Parallel Force/Position Crane Control in Marine Operations

IEEE JOURNAL OF OCEANIC ENGINEERING, VOL. 31, NO. 3, JULY 2006 599 Parallel Force/Position Crane Control in Marine Operations Bjørn Skaare and Olav Egeland, Senior Member, IEEE Abstract—A parallel force/position controller is proposed for the control of loads through the wave zone in marine operations. The controller structure has similarities to the parallel force/position control scheme used in robotics. The parallel force/position controller is tested for crane control in simulations and model experiments and the results are presented in this paper. To evaluate the performance of the proposed controller, we study three different control strategies for control of loads through the wave zone: active heave compensation, wave synchronization, and parallel force/position control. The parallel force/position controller gave improved results, in particular, a significant improvement of the minimum value of the wire tension, which is important to avoid snatch loads that may break the wire. The three strategies are tested and compared in simulations and experiments. Index Terms—Crane control, heave compensation, hydrodynamic loads, parallel force/position control, water entry, water exit, wave synchronization. I. INTRODUCTION CRITICAL phase of an offshore crane operation is when the load goes through the splash zone. The performance of the crane system through the splash zone will depend to a large extent on the crane control system. Currently, crane systems are not adequate for operation at high sea states seen during winter conditions in the North Sea. Offshore oil and gas fields will be developed to a large extent with all processing equipment on the seabed in the years to come. This means that high operability on the subsea intervention is required, which in the North Sea and other exposed areas implies underwater intervention in harsh weather conditions. Maintenance, repair, and replacement of equipment are, therefore, also required during these demanding weather conditions. Marine cranes are presently equipped with passive and/or active heave-compensation systems, where the goal is to keep the load motion unaffected by the vessel motion. These systems have been used in the industry for years and are discussed in, e.g., [1]–[5]. A Manuscript received May 22, 2004; accepted December 23, 2005. This work was supported by Norsk Hydro ASA, Oslo, Norway. Associate Editor: H. Maeda. B. Skaare is with Hydro Oil and Energy, Research Centre Bergen, Bergen 5020, Norway (e-mail: [email protected]). O. Egeland is with the Centre of Ships and Ocean Structures, NTNU, Marine Technology Centre, Trondheim N-7491, Norway (e-mail: Olav.Egeland@itk. ntnu.no). Color versions of one or more of the figures in this paper are available online at http://ieeexplore.ieee.org. Digital Object Identifier 10.1109/JOE.2006.880394 The concept of wave synchronization was introduced in [6]–[8], where the load was lowered through the wave zone with a constant speed relative to the waves, to minimize the hydrodynamic forces on the load. An extension of this paper was presented in [9], where wave synchronization was obtained from feedforward control from a wave observer based on measurements of the vessel and load accelerations and the wire tension, to avoid wave-amplitude measurements. Wave synchronization reduces the hydrodynamic forces, but the resulting inertial forces of the load may give large oscillations in the wire tension if the load is heavy and the waves are large. An augmented impedance control scheme denoted as inertance control was introduced for crane control of loads through the wave zone in [10] and [11]. A combination of force and position control is an established control method in the robotics literature. In [12] and [13], it was termed feedforward motion in a force-controlled direction, and in [14]–[16], it was termed parallel force/position control. A parallel force/position control scheme will typically employ a frequency-separation scheme where, loosely speaking, position control dominates in high-frequency range, while force control dominates at low frequencies. This concept will be made clear in the following. The focus of this paper is parallel force/position control of loads through the splash zone. A simulation study with parallel/ force position control of a load through the splash zone was presented in [17]. In this paper, the results from model scale experiments with a crane vessel with a moonpool are presented. In the model scale experiments, the three control strategies, parallel force/position control, active heave compensation, and wave synchronization, are tested during water entry and water exit in regular and irregular waves. The performance of the different control strategies are compared with important key values with respect to the wire tension and the payload acceleration; namely, the minimum value, the maximum value, and the standard deviation are compared for filtered and unfiltered data. II. PRELIMINARIES A. Forces on a Load in Waves The forces acting on a load that is lowered through the wave zone are given from Newton’s second law as (1) where is the mass of the load, is the acceleration of the load, is the acceleration of gravity, is the wire force, and 0364-9059/$20.00 © 2006 IEEE 600 IEEE JOURNAL OF OCEANIC ENGINEERING, VOL. 31, NO. 3, JULY 2006 represents the hydrodynamic forces on the load. The wire force