Dynamic positioning in ice for offshore vessels: Results from the Arctic DP Project

Transcription

1 Dynamic positioning in ice for offshore vessels: Results from the Arctic DP Project 2 nd Annual Arctic Marine Logistics & Infrastructure Forum Amsterdam, Tuesday 26 th November Illustration: Bjarne Stenberg; Copyright: NTNU Prof. Roger Skjetne Department of Marine Technology Norwegian University of Science and Technology

2 A RCN sponsored KMB project January 2010 December 2014 RCN KMB project Arctic DP: Safe and green dynamic positioning operations of offshore vessels in an arctic environment

3 Principal researchers at NTNU Roger Skjetne Professor Key researcher on marine control engineering. Project manager. Dept. Marine Technology, NTNU Lars Imsland Professor Key researcher on control system theory. Dept. Engineering Cybernetics, NTNU Sveinung Løset Professor Key researcher on ice mechanics and Arctic technology. Dept. Civil and Transport Engineering, NTNU Biao Su Post-doc Key researcher on maneuverability of ships in ice. Project assistant. Dept. Marine Technology, NTNU

4 Industrial points of contact Nils Albert Jenssen Dr.Ing. Vice President Business Development Kongsberg Maritime AS Glenn-Ole Kaasa Stian Ruud Dr.Ing. Principal Researcher Statoil Research Center Porsgrunn Statoil ASA Cand. Real and NTNU Ph.d. candidate Principle Engineer Det Norske Veritas AS

5 ACEX: IODP expedition 302

6 ACEX: IODP expedition 302 Video courtesy: Captured by IODP ACEX mission. Digitized by AKAC Inc. (Evan Martin)

7 Open water DP

8 3 DOF DP model by T. I. Fossen (2011) Position and heading:...and velocities: Kinematic equation: Kinetic equation: Current forces: Wind and wave forces assumed additive:

9 Current compensation: Feedback integral action We don t really measure º r Simplification for control design: Slowly varying bias force due to currents. Compensated well by integral action in a PID feedback control law: From the integral action on the bias b(t), a «current speed and direction» can be deduced. However, it is frequently referred to as the «DP current» since it really accounts for: Ocean currents. Wave drift forces. Unmodeled wind force dynamics. Unmodeled thruster dynamics. Other unmodeled dynamics.

10 Wind compensation: Feedforward cancellation Assume a static relationship between wind velocity and force To compensate wind force, use feedforward: Measure the wind speed and direction in the mast and assume uniform wind. Calculate corrected wind coefficients based on sensors location above sea level. Calculate an estimate of the wind force/moment: Compensate by directly opposing the wind:

11 Wave compensation: Integral action and filtering Fossen (2011): 1st-order wave-induced loads: wave frequency (WF) motion observed as zero mean oscillatory motions. 2nd-order wave-induced loads: wave drift forces observed as non-zero slowly-varying components. This motivates: Wave forces are highly periodic and described according to a wave spectrum: Courtesy: Fossen (2011) Courtesy: Fossen (2011)

12 Wave compensation: Integral action and filtering For control design it is common to assume that the 1st order wave loads only produce an oscillatory zero-mean wave motion Total ship motion: Slowly-varying wave drift forces incorporated into the bias, yielding: Courtesy: Fossen (2011) Compensation of 2nd-order wave loads: INTEGRAL ACTION Handling of 1st-order wave loads: FILTERING

13 Wave compensation: Integral action and filtering Not desired to let the wave motion enter the feedback loop, thus requiring filtering. Used for PID feedback: Frequency response: or alternatively: Courtesy: Fossen (2011)

14 Review: Open water DP Use thrusters to compensate environmental forces: Typically PID-like control. Current: Feedback by PID, especially integral action (I). Waves: 1st order wave-induced loads: Filtering to avoid that enter the feedback loop. 2nd order wave-induced loads: Feedback by PID. Wind: Feedforward compensation from measurements of wind speed and direction. REACTIVE CONTROL STRATEGIES

15 Arctic DP Ice loads Ice

16 Effective stationkeeping in ice 1. Effective ship design. 2. An effective Ice Management system. 3. Effective strategies for the DP control system to compensate ice forces.

