D5.1 Route network and performance of merchant vessels

Size: px

Start display at page:

Download "D5.1 Route network and performance of merchant vessels"

Maurice Mills
5 years ago
Views:

1 Route network and performance of merchant vessels Work Package: Deliverable type: WP5

1 SEVENTH FRAMEWORK PROGRAMME THEME 7 Transport IceWin Innovative Icebreaking Concepts for Winter Navigation Grant no D5.1 Route network and performance of merchant vessels Work Package: Deliverable type: WP5 Report Contractual date of delivery: Actual date of delivery: Author(s): Robin Berglund, VTT Arto Nokelainen, VTT

2 List of Beneficiaries Beneficiary Number * Beneficiary name Beneficiary short name Country Date enter project Date exit project 1 (coordinator) 2 Valtion Teknillinen Tutkimuskeskus Hama Investeeringud VTT FI 1 24 HI EST TML TML BEL Aker Arctic Technology AARC FI 1 24

3 Executive Summary This report describes the work done in WP5 regarding performance of merchant vessels and route network modelling. The main data source for the work is ship speed and position data obtained from the AIS network during winter 2010 in the Gulf of Finland. The ships are grouped into classes depending on their type, machine power and ice class. Their speed is measured as a function of ice conditions obtained from corresponding ice charts. The relative position of the ships and icebreakers are identified, and situations where ships proceed behind icebreakers or other ships (in a convoy) are separately analysed. The result is an estimate of how ice conditions, as measured by the average ice thickness, do affect the performance (speed) of individual ships. Also the need for icebreaker assistance can be estimated as a function of ice thickness. This data will then be used as input to the simulation model, where the effect of different ice conditions, agreement systems, and traffic densities, will be simulated. The modelling work will be continued in WP6 to achieve a realistic model of the ship traffic and assistance need as a function of winters of different severity. 3 of 27 pages

5 Contents Executive Summary...3 Terminology Route network AIS data collection and preprocessing Analysis of ship states Temporal grouping Spatial grouping Output for the simulation model Interface functions Examples Simulating different ice conditions of 27 pages

6 Terminology AIS IBNet nm SOG The Automatic Identification System (AIS) is a system used by ships and Vessel Traffic Services (VTS) principally for identification and locating vessels. AIS provides a means for ships to electronically exchange ship data including: identification, position, course and speed with other nearby ships and VTS stations. Traffic information system for icebreakers. Used operationally on board all Swedish and Finnish icebreakers in the Baltic Sea. Developed and maintained by VTT. Nautical mile (1.852 km) Speed over ground 6 of 27 pages

7 1 Route network The simulation will be based on a simplified route network. The route network is designed based on previous work (ICOMOB-project) and on actual routes used by the ships. This information is available from the AIS network which collects position and speed from ships. VTT has gathered this information into a database which is then used as the basis for the analysis of ship performance in ice. Fig. 1. Route network in the Gulf of Finland. The route legs are shown in red. The map also shows how the observations are mapped to the route legs, i.e. all observations within a polygon with violet bordersare mapped to the leg in the middle of the polygon. 7 of 27 pages

8 Fig. 2. Eastern part of the route network Fig. 3. Western part of the route network 8 of 27 pages

9 Table 1. Route network nodes Leg From To Comment Lat1 long1 Lat2 Long2 Dist In km in nm of 27 pages

10 Sillamäe Vysotsk The AIS tracks shown in the following figures illustrate how the actual ship movements have been during two days taken as examples. Also the corresponding ice situation is shown as a comparison and as an illustration of how the ship tracks change as a consequence of changing ice conditions. Fig. 4. Route network with AIS tracks superimposed. The tracks are taken from data during Black lines are passenger ships, red lines tankers and violet lines cargo ships. 10 of 27 pages

Ice chart 19 february 2010 showing ice concentration.