becomes (2) The hydrodynamic forces on a load that is transferred through the wave zone during water entry are derived in [18] from potential theory as (3) has positive direction upwards, is the density of the where water, is the instantaneously submerged volume of the load, is the relative position between the wave and the load, is the added mass of the load as a function of suband mergence. The first term represents the Froude–Kriloff pressure, the second term is due to the hydrostatic pressure, while the last terms are the effect of the added mass and the diffraction force. is zero when is less than or equal to zero, which is the case when the load is hanging in the air. The last term is often denoted as the slamming term and is always positive. Equation (3) is derived under the assumption of a large impact speed, and the added mass terms at infinite frequency should be used. Since marine operations are not performed with those large velocities, the values of the previous terms may be debatable, but this question is outside the scope of this paper. The viscous hydrodynamic forces acting on a load through the wave zone may be determined from model tests, and can be written as Fig. 1. Sketch of experimental setup and definition of coordinate systems. motor inertia (kg/m); motor torque (Nm); payload mass (kg); payload position (m); motor position (m); wire tension (N); (4) is the nonlinear drag coefficient and is the prowhere jected efficient drag area. The total hydrodynamic forces on a load through the wave zone are found as the sum of (3) and (4) hydrodynamic force on payload (N). A sketch of the experimental system with definitions of the different coordinate systems is shown in Fig. 1. The spring in Fig. 1 is introduced to simulate a realistic wire elasticity, and the measured wire tension is given as (8) (5) The winch motor used in the experiments was an alternating current (ac) servomotor with speed control and position encoder measurements. The equation of motion for the motor and the payload can be written as (6) (7) , is the wire tension and is the wire stiffness. C. Fluid Motion in a Moonpool B. Winch-Motor Model where where , The response of the wave elevation inside a moonpool to the waves outside is derived from Bernoulli’s equation in [18], where the linearized dynamics of the wave elevation inside the moonpool was found under the assumption that the wave motion did not vary across the moonpool. The linearized moonpool dynamics was found as (9) , and motor angle (rad); radius of the pulley on the motor shaft (m); where is the wave elevation inside the moonpool, is the acceleration of gravity, is the distance from the free water level to the bottom of the vessel as seen in Fig. 1, and is the time derivative of speed potential at the bottom of the vessel. SKAARE AND EGELAND: PARALLEL FORCE/POSITION CRANE CONTROL IN MARINE OPERATIONS 601 Fig. 3. Block diagram of the ac servomotor with speed control. Fig. 4. Block diagram of the ac servomotor with an inner speed control loop and an outer position control loop. Fig. 2. Moonpool warning on board the Kingfisher vessel. It is assumed that the speed controller is given by (14) It is seen from (9) that the resonance frequency for the wave elevation inside the moonpool is given by (10) where is in the order of a few meters, so that is about 1 rad/s. Thus, the moonpool resonance frequency is within the frequency range of the typical wave motion and this fact is well known in the industry, as seen from the warning on board the Kingfisher vessel shown in Fig. 2. Experiments in [6]–[8] showed that the amplitude of the wave elevation inside the moonpool was more than twice the amplitude of the waves outside the moonpool resonance frequency. III. CONTROL STRATEGIES The block diagram of the dynamic model for the ac motor given by (6) and the speed controller given in (14) is shown in Fig. 3. is found from comparison of the The speed controller to in Fig. 3 and (13) transfer from (15) which gives (16) A. Parallel Force/Position Control The ac servomotor that is considered is equipped with an internal speed control. An experimentally determined first-order model for the transfer function from the motor reference speed to motor speed was found in [19] to be on the form (11) where the time delay was expected to be mainly due to digital communication and control within the motor drive and control units. The phase lag due to the time delay is given as (12) 1) Position Control: Position control can be introduced with the proportional controller (17) A block diagram of the position control system is shown in Fig. 4. The position loop transfer function is given as (18) and the closed-loop transfer function for the position loop is found as is small within the bandwidth of the By assuming that speed controller, the transfer function can be approximated with (13) (19) 602 IEEE JOURNAL OF OCEANIC ENGINEERING, VOL. 31, NO. 3, JULY 2006 Fig. 6. Transformed block diagram of the force-controlled ac servomotor. Fig. 5. Block diagram of parallel force/position control loop. The parameter where is selected to be (26) (20) to cancel the term containing the combined stiffness of the water and the wire in (25). The force-loop transfer function can then be written as (21) The transfer function from the wire tension sition can be found as (27) to the motor powhere (28) (29) (22) (30) 2) Force Control: The force controller is given by (31) (23) where is the