17 (Courtesy: Berg, T. E. et al, Design considerations for an Arctic intervention vessel, OTC Effective ship design

18 1. Effective ship design Ice-going capability assessment and DP-Ice Capability Plot for a double acting intervention vessel in level ice Biao Su (NTNU), Øivind K. Kjerstad (NTNU), Roger Skjetne (NTNU), and Tor E. Berg (MARINTEK) The CIVArctic vessel (Source: Mainly designed for open-water operations Ice-breaking capabilities with the stern Numerical analysis of the ice-breaking capability and maneuverability for the CIVArctic vessel. Comparison with the ice model tests carried out in the Aker Arctic ice tank in May A static DP-Ice Capability Plot by coupling the numerical simulator (towing simulations) and a thrust allocation algorithm. Collaboration with KMB CIV Arctic: Construction and intervention vessels for Arctic oil and gas.

19 Collaboration with CIV Arctic: A RCN sponsored KMB project July 2008 December 2011 Construction and intervention vessels for Arctic oil and gas Design for open water modification for ice: Hull shape to optimally serve the dual purpose. Moonpool studies. Deck areas and utility systems. Environmental footprint. Open water model tests: Calm water performance. Seakeeping and stationkeeping. Ice tank tests: Ice performance. Maneuvering. DP force. Calculations: Calm water performance. Parametric study. Operability in Arctic waters.

20 Constrained Nullspace-Based Thrust Allocation for Heading Prioritized Stationkeeping of Offshore Vessels in Ice Øivind K. Kjerstad (NTNU), Roger Skjetne (NTNU), & Bjørn O. Berge (MARINTEK) We have derived a thrust allocation algorithm with strict heading priority to ensure vessel orientation against the ice drift: The motivation is to employ thrust in a way such that the loads can be maintained. Yaw-Surge-Sway prioritization order is assumed to be favorable due to the projection of the hull and its interaction geometry. Due to the nature of ice loads they might temporally peak such that a loss of capability occurs: Quantification of the ice loads w.r.t time caters for evaluation of the capability. Investigates the capability of the CIVArctic vessel through a set of broken-ice towing data. Allocation verification on a ice load time-series with the two cases; all actuators and just azimuths. The time-series are normalized.

21 DP-Ice Capability Plot for the CIVArctic vessel DP-Ice Capability Plot parameterized by ice drift speed The simulation results reveal that the DP-Ie capability of the CIVArctic vessel in level ice is restricted by a narrow relative ice drift direction band. DP-Ice Capability Plot parameterized by ice thickness The main reason is that the hull shoulder (middle body) becomes exposed to the ice when the relative drift direction is in the range ( ). As there is no vessel dynamics in a towing experiment, the hull shoulder where the hull surface is vertical can not break the ice by bending, and is rendered to continuously crush the ice.

22 What ice load is dimensioning? The quantification levels defined was Peak load Significant load Mean load Kjerstad, Ø. K., R. Skjetne, and B. O. Berge (2013), Constrained Nullspace-Based Thrust Allocation for Heading Prioritized Stationkeeping of Offshore Vessels in Ice. In proc. 22nd Int. Conf. Port and Ocean Eng. under Arctic Conditions (POAC 2013).

23 Analysis CIV Arctic towing data CIVArctic vessel investigation Towing loads from Aker Arctic captured in a broken ice scenario Increasing concentration, thickness, relative heading and available thruster configuration affects the capability of the vessel in ice Prioritization of the DOF is clearly visible Kjerstad, Ø. K., R. Skjetne, and B. O. Berge (2013), Constrained Nullspace-Based Thrust Allocation for Heading Prioritized Stationkeeping of Offshore Vessels in Ice. In proc. 22nd Int. Conf. Port and Ocean Eng. under Arctic Conditions (POAC 2013).

24 2. An effective Ice Management system Ice Management is the sum of all activities where the objective is to reduce or avoid actions from any kind of ice feature. Essential for DP operations in ice Sea ice management Iceberg management Sea ice observation and monitoring: Ice concentration, floe size distribution, etc. Ice geometry, ice age, density, salinity, etc. Sea ice tracking (drift speed and direction, ice floe positioning) Ridge and iceberg detection and tracking. Threat evaluation (forecasting, threat assessment). Physical ice management. Procedures and training. Iceberg detection and tracking. Threat evaluation (forecasting, threat assessment). Iceberg handling (towing, etc.) Procedures and training.