11 Fig. 5. Ice chart 19 February 2010 showing ice thickness. Thin ice is seen on the Southern part of the Gulf of Finland. Fig. 6. Ice chart 19 february 2010 showing ice concentration. The concentration is 90% or above in most parts of the Gulf of Finland. (Ice concentration tells the fraction of ice cover relative to total area) 11 of 27 pages

Lines are coloured according to the speed of the ship (blue is over 10 knots, red is

12 Fig. 7. Route network with AIS tracks superimposed. The tracks are taken from data during March Lines are coloured according to the speed of the ship (blue is over 10 knots, red is below 1 knot) Fig. 8. Ice chart 19 March 2010 showing ice thickness. Thin ice is now seen on the northern part of the Gulf of Finland 12 of 27 pages

1.1 AIS data collection and preprocessing In this project the AIS data is collected from a server maintained by the Finnish Traffic Administration.

13 1.1 AIS data collection and preprocessing In this project the AIS data is collected from a server maintained by the Finnish Traffic Administration. The raw data is stored in files that then are converted to an ASCII-format (Comma Separated Values). The analysis process is described in Fig 8. Ice charts Ice charts Ice charts AIS AIS interface DBMS Pre analysis Correlation with ice Transit matrix Simulation model Fig. 8. Data flow of analysis part The pre analysis consists of the following steps: Fig. 9. Preanalysis steps. The interpolated positions are calculated at fixed 10 minute intervals. 13 of 27 pages

Before aggregating and analysing the data further, each interpolated point is complemented with information about closest fixpoint, ship and icebreaker.

14 Before aggregating and analysing the data further, each interpolated point is complemented with information about closest fixpoint, ship and icebreaker. This calculation is performed in SQL query language. Fig. 10. Steps included in the analysis of the data The database consists of 5 tables of which the shippos10 contains the interpolated points at 10 minute intervals. Static ship information is stored in the tables REGISTEREDSHIP and AISTARGET. AISTARGET contains information gathered from AIS messages containing ship parameters such as length, breadth, ship type. REGISTEREDSHIP contains ship data gathered from the icebreaker system IBNet. This register is maintained by the Finnish and Swedish maritime authorities and contains up-to-date information of all ships that have visited Finnish or Swedish ports in wintertime up to end of May of 27 pages

15 shippos10 FK1 LegPoints LegPoint LegID LegPointLat LegPointLong SubareaID Subareas SubareaID Subareaname FK2,FK3,FK4 FK1 FK2 FK2 FK3,FK4 FK3,FK4 mmsi timestamp_utc latitude longitude sog cog head rot ibsource ibactivity length duration closestfixpoint closestfixpointdist secondstoclosestais minicethick meanicethick maxicethick iceconcentration degreeofridging closestibmmsi disttoclosestib closestibseen closestibsees iblat iblon ibsog ibcog ibhead ibrot ibsecstoclosestais inferredshiporibactivity closestshipmmsi disttoclosestship closestshipseen SOGdifftoShip ClosestLegPoint ClosestLegPointDist sogorig regionofinterest LegPoint at_rs_row at_rs_src rs_row rs_src I1 I2 I3 AISTARGET at_rs_row at_rs_src at_mmsi at_lastmsgtype at_targettype at_navigationstatus at_accuracy at_latitude at_longitude at_rot at_sog at_cog at_heading at_timestamp at_lat_binary at_long_binary at_timestamp1 at_year at_month at_day at_hour at_minute at_second at_localtime at_timestamp4 at_callsign at_imo at_name at_draught at_length at_width at_shiptype at_destination at_etatime at_eta_binary at_ref_binary at_timestamp5 at_state at_eta at_eta_precision at_dest_po_row at_dest_po_src at_plottarea at_automatched at_usermatched at_loadtime at_hidden at_hidetime at_deactivation_area at_deactivation_time at_sog_max at_source PK PK I4 I1 I2 I3 REGISTEREDSHIP rs_row rs_src rs_iris_no rs_new_row rs_new_src rs_name rs_callsign rs_lloydsid rs_mmsi rs_mkh_id rs_sjov_id rs_grt rs_nrt rs_dwt rs_length rs_breadth rs_draft rs_speed rs_machinepower rs_iceclass rs_build_year rs_ibnet_shiptype rs_typetext rs_nationality rs_classification_society rs_phone1 rs_phone2 rs_phone3 rs_fax rs_telex rs_captain1 rs_captain2 rs_owner rs_ownerphone rs_change_source rs_changetime rc_createtime rs_loadtime rs_verifystatus rs_ts rs_outtime rs_ow rs_ow_stdev Fig. 11. Database diagram showing essential tables in IceWinGOF database. The positions are stored in the shippos10 table which contains about 4.5 million records. 1.2 Analysis of ship states To determine performance of ships, their states with respect to other ships and icebreakers, have to be determined. In the analysis the following ship states are identified: Table 2 Ship states identified from the AIS data Name Description Basis for rule In port Ship in port or in the vicinity of the port. The speed and behaviour of the ship is more dependent on other factors than ice conditions - thus ship Experience. Extent of area is based on detailed regional statistics. 15 of 27 pages