desired wire force, is the measured wire tension, is the integration time constant, and is a gain to be determined later. In the parallel force/position control schemes in robotics, the gain is selected as the inverted spring stiffness. The linearized hydrodynamic forces on a half-submerged payload can be written as (24) The block diagram with the force controller (23) based on the motor and model (6) and (7), the wire tension (8), and the position controller (17) is shown in Fig. 5. The block diagram in Fig. 5 can be transformed into the block diagram shown in Fig. 6. The force-loop transfer function can be found from the block diagram in Fig. 6 as (25) (32) contains the transfer function from Note that wire tension to motor position, which is normally small in magnitude for low frequencies. When is small in the sense , it can be neglected in the analysis. that From (27), it is seen that the force-loop transfer function can for low frequencies. be approximated with The integration time constant is chosen to obtain sufficient stability margins. The value of will, therefore, depend on the characteristics of the ac motor and the load parameters , , , and , and the wire elasticity . 3) Parallel Control Scheme: The following parallel force/ position control scheme is proposed for control of loads through the wave zone: (33) (34) where is the estimated vessel motion, is an offset in the desired force to lift or lower the payload, is the submerged is a ramp function from 0 to 1, volume of the payload, and SKAARE AND EGELAND: PARALLEL FORCE/POSITION CRANE CONTROL IN MARINE OPERATIONS 603 TABLE I NUMERICAL DATA FOR EXPERIMENTAL AND FULL-SCALE MODEL starting when the enable signal is set to 1. The sign of the ramp is negative during water entry and positive during water exit. B. Active Heave Compensation The active heave-compensation and wave-synchronization control strategies presented here are identical with those described in [6]–[8]. These strategies are performed with feedforward components from the vessel motion and/or the wave elevation to the motor-speed reference signal. The motor-speed reference signal during active heave compensation is given as Fig. 7. Experimental vessel with moonpool, crane system, and payload. (35) where is the desired motor speed, is the commanded constant motor speed, and is the estimated vessel heave speed. C. Wave Synchronization The desired motor-speed reference signal during wave synchronization is given as (36) where and is the estimated wave speed, is the payload position, (37) Fig. 8. Payload used in experiments. (38) where is the vertical position of the bottom of the vessel is the vertical position where the blending of the acand tive heave-compensation signal with the wave-synchronization signal starts. The purpose of the blending is to avoid unnecessary oscillations in the wire force when the payload is in the air. represents the decay of the wave motion with water depth, which inside a moonpool starts at the bottom of the ship at the vertical position . IV. CASE STUDY the experimental model and the full-scale model are shown in Table I. A picture of the vessel is shown in Fig. 7, and a picture of the load is shown in Fig. 8. The crane system consists of an electric ac motor connected to a payload by a wire that runs over a pulley suspended in a spring as shown in Fig. 1. The spring was designed to simulate a realistic wire elasticity in the scale model with the numerical value shown in Table I. Note that the experimental model does not contain a passive heave-compensation system. Further information about the experimental model is given in [19]–[22]. A. Experimental Model B. Instrumentation The experimental model is a barge with a moonpool, scaled 1 : 30 with respect to the Froude number. Numerical values for The vessel model is equipped with a camera to record the load and wave motion inside the moonpool and an accelerom- 604 IEEE JOURNAL OF OCEANIC ENGINEERING, VOL. 31, NO. 3, JULY 2006 Fig. 9. Bode plot of the transfer functions h (s), h (s), and h (s) with parameter values TABLE II PAYLOAD PARAMETERS k = 10 and T = 0.048 s. The corresponding slamming term of a half-submerged load becomes (41) eter to measure vertical vessel acceleration. Two wave meters are mounted inside the moonpool and one wave meter is placed in the basin, away from the vessel. The payload is equipped with an accelerometer that measures the vertical load acceleration. The load is connected to the winch wire through a force ring that measures the wire tension. The nonlinear viscous drag coefficient for a sphere is given in [24] as 0.47. The numerical values of the parameters for the half-submerged payload and wire are summarized in Table II. C. Linearized Hydrodynamic Forces The parameters of the internal speed controller for the ac motor given by (11) and (12) was given in [19] as The payload to be considered in the experiment is a sphere of diameter 0.09 m with the parameters given in Table I. The values of the hydrodynamic parameters are found from linearization of (5) for a half-submerged load around the relative 0.18 m/s. This correspeed between the wave and the load sponds to 1 m/s for a full-scale system. The high-frequency value of the added mass for a half-submerged sphere is given in [23] as (39) where is the diameter of the sphere, is the density of water, and is the submerged volume of the payload. It is, therefore, assumed that the added mass of the load is