 nd GHG What might be happening now Annual CO2 emissions What might be likely? Return of BAU EEDI/SEEMP 2010 2030 2050
What to measure? Fuel consumption X Cf Payload x distance Fuel consumption X Cf dwt x F x distance = Operational Eff. Steamed? Great circle? = Normalised Operational Eff. sector average
Deriving fleet technical and operational characteristics
Overview of method Input Output Voyage/route maps S-AIS database Clarksons World Fleet Register Calculation Sorted into ship types Missing data algorithms Extract voyage/ operational detail Resistance and propulsion model Individual ship s operation statistics Energy efficiency calculations Fuel consumption calculations Aggregate statistics Individual ship statistics Literature
Play movie all VLCC Thanks to Martin Austwick
Play movie aframax voyages
Validation of Fuel Consumption Calculation Loaded: Ballast: 3 6 2.5 5 Fuel consumption, metric tonnes/hr 2 1.5 1 0.5 Fuel consumption, metric tonnes/hr 4 3 2 1 Blue = Estimated Green = Measured 0 0 5 10 15 20 Ship speed, knots 0 0 5 10 15 20 Ship speed, knots
Findings
VLCC 2-3gCO2/tenm
!
Operational efficiency = CO2 emitted p.a. / transport work done 1.5-8 gco2/tenm VLCC
Findings: dwt (tonnes) overall efficiency gco2/tnm Crude oil tankers >= < IMO 2 nd GHG (2007) calculated OE, filtered (2011) calculated NOE (2011) 80000 120000 10.9 12.8 10.8 120000 200000 8.1 8.5 6.0 200000 + 5.4 6.4 4.3
Ra3o*of*average*opera3ng*speed*to*design*speed* 1.1% 0.9% gco2/ceunm* 0.7% 0.5% 0.3% 0.1% 200" Normalised*opera8onal*efficiency* Rest%of%fleet% Wallenius%Lines%AB% gco2/ceunm* 180" 0% 5000% 10000% 15000% 20000% 25000% 30000% 35000% 40000% 45000% 50000%!0.1% 160" Dwt*tonnes* 140" 120" 100" 80" 60" 40" 20" Rest"of"fleet" Wallenius"Lines"AB" 0" 0" 5000" 10000" 15000" 20000" 25000" 30000" 35000" 40000" 45000" 50000" Dwt*tonnes*
Thank you to INTERTANKO, and to LCS members, particularly the management board: www.lowcarbonshipping.co.uk
Questions - Are we looking at the right variables? - What is the right mix of technical and commercial? - What other analysis using this data would be interesting? - Can you align what s useful for your commercial purposes to the MRV debate? - How are energy efficiency measurements best shared: - Within an organisation? - With other stakeholders?
Extra details
Basis for estimating FC op Power out (P me x %MCR) Power out (P ae x %MCR) PC Power Required (speed, payload) 1. Estimate power required in design specification (Holtrop & Mennen) 2. Estimate power required in given state (speed, payload, fouling/deterioration, weather) 3. Apply delta to installed power and design %MCR 4. Calculate new %MCR and corresponding SFOC 5. Calculate fuel consumption in given state
Bottom up estimates - Information required: What is the annual transport work done (tnm)? What is the annual fuel consumption (t/pa)? Operational - Time in ballast/ loaded - What speed(s)? - How much payload is carried? Technical - Fuel consumption in design condition - Off design (draught, speed) effects - Weather (wind, waves, currents) - Hull fouling and engine wear - Auxiliary load
Design condition assumption P me MCR% x P me V d V max Assumes values quoted in IMO 2 nd GHG for design MCR% Could use TPD, but no transparency
Estimating annual carbon emissions per ship Total across all operating states i Main (propulsion) Aux Time spent i C = (P me_i.sfc me_i.c f + P ae_i.sfc ae_i.c f ).D i.24 Power output of main engine Specific fuel consumption Fuel carbon factor
AIS Reported data Lat/lon Speed over ground Heading Course Port proximity Elapsed time between messages Infrequent Message data ETA Destination Draught Classify vessel state using static machine learning model (trained on vessel fixture data) In port/first message out of port Loitering In transit Align modelled network with reported port calls from AIS Time stamped O-D matrix for each vessel Remove anomalous states and resolve port/ loitering states to port locations Normalized Vessel Network Speed profile on each voyage Draught condition on a subset Aggregate network to 10 speed and draught states Aggregated operational profile per vessel - Speed/Draught/ Period
Deterioration and weather impacts 9% Hull and propeller fouling increase resistance Machinery wear can increase SFOC Coating, sea area, maintenance specific, all unknown Simplistic approach based on empirical data
Estimating fuel consumption in the design condition This Study Vessel details: IMO number, Built year, owner, flag Hull characteristics: L,B,T,Dwt,GT,TEU Engine characteristics: Installed power, make/model, SFOC, TPD