25 Regional ice surveillance Iceberg detection and tracking Ice Management Ad-hoc communication network Ice breaker II Football field size ice floes Ice breaker I County size ice floes Protected vessel Ice breaker III Cuts ice into small pieces Ice drift Online ice drift monitoring units Unmanned aerial utility vehicle Local and regional aerial ice observation Local underwater ice observation

26 3. Effective Arctic DP control strategies Reactive control and/or Proactive control?

27 3. Effective Arctic DP control strategies Reactive control strategies: A. Feedback from pos-ref and gyrocompasses, using an ice characteristic model for integral action. B. Feedforward cancellation of ice loads by (acceleration) measurements. Proactive control strategy: C. Feedforward using a high-fidelity predictive system (with inputs from an ice observation system).

28 3. Effective Arctic DP control strategies Reactive control strategies: A. Feedback from pos-ref and gyrocompasses, using an ice characteristic model integral action. B. Feedforward cancellation of ice loads by (acceleration) measurements. Proactive control strategy: C. Feedforward using a high-fidelity predictive system (with inputs from an ice observation system).

29 A) Try 1: Compensation by integral action and filtering (as for current and waves) Requires no new instrumentation, just redesign/retune the internal SW model. An observer (Kalman filter) is a lowpass state estimator that theoretically eliminates the phase lag by using both the input control vector, the output measurement, and a model to predict the «near future» behavior. Adapted from: Røset, E. S. (2009) Successful for waves having an oscillatory (periodic) nature, thus being modeled well by a linear model only driven by white noise (without physical inputs). However, for highly nonlinear and highly varying ice loads, we should ideally have available an accurate dynamic model together with measurements of the driving forces. Using a simplified linear ice model driven by white noise, as for waves, may yield reasonable results in normal conditions, but may fail in adverse conditions. Since in this setup the ice loads are «observed» through position and velocity motion measurements, it is likely that for large and rapid steps in ice forces: the vessel will gain momentum and start to move, an «unrealistic» observer will not estimate the load steps fast enough, and the feedback control system must use unnecessary energy on stopping the motion before correcting the position offsets. Jenssen, N.A., Muddesitti, S., Phillips, D., and Backstrom, K. (2009). Dp in ice conditions, in proc. Dynamic Positioning Conf., Houston, USA, Oct Hals, T., and Efraimsson, F. (2010). DP Ice Model Test of Arctic Drillship, in proc. Dynamic Positioning Conf., Houston, USA, Oct

30 A) Try 2: Compensation by replacing the integral action with an Ice Load characteristics model - using existing sensors When broken managed sea ice and a stationary ship interacts, a significant spatial footprint in the surrounding upstream ice cover is created as the ice floes are forced around the obstructing vessel. Generally, the ice load can be separated into three load components: Loads related to the interaction and around-hull transport of the affected and accumulated ice masses. This component is slowly varying and mainly dependent on the oblique angle and ice concentration. Loads associated with abnormal spatial and temporal events in the affected mass. This can occur due to force chains, unfortunate floe configurations, or abnormal ice features such as large floes, multiyear ice, or iceberg remnants. Loads related to compaction and pressure in the ice cover. This occurs either due to large scale environmental effects or affected mass boundary interaction.

31 3. Effective Arctic DP control strategies Reactive control strategies: A. Feedback from pos-ref and gyrocompasses, using an ice characteristic model integral action. B. Feedforward cancellation of ice loads by (acceleration) measurements. Proactive control strategy: C. Feedforward using a high-fidelity predictive system (with inputs from an ice observation system).

32 B) Compensation by feedforward compensation from acceleration measurements Since force and acceleration are proportional, the idea is to use accelerometers to measure and estimate the external forces acting on the vessel. Multiplying the acceleration with the mass and subtracting the control input gives (conceptually) the external resulting force directly by measurements. Practical challenges to be overcome: measurement noise, alignment errors due to mounting, g-compensation, scaling errors, and slowly drifting bias. Kongsberg Maritime MRU 5+ Includes three axes high quality MEMS rate gyros and linear accelerometers. Kjerstad, Ø. K., Skjetne, R., and Jenssen, N. A. (2011), Disturbance rejection in dynamic systems by use of acceleration feedforward: Application to dynamic positioning, in Proc. 18 th IFAC World Congress, Milano, Italy, Aug. Sept Kjerstad, Ø. K. and R. Skjetne (2012), Observer design with disturbance rejection by acceleration feedforward. In proc. 7th IFAC Symposium on Robust Control Design (Rocond'12), Aalborg, Denmark, June 20-22, Nguyen, D. T., A. H. Sørbø and A. J. Sørensen (2009). Modelling and Control for Dynamic Positioned Vessels in Level Ice. In Proceedings of 8th Conference on Manoeuvring and Control of Marine Craft (MCMC 2009), pp , September 16-18, Guarujá, Brazil.