proportional to the submerged volume of the load (40) D. Control System rad/s kg s (42) (43) (44) The parameters for the parallel force/position controller was selected as 10, 0.048 s, and the force controller gain was chosen according to (26). Bode plots with transfer functions , , and are shown in Fig. 9. The po9.81 rad/s, while the sition loop crosses the 0 dB line at 6.51rad/s. Note crossover frequency for the force loop is that both control loops are above the resonance frequency of the moonpool given in Table I. It should be noted that for the ac motor, in contradiction to the case with an hydraulic motor in [17], the transfer function has an effect on the transfer function . This is seen from the Bode plot of in Fig. 9, where is large SKAARE AND EGELAND: PARALLEL FORCE/POSITION CRANE CONTROL IN MARINE OPERATIONS Fig. 10. Simulated nondimensional motion of the wave and the bottom of the load during water entry in irregular waves for the different control strategies: parallel force/position control (at the top), active heave compensation (in the middle), and wave synchronization (at the bottom). 605 Fig. 11. Simulated nondimensional hydrodynamic force during water entry in irregular waves for the different control strategies: parallel force/position control (at the top), active heave compensation (in the middle), and wave synchronization (at the bottom). in magnitude for high frequencies and the observed high-frequency resonance does not correspond exactly to the frequency given in (28). E. Simulations Simulations of the system during water entry with the different control strategies, active heave compensation, wave synchronization, and parallel force/position control are shown in Figs. 10–12. The system is exposed to irregular waves generated from the Joint North Sea Wave Atmosphere Program (JONSWAP) wave 1.3 s and signifispectrum with wave period cant wave height 0.1 m. The plots from the simulations and the experiments in Section V are made nondimensional by scaling the time according to the Bis scaling, where the nondimensional time is given as (45) where is the time in seconds, and is given as is the acceleration of gravity, (46) Fig. 12. Simulated nondimensional wire force during water entry in irregular waves for the different control strategies: parallel force/position control (at the top), active heave compensation (in the middle), and wave synchronization (at the bottom). 606 IEEE JOURNAL OF OCEANIC ENGINEERING, VOL. 31, NO. 3, JULY 2006 Fig. 13. Bode plot of the transfer functions h (s) with increased integration time constant and notch filtering with different values of the ramp function. where is the maximum projected area of the load in the -direction. The corresponding nondimensional position and force is then given as (47) (48) where is the density of water, is the position in meter, and is the force in newtons. Nondimensional plots of the motion of the wave and the bottom of the load are shown in Fig. 10. It is seen that the load is “dancing” on top of the wave motion from nondimensional time 150 to 230, after which the load is lowered through the splash zone. The corresponding hydrodynamic force and the wire force are shown in Figs. 11 and 12, respectively. Note that both the wave-synchronization and heave-compensation approach have negative wire force, which is not allowed according to Det Norske Veritas, Oslo, Norway, (DNV) rules [25]. F. Implementation Aspects The parameters of the parallel force/position controller used in the experiments were 10, , and 0.075 s. The value of the integration time constant is higher than the value used in the analysis and simulations 0.048 s. The reason is that the problems with the wire resonance occurred with this value of the integration time constant, even when the wire tension was filtered. The problem with the wire resonance can be avoided with use of a passive heave-compensation system with large elasticity, which will move the wire resonance frequency away from the frequency area of the measurement noise. This is assumed to give a better performance both for the wave-synchronization strategy and the parallel force/position control strategy. The vessel heave position and speed estimator and the waveamplitude speed estimator used in the experiments are identical with the estimators described in [6]–[8] and [20]. Since the noise on the wire-tension measurements was within the frequency area of the wire resonance, the wire-tension measurement was filtered with the second-order notch filter (49) and since the wire resonance is changing with submergence of the load due to the added mass, the notch filter is chosen as (50) 37 rad/s and 20 rad/s were found experiwhere mentally to give good performance. Ideally, the notch frequency should be adapted online to match the observed resonance. A with the Bode plot of the force-loop transfer function notch filter (50) and integration time constant 0.075 s is shown in Fig. 13 for different values of the ramp function . The filtering and increase of the integration time constant resulted in a reduction of the bandwidth of the force controller to SKAARE AND EGELAND: PARALLEL FORCE/POSITION CRANE CONTROL IN MARINE OPERATIONS TABLE III WIRE-TENSION VALUES FOR WATER ENTRY IN REGULAR WAVES TABLE IV FILTERED WIRE-TENSION VALUES FOR WATER ENTRY IN REGULAR WAVES TABLE V PAYLOAD-ACCELERATION VALUES FOR WATER ENTRY IN REGULAR WAVES 607 different strategies for this excitation. The wave