Risk calculation project

Transcription

1 Risk calculation project Jon-Arve Røyset Helsinki,

2 Probability system Establish a long-term data collection for the analysis of probability of ship accidents in Norwegian waters Trend analysis Uniform data AIS data Ship register Accidents DSS,Dama,SD Safe Seanet, PEC Agr. Geo boundaries VTS Data marts Pilot - DNV GL Veracity - Data mangement, data quality, harmonization,... Data marts Data marts Data marts Data export Reports Meteorology Web and BIpresentation Mitigation measures Supporting tables Spes. analysis Havbas e Environmental risk Risk module Other Existing and future models

3 Results and deliveries With the help of the system, one can easily provide overviews and reports in relation to: Ship activity and trends in Norwegian sea areas Change in ship activity Probability of accidents and oil spill Changes in likelihood of accidents and oil spill Overview of accident types Use of pilot related to accidents Everything presented through a common web interface Develop method report open risk

4 Risk level trends Stakeholders: NCA, other Norwegian authorities, public? Purpose: To monitor trends in risk level in Norwegian waters and report as appropriate. Goal: To be able to identify risk level trends as the basis for further analyzes and to take expedient actions if needed. [LIJ1]

5 KV.B, KV.S, NA, Public? To clearly highlight rate of change in the risk level based on set criteria To be able to easily generate risk maps with main changes - short/long term planning Green Increased risk Red Decreased risk Date1 Date 2 Vessel type Risk type Region Vessel type Risk type >= 1. < GT 4999 GT 9999 GT GT GT GT GT 01 Oljetankere 1 % 2 % 0 % 1 % 2 % 7 % 0 % 02 Kjemikalie- /produkttankere 1 % -9 % 3 % 7 % 1 % 0 % 0 % 03 Gasstankere 0 % 3 % 1 % 1 % 1 % 0 % 1 % 04 Bulkskip 1 % -3 % -2 % -7 % 1 % 2 % 0 % 05 Stykkgodsskip 1 % 7 % 6 % 1 % 0 % 0 % 0 % 06 Konteinerskip 0 % 0 % 4 % 1 % 0 % 0 % 0 % 07 Ro Ro last 0 % 2 % 2 % 0 % 0 % 0 % 0 % 08 Kjøle-/fryseskip 0 % 8 % 1 % 0 % 0 % 0 % 0 % 09 Passasjer 2 % 8 % 5 % 6 % 4 % 3 % 1 % 10 Offshore supply skip skip 1 % -5 % -7 % 0 % 0 % 0 % 0 % 11 Andre offshore service 3 % -1 % 1 % 2 % 0 % 0 % 0 % 12 Andre aktiviteter 2 % 1 % 3 % 1 % 0 % 0 % 0 % 13 Fiskefartøy -4 % -4 % 1 % 0 % 0 % 0 % 0 %

6 UC-1,9 KV.B, KV.S, NA, Public? To identify the high and low risk for spill areas in Norwegian waters To identify the high and low risk for spill areas in Norwegian waters Vessel type Fuel/Cargo type

7 UC-1,4 KV.B, KV.S, NA, Public? Monitor trends in reported accidents and present results in multiple ways In order to be able to identify trends and take expedient actions Number of accidents within each cell Vessel type Accident type

8 UC-1,6 KV.B, KV.S, NA, Public? Monitor trends in reported accidents with oil spill and present results in multiple ways Number of accidents with oil spill within each cell / Oil spill volume In order to be able to identify trends and take expedient actions No./Volume Vessel type Fuel/Cargo type

9 UC-1,5 KV.B, KV.S, NA, Public? Monitor trends in reported accidents with resulting injury or loss of life Number of accidents with resulting injury or loss of life for respective cell injuries/fatalities In order to be able to identify trends and take expedient actions Number of accidents with resulting injury or loss of life for region Oljetankere Kjemikalie-/produkttankere Gasstankere Bulkskip Stykkgodsskip Konteinerskip Ro Ro last Kjøle-/fryseskip Passasjer Number of accidents with resulting injury or loss of life for region injuries/fatalities region Offshore supply skip Andre offshore service skip Andre aktiviteter Fiskefartøy 2008 region 2040 Vessel type injuries/fatalities 7. >= 1. < GT 4999 GT 9999 GT GT GT GT GT 01 Oljetankere 1 % 2 % 0 % 1 % 2 % 7 % 0 % 02 Kjemikalie- /produkttankere 1 % -9 % 3 % 7 % 1 % 0 % 0 % 03 Gasstankere 0 % 3 % 1 % 1 % 1 % 0 % 1 % 04 Bulkskip 1 % -3 % -2 % -7 % 1 % 2 % 0 % 05 Stykkgodsskip 1 % 7 % 6 % 1 % 0 % 0 % 0 % 06 Konteinerskip 0 % 0 % 4 % 1 % 0 % 0 % 0 % 07 Ro Ro last 0 % 2 % 2 % 0 % 0 % 0 % 0 % 08 Kjøle-/fryseskip 0 % 8 % 1 % 0 % 0 % 0 % 0 % 09 Passasjer 2 % 8 % 5 % 6 % 4 % 3 % 1 % Offshore supply skip skip 1 % -5 % -7 % 0 % 0 % 0 % 0 % 11 Andre offshore service 3 % -1 % 1 % 2 % 0 % 0 % 0 % 12 Andre aktiviteter 2 % 1 % 3 % 1 % 0 % 0 % 0 % 13 Fiskefartøy -4 % -4 % 1 % 0 % 0 % 0 % 0 %

10 UC-1,7 KV.B, KV.S, NA, Public? Identify the reason for accidents based on ship movements and immediate actions In order to be able to identify trends and take expedient actions Number of accidents per cause Ship losses and causes Grounding Fire Collision Foundering Ice damage region 2040 Vessel type Cause of accident Loss type

11 UC-1,11 KV.B, KV.S, NA, Public? To establish an overview of the use of pilots/farledsbevis connect to voyage In order to be able to identify trends and take expedient actions Without pilot With pilot Vessel type With Pilot

12 12 High level risk methodology and focus areas

13 Single ship calculations The risk model shall be location-based, i.e. the calculations are carried out individually for each position, such that the results can be defined as functions of position Based on the recorded position of AIS messages a GIS will be used to draw ship tracks illustrating traffic patterns. Ship tracks are lines drawn between AIS points recorded for each vessel based on the route the vessel has sailed, as shown in the Figure below, and later aggregated within the grid cells for use in the risk calculations. The model aims to operate on a single ship at a time.