33 AFF challenges Separating linear accelerations, rotational accelerations, and rotational rate contributions in the measurements. Important for reconstructing the full acceleration vector. Handling accelerometer errors and S/N. Redundancy and safety w.r.t feedforward compensation. Consider for instance a typical IMU acceleration measurement: Suppose alignment, measurement separation, gravity, and noise are handled and compensated: Then in the closed-loop system dynamics the rapidly changing ice disturbance force will be replaced by a nice time-varying accelerator bias: MRU 5+ motion reference unit. Includes three axes high quality MEMS rate gyros and linear accelerometers. Courtesy: Kongsberg Maritime Kjerstad, Ø. K., Skjetne, R., and Jenssen, N. A. (2011), Disturbance rejection in dynamic systems by use of acceleration feedforward: Application to dynamic positioning, in Proc. 18 th IFAC World Congress, Milano, Italy, Aug. Sept

34 AFF experimental platform on R/V Gunnerus MRU 5+ motion reference unit. Includes three axes high quality MEMS rate gyros and linear accelerometers. Courtesy: Kongsberg Maritime

35 The basic idea of the AFF regulator Consider the mass-damper system: In nominal case: typical PID-like regulator: Assume disturbance is globally Lipschitz: Acceleration feedforward: Ideally: Feedforward: Closed loop: Kjerstad, Ø. K., Skjetne, R., and Jenssen, N. A. (2011), Disturbance rejection in dynamic systems by use of acceleration feedforward: Application to dynamic positioning, in Proc. 18 th IFAC World Congress, Milano, Italy, Aug. Sept

36 The basic idea of the AFF DP observer DP vessel: Rewrite to: An observer with AFF: Closed-loop system: Kjerstad, Ø. K. and R. Skjetne (2012), Observer design with disturbance rejection by acceleration feedforward. In proc. 7th IFAC Symposium on Robust Control Design (Rocond'12), Aalborg, Denmark, June 20-22, 2012.

37 NOTE! Reactive ice compensation needs Ice Management! Reactive ice compensation: Based on instantaneous or slightly delayed sensing. This implies: Ice loads that accumulate cannot grow beyond the maximum thruster/mooring capacity. Ice loads cannot grow faster than thrust rate limit. Thus: Physical Ice Management with guaranteed success of protecting the Protected Vessel is of utmost importance.

38 3. Effective Arctic DP control strategies Reactive control strategies: A. Feedback from pos-ref and gyrocompasses, using an ice characteristic model integral action. B. Feedforward cancellation of ice loads by (acceleration) measurements. Proactive control strategy: C. Feedforward using a high-fidelity predictive system (with inputs from an ice observation system).

39 C) Compensation by feedforward from a supervisory control system Suppose we let the DP system know in advance what ice loads to encounter. DP system can then plan its motion for optimal handling of ice floe impacts while doing adequate stationkeeping. Vidar Viking stationkeeping maneuvers. One can then use high-fidelity numerical ice-ship simulators to predict the future load variations. The predicted loads can be used in a feedforward compensation strategy. A supervisory predictive system requires full overview of the ice field, including detailed high-fidelity knowledge of the local ice properties.

40 IM: Top-level feedback system IM Supervisory System IM ILRS Ice breakers ++ DP ctrl Arctic ice-infested environment Wind Waves Current Thr.sys DP Vessel Sensors Ice observation system: Mobile sensor network

41 Numerical simulation of DP in ice - the work by Ivan Metrikin and Sveinung Løset. Metrikin, I., Løset, S., Jenssen, N.A., and Kerkeni, S. (2012). Numerical Simulation of Dynamic Positioning in Ice, DP Conf., Houston, USA, Oct Metrikin, I., Lu, W., Lubbad, R., Løset, S. and Kashafutdinov, M. (2012). Numerical Simulation of a Floater in a Broken-Ice Field Part I: Model Description. OMAE2012, Rio de Janeiro, Brazil, July 1-6. Metrikin, I., Borzov, A., Lubbad, R., Løset, S. (2012). Numerical Simulation of a Floater in a Broken-Ice Field Part II: Comparative Study of Physics Engines. OMAE2012, Rio de Janeiro, Brazil, July 1-6.