elevation inside the moonpool for irregular waves is rather random, and a large number of experiments are necessary before a conclusion is made. Tables III–XVIII and Figs. 14, 17, 20, and 23 in the following sections are presented with both raw and filtered nondimensional experimental data. The filtered data contains mainly the frequency components in the frequency band between 0.5–1.5 Hz, where the significant wave motion is located. The unfiltered data will contain measurement noise, but should not be totally neglected since impulsive hydrodynamic loads may occur. Two experimental runs were done for water entry and water exit, for each control strategy and for each wave system. Some important numerical values, averaged over the runs, are shown in Tables III–XVIII. Plots for the first run of the experiments are shown in Figs. 14–25. For each type of experiment, there are four tables containing key data related to the wire tension and payload accelerations. The data in the tables are calculated for the time interval shown in the corresponding figures and do not include transient parts. The symbols used in the tables are as follows: standard deviation of a highpass filtered signal; minimum value; TABLE VI FILTERED PAYLOAD-ACCELERATION VALUES FOR WATER ENTRY IN REGULAR WAVES maximum value; wire tension, raw; wire tension, filtered; payload acceleration, raw; payload acceleration, filtered; force-control mode; active heave-compensation mode; 5.00 rad/s, 4.96 rad/s, and 4.92 rad/s, for 0, 0.5, and 1, respectively. V. COMPARATIVE EXPERIMENTAL STUDY This section contains the experimental results for water entry and water exit of a sphere in regular and irregular waves. The regular waves were of amplitude 0.05 m with the moonpool res1.3 s, while the irregular waves were genonance period erated from JONSWAP spectrum waves with significant wave height 0.05 m and characteristic wave period 1.3 s. The JONSWAP spectrum waves were generated with wave 2.6 s. components with wave periods in the range 0.4 s To test the different control strategies for harsh conditions, the resonance period of the moonpool was chosen to obtain large waves inside the moonpool. The average moonpool wave elevation for regular waves was more than twice the largest moonpool wave elevation in the experiments conducted in [6]–[8] and [20]. The analysis of the performance of the different control strategies will be based on measurements of the wire tension and the payload accelerations. The experiments in regular waves should be given most attention when analyzing the performance, since the wave elevation inside the moonpool was identical for the wave-synchronization mode; percentage improvement with the parallel force/position control strategy relative to the active heave-compensation strategy; percentage improvement with the parallel force/position control strategy relative to the wave-synchronization strategy. The most critical parameter is assumed to be the minimum tension in the wire since negative wire tension will give snatch loads that may break the wire. A. Water Entry in Regular Waves The key data for the wire tension during water entry in regular waves are presented in Tables III and IV, while the corresponding data for the payload acceleration are shown in Tables V and VI. The force-control strategy shows improvements for all key values related to the wire tension when compared to both active heave compensation and wave synchronization for both raw and filtered data. The largest improvements are for the minimum value of the wire tension for both filtered and unfiltered data, 608 IEEE JOURNAL OF OCEANIC ENGINEERING, VOL. 31, NO. 3, JULY 2006 Fig. 14. Nondimensional raw (cyan) and filtered (red) wire force during water entry in regular waves for the different control strategies: Parallel force/position control (at the top), active heave compensation (in the middle), and wave synchronization (at the bottom). Note that the payload is “dancing” on top of the wave motion until the ramp signal for submergence of the load is given around dimensionless time t 4100 during parallel force/position control. Fig. 15. Nondimensional motor position during water entry in regular waves for the different control strategies: Parallel force/position control (at the top), active heave compensation (in the middle), and wave synchronization (at the bottom). = TABLE IX PAYLOAD-ACCELERATION VALUES FOR WATER ENTRY IN IRREGULAR WAVES TABLE VII WIRE-TENSION VALUES FOR WATER ENTRY IN IRREGULAR WAVES TABLE VIII FILTERED WIRE-TENSION VALUES FOR WATER ENTRY IN IRREGULAR WAVES while the smallest improvements are for the maximum value of the wire tension, especially when compared to active heave compensation. The improvements of the force-control strategy with respect to payload accelerations were smaller than for the wire tension. The largest improvements were obtained for the filtered and unfiltered maximum payload acceleration, while the minimum payload acceleration gave smallest improvement. Active heave compensation gave better performance with respect to minimum unfiltered payload acceleration. TABLE X FILTERED PAYLOAD-ACCELERATION VALUES FOR WATER ENTRY IN IRREGULAR WAVES Figs. 14–16 show the wire force, motor position, and the wave elevation inside the moonpool during the first run with water entry in regular waves for the different control strategies. It is seen from Figs. 14 and 15 that the payload is “dancing” on top of the wave motion