14 14 What have we done so far Work load rating 1 to These risk model is influenced by the methodology used in DNV GL s NavRisk tool, DNV GL s FARGE project, IALA s Waterway Risk Assessment (IWRAP tool) and Be-Aware methodology. The essentials of these models have been used to develop the new risk model, but with new innovation and calibration. Innovation here we want to focus our efforts! Powered grounding model DNV GL (COWI) 6 Drift grounding model 3 Collision model 6 COWI DNV GL Approach: Critical situations Approach: Drift based on metocean data Approach: Critical situations (safety domain/ellipse) DNV GL Fire/explosion and foundering model 1 Consequence model Oil Consequence model outflow 3 Loss of lives 3 DNV GL DNV GL Approach: NavRisk (sailed distance) Approach: NavRisk + updates Approach: NavRisk + updates -- Contact model?

15 The overall process of calculating risk The calculation process will be fully automatic, with no manual input for execution. Manual input should only be needed for tool development and updates of parameters Accident frequencies are calculated for the following types of accidents; grounding (driftand powered grounding), collision (head-on, overtaking and crossing), fire/explosion and foundering Figure 6-1 High-level illustration of the risk calculation process

16 16 Powered grounding model

17 General modelling approach for powered grounding Simplified calculation formula: Number of critical situations x probability of grounding, given that ship is in critical situation (course not changed before impact) This calculation will primarily be based on the IWRAP method, but the calculation will be done for individual ships, not merged traffic in lanes (which is the case for IWRAP)

18 18 Modelling principle Requirement: Automatic calculations, no manual input of legs, waypoints etc. Powered grounding frequency = Sum of N (Number of critical situations) x Pc (Causation probability) Pc models the vessels and the officer of the watch s ability to perform evasive manoeuvres in the event of potential critical situation.

19 19 Number of critical situations Powered grounding frequency = Sum of N (Number of critical situations) x Pc (Causation probability) Type 1 Vessel do not turn - Watch Officer asleep - Technical (rudder/ steering gear) Type 2 Vessel deviation from route - Watch Officer misjudgement (complexity, time etc.) - Current, waves etc. - Evasive manoeuvres to avoid other ship

20 20 Number of critical situations - Type 1 Downscaled AIS high resolution data 30 s Due to noise in AIS data Points with low speed to be removed to exclude harbor turnings Half year of 2015 data General cargo carriers Critical turns Turns with Rate of Turn (RoT) > 0.1 o (1/radius of curve) 30 min vector at turns Check if the vector ends to land / shallow water Where ships turn to avoid hitting land Powered grounding frequency = Sum of N (Number of critical situations) x Pc (Causation probability) Process: 1. One ship analysed in the first stage 2. Scaling up to all vessels in the test area 3. Scaling to Norwegian coast and all vessels o Challenge with large amount of data r 1 r = 0

21 11 21 December 2013 Drift grounding

22 22 Drift grounding model Leading indicators and input: Distance sailed. Engine breakdown, leading to drift grounding, may be closely related to time at sea or distance sailed. Calculations; the drift grounding calculations should take into consideration: Distance to shore. Grounding frequency at sea must reflect the distance between the ship s route and the all potential grounding points. Location and capacity of emergency tug services Metocean data (wind roses, currents) Simplified calculation formula: Number of distance sailed (or operation time) x probability of engine malfunction x probability of ship drifts to shore (considering failure to recover ship, failure of tugs etc.) This calculation will primarily be based on the DNV GL FARGE [1] method and the IWRAP method [2], and the calculation will be done for individual ships.

23 23 General modelling approach for drift grounding. Should the methodology development for grounding fail due to unforeseen reasons (data extensiveness, complexity of modelling, time constraints etc.), the risk model will use one of these two alternatives (back-up methods): Alternative 1: Utilise the DNV GL NavRisk methodology for grounding. The above method for identifying number of critical situations (dangerous courses), in combination with distance to shore, will be used as input to the adjustment factor for the simplified calculation formula: Number of nautical miles sailed x probability of grounding per nautical mile x adjustment factor. Alternative 2: Use DNV GL NavRisk methodology for grounding in its current form, with no changes. The script for this method is existing, thus it can be implemented with minimum efforts.

24 24 Collision model

25 25

26 26 Collision model Collision frequency = Frequency of critical situations * causation factor a dependent probability of an accident given a critical situation x y, Forward B Model utilized AIS tracks directly oa L Vessels are modelled as ellipses

27 27 Critical situations A critical situation as two vessels being within 0.2 nautical miles of each other, and matching one of three criteria 0.2 nautical miles can be discussed 1) Their course difference being equal to overtaking 2) Their course difference being equal a meeting situation. 3) a crossing situation a. the two ship tracks intersect; or b. extrapolation of the two ships current position and heading forward in time leads to an intersection 1) 2) 3)

28 28 Method description summery

29 Method description Acute pollution and loss of life