16 speed etc. cannot be used for performance estimation. To make analysis simpler, the port area is extended to encompass areas where ship pilots etc. are taken on board. Assisted Under icebreaker assistance. An icebreaker is moving in front of the ship and breaks a channel in the ice for the ship. An icebreaker can assist several ships in a convoy. In the icebreaker reporting system IBNet the Finnish and Swedish icebreakers report when they are assisting and what ship they are assisting. This information is not available for icebreakers from other countries, however. The rules based on distances and relative speed and angle can be determined comparing IBnet state with AIS data. Towed After ship near (SH2) After ship far (SH4) Freely moving Waiting The ship is being towed by an icebreaker The ship moves behind another ship (not an icebreaker) seen in an angle of degrees at a distance less than 2 km The ship moves behind another ship (not an icebreaker) seen in an angle of degrees at a distance less than 4km but more than 2 km The ship has a speed of 1 knot or more and is not in any of the states listed above. The ship has a speed less than 1 knot and is freely moving or after another ship (not in port). The problem is to infer the reason for waiting: Of interest is to identify when the ship is waiting for icebreaker escort - other waitings (like waiting for berth Towing state can be determined based on distance and same speed. Rules are determined by comparing IBnet data with AIS data. Statistical data for ships moving in a channel Statistical data for ships moving in a channel Experience Rule of speed less than 1 knot is derived from IBnet data. The ice field may sometimes drift quite fast, therefore a ship stuck in ice may have a SOG > 0 knots.. 16 of 27 pages

17 Frequency IceWin allocation or engine malfunction etc. ) are not of interest here. When determining the limits for identifying a ship in assisted state, a comparison with IBNet data from the Bay of Bothnia has been done. When a ship has been towed (i.e. there is a towing cable between the icebreaker and the ship) the distance has been less than 0.21 km in 95% of the cases. For ships being assisted, the distance histogram is as follows: Histogram % % % % % Frequency Cumulative % % % Bin Fig. 12. Distribution of distance between icebreaker and ship for ships being marked as assisted in IBNet. Based on this kind of analysis, the following rules have been applied: A ship is being towed if the distance to the icebreaker is less than 0.21 km and the angle in which the ib is seen is degree and speed difference is knots. A ship is being assisted if the distance to closest ib is < 4 km, and the icebreaker is seen in an angle of +-20 degrees In addition, there are two more states identifying if the ship is moving after another ship After ship near (SH2) A ship sees another ship in front of itself at a distance < 2 km, and the other ship is seen in an angle of +-20 degrees 17 of 27 pages

18 After ship far (SH4) A ship sees another ship in front of itself at a distance < 4 km, and the other ship is seen in an angle of +-20 degrees Temporal grouping The data is aggregated in 10 day periods according to Table 3 Table 3 Periods used when aggregating the data Period Startdate Enddate Spatial grouping The observations are mapped to the route network legs using rule of closest routepoint. Usually a leg is assigned one legpoint, which is the midpoint of the route. Some legs have been assigned 3 legpoints (those in the middle of the Gulf of Finland) to achieve a better association of real traffic and the theoretical network. The areas are shown in Fig 1 to 3. To prevent association of points too far from the leg, an additional condition is that the point shall not be more distant that 20 km from the closest legpoint. The legs are also grouped into 9 subareas (North, Mid, South and for each of these West, Mid and East). This grouping can easily be changed by updating the LegPoints database table if needed. Another coarse grouping into regions can be done based on longitude limits as follows: The region is determined by the longitude of the midpoint of the route leg. The longitude limits are as follows: Region name Min longitude (deg) Max longitude (deg) Min (when the observations are divided into 3 bins of equal size) Max (when the observations are divided into 3 bins of equal size) 18 of 27 pages