until the signal for submergence of the load is given around dimensionless time 4100. When the submergence of the load starts, the load is transferred through the splash zone with a steep ramp that is not particularly affected by the wave motion, as seen in Fig. 15. The reason is that an offset in force before the load hits the water causes the steep ramp, and SKAARE AND EGELAND: PARALLEL FORCE/POSITION CRANE CONTROL IN MARINE OPERATIONS 609 Fig. 16. Nondimensional wave elevation inside the moonpool during water entry in regular waves for the different control strategies: Parallel force/position control (at the top), active heave compensation (in the middle), and wave synchronization (at the bottom). Fig. 17. Nondimensional raw (cyan) and filtered (red) wire force during water entry in irregular waves for the different control strategies: Parallel force/position control (at the top), active heave compensation (in the middle), and wave synchronization (at the bottom). TABLE XI WIRE-TENSION VALUES FOR WATER EXIT IN REGULAR WAVES TABLE XIII PAYLOAD-ACCELERATION VALUES FOR WATER EXIT IN REGULAR WAVES TABLE XII FILTERED WIRE-TENSION VALUES FOR WATER EXIT IN REGULAR WAVES TABLE XIV FILTERED PAYLOAD-ACCELERATION VALUES FOR WATER EXIT IN REGULAR WAVES that the real hydrodynamic forces are smaller than the theoretical hydrodynamic forces used in the simulations. The change of desired wire force was set to occur within one second to submerge the load within one wave period. B. Water Entry in Irregular Waves The key data for the wire tension during water entry in irregular waves are presented in Tables VII and VIII, while the corresponding data for the payload acceleration are shown in Tables IX and X. The force-control strategy shows improvements at most key values related to the wire tension when compared to active heave compensation and wave synchronization. The largest improvements are for the minimum value of the wire tension for both filtered and unfiltered data. The wave-synchronization strategy shows best performance with respect to the standard deviation of the tension for both filtered and unfiltered data, while the active heave-compensation strategy shows best performance with respect to maximum value of the filtered wire tension. The improvements of the force-control strategy with respect to payload accelerations were rather negative for water entry in 610 IEEE JOURNAL OF OCEANIC ENGINEERING, VOL. 31, NO. 3, JULY 2006 Fig. 18. Nondimensional motor position during water entry in irregular waves for the different control strategies: Parallel force/position control (at the top), active heave compensation (in the middle), and wave synchronization (at the bottom). TABLE XV WIRE-TENSION VALUES FOR WATER EXIT IN IRREGULAR WAVES TABLE XVI FILTERED WIRE-TENSION VALUES FOR WATER EXIT IN IRREGULAR WAVES irregular waves. The active heave-compensation strategy and the wave-synchronization strategy showed better performance with respect to the payload acceleration. Figs. 17–19 show the wire force, motor position, and the wave elevation inside the moonpool during the first run with water entry in irregular waves for the different control strategies. Note that the wave motion inside the moonpool is larger at the time the payload is lowered through the wave zone during the forcecontrol strategy than for the other control strategies. This can explain some of the reasons for the poor performance of the force-control strategy with respect to payload accelerations in irregular waves. Fig. 19. Nondimensional wave elevation inside the moonpool during water entry in irregular waves for the different control strategies: Parallel force/position control (at the top), active heave compensation (in the middle), and wave synchronization (at the bottom). TABLE XVII PAYLOAD-ACCELERATION VALUES FOR WATER EXIT IN IRREGULAR WAVES TABLE XVIII PAYLOAD-ACCELERATION VALUES FOR WATER EXIT IN IRREGULAR WAVES C. Water Exit in Regular Waves The key data for the wire tension during water exit in regular waves are presented in Tables XI and XII, while the corresponding data for the payload acceleration are shown in Tables XIII and XIV. The improvements of the force-control strategy with respect to the other control strategies are similar as for water entry in regular waves. The exception is that the maximum value of the filtered wire tension is smaller for active heave compensation than for force control. The minimum value of the wire tension is decreased for all control strategies compared to water entry in SKAARE AND EGELAND: PARALLEL FORCE/POSITION CRANE CONTROL IN MARINE OPERATIONS Fig. 20. Nondimensional raw (cyan) and filtered (red) wire force during water exit in regular waves for the different control strategies: Parallel force/position control (at the top), active heave compensation (in the middle), and wave synchronization (at the bottom). 