42 Ice intelligence: A better instrumented Arctics! Courtesy: Met.no Illustration: Bjarne Stenberg

43 Ice imagery for Ice Management Thresholding Greyscale image Pixel frequency vs. greyscale value. Ice concentration: 71.43%

44 Sobel Boundary detection Prewitt Canny IC = 65.81% Number of floes: 1732 IC = 65.04% Number of floes: 1647 IC = 62.25% Number of floes: 1392

45 Ice floe identification by snake algorithm The initial contour should approach the floe boundary, and its center should be located close to the floe center. Initial contours Result

46 Experiments Overall ice tank image Ice concentration: 76.96%

47 Ice floe size distribution

48 Ice floe identification Rectangularization: the minimum area bounding rectangle for each ice floe is found to complete them. Ice concentration: 83.23% The rectangularized floes will be more or less larger than the actual segmented floes and some floes will overlap

49 Test on HSVA video Problem: Light refection effect Uniform parameters

50 Test on HSVA video Maximum floe size entering the protected vessel

51 Sea ice image Non uniform illumination, shadows, impurities; variance and irregular of floe size and shape. The initial contour with circle shape sometimes could not approach to the true floe boundary.

52 Sea ice image Sea ice image

53 Sea ice image Binary image (thresholding) Dark floes are missed

54 Sea ice image Binary image (k-means) More floes are found

55 Sea ice image Ice floe Gray scale Brash K-means GVF Snake Slush Open water (identify light & dark floes, initial contours)

56 Sea ice image Floe & Brash

57 Sea ice image Ice floe size distribution

58 Underwater ice observation system Courtesy: Autonomous Undersea Vehicle Applications Center Jørgensen, U. and Skjetne, R. (2012). Autonomous Estimation of Drifting Ice Topography Using Continuous Steepest Descent Gradient Minimization, in Proc. American Control Conf., AACC, Montréal, Canada, June Jørgensen, U. and R. Skjetne (2013), Dynamic estimation of a drifting ice topography over a 2D area using mobile underwater measurements. In proc. 22nd Int. Conf. Port and Ocean Eng. under Arctic Conditions (POAC 2013).

59 Underwater sea-ice topography observation Underwater monitoring of sea ice Guidance algorithms for determining the optimal path during online monitoring of drifting ice. Guidance algorithms. Detection and tracking of ice features like ridges and icebergs. Collision avoidance. Courtesy: NASA Integration of the guidance problem with a higher-level sea-ice estimator. 59

60 Underwater iceberg monitoring Underwater iceberg survey Map underwater iceberg geometries by using an AUV. Forecast iceberg drift in ocean currents. Calculate roll-stability of the iceberg for safer towing operation. Courtesy: National Geographic Online monitoring of the iceberg stability during towing operations. Courtesy: Kongsberg Maritime 60

61 Real seabed topography example: Measurements of a shipwreck site close to Ormen Lange. Measured with DeltaT multi-beam echo (MBE) sounder placed on an ROV.

62 Experiences gained Operation from Oden: USBL transducer deployed from side of Oden. 3 meters below water due to cable length. Possible interference from Oden hull. The position and heading from Oden could have been used to obtain navigation-reference on screen. Dedicated personnel for ROV operation and sonar/navigation computer operation. Software crash on sonar/navigation computer lead to problems. Courtesy: Roy Wollvik Highly affected by large ocean currents. 62

63 Experiences gained Operation from remote iceberg-station: Cancelled in the last minute due to ice conditions. Transport and setup of all equipment on ice complicates operation. Hoisting of equipment to ice floe. Not all equipment was waterproof. Grounding of generator in water. Deployment of transducer. Deployment and recovery may be risky. Deployment by cutting hole in ice is time consuming due to size of vehicle. Deployment/recovery from ice-edge can be risky due to possible poor ice conditions at ice-edge. Courtesy: Per Frejvall 63

64 Experiences gained Operation from workboat: Limited space for equipment and personnel. Need a free-floating iceberg due to risk of getting stuck in ice with workboat. Would increase probability of getting good data since it is possible to maneuver relatively fast around the iceberg. 64