19 West Mid East Output for the simulation model The purpose of the analysis is to provide the simulation model with realistic data regarding ship performance in ice and about the ice conditions in the Gulf of Finland. One challenge is that real data is exactly applicable only to the ice conditions during the season that occurred. Applicability to other seasons is then possible only by making a ship performance model. This work will be done as part of the Simulation model design (WP 6) Interface functions The simulation model is provided with the following functions implemented as look-up tables. 1) Ship speed as a function of ice conditions. The ship speed is expressed relative to the open water speed, as this is found to be a first approximation how to normalise the effect of ice on different ships [1] v = v r (G(s i ), h ice ) * v ow (s i ) (1) where v r = relative ship speed (speed relative to open water speed) G = parameter vector describing the ship group to which the individual ship belongs. This parameter vector consists of ship type (passenger ship, cargo ship or tanker), machine power class (< 5000 kw, kw, kw, kw and > kw), and iceclass (II, IC, IB, IA, IAS) s i = individual ship h ice = ice thickness class (0-10 cm, cm, cm, cm, cm and over 50 cm ) v ow = Open water speed of the ship determined from the AIS data by taking the average value of the ship speed when the ship is not in ice and moving faster than 5 knots. 19 of 27 pages

20 2) Estimate of probability of needing icebreaker assistance as a function of region and time period. The basis for the estimate is the ratio between distances in different states per total distance. As the data is processed to exactly 10 minute intervals, an estimate of the total distance is sum of the SOG -values. (The distance in nm during 10 minute interval is SOG/6 (distance = speed*time). The sum of these SOG:s divided by 6 would give the total distance travelled in nm). Thus this estimate tells that on average, during the period, the ship has been assisted a miles on the leg, where a is l * p and l is the total length the ship has travelled on that leg during the period p ass10 = p ass10 (G(s i ), R, T) (2) where p ass10 = ratio of icebreaking assistance distance as measured during winter G = parameter vector describing the ship group to which the individual ship belongs. This parameter vector consists of ship type (passenger ship, cargo ship or tanker), machine power class (< 5000 kw, kw, kw, kw and > kw), and iceclass (II, IC, IB, IA, IAS) s i = individual ship R =region (Western, Middle, Eastern part of the Gulf of Finland). The region is based on longitude limits. OR other suitable criteria. T = Time period. The winter is divided into 10 day periods (see Table 3) 3) Estimate of probability of needing convoy assistance as a function of region and time period. : p conv10 = p conv10 (G(s i ), R, T) (3) where p conv10 = Ratio: (icebreaking assistance or being after another ship at a distance less than 4 km) / total distance as measured during winter G = parameter vector describing the ship group to which the individual ship belongs. This parameter vector consists of ship type (passenger ship, cargo ship or tanker), machine power class (< 5000 kw, kw, kw, kw and > kw), and iceclass (II, IC, IB, IA, IAS) 20 of 27 pages

21 s i = individual ship R =region (Western, Middle, Eastern part of the Gulf of Finland). The region is based on longitude limits OR other suitable criteria. T = Time period. The winter is divided into 10 day periods (see Table 3) 4) Estimate of probability of needing icebreaker assistance as a function of ship group and ice thickness. Thus this estimate tells that on average, the ship has been assisted a miles, where a is l * p and l is the total length the ship has travelled in the given ice thickness p assice = p assice (G(s i ),h ice ) (4) where p assice = ratio of icebreaking assistance distance to total distance. G = parameter vector describing the ship group to which the individual ship belongs. This parameter vector consists of ship type (passenger ship, cargo ship or tanker), machine power class (< 5000 kw, kw, kw, kw and > kw), and iceclass (II, IC, IB, IA, IAS) h ice = ice thickness class (0-10 cm, cm, cm, cm, cm and over 50 cm ) 5) Estimate of probability of needing convoy assistance as a function of ship group and ice thickness: p convice = p convice (G(s i ),h ice ) (5) where p convice = Ratio: (icebreaking assistance or being after another ship at a distance less than 4 km) / total distance G = parameter vector describing the ship group to which the individual ship belongs. This parameter vector consists of ship type (passenger ship, cargo ship or tanker), machine power class (< 5000 kw, kw, kw, kw and > kw), and iceclass (II, IC, IB, IA, IAS) h ice = ice thickness class (0-10 cm, cm, cm, cm, cm and over 50 cm ) 21 of 27 pages