611 Fig. 21. Nondimensional motor position during water exit in regular waves for the different control strategies: Parallel force/position control (at the top), active heave compensation (in the middle), and wave synchronization (at the bottom). regular waves. There is a possibility for slack wire both for the active heave-compensation and wave-synchronization strategy at this sea state. Figs. 20–22 show the wire force, motor position, and the wave elevation inside the moonpool during the first run with water exit in regular waves for the different control strategies. D. Water Exit in Irregular Waves The key data for the wire tension during water exit in irregular waves are presented in Tables XV and XVI, while the corresponding data for the payload acceleration are shown in Tables XVII and XVIII. The improvements of the force-control strategy are largest for the filtered and unfiltered minimum value of the wire tension compared to active heave compensation, and for the minimum value of the unfiltered wire tension compared to wave synchronization. The smallest improvements were found for the maximum value of both the filtered and unfiltered wire tension. The improvements of the force-control strategy with respect to payload accelerations were largest with respect to the maximum value of the filtered and unfiltered payload acceleration. The performance of the force controller was poorer with respect to the filtered and unfiltered standard deviation of the payload acceleration compared to wave synchronization. Figs. 23–25 show the wire force, the motor position, and the wave elevation inside the moonpool during the first run with water exit in irregular waves for the different control strategies. Fig. 22. Nondimensional wave elevation inside the moonpool during water exit in irregular waves for the different control strategies: Parallel force/position control (at the top), active heave compensation (in the middle), and wave synchronization (at the bottom). 612 IEEE JOURNAL OF OCEANIC ENGINEERING, VOL. 31, NO. 3, JULY 2006 Fig. 23. Nondimensional raw (cyan) and filtered (red) wire force during water exit in irregular waves for the different control strategies: Parallel force/position control (at the top), active heave compensation (in the middle), and wave synchronization (at the bottom). Fig. 24. Nondimensional motor position during water exit in irregular waves for the different control strategies: Parallel force/position control (at the top), active heave compensation (in the middle), and wave synchronization (at the bottom). Similar to the first run with water entry in irregular waves, we note that the wave elevation inside the moonpool shown in Fig. 25 is different for the different control strategies, and will have an influence on the performance in Tables XV–XVIII. VI. DISCUSSION AND CONCLUSION The parallel force/position control strategy showed best performance of the different control strategies with respect to the minimum wire tension, filtered and unfiltered, in all runs. The averaged improvements over two runs in regular waves were in the range of 51%–1061%. This implies that the weather window for marine crane operations can be increased with use of a parallel force/position control strategy. For the averaged results over two runs in regular waves, the force-control strategy showed better performance than the active heave-compensation strategy and the wave-synchronization strategy in all key values with respect to both wire tension and payload acceleration, for both filtered and unfiltered data, during both water entry and water exit, with two minor exceptions. Wave synchronization during water exit in regular waves gave 3.03% better performance than the force-control strategy with respect to the maximum filtered wire tension, while active heave compensation during water entry in regular waves gave 2.81% better performance than the force-control strategy with respect to the minimum value of the filtered payload acceleration. The performance of the force-control strategy and the wavesynchronization strategy is assumed to increase when used in combination with a passive heave-compensation system that can Fig. 25. Nondimensional wave elevation inside the moonpool during water exit in irregular waves for the different control strategies: Parallel force/position control (at the top), active heave compensation (in the middle), and wave synchronization (at the bottom). SKAARE AND EGELAND: PARALLEL FORCE/POSITION CRANE CONTROL IN MARINE OPERATIONS move the resonance frequency of the wire away from the frequency area of the measurement noise. This implies that the notch filtering used in both control strategies can be avoided, and that the integration time constant in the force controller can be decreased. The wave-synchronization strategy could be further improved by using position control instead of speed control, and thereby avoiding the differentiation of the wave elevation measurements. The parallel force/position controller is rather straightforward to implement on a full-scale model equipped with an active heave-compensation system, since measurement of the wire tension is normally available. In general, the controller requires knowledge of the buoyancy of the load with submergence, but for loads with increasing buoyancy with increasing submergence only the submerged buoyancy of the load is necessary. REFERENCES [1] in Proc. Int. Offshore Crane Conf., Kristiansand, Norway, 1998, Norwegian Society of Lifting Technology (NSLT). [2] in Proc. 5th North Sea Offshore Crane Conf., Aberdeen, U.K., 2000, Norwegian Society of Lifting Technology (NSLT). [3] Underwater Lifting Operations. Stavanger, Norway: Norwegian Society of Lifting Technology (NSLT), 2000. [4] U. Korde, “Active heave compensation on drill-ships in irregular waves,” Ocean Eng., vol. 25, pp. 541–561, 1998. [5] P. Sandvik, “Methods for specification of heave compensator performance” MARINTEK, Trondheim, Norway, Tech. Rep. MT87-0194, 1987. [6] S. I. Sagatun, T. A. Johansen, T. I. Fossen, and