65 Real seabed topography example: Measurements of a shipwreck site close to Ormen Lange. Measured with DeltaT multi-beam echo (MBE) sounder placed on an ROV. Raw data from MBE Each point (x,y,z) is the depth measured by one single ray from the MBE The final product Used as input in simulation Processed data: Rotated Discretized into a rectangular grid

66 Dynamic Estimation of Drifting Ice Topography Over a 2D Area Using Underwater Measurements Ulrik Jørgensen (NTNU) & Roger Skjetne (NTNU) Objective: Generate an estimate of the ice topography by using underwater measurements Initial state of simulation Stems are current measurements Top layer is estimated topography Bottom layer is the true topography Final state of simulation Stems are current measurements

67 3D topographic model: ϕ xy ( sc, ) ( kκ ( ) ( )) 0 t ξ + κ0 y t ξ p q A cos s () l s () = l= 0 k= 0 + B sin s l s kl x x x y y ( kκ ( ) ( )) 0 ( t) ξ + κ0 y ( t) ξ kl x x x y y c = 0x { A, B } kl 0 y kl : Actual unknown coefficients T st ( ) = sx s y : unknown displacement κ, κ : known fundamental wavenumbers in x and y direction ξ ( x, y ) ( x y ) { } 1 1,,, : known points of measurements n m Jørgensen, U. and R. Skjetne (2013), Dynamic estimation of a drifting ice topography over a 2D area using mobile underwater measurements. In proc. 22nd Int. Conf. Port and Ocean Eng. under Arctic Conditions (POAC 2013).

68 2D case: Real topography Simulation: SCICEX-99 dataseries:

69 Concluding remarks on reactive control We have chosen to pursue several paths towards the goal: A. Development of an Ice Characteristics control model: A simple Bias model facilitates integral action (adaptive feedback), but may be too simple (assumption of constant bias is too invalid). A further developed Ice Load characteristics, e.g. an Ice Accumulation load model, may need some other state measurements/estimates and facilitate adaptive parameter estimation. Other techniques such as parameter resetting may be necessary to account for abrupt changes in the ice loads. B. Feedforward compensation by use of load measurements: Acceleration measurements is in the ideal case a very good candidate for load disturbance rejection. Effective disturbance rejection for both feedback control law and observer algorithms. Challenges related to signal-to-noise level, alignment errors, gravity compensation, bias drift, and sensor faults.

70 Concluding remarks on proactive control We want to measure and estimate: ice concentration, floe size distribution, ice floe identification (position, area, shape approximation), ice thickness (through below- and above-surface topographies), ice drift velocity, ++ over the local area of DP operation, as it varies with time. Then high-fidelity numerical simulator tools can be applied to forecast the incoming ice loads in the near future. Can be used to: 1. Prepare and set up compensating thrust as a feedforward strategy. 2. Guide the DP vessel actively through the incoming ice stream within allowable operating circle, to minimize loads. A Supervisory (Ice Management) control system will also facilitate a better instrumented operations center, with decision support tools based on an advanced Ice Intelligence Systems and numerical simulation tools.