22 For validation of the model, the waiting times of the ships are also measured (6) To remove waiting times caused by other factors than ice condition dependent reasons, average waiting time values from ice free periods can be subtracted from the times measured during other times. This then assumes that the (average) waiting times can be expressed as a sum of waitings due to ice conditions and waitings due to other reasons. 6) Waiting time estimate : w = w10 (G(s i ), R, T) (6) w10 = waiting time ratio, i.e. waiting time per total time in open sea areas (ice or open water) G = parameter vector describing the ship group to which the individual ship belongs. This parameter vector consists of ship type (passenger ship, cargo ship or tanker), machine power class (< 5000 kw, kw, kw, kw and > kw), and iceclass (II, IC, IB, IA, IAS) s i = individual ship R =region (Western, Middle, Eastern part of the Gulf of Finland). The region is based on longitude limits OR other suitable criteria T = Time period. The winter is divided into 10 day periods (see Table 3) 7) Average ice conditions per route leg and period during winter 2010: h ice = h ice (L, T) (7) h ice = ice thickness class (0-10 cm, cm, cm, cm, cm and over 50 cm ) L = Leg number T= Time period. The winter is divided into 10 day periods (see Table 3) Examples In the following some examples of the analysed data is given 1) Relative ship speed as a function of ice conditions 22 of 27 pages

icebreaker or convoy assistance as a function of ship group and ice thickness. Fig. 14.

23 Fig. 13. Relative ship speed per iceclass and machine power as a function of ice thickness 2) Estimate of probability of not needing icebreaker or convoy assistance as a function of ship group and ice thickness. Fig. 14. Relative distance travelled freely moving depending on machine power and ice class as a function of ice thickness. 3) Estimate of probability of not needing icebreaker or convoy assistance as a function of ship group and ice thickness - only moving ships included. 23 of 27 pages

24 Fig. 15. Relative distance travelled freely moving depending on machine power and ice class as a function of ice thickness, sog > 1 4) Estimate of probability of needing icebreaker assistance as a function of ice class, machine power and ice thickness - only moving ships included. 24 of 27 pages

25 Fig. 16. Relativedistance travelled assisted depending on machine power and ice class as a function of ice thickness. 5) Estimate of probability of needing icebreaker assistance as a function of ship type, ice class, machine power and ice thickness - only moving ships included. Fig. 17. Relative distance travelled assisted depending on ship type, machine power and ice class as a function of ice thickness. 25 of 27 pages

26 1.3.3 Simulating different ice conditions The ideal way of simulating different ice conditions would be to measure performance during several winters of varying severity and then applying these values to the simulation model. In the lack of this kind of data, the approach here is to use different parts of the Gulf of Finland as measured during the winter 2010 as representative samples of the conditions during winters of varying severity. Thus applying the conditions in the Eastern part to the whole sea area would indicate conditions during a severe winter. The weakness of this approach is that the conditions in the Eastern part would not be changed during simulation of a severe winter, thus additional modelling has to be applied to achieve this. As the speed and assistance probabilities have been estimated as a function of ice thickness, a suggestion is to increase the average thickness on all legs by one step to simulate more severe conditions. The average thickness per leg during the most severe period in a severe winter could also be estimated using statistical data from the Ice Service regarding ice conditions in the Gulf of Finland during a severe winter. As part of the model design in WP6, modelling techniques such as regression analysis will be used to create a simplified model of ship performance as a function of ice conditions and ship parameters. This will enable improved estimation of ship performance also for ships that have very few observations when going in ice. 26 of 27 pages

27 References [1] Riska, K., Wilhelmson, M., Englund, K. and Leiviskä, T Performance of Merchant Vessels in Ice in the Baltic. Winter Navigation Research Board, report 52, Helsinki 27 of 27 pages

VINTERSJÖFARTSFORSKNING

STYRELSEN FÖR VINTERSJÖFARTSFORSKNING WINTER NAVIGATION RESEARCH BOARD Research Report No 84 Leena Vedenpää OBSERV OBSERVATIONS OF SHIP ICE PERFORMANCE IN THE BALTIC Winter 2012 Finnish Transport Safety