F. G. Nielsen, “Wave synchronizing crane control during water entry in offshore moonpool operations,” in Proc. IEEE Int. Conf. Control Appl., Glasgow, Scotland, 2002, pp. 174–179. [7] T. Johansen, T. Fossen, S. Sagatun, and F. Nielsen, “Wave synchronizing crane control during water entry in offshore moonpool operations: Experimental results,” IEEE J. Ocean. Eng., vol. 28, no. 4, pp. 720–728, Oct. 2003. [8] T. A. Johansen, T. I. Fossen, S. I. Sagatun, and F. G. Nielsen, “Wave synchronizing crane control during water entry in offshore moonpool operations-experimental results,” Modeling, Identification Control, vol. 25, pp. 29–44, 2004. [9] B. Skaare, O. Egeland, and S. Sagatun, “Adaptive wave synchronization for lowering of crane load,” in Proc. 8th IFAC Conf. Manoeuvering Control Mar. Crafts, Girona, Spain, 2003, pp. 197–202. [10] S. I. Sagatun, T. I. Fossen, and K. P. Lindegaard, “Inertance control of underwater installations,” in Proc. 5th IFAC Conf. Control Appl. Marine Syst., Glasgow, Scotland, 2001. [11] S. I. Sagatun, “Active control of underwater installation,” IEEE Trans. Control Syst. Technol., vol. 10, pp. 743–749, 2002. [12] J. De Schutter, “Compliant motion ii. A control approach based on external control loops,” Int. J. Robot. Res., pp. 7–18, 1988. [13] ——, “Improved force control laws for advanced tracking applications,” in Proc. 1988 IEEE Int. Conf Robot Autom., Philadelphia, PA, 1988. [14] S. Chiaverini, B. Siciliano, and L. Villani, “A stable force/position controller for robot manipulators,” in Proc. 31st IEEE Conf. Decision Control, Tucson, AZ, Dec. 1992, pp. 1869–1874. [15] S. Chiaverini and L. Sciavicco, “The parallel approach to force/position control of robotic manipulators,” IEEE Trans. Robot. Autom., vol. 9, no. 4, pp. 361–373, Aug. 1993. [16] B. Siciliano, “Parallel force/position control of robot manipulators,” in Proc. 7th Int. Symp. Robot. Res., G. Giralt and G. Hirzinger, Eds., London, U.K., 1995, pp. 78–89. 613 [17] B. Skaare and O. Egeland, “Force control of a load through the splash zone,” in Proc. 8th IFAC Conf. Control Appl. Mar. Syst., Ancona, Italy, Jul. 2004, pp. 227–232. [18] O. M. Faltinsen, Sea Loads on Ships and Offshore Structures, 1 ed. Cambridge, U.K.: Cambridge Univ. Press, 1990. [19] T. Fossen and T. Johansen, “Modeling and identification of offshore crane-rig system” NTNU, Tech. Rep. 2001-12-T, 2001 [Online]. Available: http://www.itk.ntnu.no/research/HydroLab/reports.htm [20] T. Johansen and T. Fossen, “Observer and controller design for an offshore Crane moonpool system” NTNU, Tech. Rep. 2001-13-T, May 2002 [Online]. Available: http://www.itk.ntnu.no/research/HydroLab/ reports.htm [21] T. Fossen and S. Sagatun, “Vessel-crane monpool specifications (hydrolaunch)” NTNU, Tech. Rep. 2000-38-L, 2000 [Online]. Available: http://www.itk.ntnu.no/research/HydroLab/reports.htm [22] T. Fossen and S. Sagatun, “Hydrolaunch free decay tests” NTNU, Tech. Rep. 2001-18-T, 2001 [Online]. Available: http://www.itk.ntnu.no/research/HydroLab/reports.htm [23] B. Pettersen, “Lecture notes, sin 1501 marin hydrodynamikk og konstruksjonsteknikk,” in Lecture Notes UK-01-75. Trondheim, Norway: Norwegian University of Science & Technology, Aug. 2001. [24] F. M. White, Fluid Mechanics, 4th ed. New York: McGraw-Hill, 1999. [25] Det Norske Veritas (DNV), “Rules for planning and execution of marine operations” Høvik, Norway, pt. 1, ch. 2, ed. 1, 2000. Bjørn Skaare was born in Oslo, Norway, in 1974. He received the M.Sc. degree in engineering cybernetics from the Norwegian University of Science and Technology (NTNU), Trondheim, Norway, and the University of California at Santa Barbara (UCSB), Santa Barbara, in 1999 and the Ph.D. degree in engineering cybernetics from NTNU, in 2004. He worked as a Software Developer at Aston Technology in 2000. He is currently a Senior Engineer at Hydro Oil and Energy, Research Centre Bergen, Bergen, Norway. His research interests are in modeling, simulation, and control of marine systems. Olav Egeland (S’85–M’86–SM’01) received Siv.Ing. and Dr.Ing. degrees from the Department of Engineering Cybernetics, the Norwegian University of Science and Technology NTNU, Trondheim, Norway, in 1984 and 1987, respectively. He has been a Professor at the Department of Engineering Cybernetics, NTNU, since 1989. In the academic year 1988–1989, he was at the German Aerospace Center in Oberpfaffenhofen, Germany. From 1996 to 1998, he was a Head of the Department of Engineering Cybernetics, a Vice-Dean of Faculty of Electrical Engineering and Telecommunications, and a Member of the Research Committee for Science and Technology at NTNU. He has supervised the graduation of 75 Siv.Ing. and 19 Dr.Ing., and was a Program Manager of the Strategic University Program in Marine Cybernetics at NTNU. Currently, he is a Coordinator of the Control Activity of the Centre of Ships and Ocean Structures. He has wide experience as a consultant for industry, and is a Cofounder of Marine Cybernetics, which is a company at the NTNU incubator. His research interests are in modeling, simulation, and control of mechanical systems with applications to robotics and marine systems. Dr. Egeland was an Associate Editor of the IEEE TRANSACTIONS ON AUTOMATIC CONTROL from 1996 to 1999, and the European Journal of Control, from 1998 to 2000. He received the Automatica Prize Paper Award in 1996, and the 2000 IEEE TRANSACTIONS ON CONTROL SYSTEMS TECHNOLOGY Outstanding Paper Award.