71 References Metrikin, I., Løset, S., Jenssen, N.A., and Kerkeni, S. (2012). Numerical Simulation of Dynamic Positioning in Ice, DP Conference, Houston, USA, Oct Scibilia, F., Jørgensen, U. and Skjetne, R. (2012). AUV Guidance System for Subsurface Ice Intelligence in Proc. 31st International Conference on Ocean, Offshore and Arctic Engineering (OMAE), Rio de Janeiro, Brazil, July Metrikin, I., Lu, W., Lubbad, R., Løset, S. and Kashafutdinov, M. (2012). Numerical Simulation of a Floater in a Broken-Ice Field Part I: Model Description. Proceedings of the ASME st International Conference on Ocean, Offshore and Arctic Engineering (OMAE2012), Rio de Janeiro, Brazil, July Paper number OMAE Metrikin, I., Borzov, A., Lubbad, R., Løset, S. (2012). Numerical Simulation of a Floater in a Broken-Ice Field Part II: Comparative Study of Physics Engines. Proceedings of the ASME st International Conference on Ocean, Offshore and Arctic Engineering (OMAE2012), Rio de Janeiro, Brazil, July Paper number OMAE Zhang, Q., van der Werff, S., Metrikin, I., Løset, S., Skjetne, R. (2012). Image Processing for the Analysis of an Evolving Broken-Ice Field in Model Testing. Proceedings of the ASME st International Conference on Ocean, Offshore and Arctic Engineering (OMAE2012), Rio de Janeiro, Brazil, July Paper number OMAE Zhang, Q., R. Skjetne, S. Løset, and A. Marchenko (2012). Digital image processing for sea ice observations in support to Arctic DP operations. In proc. 31st Int. Conf. Ocean, Offshore and Arctic Eng. (OMAE 2012), Rio de Janeiro, Brazil, July 1-6, Scibilia, F., U. Jørgensen, and R. Skjetne (2012), AUV Guidance System for Dynamic Trajectory Generation, in proc. IFAC Workshop on Navigation, Guidance and Control of Underwater Vehicles (NGCUV 2012), Porto, Portugal, April 10-12, Jørgensen, U. and Skjetne, R. (2012), Autonomous Estimation of Drifting Ice Topography Using Continuous Steepest Descent Gradient Minimization, in Proc. American Control Conf., AACC, Montréal, Canada, June Kjerstad, Ø. K. and R. Skjetne (2012), Observer design with disturbance rejection by acceleration feedforward. In proc. 7th IFAC Symposium on Robust Control Design (Rocond'12), Aalborg, Denmark, June 20-22, Kjerstad, Ø. K., Skjetne, R., and Jenssen, N. A. (2011), Disturbance rejection by acceleration feedforward: Application to dynamic positioning, 18th IFAC World Congress, Milano, Italy. Fossen, T. I. (2011), Handbook of Marine Craft Hydrodynamics and Motion Control, ISBN John Wiley & Sons Ltd, UK. Haugen, J., Imsland, L., Løset, S. and Skjetne, R. (2011). Ice Observer System for Ice Management Operations. Proc. 21st Int. Offshore (Ocean) and Polar Eng. Conf., Maui, Hawaii, USA, June 19-24, Hals, T., and Efraimsson, F. (2010). DP Ice Model Test of Arctic Drillship, DP Conference, Houston, USA, Oct Eik, K. J. (2010), Ice Management in Arctic Offshore Operations and Field Developments, Ph.d. thesis, NTNU. Jenssen, N. A., Muddesitti, S., Phillips, D., and Backstrom, K. (2009), DP In Ice Conditions, DP Conference, Houston, USA. Nguyen, D. T., A. H. Sørbø, and A. J. Sørensen (2009), Modelling and Control for Dynamic Positioned Vessels in Level Ice, in proc. 8th Conf. Manoeuvring and Control of Marine Craft (MCMC 2009), Sept , Guarujá, Brazil. Røset, E. S. (2009), Dynamic Positioning of Marine Vessels in Level Ice, M.Sc. Thesis, NTNU. Zhou, L. (2012), Numerical and Experimental Investigation of Station-keeping in Level Ice, Ph.d. thesis, CeSOS, NTNU. Kjerstad, Ø. K., R. Skjetne, and B. O. Berge (2013), Constrained Nullspace-Based Thrust Allocation for Heading Prioritized Stationkeeping of Offshore Vessels in Ice. In proc. 22nd Int. Conf. Port and Ocean Eng. under Arctic Conditions (POAC 2013). Su, B., Ø. K. Kjerstad, R. Skjetne, and T. E. Berg (2013), Ice-going capability assessment and DP-Ice Capability Plot for a double acting intervention vessel in level ice. In proc. 22nd Int. Conf. Port and Ocean Eng. under Arctic Conditions (POAC 2013). Jørgensen, U. and R. Skjetne (2013), Dynamic estimation of a drifting ice topography over a 2D area using mobile underwater measurements. In proc. 22nd Int. Conf. Port and Ocean Eng. under Arctic Conditions (POAC 2013). Haugen, J. and L. Imsland (2013), Optimization-Based Autonomous Remote Sensing of Surface Objects Using an Unmanned Aerial Vehicle. In proc. 12th biannual European Control Conference (ECC'13). Zhang, Q., R. Skjetne, and B. Su (2013), Automatic image segmentation for boundary detection of apparently connected sea-ice floes. In proc. 22nd Int. Conf. Port and Ocean Engineering under Arctic Conditions (POAC 2013).

72 Arctic DP Illustration: Bjarne Stenberg